text
stringlengths 313
1.33M
|
---|
# Sound
## Introduction
If a tree falls in a forest (see ) and no one is there to hear it, does it make a sound? The answer to this old philosophical question depends on how you define sound. If sound only exists when someone is around to perceive it, then the falling tree produced no sound. However, in physics, we know that colliding objects can disturb the air, water or other matter surrounding them. As a result of the collision, the surrounding particles of matter began vibrating in a wave-like fashion. This is a sound wave. Consequently, if a tree collided with another object in space, no one would hear it, because no sound would be produced. This is because, in space, there is no air, water or other matter to be disturbed and produce sound waves. In this chapter, we’ll learn more about the wave properties of sound, and explore hearing, as well as some special uses for sound. |
# Sound
## Speed of Sound, Frequency, and Wavelength
### Section Key Terms
### Properties of Sound Waves
Sound is a wave. More specifically, sound is defined to be a disturbance of matter that is transmitted from its source outward. A disturbance is anything that is moved from its state of equilibrium. Some sound waves can be characterized as periodic waves, which means that the atoms that make up the matter experience simple harmonic motion.
A vibrating string produces a sound wave as illustrated in , , and . As the string oscillates back and forth, part of the string’s energy goes into compressing and expanding the surrounding air. This creates slightly higher and lower pressures. The higher pressure... regions are compressions, and the low pressure regions are rarefactions. The pressure disturbance moves through the air as longitudinal waves with the same frequency as the string. Some of the energy is lost in the form of thermal energy transferred to the air. You may recall from the chapter on waves that areas of compression and rarefaction in longitudinal waves (such as sound) are analogous to crests and troughs in transverse waves.
The amplitude of a sound wave decreases with distance from its source, because the energy of the wave is spread over a larger and larger area. But some of the energy is also absorbed by objects, such as the eardrum in , and some of the energy is converted to thermal energy in the air. shows a graph of gauge pressure versus distance from the vibrating string. From this figure, you can see that the compression of a longitudinal wave is analogous to the peak of a transverse wave, and the rarefaction of a longitudinal wave is analogous to the trough of a transverse wave. Just as a transverse wave alternates between peaks and troughs, a longitudinal wave alternates between compression and rarefaction.
### The Speed of Sound
The speed of sound varies greatly depending upon the medium it is traveling through. The speed of sound in a medium is determined by a combination of the medium’s rigidity (or compressibility in gases) and its density. The more rigid (or less compressible) the medium, the faster the speed of sound. The greater the density of a medium, the slower the speed of sound. The speed of sound in air is low, because air is compressible. Because liquids and solids are relatively rigid and very difficult to compress, the speed of sound in such media is generally greater than in gases. shows the speed of sound in various media. Since temperature affects density, the speed of sound varies with the temperature of the medium through which it’s traveling to some extent, especially for gases.
### The Relationship Between the Speed of Sound and the Frequency and Wavelength of a Sound Wave
Sound, like all waves, travels at certain speeds through different media and has the properties of frequency and wavelength. Sound travels much slower than light—you can observe this while watching a fireworks display (see ), since the flash of an explosion is seen before its sound is heard.
The relationship between the speed of sound, its frequency, and wavelength is the same as for all waves:
where v is the speed of sound (in units of m/s), f is its frequency (in units of hertz), and
is its wavelength (in units of meters). Recall that wavelength is defined as the distance between adjacent identical parts of a wave. The wavelength of a sound, therefore, is the distance between adjacent identical parts of a sound wave. Just as the distance between adjacent crests in a transverse wave is one wavelength, the distance between adjacent compressions in a sound wave is also one wavelength, as shown in . The frequency of a sound wave is the same as that of the source. For example, a tuning fork vibrating at a given frequency would produce sound waves that oscillate at that same frequency. The frequency of a sound is the number of waves that pass a point per unit time.
One of the more important properties of sound is that its speed is nearly independent of frequency. If this were not the case, and high-frequency sounds traveled faster, for example, then the farther you were from a band in a football stadium, the more the sound from the low-pitch instruments would lag behind the high-pitch ones. But the music from all instruments arrives in cadence independent of distance, and so all frequencies must travel at nearly the same speed.
Recall that
, and in a given medium under fixed temperature and humidity, v is constant. Therefore, the relationship between f and
is inverse: The higher the frequency, the shorter the wavelength of a sound wave.
The speed of sound can change when sound travels from one medium to another. However, the frequency usually remains the same because it is like a driven oscillation and maintains the frequency of the original source. If v changes and f remains the same, then the wavelength
must change. Since
, the higher the speed of a sound, the greater its wavelength for a given frequency.
### Practice Problems
### Check Your Understanding
### Section Summary
1. Sound is one type of wave.
2. Sound is a disturbance of matter that is transmitted from its source outward in the form of longitudinal waves.
3. The relationship of the speed of sound v, its frequency f, and its wavelength
is given by
, which is the same relationship given for all waves.
4. The speed of sound depends upon the medium through which the sound wave is travelling.
5. In a given medium at a specific temperature (or density), the speed of sound v is the same for all frequencies and wavelengths.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Sound
## Sound Intensity and Sound Level
### Section Key Terms
### Amplitude, Loudness and Energy of a Sound Wave
In a quiet forest, you can sometimes hear a single leaf fall to the ground. But in a traffic jam filled with honking cars, you may have to shout just so the person next to you can hear .The loudness of a sound is related to how energetically its source is vibrating. In cartoons showing a screaming person, the cartoonist often shows an open mouth with a vibrating uvula (the hanging tissue at the back of the mouth) to represent a loud sound coming from the throat. shows such a cartoon depiction of a bird loudly expressing its opinion.
A useful quantity for describing the loudness of sounds is called sound intensity. In general, the intensity of a wave is the power per unit area carried by the wave. Power is the rate at which energy is transferred by the wave. In equation form, intensity I is
where P is the power through an area A. The SI unit for I is W/m2. The intensity of a sound depends upon its pressure amplitude. The relationship between the intensity of a sound wave and its pressure amplitude (or pressure variation Δp) is
where ρ is the density of the material in which the sound wave travels, in units of kg/m3, and v is the speed of sound in the medium, in units of m/s. Pressure amplitude has units of pascals (Pa) or N/m2. Note that Δp is half the difference between the maximum and minimum pressure in the sound wave.
We can see from the equation that the intensity of a sound is proportional to its amplitude squared. The pressure variation is proportional to the amplitude of the oscillation, and so I varies as (Δp)2. This relationship is consistent with the fact that the sound wave is produced by some vibration; the greater its pressure amplitude, the more the air is compressed during the vibration. Because the power of a sound wave is the rate at which energy is transferred, the energy of a sound wave is also proportional to its amplitude squared.
### The Decibel Scale
You may have noticed that when people talk about the loudness of a sound, they describe it in units of decibels rather than watts per meter squared. While sound intensity (in W/m2) is the SI unit, the sound intensity level in decibels (dB) is more relevant for how humans perceive sounds. The way our ears perceive sound can be more accurately described by the logarithm of the intensity of a sound rather than the intensity of a sound directly. The sound intensity level β is defined to be
where I is sound intensity in watts per meter squared, and I0 = 10–12 W/m2 is a reference intensity. I0 is chosen as the reference point because it is the lowest intensity of sound a person with normal hearing can perceive. The decibel level of a sound having an intensity of 10–12 W/m2 is β = 0 dB, because log10 1 = 0. That is, the threshold of human hearing is 0 decibels.
Each factor of 10 in intensity corresponds to 10 dB. For example, a 90 dB sound compared with a 60 dB sound is 30 dB greater, or three factors of 10 (that is, 103 times) as intense. Another example is that if one sound is 107 as intense as another, it is 70 dB higher.
Since β is defined in terms of a ratio, it is unit-less. The unit called decibel (dB) is used to indicate that this ratio is multiplied by 10. The sound intensity level is not the same as sound intensity—it tells you the level of the sound relative to a reference intensity rather than the actual intensity.
### Solving Sound Wave Intensity Problems
### Practice Problems
### Hearing and Voice
People create sounds by pushing air up through their lungs and through elastic folds in the throat called vocal cords. These folds open and close rhythmically, creating a pressure buildup. As air travels up and past the vocal cords, it causes them to vibrate. This vibration escapes the mouth along with puffs of air as sound. A voice changes in pitch when the muscles of the larynx relax or tighten, changing the tension on the vocal chords. A voice becomes louder when air flow from the lungs increases, making the amplitude of the sound pressure wave greater.
Hearing is the perception of sound. It can give us plenty of information—such as pitch, loudness, and direction. Humans can normally hear frequencies ranging from approximately 20 to 20,000 Hz. Other animals have hearing ranges different from that of humans. Dogs can hear sounds as high as 45,000 Hz, whereas bats and dolphins can hear up to 110,000 Hz sounds. You may have noticed that dogs respond to the sound of a dog whistle which produces sound out of the range of human hearing.
Sounds below 20 Hz are called infrasound, whereas those above 20,000 Hz are ultrasound. The perception of frequency is called pitch, and the perception of intensity is called loudness.
The way we hear involves some interesting physics. The sound wave that hits our ear is a pressure wave. The ear converts sound waves into electrical nerve impulses, similar to a microphone.
shows the anatomy of the ear with its division into three parts: the outer ear or ear canal; the middle ear, which runs from the eardrum to the cochlea; and the inner ear, which is the cochlea itself. The body part normally referred to as the ear is technically called the pinna.
The outer ear, or ear canal, carries sound to the eardrum protected inside of the ear. The middle ear converts sound into mechanical vibrations and applies these vibrations to the cochlea. The lever system of the middle ear takes the force exerted on the eardrum by sound pressure variations, amplifies it and transmits it to the inner ear via the oval window. Two muscles in the middle ear protect the inner ear from very intense sounds. They react to intense sound in a few milliseconds and reduce the force transmitted to the cochlea. This protective reaction can also be triggered by your own voice, so that humming during a fireworks display, for example, can reduce noise damage.
shows the middle and inner ear in greater detail. As the middle ear bones vibrate, they vibrate the cochlea, which contains fluid. This creates pressure waves in the fluid that cause the tectorial membrane to vibrate. The motion of the tectorial membrane stimulates tiny cilia on specialized cells called hair cells. These hair cells, and their attached neurons, transform the motion of the tectorial membrane into electrical signals that are sent to the brain.
The tectorial membrane vibrates at different positions based on the frequency of the incoming sound. This allows us to detect the pitch of sound. Additional processing in the brain also allows us to determine which direction the sound is coming from (based on comparison of the sound’s arrival time and intensity between our two ears).
### Check Your Understanding
### Section Summary
1. The intensity of a sound is proportional to its amplitude squared.
2. The energy of a sound wave is also proportional to its amplitude squared.
3. Sound intensity level in decibels (dB) is more relevant for how humans perceive sounds than sound intensity (in W/m2), even though sound intensity is the SI unit.
4. Sound intensity level is not the same as sound intensity—it tells you the level of the sound relative to a reference intensity rather than the actual intensity.
5. Hearing is the perception of sound and involves that transformation of sound waves into vibrations of parts within the ear. These vibrations are then transformed into neural signals that are interpreted by the brain.
6. People create sounds by pushing air up through their lungs and through elastic folds in the throat called vocal cords.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
### References
Nave, R. Vocal sound production—HyperPhysics. Retrieved from http://hyperphysics.phy-astr.gsu.edu/hbase/music/voice.html |
# Sound
## Doppler Effect and Sonic Booms
### Section Key Terms
### The Doppler Effect of Sound Waves
The Doppler effect is a change in the observed pitch of a sound, due to relative motion between the source and the observer. An example of the Doppler effect due to the motion of a source occurs when you are standing still, and the sound of a siren coming from an ambulance shifts from high-pitch to low-pitch as it passes by. The closer the ambulance is to you, the more sudden the shift. The faster the ambulance moves, the greater the shift. We also hear this shift in frequency for passing race cars, airplanes, and trains. An example of the Doppler effect with a stationary source and moving observer is if you ride a train past a stationary warning bell, you will hear the bell’s frequency shift from high to low as you pass by.
What causes the Doppler effect? Let’s compare three different scenarios: Sound waves emitted by a stationary source (), sound waves emitted by a moving source (), and sound waves emitted by a stationary source but heard by moving observers (). In each case, the sound spreads out from the point where it was emitted.
If the source and observers are stationary, then observers on either side see the same wavelength and frequency as emitted by the source. But if the source is moving and continues to emit sound as it travels, then the air compressions (crests) become closer together in the direction in which it’s traveling and farther apart in the direction it’s traveling away from. Therefore, the wavelength is shorter in the direction the source is moving (on the right in ), and longer in the opposite direction (on the left in ).
Finally, if the observers move, as in , the frequency at which they receive the compressions changes. The observer moving toward the source receives them at a higher frequency (and therefore shorter wavelength), and the person moving away from the source receives them at a lower frequency (and therefore longer wavelength).
We know that wavelength and frequency are related by
where v is the fixed speed of sound. The sound moves in a medium and has the same speed v in that medium whether the source is moving or not. Therefore, f multiplied by
is a constant. Because the observer on the right in receives a shorter wavelength, the frequency she perceives must be higher. Similarly, the observer on the left receives a longer wavelength and therefore perceives a lower frequency.
The same thing happens in . A higher frequency is perceived by the observer moving toward the source, and a lower frequency is perceived by an observer moving away from the source. In general, then, relative motion of source and observer toward one another increases the perceived frequency. Relative motion apart decreases the perceived frequency. The greater the relative speed is, the greater the effect.
For a stationary observer and a moving source of sound, the frequency (fobs) of sound perceived by the observer is
where fs is the frequency of sound from a source, vs is the speed of the source along a line joining the source and observer, and vw is the speed of sound. The minus sign is used for motion toward the observer and the plus sign for motion away from the observer.
Note that the greater the speed of the source, the greater the Doppler effect. Similarly, for a stationary source and moving observer, the frequency perceived by the observer fobs is given by
where vobs is the speed of the observer along a line joining the source and observer. Here the plus sign is for motion toward the source, and the minus sign is for motion away from the source.
### Sonic Booms
What happens to the sound produced by a moving source, such as a jet airplane, that approaches or even exceeds the speed of sound? Suppose a jet airplane is coming nearly straight at you, emitting a sound of frequency fs. The greater the plane’s speed, vs, the greater the Doppler shift and the greater the value of fobs. Now, as vs approaches the speed of sound, vw, fobs approaches infinity, because the denominator in
approaches zero.
This result means that at the speed of sound, in front of the source, each wave is superimposed on the previous one because the source moves forward at the speed of sound. The observer gets them all at the same instant, and so the frequency is theoretically infinite. If the source exceeds the speed of sound, no sound is received by the observer until the source has passed, so that the sounds from the source when it was approaching are stacked up with those from it when receding, creating a sonic boom. A sonic boom is a constructive interference of sound created by an object moving faster than sound.
An aircraft creates two sonic booms, one from its nose and one from its tail (see ). During television coverage of space shuttle landings, two distinct booms could often be heard. These were separated by exactly the time it would take the shuttle to pass by a point. Observers on the ground often do not observe the aircraft creating the sonic boom, because it has passed by before the shock wave reaches them. If the aircraft flies close by at low altitude, pressures in the sonic boom can be destructive enough to break windows. Because of this, supersonic flights are banned over populated areas of the United States.
### Solving Problems Using the Doppler Shift Formula
### Practice Problems
### Check Your Understanding
### Section Summary
1. The Doppler effect is a shift in the observed frequency of a sound due to motion of either the source or the observer.
2. The observed frequency is greater than the actual source’s frequency when the source and the observer are moving closer together, either by the source moving toward the observer or the observer moving toward the source.
3. A sonic boom is constructive interference of sound created by an object moving faster than sound.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Sound
## Sound Interference and Resonance
### Section Key Terms
### Resonance and Beats
Sit in front of a piano sometime and sing a loud brief note at it while pushing down on the sustain pedal. It will sing the same note back at you—the strings that have the same frequencies as your voice, are resonating in response to the forces from the sound waves that you sent to them. This is a good example of the fact that objects—in this case, piano strings—can be forced to oscillate but oscillate best at their natural frequency.
A driving force (such as your voice in the example) puts energy into a system at a certain frequency, which is not necessarily the same as the natural frequency of the system. Over time the energy dissipates, and the amplitude gradually reduces to zero- this is called damping. The natural frequency is the frequency at which a system would oscillate if there were no driving and no damping force. The phenomenon of driving a system with a frequency equal to its natural frequency is called resonance, and a system being driven at its natural frequency is said to resonate.
Most of us have played with toys where an object bobs up and down on an elastic band, something like the paddle ball suspended from a finger in . At first you hold your finger steady, and the ball bounces up and down with a small amount of damping. If you move your finger up and down slowly, the ball will follow along without bouncing much on its own. As you increase the frequency at which you move your finger up and down, the ball will respond by oscillating with increasing amplitude. When you drive the ball at its natural frequency, the ball’s oscillations increase in amplitude with each oscillation for as long as you drive it. As the driving frequency gets progressively higher than the resonant or natural frequency, the amplitude of the oscillations becomes smaller, until the oscillations nearly disappear and your finger simply moves up and down with little effect on the ball.
Another example is that when you tune a radio, you adjust its resonant frequency so that it oscillates only at the desired station’s broadcast (driving) frequency. Also, a child on a swing is driven (pushed) by a parent at the swing’s natural frequency to reach the maximum amplitude (height). In all of these cases, the efficiency of energy transfer from the driving force into the oscillator is best at resonance.
All sound resonances are due to constructive and destructive interference. Only the resonant frequencies interfere constructively to form standing waves, while others interfere destructively and are absent. From the toot made by blowing over a bottle to the recognizability of a great singer’s voice, resonance and standing waves play a vital role in sound.
Interference happens to all types of waves, including sound waves. In fact, one way to support that something is a wave is to observe interference effects. shows a set of headphones that employs a clever use of sound interference to cancel noise. To get destructive interference, a fast electronic analysis is performed, and a second sound is introduced with its maxima and minima exactly reversed from the incoming noise.
In addition to resonance, superposition of waves can also create beats. Beats are produced by the superposition of two waves with slightly different frequencies but the same amplitude. The waves alternate in time between constructive interference and destructive interference, giving the resultant wave an amplitude that varies over time. (See the resultant wave in ).
This wave fluctuates in amplitude, or beats, with a frequency called the beat frequency. The equation for beat frequency is
where f1 and f2 are the frequencies of the two original waves. If the two frequencies of sound waves are similar, then what we hear is an average frequency that gets louder and softer at the beat frequency.
### Fundamental Frequency and Harmonics
Suppose we hold a tuning fork near the end of a tube that is closed at the other end, as shown in , , and . If the tuning fork has just the right frequency, the air column in the tube resonates loudly, but at most frequencies it vibrates very little. This means that the air column has only certain natural frequencies. The figures show how a resonance at the lowest of these natural frequencies is formed. A disturbance travels down the tube at the speed of sound and bounces off the closed end. If the tube is just the right length, the reflected sound arrives back at the tuning fork exactly half a cycle later, and it interferes constructively with the continuing sound produced by the tuning fork. The incoming and reflected sounds form a standing wave in the tube as shown.
The standing wave formed in the tube has its maximum air displacement (an antinode) at the open end, and no displacement (a node) at the closed end. Recall from the last chapter on waves that motion is unconstrained at the antinode, and halted at the node. The distance from a node to an antinode is one-fourth of a wavelength, and this equals the length of the tube; therefore,
. This same resonance can be produced by a vibration introduced at or near the closed end of the tube, as shown in .
Since maximum air displacements are possible at the open end and none at the closed end, there are other, shorter wavelengths that can resonate in the tube see ). Here the standing wave has three-fourths of its wavelength in the tube, or
, so that
. There is a whole series of shorter-wavelength and higher-frequency sounds that resonate in the tube.
We use specific terms for the resonances in any system. The lowest resonant frequency is called the fundamental, while all higher resonant frequencies are called overtones. All resonant frequencies are multiples of the fundamental, and are called harmonics. The fundamental is the first harmonic, the first overtone is the second harmonic, and so on. shows the fundamental and the first three overtones (the first four harmonics) in a tube closed at one end.
The fundamental and overtones can be present at the same time in a variety of combinations. For example, the note middle C on a trumpet sounds very different from middle C on a clarinet, even though both instruments are basically modified versions of a tube closed at one end. The fundamental frequency is the same (and usually the most intense), but the overtones and their mix of intensities are different. This mix is what gives musical instruments (and human voices) their distinctive characteristics, whether they have air columns, strings, or drumheads. In fact, much of our speech is determined by shaping the cavity formed by the throat and mouth and positioning the tongue to adjust the fundamental and combination of overtones.
### Open-Pipe and Closed-Pipe Resonators
The resonant frequencies of a tube closed at one end (known as a closed-pipe resonator) are
where f1 is the fundamental, f3 is the first overtone, and so on. Note that the resonant frequencies depend on the speed of sound v and on the length of the tube L.
Another type of tube is one that is open at both ends (known as an open-pipe resonator). Examples are some organ pipes, flutes, and oboes. The air columns in tubes open at both ends have maximum air displacements at both ends. (See ). Standing waves form as shown.
The resonant frequencies of an open-pipe resonator are
where f1 is the fundamental, f2 is the first overtone, f3 is the second overtone, and so on. Note that a tube open at both ends has a fundamental frequency twice what it would have if closed at one end. It also has a different spectrum of overtones than a tube closed at one end. So if you had two tubes with the same fundamental frequency but one was open at both ends and the other was closed at one end, they would sound different when played because they have different overtones.
Middle C, for example, would sound richer played on an open tube since it has more overtones. An open-pipe resonator has more overtones than a closed-pipe resonator because it has even multiples of the fundamental as well as odd, whereas a closed tube has only odd multiples.
In this section we have covered resonance and standing waves for wind instruments, but vibrating strings on stringed instruments also resonate and have fundamentals and overtones similar to those for wind instruments.
### Solving Problems Involving Harmonic Series and Beat Frequency
### Practice Problems
### Check Your Understanding
### Section Summary
1. A system’s natural frequency is the frequency at which the system will oscillate if not affected by driving or damping forces.
2. A periodic force driving a harmonic oscillator at its natural frequency produces resonance. The system is said to resonate.
3. Beats occur when waves of slightly different frequencies are superimposed.
4. In air columns, the lowest-frequency resonance is called the fundamental, whereas all higher resonant frequencies are called overtones. Collectively, they are called harmonics.
5. The resonant frequencies of a tube closed at one end are
, where f is the fundamental and L is the length of the tube.
6. The resonant frequencies of a tube open at both ends are
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Performance Task
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Light
## Introduction
The beauty of a coral reef, the warm radiance of sunshine, the sting of sunburn, the X-ray revealing a broken bone, even microwave popcorn—all are brought to us by electromagnetic waves. The list of the various types of electromagnetic waves, ranging from radio transmission waves to nuclear gamma-ray (γ-ray) emissions, is interesting in itself.
Even more intriguing is that all of these different phenomena are manifestations of the same thing—electromagnetic waves (see ). What are electromagnetic waves? How are they created, and how do they travel? How can we understand their widely varying properties? What is the relationship between electric and magnetic effects? These and other questions will be explored. |
# Light
## The Electromagnetic Spectrum
### Section Key Terms
### The Electromagnetic Spectrum
We generally take light for granted, but it is a truly amazing and mysterious form of energy. Think about it: Light travels to Earth across millions of kilometers of empty space. When it reaches us, it interacts with matter in various ways to generate almost all the energy needed to support life, provide heat, and cause weather patterns. Light is a form of electromagnetic radiation (EMR). The term light usually refers to visible light, but this is not the only form of EMR. As we will see, visible light occupies a narrow band in a broad range of types of electromagnetic radiation.
Electromagnetic radiation is generated by a moving electric charge, that is, by an electric current. As you will see when you study electricity, an electric current generates both an electric field, E, and a magnetic field, B. These fields are perpendicular to each other. When the moving charge oscillates, as in an alternating current, an EM wave is propagated. shows how an electromagnetic wave moves away from the source—indicated by the ~ symbol.
From your study of sound waves, recall these features that apply to all types of waves:
1. Wavelength—The distance between two wave crests or two wave troughs, expressed in various metric measures of distance
2. Frequency—The number of wave crests that pass a point per second, expressed in hertz (Hz or s–1)
3. Amplitude: The height of the crest above the null point
As mentioned, electromagnetic radiation takes several forms. These forms are characterized by a range of frequencies. Because frequency is inversely proportional to wavelength, any form of EMR can also be represented by its range of wavelengths. shows the frequency and wavelength ranges of various types of EMR. With how many of these types are you familiar?
Take a few minutes to study the positions of the various types of radiation on the EM spectrum, above. The narrow band that is visible light extends from lower-frequency red light to higher-frequency violet light. Frequencies just below the visible are called (below red) and those just above are ultraviolet (beyond violet). Radio waves, which overlap with the frequencies used for media broadcasts of TV and radio signals, occupy frequencies even lower than infrared (IR). The microwave radiation that you see on the diagram is the same radiation that is used in a microwave oven. What we feel as radiant heat is also a form of low-frequency EMR. The high-frequency radiation to the right of ultraviolet (UV) includes X-rays and gamma (γ) rays.
### Characteristics of Electromagnetic Radiation
All the EM waves mentioned above are basically the same form of radiation. They can all travel across empty space, and they all travel at the speed of light in a vacuum. The basic difference between types of radiation is their differing frequencies. Each frequency has an associated wavelength. As frequency increases across the spectrum, wavelength decreases. Energy also increases with frequency. Because of this, higher frequencies penetrate matter more readily. Some of the properties and uses of the various EM spectrum bands are listed in .
The narrow band of visible light is a combination of the colors of the rainbow. shows the section of the EM spectrum that includes visible light. The frequencies corresponding to these wavelengths are
at the red end to
at the violet end. This is a very narrow range, considering that the EM spectrum spans about 20 orders of magnitude.
As a child, you probably learned the color wheel, shown on the left in . It helps if you know what color results when you mix different colors of paint together. Mixing two of the primary colors—magenta, yellow, or cyan—together results in a secondary color. For example, mixing cyan and yellow makes green. This is called subtractive color mixing. Mixing different colors of light together is quite different. The diagram on the right shows additive color mixing. In this case, the primary colors are red, green, and blue, and the secondary colors are cyan, magenta, and yellow. Mixing pigments and mixing light are different because materials absorb light by a different set of rules than does the perception of light by the eye. Notice that, when all colors are subtracted, the result is no color, or black. When all colors are added, the result is white light. We see the reverse of this when white sunlight is separated into the visible spectrum by a prism or by raindrops when a rainbow appears in the sky.
Humans have found uses for every part of the electromagnetic spectrum. We will take a look at the uses of each range of frequencies, beginning with visible light. Most of our uses of visible light are obvious; without it our interaction with our surroundings would be much different. We might forget that nearly all of our food depends on the photosynthesis process in plants, and that the energy for this process comes from the visible part of the spectrum. Without photosynthesis, we would also have almost no oxygen in the atmosphere.
The low-frequency, infrared region of the spectrum has many applications in media broadcasting. Television, radio, cell phone, and remote-control devices all broadcast and/or receive signals with these wavelengths. AM and FM radio signals are both low-frequency radiation. They are in different regions of the spectrum, but that is not their basic difference. AM and FM are abbreviations for and . Information in AM signals has the form of changes in amplitude of the radio waves; information in FM signals has the form of changes in wave frequency.
Another application of long-wavelength radiation is found in microwave ovens. These appliances cook or warm food by irradiating it with EM radiation in the microwave frequency range. Most kitchen microwaves use a frequency of
Hz. These waves have the right amount of energy to cause polar molecules, such as water, to rotate faster. Polar molecules are those that have a partial charge separation. The rotational energy of these molecules is given up to surrounding matter as heat. The first microwave ovens were called Radaranges because they were based on radar technology developed during World War II.
Radar uses radiation with wavelengths similar to those of microwaves to detect the location and speed of distant objects, such as airplanes, weather formations, and motor vehicles. Radar information is obtained by receiving and analyzing the echoes of microwaves reflected by an object. The speed of the object can be measured using the Doppler shift of the returning waves. This is the same effect you learned about when you studied sound waves. Like sound waves, EM waves are shifted to higher frequencies by an object moving toward an observer, and to lower frequencies by an object moving away from the observer. Astronomers use this same Doppler effect to measure the speed at which distant galaxies are moving away from us. In this case, the shift in frequency is called the red shift, because visible frequencies are shifted toward the lower-frequency, red end of the spectrum.
Exposure to any radiation with frequencies greater than those of visible light carries some health hazards. All types of radiation in this range are known to cause cell damage. The danger is related to the high energy and penetrating ability of these EM waves. The likelihood of being harmed by any of this radiation depends largely on the amount of exposure. Most people try to reduce exposure to UV radiation from sunlight by using sunscreen and protective clothing. Physicians still use X-rays to diagnose medical problems, but the intensity of the radiation used is extremely low. shows an X-ray image of a patient’s chest cavity.
One medical-imaging technique that involves no danger of exposure is magnetic resonance imaging (MRI). MRI is an important imaging and research tool in medicine, producing highly detailed two- and three-dimensional images. Radio waves are broadcast, absorbed, and reemitted in a resonance process that is sensitive to the density of nuclei, usually hydrogen nuclei—protons.
### Check Your Understanding
### Section Summary
1. The electromagnetic spectrum is made up of a broad range of frequencies of electromagnetic radiation.
2. All frequencies of EM radiation travel at the same speed in a vacuum and consist of an electric field and a magnetic field. The types of EM radiation have different frequencies and wavelengths, and different energies and penetrating ability.
### Concept Items
### Critical Thinking
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Light
## The Behavior of Electromagnetic Radiation
### Section Key Terms
### Types of Electromagnetic Wave Behavior
In a vacuum, all electromagnetic radiation travels at the same incredible speed of 3.00 × 108 m/s, which is equal to 671 million miles per hour. This is one of the fundamental physical constants. It is referred to as the speed of light and is given the symbol c. The space between celestial bodies is a near vacuum, so the light we see from the Sun, stars, and other planets has traveled here at the speed of light. Keep in mind that all EM radiation travels at this speed. All the different wavelengths of radiation that leave the Sun make the trip to Earth in the same amount of time. That trip takes 8.3 minutes. Light from the nearest star, besides the Sun, takes 4.2 years to reach Earth, and light from the nearest galaxy—a dwarf galaxy that orbits the Milky Way—travels 25,000 years on its way to Earth. You can see why we call very long distances astronomical.
When light travels through a physical medium, its speed is always less than the speed of light. For example, light travels in water at three-fourths the value of c. In air, light has a speed that is just slightly slower than in empty space: 99.97 percent of c. Diamond slows light down to just 41 percent of c. When light changes speeds at a boundary between media, it also changes direction. The greater the difference in speeds, the more the path of light bends. In other chapters, we look at this bending, called refraction, in greater detail. We introduce refraction here to help explain a phenomenon called thin-film interference.
Have you ever wondered about the rainbow colors you often see on soap bubbles, oil slicks, and compact discs? This occurs when light is both refracted by and reflected from a very thin film. The diagram shows the path of light through such a thin film. The symbols n1, n2, and n3 indicate that light travels at different speeds in each of the three materials. Learn more about this topic in the chapter on diffraction and interference.
shows the result of thin film interference on the surface of soap bubbles. Because ray 2 travels a greater distance, the two rays become out of phase. That is, the crests of the two emerging waves are no longer moving together. This causes interference, which reinforces the intensity of the wavelengths of light that create the bands of color. The color bands are separated because each color has a different wavelength. Also, the thickness of the film is not uniform, and different thicknesses cause colors of different wavelengths to interfere in different places. Note that the film must be very, very thin—somewhere in the vicinity of the wavelengths of visible light.
You have probably experienced how polarized sunglasses reduce glare from the surface of water or snow. The effect is caused by the wave nature of light. Looking back at , we see that the electric field moves in only one direction perpendicular to the direction of propagation. Light from most sources vibrates in all directions perpendicular to propagation. Light with an electric field that vibrates in only one direction is called polarized. A diagram of polarized light would look like .
Polarized glasses are an example of a polarizing filter. These glasses absorb most of the horizontal light waves and transmit the vertical waves. This cuts down glare, which is caused by horizontal waves. shows how waves traveling along a rope can be used as a model of how a polarizing filter works. The oscillations in one rope are in a vertical plane and are said to be vertically polarized. Those in the other rope are in a horizontal plane and are horizontally polarized. If a vertical slit is placed on the first rope, the waves pass through. However, a vertical slit blocks the horizontally polarized waves. For EM waves, the direction of the electric field oscillation is analogous to the disturbances on the ropes.
Light can also be polarized by reflection. Most of the light reflected from water, glass, or any highly reflective surface is polarized horizontally. shows the effect of a polarizing lens on light reflected from the surface of water.
### Quantitative Treatment of Electromagnetic Waves
We can use the speed of light, c, to carry out several simple but interesting calculations. If we know the distance to a celestial object, we can calculate how long it takes its light to reach us. Of course, we can also make the reverse calculation if we know the time it takes for the light to travel to us. For an object at a very great distance from Earth, it takes many years for its light to reach us. This means that we are looking at the object as it existed in the distant past. The object may, in fact, no longer exist. Very large distances in the universe are measured in light years. One light year is the distance that light travels in one year, which is
kilometers or
miles (…and 1012 is a trillion!).
A useful equation involving c is
where f is frequency in Hz, and
is wavelength in meters.
The rate at which light is radiated from a source is called luminous flux, P, and it is measured in lumens (lm). Energy-saving light bulbs, which provide more luminous flux for a given use of electricity, are now available. One of these bulbs is called a compact fluorescent lamp; another is an LED (light-emitting diode) bulb. If you wanted to replace an old incandescent bulb with an energy saving bulb, you would want the new bulb to have the same brightness as the old one. To compare bulbs accurately, you would need to compare the lumens each one puts out. Comparing wattage—that is, the electric power used—would be misleading. Both wattage and lumens are stated on the packaging.
The luminous flux of a bulb might be 2,000 lm. That accounts for all the light radiated in all directions. However, what we really need to know is how much light falls on an object, such as a book, at a specific distance. The number of lumens per square meter is called illuminance, and is given in units of lux (lx). Picture a light bulb in the middle of a sphere with a 1-m radius. The total surface of the sphere equals 4πr2 m2. The illuminance then is given by
What happens if the radius of the sphere is increased 2 m? The illuminance is now only one-fourth as great, because the r2 term in the denominator is 4 instead of 1. shows how illuminance decreases with the inverse square of the distance.
### Practice Problems
### Check Your Understanding
### Section Summary
1. EM radiation travels at different speeds in different media, produces colors on thin films, and can be polarized to oscillate in only one direction.
2. Calculations can be based on the relationship among the speed, frequency, and wavelength of light, and on the relationship among luminous flux, illuminance, and distance.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Performance Task
1. EYE SAFETY—Chemicals in this lab are poisonous if ingested. If chemicals are ingested, inform your teacher immediately.
2. FUMES—Certain chemicals or chemical reactions in this lab create a vapor that is harmful if inhaled. Follow your teacher's instructions for the use of fume hoods and other safety apparatus designed to prevent fume inhalation. Never smell or otherwise breath in any chemicals or vapors in the lab.
3. FLAMMABLE—Chemicals in this lab are highly flammable and can ignite, especially if exposed to a spark or open flame. Follow your teacher's instructions carefully on how to handle flammable chemicals. Do not expose any chemical to a flame or other heat source unless specifically instructed by your teacher.
4. HAND WASHING—Some materials may be hazardous if in extended contact with the skin. Be sure to wash your hands with soap after handling and disposing of these materials during the lab.
5. WASTE—Some things in this lab are hazardous and need to be disposed of properly. Follow your teacher's instructions for disposal of all items.
1. A large flat tray with raised sides, such as a baking tray
2. Small volumes of motor oil, lighter fluid or a penetrating oil of the type used to loosen rusty bolts, and cooking oil
3. Water
4. A camera
1. Thin-film interference causes colors to appear on the surface of a thin transparent layer. Do you expect to see a pattern to the colors?
2. How could you make a permanent record of your observations?
3. What data would you need to look up to help explain any patterns that you see?
4. What could explain colors failing to appear under some conditions?
### Test Prep Multiple Choice
### Short Answer
### Extended Response
|
# Mirrors and Lenses
## Introduction
In another moment Alice was through the glass, and had jumped lightly down into the Looking-glass room.
—Through the Looking Glass by Lewis Carol
Through the Looking Glass tells of the adventures of Alice after she steps from the real world, through a mirror, and into the virtual world. In this chapter we examine the optical meanings of real and virtual, as well as other concepts that make up the field of optics.
The light from this page or screen is formed into an image by the lens of your eyes, much as the lens of the camera that made the photograph at the beginning of this chapter. Mirrors, like lenses, can also form images, which in turn are captured by your eyes.
Optics is the branch of physics that deals with the behavior of visible light and other electromagnetic waves. For now, we concentrate on the propagation of light and its interaction with matter.
It is convenient to divide optics into two major parts based on the size of objects that light encounters. When light interacts with an object that is several times as large as the light’s wavelength, its observable behavior is similar to a ray; it does not display its wave characteristics prominently. We call this part of optics geometric optics. This chapter focuses on situations for which geometric optics is suited. |
# Mirrors and Lenses
## Reflection
### Section Key Terms
### Characteristics of Mirrors
There are three ways, as shown in , in which light can travel from a source to another location. It can come directly from the source through empty space, such as from the Sun to Earth. Light can travel to an object through various media, such as air and glass. Light can also arrive at an object after being reflected, such as by a mirror. In all these cases, light is modeled as traveling in a straight line, called a . Light may change direction when it encounters the surface of a different material (such as a mirror) or when it passes from one material to another (such as when passing from air into glass). It then continues in a straight line—that is, as a ray. The word ray comes from mathematics. Here it means a straight line that originates from some point. It is acceptable to visualize light rays as laser rays (or even science fiction depictions of ray guns).
Because light moves in straight lines, that is, as rays, and changes directions when it interacts with matter, it can be described through geometry and trigonometry. This part of optics, described by straight lines and angles, is therefore called . There are two laws that govern how light changes direction when it interacts with matter: the law of reflection, for situations in which light bounces off matter; and the law of refraction, for situations in which light passes through matter. In this section, we consider the geometric optics of reflection.
Whenever we look into a mirror or squint at sunlight glinting from a lake, we are seeing a reflection. How does the reflected light travel from the object to your eyes? The law of reflection states: The angle of reflection,
, equals the angle of incidence,
. This law governs the behavior of all waves when they interact with a smooth surface, and therefore describe the behavior of light waves as well. The reflection of light is simplified when light is treated as a ray. This concept is illustrated in , which also shows how the angles are measured relative to the line perpendicular to the surface at the point where the light ray strikes it. This perpendicular line is also called the normal line, or just the normal. Light reflected in this way is referred to as specular (from the Latin word for mirror: speculum).
We expect to see reflections from smooth surfaces, but , illustrates how a rough surface reflects light. Because the light is reflected from different parts of the surface at different angles, the rays go in many different directions, so the reflected light is diffused. Diffused light allows you to read a printed page from almost any angle because some of the rays go in different directions. Many objects, such as people, clothing, leaves, and walls, have rough surfaces and can be seen from many angles. A mirror, on the other hand, has a smooth surface and reflects light at specific angles.
When we see ourselves in a mirror, it appears that our image is actually behind the mirror. We see the light coming from a direction determined by the law of reflection. The angles are such that our image is exactly the same distance behind the mirror, di, as the distance we stand away from the mirror, do. Although these mirror images make objects appear to be where they cannot be (such as behind a solid wall), the images are not figments of our imagination. Mirror images can be photographed and videotaped by instruments and look just as they do to our eyes, which are themselves optical instruments. An image in a mirror is said to be a virtual image, as opposed to a real image. A virtual image is formed when light rays appear to diverge from a point without actually doing so.
helps illustrate how a flat mirror forms an image. Two rays are shown emerging from the same point, striking the mirror, and reflecting into the observer’s eye. The rays can diverge slightly, and both still enter the eye. If the rays are extrapolated backward, they seem to originate from a common point behind the mirror, allowing us to locate the image. The paths of the reflected rays into the eye are the same as if they had come directly from that point behind the mirror. Using the law of reflection—the angle of reflection equals the angle of incidence—we can see that the image and object are the same distance from the mirror. This is a virtual image, as defined earlier.
Some mirrors are curved instead of flat. A mirror that curves inward is called a , whereas one that curves outward is called a . Pick up a well-polished metal spoon and you can see an example of each type of curvature. The side of the spoon that holds the food is a concave mirror; the back of the spoon is a convex mirror. Observe your image on both sides of the spoon.
Ray diagrams can be used to find the point where reflected rays converge or appear to converge, or the point from which rays appear to diverge. This is called the , F. The distance from F to the mirror along the central axis (the line perpendicular to the center of the mirror’s surface) is called the , f. shows the focal points of concave and convex mirrors.
Images formed by a concave mirror vary, depending on which side of the focal point the object is placed. For any object placed on the far side of the focal point with respect to the mirror, the rays converge in front of the mirror to form a real image, which can be projected onto a surface, such as a screen or sheet of paper However, for an object located inside the focal point with respect to the concave mirror, the image is virtual. For a convex mirror the image is always virtual—that is, it appears to be behind the mirror. The ray diagrams in show how to determine the nature of the image formed by concave and convex mirrors.
The information in is summarized in .
You should be able to notice everyday applications of curved mirrors. One common example is the use of security mirrors in stores, as shown in .
Some telescopes also use curved mirrors and no lenses (except in the eyepieces) both to magnify images and to change the path of light. shows a Schmidt-Cassegrain telescope. This design uses a spherical primary concave mirror and a convex secondary mirror. The image is projected onto the focal plane by light passing through the perforated primary mirror. The effective focal length of such a telescope is the focal length of the primary mirror multiplied by the magnification of the secondary mirror. The result is a telescope with a focal length much greater than the length of the telescope itself.
A parabolic concave mirror has the very useful property that all light from a distant source, on reflection by the mirror surface, is directed to the focal point. Likewise, a light source placed at the focal point directs all the light it emits in parallel lines away from the mirror. This case is illustrated by the ray diagram in . The light source in a car headlight, for example, is located at the focal point of a parabolic mirror.
Parabolic mirrors are also used to collect sunlight and direct it to a focal point, where it is transformed into heat, which in turn can be used to generate electricity. This application is shown in .
### The Application of the Curved Mirror Equations
Curved mirrors and the images they create involve a fairly small number of variables: the mirror’s radius of curvature, R; the focal length, f; the distances of the object and image from the mirror, d and d, respectively; and the heights of the object and image, h and h, respectively. The signs of these values indicate whether the image is inverted, erect (upright), real, or virtual. We now look at the equations that relate these variables and apply them to everyday problems.
shows the meanings of most of the variables we will use for calculations involving curved mirrors.
The basic equation that describes both lenses and mirrors is the lens/mirror equation
This equation can be rearranged several ways. For example, it may be written to solve for focal length.
Magnification, m, is the ratio of the size of the image, h, to the size of the object, h. The value of m can be calculated in two ways.
This relationship can be written to solve for any of the variables involved. For example, the height of the image is given by
We saved the simplest equation for last. The radius of curvature of a curved mirror, R, is simply twice the focal length.
We can learn important information from the algebraic sign of the result of a calculation using the previous equations:
1. A negative d indicates a virtual image; a positive value indicates a real image
2. A negative h indicates an inverted image; a positive value indicates an erect image
3. For concave mirrors, f is positive; for convex mirrors, f is negative
Now let’s apply these equations to solve some problems.
### Practice Problems
### Check Your Understanding
### Section Summary
1. The angle of reflection equals the angle of incidence.
2. Plane mirrors and convex mirrors reflect virtual, erect images. Concave mirrors reflect light to form real, inverted images or virtual, erect images, depending on the location of the object.
3. Image distance, height, and other characteristics can be calculated using the lens/mirror equation and the magnification equation.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Mirrors and Lenses
## Refraction
### Section Key Terms
### The Law of Refraction
You may have noticed some odd optical phenomena when looking into a fish tank. For example, you may see the same fish appear to be in two different places (). This is because light coming to you from the fish changes direction when it leaves the tank and, in this case, light rays traveling along two different paths both reach our eyes. The changing of a light ray’s direction (loosely called bending) when it passes a boundary between materials of different composition, or between layers in single material where there are changes in temperature and density, is called refraction. Refraction is responsible for a tremendous range of optical phenomena, from the action of lenses to voice transmission through optical fibers.
Why does light change direction when passing from one material (medium) to another? It is because light changes speed when going from one material to another. This behavior is typical of all waves and is especially easy to apply to light because light waves have very small wavelengths, and so they can be treated as rays. Before we study the law of refraction, it is useful to discuss the speed of light and how it varies between different media.
The speed of light is now known to great precision. In fact, the speed of light in a vacuum, c, is so important, and is so precisely known, that it is accepted as one of the basic physical quantities, and has the fixed value
where the approximate value of 3.00
108 m/s is used whenever three-digit precision is sufficient. The speed of light through matter is less than it is in a vacuum, because light interacts with atoms in a material. The speed of light depends strongly on the type of material, given that its interaction with different atoms, crystal lattices, and other substructures varies. We define the index of refraction, n, of a material to be
where v is the observed speed of light in the material. Because the speed of light is always less than c in matter and equals c only in a vacuum, the index of refraction (plural: indices of refraction) is always greater than or equal to one.
lists the indices of refraction in various common materials.
provides an analogy for and a description of how a ray of light changes direction when it passes from one medium to another. As in the previous section, the angles are measured relative to a perpendicular to the surface at the point where the light ray crosses it. The change in direction of the light ray depends on how the speed of light changes. The change in the speed of light is related to the indices of refraction of the media involved. In the situations shown in , medium 2 has a greater index of refraction than medium 1. This difference in index of refraction means that the speed of light is less in medium 2 than in medium 1. Note that, in (a), the path of the ray moves closer to the perpendicular when the ray slows down. Conversely, in (b), the path of the ray moves away from the perpendicular when the ray speeds up. The path is exactly reversible. In both cases, you can imagine what happens by thinking about pushing a lawn mower from a footpath onto grass, and vice versa. Going from the footpath to grass, the right front wheel is slowed and pulled to the side as shown. This is the same change in direction for light when it goes from a fast medium to a slow one. When going from the grass to the footpath, the left front wheel moves faster than the others, and the mower changes direction as shown. This, too, is the same change in direction as light going from slow to fast.
The amount that a light ray changes direction depends both on the incident angle and the amount that the speed changes. For a ray at a given incident angle, a large change in speed causes a large change in direction, and thus a large change in the angle of refraction. The exact mathematical relationship is the law of refraction, or , which is stated in equation form as
In terms of speeds, Snell’s law becomes
Here, n1 and n2 are the indices of refraction for media 1 and 2, respectively, and θ1 and θ2 are the angles between the rays and the perpendicular in the respective media 1 and 2, as shown in . The incoming ray is called the and the outgoing ray is called the . The associated angles are called the angle of incidence and the angle of refraction. Later, we apply Snell’s law to some practical situations.
Dispersion is defined as the spreading of white light into the wavelengths of which it is composed. This happens because the index of refraction varies slightly with wavelength. shows how a prism disperses white light into the colors of the rainbow.
Rainbows are produced by a combination of refraction and reflection. You may have noticed that you see a rainbow only when you turn your back to the Sun. Light enters a drop of water and is reflected from the back of the drop, as shown in . The light is refracted both as it enters and as it leaves the drop. Because the index of refraction of water varies with wavelength, the light is dispersed and a rainbow is observed.
A good-quality mirror reflects more than 90 percent of the light that falls on it; the mirror absorbs the rest. But, it would be useful to have a mirror that reflects all the light that falls on it. Interestingly, we can produce total reflection using an aspect of refraction. Consider what happens when a ray of light strikes the surface between two materials, such as is shown in (a). Part of the light crosses the boundary and is refracted; the rest is reflected. If, as shown in the figure, the index of refraction for the second medium is less than the first, the ray bends away from the perpendicular. Because n1 > n2, the angle of refraction is greater than the angle of incidence—that is,
>
. Now, imagine what happens as the incident angle is increased. This causes
to increase as well. The largest the angle of refraction,
, can be is 90°, as shown in (b). The critical angle,
, for a combination of two materials is defined to be the incident angle,
, which produces an angle of refraction of 90°. That is,
is the incident angle for which
= 90°. If the incident angle,
, is greater than the critical angle, as shown in (c), then all the light is reflected back into medium 1, a condition called .
Recall that Snell’s law states the relationship between angles and indices of refraction. It is given by
When the incident angle equals the critical angle (
=
), the angle of refraction is 90° (
= 90°). Noting that sin 90° = 1, Snell’s law in this case becomes
The critical angle,
, for a given combination of materials is thus
for n1 > n2.
Total internal reflection occurs for any incident angle greater than the critical angle,
, and it can only occur when the second medium has an index of refraction less than the first. Note that the previous equation is written for a light ray that travels in medium 1 and reflects from medium 2, as shown in .
There are several important applications of total internal reflection. Total internal reflection, coupled with a large index of refraction, explains why diamonds sparkle more than other materials. The critical angle for a diamond-to-air surface is only 24.4°; so, when light enters a diamond, it has trouble getting back out (). Although light freely enters the diamond at different angles, it can exit only if it makes an angle less than 24.4° with the normal to a given surface. Facets on diamonds are specifically intended to make this unlikely, so that the light can exit only in certain places. Diamonds with very few impurities are very clear, so the light makes many internal reflections and is concentrated at the few places it can exit—hence the sparkle.
A light ray that strikes an object that consists of two mutually perpendicular reflecting surfaces is reflected back exactly parallel to the direction from which it came. This parallel reflection is true whenever the reflecting surfaces are perpendicular, and it is independent of the angle of incidence. Such an object is called a because the light bounces from its inside corner. Many inexpensive reflector buttons on bicycles, cars, and warning signs have corner reflectors designed to return light in the direction from which it originates. Corner reflectors are perfectly efficient when the conditions for total internal reflection are satisfied. With common materials, it is easy to obtain a critical angle that is less than 45°. One use of these perfect mirrors is in binoculars, as shown in . Another application is for periscopes used in submarines.
Fiber optics are one common application of total internal reflection. In communications, fiber optics are used to transmit telephone, internet, and cable TV signals, and they use the transmission of light down fibers of plastic or glass. Because the fibers are thin, light entering one is likely to strike the inside surface at an angle greater than the critical angle and, thus, be totally reflected (). The index of refraction outside the fiber must be smaller than inside, a condition that is satisfied easily by coating the outside of the fiber with a material that has an appropriate refractive index. In fact, most fibers have a varying refractive index to allow more light to be guided along the fiber through total internal reflection. Rays are reflected around corners as shown in the figure, making the fibers into tiny light pipes.
### Calculations with the Law of Refraction
The calculation problems that follow require application of the following equations:
and
These are the equations for refractive index, the mathematical statement of the law of refraction (Snell’s law), and the equation for the critical angle.
### Practice Problems
### Check Your Understanding
### Section Summary
1. The index of refraction for a material is given by the speed of light in a vacuum divided by the speed of light in that material.
2. Snell’s law states the relationship between indices of refraction, the incident angle, and the angle of refraction.
3. The critical angle,
, determines whether total internal refraction can take place, and can be calculated according to
.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Mirrors and Lenses
## Lenses
### Section Key Terms
### Characteristics of Lenses
Lenses are found in a huge array of optical instruments, ranging from a simple magnifying glass to the eye to a camera’s zoom lens. In this section, we use the law of refraction to explore the properties of lenses and how they form images.
Some of what we learned in the earlier discussion of curved mirrors also applies to the study of lenses. Concave, convex, focal point F, and focal length f have the same meanings as before, except each measurement is made from the center of the lens instead of the surface of the mirror. The convex lens shown in has been shaped so that all light rays that enter it parallel to its central axis cross one another at a single point on the opposite side of the lens. The central axis, or axis, is defined to be a line normal to the lens at its center. Such a lens is called a converging lens because of the converging effect it has on light rays. An expanded view of the path of one ray through the lens is shown in to illustrate how the ray changes direction both as it enters and as it leaves the lens. Because the index of refraction of the lens is greater than that of air, the ray moves toward the perpendicular as it enters and away from the perpendicular as it leaves. (This is in accordance with the law of refraction.) As a result of the shape of the lens, light is thus bent toward the axis at both surfaces.
Note that rays from a light source placed at the focal point of a converging lens emerge parallel from the other side of the lens. You may have heard of the trick of using a converging lens to focus rays of sunlight to a point. Such a concentration of light energy can produce enough heat to ignite paper.
shows a concave lens and the effect it has on rays of light that enter it parallel to its axis (the path taken by ray 2 in the figure is the axis of the lens). The concave lens is a diverging lens because it causes the light rays to bend away (diverge) from its axis. In this case, the lens has been shaped so all light rays entering it parallel to its axis appear to originate from the same point, F, defined to be the focal point of a diverging lens. The distance from the center of the lens to the focal point is again called the focal length, or “ƒ,” of the lens. Note that the focal length of a diverging lens is defined to be negative. An expanded view of the path of one ray through the lens is shown in to illustrate how the shape of the lens, together with the law of refraction, causes the ray to follow its particular path and diverge.
The power, P, of a lens is very easy to calculate. It is simply the reciprocal of the focal length, expressed in meters
The units of power are diopters, D, which are expressed in reciprocal meters. If the focal length is negative, as it is for the diverging lens in , then the power is also negative.
In some circumstances, a lens forms an image at an obvious location, such as when a movie projector casts an image onto a screen. In other cases, the image location is less obvious. Where, for example, is the image formed by eyeglasses? We use ray tracing for thin lenses to illustrate how they form images, and we develop equations to describe the image-formation quantitatively. These are the rules for ray tracing:
1. A ray entering a converging lens parallel to its axis passes through the focal point, F, of the lens on the other side
2. A ray entering a diverging lens parallel to its axis seems to come from the focal point, F, on the side of the entering ray
3. A ray passing through the center of either a converging or a diverging lens does not change direction
4. A ray entering a converging lens through its focal point exits parallel to its axis
5. A ray that enters a diverging lens by heading toward the focal point on the opposite side exits parallel to the axis
Consider an object some distance away from a converging lens, as shown in . To find the location and size of the image formed, we trace the paths of select light rays originating from one point on the object. In this example, the originating point is the top of a woman’s head. shows three rays from the top of the object that can be traced using the ray-tracing rules just listed. Rays leave this point traveling in many directions, but we concentrate on only a few, which have paths that are easy to trace. The first ray is one that enters the lens parallel to its axis and passes through the focal point on the other side (rule 1). The second ray passes through the center of the lens without changing direction (rule 3). The third ray passes through the nearer focal point on its way into the lens and leaves the lens parallel to its axis (rule 4). All rays that come from the same point on the top of the person’s head are refracted in such a way as to cross at the same point on the other side of the lens. The image of the top of the person’s head is located at this point. Rays from another point on the object, such as the belt buckle, also cross at another common point, forming a complete image, as shown. Although three rays are traced in , only two are necessary to locate the image. It is best to trace rays for which there are simple ray-tracing rules. Before applying ray tracing to other situations, let us consider the example shown in in more detail.
The image formed in is a real image—meaning, it can be projected. That is, light rays from one point on the object actually cross at the location of the image and can be projected onto a screen, a piece of film, or the retina of an eye.
In , the object distance, d, is greater than f. Now we consider a ray diagram for a convex lens where d f, and another diagram for a concave lens.
The examples in and represent the three possible cases—case 1, case 2, and case 3—summarized in . In the table, m is magnification; the other symbols have the same meaning as they did for curved mirrors.
Image formation by lenses can also be calculated from simple equations. We learn how these calculations are carried out near the end of this section.
Some common applications of lenses with which we are all familiar are magnifying glasses, eyeglasses, cameras, microscopes, and telescopes. We take a look at the latter two examples, which are the most complex. We have already seen the design of a telescope that uses only mirrors in . shows the design of a telescope that uses two lenses. Part (a) of the figure shows the design of the telescope used by Galileo. It produces an upright image, which is more convenient for many applications. Part (b) shows an arrangement of lenses used in many astronomical telescopes. This design produces an inverted image, which is less of a problem when viewing celestial objects.
shows the path of light through a typical microscope. Microscopes were first developed during the early 1600s by eyeglass makers in the Netherlands and Denmark. The simplest compound microscope is constructed from two convex lenses, as shown schematically in . The first lens is called the objective lens; it has typical magnification values from 5
to 100
. In standard microscopes, the objectives are mounted such that when you switch between them, the sample remains in focus. Objectives arranged in this way are described as . The second lens, the eyepiece, also referred to as the , has several lenses that slide inside a cylindrical barrel. The focusing ability is provided by the movement of both the objective lens and the eyepiece. The purpose of a microscope is to magnify small objects, and both lenses contribute to the final magnification. In addition, the final enlarged image is produced in a location far enough from the observer to be viewed easily because the eye cannot focus on objects or images that are too close.
Real lenses behave somewhat differently from how they are modeled using rays diagrams or the thin-lens equations. Real lenses produce aberrations. An aberration is a distortion in an image. There are a variety of aberrations that result from lens size, material, thickness, and the position of the object. One common type of aberration is chromatic aberration, which is related to color. Because the index of refraction of lenses depends on color, or wavelength, images are produced at different places and with different magnifications for different colors. The law of reflection is independent of wavelength, so mirrors do not have this problem. This result is another advantage for the use of mirrors in optical systems such as telescopes.
(a) shows chromatic aberration for a single convex lens, and its partial correction with a two-lens system. The index of refraction of the lens increases with decreasing wavelength, so violet rays are refracted more than red rays, and are thus focused closer to the lens. The diverging lens corrects this in part, although it is usually not possible to do so completely. Lenses made of different materials and with different dispersions may be used. For example, an achromatic doublet consisting of a converging lens made of crown glass in contact with a diverging lens made of flint glass can reduce chromatic aberration dramatically ((b)).
### Physics of the Eye
The eye is perhaps the most interesting of all optical instruments. It is remarkable in how it forms images and in the richness of detail and color they eye can detect. However, our eyes commonly need some correction to reach what is called normal vision, but should be called ideal vision instead. Image formation by our eyes and common vision correction are easy to analyze using geometric optics. shows the basic anatomy of the eye. The cornea and lens form a system that, to a good approximation, acts as a single thin lens. For clear vision, a real image must be projected onto the light-sensitive retina, which lies at a fixed distance from the lens. The lens of the eye adjusts its power to produce an image on the retina for objects at different distances. The center of the image falls on the fovea, which has the greatest density of light receptors and the greatest acuity (sharpness) in the visual field. There are no receptors at the place where the optic nerve meets the eye, which is called the blind spot. An image falling on this spot cannot be seen. The variable opening (or pupil) of the eye along with chemical adaptation allows the eye to detect light intensities from the lowest observable to 1010 times greater (without damage). Ten orders of magnitude is an incredible range of detection. Our eyes perform a vast number of functions, such as sense direction, movement, sophisticated colors, and distance. Processing of visual nerve impulses begins with interconnections in the retina and continues in the brain. The optic nerve conveys signals received by the eye to the brain.
Refractive indices are crucial to image formation using lenses. shows refractive indices relevant to the eye. The biggest change in the refractive index—and the one that causes the greatest bending of rays—occurs at the cornea rather than the lens. The ray diagram in shows image formation by the cornea and lens of the eye. The rays bend according to the refractive indices provided in . The cornea provides about two-thirds of the magnification of the eye because the speed of light changes considerably while traveling from air into the cornea. The lens provides the remaining magnification needed to produce an image on the retina. The cornea and lens can be treated as a single thin lens, although the light rays pass through several layers of material (such as the cornea, aqueous humor, several layers in the lens, and vitreous humor), changing direction at each interface. The image formed is much like the one produced by a single convex lens. This result is a case 1 image. Images formed in the eye are inverted, but the brain inverts them once more to make them seem upright.
As noted, the image must fall precisely on the retina to produce clear vision—that is, the image distance, di, must equal the lens-to-retina distance. Because the lens-to-retina distance does not change, di must be the same for objects at all distances. The eye manages to vary the distance by varying the power (and focal length) of the lens to accommodate for objects at various distances. In , you can see the small ciliary muscles above and below the lens that change the shape of the lens and, thus, the focal length.
The need for some type of vision correction is very common. Common vision defects are easy to understand, and some are simple to correct. illustrates two common vision defects. Nearsightedness, or myopia, is the inability to see distant objects clearly while close objects are in focus. The nearsighted eye overconverges the nearly parallel rays from a distant object, and the rays cross in front of the retina. More divergent rays from a close object are converged on the retina, producing a clear image. Farsightedness, or hyperopia, is the inability to see close objects clearly whereas distant objects may be in focus. A farsighted eye does not converge rays from a close object sufficiently to make the rays meet on the retina. Less divergent rays from a distant object can be converged for a clear image.
Because the nearsighted eye overconverges light rays, the correction for nearsightedness involves placing a diverging spectacle lens in front of the eye. This lens reduces the power of an eye that has too short a focal length ((a)). Because the farsighted eye underconverges light rays, the correction for farsightedness is to place a converging spectacle lens in front of the eye. This lens increases the power of an eye that has too long a focal length ((b)).
### Calculations Using Lens Equations
As promised, there are no new equations to memorize. We can use equations already presented for solving problems involving curved mirrors. Careful analysis allows you to apply these equations to lenses. Here are the equations you need
where P is power, expressed in reciprocal meters (m–1) rather than diopters (D), and f is focal length, expressed in meters (m). You also need
where, as before, do and di are object distance and image distance, respectively. Remember, this equation is usually more useful if rearranged to solve for one of the variables. For example,
The equations for magnification, m, are also the same as for mirrors
where hi and ho are the image height and object height, respectively. Remember, also, that a negative di value indicates a virtual image and a negative hi value indicates an inverted image.
These are the steps to follow when solving a lens problem:
1. Step 1. Examine the situation to determine that image formation by a lens is involved.
2. Step 2. Determine whether ray tracing, the thin-lens equations, or both should be used. A sketch is very helpful even if ray tracing is not specifically required by the problem. Write useful symbols and values on the sketch.
3. Step 3. Identify exactly what needs to be determined in the problem (identify the unknowns).
4. Step 4. Make a list of what is given or can be inferred from the problem as stated (identify the knowns). It is helpful to determine whether the situation involves a case 1, 2, or 3 image. Although these are just names for types of images, they have certain characteristics (given in ) that can be of great use in solving problems.
5. Step 5. If ray tracing is required, use the ray-tracing rules listed earlier in this section.
6. Step 6. Most quantitative problems require the use of the thin-lens equations. These equations are solved in the usual manner by substituting knowns and solving for unknowns. Several worked examples were included earlier and can serve as guides.
7. Step 7. Check whether the answer is reasonable. Does it make sense? If you identified the type of image (case 1, 2, or 3) correctly, you should assess whether your answer is consistent with the type of image, magnification, and so on.
All problems will be solved by one or more of the equations just presented, with ray tracing used only for general analysis of the problem. The steps then simplify to the following:
1. Identify the unknown.
2. Identify the knowns.
3. Choose an equation, plug in the knowns, and solve for the unknown.
Here are some worked examples:
### Practice Problems
### Check Your Understanding
### Section Summary
1. The characteristics of images formed by concave and convex lenses can be predicted using ray tracing. Characteristics include real versus virtual, inverted versus upright, and size.
2. The human eye and corrective lenses can be explained using geometric optics.
3. Characteristics of images formed by lenses can be calculated using the mirror/lens equation.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Performance Task
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Diffraction and Interference
## Introduction
Examine a compact disc under white light, noting the colors observed and their locations on the disc. Using the CD, explore the spectra of a few light sources, such as a candle flame, an incandescent bulb, and fluorescent light. If you have ever looked at the reds, blues, and greens in a sunlit soap bubble and wondered how straw-colored soapy water could produce them, you have hit upon one of the many phenomena that can only be explained by the wave character of light. That and other interesting phenomena, such as the dispersion of white light into a rainbow of colors when passed through a narrow slit, cannot be explained fully by geometric optics. In such cases, light interacts with small objects and exhibits its wave characteristics. The topic of this chapter is the branch of optics that considers the behavior of light when it exhibits wave characteristics. |
# Diffraction and Interference
## Understanding Diffraction and Interference
### Section Key Terms
### Diffraction and Interference
We know that visible light is the type of electromagnetic wave to which our eyes responds. As we have seen previously, light obeys the equation
where
m/s is the speed of light in vacuum, f is the frequency of the electromagnetic wave in Hz (or s–1), and
is its wavelength in m. The range of visible wavelengths is approximately 380 to 750 nm. As is true for all waves, light travels in straight lines and acts like a ray when it interacts with objects several times as large as its wavelength. However, when it interacts with smaller objects, it displays its wave characteristics prominently. Interference is the identifying behavior of a wave.
In , both the ray and wave characteristics of light can be seen. The laser beam emitted by the observatory represents ray behavior, as it travels in a straight line. Passing a pure, one-wavelength beam through vertical slits with a width close to the wavelength of the beam reveals the wave character of light. Here we see the beam spreading out horizontally into a pattern of bright and dark regions that are caused by systematic constructive and destructive interference. As it is characteristic of wave behavior, interference is observed for water waves, sound waves, and light waves.
That interference is a characteristic of energy propagation by waves is demonstrated more convincingly by water waves. shows water waves passing through gaps between some rocks. You can easily see that the gaps are similar in width to the wavelength of the waves and that this causes an interference pattern as the waves pass beyond the gaps. A cross-section across the waves in the foreground would show the crests and troughs characteristic of an interference pattern.
Light has wave characteristics in various media as well as in a vacuum. When light goes from a vacuum to some medium, such as water, its speed and wavelength change, but its frequency, f, remains the same. The speed of light in a medium is
, where n is its index of refraction. If you divide both sides of the equation
by n, you get
. Therefore,
, where
is the wavelength in a medium, and
where
is the wavelength in vacuum and n is the medium’s index of refraction. It follows that the wavelength of light is smaller in any medium than it is in vacuum. In water, for example, which has n = 1.333, the range of visible wavelengths is (380 nm)/1.333 to (760 nm)/1.333, or
285–570 nm. Although wavelengths change while traveling from one medium to another, colors do not, since colors are associated with frequency.
The Dutch scientist Christiaan Huygens (1629–1695) developed a useful technique for determining in detail how and where waves propagate. He used wavefronts, which are the points on a wave’s surface that share the same, constant phase (such as all the points that make up the crest of a water wave). Huygens’s principle states, “Every point on a wavefront is a source of wavelets that spread out in the forward direction at the same speed as the wave itself. The new wavefront is a line tangent to all of the wavelets.”
shows how Huygens’s principle is applied. A wavefront is the long edge that moves; for example, the crest or the trough. Each point on the wavefront emits a semicircular wave that moves at the propagation speed v. These are drawn later at a time, t, so that they have moved a distance
. The new wavefront is a line tangent to the wavelets and is where the wave is located at time t. Huygens’s principle works for all types of waves, including water waves, sound waves, and light waves. It will be useful not only in describing how light waves propagate, but also in how they interfere.
What happens when a wave passes through an opening, such as light shining through an open door into a dark room? For light, you expect to see a sharp shadow of the doorway on the floor of the room, and you expect no light to bend around corners into other parts of the room. When sound passes through a door, you hear it everywhere in the room and, thus, you understand that sound spreads out when passing through such an opening. What is the difference between the behavior of sound waves and light waves in this case? The answer is that the wavelengths that make up the light are very short, so that the light acts like a ray. Sound has wavelengths on the order of the size of the door, and so it bends around corners.
If light passes through smaller openings, often called slits, you can use Huygens’s principle to show that light bends as sound does (see ). The bending of a wave around the edges of an opening or an obstacle is called diffraction. Diffraction is a wave characteristic that occurs for all types of waves. If diffraction is observed for a phenomenon, it is evidence that the phenomenon is produced by waves. Thus, the horizontal diffraction of the laser beam after it passes through slits in is evidence that light has the properties of a wave.
Once again, water waves present a familiar example of a wave phenomenon that is easy to observe and understand, as shown in .
The fact that Huygens’s principle worked was not considered enough evidence to prove that light is a wave. People were also reluctant to accept light’s wave nature because it contradicted the ideas of Isaac Newton, who was still held in high esteem. The acceptance of the wave character of light came after 1801, when the English physicist and physician Thomas Young (1773–1829) did his now-classic double-slit experiment (see ).
When light passes through narrow slits, it is diffracted into semicircular waves, as shown in (a). Pure constructive interference occurs where the waves line up crest to crest or trough to trough. Pure destructive interference occurs where they line up crest to trough. The light must fall on a screen and be scattered into our eyes for the pattern to be visible. An analogous pattern for water waves is shown in (b). Note that regions of constructive and destructive interference move out from the slits at well-defined angles to the original beam. Those angles depend on wavelength and the distance between the slits, as you will see below.
### Calculations Involving Diffraction and Interference
The fact that the wavelength of light of one color, or monochromatic light, can be calculated from its two-slit diffraction pattern in Young’s experiments supports the conclusion that light has wave properties. To understand the basis of such calculations, consider how two waves travel from the slits to the screen. Each slit is a different distance from a given point on the screen. Thus different numbers of wavelengths fit into each path. Waves start out from the slits in phase (crest to crest), but they will end up out of phase (crest to trough) at the screen if the paths differ in length by half a wavelength, interfering destructively. If the paths differ by a whole wavelength, then the waves arrive in phase (crest to crest) at the screen, interfering constructively. More generally, if the paths taken by the two waves differ by any half-integral number of wavelengths
, then destructive interference occurs. Similarly, if the paths taken by the two waves differ by any integral number of wavelengths
, then constructive interference occurs.
shows how to determine the path-length difference for waves traveling from two slits to a common point on a screen. If the screen is a large distance away compared with the distance between the slits, then the angle
between the path and a line from the slits perpendicular to the screen (see the figure) is nearly the same for each path. That approximation and simple trigonometry show the length difference,
, to be
, where d is the distance between the slits,
To obtain constructive interference for a double slit, the path-length difference must be an integral multiple of the wavelength, or
Similarly, to obtain destructive interference for a double slit, the path-length difference must be a half-integral multiple of the wavelength, or
The number m is the order of the interference. For example, m = 4 is fourth-order interference.
shows how the intensity of the bands of constructive interference decreases with increasing angle.
Light passing through a single slit forms a diffraction pattern somewhat different from that formed by double slits. shows a single-slit diffraction pattern. Note that the central maximum is larger than those on either side, and that the intensity decreases rapidly on either side.
The analysis of single-slit diffraction is illustrated in . Assuming the screen is very far away compared with the size of the slit, rays heading toward a common destination are nearly parallel. That approximation allows a series of trigonometric operations that result in the equations for the minima produced by destructive interference.
or
When rays travel straight ahead, they remain in phase and a central maximum is obtained. However, when rays travel at an angle
relative to the original direction of the beam, each ray travels a different distance to the screen, and they can arrive in or out of phase. Thus, a ray from the center travels a distance
farther than the ray from the top edge of the slit, they arrive out of phase, and they interfere destructively. Similarly, for every ray between the top and the center of the slit, there is a ray between the center and the bottom of the slit that travels a distance
farther to the common point on the screen, and so interferes destructively. Symmetrically, there will be another minimum at the same angle below the direct ray.
Below we summarize the equations needed for the calculations to follow.
The speed of light in a vacuum, c, the wavelength of the light,
, and its frequency, f, are related as follows.
The wavelength of light in a medium,
, compared to its wavelength in a vacuum,
, is given by
To calculate the positions of constructive interference for a double slit, the path-length difference must be an integral multiple, m, of the wavelength.
where d is the distance between the slits and
is the angle between a line from the slits to the maximum and a line perpendicular to the barrier in which the slits are located. To calculate the positions of destructive interference for a double slit, the path-length difference must be a half-integral multiple of the wavelength:
For a single-slit diffraction pattern, the width of the slit, D, the distance of the first (m = 1) destructive interference minimum, y, the distance from the slit to the screen, L, and the wavelength,
, are given by
Also, for single-slit diffraction,
where
is the angle between a line from the slit to the minimum and a line perpendicular to the screen, and m is the order of the minimum.
### Practice Problems
### Check Your Understanding
### Section Summary
1. The wavelength of light varies with the refractive index of the medium.
2. Slits produce a diffraction pattern if their width and separation are similar to the wavelength of light passing through them.
3. Interference bands of a single-slit diffraction pattern can be predicted.
4. Interference bands of a double-slit diffraction pattern can be predicted.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Diffraction and Interference
## Applications of Diffraction, Interference, and Coherence
### Section Key Terms
### Wave-Based Applications of Light
In 1917, Albert Einstein was thinking about photons and excited atoms. He considered an atom excited by a certain amount of energy and what would happen if that atom were hit by a photon with the same amount of energy. He suggested that the atom would emit a photon with that amount of energy, and it would be accompanied by the original photon. The exciting part is that you would have two photons with the same energy and they would be in phase. Those photons could go on to hit other excited atoms, and soon you would have a stream of in-phase photons. Such a light stream is said to be coherent. Some four decades later, Einstein’s idea found application in a process called, light amplification by stimulated emission of radiation. Take the first letters of all the words (except by and “of”) and write them in order. You get the word (see (a)), which is the name of the device that produces such a beam of light.
Laser beams are directional, very intense, and narrow (only about 0.5 mm in diameter). These properties lead to a number of applications in industry and medicine. The following are just a few examples:
1. This chapter began with a picture of a compact disc (see ). Those audio and data-storage devices began replacing cassette tapes during the 1990s. CDs are read by interpreting variations in reflections of a laser beam from the surface.
2. Some barcode scanners use a laser beam.
3. Lasers are used in industry to cut steel and other metals.
4. Lasers are bounced off reflectors that astronauts left on the Moon. The time it takes for the light to make the round trip can be used to make precise calculations of the Earth-Moon distance.
5. Laser beams are used to produce holograms. The name hologram means entire picture (from the Greek holo-, as in holistic), because the image is three-dimensional. A viewer can move around the image and see it from different perspectives. Holograms take advantage of the wave properties of light, as opposed to traditional photography which is based on geometric optics. A holographic image is produced by constructive and destructive interference of a split laser beam.
6. One of the advantages of using a laser as a surgical tool is that it is accompanied by very little bleeding.
7. Laser eye surgery has improved the vision of many people, without the need for corrective lenses. A laser beam is used to change the shape of the lens of the eye, thus changing its focal length.
An interesting thing happens if you pass light through a large number of evenly-spaced parallel slits. Such an arrangement of slits is called a diffraction grating. An interference pattern is created that is very similar to the one formed by double-slit diffraction (see and ). A diffraction grating can be manufactured by scratching glass with a sharp tool to form a number of precisely positioned parallel lines, which act like slits. Diffraction gratings work both for transmission of light, as in , and for reflection of light, as on the butterfly wings or the Australian opal shown in , or the CD pictured in the opening illustration of this chapter. In addition to their use as novelty items, diffraction gratings are commonly used for spectroscopic dispersion and analysis of light. What makes them particularly useful is the fact that they form a sharper pattern than do double slits. That is, their bright regions are narrower and brighter, while their dark regions are darker. shows idealized graphs demonstrating the sharper pattern. Natural diffraction gratings occur in the feathers of certain birds. Tiny, fingerlike structures in regular patterns act as reflection gratings, producing constructive interference that gives the feathers colors not solely due to their pigmentation. The effect is called iridescence.
Where are diffraction gratings used? Diffraction gratings are key components of monochromators—devices that separate the various wavelengths of incoming light and allow a beam with only a specific wavelength to pass through. Monochromators are used, for example, in optical imaging of particular wavelengths from biological or medical samples. A diffraction grating can be chosen to specifically analyze a wavelength of light emitted by molecules in diseased cells in a biopsy sample, or to help excite strategic molecules in the sample with a selected frequency of light. Another important use is in optical fiber technologies where fibers are designed to provide optimum performance at specific wavelengths. A range of diffraction gratings is available for selecting specific wavelengths for such use.
Diffraction gratings are used in spectroscopes to separate a light source into its component wavelengths. When a material is heated to incandescence, it gives off wavelengths of light characteristic of the chemical makeup of the material. A pure substance will produce a spectrum that is unique, thus allowing identification of the substance. Spectroscopes are also used to measure wavelengths both shorter and longer than visible light. Such instruments have become especially useful to astronomers and chemists. shows a diagram of a spectroscope.
Light diffracts as it moves through space, bending around obstacles and interfering constructively and destructively. While diffraction allows light to be used as a spectroscopic tool, it also limits the detail we can obtain in images.
(a) shows the effect of passing light through a small circular aperture. Instead of a bright spot with sharp edges, a spot with a fuzzy edge surrounded by circles of light is obtained. This pattern is caused by diffraction similar to that produced by a single slit. Light from different parts of the circular aperture interferes constructively and destructively. The effect is most noticeable when the aperture is small, but the effect is there for large apertures, too.
How does diffraction affect the detail that can be observed when light passes through an aperture? (b) shows the diffraction pattern produced by two point light sources that are close to one another. The pattern is similar to that for a single point source, and it is just barely possible to tell that there are two light sources rather than one. If they are closer together, as in (c), you cannot distinguish them, thus limiting the detail, or resolution, you can obtain. That limit is an inescapable consequence of the wave nature of light.
There are many situations in which diffraction limits the resolution. The acuity of vision is limited because light passes through the pupil, the circular aperture of the eye. Be aware that the diffraction-like spreading of light is due to the limited diameter of a light beam, not the interaction with an aperture. Thus light passing through a lens with a diameter of D shows the diffraction effect and spreads, blurring the image, just as light passing through an aperture of diameter D does. Diffraction limits the resolution of any system having a lens or mirror. Telescopes are also limited by diffraction, because of the finite diameter, D, of their primary mirror.
### Calculations Involving Diffraction Gratings and Resolution
Early in the chapter, it was mentioned that when light passes from one medium to another, its speed and wavelength change, but its frequency remains constant. The equation
shows how to the wavelength in a given medium,
, is related to the wavelength in a vacuum,
, and the refractive index, n, of the medium. The equation is useful for calculating the change in wavelength of a monochromatic laser beam in various media. The analysis of a diffraction grating is very similar to that for a double slit. As you know from the discussion of double slits in Young’s double-slit experiment, light is diffracted by, and spreads out after passing through, each slit. Rays travel at an angle
relative to the incident direction. Each ray travels a different distance to a common point on a screen far away. The rays start in phase, and they can be in or out of phase when they reach a screen, depending on the difference in the path lengths traveled. Each ray travels a distance that differs by
from that of its neighbor, where d is the distance between slits. If
equals an integral number of wavelengths, the rays all arrive in phase, and constructive interference (a maximum) is obtained. Thus, the condition necessary to obtain constructive interference for a diffraction grating is
where d is the distance between slits in the grating,
is the wavelength of the light, and m is the order of the maximum. Note that this is exactly the same equation as for two slits separated by d. However, the slits are usually closer in diffraction gratings than in double slits, producing fewer maxima at larger angles.
Just what is the resolution limit of an aperture or lens? To answer that question, consider the diffraction pattern for a circular aperture, which, similar to the diffraction pattern of light passing through a slit, has a central maximum that is wider and brighter than the maxima surrounding it (see (a)). It can be shown that, for a circular aperture of diameter D, the first minimum in the diffraction pattern occurs at
, provided that the aperture is large compared with the wavelength of light, which is the case for most optical instruments. The accepted criterion for determining the diffraction limit to resolution based on diffraction was developed by Lord Rayleigh in the 19th century. The Rayleigh criterion for the diffraction limit to resolution states that two images are just resolvable when the center of the diffraction pattern of one is directly over the first minimum of the diffraction pattern of the other. See (b). The first minimum is at an angle of
, so that two point objects are just resolvable if they are separated by the angle
where
is the wavelength of the light (or other electromagnetic radiation) and D is the diameter of the aperture, lens, mirror, etc., with which the two objects are observed. In the expression above,
has units of radians.
### Practice Problems
### Check Your Understanding
### Section Summary
1. The focused, coherent radiation emitted by lasers has many uses in medicine and industry.
2. Characteristics of diffraction patterns produced with diffraction gratings can be determined.
3. Diffraction gratings have been incorporated in many instruments, including microscopes and spectrometers.
4. Resolution has a limit that can be predicted.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Extended Response
### Performance Task
|
# Static Electricity
## Introduction
You may have been introduced to static electricity like the child sliding down the slide in the opening photograph (). The zap that he is likely to receive if he touches a playmate or parent tends to bring home the lesson. But static electricity is more than just fun and games—it is put to use in many industries. The forces between electrically charged particles are used in technologies such as printers, pollution filters, and spray guns used for painting cars and trucks. Static electricity is the study of phenomena that involve an imbalance of electrical charge. Although creating this imbalance typically requires moving charge around, once the imbalance is created, it often remains static for a long time. The study of charge in motion is called electromagnetism and will be covered in a later chapter. What is electrical charge, how is it associated with objects, and what forces does it create? These are just some of the questions that this chapter addresses. |
# Static Electricity
## Electrical Charges, Conservation of Charge, and Transfer of Charge
### Section Key Terms
### Electric Charge
You may know someone who has an electric personality, which usually means that other people are attracted to this person. This saying is based on electric charge, which is a property of matter that causes objects to attract or repel each other. Electric charge comes in two varieties, which we call positive and negative. Like charges repel each other, and unlike charges attract each other. Thus, two positive charges repel each other, as do two negative charges. A positive charge and a negative charge attract each other.
How do we know there are two types of electric charge? When various materials are rubbed together in controlled ways, certain combinations of materials always result in a net charge of one type on one material and a net charge of the opposite type on the other material. By convention, we call one type of charge positive and the other type negative. For example, when glass is rubbed with silk, the glass becomes positively charged and the silk negatively charged. Because the glass and silk have opposite charges, they attract one another like clothes that have rubbed together in a dryer. Two glass rods rubbed with silk in this manner will repel one another, because each rod has positive charge on it. Similarly, two silk cloths rubbed in this manner will repel each other, because both cloths have negative charge. shows how these simple materials can be used to explore the nature of the force between charges.
It took scientists a long time to discover what lay behind these two types of charges. The word electric itself comes from the Greek word elektron for amber, because the ancient Greeks noticed that amber, when rubbed by fur, attracts dry straw. Almost 2,000 years later, the English physicist William Gilbert proposed a model that explained the effect of electric charge as being due to a mysterious electrical fluid that would pass from one object to another. This model was debated for several hundred years, but it was finally put to rest in 1897 by the work of the English physicist J. J. Thomson and French physicist Jean Perrin. Along with many others, Thomson and Perrin were studying the mysterious cathode rays that were known at the time to consist of particles smaller than the smallest atom. Perrin showed that cathode rays actually carried negative electrical charge. Later, Thomson’s work led him to declare, “I can see no escape from the conclusion that [cathode rays] are charges of negative electricity carried by particles of matter.”
It took several years of further experiments to confirm Thomson’s interpretation of the experiments, but science had in fact discovered the particle that carries the fundamental unit of negative electrical charge. We now know this particle as the electron.
Atoms, however, were known to be electrically neutral, which means that they carry the same amount of positive and negative charge, so their net charge is zero. Because electrons are negative, some other part of the atom must contain positive charge. Thomson put forth what is called the plum pudding model, in which he described atoms as being made of thousands of electrons swimming around in a nebulous mass of positive charge, as shown by the left-side image of . His student, Ernest Rutherford, originally believed that this model was correct and used it (along with other models) to try to understand the results of his experiments bombarding gold foils with alpha particles (i.e., helium atoms stripped of their electrons). The results, however, did not confirm Thomson’s model but rather destroyed it! Rutherford found that most of the space occupied by the gold atoms was actually empty and that almost all of the matter of each atom was concentrated into a tiny, extremely dense nucleus, as shown by the right-side image of . The atomic nucleus was later found to contain particles called protons, each of which carries a unit of positive electric charge.Protons were later found to contain sub particles called
Protons and electrons are thus the fundamental particles that carry electric charge. Each proton carries one unit of positive charge, and each electron carries one unit of negative charge. To the best precision that modern technology can provide, the charge carried by a proton is exactly the opposite of that carried by an electron. The SI unit for electric charge is the coulomb (abbreviated as “C”), which is named after the French physicist Charles Augustin de Coulomb, who studied the force between charged objects. The proton carries
and the electron carries
. The number n of protons required to make +1.00 C is
The same number of electrons is required to make −1.00 C of electric charge. The fundamental unit of charge is often represented as e. Thus, the charge on a proton is e, and the charge on an electron is −e. Mathematically,
### Conservation of Charge
Because the fundamental positive and negative units of charge are carried on protons and electrons, we would expect that the total charge cannot change in any system that we define. In other words, although we might be able to move charge around, we cannot create or destroy it. This should be true provided that we do not create or destroy protons or electrons in our system. In the twentieth century, however, scientists learned how to create and destroy electrons and protons, but they found that charge is still conserved. Many experiments and solid theoretical arguments have elevated this idea to the status of a law. The law of conservation of charge says that electrical charge cannot be created or destroyed.
The law of conservation of charge is very useful. It tells us that the net charge in a system is the same before and after any interaction within the system. Of course, we must ensure that no external charge enters the system during the interaction and that no internal charge leaves the system. Mathematically, conservation of charge can be expressed as
where
is the net charge of the system before the interaction, and
is the net charge after the interaction.
### Practice Problems
### Conductors and Insulators
Materials can be classified depending on whether they allow charge to move. If charge can easily move through a material, such as metals, then these materials are called conductors. This means that charge can be conducted (i.e., move) through the material rather easily. If charge cannot move through a material, such as rubber, then this material is called an insulator.
Most materials are insulators. Their atoms and molecules hold on more tightly to their electrons, so it is difficult for electrons to move between atoms. However, it is not impossible. With enough energy, it is possible to force electrons to move through an insulator. However, the insulator is often physically destroyed in the process. In metals, the outer electrons are loosely bound to their atoms, so not much energy is required to make electrons move through metal. Such metals as copper, silver, and aluminum are good conductors. Insulating materials include plastics, glass, ceramics, and wood.
The conductivity of some materials is intermediate between conductors and insulators. These are called semiconductors. They can be made conductive under the right conditions, which can involve temperature, the purity of the material, and the force applied to push electrons through them. Because we can control whether semiconductors are conductors or insulators, these materials are used extensively in computer chips. The most commonly used semiconductor is silicon. shows various materials arranged according to their ability to conduct electrons.
What happens if an excess negative charge is placed on a conducting object? Because like charges repel each other, they will push against each other until they are as far apart as they can get. Because the charge can move in a conductor, it moves to the outer surfaces of the object. (a) shows schematically how an excess negative charge spreads itself evenly over the outer surface of a metal sphere.
What happens if the same is done with an insulating object? The electrons still repel each other, but they are not able to move, because the material is an insulator. Thus, the excess charge stays put and does not distribute itself over the object. (b) shows this situation.
### Transfer and Separation of Charge
Most objects we deal with are electrically neutral, which means that they have the same amount of positive and negative charge. However, transferring negative charge from one object to another is fairly easy to do. When negative charge is transferred from one object to another, an excess of positive charge is left behind. How do we know that the negative charge is the mobile charge? The positive charge is carried by the proton, which is stuck firmly in the nucleus of atoms, and the atoms are stuck in place in solid materials. Electrons, which carry the negative charge, are much easier to remove from their atoms or molecules and can therefore be transferred more easily.
Electric charge can be transferred in several manners. One of the simplest ways to transfer charge is charging by contact, in which the surfaces of two objects made of different materials are placed in close contact. If one of the materials holds electrons more tightly than the other, then it takes some electrons with it when the materials are separated. Rubbing two surfaces together increases the transfer of electrons, because it creates a closer contact between the materials. It also serves to present fresh material with a full supply of electrons to the other material. Thus, when you walk across a carpet on a dry day, your shoes rub against the carpet, and some electrons are removed from the carpet by your shoes. The result is that you have an excess of negative charge on your shoes. When you then touch a doorknob, some of your excess of electrons transfer to the neutral doorknob, creating a small spark.
Touching the doorknob with your hand demonstrates a second way to transfer electric charge, which is charging by conduction. This transfer happens because like charges repel, and so the excess electrons that you picked up from the carpet want to be as far away from each other as possible. Some of them move to the doorknob, where they will distribute themselves over the outer surface of the metal. Another example of charging by conduction is shown in the top row of . A metal sphere with 100 excess electrons touches a metal sphere with 50 excess electrons, so 25 electrons from the first sphere transfer to the second sphere. Each sphere finishes with 75 excess electrons.
The same reasoning applies to the transfer of positive charge. However, because positive charge essentially cannot move in solids, it is transferred by moving negative charge in the opposite direction. For example, consider the bottom row of . The first metal sphere has 100 excess protons and touches a metal sphere with 50 excess protons, so the second sphere transfers 25 electrons to the first sphere. These 25 extra electrons will electrically cancel 25 protons so that the first metal sphere is left with 75 excess protons. This is shown in the bottom row of . The second metal sphere lost 25 electrons so it has 25 more excess protons, for a total of 75 excess protons. The end result is the same if we consider that the first ball transferred a net positive charge equal to that of 25 protons to the first ball.
In this discussion, you may wonder how the excess electrons originally got from your shoes to your hand to create the spark when you touched the doorknob. The answer is that no electrons actually traveled from your shoes to your hands. Instead, because like charges repel each other, the excess electrons on your shoe simply pushed away some of the electrons in your feet. The electrons thus dislodged from your feet moved up into your leg and in turn pushed away some electrons in your leg. This process continued through your whole body until a distribution of excess electrons covered the extremities of your body. Thus your head, your hands, the tip of your nose, and so forth all received their doses of excess electrons that had been pushed out of their normal positions. All this was the result of electrons being pushed out of your feet by the excess electrons on your shoes.
This type of charge separation is called polarization. As soon as the excess electrons leave your shoes (by rubbing off onto the floor or being carried away in humid air), the distribution of electrons in your body returns to normal. Every part of your body is again electrically neutral (i.e., zero excess charge).
The phenomenon of polarization is seen in . The child has accumulated excess positive charge by sliding on the slide. This excess charge repels itself and so becomes distributed over the extremities of the child’s body, notably in his hair. As a result, the hair stands on end, because the excess negative charge on each strand repels the excess negative charge on neighboring strands.
Polarization can be used to charge objects. Consider the two metallic spheres shown in . The spheres are electrically neutral, so they carry the same amounts of positive and negative charge. In the top picture ((a)), the two spheres are touching, and the positive and negative charge is evenly distributed over the two spheres. We then approach a glass rod that carries an excess positive charge, which can be done by rubbing the glass rod with silk, as shown in (b). Because opposite charges attract each other, the negative charge is attracted to the glass rod, leaving an excess positive charge on the opposite side of the right sphere. This is an example of charging by induction, whereby a charge is created by approaching a charged object with a second object to create an unbalanced charge in the second object. If we then separate the two spheres, as shown in (c), the excess charge is stuck on each sphere. The left sphere now has an excess negative charge, and the right sphere has an excess positive charge. Finally, in the bottom picture, the rod is removed, and the opposite charges attract each other, so they move as close together as they can get.
### Practice Problems
### Check Your Understanding
### Section Summary
1. Electric charge is a conserved quantity, which means it can be neither created nor destroyed.
2. Electric charge comes in two varieties, which are called positive and negative.
3. Charges with the same sign repel each other. Charges with opposite signs attract each other.
4. Charges can move easily in conducting material. Charges cannot move easily in an insulating material.
5. Objects can be charged in three ways: by contact, by conduction, and by induction.
6. Although a polarized object may be neutral, its electrical charge is unbalanced, so one side of the object has excess negative charge and the other side has an equal magnitude of excess positive charge.
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Static Electricity
## Coulomb's law
### Section Key Terms
More than 100 years before Thomson and Rutherford discovered the fundamental particles that carry positive and negative electric charges, the French scientist Charles-Augustin de Coulomb mathematically described the force between charged objects. Doing so required careful measurements of forces between charged spheres, for which he built an ingenious device called a .
This device, shown in , contains an insulating rod that is hanging by a thread inside a glass-walled enclosure. At one end of the rod is the metallic sphere A. When no charge is on this sphere, it touches sphere B. Coulomb would touch the spheres with a third metallic ball (shown at the bottom of the diagram) that was charged. An unknown amount of charge would distribute evenly between spheres A and B, which would then repel each other, because like charges repel. This force would cause sphere A to rotate away from sphere B, thus twisting the wire until the torsion in the wire balanced the electrical force. Coulomb then turned the knob at the top, which allowed him to rotate the thread, thus bringing sphere A closer to sphere B. He found that bringing sphere A twice as close to sphere B required increasing the torsion by a factor of four. Bringing the sphere three times closer required a ninefold increase in the torsion. From this type of measurement, he deduced that the electrical force between the spheres was inversely proportional to the distance squared between the spheres. In other words,
where r is the distance between the spheres.
An electrical charge distributes itself equally between two conducting spheres of the same size. Knowing this allowed Coulomb to divide an unknown charge in half. Repeating this process would produce a sphere with one quarter of the initial charge, and so on. Using this technique, he measured the force between spheres A and B when they were charged with different amounts of charge. These measurements led him to deduce that the force was proportional to the charge on each sphere, or
where
is the charge on sphere A, and
is the charge on sphere B.
Combining these two proportionalities, he proposed the following expression to describe the force between the charged spheres.
This equation is known as Coulomb’s law, and it describes the electrostatic force between charged objects. The constant of proportionality k is called . In SI units, the constant k has the value
The direction of the force is along the line joining the centers of the two objects. If the two charges are of opposite signs, Coulomb’s law gives a negative result. This means that the force between the particles is attractive. If the two charges have the same signs, Coulomb’s law gives a positive result. This means that the force between the particles is repulsive. For example, if both
and
are negative or if both are positive, the force between them is repulsive. This is shown in (a). If
is a negative charge and
is a positive charge (or vice versa), then the charges are different, so the force between them is attractive. This is shown in (b).
Note that Coulomb’s law applies only to charged objects that are not moving with respect to each other. The law says that the force is proportional to the amount of charge on each object and inversely proportional to the square of the distance between the objects. If we double the charge
, for instance, then the force is doubled. If we double the distance between the objects, then the force between them decreases by a factor of
. Although Coulomb’s law is true in general, it is easiest to apply to spherical objects or to objects that are much smaller than the distance between the objects (in which case, the objects can be approximated as spheres).
Coulomb’s law is an example of an inverse-square law, which means the force depends on the square of the denominator. Another inverse-square law is Newton’s law of universal gravitation, which is
. Although these laws are similar, they differ in two important respects: (i) The gravitational constant G is much, much smaller than k (
); and (ii) only one type of mass exists, whereas two types of electric charge exist. These two differences explain why gravity is so much weaker than the electrostatic force and why gravity is only attractive, whereas the electrostatic force can be attractive or repulsive.
Finally, note that Coulomb measured the distance between the spheres from the centers of each sphere. He did not explain this assumption in his original papers, but it turns out to be valid. From outside a uniform spherical distribution of charge, it can be treated as if all the charge were located at the center of the sphere.
### Practice Problems
### Check Your Understanding
### Section Summary
1. Coulomb’s law is an inverse square law and describes the electrostatic force between particles.
2. The electrostatic force between charged objects is proportional to the charge on each object and inversely proportional to the distance squared between the objects.
3. If Coulomb’s law gives a negative result, the force is attractive; if the result is positive, the force is repulsive.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Static Electricity
## Electric Field
### Section Key Terms
You may have heard of a in science fiction movies, where such fields apply forces at particular positions in space to keep a villain trapped or to protect a spaceship from enemy fire. The concept of a field is very useful in physics, although it differs somewhat from what you see in movies.
A field is a way of conceptualizing and mapping the force that surrounds any object and acts on another object at a distance without apparent physical connection. For example, the gravitational field surrounding Earth and all other masses represents the gravitational force that would be experienced if another mass were placed at a given point within the field. Michael Faraday, an English physicist of the nineteenth century, proposed the concept of an electric field. If you know the electric field, then you can easily calculate the force (magnitude and direction) applied to any electric charge that you place in the field.
An electric field is generated by electric charge and tells us the force per unit charge at all locations in space around a charge distribution. The charge distribution could be a single point charge; a distribution of charge over, say, a flat plate; or a more complex distribution of charge. The electric field extends into space around the charge distribution. Now consider placing a test charge in the field. A test charge is a positive electric charge whose charge is so small that it does not significantly disturb the charges that create the electric field. The electric field exerts a force on the test charge in a given direction. The force exerted is proportional to the charge of the test charge. For example, if we double the charge of the test charge, the force exerted on it doubles. Mathematically, saying that electric field is the force per unit charge is written as
where we are considering only electric forces. Note that the electric field is a vector field that points in the same direction as the force on the positive test charge. The units of electric field are N/C.
If the electric field is created by a point charge or a sphere of uniform charge, then the magnitude of the force between this point charge Q and the test charge is given by Coulomb’s law
where the absolute value is used, because we only consider the magnitude of the force. The magnitude of the electric field is then
This equation gives the magnitude of the electric field created by a point charge Q. The distance r in the denominator is the distance from the point charge, Q, or from the center of a spherical charge, to the point of interest.
If the test charge is removed from the electric field, the electric field still exists. To create a three-dimensional map of the electric field, imagine placing the test charge in various locations in the field. At each location, measure the force on the charge, and use the vector equation
to calculate the electric field. Draw an arrow at each point where you place the test charge to represent the strength and the direction of the electric field. The length of the arrows should be proportional to the strength of the electric field. If you join together these arrows, you obtain lines. shows an image of the three-dimensional electric field created by a positive charge.
Just drawing the electric field lines in a plane that slices through the charge gives the two-dimensional electric-field maps shown in . On the left is the electric field created by a positive charge, and on the right is the electric field created by a negative charge.
Notice that the electric field lines point away from the positive charge and toward the negative charge. Thus, a positive test charge placed in the electric field of the positive charge will be repelled. This is consistent with Coulomb’s law, which says that like charges repel each other. If we place the positive charge in the electric field of the negative charge, the positive charge is attracted to the negative charge. The opposite is true for negative test charges. Thus, the direction of the electric field lines is consistent with what we find by using Coulomb’s law.
The equation
says that the electric field gets stronger as we approach the charge that generates it. For example, at 2 cm from the charge Q (r = 2 cm), the electric field is four times stronger than at 4 cm from the charge (r = 4 cm). Looking at and again, we see that the electric field lines become denser as we approach the charge that generates it. In fact, the density of the electric field lines is proportional to the strength of the electric field!
Electric-field maps can be made for several charges or for more complicated charge distributions. The electric field due to multiple charges may be found by adding together the electric field from each individual charge. Because this sum can only be a single number, we know that only a single electric-field line can go through any given point. In other words, electric-field lines cannot cross each other.
(a) shows a two-dimensional map of the electric field generated by a charge of +q and a nearby charge of −q. The three-dimensional version of this map is obtained by rotating this map about the axis that goes through both charges. A positive test charge placed in this field would experience a force in the direction of the field lines at its location. It would thus be repelled from the positive charge and attracted to the negative charge. (b) shows the electric field generated by two charges of −q. Note how the field lines tend to repel each other and do not overlap. A positive test charge placed in this field would be attracted to both charges. If you are far from these two charges, where far means much farther than the distance between the charges, the electric field looks like the electric field from a single charge of −2q.
### Practice Problems
### Check Your Understanding
### Section Summary
1. The electric field defines the force per unit charge in the space around a charge distribution.
2. For a point charge or a sphere of uniform charge, the electric field is inversely proportional to the distance from the point charge or from the center of the sphere.
3. Electric-field lines never cross each other.
4. More force is applied to a charge in a region with many electric field lines than in a region with few electric field lines.
5. Electric field lines start at positive charges and point away from positive charges. They end at negative charges and point toward negative charges.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Static Electricity
## Electric Potential
### Section Key Terms
As you learned in studying gravity, a mass in a gravitational field has potential energy, which means it has the potential to accelerate and thereby increase its kinetic energy. This kinetic energy can be used to do work. For example, imagine you want to use a stone to pound a nail into a piece of wood. You first lift the stone high above the nail, which increases the potential energy of the stone-Earth system—because Earth is so large, it does not move, so we usually shorten this by saying simply that the potential energy of the stone increases. When you drop the stone, gravity converts the potential energy into kinetic energy. When the stone hits the nail, it does work by pounding the nail into the wood. The gravitational potential energy is the work that a mass can potentially do by virtue of its position in a gravitational field. Potential energy is a very useful concept, because it can be used with conservation of energy to calculate the motion of masses in a gravitational field.
Electric potential energy works much the same way, but it is based on the electric field instead of the gravitational field. By virtue of its position in an electric field, a charge has an electric potential energy. If the charge is free to move, the force due to the electric field causes it to accelerate, so its potential energy is converted to kinetic energy, just like a mass that falls in a gravitational field. This kinetic energy can be used to do work. The electric potential energy is the work that a charge can do by virtue of its position in an electric field.
The analogy between gravitational potential energy and electric potential energy is depicted in . On the left, the ball-Earth system gains gravitational potential energy when the ball is higher in Earth's gravitational field. On the right, the two-charge system gains electric potential energy when the positive charge is farther from the negative charge.
Let’s use the symbol
to denote gravitational potential energy. When a mass falls in a gravitational field, its gravitational potential energy decreases. Conservation of energy tells us that the work done by the gravitational field to make the mass accelerate must equal the loss of potential energy of the mass. If we use the symbol
to denote this work, then
where the minus sign reflects the fact that the potential energy of the ball decreases.
The work done by gravity on the mass is
where F is the force due to gravity, and
and
are the initial and final positions of the ball, respectively. The negative sign is because gravity points down, which we consider to be the negative direction. For the constant gravitational field near Earth’s surface,
. The change in gravitational potential energy of the mass is
Note that
is just the negative of the height h from which the mass falls, so we usually just write
.
We now apply the same reasoning to a charge in an electric field to find the electric potential energy. The change
in electric potential energy is the work done by the electric field to move a charge q from an initial position
to a final position
(
). The definition of work does not change, except that now the work is done by the electric field:
. For a charge that falls through a constant electric field E, the force applied to the charge by the electric field is
. The change in electric potential energy of the charge is thus
or
This equation gives the change in electric potential energy of a charge q when it moves from position
to position
in a constant electric field E.
shows how this analogy would work if we were close to Earth’s surface, where gravity is constant. The top image shows a charge accelerating due to a constant electric field. Likewise, the round mass in the bottom image accelerates due to a constant gravitation field. In both cases, the potential energy of the particle decreases, and its kinetic energy increases.
If the electric field is not constant, then the equation
is not valid, and deriving the electric potential energy becomes more involved. For example, consider the electric potential energy of an assembly of two point charges
and
of the same sign that are initially very far apart. We start by placing charge
at the origin of our coordinate system. This takes no electrical energy, because there is no electric field at the origin (because charge
is very far away). We then bring charge
in from very far away to a distance r from the center of charge
. This requires some effort, because the electric field of charge
applies a repulsive force on charge
. The energy it takes to assemble these two charges can be recuperated if we let them fly apart again. Thus, the charges have potential energy when they are a distance r apart. It turns out that the electric potential energy of a pair of point charges
and
a distance r apart is
To recap, if charges
and
are free to move, they can accumulate kinetic energy by flying apart, and this kinetic energy can be used to do work. The maximum amount of work the two charges can do (if they fly infinitely far from each other) is given by the equation above.
Notice that if the two charges have opposite signs, then the potential energy is negative. This means that the charges have more potential to do work when they are far apart than when they are at a distance r apart. This makes sense: Opposite charges attract, so the charges can gain more kinetic energy if they attract each other from far away than if they start at only a short distance apart. Thus, they have more potential to do work when they are far apart. summarizes how the electric potential energy depends on charge and separation.
### Electric Potential
Recall that to find the force applied by a fixed charge Q on any arbitrary test charge q, it was convenient to define the electric field, which is the force per unit charge applied by Q on any test charge that we place in its electric field. The same Strategy is used here with electric potential energy: We now define the electric potential V, which is the electric potential energy per unit charge.
Normally, the electric potential is simply called the potential or . The units for the potential are J/C, which are given the name volt (V) after the Italian physicist Alessandro Volta (1745–1827). From the equation
, the electric potential a distance r from a point charge
is
This equation gives the energy required per unit charge to bring a charge
from infinity to a distance r from a point charge
Mathematically, this is written as
Note that this equation actually represents a difference in electric potential. However, because the second term is zero, it is normally not written, and we speak of the electric potential instead of the electric potential difference, or we just say the potential difference, or voltage). Below, when we consider the electric potential energy per unit charge between two points not infinitely far apart, we speak of electric potential difference explicitly. Just remember that electric potential and electric potential difference are really the same thing; the former is used just when the electric potential energy is zero in either the initial or final charge configuration.
Coming back now to the electric potential a distance r from a point charge
, note that
can be any arbitrary point charge, so we can drop the subscripts and simply write
Now consider the electric potential near a group of charges q1, q2, and q3, as drawn in . The electric potential is derived by considering the electric field. Electric fields follow the principle of superposition and can be simply added together, so the electric potential from different charges also add together. Thus, the electric potential of a point near a group of charges is
where
are the distances from the center of charges
to the point of interest, as shown in .
Now let’s consider the electric potential in a uniform electric field. From the equation
, we see that the potential difference in going from
to
in a uniform electric field E is
Notice that a positive charge in a region with high potential will experience a force pushing it toward regions of lower potential. In this sense, potential is like pressure for fluids. Imagine a pipe containing fluid, with the fluid at one end of the pipe under high pressure and the fluid at the other end of the pipe under low pressure. If nothing prevents the fluid from flowing, it will flow from the high-pressure end to the low-pressure end. Likewise, a positive charge that is free to move will move from a region with high potential to a region with lower potential.
### Practice Problems
### Check Your Understanding
### Section Summary
1. Electric potential energy is a concept similar to gravitational potential energy: It is the potential that charges have to do work by virtue of their positions relative to each other.
2. Electric potential is the electric potential energy per unit charge.
3. The potential is always measured between two points, where one point may be at infinity.
4. Positive charges move from regions of high potential to regions of low potential.
5. Negative charges move from regions of low potential to regions of high potential.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Static Electricity
## Capacitors and Dielectrics
### Section Key Terms
### Capacitors
Consider again the X-ray tube discussed in the previous sample problem. How can a uniform electric field be produced? A single positive charge produces an electric field that points away from it, as in . This field is not uniform, because the space between the lines increases as you move away from the charge. However, if we combine a positive and a negative charge, we obtain the electric field shown in (a). Notice that, between the charges, the electric field lines are more equally spaced.
What happens if we place, say, five positive charges in a line across from five negative charges, as in ? Now the region between the lines of charge contains a fairly uniform electric field.
We can extend this idea even further and into two dimensions by placing two metallic plates face to face and charging one with positive charge and the other with an equal magnitude of negative charge. This can be done by connecting one plate to the positive terminal of a battery and the other plate to the negative terminal, as shown in . The electric field between these charged plates will be extremely uniform.
Let’s think about the work required to charge these plates. Before the plates are connected to the battery, they are neutral—that is, they have zero net charge. Placing the first positive charge on the left plate and the first negative charge on the right plate requires very little work, because the plates are neutral, so no opposing charges are present. Now consider placing a second positive charge on the left plate and a second negative charge on the right plate. Because the first two charges repel the new arrivals, a force must be applied to the two new charges over a distance to put them on the plates. This is the definition of work, which means that, compared with the first pair, more work is required to put the second pair of charges on the plates. To place the third positive and negative charges on the plates requires yet more work, and so on. Where does this work come from? The battery! Its chemical potential energy is converted into the work required to separate the positive and negative charges.
Although the battery does work, this work remains within the battery-plate system. Therefore, conservation of energy tells us that, if the potential energy of the battery decreases to separate charges, the energy of another part of the system must increase by the same amount. In fact, the energy from the battery is stored in the electric field between the plates. This idea is analogous to considering that the potential energy of a raised hammer is stored in Earth’s gravitational field. If the gravitational field were to disappear, the hammer would have no potential energy. Likewise, if no electric field existed between the plates, no energy would be stored between them.
If we now disconnect the plates from the battery, they will hold the energy. We could connect the plates to a lightbulb, for example, and the lightbulb would light up until this energy was used up. These plates thus have the capacity to store energy. For this reason, an arrangement such as this is called a capacitor. A capacitor is an arrangement of objects that, by virtue of their geometry, can store energy an electric field.
Various real capacitors are shown in . They are usually made from conducting plates or sheets that are separated by an insulating material. They can be flat or rolled up or have other geometries.
The capacity of a capacitor is defined by its capacitance C, which is given by
where Q is the magnitude of the charge on each capacitor plate, and V is the potential difference in going from the negative plate to the positive plate. This means that both Q and V are always positive, so the capacitance is always positive. We can see from the equation for capacitance that the units of capacitance are C/V, which are called farads (F) after the nineteenth-century English physicist Michael Faraday.
The equation
makes sense: A parallel-plate capacitor (like the one shown in ) the size of a football field could hold a lot of charge without requiring too much work per unit charge to push the charge into the capacitor. Thus, Q would be large, and V would be small, so the capacitance C would be very large. Squeezing the same charge into a capacitor the size of a fingernail would require much more work, so V would be very large, and the capacitance would be much smaller.
Although the equation
makes it seem that capacitance depends on voltage, in fact it does not. For a given capacitor, the ratio of the charge stored in the capacitor to the voltage difference between the plates of the capacitor always remains the same. Capacitance is determined by the geometry of the capacitor and the materials that it is made from. For a parallel-plate capacitor with nothing between its plates, the capacitance is given by
where A is the area of the plates of the capacitor and d is their separation. We use
instead of C, because the capacitor has nothing between its plates (in the next section, we’ll see what happens when this is not the case). The constant
read epsilon zero is called the permittivity of free space, and its value is
Coming back to the energy stored in a capacitor, we can ask exactly how much energy a capacitor stores. If a capacitor is charged by putting a voltage V across it for example, by connecting it to a battery with voltage V—the electrical potential energy stored in the capacitor is
Notice that the form of this equation is similar to that for kinetic energy,
.
### Practice Problems
### Dielectrics
Before working through some sample problems, let’s look at what happens if we put an insulating material between the plates of a capacitor that has been charged and then disconnected from the charging battery, as illustrated in . Because the material is insulating, the charge cannot move through it from one plate to the other, so the charge Q on the capacitor does not change. An electric field exists between the plates of a charged capacitor, so the insulating material becomes polarized, as shown in the lower part of the figure. An electrically insulating material that becomes polarized in an electric field is called a dielectric.
shows that the negative charge in the molecules in the material shifts to the left, toward the positive charge of the capacitor. This shift is due to the electric field, which applies a force to the left on the electrons in the molecules of the dielectric. The right sides of the molecules are now missing a bit of negative charge, so their net charge is positive.
All electrically insulating materials are dielectrics, but some are better dielectrics than others. A good dielectric is one whose molecules allow their electrons to shift strongly in an electric field. In other words, an electric field pulls their electrons a fair bit away from their atom, but they do not escape completely from their atom (which is why they are insulators).
shows a macroscopic view of a dielectric in a charged capacitor. Notice that the electric-field lines in the capacitor with the dielectric are spaced farther apart than the electric-field lines in the capacitor with no dielectric. This means that the electric field in the dielectric is weaker, so it stores less electrical potential energy than the electric field in the capacitor with no dielectric.
Where has this energy gone? In fact, the molecules in the dielectric act like tiny springs, and the energy in the electric field goes into stretching these springs. With the electric field thus weakened, the voltage difference between the two sides of the capacitor is smaller, so it becomes easier to put more charge on the capacitor. Placing a dielectric in a capacitor before charging it therefore allows more charge and potential energy to be stored in the capacitor. A parallel plate with a dielectric has a capacitance of
where
(kappa) is a dimensionless constant called the dielectric constant. Because
is greater than 1 for dielectrics, the capacitance increases when a dielectric is placed between the capacitor plates. The dielectric constant of several materials is shown in .
### Practice Problems
### Check Your Understanding
### Section Summary
1. The capacitance of a capacitor depends only on the geometry of the capacitor and the materials from which it is made. It does not depend on the voltage across the capacitor.
2. Capacitors store electrical energy in the electric field between their plates.
3. A dielectric material is an insulator that is polarized in an electric field.
4. Putting a dielectric between the plates of a capacitor increases the capacitance of the capacitor.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Performance Task
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Electrical Circuits
## Introduction
The flicker of numbers on a handheld calculator, nerve impulses carrying signals of vision to the brain, an ultrasound device sending a signal to a computer screen, the brain sending a message for a baby to twitch its toes, an electric train pulling into a station, a hydroelectric plant sending energy to metropolitan and rural users—these and many other examples of electricity involve electric current, which is the movement of charge. Humanity has harnessed electricity, the basis of this technology, to improve our quality of life. Whereas the previous chapter concentrated on static electricity and the fundamental force underlying its behavior, the next two chapters will be devoted to electric and magnetic phenomena involving current. In addition to exploring applications of electricity, we shall gain new insights into the workings of nature. |
# Electrical Circuits
## Ohm's law
### Section Key Terms
### Direct and Alternating Current
Just as water flows from high to low elevation, electrons that are free to move will travel from a place with low potential to a place with high potential. A battery has two terminals that are at different potentials. If the terminals are connected by a conducting wire, an electric current (charges) will flow, as shown in . Electrons will then move from the low-potential terminal of the battery (the negative end) through the wire and enter the high-potential terminal of the battery (the positive end).
Electric current is the rate at which electric charge moves. A large current, such as that used to start a truck engine, moves a large amount very quickly, whereas a small current, such as that used to operate a hand-held calculator, moves a small amount of charge more slowly. In equation form, electric current I is defined as
where
is the amount of charge that flows past a given area and
is the time it takes for the charge to move past the area. The SI unit for electric current is the ampere (A), which is named in honor of the French physicist André-Marie Ampère (1775–1836). One ampere is one coulomb per second, or
Electric current moving through a wire is in many ways similar to water current moving through a pipe. To define the flow of water through a pipe, we can count the water molecules that flow past a given section of the pipe. As shown in , electric current is very similar. We count the number of electrical charges that flow past a section of a conductor; in this case, a wire.
Assume each particle q in carries a charge
, in which case the total charge shown would be
. If these charges move past the area A in a time
, then the current would be
Note that we assigned a positive charge to the charges in . Normally, negative charges—electrons—are the mobile charge in wires, as indicated in . Positive charges are normally stuck in place in solids and cannot move freely. However, because a positive current moving to the right is the same as a negative current of equal magnitude moving to the left, as shown in , we define conventional current to flow in the direction that a positive charge would flow if it could move. Thus, unless otherwise specified, an electric current is assumed to be composed of positive charges.
Also note that one Coulomb is a significant amount of electric charge, so 5 A is a very large current. Most often you will see current on the order of milliamperes (mA).
The direction of conventional current is the direction that positive charge would flow. Depending on the situation, positive charges, negative charges, or both may move. In metal wires, as we have seen, current is carried by electrons, so the negative charges move. In ionic solutions, such as salt water, both positively charged and negatively charged ions move. This is also true in nerve cells. Pure positive currents are relatively rare but do occur. History credits American politician and scientist Benjamin Franklin with describing current as the direction that positive charges flow through a wire. He named the type of charge associated with electrons negative long before they were known to carry current in so many situations.
As electrons move through a metal wire, they encounter obstacles such as other electrons, atoms, impurities, etc. The electrons scatter from these obstacles, as depicted in . Normally, the electrons lose energy with each interaction. This energy is transferred to the wire and becomes thermal energy, which is what makes wires hot when they carry a lot of current. To keep the electrons moving thus requires a force, which is supplied by an electric field. The electric field in a wire points from the end of the wire at the higher potential to the end of the wire at the lower potential. Electrons, carrying a negative charge, move on average (or drift) in the direction opposite the electric field, as shown in .
So far, we have discussed current that moves constantly in a single direction. This is called direct current, because the electric charge flows in only one direction. Direct current is often called DC current.
Many sources of electrical power, such as the hydroelectric dam shown at the beginning of this chapter, produce alternating current, in which the current direction alternates back and forth. Alternating current is often called AC current. Alternating current moves back and forth at regular time intervals, as shown in . The alternating current that comes from a normal wall socket does not suddenly switch directions. Rather, it increases smoothly up to a maximum current and then smoothly decreases back to zero. It then grows again, but in the opposite direction until it has reached the same maximum value. After that, it decreases smoothly back to zero, and the cycle starts over again.
Devices that use AC include vacuum cleaners, fans, power tools, hair dryers, and countless others. These devices obtain the power they require when you plug them into a wall socket. The wall socket is connected to the power grid that provides an alternating potential (AC potential). When your device is plugged in, the AC potential pushes charges back and forth in the circuit of the device, creating an alternating current.
Many devices, however, use DC, such as computers, cell phones, flashlights, and cars. One source of DC is a battery, which provides a constant potential (DC potential) between its terminals. With your device connected to a battery, the DC potential pushes charge in one direction through the circuit of your device, creating a DC current. Another way to produce DC current is by using a transformer, which converts AC potential to DC potential. Small transformers that you can plug into a wall socket are used to charge up your laptop, cell phone, or other electronic device. People generally call this a charger or a battery, but it is a transformer that transforms AC voltage into DC voltage. The next time someone asks to borrow your laptop charger, tell them that you don’t have a laptop charger, but that they may borrow your converter.
### Practice Problems
### Resistance and Ohm’s Law
As mentioned previously, electrical current in a wire is in many ways similar to water flowing through a pipe. The water current that can flow through a pipe is affected by obstacles in the pipe, such as clogs and narrow sections in the pipe. These obstacles slow down the flow of current through the pipe. Similarly, electrical current in a wire can be slowed down by many factors, including impurities in the metal of the wire or collisions between the charges in the material. These factors create a resistance to the electrical current. Resistance is a description of how much a wire or other electrical component opposes the flow of charge through it.
In the 19th century, the German physicist Georg Simon Ohm (1787–1854) found experimentally that current through a conductor is proportional to the voltage drop across a current-carrying conductor.
The constant of proportionality is the resistance R of the material, which leads to
This relationship is called Ohm’s law. It can be viewed as a cause-and-effect relationship, with voltage being the cause and the current being the effect. Ohm’s law is an empirical law like that for friction, which means that it is an experimentally observed phenomenon. The units of resistance are volts per ampere, or V/A. We call a V/A an ohm, which is represented by the uppercase Greek letter omega (
). Thus,
Ohm’s law holds for most materials and at common temperatures. At very low temperatures, resistance may drop to zero (superconductivity). At very high temperatures, the thermal motion of atoms in the material inhibits the flow of electrons, increasing the resistance. The many substances for which Ohm’s law holds are called ohmic. Ohmic materials include good conductors like copper, aluminum, and silver, and some poor conductors under certain circumstances. The resistance of ohmic materials remains essentially the same for a wide range of voltage and current.
### Practice Problems
### Check Your Understanding
### Section Summary
1. Direct current is constant over time; alternating current alternates smoothly back and forth over time.
2. Electrical resistance causes materials to extract work from the current that flows through them.
3. In ohmic materials, voltage drop along a path is proportional to the current that runs through the path.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Electrical Circuits
## Series Circuits
### Section Key Terms
### Electric Circuits and Resistors
Now that we understand the concept of electric current, let’s see what we can do with it. As you are no doubt aware, the modern lifestyle relies heavily on electrical devices. These devices contain ingenious electric circuits, which are complete, closed pathways through which electric current flows. Returning to our water analogy, an electric circuit is to electric charge like a network of pipes is to water: The electric circuit guides electric charge from one point to the next, running the charge through various devices along the way to extract work or information.
Electric circuits are made from many materials and cover a huge range of sizes, as shown in . Computers and cell phones contain electric circuits whose features can be as small as roughly a billionth of a meter (a nanometer, or
). The pathways that guide the current in these devices are made by ultraprecise chemical treatments of silicon or other semiconductors. Large power systems, on the other hand, contain electric circuits whose features are on the scale of meters. These systems carry such large electric currents that their physical dimensions must be relatively large.
The pathways that form electric circuits are made from a conducting material, normally a metal in macroscopic circuits. For example, copper wires inside your school building form the electrical circuits that power lighting, projectors, screens, speakers, etc. To represent an electric circuit, we draw circuit diagrams. We use lines and symbols to represent the elements in the circuit. A simple electric circuit diagram is shown on the left side of . On the right side is an analogous water circuit, which we discuss below.
There are many different symbols that scientists and engineers use in circuit diagrams, but we will focus on four main symbols: the wire, the battery or voltage source, resistors, and the ground. The thin black lines in the electric circuit diagram represent the pathway that the electric charge must follow. These pathways are assumed to be perfect conductors, so electric charge can move along these pathways without losing any energy. In reality, the wires in circuits are not perfect, but they come close enough for our purposes.
The zigzag element labeled R is a resistor, which is a circuit element that provides a known resistance. Macroscopic resistors are often color coded to indicate their resistance, as shown in .
The red element in is a battery, with its positive and negative terminals indicated; the longer line represents the positive terminal of the battery, and the shorter line represents the negative terminal. Note that the battery icon is not always colored red; this is done in just to make it easy to identify.
Finally, the element labeled ground on the lower left of the circuit indicates that the circuit is connected to Earth, which is a large, essentially neutral object containing an infinite amount of charge. Among other things, the ground determines the potential of the negative terminal of the battery. Normally, the potential of the ground is defined to be zero:
. This means that the entire lower wire in is at a voltage of zero volts.
The electric current in is indicted by the blue line labeled I. The arrow indicates the direction in which positive charge would flow in this circuit. Recall that, in metals, electrons are mobile charge carriers, so negative charges actually flow in the opposite direction around this circuit (i.e., counterclockwise). However, we draw the current to show the direction in which positive charge would move.
On the right side of is an analogous water circuit. Water at a higher pressure leaves the top of the pump, which is like charges leaving the positive terminal of the battery. The water travels through the pipe, like the charges traveling through the wire. Next, the water goes through a sand filter, which heats up as the water squeezes through. This step is like the charges going through the resistor. When charges flow through a resistor, they do work to heat up the resistor. After flowing through the sand filter, the water has converted its potential energy into heat, so it is at a lower pressure. Likewise, the charges exiting the resistor have converted their potential energy into heat, so they are at a lower voltage. Recall that voltage is just potential energy per charge. Thus, water pressure is analogous to electric potential energy (i.e., voltage). Coming back to the water circuit again, we see that the water returns to the bottom of the pump, which is like the charge returning to the negative terminal of the battery. The water pump uses a source of energy to pump the water back up to a high pressure again, giving it the pressure required to go through the circuit once more. The water pump is like the battery, which uses chemical energy to increase the voltage of the charge up to the level of the positive terminal.
The potential energy per charge at the positive terminal of the battery is the voltage rating of the battery. This voltage is like water pressure in the upper pipe. Just like a higher pressure forces water to move toward a lower pressure, a higher voltage forces electric charge to flow toward a lower voltage. The pump takes water at low pressure and does work on it, ejecting water at a higher pressure. Likewise, a battery takes charge at a low voltage, does work on it, and ejects charge at a higher voltage.
Note that the current in the water circuit of is the same throughout the circuit. In other words, if we measured the number of water molecules passing a cross-section of the pipe per unit time at any point in the circuit, we would get the same answer no matter where in the circuit we measured. The same is true of the electrical circuit in the same figure. The electric current is the same at all points in this circuit, including inside the battery and in the resistor. The electric current neither speeds up in the wires nor slows down in the resistor. This would create points where too much or too little charge would be bunched up. Thus, the current is the same at all points in the circuit shown in .
Although the current is the same everywhere in both the electric and water circuits, the voltage or water pressure changes as you move through the circuits. In the water circuit, the water pressure at the pump outlet stays the same until the water goes through the sand filter, assuming no energy loss in the pipe. Likewise, the voltage in the electrical circuit is the same at all points in a given wire, because we have assumed that the wires are perfect conductors. Thus, as indicated by the constant red color of the upper wire in , the voltage throughout this wire is constant at
. The voltage then drops as you go through the resistor, but once you reach the blue wire, the voltage stays at its new level of
all the way to the negative terminal of the battery (i.e., the blue terminal of the battery).
If we go from the blue wire through the battery to the red wire, the voltage increases from
to
. Likewise, if we go from the blue wire up through the resistor to the red wire, the voltage also goes from
to
. Thus, using Ohm’s law, we can write
Note that
is measured from the bottom of the resistor to the top, meaning that the top of the resistor is at a higher voltage than the bottom of the resistor. Thus, current flows from the top of the resistor or higher voltage to the bottom of the resistor or lower voltage.
Other possible circuit elements include capacitors and switches. These are drawn as shown on the left side of . A switch is a device that opens and closes the circuit, like a light switch. It is analogous to a valve in a water circuit, as shown on the right side of . With the switch open, no current passes through the circuit. With the switch closed, it becomes part of the wire, so the current passes through it with no loss of voltage.
The capacitor is labeled C on the left of . A capacitor in an electrical circuit is analogous to a flexible membrane in a water circuit. When the switch is closed in the circuit of , the battery forces electrical current to flow toward the capacitor, charging the upper capacitor plate with positive charge. As this happens, the voltage across the capacitor plates increases. This is like the membrane in the water circuit: When the valve is opened, the pump forces water to flow toward the membrane, making it stretch to store the excess water. As this happens, the pressure behind the membrane increases.
Now if we open the switch, the capacitor holds the voltage between its plates because the charges have nowhere to go. Likewise, if we close the valve, the water has nowhere to go and the membrane maintains the water pressure in the pipe between itself and the valve.
If the switch is closed for a long time in the electric circuit or if the valve is open for a long time in the water circuit, the current will eventually stop flowing because the capacitor or the membrane will have become completely charged. Each circuit is now in the steady state, which means that its characteristics do not change over time. In this case, the steady state is characterized by zero current, and this does not change as long as the switch or valve remains in the same position. In the steady state, no electrical current passes through the capacitor, and no water current passes through the membrane. The voltage difference between the capacitor plates will be the same as the battery voltage. In the water circuit, the pressure behind the membrane will be the same as the pressure created by the pump.
Although the circuit in may seem a bit pointless because all that happens when the switch is closed is that the capacitor charges up, it does show the capacitor’s ability to store charge. Thus, the capacitor serves as a reservoir for charge. This property of capacitors is used in circuits in many ways. For example, capacitors are used to power circuits while batteries are being charged. In addition, capacitors can serve as filters. To understand this, let’s go back to the water analogy. Suppose you have a water hose and are watering your garden. Your friend thinks he’s funny, and kinks the hose. While the hose is kinked, you experience no water flow. When he lets go, the water starts flowing again. If he does this really fast, you experience water-no water-water-no water, and that’s really no way to water your garden. Now imagine that the hose is filling up a big bucket, and you are watering from the bottom of the bucket. As long as you had water in your bucket to begin with and your friend doesn’t kink the water hose for too long, you would be able to water your garden without the interruptions. Your friend kinking the water hose is filtered by the big bucket’s supply of water, so it does not impact your ability to water the garden. We can think of the interruptions in the current (be it water or electrical current) as noise. Capacitors act in an analogous way as the water bucket to help filter out the noise. Capacitors have so many uses that it is very rare to find an electronic circuit that does not include some capacitors.
### Resistors in Series and Equivalent Resistance
Now that we have a basic idea of how electrical circuits work, let’s see what happens in circuits with more than one circuit element. In this section, we look at resistors in series. Components connected in series are connected one after the other in the same branch of a circuit, such as the resistors connected in series on the left side of .
We will now try to find a single resistance that is equivalent to the three resistors in series on the left side of . An equivalent resistor is a resistor that has the same resistance as the combined resistance of a set of other resistors. In other words, the same current will flow through the left and right circuits in if we use the equivalent resistor in the right circuit.
According to Ohm’s law, the voltage drop V across a resistor when a current flows through it is
where I is the current in amperes (A) and R is the resistance in ohms (
). Another way to think of this is that V is the voltage necessary to make a current I flow through a resistance R. Applying Ohm’s law to each resistor on the left circuit of , we find that the voltage drop across
is
, that across
is
, and that across
is
. The sum of these voltages equals the voltage output of the battery, that is
You may wonder why voltages must add up like this. One way to understand this is to go once around the circuit and add up the successive changes in voltage. If you do this around a loop and get back to the starting point, the total change in voltage should be zero, because you end up at the same place that you started. To better understand this, consider the analogy of going for a stroll through some hilly countryside. If you leave your car and walk around, then come back to your car, the total height you gained in your stroll must be the same as the total height you lost, because you end up at the same place as you started. Thus, the gravitational potential energy you gain must be the same as the gravitational potential energy you lose. The same reasoning holds for voltage in going around an electric circuit. Let’s apply this reasoning to the left circuit in . We start just below the battery and move up through the battery, which contributes a voltage gain of
. Next, we got through the resistors. The voltage drops by
in going through resistor
, by
in going through resistor
, and by
in going through resistor
. After going through resistor
, we arrive back at the starting point, so we add up these four changes in voltage and set the sum equal to zero. This gives
which is the same as the previous equation. Note that the minus signs in front of
are because these are voltage drops, whereas
is a voltage rise.
Ohm’s law tells us that
,
, and
. Inserting these values into equation
gives
Applying this same logic to the right circuit in
gives
Dividing the equation
by
, we get
This shows that the equivalent resistance for a series of resistors is simply the sum of the resistances of each resistor. In general, N resistors connected in series can be replaced by an equivalent resistor with a resistance of
### Check your Understanding
### Section Summary
1. Circuit diagrams are schematic representations of electric circuits.
2. Resistors in series are resistors that are connected head to tail.
3. The same current runs through all resistors in series; however, the voltage drop across each resistor can be different.
4. The voltage is the same at every point in a given wire.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Electrical Circuits
## Parallel Circuits
### Section Key Terms
### Resistors in Parallel
In the previous section, we learned that resistors in series are resistors that are connected one after the other. If we instead combine resistors by connecting them next to each other, as shown in , then the resistors are said to be connected in parallel. Resistors are in parallel when both ends of each resistor are connected directly together.
Note that the tops of the resistors are all connected to the same wire, so the voltage at the top of the each resistor is the same. Likewise, the bottoms of the resistors are all connected to the same wire, so the voltage at the bottom of each resistor is the same. This means that the voltage drop across each resistor is the same. In this case, the voltage drop is the voltage rating V of the battery, because the top and bottom wires connect to the positive and negative terminals of the battery, respectively.
Although the voltage drop across each resistor is the same, we cannot say the same for the current running through each resistor. Thus,
are not necessarily the same, because the resistors
do not necessarily have the same resistance.
Note that the three resistors in provide three different paths through which the current can flow. This means that the equivalent resistance for these three resistors must be less than the smallest of the three resistors. To understand this, imagine that the smallest resistor is the only path through which the current can flow. Now add on the alternate paths by connecting other resistors in parallel. Because the current has more paths to go through, the overall resistance (i.e., the equivalent resistance) will decrease. Therefore, the equivalent resistance must be less than the smallest resistance of the parallel resistors.
To find the equivalent resistance
of the three resistors
, we apply Ohm’s law to each resistor. Because the voltage drop across each resistor is V, we obtain
or
We also know from conservation of charge that the three currents
must add up to give the current I that goes through the battery. If this were not true, current would have to be mysteriously created or destroyed somewhere in the circuit, which is physically impossible. Thus, we have
Inserting the expressions for
into this equation gives
or
This formula is just Ohm’s law, with the factor in parentheses being the equivalent resistance.
Thus, the equivalent resistance for three resistors in parallel is
The same logic works for any number of resistors in parallel, so the general form of the equation that gives the equivalent resistance of N resistors connected in parallel is
### Practice Problems
### Resistors in Parallel and in Series
More complex connections of resistors are sometimes just combinations of series and parallel. Combinations of series and parallel resistors can be reduced to a single equivalent resistance by using the technique illustrated in . Various parts are identified as either series or parallel, reduced to their equivalents, and further reduced until a single resistance is left. The process is more time consuming than difficult.
Let’s work through the four steps in to reduce the seven resistors to a single equivalent resistor. To avoid distracting algebra, we’ll assume each resistor is 10
. In step 1, we reduce the two sets of parallel resistors circled by the blue dashed loop. The upper set has three resistors in parallel and will be reduced to a single equivalent resistor
. The lower set has two resistors in parallel and will be reduced to a single equivalent resistor
. Using the equation for the equivalent resistance of resistors in parallel, we obtain
These two equivalent resistances are encircled by the red dashed loop following step 1. They are in series, so we can use the equation for the equivalent resistance of resistors in series to reduce them to a single equivalent resistance
. This is done in step 2, with the result being
The equivalent resistor
appears in the green dashed loop following step 2. This resistor is in parallel with resistor
, so the pair can be replaced by the equivalent resistor
, which is given by
This is done in step 3. The resistor
is in series with the resistor
, as shown in the purple dashed loop following step 3. These two resistors are combined in the final step to form the final equivalent resistor
, which is
Thus, the entire combination of seven resistors may be replaced by a single resistor with a resistance of about 14.5
.
That was a lot of work, and you might be asking why we do it. It’s important for us to know the equivalent resistance of the entire circuit so that we can calculate the current flowing through the circuit. Ohm’s law tells us that the current flowing through a circuit depends on the resistance of the circuit and the voltage across the circuit. But to know the current, we must first know the equivalent resistance.
Here is a general approach to find the equivalent resistor for any arbitrary combination of resistors:
1. Identify a group of resistors that are only in parallel or only in series.
2. For resistors in series, use the equation for the equivalent resistance of resistors in series to reduce them to a single equivalent resistance. For resistors in parallel, use the equation for the equivalent resistance of resistors in parallel to reduce them to a single equivalent resistance.
3. Draw a new circuit diagram with the resistors from step 1 replaced by their equivalent resistor.
4. If more than one resistor remains in the circuit, return to step 1 and repeat. Otherwise, you are finished.
### Practice Problems
### Check Your Understanding
### Section Summary
1. The equivalent resistance of a group of N identical resistors R connected in parallel is R/N.
2. Connecting resistors in parallel provides more paths for the current to go through, so the equivalent resistance is always less than the smallest resistance of the parallel resistors.
3. The same voltage drop occurs across all resistors in parallel; however, the current through each resistor can differ.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Electrical Circuits
## Electric Power
### Section Key Terms
Power is associated by many people with electricity. Every day, we use electric power to run our modern appliances. Electric power transmission lines are visible examples of electricity providing power. We also use electric power to start our cars, to run our computers, or to light our homes. Power is the rate at which energy of any type is transferred; electric power is the rate at which electric energy is transferred in a circuit. In this section, we’ll learn not only what this means, but also what factors determine electric power.
To get started, let’s think of light bulbs, which are often characterized in terms of their power ratings in watts. Let us compare a 25-W bulb with a 60-W bulb (see ). Although both operate at the same voltage, the 60-W bulb emits more light intensity than the 25-W bulb. This tells us that something other than voltage determines the power output of an electric circuit.
Incandescent light bulbs, such as the two shown in , are essentially resistors that heat up when current flows through them and they get so hot that they emit visible and invisible light. Thus the two light bulbs in the photo can be considered as two different resistors. In a simple circuit such as a light bulb with a voltage applied to it, the resistance determines the current by Ohm’s law, so we can see that current as well as voltage must determine the power.
The formula for power may be found by dimensional analysis. Consider the units of power. In the SI system, power is given in watts (W), which is energy per unit time, or J/s
Recall now that a voltage is the potential energy per unit charge, which means that voltage has units of J/C
We can rewrite this equation as
and substitute this into the equation for watts to get
But a Coulomb per second (C/s) is an electric current, which we can see from the definition of electric current,
, where
Q is the charge in coulombs and
t is time in seconds. Thus, equation above tells us that electric power is voltage times current, or
This equation gives the electric power consumed by a circuit with a voltage drop of V and a current of I.
For example, consider the circuit in . From Ohm’s law, the current running through the circuit is
Thus, the power consumed by the circuit is
Where does this power go? In this circuit, the power goes primarily into heating the resistor in this circuit.
In calculating the power in the circuit of , we used the resistance and Ohm’s law to find the current. Ohm’s law gives the current:
, which we can insert into the equation for electric power to obtain
This gives the power in terms of only the voltage and the resistance.
We can also use Ohm’s law to eliminate the voltage in the equation for electric power and obtain an expression for power in terms of just the current and the resistance. If we write Ohm’s law as
and use this to eliminate V in the equation
, we obtain
This gives the power in terms of only the current and the resistance.
Thus, by combining Ohm’s law with the equation
for electric power, we obtain two more expressions for power: one in terms of voltage and resistance and one in terms of current and resistance. Note that only resistance (not capacitance or anything else), current, and voltage enter into the expressions for electric power. This means that the physical characteristic of a circuit that determines how much power it dissipates is its resistance. Any capacitors in the circuit do not dissipate electric power—on the contrary, capacitors either store electric energy or release electric energy back to the circuit.
To clarify how voltage, resistance, current, and power are all related, consider , which shows the formula wheel. The quantities in the center quarter circle are equal to the quantities in the corresponding outer quarter circle. For example, to express a potential V in terms of power and current, we see from the formula wheel that
.
### Practice Problems
### Check your Understanding
### Section Summary
1. Electric power is dissipated in the resistances of a circuit. Capacitors do not dissipate electric power.
2. Electric power is proportional to the voltage and the current in a circuit.
3. Ohm’s law provides two extra expressions for electric power: one that does not involve current and one that does not involve voltage.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Performance Task
### Test Prep Extended Response
|
# Magnetism
## Introduction
You may have encountered magnets for the first time as a small child playing with magnetic toys or refrigerator magnets. At the time, you likely noticed that two magnets that repulse each other will attract each other if you flip one of them around. The force that acts across the air gaps between magnets is the same force that creates wonders such as the Aurora Borealis. In fact, magnetic effects pervade our lives in myriad ways, from electric motors to medical imaging and computer memory. In this chapter, we introduce magnets and learn how they work and how magnetic fields and electric currents interact. |
# Magnetism
## Magnetic Fields, Field Lines, and Force
### Section Key Terms
### Magnets and Magnetization
People have been aware of magnets and magnetism for thousands of years. The earliest records date back to ancient times, particularly in the region of Asia Minor called Magnesia—the name of this region is the source of words like magnet. Magnetic rocks found in Magnesia, which is now part of western Turkey, stimulated interest during ancient times. When humans first discovered magnetic rocks, they likely found that certain parts of these rocks attracted bits of iron or other magnetic rocks more strongly than other parts. These areas are called the poles of a magnet. A magnetic pole is the part of a magnet that exerts the strongest force on other magnets or magnetic material, such as iron. For example, the poles of the bar magnet shown in are where the paper clips are concentrated.
If a bar magnet is suspended so that it rotates freely, one pole of the magnet will always turn toward the north, with the opposite pole facing south. This discovery led to the compass, which is simply a small, elongated magnet mounted so that it can rotate freely. An example of a compass is shown . The pole of the magnet that orients northward is called the north pole, and the opposite pole of the magnet is called the south pole.
The discovery that one particular pole of a magnet orients northward, whereas the other pole orients southward allowed people to identify the north and south poles of any magnet. It was then noticed that the north poles of two different magnets repel each other, and likewise for the south poles. Conversely, the north pole of one magnet attracts the south pole of other magnets. This situation is analogous to that of electric charge, where like charges repel and unlike charges attract. In magnets, we simply replace charge with pole: Like poles repel and unlike poles attract. This is summarized in , which shows how the force between magnets depends on their relative orientation.
Consider again the fact that the pole of a magnet that orients northward is called the north pole of the magnet. If unlike poles attract, then the magnetic pole of Earth that is close to the geographic North Pole must be a magnetic south pole! Likewise, the magnetic pole of Earth that is close to the geographic South Pole must be a magnetic north pole. This situation is depicted in , in which Earth is represented as containing a giant internal bar magnet with its magnetic south pole at the geographic North Pole and vice versa. If we were to somehow suspend a giant bar magnet in space near Earth, then the north pole of the space magnet would be attracted to the south pole of Earth’s internal magnet. This is in essence what happens with a compass needle: Its magnetic north pole is attracted to the magnet south pole of Earth’s internal magnet.
What happens if you cut a bar magnet in half? Do you obtain one magnet with two south poles and one magnet with two north poles? The answer is no: Each half of the bar magnet has a north pole and a south pole. You can even continue cutting each piece of the bar magnet in half, and you will always obtain a new, smaller magnet with two opposite poles. As shown in , you can continue this process down to the atomic scale, and you will find that even the smallest particles that behave as magnets have two opposite poles. In fact, no experiment has ever found any object with a single magnetic pole, from the smallest subatomic particle such as electrons to the largest objects in the universe such as stars. Because magnets always have two poles, they are referred to as magnetic dipoles—di means two. Below, we will see that magnetic dipoles have properties that are analogous to electric dipoles.
Only certain materials, such as iron, cobalt, nickel, and gadolinium, exhibit strong magnetic effects. Such materials are called ferromagnetic, after the Latin word ferrum for iron. Other materials exhibit weak magnetic effects, which are detectable only with sensitive instruments. Not only do ferromagnetic materials respond strongly to magnets—the way iron is attracted to magnets—but they can also be magnetized themselves—that is, they can be induced to be magnetic or made into permanent magnets (). A permanent magnet is simply a material that retains its magnetic behavior for a long time, even when exposed to demagnetizing influences.
When a magnet is brought near a previously unmagnetized ferromagnetic material, it causes local magnetization of the material with unlike poles closest, as in the right side of . This causes an attractive force, which is why unmagnetized iron is attracted to a magnet.
What happens on a microscopic scale is illustrated in Figure 7(a). Regions within the material called domains act like small bar magnets. Within domains, the magnetic poles of individual atoms are aligned. Each atom acts like a tiny bar magnet. Domains are small and randomly oriented in an unmagnetized ferromagnetic object. In response to an external magnetic field, the domains may grow to millimeter size, aligning themselves, as shown in Figure 7(b). This induced magnetization can be made permanent if the material is heated and then cooled, or simply tapped in the presence of other magnets.
Conversely, a permanent magnet can be demagnetized by hard blows or by heating it in the absence of another magnet. Increased thermal motion at higher temperature can disrupt and randomize the orientation and size of the domains. There is a well-defined temperature for ferromagnetic materials, which is called the Curie temperature, above which they cannot be magnetized. The Curie temperature for iron is 1,043 K (770 ), which is well above room temperature. There are several elements and alloys that have Curie temperatures much lower than room temperature and are ferromagnetic only below those temperatures.
### Magnetic Fields
We have thus seen that forces can be applied between magnets and between magnets and ferromagnetic materials without any contact between the objects. This is reminiscent of electric forces, which also act over distances. Electric forces are described using the concept of the electric field, which is a force field around electric charges that describes the force on any other charge placed in the field. Likewise, a magnet creates a magnetic field around it that describes the force exerted on other magnets placed in the field. As with electric fields, the pictorial representation of magnetic field lines is very useful for visualizing the strength and direction of the magnetic field.
As shown in , the direction of magnetic field lines is defined to be the direction in which the north pole of a compass needle points. If you place a compass near the north pole of a magnet, the north pole of the compass needle will be repelled and point away from the magnet. Thus, the magnetic field lines point away from the north pole of a magnet and toward its south pole.
Magnetic field lines can be mapped out using a small compass. The compass is moved from point to point around a magnet, and at each point, a short line is drawn in the direction of the needle, as shown in . Joining the lines together then reveals the path of the magnetic field line. Another way to visualize magnetic field lines is to sprinkle iron filings around a magnet. The filings will orient themselves along the magnetic field lines, forming a pattern such as that shown on the right in .
When two magnets are brought close together, the magnetic field lines are perturbed, just as happens for electric field lines when two electric charges are brought together. Bringing two north poles together—or two south poles—will cause a repulsion, and the magnetic field lines will bend away from each other. This is shown in , which shows the magnetic field lines created by the two closely separated north poles of a bar magnet. When opposite poles of two magnets are brought together, the magnetic field lines join together and become denser between the poles. This situation is shown in .
Like the electric field, the magnetic field is stronger where the lines are denser. Thus, between the two north poles in , the magnetic field is very weak because the density of the magnetic field is almost zero. A compass placed at that point would essentially spin freely if we ignore Earth’s magnetic field. Conversely, the magnetic field lines between the north and south poles in are very dense, indicating that the magnetic field is very strong in this region. A compass placed here would quickly align with the magnetic field and point toward the south pole on the right.
Note that magnets are not the only things that make magnetic fields. Early in the nineteenth century, people discovered that electrical currents cause magnetic effects. The first significant observation was by the Danish scientist Hans Christian Oersted (1777–1851), who found that a compass needle was deflected by a current-carrying wire. This was the first significant evidence that the movement of electric charges had any connection with magnets. An electromagnet is a device that uses electric current to make a magnetic field. These temporarily induced magnets are called electromagnets. Electromagnets are employed for everything from a wrecking yard crane that lifts scrapped cars to controlling the beam of a 90-km-circumference particle accelerator to the magnets in medical-imaging machines (see ).
The magnetic field created by an electric current in a long straight wire is shown in . The magnetic field lines form concentric circles around the wire. The direction of the magnetic field can be determined using the right-hand rule. This rule shows up in several places in the study of electricity and magnetism. Applied to a straight current-carrying wire, the right-hand rule says that, with your right thumb pointed in the direction of the current, the magnetic field will be in the direction in which your right fingers curl, as shown in . If the wire is very long compared to the distance r from the wire, the strength B of the magnetic field is given by
where I is the current in the wire in amperes. The SI unit for magnetic field is the tesla (T). The symbol
—read “mu-zero”—is a constant called the “permeability of free space” and is given by
Now imagine winding a wire around a cylinder with the cylinder then removed. The result is a wire coil, as shown in . This is called a solenoid. To find the direction of the magnetic field produced by a solenoid, apply the right-hand rule to several points on the coil. You should be able to convince yourself that, inside the coil, the magnetic field points from left to right. In fact, another application of the right-hand rule is to curl your right-hand fingers around the coil in the direction in which the current flows. Your right thumb then points in the direction of the magnetic field inside the coil: left to right in this case.
Each loop of wire contributes to the magnetic field inside the solenoid. Because the magnetic field lines must form closed loops, the field lines close the loop outside the solenoid. The magnetic field lines are much denser inside the solenoid than outside the solenoid. The resulting magnetic field looks very much like that of a bar magnet, as shown in . The magnetic field strength deep inside a solenoid is
where N is the number of wire loops in the solenoid and
is the length of the solenoid.
### Magnetic Force
If a moving electric charge, that is electric current, produces a magnetic field that can exert a force on another magnet, then the reverse should be true by Newton’s third law. In other words, a charge moving through the magnetic field produced by another object should experience a force—and this is exactly what we find. As a concrete example, consider , which shows a charge q moving with velocity
through a magnetic field
between the poles of a permanent magnet. The magnitude F of the force experienced by this charge is
where
is the angle between the velocity of the charge and the magnetic field.
The direction of the force may be found by using another version of the right-hand rule: First, we join the tails of the velocity vector and a magnetic field vector, as shown in step 1 of . We then curl our right fingers from
to
, as indicated in step (2) of . The direction in which the right thumb points is the direction of the force. For the charge in , we find that the force is directed into the page.
Note that the factor
in the equation
means that zero force is applied on a charge that moves parallel to a magnetic field because
and
. The maximum force a charge can experience is when it moves perpendicular to the magnetic field, because
and
Instead of a single charge moving through a magnetic field, consider now a steady current I moving through a straight wire. If we place this wire in a uniform magnetic field, as shown in , what is the force on the wire or, more precisely, on the electrons in the wire? An electric current involves charges that move. If the charges q move a distance
in a time t, then their speed is
Inserting this into the equation
gives
The factor q/t in this equation is nothing more than the current in the wire. Thus, using
, we obtain
This equation gives the force on a straight current-carrying wire of length
in a magnetic field of strength B. The angle
is the angle between the current vector and the magnetic field vector. Note that
is the length of wire that is in the magnetic field and for which
as shown in .
The direction of the force is determined in the same way as for a single charge. Curl your right fingers from the vector for I to the vector for B, and your right thumb will point in the direction of the force on the wire. For the wire shown in , the force is directed into the page.
Throughout this section, you may have noticed the symmetries between magnetic effects and electric effects. These effects all fall under the umbrella of electromagnetism, which is the study of electric and magnetic phenomena. We have seen that electric charges produce electric fields, and moving electric charges produce magnetic fields. A magnetic dipole produces a magnetic field, and, as we will see in the next section, moving magnetic dipoles produce an electric field. Thus, electricity and magnetism are two intimately related and symmetric phenomena.
### Practice Problems
### Check Your Understanding
### Section Summary
1. All magnets have two poles: a north pole and a south pole. If the magnet is free to move, its north pole orients itself toward the geographic North Pole of Earth, and the south pole orients itself toward the geographic South Pole of Earth.
2. A repulsive force occurs between the north poles of two magnets and likewise for two south poles. However, an attractive force occurs between the north pole of one magnet and the south pole of another magnet.
3. A charged particle moving through a magnetic field experiences a force whose direction is determined by the right-hand rule.
4. An electric current generates a magnetic field.
5. Electromagnets are magnets made by passing a current through a system of wires.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Magnetism
## Motors, Generators, and Transformers
### Section Key Terms
### Electric Motors, Generators, and Transformers
As we learned previously, a current-carrying wire in a magnetic field experiences a force—recall
. Electric motors, which convert electrical energy into mechanical energy, are the most common application of magnetic force on current-carrying wires. Motors consist of loops of wire in a magnetic field. When current is passed through the loops, the magnetic field exerts a torque on the loops, which rotates a shaft. Electrical energy is converted to mechanical work in the process. shows a schematic drawing of an electric motor.
Let us examine the force on each segment of the loop in to find the torques produced about the axis of the vertical shaft—this will lead to a useful equation for the torque on the loop. We take the magnetic field to be uniform over the rectangular loop, which has width w and height
as shown in the figure. First, consider the force on the top segment of the loop. To determine the direction of the force, we use the right-hand rule. The current goes from left to right into the page, and the magnetic field goes from left to right in the plane of the page. Curl your right fingers from the current vector to the magnetic field vector and your right thumb points down. Thus, the force on the top segment is downward, which produces no torque on the shaft. Repeating this analysis for the bottom segment—neglect the small gap where the lead wires go out—shows that the force on the bottom segment is upward, again producing no torque on the shaft.
Consider now the left vertical segment of the loop. Again using the right-hand rule, we find that the force exerted on this segment is perpendicular to the magnetic field, as shown in . This force produces a torque on the shaft. Repeating this analysis on the right vertical segment of the loop shows that the force on this segment is in the direction opposite that of the force on the left segment, thereby producing an equal torque on the shaft. The total torque on the shaft is thus twice the toque on one of the vertical segments of the loop.
To find the magnitude of the torque as the wire loop spins, consider , which shows a view of the wire loop from above. Recall that torque is defined as
where F is the applied force, r is the distance from the pivot to where the force is applied, and θ is the angle between r and F. Notice that, as the loop spins, the current in the vertical loop segments is always perpendicular to the magnetic field. Thus, the equation
gives the magnitude of the force on each vertical segment as
The distance r from the shaft to where this force is applied is w/2, so the torque created by this force is
Because there are two vertical segments, the total torque is twice this, or
If we have a multiple loop with N turns, we get N times the torque of a single loop. Using the fact that the area of the loop is
the expression for the torque becomes
This is the torque on a current-carrying loop in a uniform magnetic field. This equation can be shown to be valid for a loop of any shape.
From the equation
we see that the torque is zero when
As the wire loop rotates, the torque increases to a maximum positive torque of
when
The torque then decreases back to zero as the wire loop rotates to
From
to
the torque is negative. Thus, the torque changes sign every half turn, so the wire loop will oscillate back and forth.
For the coil to continue rotating in the same direction, the current is reversed as the coil passes through
using automatic switches called brushes, as shown in .
Consider now what happens if we run the motor in reverse; that is, we attach a handle to the shaft and mechanically force the coil to rotate within the magnetic field, as shown in . As per the equation
—where
is the angle between the vectors
and
in the wires of the loop experience a magnetic force because they are moving in a magnetic field. Again using the right-hand rule, where we curl our fingers from vector
to vector
, we find that charges in the top and bottom segments feel a force perpendicular to the wire, which does not cause a current. However, charges in the vertical wires experience forces parallel to the wire, causing a current to flow through the wire and through an external circuit if one is connected. A device such as this that converts mechanical energy into electrical energy is called a generator.
Because current is induced only in the side wires, we can find the induced emf by only considering these wires. As explained in Induced Current in a Wire, motional emf in a straight wire moving at velocity v through a magnetic field B is
where the velocity is perpendicular to the magnetic field. In the generator, the velocity makes an angle
with B (see ), so the velocity component perpendicular to B is
Thus, in this case, the emf induced on each vertical wire segment is
and they are in the same direction. The total emf around the loop is then
Although this expression is valid, it does not give the emf as a function of time. To find how the emf evolves in time, we assume that the coil is rotated at a constant angular velocity
The angle
is related to the angular velocity by
so that
Recall that tangential velocity v is related to angular velocity
by
Here,
, so that
and
Noting that the area of the loop is
and allowing for N wire loops, we find that
is the emf induced in a generator coil of N turns and area A rotating at a constant angular velocity
in a uniform magnetic field B. This can also be expressed as
where
is the maximum (peak) emf.
shows a generator connected to a light bulb and a graph of the emf vs. time. Note that the emf oscillates from a positive maximum of
to a negative maximum of
In between, the emf goes through zero, which means that zero current flows through the light bulb at these times. Thus, the light bulb actually flickers on and off at a frequency of 2f, because there are two zero crossings per period. Since alternating current such as this is used in homes around the world, why do we not notice the lights flickering on and off? In the United States, the frequency of alternating current is 60 Hz, so the lights flicker on and off at a frequency of 120 Hz. This is faster than the refresh rate of the human eye, so you don’t notice the flicker of the lights. Also, other factors prevent various different types of light bulbs from switching on and off so fast, so the light output is smoothed out a bit.
In real life, electric generators look a lot different than the figures in this section, but the principles are the same. The source of mechanical energy that turns the coil can be falling water—hydropower—steam produced by the burning of fossil fuels, or the kinetic energy of wind. shows a cutaway view of a steam turbine; steam moves over the blades connected to the shaft, which rotates the coil within the generator.
Another very useful and common device that exploits magnetic induction is called a transformer. Transformers do what their name implies—they transform voltages from one value to another; the term voltage is used rather than emf because transformers have internal resistance. For example, many cell phones, laptops, video games, power tools, and small appliances have a transformer built into their plug-in unit that changes 120 V or 240 V AC into whatever voltage the device uses. shows two different transformers. Notice the wire coils that are visible in each device. The purpose of these coils is explained below.
shows a laminated-coil transformer, which is based on Faraday’s law of induction and is very similar in construction to the apparatus Faraday used to demonstrate that magnetic fields can generate electric currents. The two wire coils are called the primary and secondary coils. In normal use, the input voltage is applied across the primary coil, and the secondary produces the transformed output voltage. Not only does the iron core trap the magnetic field created by the primary coil, but also its magnetization increases the field strength, which is analogous to how a dielectric increases the electric field strength in a capacitor. Since the input voltage is AC, a time-varying magnetic flux is sent through the secondary coil, inducing an AC output voltage.
For the transformer shown in , the output voltage
from the secondary coil depends almost entirely on the input voltage
across the primary coil and the number of loops in the primary and secondary coils. Faraday’s law of induction for the secondary coil gives its induced output voltage
to be
where
is the number of loops in the secondary coil and
is the rate of change of magnetic flux. The output voltage equals the induced emf
provided coil resistance is small—a reasonable assumption for transformers. The cross-sectional area of the coils is the same on each side, as is the magnetic field strength, and so
is the same on each side. The input primary voltage
is also related to changing flux by
Taking the ratio of these last two equations yields the useful relationship
This is known as the transformer equation. It simply states that the ratio of the secondary voltage to the primary voltage in a transformer equals the ratio of the number of loops in secondary coil to the number of loops in the primary coil.
### Transmission of Electrical Power
Transformers are widely used in the electric power industry to increase voltages—called step-up transformers—before long-distance transmission via high-voltage wires. They are also used to decrease voltages—called step-down transformers—to deliver power to homes and businesses. The overwhelming majority of electric power is generated by using magnetic induction, whereby a wire coil or copper disk is rotated in a magnetic field. The primary energy required to rotate the coils or disk can be provided by a variety of means. Hydroelectric power plants use the kinetic energy of water to drive electric generators. Coal or nuclear power plants create steam to drive steam turbines that turn the coils. Other sources of primary energy include wind, tides, or waves on water.
Once power is generated, it must be transmitted to the consumer, which often means transmitting power over hundreds of kilometers. To do this, the voltage of the power plant is increased by a step-up transformer, that is stepped up, and the current decreases proportionally because
The lower current
in the transmission wires reduces the Joule losses, which is heating of the wire due to a current flow. This heating is caused by the small, but nonzero, resistance
of the transmission wires. The power lost to the environment through this heat is
which is proportional to the current squared in the transmission wire. This is why the transmitted current
must be as small as possible and, consequently, the voltage must be large to transmit the power
Voltages ranging from 120 to 700 kV are used for transmitting power over long distances. The voltage is stepped up at the exit of the power station by a step-up transformer, as shown in .
Once the power has arrived at a population or industrial center, the voltage is stepped down at a substation to between 5 and 30 kV. Finally, at individual homes or businesses, the power is stepped down again to 120, 240, or 480 V. Each step-up and step-down transformation is done with a transformer designed based on Faradays law of induction. We’ve come a long way since Queen Elizabeth asked Faraday what possible use could be made of electricity.
### Check Your Understanding
### Section Summary
1. Electric motors contain wire loops in a magnetic field. Current is passed through the wire loops, which forces them to rotate in the magnetic field. The current is reversed every half rotation so that the torque on the loop is always in the same direction.
2. Electric generators contain wire loops in a magnetic field. An external agent provides mechanical energy to force the loops to rotate in the magnetic field, which produces an AC voltage that drives an AC current through the loops.
3. Transformers contain a ring made of magnetic material and, on opposite sides of the ring, two windings of wire wrap around the ring. A changing current in one wire winding creates a changing magnetic field, which is trapped in the ring and thus goes through the second winding and induces an emf in the second winding. The voltage in the second winding is proportional to the ratio of the number of loops in each winding.
4. Transformers are used to step up and step down the voltage for power transmission.
5. Over long distances, electric power is transmitted at high voltage to minimize the current and thereby minimize the Joule losses due to resistive heating.
### Concept Items
### Critical Thinking
### Test Prep Multiple Choice
### Test Prep Short Answer
### Performance Task
### Test Prep Extended Response
|
# Magnetism
## Electromagnetic Induction
### Section Key Terms
### Changing Magnetic Fields
In the preceding section, we learned that a current creates a magnetic field. If nature is symmetrical, then perhaps a magnetic field can create a current. In 1831, some 12 years after the discovery that an electric current generates a magnetic field, English scientist Michael Faraday (1791–1862) and American scientist Joseph Henry (1797–1878) independently demonstrated that magnetic fields can produce currents. The basic process of generating currents with magnetic fields is called induction; this process is also called magnetic induction to distinguish it from charging by induction, which uses the electrostatic Coulomb force.
When Faraday discovered what is now called Faraday’s law of induction, Queen Victoria asked him what possible use was electricity. “Madam,” he replied, “What good is a baby?” Today, currents induced by magnetic fields are essential to our technological society. The electric generator—found in everything from automobiles to bicycles to nuclear power plants—uses magnetism to generate electric current. Other devices that use magnetism to induce currents include pickup coils in electric guitars, transformers of every size, certain microphones, airport security gates, and damping mechanisms on sensitive chemical balances.
One experiment Faraday did to demonstrate magnetic induction was to move a bar magnet through a wire coil and measure the resulting electric current through the wire. A schematic of this experiment is shown in . He found that current is induced only when the magnet moves with respect to the coil. When the magnet is motionless with respect to the coil, no current is induced in the coil, as in . In addition, moving the magnet in the opposite direction (compare with ) or reversing the poles of the magnet (compare with ) results in a current in the opposite direction.
### Induced Electromotive Force
If a current is induced in the coil, Faraday reasoned that there must be what he called an electromotive force pushing the charges through the coil. This interpretation turned out to be incorrect; instead, the external source doing the work of moving the magnet adds energy to the charges in the coil. The energy added per unit charge has units of volts, so the electromotive force is actually a potential. Unfortunately, the name electromotive force stuck and with it the potential for confusing it with a real force. For this reason, we avoid the term electromotive force and just use the abbreviation emf, which has the mathematical symbol
The emf may be defined as the rate at which energy is drawn from a source per unit current flowing through a circuit. Thus, emf is the energy per unit charge added by a source, which contrasts with voltage, which is the energy per unit charge released as the charges flow through a circuit.
To understand why an emf is generated in a coil due to a moving magnet, consider , which shows a bar magnet moving downward with respect to a wire loop. Initially, seven magnetic field lines are going through the loop (see left-hand image). Because the magnet is moving away from the coil, only five magnetic field lines are going through the loop after a short time
(see right-hand image). Thus, when a change occurs in the number of magnetic field lines going through the area defined by the wire loop, an emf is induced in the wire loop. Experiments such as this show that the induced emf is proportional to the rate of change of the magnetic field. Mathematically, we express this as
where
is the change in the magnitude in the magnetic field during time
and A is the area of the loop.
Note that magnetic field lines that lie in the plane of the wire loop do not actually pass through the loop, as shown by the left-most loop in . In this figure, the arrow coming out of the loop is a vector whose magnitude is the area of the loop and whose direction is perpendicular to the plane of the loop. In , as the loop is rotated from
to
the contribution of the magnetic field lines to the emf increases. Thus, what is important in generating an emf in the wire loop is the component of the magnetic field that is perpendicular to the plane of the loop, which is
This is analogous to a sail in the wind. Think of the conducting loop as the sail and the magnetic field as the wind. To maximize the force of the wind on the sail, the sail is oriented so that its surface vector points in the same direction as the winds, as in the right-most loop in . When the sail is aligned so that its surface vector is perpendicular to the wind, as in the left-most loop in , then the wind exerts no force on the sail.
Thus, taking into account the angle of the magnetic field with respect to the area, the proportionality
becomes
Another way to reduce the number of magnetic field lines that go through the conducting loop in is not to move the magnet but to make the loop smaller. Experiments show that changing the area of a conducting loop in a stable magnetic field induces an emf in the loop. Thus, the emf produced in a conducting loop is proportional to the rate of change of the product of the perpendicular magnetic field and the loop area
where
is the perpendicular magnetic field and A is the area of the loop. The product
is very important. It is proportional to the number of magnetic field lines that pass perpendicularly through a surface of area A. Going back to our sail analogy, it would be proportional to the force of the wind on the sail. It is called the magnetic flux and is represented by
.
The unit of magnetic flux is the weber (Wb), which is magnetic field per unit area, or T/m2. The weber is also a volt second (Vs).
The induced emf is in fact proportional to the rate of change of the magnetic flux through a conducting loop.
Finally, for a coil made from N loops, the emf is N times stronger than for a single loop. Thus, the emf induced by a changing magnetic field in a coil of N loops is
The last question to answer before we can change the proportionality into an equation is “In what direction does the current flow?” The Russian scientist Heinrich Lenz (1804–1865) explained that the current flows in the direction that creates a magnetic field that tries to keep the flux constant in the loop. For example, consider again . The motion of the bar magnet causes the number of upward-pointing magnetic field lines that go through the loop to decrease. Therefore, an emf is generated in the loop that drives a current in the direction that creates more upward-pointing magnetic field lines. By using the right-hand rule, we see that this current must flow in the direction shown in the figure. To express the fact that the induced emf acts to counter the change in the magnetic flux through a wire loop, a minus sign is introduced into the proportionality
, which gives Faraday’s law of induction.
Lenz’s law is very important. To better understand it, consider , which shows a magnet moving with respect to a wire coil and the direction of the resulting current in the coil. In the top row, the north pole of the magnet approaches the coil, so the magnetic field lines from the magnet point toward the coil. Thus, the magnetic field
pointing to the right increases in the coil. According to Lenz’s law, the emf produced in the coil will drive a current in the direction that creates a magnetic field
inside the coil pointing to the left. This will counter the increase in magnetic flux pointing to the right. To see which way the current must flow, point your right thumb in the desired direction of the magnetic field
and the current will flow in the direction indicated by curling your right fingers. This is shown by the image of the right hand in the top row of . Thus, the current must flow in the direction shown in Figure 4(a).
In Figure 4(b), the direction in which the magnet moves is reversed. In the coil, the right-pointing magnetic field
due to the moving magnet decreases. Lenz’s law says that, to counter this decrease, the emf will drive a current that creates an additional right-pointing magnetic field
in the coil. Again, point your right thumb in the desired direction of the magnetic field, and the current will flow in the direction indicate by curling your right fingers (Figure 4(b)).
Finally, in Figure 4(c), the magnet is reversed so that the south pole is nearest the coil. Now the magnetic field
points toward the magnet instead of toward the coil. As the magnet approaches the coil, it causes the left-pointing magnetic field in the coil to increase. Lenz’s law tells us that the emf induced in the coil will drive a current in the direction that creates a magnetic field pointing to the right. This will counter the increasing magnetic flux pointing to the left due to the magnet. Using the right-hand rule again, as indicated in the figure, shows that the current must flow in the direction shown in Figure 4(c).
### Practice Problems
### Check Your Understanding
### Section Summary
1. Faraday’s law of induction states that a changing magnetic flux that occurs within an area enclosed by a conducting loop induces an electric current in the loop.
2. Lenz’ law states that an induced current flows in the direction such that it opposes the change that induced it.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# The Quantum Nature of Light
## Introduction
At first glance, the quantum nature of light can be a strange and bewildering concept. Between light acting as discrete chunks, massless particles providing momenta, and fundamental particles behaving like waves, it may often seem like something out of Alice in Wonderland.
For many, the study of this branch of physics can be as enthralling as Lewis Carroll’s classic novel. Recalling the works of legendary characters and brilliant scientists such as Einstein, Planck, and Compton, the study of light’s quantum nature will provide you an interesting tale of how a clever interpretation of some small details led to the most important discoveries of the past 150 years. From the electronics revolution of the twentieth century to our future progress in solar energy and space exploration, the quantum nature of light should yield a rabbit hole of curious consequence, within which lie some of the most fascinating truths of our time. |
# The Quantum Nature of Light
## Planck and Quantum Nature of Light
### Section Key Terms
### Blackbodies
Our first story of curious significance begins with a T-shirt. You are likely aware that wearing a tight black T-shirt outside on a hot day provides a significantly less comfortable experience than wearing a white shirt. Black shirts, as well as all other black objects, will absorb and re-emit a significantly greater amount of radiation from the sun. This shirt is a good approximation of what is called a blackbody.
A perfect blackbody is one that absorbs and re-emits all radiated energy that is incident upon it. Imagine wearing a tight shirt that did this! This phenomenon is often modeled with quite a different scenario. Imagine carving a small hole in an oven that can be heated to very high temperatures. As the temperature of this container gets hotter and hotter, the radiation out of this dark hole would increase as well, re-emitting all energy provided it by the increased temperature. The hole may even begin to glow in different colors as the temperature is increased. Like a burner on your stove, the hole would glow red, then orange, then blue, as the temperature is increased. In time, the hole would continue to glow but the light would be invisible to our eyes. This container is a good model of a perfect blackbody.
It is the analysis of blackbodies that led to one of the most consequential discoveries of the twentieth century. Take a moment to carefully examine . What relationships exist? What trends can you see? The more time you spend interpreting this figure, the closer you will be to understanding quantum physics!
### Understanding Blackbody Graphs
is a plot of radiation intensity against radiated wavelength. In other words, it shows how the intensity of radiated light changes when a blackbody is heated to a particular temperature.
It may help to just follow the bottom-most red line labeled 3,000 K, red hot. The graph shows that when a blackbody acquires a temperature of 3,000 K, it radiates energy across the electromagnetic spectrum. However, the energy is most intensely emitted at a wavelength of approximately 1000 nm. This is in the infrared portion of the electromagnetic spectrum. While a body at this temperature would appear red-hot to our eyes, it would truly appear ‘infrared-hot’ if we were able to see the entire spectrum.
A few other important notes regarding :
1. As temperature increases, the total amount of energy radiated increases. This is shown by examining the area underneath each line.
2. Regardless of temperature, all red lines on the graph undergo a consistent pattern. While electromagnetic radiation is emitted throughout the spectrum, the intensity of this radiation peaks at one particular wavelength.
3. As the temperature changes, the wavelength of greatest radiation intensity changes. At 4,000 K, the radiation is most intense in the yellow-green portion of the spectrum. At 6,000 K, the blackbody would radiate white hot, due to intense radiation throughout the visible portion of the electromagnetic spectrum. Remember that white light is the emission of all visible colors simultaneously.
4. As the temperature increases, the frequency of light providing the greatest intensity increases as well. Recall the equation Because the speed of light is constant, frequency and wavelength are inversely related. This is verified by the leftward movement of the three red lines as temperature is increased.
While in science it is important to categorize observations, theorizing as to why the observations exist is crucial to scientific advancement. Why doesn’t a blackbody emit radiation evenly across all wavelengths? Why does the temperature of the body change the peak wavelength that is radiated? Why does an increase in temperature cause the peak wavelength emitted to decrease? It is questions like these that drove significant research at the turn of the twentieth century. And within the context of these questions, Max Planck discovered something of tremendous importance.
### Planck’s Revolution
The prevailing theory at the time of Max Planck’s discovery was that intensity and frequency were related by the equation This equation, derived from classical physics and using wave phenomena, infers that as wavelength increases, the intensity of energy provided will decrease with an inverse-squared relationship. This relationship is graphed in and shows a troubling trend. For starters, it should be apparent that the graph from this equation does not match the blackbody graphs found experimentally. Additionally, it shows that for an object of any temperature, there should be an infinite amount of energy quickly emitted in the shortest wavelengths. When theory and experimental results clash, it is important to re-evaluate both models. The disconnect between theory and reality was termed the ultraviolet catastrophe.
Due to concerns over the ultraviolet catastrophe, Max Planck began to question whether another factor impacted the relationship between intensity and wavelength. This factor, he posited, should affect the probability that short wavelength light would be emitted. Should this factor reduce the probability of short wavelength light, it would cause the radiance curve to not progress infinitely as in the classical theory, but would instead cause the curve to precipitate back downward as is shown in the 5,000 K, 4,000 K, and 3,000 K temperature lines of the graph in . Planck noted that this factor, whatever it may be, must also be dependent on temperature, as the intensity decreases at lower and lower wavelengths as the temperature increases.
The determination of this probability factor was a groundbreaking discovery in physics, yielding insight not just into light but also into energy and matter itself. It would be the basis for Planck’s 1918 Nobel Prize in Physics and would result in the transition of physics from classical to modern understanding. In an attempt to determine the cause of the probability factor, Max Planck constructed a new theory. This theory, which created the branch of physics called quantum mechanics, speculated that the energy radiated by the blackbody could exist only in specific numerical, or quantum, states. This theory is described by the equation where n is any nonnegative integer (0, 1, 2, 3, …) and h is Planck’s constant, given by and f is frequency.
Through this equation, Planck’s probability factor can be more clearly understood. Each frequency of light provides a specific quantized amount of energy. Low frequency light, associated with longer wavelengths would provide a smaller amount of energy, while high frequency light, associated with shorter wavelengths, would provide a larger amount of energy. For specified temperatures with specific total energies, it makes sense that more low frequency light would be radiated than high frequency light. To a degree, the relationship is like pouring coins through a funnel. More of the smaller pennies would be able to pass through the funnel than the larger quarters. In other words, because the value of the coin is somewhat related to the size of the coin, the probability of a quarter passing through the funnel is reduced!
Furthermore, an increase in temperature would signify the presence of higher energy. As a result, the greater amount of total blackbody energy would allow for more of the high frequency, short wavelength, energies to be radiated. This permits the peak of the blackbody curve to drift leftward as the temperature increases, as it does from the 3,000 K to 4,000 K to 5,000 K values. Furthering our coin analogy, consider a wider funnel. This funnel would permit more quarters to pass through and allow for a reduction in concern about the probability factor.
In summary, it is the interplay between the predicted classical model and the quantum probability that creates the curve depicted in . Just as quarters have a higher currency denomination than pennies, higher frequencies come with larger amounts of energy. However, just as the probability of a quarter passing through a fixed diameter funnel is reduced, so is the probability of a high frequency light existing in a fixed temperature object. As is often the case in physics, it is the balancing of multiple incredible ideas that finally allows for better understanding.
### Quantization
It may be helpful at this point to further consider the idea of quantum states. Atoms, molecules, and fundamental electron and proton charges are all examples of physical entities that are quantized—that is, they appear only in certain discrete values and do not have every conceivable value. On the macroscopic scale, this is not a revolutionary concept. A standing wave on a string allows only particular harmonics described by integers. Going up and down a hill using discrete stair steps causes your potential energy to take on discrete values as you move from step to step. Furthermore, we cannot have a fraction of an atom, or part of an electron’s charge, or 14.33 cents. Rather, everything is built of integral multiples of these substructures.
That said, to discover quantum states within a phenomenon that science had always considered continuous would certainly be surprising. When Max Planck was able to use quantization to correctly describe the experimentally known shape of the blackbody spectrum, it was the first indication that energy was quantized on a small scale as well. This discovery earned Planck the Nobel Prize in Physics in 1918 and was such a revolutionary departure from classical physics that Planck himself was reluctant to accept his own idea. The general acceptance of Planck’s energy quantization was greatly enhanced by Einstein’s explanation of the photoelectric effect (discussed in the next section), which took energy quantization a step further.
### Practice Problems
### Check Your Understanding
### Section Summary
1. A blackbody will radiate energy across all wavelengths of the electromagnetic spectrum.
2. Radiation of a blackbody will peak at a particular wavelength, dependent on the temperature of the blackbody.
3. Analysis of blackbody radiation led to the field of quantum mechanics, which states that radiated energy can only exist in discrete quantum states.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# The Quantum Nature of Light
## Einstein and the Photoelectric Effect
### Section Key Terms
### The Photoelectric Effect
When light strikes certain materials, it can eject electrons from them. This is called the photoelectric effect, meaning that light (photo) produces electricity. One common use of the photoelectric effect is in light meters, such as those that adjust the automatic iris in various types of cameras. Another use is in solar cells, as you probably have in your calculator or have seen on a rooftop or a roadside sign. These make use of the photoelectric effect to convert light into electricity for running different devices.
### Revolutionary Properties of the Photoelectric Effect
When Max Planck theorized that energy was quantized in a blackbody radiator, it is unlikely that he would have recognized just how revolutionary his idea was. Using tools similar to the light meter in , it would take a scientist of Albert Einstein’s stature to fully discover the implications of Max Planck’s radical concept.
Through careful observations of the photoelectric effect, Albert Einstein realized that there were several characteristics that could be explained only if EM radiation is itself quantized. While these characteristics will be explained a bit later in this section, you can already begin to appreciate why Einstein’s idea is very important. It means that the apparently continuous stream of energy in an EM wave is actually not a continuous stream at all. In fact, the EM wave itself is actually composed of tiny quantum packets of energy called photons.
In equation form, Einstein found the energy of a photon or photoelectron to be
where E is the energy of a photon of frequency f and h is Planck’s constant. A beam from a flashlight, which to this point had been considered a wave, instead could now be viewed as a series of photons, each providing a specific amount of energy see . Furthermore, the amount of energy within each individual photon is based upon its individual frequency, as dictated by As a result, the total amount of energy provided by the beam could now be viewed as the sum of all frequency-dependent photon energies added together.
Just as with Planck’s blackbody radiation, Einstein’s concept of the photon could take hold in the scientific community only if it could succeed where classical physics failed. The photoelectric effect would be a key to demonstrating Einstein’s brilliance.
Consider the following five properties of the photoelectric effect. All of these properties are consistent with the idea that individual photons of EM radiation are absorbed by individual electrons in a material, with the electron gaining the photon’s energy. Some of these properties are inconsistent with the idea that EM radiation is a simple wave. For simplicity, let us consider what happens with monochromatic EM radiation in which all photons have the same energy hf.
1. If we vary the frequency of the EM radiation falling on a clean metal surface, we find the following: For a given material, there is a threshold frequency f0 for the EM radiation below which no electrons are ejected, regardless of intensity. Using the photon model, the explanation for this is clear. Individual photons interact with individual electrons. Thus if the energy of an individual photon is too low to break an electron away, no electrons will be ejected. However, if EM radiation were a simple wave, sufficient energy could be obtained simply by increasing the intensity.
2. Once EM radiation falls on a material, electrons are ejected without delay. As soon as an individual photon of sufficiently high frequency is absorbed by an individual electron, the electron is ejected. If the EM radiation were a simple wave, several minutes would be required for sufficient energy to be deposited at the metal surface in order to eject an electron.
3. The number of electrons ejected per unit time is proportional to the intensity of the EM radiation and to no other characteristic. High-intensity EM radiation consists of large numbers of photons per unit area, with all photons having the same characteristic energy, hf. The increased number of photons per unit area results in an increased number of electrons per unit area ejected.
4. If we vary the intensity of the EM radiation and measure the energy of ejected electrons, we find the following: The maximum kinetic energy of ejected electrons is independent of the intensity of the EM radiation. Instead, as noted in point 3 above, increased intensity results in more electrons of the same energy being ejected. If EM radiation were a simple wave, a higher intensity could transfer more energy, and higher-energy electrons would be ejected.
5. The kinetic energy KE of an ejected electron equals the photon energy minus the binding energy BE of the electron in the specific material. An individual photon can give all of its energy to an electron. The photon’s energy is partly used to break the electron away from the material. The remainder goes into the ejected electron’s kinetic energy. In equation form, this is given by
where is the maximum kinetic energy of the ejected electron, is the photon’s energy, and BE is the binding energy of the electron to the particular material. This equation explains the properties of the photoelectric effect quantitatively and demonstrates that BE is the minimum amount of energy necessary to eject an electron. If the energy supplied is less than BE, the electron cannot be ejected. The binding energy can also be written as where is the threshold frequency for the particular material. shows a graph of maximum versus the frequency of incident EM radiation falling on a particular material.
### Practice Problems
### Technological Applications of the Photoelectric Effect
While Einstein’s understanding of the photoelectric effect was a transformative discovery in the early 1900s, its presence is ubiquitous today. If you have watched streetlights turn on automatically in response to the setting sun, stopped elevator doors from closing simply by putting your hands between them, or turned on a water faucet by sliding your hands near it, you are familiar with the electric eye, a name given to a group of devices that use the photoelectric effect for detection.
All these devices rely on photoconductive cells. These cells are activated when light is absorbed by a semi-conductive material, knocking off a free electron. When this happens, an electron void is left behind, which attracts a nearby electron. The movement of this electron, and the resultant chain of electron movements, produces a current. If electron ejection continues, further holes are created, thereby increasing the electrical conductivity of the cell. This current can turn switches on and off and activate various familiar mechanisms.
One such mechanism takes place where you may not expect it. Next time you are at the movie theater, pay close attention to the sound coming out of the speakers. This sound is actually created using the photoelectric effect! The audiotape in the projector booth is a transparent piece of film of varying width. This film is fed between a photocell and a bright light produced by an exciter lamp. As the transparent portion of the film varies in width, the amount of light that strikes the photocell varies as well. As a result, the current in the photoconductive circuit changes with the width of the filmstrip. This changing current is converted to a changing frequency, which creates the soundtrack commonly heard in the theater.
### Check Your Understanding
### Section Summary
1. The photoelectric effect is the process in which EM radiation ejects electrons from a material.
2. Einstein proposed photons to be quanta of EM radiation having energy where f is the frequency of the radiation.
3. All EM radiation is composed of photons. As Einstein explained, all characteristics of the photoelectric effect are due to the interaction of individual photons with individual electrons.
4. The maximum kinetic energy KE of ejected electrons (photoelectrons) is given by where hf is the photon energy and BE is the binding energy (or work function) of the electron in the particular material.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# The Quantum Nature of Light
## The Dual Nature of Light
### Section Key Terms
### Photon Momentum
Do photons abide by the fundamental properties of physics? Can packets of electromagnetic energy possibly follow the same rules as a ping-pong ball or an electron? Although strange to consider, the answer to both questions is yes.
Despite the odd nature of photons, scientists prior to Einstein had long suspected that the fundamental particle of electromagnetic radiation shared properties with our more macroscopic particles. This is no clearer than when considering the photoelectric effect, where photons knock electrons out of a substance. While it is strange to think of a massless particle exhibiting momentum, it is now a well-established fact within the scientific community. shows macroscopic evidence of photon momentum.
shows a comet with two prominent tails. Comet tails are composed of gases and dust evaporated from the body of the comet and ionized gas. What most people do not know about the tails is that they always point away from the Sun rather than trailing behind the comet. This can be seen in the diagram.
Why would this be the case? The evidence indicates that the dust particles of the comet are forced away from the Sun when photons strike them. Evidently, photons carry momentum in the direction of their motion away from the Sun, and some of this momentum is transferred to dust particles in collisions. The blue tail is caused by the solar wind, a stream of plasma consisting primarily of protons and electrons evaporating from the corona of the Sun.
### Momentum, The Compton Effect, and Solar Sails
Momentum is conserved in quantum mechanics, just as it is in relativity and classical physics. Some of the earliest direct experimental evidence of this came from the scattering of X-ray photons by electrons in substances, a phenomenon discovered by American physicist Arthur H. Compton (1892–1962). Around 1923, Compton observed that X-rays reflecting from materials had decreased energy and correctly interpreted this as being due to the scattering of the X-ray photons by electrons. This phenomenon could be handled as a collision between two particles—a photon and an electron at rest in the material. After careful observation, it was found that both energy and momentum were conserved in the collision. See . For the discovery of this conserved scattering, now known as the Compton effect, Arthur Compton was awarded the Nobel Prize in 1929.
Shortly after the discovery of Compton scattering, the value of the photon momentum,
was determined by Louis de Broglie. In this equation, called the de Broglie relation, h represents Planck’s constant and λ is the photon wavelength.
We can see that photon momentum is small, since and h is very small. It is for this reason that we do not ordinarily observe photon momentum. Our mirrors do not recoil when light reflects from them, except perhaps in cartoons. Compton saw the effects of photon momentum because he was observing X-rays, which have a small wavelength and a relatively large momentum, interacting with the lightest of particles, the electron.
### Practice Problems
### Particle-Wave Duality
We have long known that EM radiation is like a wave, capable of interference and diffraction. We now see that light can also be modeled as particles—massless photons of discrete energy and momentum. We call this twofold nature the particle-wave duality, meaning that EM radiation has properties of both particles and waves. This may seem contradictory, since we ordinarily deal with large objects that never act like both waves and particles. An ocean wave, for example, looks nothing like a grain of sand. However, this so-called duality is simply a term for properties of the photon analogous to phenomena we can observe directly, on a macroscopic scale. See . If this term seems strange, it is because we do not ordinarily observe details on the quantum level directly, and our observations yield either particle-like or wave-like properties, but never both simultaneously.
Since we have a particle-wave duality for photons, and since we have seen connections between photons and matter in that both have momentum, it is reasonable to ask whether there is a particle-wave duality for matter as well. If the EM radiation we once thought to be a pure wave has particle properties, is it possible that matter has wave properties? The answer, strangely, is yes. The consequences of this are tremendous, as particle-wave duality has been a constant source of scientific wonder during the twentieth and twenty-first centuries.
### Check Your Understanding
### Section Summary
1. Compton scattering provided evidence that photon-electron interactions abide by the principles of conservation of momentum and conservation of energy.
2. The momentum of individual photons, quantified by , can be used to explain observations of comets and may lead to future space technologies.
3. Electromagnetic waves and matter have both wave-like and particle-like properties. This phenomenon is defined as particle-wave duality.
### Key Equations
### Concept Items
### Critical Thinking
### Problems
### Test Prep Multiple Choice
### Test Prep Short Answer
### Performance Task
### Test Prep Extended Response
|
# The Atom
## Introduction
From childhood on, we learn that atoms are a substructure of all things around us, from the air we breathe to the autumn leaves that blanket a forest trail. Invisible to the eye, the atoms have properties that are used to explain many phenomena—a theme found throughout this text. In this chapter, we discuss the discovery of atoms and their own substructures. We will then learn about the forces that keep them together and the tremendous energy they release when we break them apart. Finally, we will see how the knowledge and manipulation of atoms allows us to better understand geology, biology, and the world around us. |
# The Atom
## The Structure of the Atom
### Section Key Terms
How do we know that atoms are really there if we cannot see them with our own eyes? While often taken for granted, our knowledge of the existence and structure of atoms is the result of centuries of contemplation and experimentation. The earliest known speculation on the atom dates back to the fifth century B.C., when Greek philosophers Leucippus and Democritus contemplated whether a substance could be divided without limit into ever smaller pieces. Since then, scientists such as John Dalton (1766–1844), Amadeo Avogadro (1776–1856), and Dmitri Mendeleev (1834–1907) helped to discover the properties of that fundamental structure of matter. While much could be written about any number of important scientific philosophers, this section will focus on the role played by Ernest Rutherford (1871–1937). Though his understanding of our most elemental matter is rooted in the success of countless prior investigations, his surprising discovery about the interior of the atom is most fundamental in explaining so many well-known phenomena.
### Rutherford’s Experiment
In the early 1900’s, the was the accepted model of the atom. Proposed in 1904 by J. J. Thomson, the model suggested that the atom was a spherical ball of positive charge, with negatively charged electrons scattered evenly throughout. In that model, the positive charges made up the pudding, while the electrons acted as isolated plums. During its short life, the model could be used to explain why most particles were neutral, although with an unbalanced number of plums, electrically charged atoms could exist.
When Ernest Rutherford began his gold foil experiment in 1909, it is unlikely that anyone would have expected that the plum pudding model would be challenged. However, using a radioactive source, a thin sheet of gold foil, and a phosphorescent screen, Rutherford would uncover something so great that he would later call it “the most incredible event that has ever happened to me in my life”[James, L. K. (1993). Nobel Laureates in Chemistry, 1901–1992. Washington, DC: American Chemical Society.]
The experiment that Rutherford designed is shown in . As you can see in, a radioactive source was placed in a lead container with a hole in one side to produce a beam of positively charged helium particles, called alpha particles. Then, a thin gold foil sheet was placed in the beam. When the high-energy alpha particles passed through the gold foil, they were scattered. The scattering was observed from the bright spots they produced when they struck the phosphor screen.
The expectation of the plum pudding model was that the high-energy alpha particles would be scattered only slightly by the presence of the gold sheet. Because the energy of the alpha particles was much higher than those typically associated with atoms, the alpha particles should have passed through the thin foil much like a supersonic bowling ball would crash through a few dozen rows of bowling pins. Any deflection was expected to be minor, and due primarily to the electrostatic Coulomb force between the alpha particles and the foil’s interior electric charges.
However, the true result was nothing of the sort. While the majority of alpha particles passed through the foil unobstructed, Rutherford and his collaborators Hans Geiger and Ernest Marsden found that alpha particles occasionally were scattered to large angles, and some even came back in the direction from which they came! The result, called Rutherford scattering, implied that the gold nuclei were actually very small when compared with the size of the gold atom. As shown in , the dense nucleus is surrounded by mostly empty space of the atom, an idea verified by the fact that only 1 in 8,000 particles was scattered backward.
Although the results of the experiment were published by his colleagues in 1909, it took Rutherford two years to convince himself of their meaning. Rutherford later wrote: “It was almost as incredible as if you fired a 15-inch shell at a piece of tissue paper and it came back and hit you. On consideration, I realized that this scattering backwards ... [meant] ... the greatest part of the mass of the atom was concentrated in a tiny nucleus.” In 1911, Rutherford published his analysis together with a proposed model of the atom, which was in part based on Geiger’s work from the previous year. As a result of the paper, the size of the nucleus was determined to be about
m, or 100,000 times smaller than the atom. That implies a huge density, on the order of
g/cm3, much greater than any macroscopic matter.
Based on the size and mass of the nucleus revealed by his experiment, as well as the mass of electrons, Rutherford proposed the planetary model of the atom. The planetary model of the atom pictures low-mass electrons orbiting a large-mass nucleus. The sizes of the electron orbits are large compared with the size of the nucleus, and most of the atom is a vacuum. The model is analogous to how low-mass planets in our solar system orbit the large-mass Sun. In the atom, the attractive Coulomb force is analogous to gravitation in the planetary system (see ).
### Absorption and Emission Spectra
In 1900, Max Planck recognized that all energy radiated from a source is emitted by atoms in quantum states. How would that radical idea relate to the interior of an atom? The answer was first found by investigating the spectrum of light or emission spectrum produced when a gas is highly energized.
shows how to isolate the emission spectrum of one such gas. The gas is placed in the discharge tube at the left, where it is energized to the point at which it begins to radiate energy or emit light. The radiated light is channeled by a thin slit and then passed through a diffraction grating, which will separate the light into its constituent wavelengths. The separated light will then strike the photographic film on the right.
The line spectrum shown in part (b) of is the output shown on the film for excited iron. Note that this spectrum is not continuous but discrete. In other words, only particular wavelengths are emitted by the iron source. Why would that be the case?
The spectrum of light created by excited iron shows a variety of discrete wavelengths emitted within the visible spectrum. Each element, when excited to the appropriate degree, will create a discrete emission spectrum as in part (b) of . However, the wavelengths emitted will vary from element to element. The emission spectrum for iron was chosen for solely because a substantial portion of its emission spectrum is within the visible spectrum. shows the emission spectrum for hydrogen. Note that, while discrete, a large portion of hydrogen emission takes place in the ultraviolet and infrared regions.
Just as an emission spectrum shows all discrete wavelengths emitted by a gas, an absorption spectrum will show all light that is absorbed by a gas. Black lines exist where the wavelengths are absorbed, with the remainder of the spectrum lit by light is free to pass through. What relationship do you think exists between the black lines of a gas’s absorption spectrum and the colored lines of its emission spectrum? shows the absorption spectrum of the Sun. The black lines are called Fraunhofer lines, and they correspond to the wavelengths absorbed by gases in the Sun’s exterior.
### Bohr’s Explanation of the Hydrogen Spectrum
To tie the unique signatures of emission spectra to the composition of the atom itself would require clever thinking. Niels Bohr (1885–1962), a Danish physicist, did just that, by making immediate use of Rutherford’s planetary model of the atom. Bohr, shown in , became convinced of its validity and spent part of 1912 at Rutherford’s laboratory. In 1913, after returning to Copenhagen, he began publishing his theory of the simplest atom, hydrogen, based on Rutherford’s planetary model.
Bohr was able to derive the formula for the hydrogen spectrum using basic physics, the planetary model of the atom, and some very important new conjectures. His first conjecture was that only certain orbits are allowed: In other words, in an atom, the orbits of electrons are quantized. Each quantized orbit has a different distinct energy, and electrons can move to a higher orbit by absorbing energy or drop to a lower orbit by emitting energy. Because of the quantized orbits, the amount of energy emitted or absorbed must also be quantized, producing the discrete spectra seen in and . In equation form, the amount of energy absorbed or emitted can be found as
where
refers to the energy of the initial quantized orbit, and
refers to the energy of the final orbits. Furthermore, the wavelength emitted can be found using the equation
and relating the wavelength to the frequency found using the equation
, where v corresponds to the speed of light.
It makes sense that energy is involved in changing orbits. For example, a burst of energy is required for a satellite to climb to a higher orbit. What is not expected is that atomic orbits should be quantized. Quantization is not observed for satellites or planets, which can have any orbit, given the proper energy (see ).
shows an energy-level diagram, a convenient way to display energy states. Each of the horizontal lines corresponds to the energy of an electron in a different orbital. Energy is plotted vertically with the lowest or ground state at the bottom and with excited states above. The vertical arrow downwards shows energy being emitted out of the atom due to an electron dropping from one excited state to another. That would correspond to a line shown on the atom’s emission spectrum. The Lyman series shown in results from electrons dropping to the ground state, while the Balmer and Paschen series result to electrons dropping to the n = 2 and n = 3 states, respectively.
### Energy and Wavelength of Emitted Hydrogen Spectra
The energy associated with a particular orbital of a hydrogen atom can be found using the equation
where n corresponds to the orbital value from the atom’s nucleus. The negative value in the equation is based upon a baseline energy of zero when the electron is infinitely far from the atom. As a result, the negative value shows that energy is necessary to free the electron from its orbital state. The minimum energy to free the electron is also referred to as its binding energy. The equation is only valid for atoms with single electrons in their orbital shells (like hydrogen). For ionized atoms similar to hydrogen, the following formula may be used.
Please note that
corresponds to –13.6 eV, as mentioned earlier. Additionally, refers to the atomic number of the element studied. The atomic number is the number of protons in the nucleus—it is different for each element. The above equation is derived from some basic physics principles, namely conservation of energy, conservation of angular momentum, Coulomb’s law, and centripetal force. There are three derivations that result in the orbital energy equations, and they are shown below. While you can use the energy equations without understanding the derivations, they will help to remind you of just how valuable those fundamental concepts are.
### Derivation 1 (Finding the Radius of an Orbital)
One primary difference between the planetary model of the solar system and the planetary model of the atom is the cause of the circular motion. While gravitation causes the motion of orbiting planets around an interior star, the Coulomb force is responsible for the circular shape of the electron’s orbit. The magnitude of the centripetal force is
, while the magnitude of the Coulomb force is
. The assumption here is that the nucleus is more massive than the stationary electron, and the electron orbits about it. That is consistent with the planetary model of the atom. Equating the Coulomb force and the centripetal force,
which yields
### Derivation 2 (Finding the Velocity of the Orbiting Electron)
Bohr was clever enough to find a way to calculate the electron orbital energies in hydrogen. That was an important first step that has been improved upon, but it is well worth repeating here, because it does correctly describe many characteristics of hydrogen. Assuming circular orbits, Bohr proposed that the angular momentum L of an electron in its orbit is also quantized, that is, it has only specific, discrete values. The value for L is given by the formula
where L is the angular momentum, m is the electron’s mass, r is the radius of the n th orbit, and h is Planck’s constant. Note that angular momentum is
. For a small object at a radius r,
, and
, so that
Quantization says that the value of mvr can only be equal to h / 2, 2h / 2, 3h / 2, etc. At the time, Bohr himself did not know why angular momentum should be quantized, but by using that assumption, he was able to calculate the energies in the hydrogen spectrum, something no one else had done at the time.
### Derivation 3 (Finding the Energy of the Orbiting Electron)
To get the electron orbital energies, we start by noting that the electron energy is the sum of its kinetic and potential energy.
Kinetic energy is the familiar
, assuming the electron is not moving at a relativistic speed. Potential energy for the electron is electrical, or
, where V is the potential due to the nucleus, which looks like a point charge. The nucleus has a positive charge
; thus,
, recalling an earlier equation for the potential due to a point charge from the chapter on Electricity and Magnetism. Since the electron’s charge is negative, we see that
Substituting the expressions for KE and PE,
Now we solve for r and v using the equation for angular momentum
, giving
and
Substituting the expression for r and v into the above expressions for energy (KE and PE), and performing algebraic manipulation, yields
for the orbital energies of hydrogen-like atoms. Here, E is the ground-state energy (n = 1) for hydrogen (Z = 1) and is given by
Thus, for hydrogen,
The relationship between orbital energies and orbital states for the hydrogen atom can be seen in .
The wavelength of light emitted by an atom can also be determined through basic derivations. Let us consider the energy of a photon emitted from a hydrogen atom in a downward transition, given by the equation
Substituting
, we get
Dividing both sides of the equation by hc gives us an expression for
,
It can be shown that
where R is the Rydberg constant.
Simplified, the formula for determining emitted wavelength can now be written as
### Limits of Bohr’s Theory and the Quantum Model of the Atom
There are limits to Bohr’s theory. It does not account for the interaction of bound electrons, so it cannot be fully applied to multielectron atoms, even one as simple as the two-electron helium atom. Bohr’s model is what we call semiclassical. The orbits are quantized (nonclassical) but are assumed to be simple circular paths (classical). As quantum mechanics was developed, it became clear that there are no well-defined orbits; rather, there are clouds of probability. Additionally, Bohr’s theory did not explain that some spectral lines are doublets or split into two when examined closely. While we shall examine a few of those aspects of quantum mechanics in more detail, it should be kept in mind that Bohr did not fail. Rather, he made very important steps along the path to greater knowledge and laid the foundation for all of atomic physics that has since evolved.
### DeBroglie’s Waves
Following Bohr’s initial work on the hydrogen atom, a decade was to pass before Louis de Broglie proposed that matter has wave properties. The wave-like properties of matter were subsequently confirmed by observations of electron interference when scattered from crystals. Electrons can exist only in locations where they interfere constructively. How does that affect electrons in atomic orbits? When an electron is bound to an atom, its wavelength must fit into a small space, something like a standing wave on a string (see ). Orbits in which an electron can constructively interfere with itself are allowed. All orbits in which constructive interference cannot occur are not able to exist. Thus, only certain orbits are allowed. The wave nature of an electron, according to de Broglie, is why the orbits are quantized!
For a circular orbit, constructive interference occurs when the electron’s wavelength fits neatly into the circumference, so that wave crests always align with crests and wave troughs align with troughs, as shown in (b). More precisely, when an integral multiple of the electron’s wavelength equals the circumference of the orbit, constructive interference is obtained. In equation form, the condition for constructive interference and an allowed electron orbit is
where
is the electron’s wavelength and r is the radius of that circular orbit. shows the third and fourth orbitals of a hydrogen atom.
### Heisenberg Uncertainty
How does determining the location of an electron change its trajectory? The answer is fundamentally important—measurement affects the system being observed. It is impossible to measure a physical quantity exactly, and greater precision in measuring one quantity produces less precision in measuring a related quantity. It was Werner Heisenberg who first stated that limit to knowledge in 1929 as a result of his work on quantum mechanics and the wave characteristics of all particles (see ).
For example, you can measure the position of a moving electron by scattering light or other electrons from it. However, by doing so, you are giving the electron energy, and therefore imparting momentum to it. As a result, the momentum of the electron is affected and cannot be determined precisely. This change in momentum could be anywhere from close to zero up to the relative momentum of the electron (
). Note that, in this case, the particle is an electron, but the principle applies to any particle.
Viewing the electron through the model of wave-particle duality, Heisenberg recognized that, because a wave is not located at one fixed point in space, there is an uncertainty associated with any electron’s position. That uncertainty in position,
, is approximately equal to the wavelength of the particle. That is,
. There is an interesting trade-off between position and momentum. The uncertainty in an electron’s position can be reduced by using a shorter-wavelength electron, since
. But shortening the wavelength increases the uncertainty in momentum, since
. Conversely, the uncertainty in momentum can be reduced by using a longer-wavelength electron, but that increases the uncertainty in position. Mathematically, you can express the trade-off by multiplying the uncertainties. The wavelength cancels, leaving
Therefore, if one uncertainty is reduced, the other must increase so that their product is
. With the use of advanced mathematics, Heisenberg showed that the best that can be done in a simultaneous measurement of position and momentum is
That relationship is known as the Heisenberg uncertainty principle.
### The Quantum Model of the Atom
Because of the wave characteristic of matter, the idea of well-defined orbits gives way to a model in which there is a cloud of probability, consistent with Heisenberg’s uncertainty principle. shows how the principle applies to the ground state of hydrogen. If you try to follow the electron in some well-defined orbit using a probe that has a wavelength small enough to measure position accurately, you will instead knock the electron out of its orbit. Each measurement of the electron’s position will find it to be in a definite location somewhere near the nucleus. Repeated measurements reveal a cloud of probability like that in the figure, with each speck the location determined by a single measurement. There is not a well-defined, circular-orbit type of distribution. Nature again proves to be different on a small scale than on a macroscopic scale.
### Check Your Understanding
### Section Summary
1. Rutherford’s gold foil experiment provided evidence that the atom is composed of a small, dense nucleus with electrons occupying the mostly empty space around it.
2. Analysis of emission spectra shows that energy is emitted from energized gas in discrete quantities.
3. The Bohr model of the atom describes electrons existing in discrete orbits, with discrete energies emitted and absorbed as the electrons decrease and increase in orbital energy.
4. The energy emitted or absorbed by an electron as it changes energy state can be determined with the equation
, where
.
5. The wavelength of energy absorbed or emitted by an electron as it changes energy state can be determined by the equation
, where
.
6. Described as an electron cloud, the quantum model of the atom is the result of de Broglie waves and Heisenberg’s uncertainty principle.
### Key Equations
### Concept Items
### Critical Thinking
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# The Atom
## Nuclear Forces and Radioactivity
### Section Key Terms
There is an ongoing quest to find the substructures of matter. At one time, it was thought that atoms would be the ultimate substructure. However, just when the first direct evidence of atoms was obtained, it became clear that they have a substructure and a tiny nucleus. The nucleus itself has spectacular characteristics. For example, certain nuclei are unstable, and their decay emits radiations with energies millions of times greater than atomic energies. Some of the mysteries of nature, such as why the core of Earth remains molten and how the Sun produces its energy, are explained by nuclear phenomena. The exploration of radioactivity and the nucleus has revealed new fundamental particles, forces, and conservation laws. That exploration has evolved into a search for further underlying structures, such as quarks. In this section, we will explore the fundamentals of the nucleus and nuclear radioactivity.
### The Structure of the Nucleus
At this point, you are likely familiar with the neutron and proton, the two fundamental particles that make up the nucleus of an atom. Those two particles, collectively called nucleons, make up the small interior portion of the atom. Both particles have nearly the same mass, although the neutron is about two parts in 1,000 more massive. The mass of a proton is equivalent to 1,836 electrons, while the mass of a neutron is equivalent to that of 1,839 electrons. That said, each of the particles is significantly more massive than the electron.
When describing the mass of objects on the scale of nucleons and atoms, it is most reasonable to measure their mass in terms of atoms. The atomic mass unit (u) was originally defined so that a neutral carbon atom would have a mass of exactly 12 u. Given that protons and neutrons are approximately the same mass, that there are six protons and six neutrons in a carbon atom, and that the mass of an electron is minuscule in comparison, measuring this way allows for both protons and neutrons to have masses close to 1 u. shows the mass of protons, neutrons, and electrons on the new scale.
Another other useful mass unit on the atomic scale is the
. While rarely used in most contexts, it is convenient when one uses the equation
, as will be addressed later in this text.
To more completely characterize nuclei, let us also consider two other important quantities: the atomic number and the mass number. The atomic number, Z, represents the number of protons within a nucleus. That value determines the elemental quality of each atom. Every carbon atom, for instance, has a Z value of 6, whereas every oxygen atom has a Z value of 8. For clarification, only oxygen atoms may have a Z value of 8. If the Z value is not 8, the atom cannot be oxygen.
The mass number, A, represents the total number of protons and neutrons, or nucleons, within an atom. For an ordinary carbon atom the mass number would be 12, as there are typically six neutrons accompanying the six protons within the atom. In the case of carbon, the mass would be exactly 12 u. For oxygen, with a mass number of 16, the atomic mass is 15.994915 u. Of course, the difference is minor and can be ignored for most scenarios. Again, because the mass of an electron is so small compared to the nucleons, the mass number and the atomic mass can be essentially equivalent. shows an example of Lithium-7, which has an atomic number of 3 and a mass number of 7.
How does the mass number help to differentiate one atom from another? If each atom of carbon has an atomic number of 6, then what is the value of including the mass number at all? The intent of the mass number is to differentiate between various isotopes of an atom. The term isotope refers to the variation of atoms based upon the number of neutrons within their nucleus. While it is most common for there to be six neutrons accompanying the six protons within a carbon atom, it is possible to find carbon atoms with seven neutrons or eight neutrons. Those carbon atoms are respectively referred to as carbon-13 and carbon-14 atoms, with their mass numbers being their primary distinction. The isotope distinction is an important one to make, as the number of neutrons within an atom can affect a number of its properties, not the least of which is nuclear stability.
To more easily identify various atoms, their atomic number and mass number are typically written in a form of representation called the nuclide. The nuclide form appears as follows:
, where X is the atomic symbol and N represents the number of neutrons.
Let us look at a few examples of nuclides expressed in the
notation. The nucleus of the simplest atom, hydrogen, is a single proton, or
(the zero for no neutrons is often omitted). To check the symbol, refer to the periodic table—you see that the atomic number Z of hydrogen is 1. Since you are given that there are no neutrons, the mass number A is also 1. There is a scarce form of hydrogen found in nature called deuterium; its nucleus has one proton and one neutron and, hence, twice the mass of common hydrogen. The symbol for deuterium is, thus,
. An even rarer—and radioactive—form of hydrogen is called tritium, since it has a single proton and two neutrons, and it is written
. The three varieties of hydrogen have nearly identical chemistries, but the nuclei differ greatly in mass, stability, and other characteristics. Again, the different nuclei are referred to as isotopes of the same element.
There is some redundancy in the symbols A, X, Z, and N. If the element X is known, then Z can be found in a periodic table. If both A and X are known, then N can also be determined by first finding Z; then, N = A – Z. Thus the simpler notation for nuclides is
which is sufficient and is most commonly used. For example, in this simpler notation, the three isotopes of hydrogen are
,
, and
. For
, should we need to know, we
can determine that Z = 92 for uranium from the periodic table, and thus, N = 238 − 92 = 146.
### Radioactivity and Nuclear Forces
In 1896, the French physicist Antoine Henri Becquerel (1852–1908) noticed something strange. When a uranium-rich mineral called pitchblende was placed on a completely opaque envelope containing a photographic plate, it darkened spots on the photographic plate.. Becquerel reasoned that the pitchblende must emit invisible rays capable of penetrating the opaque material. Stranger still was that no light was shining on the pitchblende, which means that the pitchblende was emitting the invisible rays continuously without having any energy input! There is an apparent violation of the law of conservation of energy, one that scientists can now explain using Einstein’s famous equation
It was soon evident that Becquerel’s rays originate in the nuclei of the atoms and have other unique characteristics.
To this point, most reactions you have studied have been chemical reactions, which are reactions involving the electrons surrounding the atoms. However, two types of experimental evidence implied that Becquerel’s rays did not originate with electrons, but instead within the nucleus of an atom.
First, the radiation is found to be only associated with certain elements, such as uranium. Whether uranium was in the form of an element or compound was irrelevant to its radiation. In addition, the presence of radiation does not vary with temperature, pressure, or ionization state of the uranium atom. Since all of those factors affect electrons in an atom, the radiation cannot come from electron transitions, as atomic spectra do.
The huge energy emitted during each event is the second piece of evidence that the radiation cannot be atomic. Nuclear radiation has energies on the order of 106 eV per event, which is much greater than typical atomic energies that are a few eV, such as those observed in spectra and chemical reactions, and more than ten times as high as the most energetic X-rays.
But why would reactions within the nucleus take place? And what would cause an apparently stable structure to begin emitting energy? Was there something special about Becquerel’s uranium-rich pitchblende? To answer those questions, it is necessary to look into the structure of the nucleus. Though it is perhaps surprising, you will find that many of the same principles that we observe on a macroscopic level still apply to the nucleus.
### Nuclear Stability
A variety of experiments indicate that a nucleus behaves something like a tightly packed ball of nucleons, as illustrated in . Those nucleons have large kinetic energies and, thus, move rapidly in very close contact. Nucleons can be separated by a large force, such as in a collision with another nucleus, but strongly resist being pushed closer together. The most compelling evidence that nucleons are closely packed in a nucleus is that the radius of a nucleus, r, is found to be approximately
where
1.2 femtometer (fm) and A is the mass number of the nucleus.
Note that
. Since many nuclei are spherical, and the volume of a sphere is
, we see that
—that is, the volume of a nucleus is proportional to the number of nucleons in it. That is what you expect if you pack nucleons so close that there is no empty space between them.
So what forces hold a nucleus together? After all, the nucleus is very small and its protons, being positive, should exert tremendous repulsive forces on one another. Considering that, it seems that the nucleus would be forced apart, not together!
The answer is that a previously unknown force holds the nucleus together and makes it into a tightly packed ball of nucleons. This force is known as the strong nuclear force. The strong force has such a short range that it quickly fall to zero over a distance of only 10–15 meters. However, like glue, it is very strong when the nucleons get close to one another.
The balancing of the electromagnetic force with the nuclear forces is what allows the nucleus to maintain its spherical shape. If, for any reason, the electromagnetic force should overcome the nuclear force, components of the nucleus would be projected outward, creating the very radiation that Becquerel discovered!
Understanding why the nucleus would break apart can be partially explained using . The balance between the strong nuclear force and the electromagnetic force is a tenuous one. Recall that the attractive strong nuclear force exists between any two nucleons and acts over a very short range while the weaker repulsive electromagnetic force only acts between protons, although over a larger range. Considering the interactions, an imperfect balance between neutrons and protons can result in a nuclear reaction, with the result of regaining equilibrium.
The radiation discovered by Becquerel was due to the large number of protons present in his uranium-rich pitchblende. In short, the large number of protons caused the electromagnetic force to be greater than the strong nuclear force. To regain stability, the nucleus needed to undergo a nuclear reaction called alpha (α) decay.
### The Three Types of Radiation
Radioactivity refers to the act of emitting particles or energy from the nucleus. When the uranium nucleus emits energetic nucleons in Becquerel’s experiment, the radioactive process causes the nucleus to alter in structure. The alteration is called radioactive decay. Any substance that undergoes radioactive decay is said to be radioactive. That those terms share a root with the term radiation should not be too surprising, as they all relate to the transmission of energy.
### Alpha Decay
Alpha decay refers to the type of decay that takes place when too many protons exist in the nucleus. It is the most common type of decay and causes the nucleus to regain equilibrium between its two competing internal forces. During alpha decay, the nucleus ejects two protons and two neutrons, allowing the strong nuclear force to regain balance with the repulsive electromagnetic force. The nuclear equation for an alpha decay process can be shown as follows.
Three things to note as a result of the above equation:
1. By ejecting an alpha particle, the original nuclide decreases in atomic number. That means that Becquerel’s uranium nucleus, upon decaying, is actually transformed into thorium, two atomic numbers lower on the periodic table! The process of changing elemental composition is called transmutation.
2. Note that the two protons and two neutrons ejected from the nucleus combine to form a helium nucleus. Shortly after decay, the ejected helium ion typically acquires two electrons to become a stable helium atom.
3. Finally, it is important to see that, despite the elemental change, physical conservation still takes place. The mass number of the new element and the alpha particle together equal the mass number of the original element. Also, the net charge of all particles involved remains the same before and after the transmutation.
### Beta Decay
Like alpha decay, beta ( also takes place when there is an imbalance between neutrons and protons within the nucleus. For beta decay, however, a neutron is transformed into a proton and electron or vice versa. The transformation allows for the total mass number of the atom to remain the same, although the atomic number will increase by one (or decrease by one). Once again, the transformation of the neutron allows for a rebalancing of the strong nuclear and electromagnetic forces. The nuclear equation for a beta decay process is shown below.
The symbol
in the equation above stands for a high-energy particle called the neutrino. A nucleus may also emit a positron, and in that case Z decreases and N increases. It is beyond the scope of this section and will be discussed in further detail in the chapter on particles. It is worth noting, however, that the mass number and charge in all beta-decay reactions are conserved.
### Gamma Decay
Gamma decay is a unique form of radiation that does not involve balancing forces within the nucleus. Gamma decay occurs when a nucleus drops from an excited state to the ground state. Recall that such a change in energy state will release energy from the nucleus in the form of a photon. The energy associated with the photon emitted is so great that its wavelength is shorter than that of an X-ray. Its nuclear equation is as follows.
### Properties of Radiation
The charges of the three radiated particles differ. Alpha particles, with two protons, carry a net charge of +2. Beta particles, with one electron, carry a net charge of –1. Meanwhile, gamma rays are solely photons, or light, and carry no charge. The difference in charge plays an important role in how the three radiations affect surrounding substances.
Alpha particles, being highly charged, will quickly interact with ions in the air and electrons within metals. As a result, they have a short range and short penetrating distance in most materials. Beta particles, being slightly less charged, have a larger range and larger penetrating distance. Gamma rays, on the other hand, have little electric interaction with particles and travel much farther. Two diagrams below show the importance of difference in penetration. shows the distance of radiation penetration, and shows the influence various factors have on radiation penetration distance.
### Check Your Understanding
### Section Summary
1. The structure of the nucleus is defined by its two nucleons, the neutron and proton.
2. Atomic numbers and mass numbers are used to differentiate between various atoms and isotopes. Those numbers can be combined into an easily recognizable form called a nuclide.
3. The size and stability of the nucleus is based upon two forces: the electromagnetic force and strong nuclear force.
4. Radioactive decay is the alteration of the nucleus through the emission of particles or energy.
5. Alpha decay occurs when too many protons exist in the nucleus. It results in the ejection of an alpha particle, as described in the equation
.
6. Beta decay occurs when too many neutrons (or protons) exist in the nucleus. It results in the transmutation of a neutron into a proton, electron, and neutrino. The decay is expressed through the equation
. (Beta decay may also transform a proton into a neutron.)
7. Gamma decay occurs when a nucleus in an excited state move to a more stable state, resulting in the release of a photon. Gamma decay is represented with the equation
.
8. The penetration distance of radiation depends on its energy, charge, and type of material it encounters.
### Key Equations
### Critical Thinking
### Test Prep Short Answer
### Test Prep Extended Response
|
# The Atom
## Half Life and Radiometric Dating
### Section Key Terms
### Half-Life and the Rate of Radioactive Decay
Unstable nuclei decay. However, some nuclides decay faster than others. For example, radium and polonium, discovered by Marie and Pierre Curie, decay faster than uranium. That means they have shorter lifetimes, producing a greater rate of decay. Here we will explore half-life and activity, the quantitative terms for lifetime and rate of decay.
Why do we use the term like half-life rather than lifetime? The answer can be found by examining , which shows how the number of radioactive nuclei in a sample decreases with time. The time in which half of the original number of nuclei decay is defined as the half-life,
. After one half-life passes, half of the remaining nuclei will decay in the next half-life. Then, half of that amount in turn decays in the following half-life. Therefore, the number of radioactive nuclei decreases from N to N / 2 in one half-life, to N / 4 in the next, to N / 8 in the next, and so on. Nuclear decay is an example of a purely statistical process.
The following equation gives the quantitative relationship between the original number of nuclei present at time zero
and the number
at a later time t
where e = 2.71828... is the base of the natural logarithm, and
is the decay constant for the nuclide. The shorter the half-life, the larger is the value of
, and the faster the exponential
decreases with time. The decay constant can be found with the equation
### Activity, the Rate of Decay
What do we mean when we say a source is highly radioactive? Generally, it means the number of decays per unit time is very high. We define activity R to be the rate of decay expressed in decays per unit time. In equation form, this is
where
is the number of decays that occur in time
.
Activity can also be determined through the equation
which shows that as the amount of radiative material (N) decreases, the rate of decay decreases as well.
The SI unit for activity is one decay per second and it is given the name becquerel (Bq) in honor of the discoverer of radioactivity. That is,
Activity R is often expressed in other units, such as decays per minute or decays per year. One of the most common units for activity is the curie (Ci), defined to be the activity of 1 g of 226Ra, in honor of Marie Curie’s work with radium. The definition of the curie is
or
decays per second.
### Radiometric Dating
Radioactive dating or radiometric dating is a clever use of naturally occurring radioactivity. Its most familiar application is carbon-14 dating. Carbon-14 is an isotope of carbon that is produced when solar neutrinos strike
particles within the atmosphere. Radioactive carbon has the same chemistry as stable carbon, and so it mixes into the biosphere, where it is consumed and becomes part of every living organism. Carbon-14 has an abundance of 1.3 parts per trillion of normal carbon, so if you know the number of carbon nuclei in an object (perhaps determined by mass and Avogadro’s number), you can multiply that number by
to find the number of
nuclei within the object. Over time, carbon-14 will naturally decay back to
with a half-life of 5,730 years (note that this is an example of beta decay). When an organism dies, carbon exchange with the environment ceases, and
is not replenished. By comparing the abundance of
in an artifact, such as mummy wrappings, with the normal abundance in living tissue, it is possible to determine the artifact’s age (or time since death). Carbon-14 dating can be used for biological tissues as old as 50 or 60 thousand years, but is most accurate for younger samples, since the abundance of
nuclei in them is greater.
One of the most famous cases of carbon-14 dating involves the Shroud of Turin, a long piece of fabric purported to be the burial shroud of Jesus (see ). This relic was first displayed in Turin in 1354 and was denounced as a fraud at that time by a French bishop. Its remarkable negative imprint of an apparently crucified body resembles the then-accepted image of Jesus. As a result, the relic has been remained controversial throughout the centuries. Carbon-14 dating was not performed on the shroud until 1988, when the process had been refined to the point where only a small amount of material needed to be destroyed. Samples were tested at three independent laboratories, each being given four pieces of cloth, with only one unidentified piece from the shroud, to avoid prejudice. All three laboratories found samples of the shroud contain 92 percent of the
found in living tissues, allowing the shroud to be dated (see ).
### Section Summary
1. Radioactive half-life is the time it takes a sample of nuclei to decay to half of its original amount.
2. The rate of radioactive decay is defined as the sample’s activity, represented by the equation
.
3. Knowing the half-life of a radioactive isotope allows for the process of radioactive dating to determine the age of a material.
4. If the half-life of a material is known, the age of the material can be found using the equation
.
5. The age of organic material can be determined using the decay of the carbon-14 isotope, while the age of rocks can be determined using the decay of uranium-238.
### Key Equations
### Critical Thinking
### Test Prep Short Answer
|
# The Atom
## Nuclear Fission and Fusion
### Section Key Terms
The previous section dealt with naturally occurring nuclear decay. Without human intervention, some nuclei will change composition in order to achieve a stable equilibrium. This section delves into a less-natural process. Knowing that energy can be emitted in various forms of nuclear change, is it possible to create a nuclear reaction through our own intervention? The answer to this question is yes. Through two distinct methods, humankind has discovered multiple ways of manipulating the atom to release its internal energy.
### Nuclear Fission
In simplest terms, nuclear fission is the splitting of an atomic bond. Given that it requires great energy separate two nucleons, it may come as a surprise to learn that splitting a nucleus can release vast potential energy. And although it is true that huge amounts of energy can be released, considerable effort is needed to do so in practice.
An unstable atom will naturally decay, but it may take millions of years to do so. As a result, a physical catalyst is necessary to produce useful energy through nuclear fission. The catalyst typically occurs in the form of a free neutron, projected directly at the nucleus of a high-mass atom.
As shown in , a neutron strike can cause the nucleus to elongate, much like a drop of liquid water. This is why the model is known as the liquid drop model. As the nucleus elongates, nucleons are no longer so tightly packed, and the repulsive electromagnetic force can overcome the short-range strong nuclear force. The imbalance of forces can result in the two ends of the drop flying apart, with some of the nuclear binding energy released to the surroundings.
As you can imagine, the consequences of the nuclei splitting are substantial. When a nucleus is split, it is not only energy that is released, but a small number of neutrons as well. Those neutrons have the potential to cause further fission in other nuclei, especially if they are directed back toward the other nuclei by a dense shield or neutron reflector (see part (d) of ).
However, not every neutron produced by fission induces further fission. Some neutrons escape the fissionable material, while others interact with a nucleus without making it split. We can enhance the number of fissions produced by neutrons by having a large amount of fissionable material as well as a neutron reflector. The minimum amount necessary for self-sustained fission of a given nuclide is called its critical mass. Some nuclides, such as 239Pu, produce more neutrons per fission than others, such as 235U. Additionally, some nuclides are easier to make fission than others. In particular, 235U and 239Pu are easier to fission than the much more abundant 238U. Both factors affect critical mass, which is smallest for 239Pu. The self-sustained fission of nuclei is commonly referred to as a chain reaction, as shown in .
A chain reaction can have runaway results. If each atomic split results in two nuclei producing a new fission, the number of nuclear reactions will increase exponentially. One fission will produce two atoms, the next round of fission will create four atoms, the third round eight atoms, and so on. Of course, each time fission occurs, more energy will be emitted, further increasing the power of the atomic reaction. And that is just if two neutrons create fission reactions each round. Perhaps you can now see why so many people consider atomic energy to be an exciting energy source!
To make a self-sustained nuclear fission reactor with 235U, it is necessary to slow down the neutrons. Water is very effective at this, since neutrons collide with protons in water molecules and lose energy. shows a schematic of a reactor design called the pressurized water reactor.
Control rods containing nuclides that very strongly absorb neutrons are used to adjust neutron flux. To produce large amounts of power, reactors contain hundreds to thousands of critical masses, and the chain reaction easily becomes self-sustaining. Neutron flux must be carefully regulated to avoid an out-of-control exponential increase in the rate of fission.
Control rods help prevent overheating, perhaps even a meltdown or explosive disassembly. The water that is used to slow down neutrons, necessary to get them to induce fission in 235U, and achieve criticality, provides a negative feedback for temperature increase. In case the reactor overheats and boils the water to steam or is breached, the absence of water kills the chain reaction. Considerable heat, however, can still be generated by the reactor’s radioactive fission products. Other safety features, thus, need to be incorporated in the event of a loss of coolant accident, including auxiliary cooling water and pumps.
### Energies in Nuclear Fission
The following are two interesting facts to consider:
1. The average fission reaction produces 200 MeV of energy.
2. If you were to measure the mass of the products of a nuclear reaction, you would find that their mass was slightly less than the mass of the original nucleus.
How are those things possible? Doesn’t the fission reaction’s production of energy violate the conservation of energy? Furthermore, doesn’t the loss in mass in the reaction violate the conservation of mass? Those are important questions, and they can both be answered with one of the most famous equations in scientific history.
Recall that, according to Einstein’s theory, energy and mass are essentially the same thing. In the case of fission, the mass of the products is less than that of the reactants because the missing mass appears in the form of the energy released in the reaction, with a constant value of c2 Joules of energy converted for each kilogram of material. The value of c2 is substantial—from Einstein’s equation, the amount of energy in just 1 gram of mass would be enough to support the average U.S. citizen for more than 270 years! The example below will show you how a mass-energy transformation of this type takes place.
### Nuclear Fusion
Nuclear fusion is defined as the combining, or fusing, of two nuclei and, the combining of nuclei also results in an emission of energy. For many, the concept is counterintuitive. After all, if energy is released when a nucleus is split, how can it also be released when nucleons are combined together? The difference between fission and fusion, which results from the size of the nuclei involved, will be addressed next.
Remember that the structure of a nucleus is based on the interplay of the compressive nuclear strong force and the repulsive electromagnetic force. For nuclei that are less massive than iron, the nuclear force is actually stronger than that of the Coulomb force. As a result, when a low-mass nucleus absorbs nucleons, the added neutrons and protons bind the nucleus more tightly. The increased nuclear strong force does work on the nucleus, and energy is released.
Once the size of the created nucleus exceeds that of iron, the short-ranging nuclear force does not have the ability to bind a nucleus more tightly, and the emission of energy ceases. In fact, for fusion to occur for elements of greater mass than iron, energy must be added to the system! shows an energy-mass curve commonly used to describe nuclear reactions. Notice the location of iron (Fe) on the graph. All low-mass nuclei to the left of iron release energy through fusion, while all high-mass particles to the right of iron produce energy through fission.
The major obstruction to fusion is the Coulomb repulsion force between nuclei. Since the attractive nuclear force that can fuse nuclei together is short ranged, the repulsion of like positive charges must be overcome in order to get nuclei close enough to induce fusion. shows an approximate graph of the potential energy between two nuclei as a function of the distance between their centers. The graph resembles a hill with a well in its center. A ball rolled to the left must have enough kinetic energy to get over the hump before it falls into the deeper well with a net gain in energy. So it is with fusion. If the nuclei are given enough kinetic energy to overcome the electric potential energy due to repulsion, then they can combine, release energy, and fall into a deep well. One way to accomplish that end is to heat fusion fuel to high temperatures so that the kinetic energy of thermal motion is sufficient to get the nuclei together.
You might think that, in our Sun, nuclei are constantly coming into contact and fusing. However, this is only partially true. Only at the Sun’s core are the particles close enough and the temperature high enough for fusion to occur!
In the series of reactions below, the Sun produces energy by fusing protons, or hydrogen nuclei (
, by far the Sun’s most abundant nuclide) into helium nuclei
. The principal sequence of fusion reactions forms what is called the proton-proton cycle
where
stands for a positron and
is an electron neutrino. The energy in parentheses is released by the reaction. Note that the first two reactions must occur twice for the third to be possible, so the cycle consumes six protons (
) but gives back two. Furthermore, the two positrons produced will find two electrons and annihilate to form four more
rays, for a total of six. The overall cycle is thus
where the 26.7 MeV includes the annihilation energy of the positrons and electrons and is distributed among all the reaction products. The solar interior is dense, and the reactions occur deep in the Sun where temperatures are highest. It takes about 32,000 years for the energy to diffuse to the surface and radiate away. However, the neutrinos can carry their energy out of the Sun in less than two seconds, because they interact so weakly with other matter. Negative feedback in the Sun acts as a thermostat to regulate the overall energy output. For instance, if the interior of the Sun becomes hotter than normal, the reaction rate increases, producing energy that expands the interior. The expansion cools it and lowers the reaction rate. Conversely, if the interior becomes too cool, it contracts, increasing the temperature and therefore the reaction rate (see ). Stars like the Sun are stable for billions of years, until a significant fraction of their hydrogen has been depleted.
### Nuclear Weapons and Nuclear Power
The world was in political turmoil when fission was discovered in 1938. Compounding the troubles, the possibility of a self-sustained chain reaction was immediately recognized by leading scientists the world over. The enormous energy known to be in nuclei, but considered inaccessible, now seemed to be available on a large scale.
Within months after the announcement of the discovery of fission, Adolf Hitler banned the export of uranium from newly occupied Czechoslovakia. It seemed that the possible military value of uranium had been recognized in Nazi Germany, and that a serious effort to build a nuclear bomb had begun.
Alarmed scientists, many of whom fled Nazi Germany, decided to take action. None was more famous or revered than Einstein. It was felt that his help was needed to get the American government to make a serious effort at constructing nuclear weapons as a matter of survival. Leo Szilard, a Hungarian physicist who had emigrated to America, took a draft of a letter to Einstein, who, although a pacifist, signed the final version. The letter was for President Franklin Roosevelt, warning of the German potential to build extremely powerful bombs of a new type. It was sent in August of 1939, just before the German invasion of Poland that marked the start of World War II.
It was not until December 6, 1941, the day before the Japanese attack on Pearl Harbor, that the United States made a massive commitment to building a nuclear bomb. The top secret Manhattan Project was a crash program aimed at beating the Germans. It was carried out in remote locations, such as Los Alamos, New Mexico, whenever possible, and eventually came to cost billions of dollars and employ the efforts of more than 100,000 people. J. Robert Oppenheimer (1904–1967), a talented physicist, was chosen to head the project. The first major step was made by Enrico Fermi and his group in December 1942, when they completed the first self-sustaining nuclear reactor. This first atomic pile, built in a squash court at the University of Chicago, proved that a fission chain reaction was possible.
Plutonium was recognized as easier to fission with neutrons and, hence, a superior fission material very early in the Manhattan Project. Plutonium availability was uncertain, and so a uranium bomb was developed simultaneously. shows a gun-type bomb, which takes two subcritical uranium masses and shoots them together. To get an appreciable yield, the critical mass must be held together by the explosive charges inside the cannon barrel for a few microseconds. Since the buildup of the uranium chain reaction is relatively slow, the device to bring the critical mass together can be relatively simple. Owing to the fact that the rate of spontaneous fission is low, a neutron source is at the center the assembled critical mass.
Plutonium’s special properties necessitated a more sophisticated critical mass assembly, shown schematically in . A spherical mass of plutonium is surrounded by shaped charges (high explosives that focus their blast) that implode the plutonium, crushing it into a smaller volume to form a critical mass. The implosion technique is faster and more effective, because it compresses three-dimensionally rather than one-dimensionally as in the gun-type bomb. Again, a neutron source is included to initiate the chain reaction.
Owing to its complexity, the plutonium bomb needed to be tested before there could be any attempt to use it. On July 16, 1945, the test named Trinity was conducted in the isolated Alamogordo Desert in New Mexico, about 200 miles south of Los Alamos (see ). A new age had begun. The yield of the Trinity device was about 10 kilotons (kT), the equivalent of 5,000 of the largest conventional bombs.
Although Germany surrendered on May 7, 1945, Japan had been steadfastly refusing to surrender for many months, resulting large numbers of civilian and military casualties. Invasion plans by the Allies estimated a million casualties of their own and untold losses of Japanese lives. The bomb was viewed as a way to end the war. The first bomb used was a gun-type uranium bomb dropped on Hiroshima on August 6 by the United States. Its yield of about 15 kT destroyed the city and killed an estimated 80,000 people, with 100,000 more being seriously injured. The second bomb was an implosion-type plutonium bomb dropped on Nagasaki only three days later. Its 20-kT yield killed at least 50,000 people, something less than Hiroshima because of the hilly terrain and the fact that it was a few kilometers off target. The Japanese were told that one bomb a week would be dropped until they surrendered unconditionally, which they did on August 14. In actuality, the United States had only enough plutonium for one more bomb, as yet unassembled.
Knowing that fusion produces several times more energy per kilogram of fuel than fission, some scientists pursued the idea of constructing a fusion bomb. The first such bomb was detonated by the United States several years after the first fission bombs, on October 31, 1952, at Eniwetok Atoll in the Pacific Ocean. It had a yield of 10 megatons (MT), about 670 times that of the fission bomb that destroyed Hiroshima. The Soviet Union followed with a fusion device of its own in August 1953, and a weapons race, beyond the aim of this text to discuss, continued until the end of the Cold War.
shows a simple diagram of how a thermonuclear bomb is constructed. A fission bomb is exploded next to fusion fuel in the solid form of lithium deuteride. Before the shock wave blows it apart,
rays heat and compress the fuel, and neutrons create tritium through the reaction
. Additional fusion and fission fuels are enclosed in a dense shell of
. At the same time that the uranium shell reflects the neutrons back into the fuel to enhance its fusion, the fast-moving neutrons cause the plentiful and inexpensive
to fission, part of what allows thermonuclear bombs to be so large.
Of course, not all applications of nuclear physics are as destructive as the weapons described above. Hundreds of nuclear fission power plants around the world attest to the fact that controlled fission is both practical and economical. Given growing concerns over global warming, nuclear power is often seen as a viable alternative to energy derived from fossil fuels.
### Section Summary
1. Nuclear fission is the splitting of an atomic bond, releasing a large amount of potential energy previously holding the atom together. The amount of energy released can be determined through the equation
.
2. Nuclear fusion is the combining, or fusing together, of two nuclei. Energy is also released in nuclear fusion as the combined nuclei are closer together, resulting in a decreased strong nuclear force.
3. Fission was used in two nuclear weapons at the conclusion of World War II: the gun-type uranium bomb and the implosion-type plutonium bomb.
4. While fission has been used in both nuclear weapons and nuclear reactors, fusion is capable of releasing more energy per reaction. As a result, fusion is a well-researched, if not yet well-controlled, energy source.
### Key Equations
### Concept Items
### Critical Thinking
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# The Atom
## Medical Applications of Radioactivity: Diagnostic Imaging and Radiation
### Section Key Terms
### Medical Applications of Nuclear Physics
Applications of nuclear physics have become an integral part of modern life. From the bone scan that detects one cancer to the radioiodine treatment that cures another, nuclear radiation has diagnostic and therapeutic effects on medicine.
### Medical Imaging
A host of medical imaging techniques employ nuclear radiation. What makes nuclear radiation so useful? First,
radiation can easily penetrate tissue; hence, it is a useful probe to monitor conditions inside the body. Second, nuclear radiation depends on the nuclide and not on the chemical compound it is in, so that a radioactive nuclide can be put into a compound designed for specific purposes. When that is done, the compound is said to be tagged. A tagged compound used for medical purposes is called a radiopharmaceutical. Radiation detectors external to the body can determine the location and concentration of a radiopharmaceutical to yield medically useful information. For example, certain drugs are concentrated in inflamed regions of the body, and their locations can aid diagnosis and treatment as seen in . Another application utilizes a radiopharmaceutical that the body sends to bone cells, particularly those that are most active, to detect cancerous tumors or healing points. Images can then be produced of such bone scans. Clever use of radioisotopes determines the functioning of body organs, such as blood flow, heart muscle activity, and iodine uptake in the thyroid gland. For instance, a radioactive form of iodine can be used to monitor the thyroid, a radioactive thallium salt can be used to follow the blood stream, and radioactive gallium can be used for cancer imaging.
Once a radioactive compound has been ingested, a device like that shown in is used to monitor nuclear activity. The device, called an Anger camera or gamma camera uses a piece of lead with holes bored through it. The gamma rays are redirected through the collimator to narrow their beam, and are then interpreted using a device called a scintillator. The computer analysis of detector signals produces an image. One of the disadvantages of this detection method is that there is no depth information (i.e., it provides a two-dimensional view of the tumor as opposed to a three-dimensional view), because radiation from any location under that detector produces a signal.
Single-photon-emission computer tomography (SPECT) used in conjunction with a CT scanner improves on the process carried out by the gamma camera. shows a patient in a circular array of SPECT detectors that may be stationary or rotated, with detector output used by a computer to construct a detailed image. The spatial resolution of this technique is poor, but the three-dimensional image created results in a marked improvement in contrast.
Positron emission tomography (or PET) scans utilize images produced by
emitters. When the emitted positron β+ encounters an electron, mutual annihilation occurs, producing two γ rays. Those
rays have identical 0.511 MeV energies (the energy comes from the destruction of an electron or positron mass) and they move directly away from each other, allowing detectors to determine their point of origin accurately (as shown in ). It requires detectors on opposite sides to simultaneously (i.e., at the same time) detect photons of 0.511 MeV energy and utilizes computer imaging techniques similar to those in SPECT and CT scans. PET is used extensively for diagnosing brain disorders. It can note decreased metabolism in certain regions that accompany Alzheimer’s disease. PET can also locate regions in the brain that become active when a person carries out specific activities, such as speaking, closing his or her eyes, and so on.
### Ionizing Radiation on the Body
We hear many seemingly contradictory things about the biological effects of ionizing radiation. It can cause cancer, burns, and hair loss, and yet it is used to treat and even cure cancer. How do we understand such effects? Once again, there is an underlying simplicity in nature, even in complicated biological organisms. All the effects of ionizing radiation on biological tissue can be understood by knowing that ionizing radiation affects molecules within cells, particularly DNA molecules. Let us take a brief look at molecules within cells and how cells operate. Cells have long, double-helical DNA molecules containing chemical patterns called genetic codes that govern the function and processes undertaken by the cells. Damage to DNA consists of breaks in chemical bonds or other changes in the structural features of the DNA chain, leading to changes in the genetic code. In human cells, we can have as many as a million individual instances of damage to DNA per cell per day. The repair ability of DNA is vital for maintaining the integrity of the genetic code and for the normal functioning of the entire organism. A cell with a damaged ability to repair DNA, which could have been induced by ionizing radiation, can do one of the following:
1. The cell can go into an irreversible state of dormancy, known as senescence.
2. The cell can commit suicide, known as programmed cell death.
3. The cell can go into unregulated cell division, leading to tumors and cancers.
Since ionizing radiation damages the DNA, ionizing radiation has its greatest effect on cells that rapidly reproduce, including most types of cancer. Thus, cancer cells are more sensitive to radiation than normal cells and can be killed by it easily. Cancer is characterized by a malfunction of cell reproduction, and can also be caused by ionizing radiation. There is no contradiction to say that ionizing radiation can be both a cure and a cause.
### Radiotherapy
Radiotherapy is effective against cancer because cancer cells reproduce rapidly and, consequently, are more sensitive to radiation. The central problem in radiotherapy is to make the dose for cancer cells as high as possible while limiting the dose for normal cells. The ratio of abnormal cells killed to normal cells killed is called the therapeutic ratio, and all radiotherapy techniques are designed to enhance that ratio. Radiation can be concentrated in cancerous tissue by a number of techniques. One of the most prevalent techniques for well-defined tumors is a geometric technique shown in . A narrow beam of radiation is passed through the patient from a variety of directions with a common crossing point in the tumor. The technique concentrates the dose in the tumor while spreading it out over a large volume of normal tissue.
Another use of radiation therapy is through radiopharmaceuticals. Cleverly, radiopharmaceuticals are used in cancer therapy by tagging antibodies with radioisotopes. Those antibodies are extracted from the patient, cultured, loaded with a radioisotope, and then returned to the patient. The antibodies are then concentrated almost entirely in the tissue they developed to fight, thus localizing the radiation in abnormal tissue. This method is used with radioactive iodine to fight thyroid cancer. While the therapeutic ratio can be quite high for such short-range radiation, there can be a significant dose for organs that eliminate radiopharmaceuticals from the body, such as the liver, kidneys, and bladder. As with most radiotherapy, the technique is limited by the tolerable amount of damage to the normal tissue.
### Radiation Dosage
To quantitatively discuss the biological effects of ionizing radiation, we need a radiation dose unit that is directly related to those effects. To do define such a unit, it is important to consider both the biological organism and the radiation itself. Knowing that the amount of ionization is proportional to the amount of deposited energy, we define a radiation dose unit called the rad. It 1/100 of a joule of ionizing energy deposited per kilogram of tissue, which is
For example, if a 50.0-kg person is exposed to ionizing radiation over her entire body and she absorbs 1.00 J, then her whole-body radiation dose is
If the same 1.00 J of ionizing energy were absorbed in her 2.00-kg forearm alone, then the dose to the forearm would be
and the unaffected tissue would have a zero rad dose. When calculating radiation doses, you divide the energy absorbed by the mass of affected tissue. You must specify the affected region, such as the whole body or forearm in addition to giving the numerical dose in rads. Although the energy per kilogram in 1 rad is small, it can still have significant effects. Since only a few eV cause ionization, just 0.01 J of ionizing energy can create a huge number of ion pairs and have an effect at the cellular level.
The effects of ionizing radiation may be directly proportional to the dose in rads, but they also depend on the type of radiation and the type of tissue. That is, for a given dose in rads, the effects depend on whether the radiation is
,
,
, X-ray, or some other type of ionizing radiation. The relative biological effectiveness (RBE) relates to the amount of biological damage that can occur from a given type of radiation and is given in for several types of ionizing radiation.
A final dose unit more closely related to the effect of radiation on biological tissue is called the roentgen equivalent man, or rem. A combination of all factors mentioned previously, the roentgen equivalent man is defined to be the dose in rads multiplied by the relative biological effectiveness.
The large-scale effects of radiation on humans can be divided into two categories: immediate effects and long-term effects. gives the immediate effects of whole-body exposures received in less than one day. If the radiation exposure is spread out over more time, greater doses are needed to cause the effects listed. Any dose less than 10 rem is called a low dose, a dose 10 to 100 rem is called a moderate dose, and anything greater than 100 rem is called a high dose.
### Section Summary
1. Medical imaging occurs when a radiopharmaceutical placed in the body provides information to an array of radiation detectors outside the body.
2. Devices utilizing medical imaging include the Anger camera, SPECT detector, and PET scan.
3. Ionizing radiation can both cure and cause cancer through the manipulation of DNA molecules.
4. Radiation dosage and its effect on the body can be measured using the quantities radiation dose unit (rad), relative biological effectiveness (RBE), and the roentgen equivalent man (rem).
### Key Equations
### Critical Thinking
### Test Prep Short Answer
### Test Prep Extended Response
### Performance Task
|
# Particle Physics
## Introduction
Following ideas remarkably similar to those of the ancient Greeks, we continue to look for smaller and smaller structures in nature, hoping ultimately to find and understand the most fundamental building blocks that exist. Atomic physics deals with the smallest units of elements and compounds. In its study, we have found a relatively small number of atoms with systematic properties, and these properties have explained a tremendous range of phenomena. Nuclear physics is concerned with the nuclei of atoms and their substructures. Here, a smaller number of components—the proton and neutron—make up all nuclei. Exploring the systematic behavior of their interactions has revealed even more about matter, forces, and energy. Particle physics deals with the substructures of atoms and nuclei and is particularly aimed at finding those truly fundamental particles that have no further substructure. Just as in atomic and nuclear physics, we have found a complex array of particles and properties with systematic characteristics analogous to the periodic table and the chart of nuclides. An underlying structure is apparent, and there is some reason to think that we are finding particles that have no substructure. Of course, we have been in similar situations before. For example, atoms were once thought to be the ultimate substructures. It is possible that we could continue to find deeper and deeper structures without ever discovering the ultimate substructure—in science there is never complete certainty. See .
The properties of matter are based on substructures called molecules and atoms. Each atom has the substructure of a nucleus surrounded by electrons, and their interactions explain atomic properties. Protons and neutrons—and the interactions between them—explain the stability and abundance of elements and form the substructure of nuclei. Protons and neutrons are not fundamental—they are composed of quarks. Like electrons and a few other particles, quarks may be the fundamental building blocks of all matter, lacking any further substructure. But the story is not complete because quarks and electrons may have substructures smaller than details that are presently observable.
This chapter covers the basics of particle physics as we know it today. An amazing convergence of topics is evolving in particle physics. We find that some particles are intimately related to forces and that nature on the smallest scale may have its greatest influence on the large scale character of the universe. It is an adventure exceeding the best science fiction because it is not only fantastic but also real. |
# Particle Physics
## The Four Fundamental Forces
### Section Key Terms
Despite the apparent complexity within the universe, there remain just four basic forces. These forces are responsible for all interactions known to science: from the very small to the very large to those that we experience in our day-to-day lives. These forces describe the movement of galaxies, the chemical reactions in our laboratories, the structure within atomic nuclei, and the cause of radioactive decay. They describe the true cause behind familiar terms like friction and the normal force. These four basic forces are known as fundamental because they alone are responsible for all observations of forces in nature. The four fundamental forces are gravity, electromagnetism, weak nuclear force, and strong nuclear force.
### Understanding the Four Forces
The gravitational force is most familiar to us because it describes so many of our common observations. It explains why a dropped ball falls to the ground and why our planet orbits the Sun. It gives us the property of weight and determines much about the motion of objects in our daily lives. Because gravitational force acts between all objects of mass and has the ability to act over large distances, the gravitational force can be used to explain much of what we observe and can even describe the motion of objects on astronomical scales! That said, gravity is incredibly weak compared to the other fundamental forces and is the weakest of all of the fundamental forces. Consider this: The entire mass of Earth is needed to hold an iron nail to the ground. Yet with a simple magnet, the force of gravity can be overcome, allowing the nail to accelerate upward through space.
The electromagnetic force is responsible for both electrostatic interactions and the magnetic force seen between bar magnets. When focusing on the electrostatic relationship between two charged particles, the electromagnetic force is known as the coulomb force. The electromagnetic force is an important force in the chemical and biological sciences, as it is responsible for molecular connections like ionic bonding and hydrogen bonding. Additionally, the electromagnetic force is behind the common physics forces of friction and the normal force. Like the gravitational force, the electromagnetic force is an inverse square law. However, the electromagnetic force does not exist between any two objects of mass, only those that are charged.
When considering the structure of an atom, the electromagnetic force is somewhat apparent. After all, the electrons are held in place by an attractive force from the nucleus. But what causes the nucleus to remain intact? After all, if all protons are positive, it makes sense that the coulomb force between the protons would repel the nucleus apart immediately. Scientists theorized that another force must exist within the nucleus to keep it together. They further theorized that this nuclear force must be significantly stronger than gravity, which has been observed and measured for centuries, and also stronger than the electromagnetic force, which would cause the protons to want to accelerate away from each other.
The strong nuclear force is an attractive force that exists between all nucleons. This force, which acts equally between proton-proton connections, proton-neutron connections, and neutron-neutron connections, is the strongest of all forces at short ranges. However, at a distance of 1013 cm, or the diameter of a single proton, the force dissipates to zero. If the nucleus is large (it has many nucleons), then the distance between each nucleon could be much larger than the diameter of a single proton.
The weak nuclear force is responsible for beta decay, as seen in the equation
Recall that beta decay is when a beta particle is ejected from an atom. In order to accelerate away from the nucleus, the particle must be acted on by a force. Enrico Fermi was the first to envision this type of force. While this force is appropriately labeled, it remains stronger than the gravitational force. However, its range is even smaller than that of the strong force, as can be seen in . The weak nuclear force is more important than it may appear at this time, as will be addressed when we discuss quarks.
### Transmitting the Four Fundamental Forces
Just as it troubled Einstein prior to formulating the gravitational field theory, the concept of forces acting over a distance had greatly troubled particle physicists. That is, how does one proton know that another exists? Furthermore, what causes one proton to make a second proton repel? Or, for that matter, what is it about a proton that causes a neutron to attract? These mysterious interactions were first considered by Hideki Yukawa in 1935 and laid the foundation for much of what we now understand about particle physics.
Hideki Yukawa’s focus was on the strong nuclear force and, in particular, its incredibly short range. His idea was a blend of particles, relativity, and quantum mechanics that was applicable to all four forces. Yukawa proposed that the nuclear force is actually transmitted by the exchange of particles, called carrier particles, and that what we commonly refer to as the force’s field consists of these carrier particles. Specifically for the strong nuclear force, Yukawa proposed that a previously unknown particle, called a pion, is exchanged between nucleons, transmitting the force between them. illustrates how a pion would carry a force between a proton and a neutron.
In Yukawa’s strong force, the carrier particle is assumed to be transmitted at the speed of light and is continually transferred between the two nucleons shown. The particle that Yukawa predicted was finally discovered within cosmic rays in 1947. Its name, the pion, stands for pi meson, where meson means medium mass; it’s a medium mass because it is smaller than a nucleon but larger than an electron. Yukawa launched the field that is now called quantum chromodynamics, and the carrier particles are now called gluons due to their strong binding power. The reason for the change in the particle name will be explained when quarks are discussed later in this section.
As you may assume, the strong force is not the only force with a carrier particle. Nuclear decay from the weak force also requires a particle transfer. In the weak force are the following three: the weak negative carrier, W–; the weak positive carrier, W+; and the zero charge carrier, Z0. As we will see, Fermi inferred that these particles must carry mass, as the total mass of the products of nuclear decay is slightly larger than the total mass of all reactants after nuclear decay.
The carrier particle for the electromagnetic force is, not surprisingly, the photon. After all, just as a lightbulb can emit photons from a charged tungsten filament, the photon can be used to transfer information from one electrically charged particle to another. Finally, the graviton is the proposed carrier particle for gravity. While it has not yet been found, scientists are currently looking for evidence of its existence (see Boundless Physics: Searching for the Graviton).
So how does a carrier particle transmit a fundamental force? shows a virtual photon transmitted from one positively charged particle to another. The transmitted photon is referred to as a virtual particle because it cannot be directly observed while transmitting the force. shows a way of graphing the exchange of a virtual photon between the two positively charged particles. This graph of time versus position is called a Feynman diagram, after the brilliant American physicist Richard Feynman (1918–1988), who developed it.
The Feynman diagram should be read from the bottom up to show the movement of particles over time. In it, you can see that the left proton is propelled leftward from the photon emission, while the right proton feels an impulse to the right when the photon is received. In addition to the Feynman diagram, Richard Feynman was one of the theorists who developed the field of quantum electrodynamics (QED), which further describes electromagnetic interactions on the submicroscopic scale. For this work, he shared the 1965 Nobel Prize with Julian Schwinger and S.I. Tomonaga. A Feynman diagram explaining the strong force interaction hypothesized by Yukawa can be seen in . Here, you can see the change in particle type due to the exchange of the pi meson.
The relative masses of the listed carrier particles describe something valuable about the four fundamental forces, as can be seen in . W bosons (consisting of
and
bosons) and Z bosons (
bosons), carriers of the weak nuclear force, are nearly 1,000 times more massive than pions, carriers of the strong nuclear force. Simultaneously, the distance that the weak nuclear force can be transmitted is approximately
times the strong force transmission distance. Unlike carrier particles, which have a limited range, the photon is a massless particle that has no limit to the transmission distance of the electromagnetic force. This relationship leads scientists to understand that the yet-unfound graviton is likely massless as well.
### Accelerators Create Matter From Energy
Before looking at all the particles that make up our universe, let us first examine some of the machines that create them. The fundamental process in creating unknown particles is to accelerate known particles, such as protons or electrons, and direct a beam of them toward a target. Collisions with target nuclei provide a wealth of information, such as information obtained by Rutherford in the gold foil experiment. If the energy of the incoming particles is large enough, new matter can even be created in the collision. The more energy input or ΔE, the more matter m can be created, according to mass energy equivalence
. Limitations are placed on what can occur by known conservation laws, such as conservation of mass-energy, momentum, and charge. Even more interesting are the unknown limitations provided by nature. While some expected reactions do occur, others do not, and still other unexpected reactions may appear. New laws are revealed, and the vast majority of what we know about particle physics has come from accelerator laboratories. It is the particle physicist’s favorite indoor sport.
Our earliest model of a particle accelerator comes from the Van de Graaff generator. The relatively simple device, which you have likely seen in physics demonstrations, can be manipulated to produce potentials as great as 50 million volts. While these machines do not have energies large enough to produce new particles, analysis of their accelerated ions was instrumental in exploring several aspects of the nucleus.
Another equally famous early accelerator is the cyclotron, invented in 1930 by the American physicist, E.O. Lawrence (1901–1958). is a visual representation with more detail. Cyclotrons use fixed-frequency alternating electric fields to accelerate particles. The particles spiral outward in a magnetic field, making increasingly larger radius orbits during acceleration. This clever arrangement allows the successive addition of electric potential energy with each loop. As a result, greater particle energies are possible than in a Van de Graaff generator.
A synchrotron is a modification of the cyclotron in which particles continually travel in a fixed-radius orbit, increasing speed each time. Accelerating voltages are synchronized with the particles to accelerate them, hence the name. Additionally, magnetic field strength is increased to keep the orbital radius constant as energy increases. A ring of magnets and accelerating tubes, as shown in , are the major components of synchrotrons. High-energy particles require strong magnetic fields to steer them, so superconducting magnets are commonly employed. Still limited by achievable magnetic field strengths, synchrotrons need to be very large at very high energies since the radius of a high-energy particle’s orbit is very large.
To further probe the nucleus, physicists need accelerators of greater energy and detectors of shorter wavelength. To do so requires not only greater funding but greater ingenuity as well. Colliding beams used at both the Fermi National Accelerator Laboratory (Fermilab; see ) near Chicago and the LHC in Switzerland are designed to reduce energy loss in particle collisions. Typical stationary particle detectors lose a large amount of energy to the recoiling target struck by the accelerating particle. By providing head-on collisions between particles moving in opposite directions, colliding beams make it possible to create particles with momenta and kinetic energies near zero. This allows for particles of greater energy and mass to be created. is a schematic representation of this effect. In addition to circular accelerators, linear accelerators can be used to reduce energy radiation losses. The Stanford Linear Accelerator Center (now called the SLAC National Accelerator Laboratory) in California is home to the largest such accelerator in the world.
### Check Your Understanding
### Section Summary
1. The four fundamental forces are gravity, the electromagnetic force, the weak nuclear force, and the strong nuclear force.
2. A variety of particle accelerators have been used to explore the nature of subatomic particles and to test predictions of particle theories.
### Concept Items
### Critical Thinking
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Particle Physics
## Quarks
### Section Key Terms
### Quarks
“The first principles of the universe are atoms and empty space. Everything else is merely thought to exist…”
“… Further, the atoms are unlimited in size and number, and they are borne along with the whole universe in a vortex, and thereby generate all composite things—fire, water, air, earth. For even these are conglomerations of given atoms. And it because of their solidity that these atoms are impassive and unalterable.”
Diogenes Laertius (summarizing the views of Democritus, circa 460–370 B.C.)
The search for fundamental particles is nothing new. Atomists of the Greek and Indian empires, like Democritus of fifth century B.C., openly wondered about the most finite components of our universe. Though dormant for centuries, curiosity about the atomic nature of matter was reinvigorated by Rutherford’s gold foil experiment and the discovery of the nucleus. By the early 1930s, scientists believed they had fully determined the tiniest constituents of matter—in the form of the proton, neutron, and electron.
This would be only partially true. At present, scientists know that there are hundreds of particles not unlike our electron and nucleons, all making up what some have termed the particle zoo. While we are confident that the electron remains fundamental, it is surrounded by a plethora of similar sounding terms, like leptons, hadrons, baryons, and mesons. Even though not every particle is considered fundamental, they all play a vital role in understanding the intricate structure of our universe.
A fundamental particle is defined as a particle with no substructure and no finite size. According to the Standard Model, there are three types of fundamental particles: leptons, quarks, and carrier particles. As you may recall, carrier particles are responsible for transmitting fundamental forces between their interacting masses. Leptons are a group of six particles not bound by the strong nuclear force, of which the electron is one. As for quarks, they are the fundamental building blocks of a group of particles called hadrons, a group that includes both the proton and the neutron.
Now for a brief history of quarks. Quarks were first proposed independently by American physicists Murray Gell-Mann and George Zweig in 1963. Originally, three quark types—or flavors—were proposed with the names up (u), down (d), and strange (s).
At first, physicists expected that, with sufficient energy, we should be able to free quarks and observe them directly. However, this has not proved possible, as the current understanding is that the force holding quarks together is incredibly great and, much like a spring, increases in magnitude as the quarks are separated. As a result, when large energies are put into collisions, other particles are created—but no quarks emerge. With that in mind, there is compelling evidence for the existence of quarks. By 1967, experiments at the SLAC National Accelerator Laboratory scattering 20-GeV electrons from protons produced results like Rutherford had obtained for the nucleus nearly 60 years earlier. The SLAC scattering experiments showed unambiguously that there were three point-like (meaning they had sizes considerably smaller than the probe’s wavelength) charges inside the proton as seen in . This evidence made all but the most skeptical admit that there was validity to the quark substructure of hadrons.
The inclusion of the strange quark with Zweig and Gell-Mann’s model concerned physicists. While the up and down quarks demonstrated fairly clear symmetry and were present in common fundamental particles like protons and neutrons, the strange quark did not have a counterpart of its own. This thought, coupled with the four known leptons at the time, caused scientists to predict that a fourth quark, yet to be found, also existed.
In 1974, two groups of physicists independently discovered a particle with this new quark, labeled charmed. This completed the second exotic quark pair, strange (s) and charmed (c). A final pair of quarks was proposed when a third pair of leptons was discovered in 1975. The existence of the bottom (b) quark and the top (t) quark was verified through experimentation in 1976 and 1995, respectively. While it may seem odd that so much time would elapse between the original quark discovery in 1967 and the verification of the top quark in 1995, keep in mind that each quark discovered had a progressively larger mass. As a result, each new quark has required more energy to discover.
One of the most confounding traits of quarks is their electric charge. Long assumed to be discrete, and specifically a multiple of the elementary charge of the electron, the electric charge of an individual quark is fractional and thus seems to violate a presumed tenet of particle physics. The fractional charge of quarks, which are
and
, are the only structures found in nature with a nonintegral number of charge
. However, note that despite this odd construction, the fractional value of the quark does not violate the quantum nature of the charge. After all, free quarks cannot be found in nature, and all quarks are bound into arrangements in which an integer number of charge is constructed. shows the six known quarks, in addition to their antiquark components, as will be discussed later in this section.
While the term flavor is used to differentiate between types of quarks, the concept of color is more analogous to the electric charge in that it is primarily responsible for the force interactions between quarks. Note—Take a moment to think about the electrostatic force. It is the electric charge that causes attraction and repulsion. It is the same case here but with a color charge. The three colors available to a quark are red, green, and blue, with antiquarks having colors of anti-red (or cyan), anti-green (or magenta), and anti-blue (or yellow).
Why use colors when discussing quarks? After all, the quarks are not actually colored with visible light. The reason colors are used is because the properties of a quark are analogous to the three primary and secondary colors mentioned above. Just as different colors of light can be combined to create white, different colors of quark may be combined to construct a particle like a proton or neutron. In fact, for each hadron, the quarks must combine such that their color sums to white! Recall that two up quarks and one down quark construct a proton, as seen in . The sum of the three quarks’ colors—red, green, and blue—yields the color white. This theory of color interaction within particles is called quantum chromodynamics, or QCD. As part of QCD, the strong nuclear force can be explained using color. In fact, some scientists refer to the color force, not the strong force, as one of the four fundamental forces. is a Feynman diagram showing the interaction between two quarks by using the transmission of a colored gluon. Note that the gluon is also considered the charge carrier for the strong nuclear force.
Note that quark flavor may have any color. For instance, in , the down quark has a red color and a green color. In other words, colors are not specific to a particle quark flavor.
### Hadrons and Leptons
Particles can be revealingly grouped according to what forces they feel between them. All particles (even those that are massless) are affected by gravity since gravity affects the space and time in which particles exist. All charged particles are affected by the electromagnetic force, as are neutral particles that have an internal distribution of charge (such as the neutron with its magnetic moment). Special names are given to particles that feel the strong and weak nuclear forces. Hadrons are particles that feel the strong nuclear force, whereas leptons are particles that do not. All particles feel the weak nuclear force. This means that hadrons are distinguished by being able to feel both the strong and weak nuclear forces. Leptons and hadrons are distinguished in other ways as well. Leptons are fundamental particles that have no measurable size, while hadrons are composed of quarks and have a diameter on the order of 10–15 m. Six particles, including the electron and neutrino, make up the list of known leptons. There are hundreds of complex particles in the hadron class, a few of which (including the proton and neutron) are listed in .
There are many more leptons, mesons, and baryons yet to be discovered and measured. The purpose of trying to uncover the smallest indivisible things in existence is to explain the world around us through forces and the interactions between particles, galaxies and objects. This is why a handful of scientists devote their life’s work to smashing together small particles.
What internal structure makes a proton so different from an electron? The proton, like all hadrons, is made up of quarks. A few examples of hadron quark composition can be seen in . As shown, each hadron is constructed of multiple quarks. As mentioned previously, the fractional quark charge in all four hadrons sums to the particle’s integral value. Also, notice that the color composition for each of the four particles adds to white. Each of the particles shown is constructed of up, down, and their antiquarks. This is not surprising, as the quarks strange, charmed, top, and bottom are found in only our most exotic particles.
You may have noticed that while the proton and neutron in are composed of three quarks, both pions are comprised of only two quarks. This refers to a final delineation in particle structure. Particles with three quarks are called baryons. These are heavy particles that can decay into another baryon. Particles with only two quarks—a-quarkanti-quark pair—are called mesons. These are particles of moderate mass that cannot decay into the more massive baryons.
Before continuing, take a moment to view . In this figure, you can see the strong force reimagined as a color force. The particles interacting in this figure are the proton and neutron, just as they were in . This reenvisioning of the strong force as an interaction between colored quarks is the critical concept behind quantum chromodynamics.
### Matter and Antimatter
Antimatter was first discovered in the form of the positron, the positively charged electron. In 1932, American physicist Carl Anderson discovered the positron in cosmic ray studies. Through a cloud chamber modified to curve the trajectories of cosmic rays, Anderson noticed that the curves of some particles followed that of a negative charge, while others curved like a positive charge. However, the positive curve showed not the mass of a proton but the mass of an electron. This outcome is shown in and suggests the existence of a positively charged version of the electron, created by the destruction of solar photons.
Antimatter is considered the opposite of matter. For most antiparticles, this means that they share the same properties as their original particles with the exception of their charge. This is why the positron can be considered a positive electron while the antiproton is considered a negative proton. The idea of an opposite charge for neutral particles (like the neutron) can be confusing, but it makes sense when considered from the quark perspective. Just as the neutron is composed of one up quark and two down quarks (of charge
and
, respectively), the antineutron is composed of one antiup quark and two antidown quarks (of charge
and
, respectively). While the overall charge of the neutron remains the same, its constituent particles do not!
A word about antiparticles: Like regular particles, antiparticles could function just fine on their own. In fact, a universe made up of antimatter may operate just as our own matter-based universe does. However, we do not know fully whether this is the case. The reason for this is annihilation. Annihilation is the process of destruction that occurs when a particle and its antiparticle interact. As soon as two particles (like a positron and an electron) coincide, they convert their masses to energy through the equation
. This mass-to-energy conversion, which typically results in photon release, happens instantaneously and makes it very difficult for scientists to study antimatter. That said, scientists have had success creating antimatter through high-energy particle collisions. Both antineutrons and antiprotons were created through accelerator experiments in 1956, and an antihydrogen atom was even created at CERN in 1995! As referenced in , the annihilation of antiparticles is currently used in medical studies to determine the location of radioisotopes.
### Completing the Standard Model of the Atom
The Standard Model of the atom refers to the current scientific view of the fundamental components and interacting forces of matter. The Standard Model () shows the six quarks that bind to form all hadrons, the six lepton particles already considered fundamental, the four carrier particles (or gauge bosons) that transmit forces between the leptons and quarks, and the recently added Higgs boson (which will be discussed shortly). This totals 17 fundamental particles, combinations of which are responsible for all known matter in our entire universe! When adding the antiquarks and antileptons, 31 components make up the Standard Model.
shows all particles within the Standard Model of the atom. Not only does this chart divide all known particles by color-coded group, but it also provides information on particle stability. Note that the color-coding system in this chart is separate from the red, green, and blue color labeling system of quarks. The first three columns represent the three families of matter. The first column, considered Family 1, represents particles that make up normal matter, constructing the protons, neutrons, and electrons that make up the common world. Family 2, represented from the charm quark to the muon neutrino, is comprised of particles that are more massive. The leptons in this group are less stable and more likely to decay. Family 3, represented by the third column, are more massive still and decay more quickly. The order of these families also conveniently represents the order in which these particles were discovered.
The Standard Model also summarizes the fundamental forces that exist as particles interact. A closer look at the Standard Model, as shown in , reveals that the arrangement of carrier particles describes these interactions.
Each of the shaded areas represents a fundamental force and its constituent particles. The red shaded area shows all particles involved in the strong nuclear force, which we now know is due to quantum chromodynamics. The blue shaded area corresponds to the electromagnetic force, while the green shaded area corresponds to the weak nuclear force, which affects all quarks and leptons. The electromagnetic force and weak nuclear force are considered united by the electroweak force within the Standard Model. Also, because definitive evidence of the graviton is yet to be found, it is not included in the Standard Model.
### The Higgs Boson
One interesting feature of the Standard Model shown in is that, while the gluon and photon have no mass, the Z and W bosons are very massive. What supplies these quickly moving particles with mass and not the gluons and photons? Furthermore, what causes some quarks to have more mass than others?
In the 1960s, British physicist Peter Higgs and others speculated that the W and Z bosons were actually just as massless as the gluon and photon. However, as the W and Z bosons traveled from one particle to another, they were slowed down by the presence of a Higgs field, much like a fish swimming through water. The thinking was that the existence of the Higgs field would slow down the bosons, causing them to decrease in energy and thereby transfer this energy to mass. Under this theory, all particles pass through the Higgs field, which exists throughout the universe. The gluon and photon travel through this field as well but are able to do so unaffected.
The presence of a force from the Higgs field suggests the existence of its own carrier particle, the Higgs boson. This theorized boson interacts with all particles but gluons and photons, transferring force from the Higgs field. Particles with large mass (like the top quark) are more likely to receive force from the Higgs boson.
While it is difficult to examine a field, it is somewhat simpler to find evidence of its carrier. On July 4, 2012, two groups of scientists at the LHC independently confirmed the existence of a Higgs-like particle. By examining trillions of proton–proton collisions at energies of 7 to 8 TeV, LHC scientists were able to determine the constituent particles that created the protons. In this data, scientists found a particle with similar mass, spin, parity, and interactions with other particles that matched the Higgs boson predicted decades prior. On March 13, 2013, the existence of the Higgs boson was tentatively confirmed by CERN. Peter Higgs and Francois Englert received the Nobel Prize in 2013 for the “theoretical discovery of a mechanism that contributes to our understanding of the origin and mass of subatomic particles.”
### Check Your Understanding
### Section Summary
1. There are three types of fundamental particles—leptons, quarks, and carrier particles.
2. Quarks come in six flavors and three colors and occur only in combinations that produce white.
3. Hadrons are thought to be composed of quarks, with baryons having three quarks and mesons having a quark and an antiquark.
4. Known particles can be divided into three major groups—leptons, hadrons, and carrier particles (gauge bosons).
5. All particles of matter have an antimatter counterpart that has the opposite charge and certain other quantum numbers. These matterantimatter pairs are otherwise very similar but will annihilate when brought together.
6. The strong force is carried by eight proposed particles called gluons, which are intimately connected to a quantum number called color—their governing theory is thus called quantum chromodynamics (QCD). Taken together, QCD and the electroweak theory are widely accepted as the Standard Model of particle physics.
### Concept Items
### Critical Thinking
### Test Prep Multiple Choice
### Test Prep Short Answer
### Test Prep Extended Response
|
# Particle Physics
## The Unification of Forces
### Section Key Terms
### Understanding the Grand Unified Theory
Present quests to show that the four basic forces are different manifestations of a single unified force that follow a long tradition. In the nineteenth century, the distinct electric and magnetic forces were shown to be intimately connected and are now collectively called the electromagnetic force. More recently, the weak nuclear force was united with the electromagnetic force. As shown in , carrier particles transmit three of the four fundamental forces in very similar ways. With these considerations in mind, it is natural to suggest that a theory may be constructed in which the strong nuclear, weak nuclear, and electromagnetic forces are all unified. The search for a correct theory linking the forces, called the Grand Unified Theory (GUT), is explored in this section.
In the 1960s, the electroweak theory was developed by Steven Weinberg, Sheldon Glashow, and Abdus Salam. This theory proposed that the electromagnetic and weak nuclear forces are identical at sufficiently high energies. At lower energies, like those in our present-day universe, the two forces remain united but manifest themselves in different ways. One of the main consequences of the electroweak theory was the prediction of three short-range carrier particles, now known as the
and
bosons. Not only were three particles predicted, but the mass of each
and
boson was predicted to be 81 GeV/c2, and that of the
boson was predicted to be 90 GeV/c2. In 1983, these carrier particles were observed at CERN with the predicted characteristics, including masses having those predicted values as given in .
How can forces be unified? They are definitely distinct under most circumstances. For example, they are carried by different particles and have greatly different strengths. But experiments show that at extremely short distances and at extremely high energies, the strengths of the forces begin to become more similar, as seen in .
As discussed earlier, the short ranges and large masses of the weak carrier bosons require correspondingly high energies to create them. Thus, the energy scale on the horizontal axis of also corresponds to shorter and shorter distances (going from left to right), with 100 GeV corresponding to approximately 10−18 m, for example. At that distance, the strengths of the electromagnetic and weak nuclear forces are the same. To test this, energies of about 100 GeV are put into the system. When this occurs, the
,
, and
carrier particles are created and released. At those and higher energies, the masses of the carrier particles become less and less relevant, and the
boson in particular resembles the massless, chargeless photon. As further energy is added, the
,
, and
particles are further transformed into massless carrier particles even more similar to photons and gluons.
The extremely short distances and high energies at which the electroweak force becomes identical with the strong nuclear force are not reachable with any conceivable human-built accelerator. At energies of about 1014 GeV (16,000 J per particle), distances of about 10 to 30 m can be probed. Such energies are needed to test the theory directly, but these are about 1010 times higher than the maximum energy associated with the LHC, and the distances are about 10 to 12 smaller than any structure we have direct knowledge of. This would be the realm of various GUTs, of which there are many, since there is no constraining evidence at these energies and distances. Past experience has shown that anytime you probe so many orders of magnitude further, you find the unexpected.
While direct evidence of a GUT is not presently possible, that does not rule out the ability to assess a GUT through an indirect process. Current GUTs require various other events as a consequence of their theory. Some GUTs require the existence of magnetic monopoles, very massive individual north- and south-pole particles, which have not yet been proven to exist, while others require the use of extra dimensions. However, not all theories result in the same consequences. For example, disproving the existence of magnetic monopoles will not disprove all GUTs. Much of the science we accept in our everyday lives is based on different models, each with their own strengths and limitations. Although a particular model may have drawbacks, that does not necessarily mean that it should be discounted completely.
One consequence of GUTs that can theoretically be assessed is proton decay. Multiple current GUTs hypothesize that the stable proton should actually decay at a lifetime of 1031 years. While this time is incredibly large (keep in mind that the age of the universe is less than 14 billion years), scientists at the Super-Kamiokande in Japan have used a 50,000-ton tank of water to search for its existence. The decay of a single proton in the Super-Kamiokande tank would be observed by a detector, thereby providing support for the predicting GUT model. However, as of 2014, 17 years into the experiment, decay is yet to be found. This time span equates to a minimum limit on proton life of
years. While this result certainly does not support many grand unifying theories, an acceptable model may still exist.
### The Standard Model and the Big Bang
Nature is full of examples where the macroscopic and microscopic worlds intertwine. Newton realized that the nature of gravity on Earth that pulls an apple to the ground could explain the motion of the moon and planets so much farther away. Decays of tiny nuclei explain the hot interior of the Earth. Fusion of nuclei likewise explains the energy of stars. Today, the patterns in particle physics seem to be explaining the evolution and character of the universe. And the nature of the universe has implications for unexplored regions of particle physics.
In 1929, Edwin Hubble observed that all but the closest galaxies surrounding our own had a red shift in their hydrogen spectra that was proportional to their distance from us. Applying the Doppler Effect, Hubble recognized that this meant that all galaxies were receding from our own, with those farther away receding even faster. Knowing that our place in the universe was no more unique than any other, the implication was clear: The space within the universe itself was expanding. Just like pen marks on an expanding balloon, everything in the universe was accelerating away from everything else.
shows how the recession of galaxies looks like the remnants of a gigantic explosion, the famous Big Bang. Extrapolating backward in time, the Big Bang would have occurred between 13 and 15 billion years ago, when all matter would have been at a single point. From this, questions instantly arise. What caused the explosion? What happened before the Big Bang? Was there a before, or did time start then? For our purposes, the biggest question relating to the Big Bang is this: How does the Big Bang relate to the unification of the fundamental forces?
To fully understand the conditions of the very early universe, recognize that as the universe contracts to the size of the Big Bang, changes will occur. The density and temperature of the universe will increase dramatically. As particles become closer together, they will become too close to exist as we know them. The high energies will create other, more unusual particles to exist in greater abundance. Knowing this, let’s move forward from the start of the universe, beginning with the Big Bang, as illustrated in .
The Planck Epoch
—Though scientists are unable to model the conditions of the Planck Epoch in the laboratory, speculation is that at this time compressed energy was great enough to reach the immense
GeV necessary to unify gravity with all other forces. As a result, modern cosmology suggests that all four forces would have existed as one force, a hypothetical superforce as suggested by the Theory of Everything.
The Grand Unification Epoch
—As the universe expands, the temperatures necessary to maintain the superforce decrease. As a result, gravity separates, leaving the electroweak and strong nuclear forces together. At this time, the electromagnetic, weak, and strong forces are identical, matching the conditions requested in the Grand Unification Theory.
The Inflationary Epoch
—The separation of the strong nuclear force from the electroweak force during this time is thought to have been responsible for the massive inflation of the universe. Corresponding to the steep diagonal line on the left side of , the universe may have expanded by a factor of
or more in size. In fact, the expansion was so great during this time that it actually occurred faster than the speed of light! Unfortunately, there is little hope that we may be able to test the inflationary scenario directly since it occurs at energies near
GeV, vastly greater than the limits of modern accelerators.
The Electroweak Epoch
—Now separated from both gravity and the strong nuclear force, the electroweak force exists as a singular force during this time period. As stated earlier, scientists are able to create the energies at this stage in the universe’s expansion, needing only 100 GeV, as shown in . W and Z bosons, as well as the Higgs boson, are released during this time.
The Quark Era
—During the Quark Era, the universe has expanded and temperatures have decreased to the point at which all four fundamental forces have separated. Additionally, quarks began to take form as energies decreased.
As the universe expanded, further eras took place, allowing for the existence of hadrons, leptons, and photons, the fundamental particles of the standard model. Eventually, in nucleosynthesis, nuclei would be able to form, and the basic building blocks of atomic matter could take place. Using particle accelerators, we are very much working backwards in an attempt to understand the universe. It is encouraging to see that the macroscopic conditions of the Big Bang align nicely with our submicroscopic particle theory.
### Check Your Understanding
### Section Summary
1. Attempts to show unification of the four forces are called Grand Unified Theories (GUTs) and have been partially successful, with connections proven between EM and weak forces in electroweak theory.
2. Unification of the strong force is expected at such high energies that it cannot be directly tested, but it may have observable consequences in the as-yet-unobserved decay of the proton. Although unification of forces is generally anticipated, much remains to be done to prove its validity.
### Concept Items
### Critical Thinking
### Test Prep Multiple Choice
### Test Prep Short Answer
### Performance Task
### Test Prep Extended Response
|
# Muestreo y datos
## Introducción
Probablemente se esté preguntando: “¿Cuándo y dónde voy a utilizar la estadística?”. Si lee cualquier periódico, ve la televisión o utiliza internet, verá información estadística. Hay estadísticas sobre delincuencia, deportes, educación, política y bienes raíces. Normalmente, cuando se lee un artículo de periódico o se ve un programa de noticias de televisión se da una información de muestra. Con esta información, puede tomar una decisión sobre la corrección de una declaración, afirmación o “hecho”. Los métodos estadísticos pueden ayudarlo a hacer una “mejor estimación”.
Como sin duda recibirá información estadística en algún momento de su vida, necesita conocer algunas técnicas para analizar la información de forma reflexiva. Piense en la compra de una casa o en la gestión de un presupuesto. Piense en la profesión que ha elegido. Economía, Negocios, Psicología, Educación, Biología, Derecho, Informática, Política y Desarrollo de la Primera Infancia son campos de conocimiento que requieren, al menos, un curso de Estadística.
En este capítulo se incluyen las ideas y palabras básicas de probabilidad y estadística. Pronto entenderá que la estadística y la probabilidad trabajan juntas. También aprenderá cómo se recopilan los datos y qué datos “buenos” pueden distinguirse de los “malos”. |
# Muestreo y datos
## Definiciones de estadística, probabilidad y términos clave
La ciencia de la Estadística se ocupa de la recopilación, del análisis, de la interpretación y de la presentación de datos. Vemos y utilizamos datos en nuestra vida cotidiana.
En este curso aprenderá a organizar y resumir datos. La organización y el resumen de los datos se denominan Estadística Descriptiva. Dos formas de resumir los datos son la elaboración de gráficos y el uso de números (por ejemplo, hallar un promedio). Después de haber estudiado la probabilidad y las distribuciones de probabilidad, utilizará métodos formales para sacar conclusiones de los datos “buenos”. Los métodos formales se denominan Estadística Inferencial. La inferencia estadística utiliza la probabilidad para determinar el grado de confianza que podemos tener en que nuestras conclusiones son correctas.
La interpretación eficaz de los datos (inferencia) se basa en buenos procedimientos de producción de datos y en examinarlos de forma reflexiva. Se encontrará con lo que le parecerá un exceso de fórmulas matemáticas para interpretar los datos. La meta de la Estadística no es realizar numerosos cálculos con las fórmulas, sino comprender los datos. Los cálculos se pueden hacer con una calculadora o una computadora. La comprensión debe venir de usted. Si puede comprender a fondo los fundamentos de la Estadística, podrá tener más confianza en las decisiones que tome en la vida.
### Probabilidad
La probabilidad es una herramienta matemática utilizada para estudiar el azar. Se trata de la oportunidad (la posibilidad) de que se produzca un evento. Por ejemplo, si se lanza una moneda imparcial cuatro veces, los resultados no pueden ser dos caras y dos cruces. Sin embargo, si se lanza la misma moneda 4.000 veces, los resultados se aproximarán a mitad cara y mitad cruz. La probabilidad teórica esperada de salir cara en cualquier lanzamiento es
o 0,5. Aunque los resultados de unas pocas repeticiones son inciertos, existe un patrón regular de resultados cuando hay muchas repeticiones. Tras leer sobre el estadístico inglés Karl Pearson, que lanzó una moneda 24.000 veces con un resultado de 12.012 caras, uno de los autores lanzó una moneda 2.000 veces. Los resultados fueron 996 caras. La fracción
es igual a 0,498, que está muy cerca de 0,5, la probabilidad esperada.
La teoría de la probabilidad comenzó con el estudio de los juegos de azar, como el póquer. Las predicciones adoptan la forma de probabilidades. Para predecir la probabilidad de que se produzca un terremoto, de que llueva o de que obtenga una A en este curso utilizamos las probabilidades. Los médicos utilizan la probabilidad para determinar la posibilidad de que una vacuna provoque la enfermedad que se supone que debe prevenir. Un agente de bolsa utiliza la probabilidad para determinar la tasa de rendimiento de las inversiones de un cliente. Puede utilizar la probabilidad para decidir si compra un billete de lotería o no. En su estudio de la Estadística, utilizará el poder de las Matemáticas a través de cálculos de probabilidad para analizar e interpretar sus datos.
### Términos clave
En estadística, generalmente queremos estudiar una población. Se puede pensar en una población como un conjunto de personas, cosas u objetos en estudio. Para estudiar la población seleccionamos una muestra. La idea del muestreo es seleccionar una porción (o subconjunto) de la población mayor y estudiar esa porción (la muestra) para obtener información sobre la población. Los datos son el resultado de un muestreo de una población.
Como se necesita mucho tiempo y dinero para examinar toda una población, el muestreo es una técnica muy práctica. Si desea calcular el promedio general de calificaciones de su escuela, tendría sentido seleccionar una muestra de estudiantes que asisten a la escuela. Los datos recopilados de la muestra serían los promedios de las calificaciones de los estudiantes. En las elecciones presidenciales se toman muestras de sondeos de opinión de 1.000 a 2.000 personas. Se supone que el sondeo de opinión representa el punto de vista de las personas de todo el país. Los fabricantes de bebidas carbonatadas en lata toman muestras para determinar si una lata de 16 onzas contiene 16 onzas de bebida carbonatada.
A partir de los datos de la muestra podemos calcular un estadístico. Un estadístico es un número que representa una propiedad de la muestra. Por ejemplo, si consideramos que una clase de Matemáticas es una muestra de la población de todas las clases de Matemáticas, el número promedio de puntos obtenidos por los estudiantes de esa clase de Matemáticas al final del trimestre es un ejemplo de un estadístico. La estadística es una estimación de un parámetro poblacional, en este caso la media. Un parámetro es una característica numérica de toda la población que puede estimarse mediante un estadístico. Dado que consideramos que todas las clases de Matemáticas son la población, el número promedio de puntos obtenidos por estudiante en todas las clases de Matemáticas es un ejemplo de parámetro.
Una de las principales preocupaciones en el campo de la Estadística es la precisión con la que un estadístico estima un parámetro. La precisión depende realmente de lo bien que la muestra represente a la población. La muestra debe contener las características de la población para ser una muestra representativa. En la Estadística Inferencial nos interesa tanto el estadístico de la muestra como el parámetro de la población. En un capítulo posterior utilizaremos el estadístico de la muestra para comprobar la validez del parámetro poblacional establecido.
Una variable, o variable aleatoria, que normalmente se anota con letras mayúsculas como la X y la Y, es una característica o medida que puede determinarse para cada miembro de una población. Las variables pueden ser numéricas o categóricas. Las variables numéricas toman valores con unidades iguales, como el peso en libras y el tiempo en horas. Las variables categóricas sitúan a la persona o cosa en una categoría. Si suponemos que X equivale al número de puntos obtenidos por un estudiante de Matemáticas al final de un trimestre, entonces X es una variable numérica. Si suponemos que Y es la afiliación de una persona a un partido, entonces algunos ejemplos de Y incluyen republicano, demócrata e independiente. Y es una variable categórica. Podríamos hacer algunos cálculos con valores de X (calcular el promedio de puntos obtenidos, por ejemplo), pero no tiene sentido hacer cálculos con valores de Y (calcular un promedio de afiliación a un partido no tiene sentido).
Los datos son los valores reales de la variable. Pueden ser números o palabras. El dato es un valor único.
Dos palabras que aparecen a menudo en estadística son media y proporción. Si presenta tres exámenes de sus clases de Matemáticas y obtiene calificaciones de 86, 75 y 92, calcularía su calificación media sumando las tres calificaciones de los exámenes y dividiéndolas entre tres (su calificación media sería 84,3 con un decimal). Si en su clase de Matemáticas hay 40 estudiantes y 22 son hombres y 18 son mujeres, entonces la proporción de estudiantes hombres es y la proporción de estudiantes mujeres es . La media y la proporción se tratan con más detalle en capítulos posteriores.
### Referencias
The Data and Story Library, http://lib.stat.cmu.edu/DASL/Stories/CrashTestDummies.html (consultado el 1.º de mayo de 2013).
### Repaso del capítulo
La teoría matemática de la estadística es más fácil de aprender cuando se conoce el lenguaje. Este módulo presenta términos importantes que se utilizarán a lo largo del texto.
### TAREA PARA LA CASA
Para cada uno de los ocho ejercicios siguientes, identifique: a. la población, b. la muestra, c. el parámetro, d. el estadístico, e. la variable y f. los datos. Dé ejemplos cuando sea necesario.
Use la siguiente información para responder los tres próximos ejercicios: Una instructora del Lake Tahoe Community College está interesado en el número medio de días que los estudiantes de Matemáticas del Lake Tahoe Community College se ausentan de clase durante un trimestre. |
# Muestreo y datos
## Datos, muestreo y variación de datos y muestreo
Los datos pueden proceder de una población o de una muestra. Letras minúsculas como
o
se utilizan generalmente para representar valores de datos. La mayoría de los datos se pueden clasificar en las siguientes categorías:
1. Cualitativa
2. Cuantitativa
Los datos cualitativos son el resultado de categorizar o describir los atributos de una población. Los datos cualitativos también suelen denominarse datos categóricos. El color del cabello, el tipo de sangre, la etnia, el automóvil que conduce una persona y la calle en la que vive son ejemplos de datos cualitativos (categóricos). Los datos cualitativos (categóricos) se describen con palabras o letras. Por ejemplo, el color del cabello puede ser negro, castaño oscuro, castaño claro, rubio, gris o rojo. El tipo de sangre puede ser AB+, O– o B+. Los investigadores prefieren los datos cuantitativos a los cualitativos (categóricos) porque se prestan más al análisis matemático. Por ejemplo, no tiene sentido hallar un color de cabello o un tipo de sangre promedio.
Los datos cuantitativos son siempre números. Los datos cuantitativos son el resultado de contar o medir los atributos de una población. La cantidad de dinero, la frecuencia del pulso, el peso, el número de personas que viven en su ciudad y el número de estudiantes que cursan Estadística son ejemplos de datos cuantitativos. Los datos cuantitativos pueden ser discretos o continuos.
Todos los datos que son el resultado de contar se denominan datos discretos cuantitativos. Estos datos solo adoptan ciertos valores numéricos. Si cuenta el número de llamadas telefónicas que recibe cada día de la semana, puede obtener valores como cero, uno, dos o tres.
Los datos que no solo se componen de números para contar, sino que pueden incluir fracciones, decimales o números irracionales, se denominan datos cuantitativos continuos. Los datos continuos suelen ser el resultado de mediciones como longitudes, pesos o tiempos. Una lista de la duración en minutos de todas las llamadas telefónicas que realiza en una semana, con números como 2,4; 7,5; u 11,0, sería un dato cuantitativo continuo.
### Discusión de datos cualitativos
A continuación se muestran tablas que comparan el número de estudiantes a tiempo parcial y a tiempo completo en De Anza College y Foothill College inscritos para el trimestre de primavera de 2010. Las tablas muestran recuentos (frecuencias) y porcentajes o proporciones (frecuencias relativas). Las columnas de porcentajes facilitan la comparación de las mismas categorías en los institutos universitarios. Suele ser útil mostrar porcentajes junto con números, pero es especialmente importante cuando se comparan conjuntos de datos que no tienen los mismos totales, como las inscripciones totales de ambos institutos universitarios en este ejemplo. Observe que el porcentaje de estudiantes a tiempo parcial del Foothill College es mucho mayor que el del De Anza College.
Las tablas son una buena forma de organizar y mostrar datos. Pero los gráficos pueden ser aun más útiles para entender los datos. No hay reglas estrictas en cuanto a los gráficos que hay que utilizar. Dos gráficos que se utilizan para mostrar datos cualitativos (categóricos) son los gráficos circulares y los de barras.
En un gráfico circular las categorías de datos se representan mediante cuñas en un círculo y su tamaño es proporcional al porcentaje de personas de cada categoría.
En un gráfico de barras la longitud de la barra para cada categoría es proporcional al número o porcentaje de personas en cada categoría. Las barras pueden ser verticales u horizontales.
Un diagrama de Pareto está formado por barras que se ordenan por el tamaño de la categoría (de mayor a menor).
Observe la y la y determine qué gráfico (circular o de barras) cree que muestra mejor las comparaciones.
Es una buena idea observar una variedad de gráficos para ver cuál es el más útil para mostrar los datos. Según los datos y el contexto, podemos elegir el “mejor” gráfico. Nuestra elección también depende del uso que hagamos de los datos.
### Porcentajes que suman más (o menos) que el 100 %
A veces, los porcentajes suman más del 100 % (o menos del 100 %). En el gráfico, los porcentajes suman más del 100 % porque los estudiantes pueden estar en más de una categoría. Un gráfico de barras es apropiado para comparar el tamaño relativo de las categorías. No se puede utilizar un gráfico circular. Tampoco podía utilizarse si los porcentajes sumaban menos del 100 %.
### Omisión de categorías/falta de datos
La tabla muestra el origen étnico de los estudiantes pero falta la categoría “otros/desconocidos”. En esta categoría se ubican las personas que no se consideraron incluidas en ninguna de las categorías étnicas o que se negaron a responder. Observe que las frecuencias no suman el número total de estudiantes. En esta situación, cree un gráfico de barras y no un gráfico circular.
El siguiente gráfico es igual que el anterior, pero se ha incluido el porcentaje de “otros/desconocidos” (9,6 %). La categoría “otros/desconocidos” es grande en comparación con algunas de las otras categorías (nativos de Estados Unidos, 0,6 %, isleños del Pacífico, 1,0 %). Es importante saber esto cuando pensamos en lo que nos dicen los datos.
Este gráfico de barras particular en la puede ser difícil de entender visualmente. El gráfico de la es un diagrama de Pareto. El diagrama de Pareto tiene las barras ordenadas de mayor a menor y es más fácil de leer e interpretar.
### Gráficos circulares: no faltan datos
Los siguientes gráficos circulares incluyen la categoría “otros/desconocidos” (ya que los porcentajes deben sumar el 100 %). El gráfico en la (b) está organizado por el tamaño de cada porción, lo que lo convierte en un gráfico visualmente más informativo que el gráfico sin clasificar en la (a).
### Muestreo
Recopilar información sobre toda una población suele ser demasiado costoso o prácticamente imposible. En cambio, utilizamos una muestra de la población. Una muestra debe tener las mismas características que la población que representa. La mayoría de los estadísticos utilizan varios métodos de muestreo aleatorio para intentar alcanzar esta meta. En esta sección se describen algunos de los métodos más comunes. Existen varios métodos de muestreo aleatorio. En cada forma de muestreo aleatorio, cada miembro de una población tiene inicialmente la misma probabilidad de que lo seleccionen para la muestra. Cada método tiene sus pros y sus contras. El método más fácil de describir se llama muestra aleatoria simple. Cualquier grupo de n personas tiene la misma probabilidad de que lo seleccionen que cualquier otro grupo de n personas si se utiliza la técnica de muestreo aleatorio simple. En otras palabras, cada muestra del mismo tamaño tiene la misma probabilidad de que la seleccionen.
Además del muestreo aleatorio simple, existen otras formas de muestreo que implican un proceso de azar para obtener la muestra. Otros métodos de muestreo aleatorio bien conocidos son la muestra estratificada, la muestra por conglomerados y la muestra sistemática.
Para seleccionar una muestra estratificada, hay que dividir la población en grupos llamados estratos y, a continuación, tomar un número proporcional de cada estrato. Por ejemplo, podría estratificar (agrupar) la población de su instituto universitario por departamentos y luego seleccionar una muestra aleatoria simple proporcional de cada estrato (cada departamento) para obtener una muestra aleatoria estratificada. Para seleccionar una muestra aleatoria simple de cada departamento, numere cada miembro del primer departamento, numere cada miembro del segundo departamento y haga lo mismo con los departamentos restantes. Luego, utilice un muestreo aleatorio simple para seleccionar números proporcionales del primer departamento y haga lo mismo con cada uno de los departamentos restantes. Esos números seleccionados del primer departamento y del segundo departamento, y así sucesivamente, representan los miembros que componen la muestra estratificada.
Para seleccionar una muestra por conglomerados hay que dividir la población en conglomerados (grupos) y luego seleccionar al azar algunos de los conglomerados. Todos los miembros de estos grupos están en la muestra por conglomerados. Por ejemplo, si toma una muestra aleatoria de cuatro departamentos de la población de su instituto universitario, los cuatro departamentos constituyen la muestra por conglomerados. Divida el profesorado de su instituto universitario por departamento. Los departamentos son los conglomerados. Numere cada departamento y, a continuación, elija cuatro números diferentes mediante un muestreo aleatorio simple. Todos los miembros de los cuatro departamentos con esos números son la muestra de conglomerado.
Para seleccionar una muestra sistemática, seleccione al azar un punto de partida y tome cada n.ª (enésima) pieza de datos de una lista de la población. Por ejemplo, supongamos que tiene que hacer una encuesta telefónica. Su directorio telefónico contiene 20.000 listas de residencias. Debe seleccionar 400 nombres para la muestra. Numere la población de 1 a 20.000 y luego utilice una muestra aleatoria simple para seleccionar un número que represente el primer nombre de la muestra. Luego, elija cada quincuagésimo nombre hasta que tenga un total de 400 nombres (puede que tenga que volver al principio de su lista de teléfonos). El muestreo sistemático se elige con frecuencia porque es un método sencillo.
Un tipo de muestreo que no es aleatorio es el muestreo de conveniencia. El muestreo de conveniencia implica el uso de resultados que están fácilmente disponibles. Por ejemplo, una tienda de softwares realiza un estudio de mercadeo mediante entrevistas con los clientes potenciales que se encuentran en la tienda mirando softwares disponibles. Los resultados del muestreo de conveniencia pueden ser muy buenos en algunos casos y muy sesgados (favorecer ciertos resultados) en otros.
El muestreo de datos debe hacerse con mucho cuidado. Recolectar datos sin cuidado puede causar resultados devastadores. Las encuestas enviadas por correo a los hogares y luego devueltas pueden estar muy sesgadas (pueden favorecer a un determinado grupo). Es mejor que la persona que realiza la encuesta seleccione la muestra de encuestados.
El muestreo aleatorio verdadero se realiza con reemplazo. Es decir, una vez que se selecciona un miembro, ese miembro vuelve a la población y, por tanto, lo pueden escoger más de una vez. Sin embargo, por razones prácticas, en la mayoría de las poblaciones el muestreo aleatorio simple se realiza sin reemplazo. Las encuestas suelen hacerse sin reemplazo. Es decir, un miembro de la población solo lo pueden seleccionar una vez. La mayoría de las muestras se toman de poblaciones grandes y la muestra tiende a ser pequeña en comparación con la población. En este caso, el muestreo sin reemplazo es, aproximadamente, igual al muestreo con reemplazo, ya que la probabilidad de seleccionar a la misma persona más de una vez con reemplazo es muy baja.
En una población universitaria de 10.000 personas, supongamos que se quiere seleccionar una muestra de 1.000 al azar para una encuesta. Para cualquier muestra particular de 1.000, si se hace un muestreo con reemplazo,
1. la probabilidad de seleccionar la primera persona es de 1.000 entre 10.000 (0,1000);
2. la probabilidad de seleccionar una segunda persona diferente para esta muestra es de 999 entre 10.000 (0,0999);
3. la probabilidad de volver a seleccionar a la misma persona es de 1 entre 10.000 (muy baja).
Si se trata de un muestreo sin reemplazo,
1. la probabilidad de seleccionar la primera persona para cualquier muestra específica es de 1.000 entre 10.000 (0,1000);
2. la probabilidad de seleccionar una segunda persona diferente es de 999 entre 9.999 (0,0999);
3. no se sustituye la primera persona antes de seleccionar la siguiente.
Compare las fracciones 999/10.000 y 999/9.999. Para lograr más exactitud, lleve las respuestas decimales a cuatro cifras. Con cuatro decimales, estos números son equivalentes (0,0999).
El muestreo sin reemplazo en vez del muestreo con reemplazo se convierte en una cuestión matemática solo cuando la población es pequeña. Por ejemplo, si la población es de 25 personas, la muestra es de diez y se realiza un muestreo con reemplazo para cualquier muestra particular, entonces la probabilidad de seleccionar la primera persona es de diez entre 25, y la probabilidad de seleccionar una segunda persona diferente es de nueve entre 25 (se reemplaza la primera persona).
Si se hace una muestra sin reemplazo, la probabilidad de seleccionar la primera persona es de diez entre 25, y la probabilidad de seleccionar la segunda persona (que es diferente) es de nueve entre 24 (no se reemplaza la primera persona).
Compare las fracciones 9/25 y 9/24. Con cuatro decimales, 9/25 = 0,3600 y 9/24 = 0,3750. Con cuatro decimales, estos números no son equivalentes.
Al analizar los datos, es importante tener en cuenta los errores de muestreo y los errores ajenos al muestreo. El propio proceso de muestreo provoca errores de muestreo. Por ejemplo, la muestra puede no ser lo suficientemente grande. Los factores no relacionados con el proceso de muestreo provocan errores ajenos al muestreo. Un dispositivo de recuento defectuoso puede causar un error ajeno al muestreo.
En realidad, una muestra nunca será exactamente representativa de la población, por lo que siempre habrá algún error de muestreo. Por regla general, cuanto mayor sea la muestra, menor será el error de muestreo.
En estadística, se crea un sesgo de muestreo cuando se recopila una muestra de una población y algunos de sus miembros no tienen la misma probabilidad de que los seleccionen que otros (recuerde que cada miembro de la población debe tener la misma probabilidad de que lo seleccionen). Cuando se produce un sesgo de muestreo, se pueden extraer conclusiones incorrectas sobre la población que se está estudiando.
### Evaluación crítica
Tenemos que evaluar los estudios estadísticos que leemos de forma crítica y analizarlos antes de aceptar sus resultados. Los problemas más comunes que hay que tener en cuenta son:
1. Problemas con las muestras: una muestra debe ser representativa de la población. Una muestra que no es representativa de la población está sesgada. Las muestras sesgadas que no son representativas de la población dan resultados inexactos y no válidos.
2. Muestras autoseleccionadas: las respuestas de las personas que deciden responder, como las encuestas telefónicas, suelen ser poco fiables.
3. Problemas de tamaño de la muestra: las muestras demasiado pequeñas pueden ser poco fiables. Si es posible, las muestras más grandes son mejores. En algunas situaciones, es inevitable contar con muestras pequeñas y, aun así, se pueden usar para sacar conclusiones. Ejemplos: pruebas de choques de automóviles o pruebas médicas para detectar condiciones poco comunes.
4. Influencia indebida: recopilar datos o hacer preguntas de forma que influyan en la respuesta.
5. Falta de respuesta o negativa del sujeto a participar: las respuestas recogidas pueden dejar de ser representativas de la población. A menudo, personas con fuertes opiniones positivas o negativas pueden responder las encuestas, lo que puede afectar los resultados.
6. Causalidad: una relación entre dos variables no significa que una cause la otra. Pueden estar relacionadas (correlacionadas) debido a su relación a través de una variable diferente.
7. Estudios autofinanciados o de interés propio: estudio realizado por una persona u organización para respaldar su afirmación. ¿El estudio es imparcial? Lea atentamente el estudio para evaluar el trabajo. No asuma automáticamente que el estudio es bueno, pero tampoco asuma automáticamente que es deficiente. Valórelo por sus méritos y el trabajo realizado.
8. Uso engañoso de datos: gráficos mal presentados, datos incompletos o falta de contexto.
9. Confusión: cuando los efectos de múltiples factores sobre una respuesta no se pueden separar. Los factores de confusión dificultan o impiden sacar conclusiones válidas sobre el efecto de cada uno de ellos.
Si examinamos dos muestras que representen a la misma población, aunque utilicemos métodos de muestreo aleatorio para las muestras, no serán exactamente iguales. Al igual que hay variación en los datos, hay variación en las muestras. A medida que se acostumbre a la toma de muestras, la variabilidad empezará a parecer natural.
### Variación de los datos
La variación está presente en cualquier conjunto de datos. Por ejemplo, las latas de bebida de 16 onzas pueden contener más o menos de 16 onzas de líquido. En un estudio, se midieron ocho latas de 16 onzas y produjeron la siguiente cantidad (en onzas) de bebida:
Las medidas de la cantidad de bebida en una lata de 16 onzas pueden variar porque diferentes personas hacen las mediciones o porque no se puso la cantidad exacta, 16 onzas de líquido, en las latas. Los fabricantes realizan regularmente pruebas para determinar si la cantidad de bebida en una lata de 16 onzas está dentro del rango deseado.
Tenga en cuenta que, al tomar los datos, estos pueden variar en cierta medida con respecto a los datos que otra persona está tomando para el mismo fin. Esto es completamente natural. Sin embargo, si dos o más de ustedes toman los mismos datos y obtienen resultados muy diferentes, es hora de que usted y los demás reevalúen sus métodos de toma de datos y su exactitud.
### Variación en las muestras
Ya se ha mencionado anteriormente que dos o más muestras de la misma población, tomadas al azar y que se aproximen a las mismas características de la población serán probablemente diferentes entre sí. Supongamos que Doreen y Jung deciden estudiar la cantidad promedio de tiempo que los estudiantes de su instituto universitario duermen cada noche. Doreen y Jung toman cada uno muestras de 500 estudiantes. Doreen utiliza el muestreo sistemático y Jung el muestreo por conglomerados. La muestra de Doreen será diferente a la de Jung. Aunque Doreen y Jung utilizaran el mismo método de muestreo, con toda probabilidad sus muestras serían diferentes. Sin embargo, ninguno de los dos estaría equivocado.
Piense en lo que contribuye a que las muestras de Doreen y Jung sean diferentes.
Si Doreen y Jung tomaran muestras más grandes (es decir, el número de valores de los datos se incrementa), los resultados de su muestra (la cantidad promedio de tiempo que duerme un estudiante) podrían estar más cerca del promedio real de la población. Pero aun así, sus muestras serían, con toda probabilidad, diferentes entre sí. Nunca se insistirá lo suficiente en esta variabilidad en las muestras.
### Tamaño de la muestra
Es importante el tamaño de la muestra (a menudo llamado número de observaciones, normalmente con el símbolo n). Los ejemplos que ha visto en este libro hasta ahora han sido pequeños. Muestras de solo unos cientos de observaciones, o incluso más pequeñas, son suficientes para muchos propósitos. En los sondeos, las muestras que van de 1.200 a 1.500 observaciones se consideran suficientemente grandes y buenas si la encuesta es aleatoria y está bien hecha. Más adelante veremos que hasta los tamaños de muestra mucho más pequeños darán muy buenos resultados. Aprenderá por qué cuando estudie intervalos de confianza.
Tenga en cuenta que muchas muestras grandes están sesgadas. Por ejemplo, las encuestas con llamadas están invariablemente sesgadas porque la gente decide responder o no.
### Referencias
Gallup-Healthways Well-Being Index. http://www.well-beingindex.com/default.asp (consultado el 1.º de mayo de 2013).
Gallup-Healthways Well-Being Index. http://www.well-beingindex.com/methodology.asp (consultado el 1.º de mayo de 2013).
Gallup-Healthways Well-Being Index. http://www.gallup.com/poll/146822/gallup-healthways-index-questions.aspx (consultado el 1.º de mayo de 2013).
Datos de http://www.bookofodds.com/Relationships-Society/Articles/A0374-How-George-Gallup-Picked-the-President
Dominic Lusinchi, “President’ Landon and the 1936 Literary Digest Poll: Were Automobile and Telephone Owners to Blame?” Social Science History 36, n.º 1: 23-54 (2012), http://ssh.dukejournals.org/content/36/1/23.abstract (consultado el 1.º de mayo de 2013).
“The Literary Digest Poll,” Virtual Laboratories in Probability and Statistics http://www.math.uah.edu/stat/data/LiteraryDigest.html (consultado el 1.º de mayo de 2013).
“Gallup Presidential Election Trial-Heat Trends, 1936-2008”, Gallup Politics http://www.gallup.com/poll/110548/gallup-presidential-election-trialheat-trends-19362004.aspx#4 (consultado el 1.º de mayo de 2013).
The Data and Story Library, http://lib.stat.cmu.edu/DASL/Datafiles/USCrime.html (consultado el 1.º de mayo de 2013).
LBCC Distance Learning (DL) program data in 2010-2011, http://de.lbcc.edu/reports/2010-11/future/highlights.html#focus (consultado el 1.º de mayo de 2013).
Datos de The Mercury News de San José
### Repaso del capítulo
Los datos son elementos individuales de información que provienen de una población o muestra. Los datos se clasifican en cualitativos (categóricos), cuantitativos continuos o cuantitativos distintos.
Como no es práctico medir toda la población en un estudio, los investigadores utilizan muestras para representar a la población. Una muestra aleatoria es un grupo representativo de la población elegido mediante un método que da a cada persona de la población la misma oportunidad de que la incluyan en la muestra. Los métodos de muestreo aleatorio incluyen muestreo aleatorio simple, muestreo estratificado, muestreo por conglomerados y muestreo sistemático. El muestreo de conveniencia es un método no aleatorio de elección de una muestra que suele producir datos sesgados.
Las muestras que contienen personas diferentes generan datos diferentes. Esto es así incluso cuando las muestras están bien elegidas y son representativas de la población. Cuando se seleccionan adecuadamente, las muestras más grandes modelan la población con más precisión que las más pequeñas. Hay muchos problemas potenciales que pueden afectar la fiabilidad de una muestra. Los datos estadísticos se deben analizar críticamente, no simplemente aceptarlos.
### TAREA PARA LA CASA
En los siguientes ejercicios identifique el tipo de datos que se utilizaría para describir una respuesta (cuantitativa discreta, cuantitativa continua o cualitativa) y dé un ejemplo de los datos.
Use la siguiente información para responder los dos próximos ejercicios: Se realizó un estudio para determinar la edad de los residentes que utilizan un parque local en San José y el número de veces por semana que van y la duración (cantidad de tiempo). Se seleccionó al azar la primera casa del vecindario que rodea el parque y luego se entrevistó a una de cada 8.ª casa del vecindario que rodea el parque. |
# Muestreo y datos
## Niveles de medición
Una vez que tenga un conjunto de datos, tendrá que organizarlos para poder analizar la frecuencia con la que aparece cada dato en el conjunto. Sin embargo, al calcular la frecuencia, es posible que tenga que redondear sus respuestas para que sean lo más precisas posible.
### Niveles de medición
La forma de medir un conjunto de datos se denomina nivel de medición. Los procedimientos estadísticos correctos dependen de que el investigador esté familiarizado con los niveles de medición. No todas las operaciones estadísticas se pueden usar con todos los conjuntos de datos. Los datos se pueden clasificar en cuatro niveles de medición. Son (de menor a mayor nivel):
1. Nivel de escala nominal
2. Nivel de escala ordinal
3. Nivel de escala de intervalos
4. Nivel de escala de cociente
Los datos que se miden mediante una escala nominal son cualitativos (categóricos). Categorías, colores, nombres, etiquetas y alimentos favoritos junto con las respuestas de sí o no son ejemplos de datos de nivel nominal. Los datos de escala nominal no están ordenados. Por ejemplo, intentar clasificar a las personas según su comida favorita no tiene ningún sentido. Poner la pizza en primer lugar y el sushi en segundo no tiene sentido.
Las compañías de teléfonos inteligentes son otro ejemplo de datos de escala nominal. Los datos son los nombres de las compañías que fabrican teléfonos inteligentes, pero no hay un orden consensuado de estas marcas, aunque la gente pueda tener preferencias personales. Los datos de escala nominal no se pueden usar en cálculos.
Los datos que se miden con una escala ordinal son similares a los datos de la escala nominal, pero hay una gran diferencia. Los datos de la escala ordinal se pueden ordenar. Un ejemplo de datos de escala ordinal es una lista de los cinco mejores parques nacionales de Estados Unidos. Los cinco principales parques nacionales de Estados Unidos se pueden clasificar del uno al cinco, pero no podemos medir las diferencias entre los datos.
Otro ejemplo de uso de la escala ordinal es una encuesta sobre un crucero en la que las respuestas son “excelente”, “bueno”, “satisfactorio” e “insatisfactorio”. Estas respuestas están ordenadas de la respuesta más deseada a la menos deseada. Pero las diferencias entre dos datos no se pueden medir. Al igual que los datos de la escala nominal, los datos de la escala ordinal no se pueden usar en cálculos.
Los datos que se miden con la escala de intervalos son similares a los datos de nivel ordinal porque tienen un orden definido, pero hay una diferencia entre los datos. Las diferencias entre los datos de la escala de intervalos se pueden medir aunque los datos no tengan un punto de partida.
Las escalas de temperatura como Celsius (C) y Fahrenheit (F) se miden utilizando la escala de intervalos. En ambas medidas de temperatura, 40° es igual a 100° menos 60°. Las diferencias tienen sentido. Pero los 0 grados no porque, en ambas escalas, el 0 no es la temperatura mínima absoluta. Existen temperaturas como –10 °F y –15 °C que son más frías que el 0.
Los datos a nivel de intervalo pueden utilizarse en cálculos, pero no se puede hacer un tipo de comparación. 80 °C no es cuatro veces más caliente que 20 °C (ni 80 °F es cuatro veces más caliente que 20 °F). El cociente de 80 a 20 (o de cuatro a uno) no tiene sentido.
Los datos que se miden con la escala de cociente se encargan del problema de las proporciones y ofrecen más información. Los datos de la escala de cociente son como los datos de la escala de intervalos, pero tienen un punto 0 y se pueden calcular cocientes. Por ejemplo, las calificaciones de cuatro exámenes finales de Estadística de opción múltiple son 80, 68, 20 y 92 (sobre 100 puntos posibles). Los exámenes son calificados por máquina.
Los datos se pueden ordenar de menor a mayor: 20, 68, 80, 92.
Las diferencias entre los datos tienen un significado. La calificación de 92 es superior a la de 68 por 24 puntos. Se pueden calcular cocientes. La calificación más baja es 0. Así que 80 es cuatro veces 20. La calificación de 80 es cuatro veces mejor que la de 20.
### Frecuencia
Se les preguntó a veinte estudiantes cuántas horas trabajaban al día. Sus respuestas, en horas, son las siguientes: .
La enumera los diferentes valores de los datos en orden ascendente y sus frecuencias.
Una frecuencia es el número de veces que se produce un valor de los datos. Según la , hay tres estudiantes que trabajan dos horas, cinco estudiantes que trabajan tres horas y así sucesivamente. La suma de los valores de la columna de frecuencia, 20, representa el número total de estudiantes incluidos en la muestra.
Una frecuencia relativa es el cociente (fracción o proporción) entre el número de veces que se produce un valor de los datos en el conjunto de todos los resultados y el número total de resultados. Para hallar las frecuencias relativas, divida cada frecuencia entre el número total de estudiantes de la muestra, en este caso, 20. Las frecuencias relativas se pueden escribir como fracciones, porcentajes o decimales.
La suma de los valores de la columna de frecuencia relativa de la es
, o 1.
La frecuencia relativa acumulada es la acumulación de las frecuencias relativas anteriores. Para hallar las frecuencias relativas acumuladas se suman todas las frecuencias relativas anteriores a la frecuencia relativa de la fila actual, como se muestra en la .
La última entrada de la columna de frecuencia relativa acumulada es uno, lo que indica que se ha acumulado el cien por ciento de los datos.
La representa las alturas, en pulgadas, de una muestra de 100 hombres jugadores de fútbol semiprofesionales.
Los datos de esta tabla se han agrupado en los siguientes intervalos:
1. de 59,95 a 61,95 pulgadas
2. de 61,95 a 63,95 pulgadas
3. de 63,95 a 65,95 pulgadas
4. de 65,95 a 67,95 pulgadas
5. de 67,95 a 69,95 pulgadas
6. de 69,95 a 71,95 pulgadas
7. de 71,95 a 73,95 pulgadas
8. de 73,95 a 75,95 pulgadas
En esta muestra hay cinco jugadores cuyas alturas están dentro del intervalo de 59,95 a 61,95 pulgadas, tres dentro del intervalo de 61,95 a 63,95 pulgadas, 15 dentro del intervalo de 63,95 a 65,95 pulgadas, 40 dentro del intervalo de 65,95 a 67,95 pulgadas, 17 dentro del intervalo de 67,95 a 69,95 pulgadas, 12 jugadores dentro del intervalo de 69,95 a 71,95, siete dentro del intervalo de 71,95 a 73,95 y un jugador cuya altura está dentro del intervalo de 73,95 a 75,95. Todas las alturas caen entre los puntos finales de un intervalo y no en los puntos finales.
### Referencias
“State & County QuickFacts”, U.S. Census Bureau. http://quickfacts.census.gov/qfd/download_data.html (consultado el 1.º de mayo de 2013).
“State & County QuickFacts: Quick, easy access to facts about people, business, and geography”, U.S. Census Bureau. http://quickfacts.census.gov/qfd/index.html (consultado el 1.º de mayo de 2013).
“Table 5: Direct hits by mainland United States Hurricanes (1851-2004)”, National Hurricane Center, http://www.nhc.noaa.gov/gifs/table5.gif (consultado el 1.º de mayo de 2013).
“Levels of Measurement”, http://infinity.cos.edu/faculty/woodbury/stats/tutorial/Data_Levels.htm (consultado el 1.º de mayo de 2013).
Courtney Taylor, “Levels of Measurement”, about.com, http://statistics.about.com/od/HelpandTutorials/a/Levels-Of-Measurement.htm (consultado el 1.º de mayo de 2013).
David Lane. “Levels of Measurement”, Connexions, http://cnx.org/content/m10809/latest/ (consultado el 1.º de mayo de 2013).
### Repaso del capítulo
Algunos cálculos generan números que son artificialmente precisos. No es necesario informar de un valor con ocho decimales cuando las medidas que generaron ese valor solo eran precisas hasta la décima más cercana. Redondee su respuesta final con un decimal más de los que había en los datos originales. Esto significa que si tiene datos medidos a la décima más cercana de una unidad, presente la estadística final a la centésima más cercana.
Además de redondear sus respuestas, puede medir sus datos utilizando los siguientes cuatro niveles de medición.
1. Nivel de escala nominal: datos que no se pueden ordenar ni usar en cálculos
2. Nivel de escala ordinal: datos que se pueden ordenar; las diferencias no se pueden medir
3. Nivel de escala de intervalos: datos con un orden definido pero sin punto de partida; las diferencias se pueden medir, pero no como si fuera un cociente.
4. Nivel de escala de cociente: datos con un punto de partida que se puede ordenar; las diferencias tienen significado y se pueden calcular cocientes.
Al organizar los datos, es importante saber cuántas veces aparece un valor. ¿Cuántos estudiantes de Estadística estudian cinco horas o más para un examen? ¿Qué porcentaje de familias de nuestra manzana tiene dos mascotas? La frecuencia, la frecuencia relativa y la frecuencia relativa acumulada son medidas que responden preguntas como estas.
### TAREA PARA LA CASA
Use la siguiente información para responder los próximos dos ejercicios: la contiene datos sobre los huracanes que han impactado directamente a EE. UU. entre 1851 y 2004. Un huracán recibe una categoría de fuerza basada en la velocidad mínima del viento generada por la tormenta. |
# Muestreo y datos
## Diseño experimental y ética
¿La aspirina reduce el riesgo de infarto? ¿Una marca de abono es más eficaz para el cultivo de rosas que otra? ¿El cansancio es tan peligroso para un conductor como la influencia del alcohol? Este tipo de preguntas se responden con experimentos aleatorios. En este módulo aprenderá aspectos importantes del diseño experimental. Un diseño adecuado del estudio garantiza la obtención de datos fiables y precisos.
El propósito de un experimento es investigar la relación entre dos variables. Cuando una variable provoca un cambio en otra, llamamos a la primera variable la variable independiente o explicativa. La variable afectada se llama variable dependiente o variable de respuesta: estímulo, respuesta. En un experimento aleatorio, el investigador manipula los valores de la variable explicativa y mide los cambios resultantes en la variable de respuesta. Los diferentes valores de la variable explicativa se denominan tratamientos. Una unidad experimental es un único objeto o persona que se va a medir.
Quiere investigar la eficacia de la vitamina E en la prevención de enfermedades. Usted recluta a un grupo de sujetos y les pregunta si toman regularmente vitamina E. Observa que los sujetos que toman vitamina E, en promedio, presentan una salud mejor que quienes no la toman. ¿Esto prueba que la vitamina E es eficaz en la prevención de enfermedades? No es así. Hay muchas diferencias entre los dos grupos comparados, además del consumo de vitamina E. Las personas que toman vitamina E con regularidad suelen tomar otras medidas para mejorar su salud: ejercicio, dieta, otros suplementos vitamínicos, elección de no fumar, etc. Cualquiera de estos factores podría estar influyendo en la salud. Como se ha descrito, este estudio no demuestra que la vitamina E sea la clave para la prevención de enfermedades.
Las variables adicionales que pueden enturbiar un estudio se denominan variables ocultas. Para demostrar que la variable explicativa provoca un cambio en la variable de respuesta, es necesario aislar la variable explicativa. La investigadora debe diseñar su experimento de forma que solo haya una diferencia entre los grupos que se comparan: los tratamientos previstos. Esto se consigue mediante la asignación aleatoria de unidades experimentales a grupos de tratamiento. Cuando los sujetos se asignan a los tratamientos de forma aleatoria, todas las variables ocultas potenciales se reparten por igual entre los grupos. En este punto, la única diferencia entre los grupos es la impuesta por el investigador. Los diferentes resultados medidos en la variable de respuesta, por tanto, deben ser una consecuencia directa de los diferentes tratamientos. De este modo, un experimento puede demostrar una conexión causa-efecto entre las variables explicativas y las de respuesta.
El poder de la sugestión puede tener una importante influencia en el resultado de un experimento. Los estudios han demostrado que la expectativa del participante en el estudio puede ser tan importante como el medicamento real. En un estudio sobre fármacos que mejoran el desempeño, los investigadores señalaron:
Los resultados mostraron que creer que se había tomado la sustancia provocaba tiempos de [desempeño] casi tan rápidos como los asociados al consumo del propio fármaco. Por el contrario, la toma del fármaco sin conocimiento no produjo un aumento significativo del desempeño. (McClung, M. Collins, D. “Because I know it will!”: placebo effects of an ergogenic aid on athletic performance. Journal of Sport & Exercise Psychology. Junio de 2007. 29(3):382-94. Web. 30 de abril de 2013).
Cuando la participación en un estudio provoca una respuesta física del participante, es difícil aislar los efectos de la variable explicativa. Para contrarrestar el poder de la sugestión, los investigadores reservaron un grupo de tratamiento como grupo de control. Este grupo recibe un tratamiento placebo, es decir, un tratamiento que no puede influir en la variable de respuesta. El grupo de control ayuda a los investigadores a equilibrar los efectos de estar en un experimento con los efectos de los tratamientos activos. Por supuesto, si usted participa en un estudio y sabe que está recibiendo una píldora que no contiene ningún medicamento real, entonces el poder de la sugestión ya no es un factor. Que un experimento aleatorio sea ciego preserva el poder de la sugestión. Cuando una persona participa en un estudio de investigación ciego, no sabe quién recibe el tratamiento activo y quién el placebo. Un experimento doble ciego es aquel en el que tanto los sujetos como los investigadores que participan en él no conocen la información del fármaco.
### Referencias
“Vitamin E and Health”, Nutrition Source, Harvard School of Public Health, http://www.hsph.harvard.edu/nutritionsource/vitamin-e/ (consultado el 1.º de mayo de 2013).
Stan Reents. “Don’t Underestimate the Power of Suggestion,” athleteinme.com, http://www.athleteinme.com/ArticleView.aspx?id=1053 (consultado el 1.º de mayo de 2013).
Ankita Mehta. “Daily Dose of Aspiring Helps Reduce Heart Attacks: Study,” International Business Times, 21 de julio de 2011. También disponible en línea en http://www.ibtimes.com/daily-dose-aspirin-helps-reduce-heart-attacks-study-300443 (consultado el 1.º de mayo de 2013).
The Data and Story Library, http://lib.stat.cmu.edu/DASL/Stories/ScentsandLearning.html (consultado el 1.º de mayo de 2013).
M. L. Jacskon et al., “Cognitive Components of Simulated Driving Performance: Sleep Loss effect and Predictors”, Accident Analysis and Prevention Journal, Enero n.º 50 (2013), http://www.ncbi.nlm.nih.gov/pubmed/22721550 (consultado el 1.º de mayo de 2013).
“Earthquake Information by Year”, U.S. Geological Survey. http://earthquake.usgs.gov/earthquakes/eqarchives/year/ (consultado el 1.º de mayo de 2013).
“Fatality Analysis Report Systems (FARS) Encyclopedia”, National Highway Traffic and Safety Administration. http://www-fars.nhtsa.dot.gov/Main/index.aspx (consultado el 1.º de mayo de 2013).
Datos de www.businessweek.com (consultado el 1.º de mayo de 2013).
Datos de www.forbes.com (consultado el 1.º de mayo de 2013).
“America’s Best Small Companies”, http://www.forbes.com/best-small-companies/list/ (consultado el 1.º de mayo de 2013).
U.S. Department of Health and Human Services, Code of Federal Regulations Title 45 Public Welfare Department of Health and Human Services Part 46 Protection of Human Subjects, revisado el 15 de enero de 2009. Section 46.111:Criteria for IRB Approval of Research.
“April 2013 Air Travel Consumer Report”, U.S. Department of Transportation, 11 de abril (2013), http://www.dot.gov/airconsumer/april-2013-air-travel-consumer-report (consultado el 1.º de mayo de 2013).
Lori Alden, “Statistics can be Misleading”, econoclass.com, http://www.econoclass.com/misleadingstats.html (consultado el 1.º de mayo de 2013).
María de los A. Medina, “Ethics in Statistics”, basado en “Building an Ethics Module for Business, Science, and Engineering Students” de José A. Cruz-Cruz y William Frey, Connexions, http://cnx.org/content/m15555/latest/ (consultado el 1.º de mayo de 2013).
### Repaso del capítulo
Un estudio de diseño deficiente no producirá datos fiables. Hay ciertos componentes clave que deben incluirse en cada experimento. Para eliminar las variables ocultas los sujetos deben ser asignados aleatoriamente a diferentes grupos de tratamiento. Uno de los grupos debe actuar como grupo de control, con lo que se demuestra lo que ocurre cuando no se aplica el tratamiento activo. Los participantes del grupo de control reciben un tratamiento placebo que es exactamente igual a los tratamientos activos, pero que no puede influir en la variable de respuesta. Para preservar la integridad del placebo, tanto los investigadores como los sujetos pueden estar sin conocimiento del fármaco. Cuando un estudio se diseña correctamente la única diferencia entre los grupos de tratamiento es la impuesta por el investigador. Por lo tanto, cuando los grupos responden de forma diferente a los distintos tratamientos, la diferencia debe ser por la influencia de la variable explicativa.
“Un problema de ética surge cuando se plantea una acción que le beneficia a usted o a alguna causa que apoya, perjudica o reduce los beneficios de otras personas y viola alguna norma” (Andrew Gelman, “Open Data and Open Methods”, Ethics and Statistics, http://www.stat.columbia.edu/~gelman/research/published/ChanceEthics1.pdf [consultado el 1.º de mayo de 2013]). Las violaciones de la ética en las estadísticas no siempre son fáciles de detectar. Asociaciones profesionales y agencias federales publican directrices sobre la conducta adecuada. Es importante que aprenda los procedimientos estadísticos básicos para que pueda reconocer un análisis de datos adecuado. |
# Estadística descriptiva
## Introducción
Una vez que haya recopilado los datos, ¿qué hará con ellos? Los datos se pueden describir y presentar en muchos formatos diferentes. Por ejemplo, supongamos que está interesado en comprar una casa en una zona determinada. Es posible que no tenga ni idea de los precios de las viviendas, por lo que puede pedirle a su agente inmobiliario que le dé un conjunto de datos de muestra de los precios. Mirar todos los precios de la muestra suele ser abrumador. Una mejor forma sería observar la mediana del precio y la variación de los precios. La mediana y la variación son solo dos formas que aprenderá para describir los datos. Su agente también puede proporcionarle un gráfico de los datos.
En este capítulo estudiará las formas numéricas y gráficas de describir y mostrar sus datos. Esta área de la estadística se llama “Estadística Descriptiva”. Aprenderá a calcular y, lo que es más importante, a interpretar estas medidas y gráficos.
Un gráfico estadístico es una herramienta que ayuda a conocer la forma o la distribución de una muestra o de una población. Un gráfico puede ser una forma más eficaz de presentar los datos que una masa de números porque podemos ver dónde se agrupan los datos y dónde hay solo unos pocos valores de datos. Los periódicos e internet utilizan gráficos para mostrar tendencias y permitir a los lectores comparar rápidamente datos y cifras. Los estadísticos suelen hacer primero un gráfico de los datos para hacerse una idea de lo que arrojan. Luego, se pueden aplicar herramientas más formales.
Algunos de los tipos de gráficos que se utilizan para resumir y organizar los datos son el diagrama de puntos, el gráfico de barras, el histograma, el diagrama de tallo y hojas, el polígono de frecuencias (un tipo de gráfico de líneas discontinuas), el gráfico circular y el diagrama de caja. En este capítulo veremos brevemente gráficos de tallo y hoja, gráficos de líneas y gráficos de barras, así como polígonos de frecuencia y gráficos de series temporales. Haremos hincapié en los histogramas y los diagramas de caja. |
# Estadística descriptiva
## Datos mostrados
### Gráficos de tallo y hoja (gráfico de tallo), gráficos de líneas y gráficos de barras
Un gráfico sencillo, el gráfico de tallo y hoja o gráfico de tallo, procede del campo del análisis exploratorio de datos. Es una buena opción cuando los conjuntos de datos son pequeños. Para crear el gráfico, divida cada observación de datos en un tallo y una hoja. La hoja consta de un último dígito significativo. Por ejemplo, 23 tiene el tallo dos y la hoja tres. El número 432 tiene el tallo 43 y la hoja dos. Asimismo, el número 5.432 tiene el tallo 543 y la hoja dos. El decimal 9,3 tiene el tallo nueve y la hoja tres. Escriba los tallos en una línea vertical de menor a mayor. Dibuje una línea vertical a la derecha de los tallos. Luego, escriba las hojas en orden creciente junto a su correspondiente tallo.
El diagrama de tallo es una forma rápida de representar datos gráficamente y ofrece una imagen exacta de la información. Hay que buscar un patrón general y los valores atípicos. Un valor atípico es una observación de datos que no se ajusta al resto de los datos. A veces se le llama valor extremo. Cuando grafique un valor atípico parecerá que no se ajusta al patrón del gráfico. Algunos valores atípicos se deben a errores (por ejemplo, anotar 50 en vez de 500), mientras que otros pueden indicar que está ocurriendo algo inusual. Para explicar los valores atípicos se necesita información de fondo, por lo que los trataremos con más detalle más adelante.
Otro tipo de gráfico que resulta útil para valores de datos específicos es el gráfico de líneas. En el gráfico de líneas en particular que se muestra en el , el eje (eje horizontal) está formado por los valores de los datos y el eje (eje vertical) por puntos de frecuencia. Los puntos de frecuencia se conectan mediante segmentos de la línea.
Los gráficos de barras están formados por barras separadas entre sí. Las barras pueden ser rectángulos o recuadros rectangulares (usados en representaciones tridimensionales), y pueden ser verticales u horizontales. El gráfico de barras que se muestra en el tiene los grupos de edad representados en el eje y las proporciones en el eje .
### Histogramas, polígonos de frecuencia y gráficos de series temporales
Para la mayor parte del trabajo que se realiza en este libro se utilizará un histograma para mostrar los datos. Una de las ventajas de un histograma es que puede mostrar fácilmente grandes conjuntos de datos. Una regla general es utilizar un histograma cuando el conjunto de datos consta de 100 valores o más.
Un histograma está formado por recuadros contiguos (adyacentes). Tiene un eje horizontal y otro vertical. El eje horizontal está identificado con lo que representan los datos (por ejemplo, la distancia de su casa a la escuela). El eje vertical está identificado como frecuencia o frecuencia relativa (o porcentaje de frecuencia o probabilidad). El gráfico tendrá la misma forma con cualquiera de las dos etiquetas. El histograma (al igual que el diagrama de tallo) puede darle la forma de los datos, el centro y la dispersión de los datos.
La frecuencia relativa es igual a la frecuencia de un valor observado de los datos dividida entre el número total de valores de los datos en la muestra. (Recuerde que la frecuencia se define como el número de veces que se produce una respuesta). Si:
1. f = frecuencia
2. n = número total de valores de datos (o la suma de las frecuencias individuales) y
3. RF = frecuencia relativa,
entonces:
Por ejemplo, si tres estudiantes de la clase de Inglés del Sr. Ahab compuesta por 40 estudiantes obtuvieron del 90 % al 100 %, entonces, f = 3, n = 40 y RF = = = 0,075. El 7,5 % de los estudiantes obtuvieron del 90 % al 100 %. Del 90 % al 100 % son medidas cuantitativas.
Para construir un histograma, primero hay que decidir cuántas barras o intervalos (también llamados clases) representan los datos. Muchos histogramas constan de cinco a 15 barras o clases para mayor claridad. Hay que elegir el número de barras. Elija un punto de partida para que el primer intervalo sea menor que el valor más pequeño de los datos. Un punto de partida conveniente es un valor inferior llevado a un decimal más que el valor con más decimales. Por ejemplo, si el valor con más decimales es 6,1 y este es el valor más pequeño, un punto de partida conveniente es 6,05 (6,1 – 0,05 = 6,05). Decimos que 6,05 tiene más precisión. Si el valor con más decimales es 2,23 y el valor más bajo es 1,5, un punto de partida conveniente es 1,495 (1,5 – 0,005 = 1,495). Si el valor con más decimales es 3,234 y el valor más bajo es 1,0, un punto de partida conveniente es 0,9995 (1,0 – 0,0005 = 0,9995). Si todos los datos son enteros y el valor más pequeño es dos, un punto de partida conveniente es 1,5 (2 – 0,5 = 1,5). Además, cuando el punto de partida y otros límites se llevan a un decimal adicional, ningún valor de los datos caerá en un límite. Los dos siguientes ejemplos detallan cómo construir un histograma utilizando datos continuos y cómo crear un histograma utilizando datos discretos.
### Polígonos de frecuencia
Los polígonos de frecuencias son análogos a los gráficos de líneas y, al igual que los gráficos de líneas facilitan la interpretación visual de los datos continuos, también lo hacen los polígonos de frecuencias.
Para construir un polígono de frecuencias, primero hay que examinar los datos y decidir el número de intervalos, o intervalos de clase, que se van a utilizar en los ejes x y y. Después de elegir los rangos apropiados, comience a trazar los puntos de datos. Después de trazar todos los puntos, dibuje segmentos de línea para conectarlos.
Los polígonos de frecuencia son útiles para comparar distribuciones. Esto se consigue superponiendo los polígonos de frecuencia dibujados para diferentes conjuntos de datos.
### Construcción de un gráfico de series temporales
Supongamos que queremos estudiar el rango de temperaturas de una región durante todo un mes. Todos los días a mediodía anotamos la temperatura y la anotamos en un registro. Con estos datos se podrían realizar diversos estudios estadísticos. Podemos hallar la media o la mediana de la temperatura del mes. Podemos construir un histograma que muestre el número de días en que las temperaturas alcanzan un determinado rango de valores. Sin embargo, todos estos métodos ignoran una parte de los datos que hemos recopilado.
Una característica de los datos que podemos considerar es la del tiempo. Dado que cada fecha se empareja con la lectura de la temperatura del día, no tenemos que pensar que los datos son aleatorios. En cambio, podemos utilizar los tiempos indicados para imponer un orden cronológico a los datos. Un gráfico que reconoce esta ordenación y muestra la evolución de la temperatura a medida que avanza el mes se denomina gráfico de series temporales.
Para construir un gráfico de series temporales debemos observar las dos partes de nuestro conjunto de datos emparejados. Comenzamos con un sistema de coordenadas cartesianas estándar. El eje horizontal se utiliza para trazar la fecha o los incrementos de tiempo, y el eje vertical se utiliza para trazar los valores de la variable que estamos midiendo. De este modo, hacemos que cada punto del gráfico corresponda a una fecha y a una cantidad medida. Los puntos del gráfico suelen estar conectados por líneas rectas en el orden en que se producen.
### Usos de un gráfico de series temporales
Los gráficos de series temporales son herramientas importantes en diversas aplicaciones de la estadística. Cuando se registran los valores de una misma variable durante un largo periodo, a veces, es difícil discernir cualquier tendencia o patrón. Sin embargo, una vez que los mismos puntos de datos se muestran gráficamente, algunas características saltan a la vista. Los gráficos de series temporales facilitan la detección de tendencias.
### Cómo NO mentir con las estadísticas
Es importante recordar que la razón por la que desarrollamos una variedad de métodos para presentar los datos es para comprender el tema de lo que las observaciones representan. Queremos tener una "sensación" de los datos. ¿Las observaciones son todas muy parecidas o están repartidas en un amplio rango de valores, están agrupadas en un extremo del espectro o están distribuidas uniformemente, etc.? Intentamos obtener una representación visual de los datos numéricos. En breve desarrollaremos medidas matemáticas formales de los datos, pero nuestra presentación gráfica visual puede decir mucho. Desgraciadamente, también puede decir muchas cosas que distraen, confunden y simplemente son erróneas en cuanto a la impresión que lo visual deja. Hace muchos años, Darrell Huff escribió el libro How to Lie with Statistics [Cómo mentir con estadísticas]. Ha tenido más de 25 ediciones y ha vendido más de un millón y medio de ejemplares. Su perspectiva era dura y utilizaba muchos ejemplos reales destinados a engañar. Quería hacer que la gente fuera consciente de ese engaño, pero quizás lo más importante era educar para que otros no cometieran los mismos errores inadvertidamente.
De nuevo, el objetivo es ilustrar con imágenes que cuenten la historia de los datos. Los gráficos circulares tienen una serie de problemas comunes cuando se utilizan para transmitir el mensaje de los datos. Demasiados trozos del pastel abruman al lector. Más de quizás cinco o seis categorías deberían dar una idea de la importancia relativa de cada trozo. Al fin y al cabo, este es el objetivo de un gráfico circular: qué subconjunto importa más en relación con los demás. Si hay más componentes que esto, tal vez sea mejor un enfoque alternativo o tal vez algunos puedan consolidarse en una categoría "otros". Los gráficos circulares no pueden mostrar los cambios a lo largo del tiempo, aunque vemos que esto se intenta con demasiada frecuencia. En los documentos financieros federales, estatales y municipales se suelen presentar gráficos circulares para mostrar los componentes de los ingresos de los que dispone el órgano de gobierno para su consignación: impuesto sobre la renta, impuesto sobre las ventas, impuestos sobre los vehículos de motor, etc. En sí misma es una información interesante y se puede hacer muy bien con un gráfico circular. El error se produce cuando se ponen dos años uno al lado del otro. Como los ingresos totales cambian de un año a otro, pero el tamaño del pastel es fijo, no se proporciona ninguna información real y no se puede comparar de forma significativa el tamaño relativo de cada trozo del pastel.
Los histogramas pueden ser muy útiles para entender los datos. Si se presentan correctamente, pueden ser una forma visual rápida de presentar las probabilidades de las diferentes categorías mediante la simple visualización de la comparación de las áreas relativas en cada categoría. Aquí el error, intencionado o no, es variar la amplitud de las categorías. Por supuesto, esto hace imposible la comparación con las demás categorías. Adorna la importancia de la categoría con un ancho ampliado porque tiene un área mayor, de forma inapropiada, y así "dice" visualmente que esa categoría tiene una mayor probabilidad de ocurrencia.
Los gráficos de series temporales tal vez sean de los que más se abusa. Un gráfico de alguna variable a lo largo del tiempo nunca debe presentarse en ejes que cambien en parte de la página, ya sea en la dimensión vertical u horizontal. Tal vez se cambie el marco temporal de años a meses. Probablemente esto se haga para ahorrar espacio o porque los datos mensuales no estaban disponibles para los primeros años. En cualquier caso, esto confunde la presentación y destruye cualquier valor del gráfico. Si esto no se hace para confundir a propósito al lector, entonces ciertamente es un trabajo perezoso o descuidado.
Cambiar las unidades de medida del eje puede suavizar o acentuar una caída. Si quiere mostrar grandes cambios, mida la variable en unidades pequeñas, centavos en lugar de miles de dólares. Y, por supuesto, para continuar con el fraude, asegúrese de que el eje no comienza en cero, cero. Si comienza en cero, cero entonces se hace evidente que el eje ha sido manipulado.
Tal vez tenga un cliente al que le preocupa la volatilidad de la cartera que usted gestiona. Una forma fácil de presentar los datos es utilizar periodos largos en el gráfico de la serie temporal. Utilice meses o, mejor, trimestres en lugar de datos diarios o semanales. Si eso no consigue reducir la volatilidad, entonces separe el eje temporal en relación con el eje de la tasa de rendimiento o de la valoración de la cartera. Si quiere mostrar un crecimiento dramático "rápido", entonces reduzca el eje temporal. Cualquier crecimiento positivo mostrará tasas de crecimiento visualmente "altas". Tenga en cuenta que si el crecimiento es negativo, este truco mostrará que la cartera se está hundiendo a un ritmo dramático.
Una vez más, el objetivo de la Estadística Descriptiva es transmitir imágenes significativas que cuenten la historia de los datos. La manipulación intencionada es un fraude y una falta de ética en el peor de los casos, pero incluso en el mejor, cometer este tipo de errores llevará a la confusión del análisis.
### Referencias
Burbary, Ken. Facebook Demographics Revisited–2001 Statistics, 2011. Disponible en línea en http://www.kenburbary.com/2011/03/facebook-demographics-revisited-2011-statistics-2/ (consultado el 21 de agosto de 2013).
“9th Annual AP Report to the Nation”. CollegeBoard, 2013. Disponible en línea en http://apreport.collegeboard.org/goals-and-findings/promoting-equity (consultado el 13 de septiembre de 2013).
“Overweight and Obesity: Adult Obesity Facts”. Centers for Disease Control and Prevention. Disponible en línea en http://www.cdc.gov/obesity/data/adult.html (consultado el 13 de septiembre de 2013).
Datos sobre los homicidios anuales en Detroit, 1961-1973, extraídos del libro de Gunst & Mason: “Regression Analysis and its Application”, Marcel Dekker
“Timeline: Guide to the U.S. Presidents: Information on every president’s birthplace, political party, term of office, and more”. Scholastic, 2013. Disponible en línea en http://www.scholastic.com/teachers/article/timeline-guide-us-presidents (consultado el 3 de abril de 2013).
“Presidents”. Fact Monster. Pearson Education, 2007. Disponible en línea en http://www.factmonster.com/ipka/A0194030.html (consultado el 3 de abril de 2013).
“Food Security Statistics”. Food and Agriculture Organization of the United Nations. Disponible en línea en http://www.fao.org/economic/ess/ess-fs/en/ (consultado el 3 de abril de 2013).
“Consumer Price Index”. United States Department of Labor: Bureau of Labor Statistics. Disponible en línea en http://data.bls.gov/pdq/SurveyOutputServlet (consultado el 3 de abril de 2013).
“CO2 emissions (kt)”. The World Bank, 2013. Disponible en línea en http://databank.worldbank.org/data/home.aspx (consultado el 3 de abril de 2013).
“Births Time Series Data”. General Register Office For Scotland, 2013. Disponible en línea en http://www.gro-scotland.gov.uk/statistics/theme/vital-events/births/time-series.html (consultado el 3 de abril de 2013).
“Demographics: Children under the age of 5 years underweight”. Indexmundi. Disponible en línea en http://www.indexmundi.com/g/r.aspx?t=50&v=2224&aml=en (consultado el 3 de abril de 2013).
Gunst, Richard, Robert Mason. Regression Analysis and Its Application: A Data-Oriented Approach. CRC Press: 1980.
“Overweight and Obesity: Adult Obesity Facts”. Centers for Disease Control and Prevention. Disponible en línea en http://www.cdc.gov/obesity/data/adult.html (consultado el 13 de septiembre de 2013).
### Repaso del capítulo
Un gráfico de tallo y hoja es una forma de representar los datos y observar la distribución. En un gráfico de tallo y hoja todos los valores de los datos de una clase son visibles. La ventaja de un gráfico de tallo y hoja es que se enumeran todos los valores, a diferencia de un histograma, que da clases de valores de datos. Un gráfico de líneas se suele usar para representar un conjunto de valores de datos en los que una cantidad varía con el tiempo. Estos gráficos son útiles para hallar tendencias. Es decir, hallar un patrón general en conjuntos de datos que incluyan temperatura, ventas, empleo, ganancias o costos de la compañía durante un periodo. Un gráfico de barras es un gráfico que utiliza barras horizontales o verticales para mostrar comparaciones entre categorías. Un eje del gráfico muestra las categorías específicas que se comparan, y el otro eje representa un valor discreto. Algunos gráficos de barras presentan las barras agrupadas en grupos de más de uno (gráficos de barras agrupados), y otros muestran las barras divididas en subpartes para mostrar el efecto acumulativo (gráficos de barras apilados). Los gráficos de barras son especialmente útiles cuando se utilizan datos categóricos.
Un histograma es una versión gráfica de una distribución de frecuencias. El gráfico consiste en barras de igual ancho dibujadas de forma adyacente. La escala horizontal representa clases de valores de datos cuantitativos y la escala vertical representa frecuencias. Las alturas de las barras corresponden a valores de frecuencia. Los histogramas se suelen utilizar para conjuntos de datos cuantitativos, continuos y de gran tamaño. Un polígono de frecuencias también se puede usar cuando se grafican grandes conjuntos de datos con puntos de datos que se repiten. Los datos suelen ir en el eje y, y la frecuencia se representa en el eje x. Los gráficos de series temporales pueden ser útiles cuando se observan grandes cantidades de datos de una variable durante un periodo.
Para los tres ejercicios siguientes utilice los datos para construir un gráfico de líneas.
### Tarea para la casa
Use la siguiente información para responder los dos próximos ejercicios: supongamos que se les pregunta a ciento once personas que compran en una tienda especial de camisetas el número de camisetas que tienen y que cuestan más de 19 dólares cada una.
|
# Estadística descriptiva
## Medidas de la ubicación de los datos
Las medidas habituales de localización son cuartiles y percentiles
Los cuartiles son percentiles especiales. El primer cuartil, Q1, es igual que el percentil 25, y el tercer cuartil, Q3, es igual que el percentil 75. La mediana, M, se denomina tanto el segundo cuartil como el percentil 50.
Para calcular cuartiles y percentiles, los datos se deben ordenar de menor a mayor. Los cuartiles dividen los datos ordenados en cuartos. Los percentiles dividen los datos ordenados en centésimas. Obtener una calificación en el percentil 90 de un examen no significa, necesariamente, que haya obtenido el 90 % en una prueba. Significa que el 90 % de las calificaciones de las pruebas son iguales o inferiores a su calificación y el 10 % de las calificaciones de las pruebas son iguales o superiores a su calificación.
Los percentiles son útiles para comparar valores. Por esta razón, universidades e institutos universitarios usan ampliamente los percentiles. Uno de los casos en los que institutos universitarios y universidades utilizan los percentiles es cuando los resultados del SAT se emplean para determinar una calificación mínima del examen que se utilizará como factor de aceptación. Por ejemplo, supongamos que Duke acepta calificaciones del SAT iguales o superiores al percentil 75. Eso se traduce en una calificación de, al menos, 1.220.
Los percentiles se utilizan sobre todo con poblaciones muy grandes. Por lo tanto, si se dijera que el 90 % de las calificaciones de las pruebas son menores (y no iguales o menores) que su calificación, sería aceptable porque eliminar un valor de datos particular no es significativo.
La mediana es un número que mide el “centro” de los datos. Se puede pensar en la mediana como el “valor medio”, pero no tiene por qué ser uno de los valores observados. Es un número que separa los datos ordenados en mitades. La mitad de los valores son iguales o menores que la mediana, y la mitad de los valores son iguales o mayores. Por ejemplo, considere los siguientes datos. 1; 11,5; 6; 7,2; 4; 8; 9; 10; 6,8; 8,3; 2; 2; 10; 1 Ordenado de menor a mayor: 1; 1; 2; 2; 4; 6; 6,8; 7,2; 8; 8,3; 9; 10; 10; 11,5
Como hay 14 observaciones, la mediana está entre el séptimo valor, 6,8, y el octavo, 7,2. Para hallar la mediana, sume los dos valores y divídalos entre dos.
La mediana es siete. La mitad de los valores son menores que siete y la mitad de los valores son mayores que siete.
Los cuartiles son números que separan los datos en cuartos. Los cuartiles pueden o no formar parte de los datos. Para hallar los cuartiles, primero hay que hallar la mediana o el segundo cuartil. El primer cuartil, Q1, es el valor central de la mitad inferior de los datos, y el tercer cuartil, Q3, es el valor central, o la mediana, de la mitad superior de los datos. Para hacerse una idea, considere el mismo conjunto de datos: 1; 1; 2; 2; 4; 6; 6,8; 7,2; 8; 8,3; 9; 10; 10; 11,5
La mediana o segundo cuartil es siete. La mitad inferior de los datos son 1; 1; 2; 2; 4; 6; 6,8. El valor central de la mitad inferior es dos. 1; 1; 2; 2; 4; 6; 6,8
El número dos, que forma parte de los datos, es el primer cuartil. Una cuarta parte de los conjuntos de valores son iguales o inferiores a dos y tres cuartas partes de los valores son superiores a dos.
La mitad superior de los datos es 7,2; 8; 8,3; 9; 10; 10; 11,5. El valor central de la mitad superior es nueve.
El tercer cuartil, Q3, es nueve. Tres cuartas partes (75 %) del conjunto de datos ordenados son menores de nueve. Una cuarta parte (25 %) del conjunto de datos ordenados son mayores de nueve. El tercer cuartil forma parte del conjunto de datos de este ejemplo.
El rango intercuartil es un número que indica la dispersión de la mitad central o del 50 % central de los datos. Es la diferencia entre el tercer cuartil (Q3) y el primer cuartil (Q1).
IQR = Q3 – Q1
El IQR puede ayudar a determinar posibles valores atípicos. Se sospecha que un valor es un posible valor atípico si está menos de (1,5)(. Los posibles valores atípicos siempre requieren una investigación más profunda.
### Una fórmula para hallar el percentil k
Si investiga un poco, hallará varias fórmulas para calcular el percentil k Aquí está una de ellas.
k = el percentil k. Puede o no formar parte de los datos.
i = el índice (clasificación o posición de un valor de datos)
n = el número total de puntos de datos u observaciones
1. Ordene los datos de menor a mayor.
2. Calcule
3. Si i es un número entero, el percentil k es el valor de los datos en la posición i en el conjunto ordenado de datos.
4. Si i no es un entero, entonces redondee i hacia arriba o redondee i hacia abajo a los enteros más cercanos. Promedia los dos valores de los datos en estas dos posiciones en el conjunto de datos ordenados. Esto es más fácil de entender con un ejemplo.
### Una fórmula para hallar el percentil de un valor en un conjunto de datos
1. Ordene los datos de menor a mayor.
2. x = el número de valores de datos contando desde la parte inferior de la lista de datos hasta, pero sin incluir, el valor de datos para el que se desea hallar el percentil.
3. y = el número de valores de datos iguales al valor de los datos para los que se quiere hallar el percentil.
4. n = el número total de datos.
5. Calcule (100). Luego, redondee al número entero más cercano.
### Interpretación de percentiles, cuartiles y mediana
Un percentil indica la posición relativa de un valor de datos cuando estos se ordenan numéricamente de menor a mayor. Los porcentajes de los valores de los datos son menores o iguales al percentil p. Por ejemplo, el 15 % de los valores de los datos son inferiores o iguales al percentil 15.
1. Los percentiles bajos corresponden siempre a valores de datos más bajos.
2. Los percentiles altos corresponden siempre a valores de datos más altos.
Un percentil puede corresponder o no a un juicio de valor sobre si es “bueno” o “deficiente”. La interpretación de si un determinado percentil es “bueno” o “deficiente” depende del contexto de la situación a la que se aplican los datos. En algunas situaciones, un percentil bajo se consideraría “bueno”; en otros contextos, un percentil alto podría considerarse “bueno”. En muchas situaciones no se aplica ningún juicio de valor.
Entender cómo interpretar correctamente los percentiles es importante no solo a la hora de describir los datos, sino también a la hora de calcular las probabilidades en capítulos posteriores de este texto.
### Referencias
Cauchon, Dennis, Paul Overberg. “Census data shows minorities now a majority of U.S. births”. USA Today, 2012. Disponible en línea en http://usatoday30.usatoday.com/news/nation/story/2012-05-17/minority-birthscensus/55029100/1 (consultado el 3 de abril de 2013).
Datos del Departamento de Comercio de Estados Unidos: Oficina del Censo de Estados Unidos. Disponible en línea en http://www.census.gov/ (consultado el 3 de abril de 2013).
“1990 Census”. United States Department of Commerce: Oficina del Censo de Estados Unidos. Disponible en línea en http://www.census.gov/main/www/cen1990.html (consultado el 3 de abril de 2013).
Datos de The Mercury News de San José.
Datos de la Revista Time; encuesta de Yankelovich Partners, Inc.
### Repaso del capítulo
Los valores que dividen un conjunto de datos ordenados en 100 partes iguales se llaman percentiles. Los percentiles se utilizan para comparar e interpretar datos. Por ejemplo, una observación en el percentil 50 sería mayor que el 50 % de las demás observaciones del conjunto. Los cuartiles dividen los datos en cuartos. El primer cuartil (Q1) es el percentil 25, el segundo cuartil (Q2 o mediana) es el percentil 50 y el tercer cuartil (Q3) es el percentil 75. El rango intercuartil, o IQR, es el rango del 50 % del centro de los valores de los datos. El IQR se encuentra restando Q1 de Q3, y puede ayudar a determinar los valores atípicos utilizando las dos expresiones siguientes.
1. Q3 + IQR(1,5)
2. Q1 – IQR(1,5)
### Revisión de la fórmula
donde i = la clasificación o posición de un valor de datos,
k = el percentil k,
n = número total de datos.
Expresión para hallar el percentil de un valor de datos:
(100)
donde x = el número de valores contando desde el final de la lista de datos hasta el valor de los datos para el que se quiere hallar el percentil, pero sin incluirlo,
y = el número de valores de datos iguales al valor de los datos para los que se quiere hallar el percentil,
n = número total de datos
Use la siguiente información para responder los próximos seis ejercicios. se preguntó a sesenta y cinco vendedores de automóviles seleccionados al azar el número de automóviles que suelen vender en una semana. Catorce personas respondieron que generalmente venden tres, diecinueve que venden cuatro, doce que venden cinco, nueve que venden seis y once que venden siete.
### Tarea para la casa
|
# Estadística descriptiva
## Medidas del centro de los datos
El “centro” de un conjunto de datos también es una forma de describir la ubicación. Las dos medidas más utilizadas del “centro” de los datos son la media (promedio) y la mediana. Para calcular el peso medio de 50 personas, sume los 50 pesos y los divide entre 50. Técnicamente es la media aritmética. Más adelante hablaremos de la media geométrica. Para hallar la mediana del peso de las 50 personas, ordene los datos y halle el número que divide los datos en dos partes iguales, lo que significa un número igual de observaciones en cada lado. El peso de 25 personas está por debajo de ese peso y 25 personas están por encima de ese peso. La mediana suele ser una mejor medida del centro cuando hay valores extremos o atípicos porque no se ve afectada por los valores numéricos precisos de los atípicos. La media es la medida más común del centro.
Cuando cada valor del conjunto de datos no es único, la media se puede calcular multiplicando cada valor distinto por su frecuencia y dividiendo después la suma por el número total de valores de los datos. La letra utilizada para representar la media muestral es una x con una barra encima (se pronuncia “barra de x”): .
La letra griega μ (se pronuncia “mu”) representa la media de la población. Uno de los requisitos para que la media muestral sea una buena estimación de la media de la población es que la muestra tomada sea realmente aleatoria.
Para ver que ambas formas de calcular la media son iguales, considere la muestra: 1; 1; 1; 2; 2; 3; 4; 4; 4; 4; 4
En el segundo cálculo, las frecuencias son 3, 2, 1 y 5.
Puede hallar rápidamente la ubicación de la mediana utilizando la expresión
.
La letra n es el número total de valores de datos en la muestra. Si n es un número impar, la mediana es el valor del centro de los datos ordenados (ordenados de menor a mayor). Si n es un número par, la mediana es igual a los dos valores del centro sumados y divididos entre dos después de ordenar los datos. Por ejemplo, si el número total de valores de datos es de 97, entonces
=
= 49. La mediana es el 49.º valor de los datos ordenados. Si el número total de valores de datos es 100, entonces
=
= 50,5. La mediana está a medio camino entre los valores 50.º y 51.º. La ubicación de la mediana y el valor de la mediana no son lo mismo. La letra M mayúscula se utiliza a menudo para representar la mediana. El siguiente ejemplo ilustra la ubicación de la mediana y su valor.
Otra medida del centro es la moda. La moda es el valor más frecuente. Puede haber más de una moda en un conjunto de datos siempre que esos valores tengan la misma frecuencia y esta sea la más alta. Un conjunto de datos con dos modas se denomina bimodal.
### Cálculo de la media aritmética de tablas de frecuencias agrupadas
Cuando solo se dispone de datos agrupados no se conocen los valores individuales de los datos (solo conocemos los intervalos y las frecuencias de los intervalos); por lo tanto, no se puede calcular una media exacta para el conjunto de datos. Lo que debemos hacer es estimar la media real calculando la media de una tabla de frecuencias. Una tabla de frecuencias es una representación de datos en la que se muestran datos agrupados junto con las frecuencias correspondientes. Para calcular la media de una tabla de frecuencias agrupadas podemos aplicar la definición básica de media: media = Simplemente tenemos que modificar la definición para que se ajuste a las restricciones de una tabla de frecuencias.
Como no conocemos los valores individuales de los datos podemos hallar el punto medio de cada intervalo. El punto medio es
. Ahora podemos modificar la definición de la media para que sea donde f = la frecuencia del intervalo y m = el punto medio del intervalo.
### Referencias
Datos del Banco Mundial, disponibles en línea en http://www.worldbank.org (consultado el 3 de abril de 2013).
“Demographics: Obesity – adult prevalence rate”. Indexmundi. Disponible en línea en http://www.indexmundi.com/g/r.aspx?t=50&v=2228&l=en (consultado el 3 de abril de 2013).
### Repaso del capítulo
La media y la mediana se pueden calcular para ayudar a hallar el “centro” de un conjunto de datos. La media es la mejor estimación para el conjunto de datos reales, pero la mediana es la mejor medida cuando un conjunto de datos contiene varios valores atípicos o extremos. La moda le indicará el dato (o los datos) que aparecen con más frecuencia en su conjunto de datos. La media, la mediana y la moda son extremadamente útiles cuando se necesita analizar datos, pero si el conjunto de datos está formado por rangos que carecen de valores específicos, la media puede parecer imposible de calcular. Sin embargo, la media se puede aproximar si se suma el límite inferior con el superior y se divide entre dos para hallar el punto medio de cada intervalo. Multiplique cada punto medio por el número de valores hallados en el rango correspondiente. Divida la suma de estos valores entre el número total de valores de datos del conjunto.
### Revisión de la fórmula
Donde f = frecuencias de intervalo y m = puntos medios de intervalo.
La media aritmética de una muestra (denominada ) es
La media aritmética de una población (denominada μ) es
Use la siguiente información para responder los próximos tres ejercicios: los siguientes datos muestran las esloras de barcos atracados en un puerto. Los datos están ordenados de menor a mayor:
Use la siguiente información para responder los próximos tres ejercicios: se preguntó a sesenta y cinco vendedores de automóviles seleccionados al azar el número de automóviles que suelen vender en una semana. Catorce personas respondieron que generalmente venden tres, diecinueve que venden cuatro, doce que venden cinco, nueve que venden seis y once que venden siete. Calcule lo siguiente:
### Tarea para la casa
### Resúmalo todo
Use la siguiente información para responder los próximos tres ejercicios: estamos interesados en el número de años que han vivido en California los estudiantes de una determinada clase de Estadística Elemental. La información de la siguiente tabla es de toda la sección.
|
# Estadística descriptiva
## Notación sigma y cálculo de la media aritmética
Fórmula de la media de la población
Fórmula de la media muestral
Esta unidad está aquí para recordarle el material que una vez estudió y en su momento dijo: “¡Estoy seguro de que nunca necesitaré esto!”.
Estas son las fórmulas de la media poblacional y de la media muestral. La letra griega μ es el símbolo de la media poblacional y es el símbolo de la media muestral. Ambas fórmulas tienen un símbolo matemático que nos indica cómo hacer los cálculos. Se llama notación Sigma porque el símbolo es la letra griega mayúscula sigma: Σ. Como todos los símbolos matemáticos nos dice lo que hay que hacer: igual que el signo más nos dice que hay que sumar y la x nos dice que hay que multiplicar. Se denominan operadores matemáticos. El símbolo Σ nos dice que hay que añadir una lista específica de números.
Supongamos que tenemos una muestra de animales del refugio de animales local y nos interesa su edad promedio. Si enumeramos cada valor, u observación, en una columna, se puede dar a cada uno un número de índice. El primer número será el número 1 y el segundo el número 2 y así sucesivamente.
Cada observación representa un animal concreto de la muestra. Purr es el animal número uno y es un gato de 9 años, Toto es el animal número 2 y es un cachorro de 1 año y así sucesivamente.
Para calcular la media, la fórmula nos dice que debemos sumar todos estos números, las edades en este caso, y luego dividir la suma entre 10, el número total de animales de la muestra.
El animal número uno, el gato Purr, se designa como X1, el animal número 2, Toto, se designa como X2 y así sucesivamente hasta Dundee que es el animal número 10 y se designa como X10.
La i de la fórmula nos indica cuál de las observaciones hay que sumar. En este caso es de X1 a X10 que son todos. Sabemos cuáles hay que añadir por la notación de indexación, la i = 1 y la n o N mayúscula de la población. Para este ejemplo la notación de indexación sería i = 1 y por tratarse de una muestra utilizamos una n pequeña en la parte superior del Σ que sería 10.
La desviación típica requiere el mismo operador matemático, por lo que sería útil recordar este conocimiento de su pasado.
La suma de las edades es de 78 y, dividiendo entre 10, la edad media de la muestra es de 7,8 años.
### Tarea para la casa
|
# Estadística descriptiva
## Media geométrica
La media (aritmética), la mediana y la moda son medidas del "centro" de los datos, la "media". Todos intentan, a su manera, medir el punto "común" dentro de los datos, el que es "normal". En el caso de la media aritmética esto se resuelve encontrando el valor del que todos los puntos están a igual distancia lineal. Podemos imaginar que todos los valores de los datos se combinan mediante la adición y luego se distribuyen a cada punto de datos en cantidades iguales. La suma de todos los valores es lo que se redistribuye en cantidades iguales de manera que la suma total sigue siendo la misma.
La media geométrica no redistribuye la suma de los valores, sino el producto de multiplicar todos los valores individuales y luego redistribuirlos en porciones iguales de manera que el producto total siga siendo el mismo. Esto se desprende de la fórmula de la media geométrica, : (Se dice “x tilde”)
donde es otro operador matemático, que nos dice que hay que multiplicar todos los números de la misma manera que la sigma griega mayúscula nos dice que sumemos todos los números . Recuerde que un exponente fraccionario pide la raíz enésima del número por lo que un exponente de 1/3 es la raíz cúbica del número.
La media geométrica responde a la pregunta "si todas las cantidades tuvieran el mismo valor, ¿cuál tendría que ser ese valor para conseguir el mismo producto?”. La media geométrica recibe su nombre del hecho de que cuando se redistribuye de esta manera los lados forman una forma geométrica en la que todos tienen la misma longitud. Para verlo, tomemos el ejemplo de los números 10, 51,2 y 8. La media geométrica es el producto de multiplicar estos tres números entre sí (4.096) y sacar la raíz cúbica porque son tres los números entre los que hay que repartir este producto. Por tanto, la media geométrica de estos tres números es 16. Esto describe un cubo de 16x16x16 y tiene un volumen de 4.096 unidades.
La media geométrica es relevante en Economía y Finanzas para tratar el crecimiento: el crecimiento de los mercados, de la inversión, de la población y de otras variables cuyo crecimiento interesa. Imagine que nuestra caja de 4096 unidades (quizás dólares) es el valor de una inversión al cabo de tres años y que los rendimientos de la inversión en porcentajes fueron los tres números de nuestro ejemplo. La media geométrica nos proporcionará la respuesta a la pregunta de cuál es la tasa promedio de rendimiento: 16 por ciento. La media aritmética de estas tres cifras es del 23,6 %. La razón de esta diferencia, 16 frente a 23,6, es que la media aritmética es aditiva y, por lo tanto, no tiene en cuenta el interés sobre el interés, el interés compuesto, implícito en el proceso de crecimiento de la inversión. La misma situación se plantea cuando se pregunta por la tasa promedio de crecimiento de una población o de las ventas o de la penetración en el mercado, etc., conociendo las tasas anuales de crecimiento. La fórmula de la tasa de rendimiento media geométrica, o de cualquier otra tasa de crecimiento, es:
Al manipular la fórmula de la media geométrica también se puede calcular la tasa promedio de crecimiento entre dos periodos conociendo solo el valor inicial y el valor final y el número de periodos, . La siguiente fórmula proporciona esta información:
Por último, observamos que la fórmula de la media geométrica requiere que todos los números sean positivos, mayores que cero. La razón, por supuesto, es que la raíz de un número negativo no está definida para su uso fuera de la teoría matemática. Sin embargo, hay formas de evitar este problema. En el caso de las tasas de rendimiento y otros problemas de crecimiento simples, podemos convertir los valores negativos en valores equivalentes positivos significativos. Imagine que los rendimientos anuales de los últimos tres años son del +12 %, –8 % y +2 %. El uso de los multiplicadores decimales equivalentes a 1,12, 0,92 y 1,02 nos permite calcular una media geométrica de 1,0167. Al restar 1 a este valor se obtiene la media geométrica de +1,67 % como tasa neta de crecimiento de la población (o rendimiento financiero). De este ejemplo se desprende que la media geométrica nos proporciona esta fórmula para calcular la tasa de rendimiento geométrica (media) de una serie de tasas de rendimiento anuales:
donde es la tasa promedio de rendimiento y es la media geométrica de los rendimientos durante un cierto número de periodos. Tenga en cuenta que la duración de cada periodo debe ser la misma.
Como regla general, hay que convertir los valores porcentuales en su equivalente decimal multiplicador. Es importante reconocer que cuando se trata de porcentajes, la media geométrica de los valores porcentuales no es igual a la media geométrica de los equivalentes del multiplicador decimal y es la media geométrica del multiplicador decimal la que es relevante.
### Revisión de la fórmula
La media geométrica
### Tarea para la casa
|
# Estadística descriptiva
## Distorsión y media, mediana y moda
Considere el siguiente conjunto de datos. 4; 5; 6; 6; 6; 7; 7; 7; 7; 7; 7; 8; 8; 8; 9; 10
Este conjunto de datos se puede representar mediante el siguiente histograma. Cada intervalo tiene un ancho de uno y cada valor se sitúa en el centro de un intervalo.
El histograma muestra una distribución simétrica de los datos. Una distribución es simétrica si se puede trazar una línea vertical en algún punto del histograma de manera que la forma a la izquierda y a la derecha de la línea vertical sean imágenes una espejo de la otra. La media, la mediana y la moda son siete para estos datos. En una distribución perfectamente simétrica, la media y la mediana son iguales. Este ejemplo tiene una moda (unimodal), y la moda es la misma que la media y la mediana. En una distribución simétrica que tiene dos modas (bimodal), las dos modas serían diferentes de la media y la mediana.
El histograma de los datos:
no es simétrico. El lado derecho parece “cortado” en comparación con el lado izquierdo. Una distribución de este tipo se denomina distorsionada a la izquierda porque se desplaza hacia la izquierda. Podemos medir formalmente la distorsión de una distribución del mismo modo que podemos medir matemáticamente el peso del centro de los datos o su "velocidad" general. La fórmula matemática de la distorsión es . Cuanto mayor sea la desviación con respecto a cero, mayor será el grado de distorsión. Si la distorsión es negativa, la distribución está distorsionada a la izquierda, como en la . Una medida positiva de la distorsión indica distorsionada a la derecha, como en la .
La media es 6,3, la mediana es 6,5 y la moda es siete. Observe que la media es menor que la mediana y ambas son menores que la moda. Tanto la media como la mediana reflejan la distorsión, pero la media lo refleja más.
El histograma de los datos:
, tampoco es simétrico. Es con distorsión a la derecha.
La media es 7,7, la mediana es 7,5 y la moda es siete. De las tres estadísticas, la media es la mayor, mientras que la moda es la menor. De nuevo, la media es la que más refleja la distorsión.
Para resumir, generalmente si la distribución de los datos está distorsionada a la izquierda, la media es menor que la mediana, que suele ser menor que la moda. Si la distribución de los datos está distorsionada a la derecha, la moda suele ser menor que la mediana, que es menor que la media.
Al igual que con la media, la mediana y la moda, y como veremos en breve, la varianza, existen fórmulas matemáticas que nos dan medidas precisas de estas características de la distribución de los datos. Volviendo a mirar la fórmula de la distorsión, vemos que se trata de una relación entre la media de los datos y las observaciones individuales al cubo.
donde es la desviación típica muestral de los datos, , y es la media aritmética y es el tamaño de la muestra.
Formalmente, la media aritmética se conoce como el primer momento de la distribución. El segundo momento que veremos es la varianza, y la distorsión es el tercer momento. La varianza mide las diferencias al cuadrado de los datos respecto a la media y la distorsión mide las diferencias al cubo de los datos respecto a la media. Mientras que una varianza nunca puede ser un número negativo, la medida de distorsión sí puede y así es como determinamos si los datos están distorsionados la derecha o a la izquierda. La distorsión de una distribución normal es cero, y cualquier dato simétrico debería tener una distorsión cercana a cero. Los valores negativos de la distorsión indican que los datos están sesgados hacia la izquierda y los valores positivos de la distorsión indican que los datos están sesgados hacia la derecha. Por izquierda distorsionada, queremos decir que la cola izquierda es larga en relación con la cola derecha. Del mismo modo, la derecha distorsionada significa que la cola derecha es larga en relación con la cola izquierda. La distorsión caracteriza el grado de asimetría de una distribución en torno a su media. Mientras que la media y la desviación típica son magnitudes dimensionales (por eso tomaremos la raíz cuadrada de la varianza) es decir, tienen las mismas unidades que las magnitudes medidas , la distorsión se define convencionalmente de forma que sea adimensional. Es un número puro que caracteriza únicamente la forma de la distribución. Un valor positivo de distorsión significa una distribución con una cola asimétrica que se extiende hacia un X más positiva y un valor negativo significa una distribución cuya cola se extiende hacia X más negativa. Una medida cero de distorsión indicará una distribución simétrica.
La distorsión y la simetría son importantes cuando hablemos de distribuciones de probabilidad en capítulos posteriores.
### Repaso del capítulo
Observar la distribución de los datos puede revelar mucho sobre la relación entre la media, la mediana y la moda. Hay tres tipos de distribuciones. Una distribución tiene una forma como la . Una distribución distorsionada a la derecha (o positiva) tiene una forma como la . Una distribución simétrica se parece a la .
### Revisión de la fórmula
Fórmula para la distorsión: Fórmula del coeficiente de variación
Use la siguiente información para responder los próximos tres ejercicios: Indique si los datos son simétricos, distorsionados a la izquierda o distorsionados a la derecha.
### Tarea para la casa
|
# Estadística descriptiva
## Medidas de la dispersión de los datos
Una característica importante de cualquier conjunto de datos es su variación. En algunos conjuntos de datos, los valores de los datos se concentran muy cerca de la media; en otros, están más dispersos de la media. La medida más común de variación, o dispersión, es la desviación típica. La desviación típica es un número que mide la distancia entre los valores de los datos y su media.
### La desviación típica
1. proporciona una medida numérica de la cantidad global de variación en un conjunto de datos y
2. se puede usar para determinar si un valor de datos determinado está cerca o lejos de la media.
### La desviación típica proporciona una medida de la variación global de un conjunto de datos
La desviación típica es siempre positiva o cero. La desviación típica es pequeña cuando todos los datos se concentran cerca de la media y muestran poca variación o dispersión. La desviación típica es mayor cuando los valores de los datos están más alejados de la media y muestran más variación.
Supongamos que estamos estudiando el tiempo que los clientes esperan en la fila de la caja del supermercado A y del supermercado B. El tiempo promedio de espera en ambos supermercados es de cinco minutos. En el supermercado A, la desviación típica del tiempo de espera es de dos minutos; en el supermercado B, la desviación típica del tiempo de espera es de cuatro minutos.
Como el supermercado B tiene una desviación típica más alta, sabemos que hay más variación en los tiempos de espera en el supermercado B. En general, los tiempos de espera en el supermercado B están más dispersos del promedio; los tiempos de espera en el supermercado A están más concentrados cerca del promedio.
### Cálculo de la desviación típica
Si x es un número, la diferencia "x menos la media" se denomina su deviación. En un conjunto de datos hay tantas desviaciones como elementos en el conjunto de datos. Las desviaciones se utilizan para calcular la desviación típica. Si los números pertenecen a una población, en símbolos una desviación es x – μ. Para los datos de la muestra, en símbolos una desviación es x –
.
El procedimiento para calcular la desviación típica depende de si los números son toda la población o son datos de una muestra. Los cálculos son similares, pero no idénticos. Por tanto, el símbolo utilizado para representar la desviación típica depende de si se calcula a partir de una población o de una muestra. La letra minúscula s representa la desviación típica de la muestra y la letra griega σ (sigma, minúscula) representa la desviación típica de la población. Si la muestra tiene las mismas características que la población, entonces s debería ser una buena estimación de σ.
Para calcular la desviación típica, tenemos que calcular primero la varianza. La varianza es el promedio de los cuadrados de las desviaciones (la x –
para una muestra, o los valores x – μ para una población). El símbolo σ2 representa la varianza de la población; la desviación típica de la población σ es la raíz cuadrada de la varianza de la población. El símbolo s2 representa la varianza de la muestra; la desviación típica de la muestra s es la raíz cuadrada de la varianza de la muestra. Puede pensar en la desviación típica como un promedio especial de las desviaciones. Formalmente, la varianza es el segundo momento de la distribución o el primer momento alrededor de la media. Recuerde que la media es el primer momento de la distribución.
Si las cifras proceden de un censo de toda la población y no de una muestra, cuando calculamos el promedio de las desviaciones al cuadrado para hallar la varianza, dividimos entre N, el número de elementos de la población. Si los datos proceden de una muestra y no de una población, al calcular el promedio de las desviaciones al cuadrado, dividimos entre , uno menos que el número de elementos de la muestra.
### Fórmulas para la desviación típica de la muestra
1.
o
o
2. Para la desviación típica de la muestra, el denominador es , es decir, el tamaño de la muestra menos 1.
### Fórmulas para la desviación típica de la población
1.
o
o
2. Para la desviación típica de la población el denominador es N, el número de elementos de la población.
En estas fórmulas, f representa la frecuencia con la que aparece un valor. Por ejemplo, si un valor aparece una vez, f es uno. Si un valor aparece tres veces en el conjunto de datos o población, f es tres. Dos observaciones importantes sobre la varianza y la desviación típica: las desviaciones se miden a partir de la media y las desviaciones se elevan al cuadrado. En principio, las desviaciones podrían medirse desde cualquier punto, sin embargo, nuestro interés es la medición desde el peso central de los datos, lo que es el valor "normal" o más habitual de la observación. Más adelante trataremos de medir lo “inusual" de una observación o de una media muestral y, por tanto, necesitamos una medida a partir de la media. La segunda observación es que las desviaciones son al cuadrado. Esto tiene dos efectos: primero, hace que las desviaciones sean todas positivas y segundo, cambia las unidades de medida de la media y de las observaciones originales. Si los datos son pesos, la media se mide en libras, pero la varianza se mide en libras al cuadrado. Una de las razones para utilizar la desviación típica es volver a las unidades de medida originales tomando la raíz cuadrada de la varianza. Además, cuando las desviaciones se elevan al cuadrado su valor aumenta en gran medida. Por ejemplo, una desviación de 10 de la media al cuadrado es 100, pero una desviación de 100 de la media es 10.000. Lo que hace esto es dar un gran peso a los valores atípicos al calcular la varianza.
### Tipos de variabilidad en las muestras
Cuando se trata de estudiar una población, a menudo se utiliza una muestra, ya sea por conveniencia o porque no es posible acceder a toda la población. La variabilidad es el término utilizado para describir las diferencias que pueden darse en estos resultados. Los tipos de variabilidad más comunes son los siguientes:
1. Variabilidad de observación o de medición
2. Variabilidad natural
3. Variabilidad inducida
4. Variabilidad de la muestra
He aquí algunos ejemplos para describir cada tipo de variabilidad.
Ejemplo 1: Variabilidad de la mediciónLa variabilidad de la medición se produce cuando hay diferencias en los instrumentos utilizados para medir o en las personas que utilizan esos instrumentos. Si recopilamos datos sobre el tiempo que tarda una pelota en caer desde una altura haciendo que los estudiantes midan el tiempo de la caída con un cronómetro, podemos experimentar una variabilidad en la medición si los dos cronómetros utilizados son de diferentes fabricantes: Por ejemplo, un cronómetro mide al segundo más cercano, mientras que el otro mide a la décima de segundo más cercana. También podemos experimentar la variabilidad de las mediciones porque dos personas diferentes recopilan los datos. Sus tiempos de reacción al pulsar el botón del cronómetro pueden ser diferentes, por lo que los resultados variarán en consecuencia. Las diferencias en los resultados pueden verse afectadas por la variabilidad de las mediciones.
Ejemplo 2: Variabilidad naturalLa variabilidad natural surge de las diferencias que se producen de forma natural porque los miembros de una población difieren entre sí. Por ejemplo, si tenemos dos plantas de maíz idénticas y las exponemos a la misma cantidad de agua y luz solar, pueden crecer a ritmos diferentes simplemente porque son dos plantas de maíz diferentes. La diferencia de resultados puede explicarse por la variabilidad natural.
Ejemplo 3: Variabilidad inducidaLa variabilidad inducida es la contrapartida de la variabilidad natural; se produce porque hemos inducido artificialmente un elemento de variación (que, por definición, no estaba presente de forma natural): Por ejemplo, asignamos personas a dos grupos diferentes para estudiar la memoria, e inducimos una variable en un grupo limitando la cantidad de sueño que tienen. La diferencia de resultados puede verse afectada por la variabilidad inducida.
Ejemplo 4: Variabilidad de la muestraLa variabilidad de la muestra se produce cuando se toman varias muestras aleatorias de la misma población. Por ejemplo, si se realizan cuatro encuestas a 50 personas seleccionadas al azar de una población determinada, las diferencias en los resultados pueden verse afectadas por la variabilidad de la muestra.
### Explicación del cálculo de la desviación típica que aparece en la tabla
Las desviaciones muestran la dispersión de los datos respecto a la media. El valor de los datos 11,5 está más alejado de la media que el valor de los datos 11, lo que se indica con las desviaciones 0,97 y 0,47. Una desviación positiva se produce cuando el valor de los datos es mayor que la media, mientras que una desviación negativa se produce cuando el valor de los datos es menor que la media. La desviación es de –1,525 para el noveno valor de los datos. Si se suman las desviaciones, la suma es siempre cero (según el , hay n = 20 desviaciones). Por lo tanto, no se puede simplemente sumar las desviaciones para obtener la dispersión de los datos. Al elevar al cuadrado las desviaciones se convierten en números positivos, y la suma también será positiva. La varianza, por tanto, es la desviación promedio al cuadrado. Al elevar al cuadrado las desviaciones, estamos penalizando en extremo las observaciones que se alejan de la media; estas observaciones tienen mayor peso en los cálculos de la varianza. Más adelante veremos que la varianza (desviación típica) desempeña un papel fundamental para determinar nuestras conclusiones en la estadística inferencial. Podemos empezar ahora utilizando la desviación típica como medida de lo "inusual": "¿Cómo te fue en el examen?" "¡Fantástico! Dos desviaciones típicas por encima de la media". Esto, como veremos, es una nota de examen excepcionalmente buena.
La varianza es una medida al cuadrado y no tiene las mismas unidades que los datos. Calcular la raíz cuadrada resuelve el problema. La desviación típica mide la dispersión en las mismas unidades que los datos.
Observe que en vez de dividir entre n = 20, el cálculo divide entre n – 1 = 20 – 1 = 19 porque los datos son una muestra. Para la varianza de la muestra, se divide entre el tamaño de la muestra menos uno (n – 1). ¿Por qué no dividir entre n? La respuesta tiene que ver con la varianza de la población. La varianza de la muestra es una estimación de la varianza de la población. Esta estimación nos obliga a utilizar una cifra estimada de la media de la población en lugar de la media real de la población. Basándose en la matemática teórica que hay detrás de estos cálculos, al dividir entre (n – 1) da una mejor estimación de la varianza de la población.
La desviación típica, s o σ, es cero o mayor que cero. La descripción de los datos con referencia a la dispersión se denomina “variabilidad”. La variabilidad de los datos depende del método con el que se obtienen los resultados; por ejemplo, por medición o por muestreo aleatorio. Cuando la desviación típica es cero, no hay dispersión; es decir, todos los valores de los datos son iguales entre sí. La desviación típica es pequeña cuando todos los datos se concentran cerca de la media, y es mayor cuando los valores de los datos muestran más variación con respecto a la media. Cuando la desviación típica es mucho mayor que cero, los valores de los datos están muy dispersos alrededor de la media; los valores atípicos pueden hacer que s o σ sean muy grandes.
### Desviación típica de las tablas de frecuencia agrupadas
Recordemos que para los datos agrupados no conocemos los valores individuales de los datos, por lo que no podemos describir el valor típico de los datos con precisión. En otras palabras, no podemos hallar la media, la mediana ni la moda exactas. Sin embargo, podemos determinar la mejor estimación de las medidas de centro al hallar la media de los datos agrupados con la fórmula donde
frecuencias de intervalo y m = puntos medios del intervalo.
Al igual que no podemos hallar la media exacta, tampoco podemos hallar la desviación típica exacta. Recuerde que la desviación típica describe numéricamente la desviación esperada que tiene un valor de datos con respecto a la media. En términos sencillos, la desviación típica nos permite comparar lo “inusual” que son los datos individuales en comparación con la media.
### Comparación de valores de diferentes conjuntos de datos
La desviación típica es útil cuando se comparan valores de datos que provienen de diferentes conjuntos de datos. Si los conjuntos de datos tienen medias y desviaciones típicas diferentes, la comparación directa de los valores de los datos puede ser engañosa.
1. Para cada valor de los datos x, calcule a cuántas desviaciones típicas de su media se encuentra el valor.
2. Utilice la fórmula: x = media + (n.º de STDEV)(de STandard DEViation o desviación típica); resuelva para n.º de STDEV.
3.
4. Compare los resultados de este cálculo.
N.º de STDEV suele llamarse “puntuación z”; podemos utilizar el símbolo z. En símbolos, las fórmulas se convierten en:
Las siguientes listas ofrecen algunos hechos que proporcionan un poco más de información sobre lo que la desviación típica nos dice sobre la distribución de los datos.
2. Al menos el 75 % de los datos están dentro de las dos desviaciones típicas de la media.
3. Al menos el 89 % de los datos están dentro de las tres desviaciones típicas de la media.
4. Al menos el 95 % de los datos están dentro de 4,5 desviaciones típicas de la media.
5. Esto se conoce como la regla de Chebyshev.
2. Aproximadamente el 68 % de los datos están dentro de una desviación típica de la media.
3. Aproximadamente el 95 % de los datos están dentro de las dos desviaciones típicas de la media.
4. Más del 99 % de los datos están dentro de las tres desviaciones típicas de la media.
5. Esto se conoce como la regla empírica.
6. Es importante señalar que esta regla solo se aplica cuando la forma de la distribución de los datos tiene forma de campana y es simétrica. Aprenderemos más sobre esto cuando estudiemos la distribución de probabilidad “normal” o “gaussiana” en capítulos posteriores.
### Coeficiente de variación
Otra forma útil de comparar distribuciones, además de las simples comparaciones de medias o desviaciones típicas, es ajustar las diferencias en la escala de los datos que se miden. Sencillamente, una gran variación en los datos con una media grande es diferente a la misma variación en los datos con una media pequeña. Para ajustar la escala de los datos subyacentes se ha desarrollado el coeficiente de variación (CV). Matemáticamente, el:
Podemos ver que esto mide la variabilidad de los datos subyacentes como un porcentaje del valor medio; el peso central del conjunto de datos. Esta medida es útil para comparar el riesgo cuando se justifica un ajuste debido a las diferencias de escala de dos conjuntos de datos. En efecto, la escala se cambia a escala común, diferencias porcentuales, y permite la comparación directa de las dos o más magnitudes de variación de diferentes conjuntos de datos.
### Referencias
Datos de Microsoft Bookshelf.
King, Bill.“Graphically Speaking”. Institutional Research, Lake Tahoe Community College. Disponible en línea en http://www.ltcc.edu/web/about/institutional-research (consultado el 3 de abril de 2013).
### Repaso del capítulo
La desviación típica puede ayudarlo a calcular la dispersión de los datos. Existen diferentes ecuaciones para calcular la desviación típica de una muestra o de una población.
1. La desviación típica nos permite comparar numéricamente datos individuales o clases con la media del conjunto de datos.
2. s =
o s =
es la fórmula para calcular la desviación típica de una muestra. Para calcular la desviación típica de una población usaríamos la media de la población, μ, y la fórmula σ =
o σ =
.
### Revisión de la fórmula
donde
Fórmulas para la desviación típica de la muestra
o
o Para la desviación típica de la muestra, el denominador es , es decir, el tamaño de la muestra – 1.
Fórmulas para la desviación típica de la población
o
o Para la desviación típica de la población, el denominador es N, el número de elementos de la población.
Use la siguiente información para responder los próximos dos ejercicios: Los siguientes datos son las distancias entre 20 tiendas minoristas y un gran centro de distribución. Las distancias están en millas. 29; 37; 38; 40; 58; 67; 68; 69; 76; 86; 87; 95; 96; 96; 99; 106; 112; 127; 145; 150
Calcule la desviación típica de las siguientes tablas de frecuencias utilizando la fórmula. Compruebe los cálculos con la TI 83/84.
### Tarea para la casa
Utilice la siguiente información para responder a los siguientes nueve ejercicios: Los parámetros de población que aparecen a continuación describen el número de estudiantes equivalentes a tiempo completo (full-time equivalent number of students, FTES) cada año en el Lake Tahoe Community College desde 1976-1977 hasta 2004-2005.
1. μ = 1.000 FTES
2. mediana = 1.014 FTES
3. σ = 474 FTES
4. primer cuartil = 528,5 FTES
5. tercer cuartil = 1.447,5 FTES
6. n = 29 años
### Resúmalo todo
Use la siguiente información para responder los próximos dos ejercicios. X = el número de días a la semana que 100 clientes utilizan un determinado centro de ejercicio. |
# Temas de probabilidad
## Introducción
A menudo es necesario “estimar” el resultado de un evento para tomar una decisión. Los políticos estudian los sondeos para estimar sus posibilidades de ganar unas elecciones. Los maestros eligen un curso de estudio particular con base en lo que creen que los estudiantes pueden comprender. Los médicos eligen los tratamientos necesarios para las distintas enfermedades con base en su evaluación de los resultados probables. Es posible que haya visitado un casino en el que las personas participan en juegos elegidos por la creencia de que la probabilidad de ganar es buena. Es posible que haya elegido sus estudios según la probable disponibilidad de trabajo.
Es más que posible que haya utilizado la probabilidad. De hecho, posiblemente tenga un sentido intuitivo de la probabilidad. La probabilidad se refiere a la posibilidad de que se produzca un evento. Cada vez que sopesa las probabilidades de hacer o no la tarea para la casa o de estudiar para un examen está utilizando la probabilidad. En este capítulo aprenderá a resolver problemas de probabilidad mediante un enfoque sistemático. |
# Temas de probabilidad
## Terminología
La probabilidad es una medida asociada a la certeza de los resultados de un determinado experimento o actividad. Un experimento es una operación planificada que se realiza en condiciones controladas. Si el resultado no está predeterminado, se dice que el experimento es fortuito. Lanzar una moneda imparcial dos veces es un ejemplo de experimento.
El producto de un experimento se llama resultado. El espacio muestral de un experimento es el conjunto de todos los resultados posibles. Tres formas de representar un espacio muestral son: hacer una lista de los posibles resultados, crear un diagrama de árbol o crear un diagrama de Venn. La letra S mayúscula se utiliza para denotar el espacio muestral. Por ejemplo, si se lanza una moneda imparcial, S = {H, T} donde H = cara y T = cruz son los resultados.
Un evento es cualquier combinación de resultados. Las letras mayúsculas como A y B representan eventos. Por ejemplo, si el experimento consiste en lanzar una moneda imparcial, el evento A podría obtener como máximo una cara. La probabilidad de un evento A se escribe P(A).
La probabilidad de cualquier resultado es la frecuencia relativa a largo plazo de ese resultado. Las probabilidades están comprendidas entre el cero y el uno, ambos inclusive (es decir, el cero y el uno y todos los números entre estos valores). P(A) = 0 significa que el evento A no puede ocurrir nunca. P(A) = 1 significa que el evento A siempre ocurre. P(A) = 0,5 significa que el evento A tiene la misma probabilidad de ocurrir que de no ocurrir. Por ejemplo, si se lanza una moneda imparcial repetidamente (de 20 a 2.000 a 20.000 veces) la frecuencia relativa de caras se acerca a 0,5 (la probabilidad de cara).
Igual de probable significa que cada resultado de un experimento ocurre con igual probabilidad. Por ejemplo, si se lanza un dado imparcial de seis lados, cada lado (1, 2, 3, 4, 5 o 6) tiene la misma probabilidad de caer que cualquier otro. Si se lanza una moneda imparcial, hay la misma probabilidad de que salga cara (H) que de que salga cruz (T). Si estima al azar la respuesta a una pregunta de verdadero-falso en un examen, tiene la misma probabilidad de seleccionar una respuesta correcta o una incorrecta.
Para calcular la probabilidad de un evento , cuente el número de resultados del evento A y divídalo entre el número total de resultados del espacio muestral. Por ejemplo, si se lanza una moneda imparcial de diez centavos y una moneda justa de cinco centavos, el espacio muestral es {HH, TH, HT, TT} donde T = cruz y H = cara. El espacio muestral tiene cuatro resultados. A = obtener una cara. Hay dos resultados que cumplen esta condición {HT, TH}, por lo que P(A) =
= 0,5.
Supongamos que lanza un dado imparcial de seis lados, con los números {1, 2, 3, 4, 5, 6} en sus lados. Supongamos que el evento E = lanzar un número que sea al menos cinco. Hay dos resultados {5, 6}. P(E) =
. Si lanzara el dado solo unas pocas veces, no se sorprendería si los resultados observados no coinciden con la probabilidad. Si se lanzara el dado un gran número de veces, se esperaría eso, en general, de las lanzadas daría un resultado de “al menos cinco”. No se puede esperar exactamente . La frecuencia relativa a largo plazo de obtener este resultado se acerca a la probabilidad teórica de a medida que el número de repeticiones aumenta.
Esta importante característica de los experimentos probabilísticos se conoce como la ley de los grandes números, que establece que, a medida que aumenta el número de repeticiones de un experimento, la frecuencia relativa obtenida tiende a acercarse cada vez más a la probabilidad teórica. Aunque los resultados no se produzcan según un patrón u orden determinado, en general, la frecuencia relativa observada a largo plazo se acerca a la probabilidad teórica (a menudo se utiliza la palabra empírica en vez de la palabra observado).
Es importante darse cuenta de que, en muchas situaciones, los resultados no son igualmente probables. Una moneda o un dado pueden ser desiguales o sesgados. Dos profesores de Matemáticas de Europa hicieron que sus estudiantes de Estadística probaran la moneda belga de un euro y descubrieron que, en 250 ensayos, se obtenía una cara el 56 % de las veces y una cruz el 44 %. Los datos parecen mostrar que la moneda no es imparcial; más repeticiones serían útiles para obtener una conclusión más precisa sobre dicho sesgo. Algunos dados pueden estar sesgados. Observe los dados de un juego que tenga en casa; los puntos de cada lado suelen ser pequeños agujeros tallados y luego pintados para que sean visibles. Sus dados pueden o no estar sesgados; es posible que los resultados se vean afectados por las ligeras diferencias de peso debido al diferente número de agujeros en las caras. Los casinos ganan mucho dinero dependiendo de los resultados de los dados, por lo que los dados de los casinos se fabrican de forma diferente para eliminar el sesgo. Los dados de casino tienen lados planos; los agujeros se rellenan completamente con pintura de la misma densidad que el material del que están hechos los dados, de modo que cada cara tiene la misma probabilidad de ocurrir. Más adelante aprenderemos técnicas para trabajar con probabilidades para eventos que no son igualmente probables.
"Un resultado es en el caso A B si el resultado está en A o está en B o está tanto en A como en B. Por ejemplo, supongamos que A = {1, 2, 3, 4, 5} y B = {4, 5, 6, 7, 8}. A B = {1, 2, 3, 4, 5, 6, 7, 8}. Observe que el 4 y el 5 NO aparecen dos veces en la lista.
"Un resultado es en el caso A B si el resultado está en A y B al mismo tiempo. Por ejemplo, que A y B sean {1, 2, 3, 4, 5} y {4, 5, 6, 7, 8}, respectivamente. Entonces A B = {4, 5}.
El complemento del evento A se denomina A′ (léase “A prima”). A′ consiste en todos los resultados que NO están en A. Observe que P(A) + P(A′) = 1. Por ejemplo, supongamos que S = {1, 2, 3, 4, 5, 6} y que A = {1, 2, 3, 4}. Entonces, A′ = {5, 6}. P(A) = , P(A′) = y P(A) + P(A′) =
= 1
La probabilidad condicional de A dada B se escribe P(AB). P(AB) es la probabilidad de que ocurra el evento A dado que el evento B ya ha ocurrido. Un condicional reduce el espacio muestral. Calculamos la probabilidad de A a partir del espacio muestral reducido B. La fórmula para calcular P(AB) es P(AB) =
donde P(B) es mayor que cero.
Por ejemplo, supongamos que lanzamos un dado imparcial de seis lados. El espacio muestral S = {1, 2, 3, 4, 5, 6}. Supongamos que A = el lado es 2 o 3 y B = el lado es par (2, 4, 6). Para calcular P(AB), contamos el número de resultados 2 o 3 en el espacio muestral B = {2, 4, 6}. Luego lo dividimos entre el número de resultados B (en vez de S).
Obtenemos el mismo resultado utilizando la fórmula. Recuerde que S tiene seis resultados.
P(AB) =
PosibilidadLas probabilidades de un evento presentan la probabilidad como un cociente entre el éxito y el fracaso. Esto es común en varios formatos de juego. Matemáticamente, la posibilidad de un evento se define como:
donde P(A) es la probabilidad de éxito y, por supuesto, 1 − P(A) es la probabilidad de fracaso. La posibilidad se expresa siempre como "numerador a denominador", por ejemplo: 2 a 1. En este caso, la probabilidad de ganar es el doble de la de perder; por ende, la probabilidad de ganar es de 0,66. Un 0,60 en la probabilidad de ganar generaría la posibilidad a favor de ganar de 3 a 2. Aunque el cálculo de la posibilidad pudiera servir en los locales de juegos de azar para determinar el monto del pago, es inútil para entender ni la probabilidad ni la teoría estadística.
Entender la terminología y los símbolosEs importante leer detenidamente cada problema para reflexionar y comprender los eventos. Entender el enunciado es el primer paso muy importante para resolver problemas de probabilidad. Vuelva a leer el problema varias veces si es necesario. Identifique claramente el evento de interés. Determine si hay una condición establecida en el enunciado que indique que la probabilidad es condicional; identifique cuidadosamente la condición, si la hay.
### Referencias
“Lista de países por continente”. Worldatlas, 2013. Disponible en línea en http://www.worldatlas.com/cntycont.htm (consultado el 2 de mayo de 2013).
### Repaso del capítulo
En este módulo hemos aprendido la terminología básica de la probabilidad. El conjunto de todos los resultados posibles de un experimento se denomina espacio muestral. Los eventos son subconjuntos del espacio muestral y se les asigna una probabilidad que es un número entre cero y uno, ambos inclusive.
### Revisión de la fórmula
A y B son eventos
P(S) = 1 donde S es el espacio muestral
0 ≤ P(A) ≤ 1
P(AB) =
Use la siguiente información para responder los próximos cuatro ejercicios. Una caja está llena de varios regalos de fiesta. Contiene 12 sombreros, 15 pitos, diez trampas para dedos y cinco bolsas de confeti. Se elegirá al azar un regalo de fiesta de la caja. Supongamos que H = el evento de sacar un sombrero. Supongamos que N = el evento de sacar un pito. Supongamos que F = el evento de sacar una trampa para dedos. Supongamos que C = el evento de sacar una bolsa de confeti.
Use la siguiente información para responder los próximos seis ejercicios. Una jarra de 150 gominolas contiene 22 rojas, 38 amarillas, 20 verdes, 28 moradas, 26 azules y el resto son anaranjadas. Se saca de la caja una gominola al azar. Supongamos que B = el evento de sacar una gominola azul. Supongamos que G = el evento de sacar una gominola verde. Supongamos que O = el evento de sacar una gominola anaranjada. Supongamos que P = el evento de sacar una gominola morada. Supongamos que R = el evento de sacar una gominola roja. Supongamos que Y = el evento de sacar una gominola amarilla.
Use la siguiente información para responder los próximos seis ejercicios. Hay 23 países en América del Norte, 12 en América del Sur, 47 en Europa, 44 en Asia, 54 en África y 14 en Oceanía (región del Océano Pacífico). Supongamos que A = el evento en el que un país esté en Asia. Supongamos que E = el evento en el que un país esté en Europa. Supongamos que F = el evento en el que un país esté en África. Supongamos que N = el evento en el que un país esté en América del Norte. Supongamos que O = el evento en el que un país esté en Oceanía. Supongamos que S = el evento en el que un país esté en América del Sur.
Use la siguiente información para responder los próximos dos ejercicios. Usted ve un juego en una feria local. Tiene que lanzar un dardo a una rueda de colores. Cada sección de la rueda de color es de igual área.
Supongamos que B = el evento de acertar al azul. Supongamos que R = el evento de acertar al rojo. Supongamos que G = el evento de acertar al verde. Supongamos que Y = el evento de acertar al amarillo.
Use la siguiente información para responder los próximos diez ejercicios. En un equipo de béisbol, hay jugadores de campo y jardineros. Algunos jugadores son grandes bateadores y otros no. Supongamos que I = el evento en el que un jugador es un jugador de campo. Supongamos que O = el evento en el que un jugador sea jardinero. Supongamos que H = el evento en el que un jugador sea un gran bateador. Supongamos que N = el evento en el que un jugador no sea un gran bateador.
Use la siguiente información para responder los próximos dos ejercicios. Usted está lanzando un cubo numérico imparcial de seis lados. Supongamos que E = el evento en el que caiga en un número par. Supongamos que M = el evento en el que caiga en un múltiplo de tres.
### Tarea para la casa
|
# Temas de probabilidad
## Eventos mutuamente excluyentes e independientes
Independiente y mutuamente excluyente no significan lo mismo.
### Eventos independientes
Dos eventos son independientes si uno de los siguientes es cierto:
1.
2.
3.
Dos eventos A y B son independientes si el conocimiento de que uno ha ocurrido no afecta la posibilidad de que ocurra el otro. Por ejemplo, los resultados de lanzar dos veces un dado imparcial son eventos independientes. El resultado de la primera lanzada no cambia la probabilidad del resultado de la segunda. Para demostrar que dos eventos son independientes, debe mostrar solo una de las condiciones anteriores. Si dos eventos NO son independientes, decimos que son dependientes.
El muestreo se puede hacer con reemplazo o sin reemplazo.
1. Con reemplazo: si cada miembro de una población es reemplazado después de ser elegido, entonces ese miembro tiene la posibilidad de ser elegido más de una vez. Cuando el muestreo se hace con reemplazo, los eventos se consideran independientes, lo que significa que el resultado de la primera elección no cambiará las probabilidades de la segunda.
2. Sin reemplazo: cuando el muestreo se hace sin reemplazo, cada miembro de una población solo lo pueden seleccionar una vez. En este caso, las probabilidades de la segunda elección se ven afectadas por el resultado de la primera. Los eventos se consideran dependientes o no independientes.
Si no se sabe si A y B son independientes o dependientes, suponga que son dependientes hasta que pueda demostrar lo contrario.
### Eventos mutuamente excluyentes
A y B son eventos mutuamente excluyentes si no pueden ocurrir al mismo tiempo. Dicho de otra manera, si A ocurrió entonces B no puede ocurrir y viceversa. Esto significa que A y B no comparten ningún resultado y
.
Por ejemplo, supongamos que el espacio muestral S = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}. Supongamos que A = {1, 2, 3, 4, 5}, B = {4, 5, 6, 7, 8}, y C = {7, 9}. A B = {4, 5}.
y no es igual a cero. Por lo tanto, A y B no son mutuamente excluyentes. A y C no tienen ningún número en común por lo que
. Por lo tanto, A y C son mutuamente excluyentes.
Si no se sabe si A y B son mutuamente excluyentes, suponga que no lo son hasta que pueda demostrar lo contrario. Los siguientes ejemplos ilustran estas definiciones y términos.
### Referencias
Lopez, Shane, Preety Sidhu. “U.S. Teachers Love Their Lives, but Struggle in the Workplace”. Gallup Wellbeing, 2013. http://www.gallup.com/poll/161516/teachers-love-lives-struggle-workplace.aspx (consultado el 2 de mayo de 2013).
Datos de Gallup. Disponible en línea en www.gallup.com/ (consultado el 2 de mayo de 2013).
### Repaso del capítulo
Dos eventos A y B son independientes si el conocimiento de que uno ha ocurrido no afecta a la posibilidad de que ocurra el otro. Si dos eventos no son independientes, decimos que son dependientes.
En el muestreo con reemplazo, cada miembro de una población se sustituye después de que lo seleccionen, por lo que ese miembro tiene la posibilidad de que lo seleccionen más de una vez, y los eventos se consideran independientes. En el muestreo sin reemplazo, cada miembro de una población solo lo pueden seleccionar una vez, y se considera que los eventos no son independientes. Cuando los eventos no comparten resultados, son mutuamente excluyentes.
### Revisión de la fórmula
Si y son independientes, y
Si y son mutuamente excluyentes, y
### Tarea para la casa
Use la siguiente información para responder los próximos 12 ejercicios. Entre enero y diciembre de 2012, Gallup entrevistó a más de 170.000 estadounidenses de 18 años o más con empleo. El gráfico que se muestra está elaborado a partir de datos recogidos por Gallup. Las calificaciones del Índice de Salud Emocional son el espacio muestral. Tomamos una muestra aleatoria de la calificación del Índice de Salud Emocional.
### Resúmalo todo
|
# Temas de probabilidad
## Dos reglas básicas de la probabilidad
Al calcular la probabilidad, hay que tener en cuenta dos reglas para determinar si dos eventos son independientes o dependientes y si son mutuamente excluyentes o no.
### La regla de multiplicación
Si A y B son dos eventos definidos en un espacio muestral, entonces . Podemos pensar que el símbolo de intersección sustituye a la palabra "y".
Esta regla también puede escribirse como:
Esta ecuación se lee como la probabilidad de A dado que B es igual a la probabilidad de A y B dividido entre la probabilidad de B.
Si A y B son independientes, entonces
. Entonces
se convierte en
porque el
si A y B son independientes.
Una forma fácil de recordar la regla de la multiplicación es que la palabra "y" significa que el evento tiene que satisfacer dos condiciones. Por ejemplo, el nombre extraído de la lista de la clase debe ser tanto una mujer como un estudiante de segundo año. Es más difícil satisfacer dos condiciones que una sola y, por supuesto, cuando multiplicamos fracciones el resultado es siempre menor. Esto refleja la creciente dificultad de satisfacer dos condiciones.
### La regla de adición
Si A y B están definidos en un espacio muestral, entonces
. Podemos pensar que el símbolo de la unión sustituye a la palabra "o". La razón por la que restamos la intersección de A y B es para no contar dos veces los elementos que están en A y B.
Si A y B se excluyen mutuamente, entonces
. Entonces
se convierte en
.
### Referencias
DiCamillo, Mark, Mervin Field. “The File Poll”. Field Research Corporation. Disponible en línea en http://www.field.com/fieldpollonline/subscribers/Rls2443.pdf (consultado el 2 de mayo de 2013).
Rider, David, “Ford support plumming, poll suggests”, The Star, 14 de septiembre de 2011. Disponible en línea en http://www.thestar.com/news/gta/2011/09/14/ford_support_plummeting_poll_suggests.html (consultado el 2 de mayo de 2013).
“Mayor’s Approval Down”. News Release by Forum Research Inc. Disponible en línea en http://www.forumresearch.com/forms/News Archives/News Releases/74209_TO_Issues_-_Mayoral_Approval_%28Forum_Research%29 %2820130320 %29.pdf (consultado el 2 de mayo de 2013).
“Roulette”. Wikipedia. Disponible en línea en http://en.wikipedia.org/wiki/Roulette (consultado el 2 de mayo de 2013).
Shin, Hyon B., Robert A. Kominski. “Language Use in the United States: 2007.” Oficina del Censo de Estados Unidos. Disponible en línea en http://www.census.gov/hhes/socdemo/language/data/acs/ACS-12.pdf (consultado el 2 de mayo de 2013).
Datos del Baseball-Almanac, 2013. Disponible en línea en www.baseball-almanac.com (consultado el 2 de mayo de 2013).
Datos de la Oficina del Censo de EE. UU.
Datos del Wall Street Journal.
Datos The Roper Center: Public Opinion Archives at the University of Connecticut. Disponible en línea en http://www.ropercenter.uconn.edu/ (consultado el 2 de mayo de 2013).
Datos de Field Research Corporation. Disponible en línea en www.field.com/fieldpollonline (consultado el 2 de mayo de 2013).
### Repaso del capítulo
Las reglas de multiplicación y de adición se utilizan para calcular la probabilidad de A y B, así como la probabilidad de A o B para dos eventos dados A, B definidos en el espacio muestral. En el muestreo con reemplazo, cada miembro de una población se sustituye después de ser elegido, por lo que ese miembro tiene la posibilidad de ser elegido más de una vez, y los eventos se consideran independientes. En el muestreo sin reemplazo, cada miembro de una población solo lo pueden seleccionar una vez, y se considera que los eventos no son independientes. A y B son eventos mutuamente excluyentes cuando no tienen ningún resultado en común.
### Revisión de la fórmula
La regla de multiplicación: P(A B) = P(AB)P(B)
La regla de adición: P(A B) = P(A) + P(B) – P(A B)
Use la siguiente información para responder los próximos diez ejercicios. El cuarenta y ocho por ciento de todos los californianos votantes registrados prefieren la cadena perpetua sin libertad condicional a la pena de muerte para una persona condenada por asesinato en primer grado. Entre los votantes latinos registrados en California, el 55 % prefiere la cadena perpetua sin libertad condicional a la pena de muerte para una persona condenada por asesinato en primer grado. El 37,6 % de los californianos son latinos.
En este problema supongamos que:
Supongamos que se selecciona al azar un californiano.
### Tarea para la casa
Use la siguiente información para responder los próximos tres ejercicios. El juego de casino, la ruleta, le permite al jugador apostar sobre la probabilidad de que una bola que gira en la rueda de la ruleta caiga en un color, número o rango de números particulares. La tabla utilizada para realizar las apuestas contiene 38 números, y cada número se asigna a un color y a un rango. |
# Temas de probabilidad
## Tablas de contingencia y árboles de probabilidad
### Tablas de contingencia
Una tabla de contingencia proporciona una forma de representar los datos que puede facilitar el cálculo de probabilidades. La tabla ayuda a determinar las probabilidades condicionales con bastante facilidad. La tabla muestra los valores de la muestra en relación con dos variables diferentes que pueden ser dependientes o contingentes entre sí. Más adelante volveremos a utilizar las tablas de contingencia, pero de otra manera.
### Diagramas de árbol
A veces, cuando los problemas de probabilidad son complejos, puede ser útil hacer un gráfico de la situación. Los diagramas de árbol pueden utilizarse para visualizar y resolver las probabilidades condicionales.
### Diagramas de árbol
Un diagrama de árbol es un tipo especial de gráfico utilizado para determinar los resultados de un experimento. Consta de “ramas” que se identifican con frecuencias o probabilidades. Los diagramas de árbol pueden hacer que algunos problemas de probabilidad sean más fáciles de visualizar y resolver. El siguiente ejemplo ilustra cómo utilizar un diagrama de árbol.
### Referencias
“Blood Types”. American Red Cross, 2013. Disponible en línea en http://www.redcrossblood.org/learn-about-blood/blood-types (consultado el 3 de mayo de 2013).
Datos del Centro Nacional de Estadísticas de Salud, que forma parte del Departamento de Salud y Servicios Humanos de Estados Unidos.
Datos del Senado de Estados Unidos. Disponible en línea en www.senate.gov (consultado el 2 de mayo de 2013).
“Human Blood Types”. Unite Blood Services, 2011. Disponible en línea en http://www.unitedbloodservices.org/learnMore.aspx (consultado el 2 de mayo de 2013).
Haiman, Christopher A., Daniel O. Stram, Lynn R. Wilkens, Malcom C. Pike, Laurence N. Kolonel, Brien E. Henderson y Loīc Le Marchand. “Ethnic and Racial Differences in the Smoking-Related Risk of Lung Cancer”. The New England Journal of Medicine, 2013. Disponible en línea en http://www.nejm.org/doi/full/10.1056/NEJMoa033250 (consultado el 2 de mayo de 2013).
Samuel, T. M. “Strange Facts about RH Negative Blood”. eHow Health, 2013. Disponible en línea en http://www.ehow.com/facts_5552003_strange-rh-negative-blood.html (consultado el 2 de mayo de 2013).
“United States: Uniform Crime Report – State Statistics from 1960-2011”. The Disaster Center. Disponible en línea en http://www.disastercenter.com/crime/ (consultado el 2 de mayo de 2013).
Datos del Departamento de Salud Pública del condado de Santa Clara.
Datos de la Sociedad Americana del Cáncer.
Datos de The Data and Story Library, 1996. Disponible en línea en http://lib.stat.cmu.edu/DASL/ (consultado el 2 de mayo de 2013).
Datos de la Administración Federal de Carreteras, que forma parte del Departamento de Transporte de Estados Unidos.
Datos de la Oficina del Censo de Estados Unidos, que forma parte del Departamento de Comercio de Estados Unidos.
Datos de USA Today.
“Environment”. The World Bank, 2013. Disponible en línea en http://data.worldbank.org/topic/environment (consultado el 2 de mayo de 2013).
“Search for Datasets”. Roper Center: Public Opinion Archives, University of Connecticut, 2013. Disponible en línea en https://ropercenter.cornell.edu/?s=Search+for+Datasets (consultado el 6 de febrero de 2019).
### Repaso del capítulo
Hay varias herramientas que pueden ayudar a organizar y clasificar datos cuando se calculan probabilidades. Las tablas de contingencia ayudan a visualizar los datos y son especialmente útiles cuando se calculan probabilidades que tienen múltiples variables dependientes.
Un diagrama de árbol utiliza ramas para mostrar los diferentes resultados de los experimentos y facilita la visualización de preguntas de probabilidad complejas. |
# Temas de probabilidad
## Diagramas de Venn
### Diagramas de Venn
Un diagrama de Venn es una imagen que representa los resultados de un experimento. Generalmente consiste en un recuadro que representa el espacio muestral S junto con círculos u óvalos. Los círculos u óvalos representan eventos. Los diagramas de Venn también nos ayudan a convertir palabras comunes del idioma en términos matemáticos que ayudan a agregar precisión.
Los diagramas de Venn deben su nombre a su inventor, John Venn, profesor de matemáticas en Cambridge y ministro anglicano. Su trabajo principal se llevó a cabo a finales de la década de 1870 y dio lugar a toda una rama de las matemáticas y a una nueva forma de abordar los problemas de lógica. Desarrollaremos las reglas de probabilidad que acabamos de abarcar utilizando esta poderosa forma de demostrar los postulados de la probabilidad, que incluye la regla de la adición, la regla de la multiplicación, la regla del complemento, la independencia y la probabilidad condicional.
La muestra la relación más básica entre estos números. En primer lugar, los números están en grupos llamados conjuntos; conjunto A y conjunto B. Algunos números están en ambos conjuntos; decimos que en el conjunto A en el conjunto B. La palabra “y” significa inclusivo, es decir, que tiene las características tanto de A como de B, o en este caso, que forma parte tanto de A como de B. Esta condición se llama INTERSECCIÓN de los dos conjuntos. Todos los miembros que forman parte de ambos conjuntos constituyen la intersección de los dos conjuntos. La intersección se escribe como donde es el símbolo matemático de la intersección. La afirmación se lee como "A interseca B". Puede recordarlo pensando en la intersección de dos calles.
También están los números que forman un grupo que, para ser miembro, el número debe estar en uno u otro grupo. El número no tiene que estar en AMBOS grupos, sino solamente en uno de los dos. Estos números se llaman la UNIÓN de los dos conjuntos y en este caso son los números 1-5 (de A exclusivamente), 7-9 (del conjunto B exclusivamente) y también el 6, que está en ambos conjuntos A y B. El símbolo de la UNIÓN es , por lo tanto los números 1-9, pero excluye los números 10, 11 y 12. Los valores 10, 11 y 12 forman parte del universo, pero no están en ninguno de los dos conjuntos.
Traducir la palabra “Y” al símbolo lógico matemático , intersección, y la palabra "O" al símbolo matemático , la unión, proporciona una forma muy precisa de discutir los temas de la probabilidad y la lógica. La terminología general de las tres áreas del diagrama de Venn en la se muestra en la .
Para resolver el tuvimos que recurrir al concepto de probabilidad condicional de la sección anterior. Allí utilizamos diagramas de árbol para seguir los cambios en las probabilidades, porque el espacio muestral cambiaba a medida que dibujábamos sin reemplazo. En resumen, la probabilidad condicional es la posibilidad de que algo ocurra dado que algún otro evento ya ha ocurrido. Dicho de otro modo, la probabilidad de que algo ocurra condicionada a la situación de que otra cosa también sea cierta. En el la probabilidad P(CPT) es la probabilidad condicional de que el estudiante extraído al azar sea socio del club, condicionada al hecho de que el estudiante también trabaje a tiempo parcial. Esto nos permite ver la relación entre los diagramas de Venn y los postulados de probabilidad.
### La regla de la suma de probabilidades
Antes conocimos la regla de la adición, pero sin la ayuda de los diagramas de Venn. Los diagramas de Venn ayudan a visualizar el proceso de recuento inherente al cálculo de la probabilidad. Para reafirmar la regla de la suma de probabilidades:
Recuerde que la probabilidad es simplemente la proporción de los objetos que nos interesan en relación con el número total de objetos. Por eso podemos ver la utilidad de los diagramas de Venn. El muestra cómo podemos utilizar los diagramas de Venn para contar el número de perros en la unión de marrón y macho recordándonos que hay que restar la intersección de marrón y macho. Podemos ver el efecto de esto directamente en las probabilidades en la regla de adición.
### La regla de la multiplicación de la probabilidad
Reformulando la regla de multiplicación de la probabilidad utilizando la notación de los diagramas de Venn, tenemos:
La regla de la multiplicación puede modificarse con un poco de álgebra en la siguiente regla condicional. A continuación, se pueden utilizar diagramas de Venn para demostrar el proceso.
La regla condicional:
Utilizando los mismos datos del de arriba, halle la probabilidad de que alguien obtenga una "B" si es un "novato".
La regla de multiplicación también debe modificarse si los dos eventos son independientes. Los eventos independientes se definen como una situación en la que la probabilidad condicional es simplemente la probabilidad del evento de interés. Formalmente, la independencia de los eventos se define como o . Al lanzar monedas, el resultado de la segunda tirada es independiente del resultado de la primera; las monedas no tienen memoria. La regla de multiplicación de la probabilidad para eventos independientes pasa a ser:
Una forma fácil de recordar esto es considerar lo que queremos decir con la palabra "y". Vemos que la regla de multiplicación ha traducido la palabra "y" a la notación Venn para intersección. Por lo tanto, el resultado debe cumplir las dos condiciones de primer año y nota de "B" en el ejemplo anterior. Es más difícil, menos probable, cumplir dos condiciones que una sola o alguna otra. Podemos intentar ver la lógica de la regla de la multiplicación de la probabilidad debido a que las fracciones multiplicadas entre sí se hacen más pequeñas.
El desarrollo de las reglas de la probabilidad con el uso de los diagramas de Venn puede mostrarse como una ayuda al querer calcular probabilidades a partir de datos dispuestos en una tabla de contingencia.
### Repaso del capítulo
Un diagrama de Venn es una imagen que representa los resultados de un experimento. Generalmente consiste en una caja que representa el espacio muestral S o universo de los objetos de interés junto con círculos u óvalos. Los círculos u óvalos representan grupos de eventos llamados conjuntos. Un diagrama de Venn es especialmente útil para visualizar la de eventos, la de eventos, y el complemento de un evento y para entender las probabilidades condicionales. Un diagrama de Venn es especialmente útil para visualizar una Intersección de dos eventos, una Unión de dos eventos o un Complemento de un evento. Un sistema de diagramas de Venn también puede ayudar a entender las probabilidades condicionales. Los diagramas de Venn conectan el cerebro y los ojos haciendo coincidir la aritmética literal con una imagen. Es importante señalar que se necesita más de un diagrama de Venn para resolver las fórmulas de reglas de probabilidad introducidas en la Sección 3.3.
Use la siguiente información para responder los próximos cuatro ejercicios. La muestra una muestra aleatoria de músicos y cómo aprendieron a tocar sus instrumentos.
### Resúmalo todo
Use la siguiente información para responder los próximos siete ejercicios. Un artículo en la , informó sobre un estudio de fumadores en California y Hawái. En una parte del informe se indicaba el origen étnico autodeclarado y la cantidad de cigarrillos por día. De las personas que fumaban como máximo diez cigarrillos al día, había 9.886 afroamericanos, 2.745 nativos de Hawái, 12.831 latinos, 8.378 japoneses americanos y 7.650 blancos. De las personas que fumaban como máximo diez cigarrillos al día, había 6.514 afroamericanos, 3.062 nativos de Hawái, 4.932 latinos, 10.680 japoneses americanos y 9.877 blancos. De las personas que fumaban como máximo diez cigarrillos al día, había 1.671 afroamericanos, 1.419 nativos de Hawái, 1.406 latinos, 4.715 japoneses americanos y 6.062 blancos. De las personas que fumaban al menos 31 cigarrillos al día, había 759 afroamericanos, 788 nativos de Hawái, 800 latinos, 2.305 japoneses americanos y 3.970 blancos.
Use la siguiente información para responder los próximos dos ejercicios. Suponga que tiene ocho cartas. Cinco son verdes y tres amarillas. Las cartas están bien barajadas.
Use la siguiente información para responder los próximos dos ejercicios. El porcentaje de conductores de EE. UU. con licencia (de un año reciente) que son mujeres es del 48,60. De las mujeres, el 5,03 % tienen 19 años o menos; el 81,36 % tienen entre 20 y 64 años; el 13,61 % tienen 65 años o más. De los conductores hombres con licencia en EE. UU., el 5,04 % tiene 19 años o menos; el 81,43 % tiene entre 20 y 64 años; el 13,53 % tiene 65 años o más.
### Tarea para la casa
Utilice la información de la La tabla muestra la afiliación a un partido político de cada uno de los 67 miembros del Senado de EE. UU. en junio de 2012, y cuándo se presentan a la reelección.
Use la siguiente información para responder los próximos dos ejercicios. La tabla de datos obtenida de muestra la información de bateo de cuatro conocidos jugadores de béisbol. Supongamos que se selecciona al azar un resultado de la tabla.
Use la siguiente información para responder los próximos dos ejercicios. Este diagrama de árbol muestra el lanzamiento de una moneda desigual seguido de la extracción de una cuenta de un vaso que contiene tres cuentas rojas (R), cuatro amarillas (Y) y cinco azules (B). Para la moneda, P(H) =
y P(T) =
donde H es cara y T es cruz. |
# Variables aleatorias discretas
## Introducción
Un estudiante responde un cuestionario de diez preguntas de verdadero-falso. Como el estudiante tenía una agenda tan apretada, no podía estudiar y estimaba al azar cada respuesta. ¿Cuál es la probabilidad de que el estudiante apruebe el examen con, al menos, el 70 %?
Hay pequeñas compañías que pueden estar interesadas en el número de llamadas telefónicas de larga distancia que hacen sus empleados en las horas pico del día. Supongamos que el promedio histórico es de 20 llamadas. ¿Cuál es la probabilidad de que los empleados hagan más de 20 llamadas de larga distancia durante las horas pico?
Estos dos ejemplos ilustran dos tipos diferentes de problemas de probabilidad que implican variables aleatorias discretas. Recordemos que los datos discretos son datos que se pueden contar, es decir, la variable aleatoria solo puede tomar valores de números enteros. Una variable aleatoria describe con palabras los resultados de un experimento estadístico. Los valores de una variable aleatoria pueden variar con cada repetición de un experimento, a menudo llamado ensayo.
### Notación de la variable aleatoria
La letra mayúscula X denota una variable aleatoria. Las letras minúsculas como x o y denotan el valor de una variable aleatoria. Si
Por ejemplo, supongamos que X = el número de caras que se obtiene al lanzar tres monedas imparciales. El espacio muestral para el lanzamiento de tres monedas imparciales es TTT; THH; HTH; HHT; HTT; THT; TTH; HHH. Entonces, x = 0, 1, 2, 3. X está en palabras y x es un número. Observe que para este ejemplo los valores de x son resultados contables. Como se pueden contar los posibles valores como números enteros que puede tomar X y los resultados son aleatorios (los valores de x 0, 1, 2, 3), X es una variable aleatoria discreta.
### Funciones de densidad de probabilidad (pdf) para una variable aleatoria
Una función de densidad de probabilidad o función de distribución de probabilidad tiene dos características:
1. Cada probabilidad está entre cero y uno, ambos inclusive.
2. La suma de las probabilidades es uno.
Una función de densidad de probabilidad es una fórmula matemática que calcula las probabilidades de determinados tipos de eventos, lo que hemos llamado experimentos. La función de densidad de probabilidad (probability density function, pdf) es como una receta mágica, en parte porque la misma fórmula suele describir tipos de eventos muy diferentes. Por ejemplo, la pdf binomial calculará las probabilidades de lanzar monedas, de las preguntas de respuesta afirmativa o negativa en un examen, de las opiniones de los votantes en una encuesta de opinión a favor o en contra, en definitiva, de cualquier evento binario. Otras funciones de densidad de probabilidad proporcionarán probabilidades para el tiempo que falta para que una pieza falle, cuándo llegará un cliente a la cabina de peaje de la autopista, el número de llamadas que llegan a una central telefónica, la tasa de crecimiento de una bacteria, etc. Existen familias enteras de funciones de densidad de probabilidad que se utilizan en una gran variedad de aplicaciones, como la medicina, los negocios y las finanzas, la física y la ingeniería, entre otras.
Para nuestro propósito aquí nos concentraremos en solo algunas funciones de densidad de probabilidad mientras desarrollamos las herramientas de la estadística inferencial.
### Fórmulas de recuento y fórmula combinatoria
Recordemos que la probabilidad del evento A, P(A), es simplemente el número de formas en que el experimento dará como resultado A, en relación con el número total de resultados posibles del experimento.
Como ecuación esto es:
Cuando observamos el espacio muestral para lanzar 3 monedas, podemos escribir fácilmente el espacio muestral completo y, por lo tanto, podemos contar fácilmente el número de eventos que cumplen nuestro resultado deseado, por ejemplo, x = 1 , donde X es la variable aleatoria definida como el número de caras.
A medida que tenemos un mayor número de elementos en el espacio muestral, como una baraja completa de 52 cartas, la posibilidad de escribir el espacio muestral se vuelve imposible.
Vemos que las probabilidades no son más que contar los eventos de cada grupo que nos interesa y dividirlos por el número de elementos del universo, o espacio muestral. Esto es bastante fácil si contamos los estudiantes de segundo año de una clase de Estadística, pero en casos más complicados enumerar todos los posibles resultados puede llevarnos toda la vida. Hay, por ejemplo, 36 resultados posibles al lanzar solo dos dados de seis caras en los que la variable aleatoria es la suma del número de puntos de las caras que miran hacia arriba. Si hubiera cuatro dados, el número total de resultados posibles sería de 1.296. Hay más de 2,5 MILLONES de posibles manos de póker de 5 cartas en una baraja estándar de 52 cartas. Evidentemente, llevar la cuenta de todas estas posibilidades y contarlas para llegar a una única probabilidad sería, en el mejor de los casos, tedioso.
Una alternativa a la enumeración del espacio muestral completo y al recuento del número de elementos que nos interesan, es saltarse el paso de enumerar el espacio muestral, y simplemente calcular el número de elementos que contiene y hacer la división correspondiente. Si buscamos una probabilidad, realmente no necesitamos ver todos y cada uno de los elementos del espacio muestral, solo necesitamos saber cuántos elementos hay. Las fórmulas de recuento se inventaron precisamente para eso. Nos indican el número de subconjuntos desordenados de un determinado tamaño que se pueden crear a partir de un conjunto de elementos únicos. Por desordenado se entiende que, por ejemplo, al repartir las cartas, no importa si tienes {as, as, as, as, rey} o {rey, as, as, as, as} o {as, rey, as, as, as} y así sucesivamente. Cada uno de estos subconjuntos es el mismo porque cada uno tiene 4 ases y un rey.
### Fórmula combinatoria
Es la fórmula que indica el número de subconjuntos desordenados únicos de tamaño x que se pueden crear a partir de n elementos únicos. La fórmula se lee "n combinatoria x". A veces se lee como "n elegir x". El signo de exclamación "!" se llama factorial y nos dice que hay que tomar todos los números desde el 1 hasta el número que precede al ! y multiplicarlos juntos, por lo que 4! es 1-2-3-4=24. Por definición 0! = 1. La fórmula se denomina fórmula combinatoria. También se llama coeficiente binomial, por razones que se aclararán en breve. Aunque este concepto matemático se comprendió mucho antes de 1653, se atribuye a Blaise Pascal el mayor mérito por la demostración que publicó en ese año. Además, desarrolló un método generalizado de cálculo de los valores de las combinatorias que conocemos como el Triángulo de Pascal. Pascal fue uno de los genios de una época de extraordinarios avances intelectuales que incluyó la obra de Galileo, René Descartes, Isaac Newton, William Shakespeare y el perfeccionamiento del método científico, la propia razón de ser del tema de este texto.
Vamos a encontrar por las malas el número total de combinaciones de los cuatro ases de una baraja de cartas si las tomamos de dos en dos. El espacio muestral sería:
S={Picas, Corazón),(Picas, Diamante),(Picas, Tréboles), (Diamante, Tréboles),(Corazón, Diamante),(Corazón, Tréboles)}
Hay 6 combinaciones; formalmente, seis subconjuntos desordenados únicos de tamaño 2 que se pueden crear a partir de 4 elementos únicos. Para utilizar la fórmula combinatoria resolveríamos la fórmula de la siguiente manera:
Si quisiéramos saber el número de manos únicas de póquer de 5 cartas que se pueden crear a partir de un mazo de 52 cartas, simplemente calcularíamos:
donde 52 es el número total de elementos únicos de los que estamos sacando y 5 es el grupo de tamaño en el que los estamos poniendo.
Con la fórmula combinatoria podemos contar el número de elementos de un espacio muestral sin tener que escribir cada uno de ellos, lo que realmente es el trabajo de toda una vida para solo el número de 5 manos de cartas de una baraja de 52. Ahora podemos aplicar esta herramienta a una función de densidad de probabilidad muy importante, la distribución hipergeométrica.
Recuerde que una función de densidad de probabilidad calcula las probabilidades por nosotros. Simplemente ponemos los números adecuados en la fórmula y obtenemos la probabilidad de eventos específicos. Sin embargo, para que estas fórmulas funcionen deben aplicarse solo a los casos para los que fueron diseñadas.
### Repaso del capítulo
Las características de una distribución de probabilidad o función de densidad (PDF) son las siguientes:
1. Cada probabilidad está entre cero y uno, ambos inclusive (inclusive significa incluir el cero y el uno).
2. La suma de las probabilidades es uno.
Use la siguiente información para responder los próximos cinco ejercicios: Una compañía quiere evaluar su tasa de deserción, es decir, el tiempo que los nuevos empleados permanecen en la compañía. A lo largo de los años han establecido la siguiente distribución de probabilidad.
Supongamos que X = el número de años que un nuevo empleado permanecerá en la compañía.
Supongamos que P(x) = la probabilidad de que un nuevo empleado permanezca en la compañía x años.
Use la siguiente información para responder los próximos seis ejercicios: Un panadero está decidiendo cuántos lotes de muffins va a hacer para vender en su panadería. Quiere hacer lo suficiente para venderlos todos y no menos. Mediante la observación, el panadero ha establecido una distribución de probabilidad.
Use la siguiente información para responder los próximos cuatro ejercicios: Ellen tiene práctica de música tres días a la semana. Practica los tres días el 85 % del tiempo, dos días el 8 % del tiempo, un día el 4 % del tiempo y ningún día el 3 % del tiempo. Se selecciona una semana al azar.
Use la siguiente información para responder los próximos cinco ejercicios: Javier es voluntario en eventos comunitarios cada mes. No realiza más de cinco eventos en un mes. Asiste exactamente a cinco eventos el 35 % del tiempo, a cuatro el 25 % del tiempo, a tres el 20 % del tiempo, a dos el 10 % del tiempo, a uno el 5 % del tiempo y a ninguno el 5 % del tiempo. |
# Variables aleatorias discretas
## Distribución hipergeométrica
La función de densidad de probabilidad más sencilla es la hipergeométrica. Es la más básica porque se crea combinando nuestro conocimiento de las probabilidades a partir de los diagramas de Venn, las reglas de adición y multiplicación y la fórmula de recuento combinatorio.
Para hallar el número de formas de obtener 2 ases de los cuatro que hay en la baraja, calculamos:
Y si no nos importara qué más tenemos en la mano para las otras tres cartas calcularíamos:
Uniendo todo esto, podemos calcular la probabilidad de obtener exactamente dos ases en una mano de póquer de 5 cartas como:
Esta solución es en realidad la distribución de probabilidad conocida como hipergeométrica. La fórmula generalizada es:
donde x = el número que nos interesa procedente del grupo con A objetos.
h(x) es la probabilidad de x aciertos, en n intentos, cuando los aciertos A (ases en este caso) están en una población que contiene N elementos. La distribución hipergeométrica es un ejemplo de distribución de probabilidad discreta porque no hay posibilidad de éxito parcial, es decir, no puede haber manos de póquer con 2 1/2 ases. Dicho de otro modo, una variable aleatoria discreta tiene que ser un número entero, o que se pueda contar, solamente. Esta distribución de probabilidad funciona en los casos en que la probabilidad de éxito cambia con cada extracción de cartas. Otra forma de decir esto es que los eventos NO son independientes. Al utilizar una baraja de cartas, estamos haciendo un muestreo SIN reemplazo. Si volvemos a poner cada carta después de haberla sacado, la distribución hipergeométrica sería una pdf inadecuada.
Para que el hipergeométrico funcione,
### Repaso del capítulo
La fórmula combinatoria puede proporcionar el número de subconjuntos únicos de tamaño x que se pueden crear a partir de n objetos únicos para ayudarnos a calcular las probabilidades. La fórmula combinatoria es
Un experimento hipergeométrico es un experimento estadístico con las siguientes propiedades:
1. Toma muestras de dos grupos.
2. Le interesa un grupo de interés, llamado primer grupo.
3. Toma una muestra sin reemplazo de los grupos combinados.
4. Cada selección no es independiente, ya que el muestreo es sin reemplazo.
Los resultados de un experimento hipergeométrico se ajustan a una distribución de probabilidad hipergeométrica. La variable aleatoria X = el número de elementos del grupo de interés.
.
### Revisión de la fórmula
Use la siguiente información para responder los próximos cinco ejercicios: Supongamos que un grupo de estudiantes de Estadística se divide en dos grupos: estudiantes de especialidad en Negocios y estudiantes de especialidad que no son en Negocios. En el grupo hay 16 especialidades en Negocios y siete que no son en Negocios. Se toma una muestra aleatoria de nueve estudiantes. Nos interesa el número de especialidades en Negocios en la muestra.
### TAREA PARA LA CASA
|
# Variables aleatorias discretas
## Distribución binomial
Una función de densidad de probabilidad más valiosa con muchas aplicaciones es la distribución binomial. Esta distribución calculará las probabilidades de cualquier proceso binomial. Un proceso binomial, a menudo llamado proceso de Bernoulli en honor a la primera persona que desarrolló plenamente sus propiedades, es cualquier caso en el que solo hay dos resultados posibles en cualquier ensayo, llamados éxitos y fracasos. Recibe su nombre del sistema numérico binario, en el que todos los números se reducen a 1 o 0, que es la base de la tecnología informática y de las grabaciones musicales en CD.
### Fórmula binomial
donde b(x) es la probabilidad de X aciertos en n ensayos cuando la probabilidad de un éxito en CUALQUIER ENSAYO es p. Y, por supuesto, q=(1-p) y es la probabilidad de un fracaso en cualquier ensayo.
Ahora podemos ver por qué la fórmula combinatoria se llama también coeficiente binomial, ya que vuelve a aparecer aquí en la función de probabilidad binomial. Para que la fórmula binomial funcione, la probabilidad de éxito en cualquier ensayo debe ser la misma de un ensayo a otro, o en otras palabras, los resultados de cada ensayo deben ser independientes. Lanzar una moneda es un proceso binomial porque la probabilidad de obtener una cara en un lanzamiento no depende de lo que haya ocurrido en los lanzamientos anteriores. (En este momento hay que señalar que utilizar p para el parámetro de la distribución binomial es una violación de la regla de que los parámetros de la población se designan con letras griegas. En muchos libros de texto se utiliza θ (pronunciado theta) en lugar de p y así es como debe ser.
Al igual que un conjunto de datos, una función de densidad de probabilidad tiene una media y una desviación típica que describe el conjunto de datos. Para la distribución binomial vienen dadas por las fórmulas:
Observe que p es el único parámetro en estas ecuaciones. La distribución binomial se considera, pues, de la familia de las distribuciones de probabilidad de un parámetro. En resumen, sabemos todo lo que hay que saber sobre la binomial una vez que conocemos p, la probabilidad de éxito en cualquier ensayo.
En la teoría de la probabilidad, en determinadas circunstancias, una distribución de probabilidad puede utilizarse para aproximar otra. Decimos que una es la distribución límite de la otra. Si hay que extraer un número pequeño de una población grande, aunque no haya reemplazo, podemos utilizar la binomial, aunque no sea un proceso binomial. Si no hay reemplazo se viola la regla de independencia del binomio. Sin embargo, podemos utilizar la binomial para aproximar una probabilidad que es realmente una distribución hipergeométrica si extraemos menos del 10 por ciento de la población, es decir, n es menos del 10 por ciento de N en la fórmula de la función hipergeométrica. El fundamento de este argumento es que al extraer un pequeño porcentaje de la población no alteramos la probabilidad de éxito de un sorteo a otro de forma significativa. Imagine que saca una carta no de una baraja de 52 cartas, sino de 6 barajas de cartas. La probabilidad de sacar un as, por ejemplo, no cambia la probabilidad condicional de lo que ocurre en una segunda extracción de la misma manera que lo haría si solo hubiera 4 ases en lugar de los 24 que hay ahora para sacar. Esta capacidad de utilizar una distribución de probabilidad para estimar otras será muy valiosa para nosotros más adelante.
Hay tres características de un experimento binomial
Los resultados de un experimento binomial se ajustan a una distribución de probabilidad binomial. La variable aleatoria X = el número de aciertos obtenidos en los n ensayos independientes.
La media, μ, y la varianza, σ2, de la distribución de probabilidad binomial son μ = np y σ2 = npq. La desviación típica, σ, es entonces σ =
.
Cualquier experimento que tenga las características tres y cuatro y en el que n = 1 se llama Ensayo de Bernoulli (llamado así por Jacob Bernoulli que los estudió ampliamente a finales de 1600. ). Un experimento binomial se produce cuando se cuenta el número de aciertos en uno o más ensayos de Bernoulli.
### Referencias
“Access to electricity (% of population)”. The World Bank, 2013. Disponible en línea en http://data.worldbank.org/indicator/EG.ELC.ACCS.ZS?order=wbapi_data_value_2009 %20wbapi_data_value%20wbapi_data_value-first&sort=asc (consultado el 15 de mayo de 2015).
“Distance Education”. Wikipedia. Disponible en línea en http://en.wikipedia.org/wiki/Distance_education (consultado el 15 de mayo de 2013).
“NBA Statistics-2013”, ESPN NBA, 2013. Disponible en línea en http://espn.go.com/nba/statistics/_/seasontype/2 (consultado el 15 de mayo de 2013).
Newport, Frank. “Americans Still Enjoy Saving Rather than Spending: Few demographic differences seen in these views other than by income”, GALLUP® Economy, 2013. Disponible en línea en http://www.gallup.com/poll/162368/americans-enjoy-saving-rather-spending.aspx (consultado el 15 de mayo de 2013).
Pryor, John H., Linda DeAngelo, Laura Palucki Blake, Sylvia Hurtado, Serge Tran. The American Freshman: National Norms Fall 2011. Los Ángeles: Cooperative Institutional Research Program at the Higher Education Research Institute at UCLA, 2011. También disponible en línea en http://heri.ucla.edu/PDFs/pubs/TFS/Norms/Monographs/TheAmericanFreshman2011.pdf (consultado el 15 de mayo de 2013).
“The World FactBook”, Central Intelligence Agency. Disponible en línea en https://www.cia.gov/library/publications/the-world-factbook/geos/af.html (consultado el 15 de mayo de 2013).
“What are the key statistics about pancreatic cancer?” American Cancer Society, 2013. Disponible en línea en http://www.cancer.org/cancer/pancreaticcancer/detailedguide/pancreatic-cancer-key-statistics (consultado el 15 de mayo de 2013).
### Repaso del capítulo
Un experimento estadístico se puede clasificar como experimento binomial si se cumplen las siguientes condiciones:
1. Hay un número fijo de ensayos, n.
2. Solo hay dos resultados posibles, denominados “acierto ” y “fallo” para cada ensayo. La letra p indica la probabilidad de acierto en un ensayo y la q la probabilidad de fallo en un ensayo.
3. Los n ensayos son independientes y se repiten utilizando condiciones idénticas.
Los resultados de un experimento binomial se ajustan a una distribución de probabilidad binomial. La variable aleatoria X = el número de aciertos obtenidos en los n ensayos independientes. La media de X se puede calcular mediante la fórmula μ = np, y la desviación típica viene dada por la fórmula σ =
.
La fórmula de la función de densidad de probabilidad binomial es
### Revisión de la fórmula
X ~ B(n, p) significa que la variable aleatoria discreta X tiene una distribución de probabilidad binomial con n ensayos y probabilidad de acierto p.
X = el número de aciertos en n ensayos independientes
n = el número de ensayos independientes
X toma los valores x = 0, 1, 2, 3, ..., n
p = la probabilidad de acierto de cualquier ensayo
q = la probabilidad de fallo de cualquier ensayo
p + q = 1
q = 1 – p
La media de X es μ = np. La desviación típica de X es σ =
.
donde P(X) es la probabilidad de X éxitos en n ensayos cuando la probabilidad de un éxito en CUALQUIER OTRO ENSAYO es p.
Use la siguiente información para responder los próximos ocho ejercicios: El Instituto de Investigación de la Educación Superior de la Universidad de California en Los Ángeles (University of California, Los Angeles, UCLA) recopiló datos de 203.967 estudiantes de primer año a tiempo completo de 270 institutos universitarios de cuatro años en EE. UU. El 71,3 % de esos estudiantes respondieron que sí, que creen que las parejas del mismo sexo deberían tener derecho a un estado civil legal. Supongamos que elige al azar a ocho estudiantes de primer año a tiempo completo de la encuesta. Le interesa saber el número de personas que creen que las parejas del mismo sexo deberían tener derecho a un estado civil legal.
### TAREA PARA LA CASA
Use la siguiente información para responder los próximos cuatro ejercicios. Recientemente, un enfermero comentó que cuando un paciente llama a la línea de asesoramiento médico para decir que tiene gripe, la probabilidad de que realmente la tenga (y no solo un desagradable resfriado) es solo del 4 %. De los siguientes 25 pacientes que llaman para decir que tienen gripe, nos interesa saber cuántos realmente la tienen.
Use la siguiente información para responder los dos próximos ejercicios: La probabilidad de que los San José Sharks ganen un partido cualquiera es de 0,3694, basándose en un historial de 13 años de 382 victorias de 1.034 partidos jugados (a partir de una fecha determinada). El próximo calendario mensual contiene 12 partidos.
Supongamos que X = el número de partidos ganados en ese mes. |
# Variables aleatorias discretas
## Distribución geométrica
La función de densidad de probabilidad geométrica se basa en lo que hemos aprendido de la distribución binomial. En este caso, el experimento continúa hasta que se produce un éxito o un fracaso, en lugar de un número determinado de ensayos. Hay tres características principales de un experimento geométrico.
1. Hay uno o más ensayos de Bernoulli con todos los fallos excepto el último, que es un acierto. En otras palabras, sigue repitiendo lo que está haciendo hasta el primer acierto. Entonces se detiene. Por ejemplo, se lanza un dardo a una diana hasta dar en ella. La primera vez que logra dar en la diana es un “acierto”, así que deja de lanzar el dardo. Puede que le lleve seis intentos hasta que acierte en la diana. Puede pensar en las pruebas como fallo, fallo, fallo, fallo, acierto, PARAR.
2. En teoría, el número de pruebas podría ser eterno.
3. La probabilidad, p, de un acierto y la probabilidad, q, de un fallo es igual para cada ensayo. p + q = 1 y q = 1 – p. Por ejemplo, la probabilidad de sacar un tres al lanzar un dado imparcial es
. Esto es cierto sin importar cuántas veces se lance el dado. Supongamos que quiere saber la probabilidad de obtener el primer tres en la quinta lanzada. En las lanzadas del uno al cuatro, no se obtiene un lado con un tres. La probabilidad de cada una de las lanzadas es q =
, la probabilidad de un fallo. La probabilidad de obtener un tres en la quinta lanzada es
= 0,0804
4. X = el número de ensayos independientes hasta el primer acierto.
### Notación para la Geometría: G = Función de distribución de probabilidad geométrica
X ~ G(p)
Lea como “X es una variable aleatoria con una distribución geométrica”. El parámetro es p; p = la probabilidad de acierto de cada ensayo.
La pdf geométrica nos dice la probabilidad de que la primera ocurrencia de acierto requiera x número de ensayos independientes, cada uno con probabilidad de acierto p. Si la probabilidad de éxito en cada ensayo es p, entonces la probabilidad de que el ensayo xésimo (de x ensayos) sea el primer acierto es:
para x = 1, 2, 3, .... El valor esperado de X, la media de esta distribución, es 1/p. Esto nos dice cuántos ensayos tenemos que esperar hasta obtener el primer acierto incluido en el recuento el ensayo que resulta en acierto. La forma anterior de la distribución geométrica se utiliza para modelar el número de ensayos hasta el primer acierto. El número de ensayos incluye el que es un acierto: x = todos los ensayos, incluido el que es un acierto. Esto se puede ver en la composición de la fórmula. Si X = número de ensayos incluido el acierto, entonces debemos multiplicar la probabilidad de fracaso, (1-p), por el número de fracasos, es decir, X-1.
Por el contrario, la siguiente forma de la distribución geométrica se utiliza para modelar el número de fallos hasta el primer éxito:
para x = 0, 1, 2, 3, .... En este caso el ensayo que es un éxito no se cuenta como un ensayo en la fórmula: x = número de fracasos. El valor esperado, la media, de esta distribución es . Esto nos indica cuántos fracasos debemos esperar antes de tener un acierto. En cualquier caso, la secuencia de probabilidades es una secuencia geométrica.
### Referencias
“Millennials: A Portrait of Generation Next”, PewResearchCenter. Disponible en línea en http://www.pewsocialtrends.org/files/2010/10/millennials-confident-connected-open-to-change.pdf (consultado el 15 de mayo de 2013).
“Millennials: Confident. Connected. Open to Change”. Executive Summary by PewResearch Social & Demographic Trends, 2013. Disponible en línea en http://www.pewsocialtrends.org/2010/02/24/millennials-confident-connected-open-to-change/ (consultado el 15 de mayo de 2013).
“Prevalence of HIV, total (% of populations ages 15-49),” The World Bank, 2013. Disponible en línea en http://data.worldbank.org/indicator/SH.DYN.AIDS.ZS?order=wbapi_data_value_2011+wbapi_data_value+wbapi_data_value-last&sort=desc (consultado el 15 de mayo de 2013).
Pryor, John H., Linda DeAngelo, Laura Palucki Blake, Sylvia Hurtado, Serge Tran. The American Freshman: National Norms Fall 2011. Los Ángeles: Cooperative Institutional Research Program at the Higher Education Research Institute at UCLA, 2011. También disponible en línea en http://heri.ucla.edu/PDFs/pubs/TFS/Norms/Monographs/TheAmericanFreshman2011.pdf (consultado el 15 de mayo de 2013).
“Summary of the National Risk and Vulnerability Assessment 2007/8: A profile of Afghanistan,” The European Union and ICON-Institute. Disponible en línea en http://ec.europa.eu/europeaid/where/asia/documents/afgh_brochure_summary_en.pdf (consultado el 15 de mayo de 2013).
“The World FactBook”, Central Intelligence Agency. Disponible en línea en https://www.cia.gov/library/publications/the-world-factbook/geos/af.html (consultado el 15 de mayo de 2013).
“UNICEF reports on Female Literacy Centers in Afghanistan established to teach women and girls basic resading [sic] and writing skills,” UNICEF Television. Video disponible en línea en http://www.unicefusa.org/assets/video/afghan-female-literacy-centers.html (consultado el 15 de mayo de 2013).
### Repaso del capítulo
Hay tres características de un experimento geométrico:
1. Hay uno o más ensayos de Bernoulli con todos los fallos excepto el último, que es un acierto.
2. En teoría, el número de pruebas podría ser eterno. Debe haber, al menos, un ensayo.
3. La probabilidad, p, de un acierto y la probabilidad, q, de un fallo son iguales para cada ensayo.
En un experimento geométrico defina la variable aleatoria discreta X como el número de ensayos independientes hasta el primer acierto. Decimos que X tiene una distribución geométrica y escribimos X ~ G(p) donde p es la probabilidad de acierto en un solo ensayo.
La media de la distribución geométrica X ~ G(p) es μ =
donde x = número de ensayos hasta el primer acierto de la fórmula
donde el número de pruebas es hasta el primer acierto incluido.
Una formulación alternativa de la distribución geométrica plantea la siguiente pregunta: ¿cuál es la probabilidad de x fracasos hasta el primer acierto? En esta formulación no se cuenta el ensayo que generó el primer acierto. La fórmula para esta presentación de la geométrica es:
El valor esperado en esta forma de la distribución geométrica es La forma más fácil de mantener estas dos formas de la distribución geométrica es recordar que p es la probabilidad de acierto y (1-p) es la probabilidad de fracaso. En la fórmula los exponentes simplemente cuentan el número de aciertos y el número de fallos del resultado deseado del experimento. Por supuesto, la suma de estos dos números debe dar el número de ensayos del experimento.
### Revisión de la fórmula
X ~ G(p) significa que la variable aleatoria discreta X tiene una distribución de probabilidad geométrica con probabilidad de acierto en un único ensayo p.
X = el número de ensayos independientes hasta el primer acierto
X toma los valores x = 1, 2, 3, ...
p = la probabilidad de acierto de cualquier ensayo
q = la probabilidad de fallo para cualquier ensayo p + q = 1 q = 1 – p
La media es μ =
.
La desviación típica es σ =
=
.
Use la siguiente información para responder los próximos seis ejercicios: El Instituto de Investigación de la Educación Superior de la Universidad de California en Los Ángeles (University of California, Los Angeles, UCLA) recopiló datos de 203.967 estudiantes de primer año a tiempo completo de 270 institutos universitarios de cuatro años en EE. UU. El 71,3 % de esos estudiantes respondieron que sí, que creen que las parejas del mismo sexo deberían tener derecho a un estado civil legal. Supongamos que selecciona al azar a un estudiante de primer año del estudio hasta que halle uno que responda “sí”. Le interesa el número de estudiantes de primer año a los que debe preguntar.
### TAREA PARA LA CASA
|
# Variables aleatorias discretas
## Distribución de Poisson
Otra distribución de probabilidad útil es la distribución de Poisson o distribución del tiempo de espera. Esta distribución se utiliza para determinar cuántos empleados de caja son necesarios para mantener el tiempo de espera en la fila a niveles especificados, cuántas líneas telefónicas son necesarias para evitar que el sistema se sobrecargue, y muchas otras aplicaciones prácticas. Una modificación de la distribución de Poisson, la Pascal, inventada hace casi cuatro siglos, es utilizada hoy en día por las compañías de telecomunicaciones de todo el mundo para los factores de carga, los niveles de conexión de los satélites y los problemas de capacidad de internet. La distribución recibe su nombre de Simeón Poisson, que la presentó en 1837 como una extensión de la distribución binomial, que veremos que se puede estimar con la Poisson.
Hay dos características principales de un experimento de Poisson.
1. La distribución de probabilidad de Poisson da la probabilidad de que se produzca un número de eventos en un intervalo fijo de tiempo o espacio si estos eventos se producen con una tasa promedio conocida.
2. Los eventos son independientes del tiempo transcurrido desde el último evento. Por ejemplo, un editor de libros podría estar interesado en el número de palabras escritas incorrectamente en un libro en particular. Puede ser que, en promedio, haya cinco palabras mal escritas en 100 páginas. El intervalo son las 100 páginas y se supone que no hay relación entre el momento en que se producen los errores ortográficos.
3. La variable aleatoria X = el número de ocurrencias en el intervalo de interés.
### Notación para el Poisson: P = Función de distribución de probabilidad de Poisson
X ~ P(μ)
Se lee como “X es una variable aleatoria con una distribución de Poisson”. El parámetro es μ (o λ); μ (o λ) = la media del intervalo de interés. La media es el número de ocurrencias que se producen por término promedio durante el periodo del intervalo.
La fórmula para calcular las probabilidades que provienen de un proceso de Poisson es:
donde P(X) es la probabilidad de X aciertos, μ es el número esperado de aciertos basado en datos históricos, e es el logaritmo natural aproximadamente igual a 2,718, y X es el número de aciertos por unidad, normalmente por unidad de tiempo.
Para utilizar la distribución de Poisson, deben cumplirse ciertos supuestos. Estos son: la probabilidad de un éxito, μ, no cambia dentro del intervalo, no puede haber éxitos simultáneos dentro del intervalo y, por último, que la probabilidad de un éxito entre intervalos es independiente, el mismo supuesto de la distribución binomial.
En cierto modo, la distribución de Poisson puede considerarse una forma inteligente de convertir una variable aleatoria continua, normalmente el tiempo, en una variable aleatoria discreta al dividir el tiempo en intervalos independientes discretos. Esta forma de pensar en la Poisson nos ayuda a entender por qué se puede utilizar para estimar la probabilidad de la variable aleatoria discreta de la distribución binomial. La Poisson pide la probabilidad de un número de aciertos durante un periodo mientras que la binomial pide la probabilidad de un número determinado de aciertos para un número dado de ensayos.
### Estimación de la distribución binomial con la distribución de Poisson
Anteriormente comprobamos que la distribución binomial proporcionaba una aproximación a la distribución hipergeométrica. Ahora hallamos que la distribución de Poisson puede proporcionar una aproximación para la binomial. Decimos que la distribución binomial se acerca a la Poisson. La distribución binomial se aproxima a la distribución de Poisson es a medida que n se hace más grande y p es pequeño, de manera que np se convierte en un valor constante. Hay varias reglas generales sobre cuándo se puede decir que se va a utilizar una Poisson para estimar una binomial. Una de ellas sugiere que np, la media de la binomial, debe ser inferior a 25. Otro autor sugiere que debería ser inferior a 7. Y otro, observando que la media y la varianza de la Poisson son ambas iguales, sugiere que np y npq, la media y la varianza de la binomial, deben ser mayores que 5. No existe una regla general aceptada sobre cuándo se puede utilizar la Poisson para estimar la binomial.
A medida que avanzamos por estas distribuciones de probabilidad, llegamos a distribuciones más sofisticadas que, en cierto sentido, contienen las distribuciones menos sofisticadas dentro de ellas. Esta proposición ha sido demostrada por los matemáticos. Esto nos lleva al nivel más alto de sofisticación en la siguiente distribución de probabilidad que puede ser usada como una aproximación a todas las que hemos discutido hasta ahora. Esta es la distribución normal.
### Referencias
“ATL Fact Sheet,” Department of Aviation at the Hartsfield-Jackson Atlanta International Airport, 2013. Disponible en línea en http://www.atl.com/about-atl/atl-factsheet/ (consultado el 6 de febrero de 2019).
Center for Disease Control and Prevention. “Teen Drivers: Fact Sheet,” Injury Prevention & Control: Motor Vehicle Safety, 2 de octubre de 2012. Disponible en línea en http://www.cdc.gov/Motorvehiclesafety/Teen_Drivers/teendrivers_factsheet.html (consultado el 15 de mayo de 2013).
“Children and Childrearing,” Ministry of Health, Labour, and Welfare. Disponible en línea en http://www.mhlw.go.jp/english/policy/children/children-childrearing/index.html (consultado el 15 de mayo de 2013).
“Eating Disorder Statistics,” South Carolina Department of Mental Health, 2006. Disponible en línea en http://www.state.sc.us/dmh/anorexia/statistics.htm (consultado el 15 de mayo de 2013).
“Giving Birth in Manila: The maternity ward at the Dr Jose Fabella Memorial Hospital in Manila, the busiest in the Philippines, where there is an average of 60 births a day”, theguardian, 2013. Disponible en línea en http://www.theguardian.com/world/gallery/2011/jun/08/philippines-health#/?picture=375471900&index=2 (consultado el 15 de mayo de 2013).
“How Americans Use Text Messaging,” Pew Internet, 2013. Disponible en línea en http://pewinternet.org/Reports/2011/Cell-Phone-Texting-2011/Main-Report.aspx (consultado el 15 de mayo de 2013).
Lenhart, Amanda. “Teens, Smartphones & Testing: Texting volume is up while the frequency of voice calling is down. About one in four teens say they own smartphones,” Pew Internet, 2012. Disponible en línea en http://www.pewinternet.org/~/media/Files/Reports/2012/PIP_Teens_Smartphones_and_Texting.pdf (consultado el 15 de mayo de 2013).
“One born every minute: the maternity unit where mothers are THREE to a bed”, MailOnline. Disponible en línea en http://www.dailymail.co.uk/news/article-2001422/Busiest-maternity-ward-planet-averages-60-babies-day-mothers-bed.html (consultado el 15 de mayo de 2013).
Vanderkam, Laura. “Stop Checking Your Email, Now”. CNNMoney, 2013. Disponible en línea en http://management.fortune.cnn.com/2012/10/08/stop-checking-your-email-now/ (consultado el 15 de mayo de 2013).
“World Earthquakes: Live Earthquake News and Highlights”, World Earthquakes, 2012. http://www.world-earthquakes.com/index.php?option=ethq_prediction (consultado el 15 de mayo de 2013).
### Repaso del capítulo
Una distribución de probabilidad de Poisson de una variable aleatoria discreta da la probabilidad de que se produzca un número de eventos en un intervalo fijo de tiempo o espacio, si estos eventos se producen a una tasa promedio conocida y con independencia del tiempo transcurrido desde el último evento. La distribución de Poisson puede utilizarse para aproximarse a la binomial, si la probabilidad de éxito es "pequeña" (menor o igual a 0,01) y el número de ensayos es "grande" (mayor o igual a 25). También se sugieren otras reglas generales por parte de diferentes autores, pero todos reconocen que la distribución de Poisson es la distribución límite de la binomial a medida que n aumenta y p se acerca a cero.
La fórmula para calcular las probabilidades que provienen de un proceso de Poisson es:
donde P(X) es la probabilidad de éxitos, μ (pronunciado mi) es el número esperado de éxitos, e es el logaritmo natural aproximadamente igual a 2,718, y X es el número de éxitos por unidad, normalmente por unidad de tiempo.
### Revisión de la fórmula
X ~ P(μ) significa que X tiene una distribución de probabilidad de Poisson donde X = el número de ocurrencias en el intervalo de interés.
X toma los valores x = 0, 1, 2, 3, ...
Se suele dar la media μ o λ.
La varianza es σ2 = μ, y la desviación típica es
.
Cuando se utiliza P(μ) para aproximar una distribución binomial, μ = np donde n representa el número de ensayos independientes y p representa la probabilidad de aciertos en un solo ensayo.
Use la siguiente información para responder los próximos seis ejercicios: en promedio, una tienda de ropa recibe 120 clientes al día.
Use la siguiente información para responder los próximos seis ejercicios: en EE. UU. mueren un promedio de ocho adolescentes al día por accidentes de tráfico. Como consecuencia, los estados de todo el país están debatiendo el aumento de la edad para conducir.
### TAREA PARA LA CASA
Use la siguiente información para responder los dos próximos ejercicios: los gatos de la señora Plum la despiertan por la noche porque quieren jugar un promedio de diez veces a la semana. Nos interesa saber el número de veces que sus gatos la despiertan cada semana. |
# Variables aleatorias continuas
## Introducción
Las variables aleatorias continuas tienen muchas aplicaciones. Los promedios de bateo en béisbol, las puntuaciones de CI, la duración de una llamada telefónica de larga distancia, la cantidad de dinero que lleva una persona, la duración de un chip de computadora, las tasas de rendimiento de una inversión y las puntuaciones de la selectividad son solo algunos ejemplos. El campo de la fiabilidad depende de una variedad de variables aleatorias continuas, al igual que todos los ámbitos del análisis de riesgos. |
# Variables aleatorias continuas
## Propiedades de las funciones de densidad de probabilidad continuas
El gráfico de una distribución de probabilidad continua es una curva. La probabilidad se representa mediante el área que está debajo de la curva. Ya conocimos este concepto cuando desarrollamos las frecuencias relativas con histogramas en el Capítulo 2. El área relativa para un rango de valores era la probabilidad de extraer al azar una observación en ese grupo. De nuevo con la distribución de Poisson del Capítulo 4, el gráfico del Ejemplo 4.14 utilizó cajas para representar la probabilidad de valores específicos de la variable aleatoria. En este caso, estábamos siendo poco estrictos porque las variables aleatorias de una distribución de Poisson son discretas, números enteros, y una caja tiene anchura. Observe que el eje horizontal, la variable aleatoria x, deliberadamente no marcó los puntos a lo largo del eje. La probabilidad de un valor específico de una variable aleatoria continua será cero porque el área bajo un punto es cero. La probabilidad es el área.
La curva se denomina función de densidad de probabilidad (abreviada como pdf). Utilizamos el símbolo f(x) para representar la curva. f(x) es la función que corresponde al gráfico; utilizamos la función de densidad f(x) para dibujar el gráfico de la distribución de probabilidad.
El área debajo de la curva viene dada por una función diferente llamada función de distribución acumulativa (cdf). La función de distribución acumulativa se utiliza para evaluar la probabilidad como área. Matemáticamente, la función de densidad de probabilidad acumulada es la integral de la pdf, y la probabilidad entre dos valores de una variable aleatoria continua será la integral de la pdf entre estos dos valores: el área bajo la curva entre estos valores. Recuerde que el área bajo la pdf para todos los valores posibles de la variable aleatoria es uno, la certeza. Por tanto, la probabilidad puede verse como el porcentaje relativo de certeza entre los dos valores de interés.
Hallaremos el área que representa la probabilidad mediante geometría, fórmulas, tecnología o tablas de probabilidad. En general, el cálculo integral es necesario para hallar el área bajo la curva de muchas funciones de densidad de probabilidad. Cuando usamos fórmulas para hallar el área en este libro de texto, las fórmulas fueron halladas mediante técnicas del cálculo integral.
Hay muchas distribuciones de probabilidad continuas. Cuando se utiliza una distribución de probabilidad continua para modelar la probabilidad, la distribución utilizada se selecciona para modelar y ajustarse a la situación particular de la mejor manera.
En este capítulo y en el siguiente estudiaremos la distribución uniforme, la exponencial y la normal. Los siguientes gráficos ilustran estas distribuciones.
Para distribuciones de probabilidad continuas, PROBABILIDAD = ÁREA.
### Repaso del capítulo
La función de densidad de probabilidad (pdf) se utiliza para describir probabilidades de variables aleatorias continuas. El área debajo de la curva de densidad entre dos puntos corresponde a la probabilidad de que la variable se sitúe entre esos dos valores. En otras palabras, el área debajo de la curva de densidad entre los puntos a y b es igual a P(a < x < b). La función de distribución acumulativa (cdf) da la probabilidad como un área. Si X es una variable aleatoria continua, la función de densidad de probabilidad (pdf), f(x) se utiliza para dibujar el gráfico de la distribución de probabilidad. El área total debajo del gráfico de f(x) es uno. El área debajo del gráfico de f(x) y entre los valores a y b da la probabilidad P(a < x < b).
La función de distribución acumulativa (cdf) de X se define por P (X ≤ x). Es una función de x que da la probabilidad de que la variable aleatoria sea menor o igual que x.
### Revisión de la fórmula
Función de densidad de probabilidad (pdf) f(x):
1. f(x) ≥ 0
2. El área total debajo de la curva f(x) es uno.
Función de distribución acumulativa (cdf): P(X ≤ x)
### Tarea para la casa
Para cada problema de probabilidad y percentil, haga el dibujo. |
# Variables aleatorias continuas
## La distribución uniforme
La distribución uniforme es una distribución de probabilidad continua y se refiere a eventos que tienen la misma probabilidad de ocurrir. Cuando se resuelven problemas que tienen una distribución uniforme, hay que tener en cuenta si los datos son inclusivos o excluyentes de los extremos.
El enunciado matemático de la distribución uniforme es
f(x) =
para a ≤ x ≤ b
donde a = el menor valor de x y b = el mayor valor de x.
Las fórmulas para la media teórica y la desviación típica son
y
### Repaso del capítulo
Si X tiene una distribución uniforme donde a < x < b o a ≤ x ≤ b, entonces X toma valores entre a y b (puede incluir a y b). Todos los valores x son igualmente probables. Escribimos X ∼ U(a, b). La media de X es
. La desviación típica de X es
. La función de densidad de probabilidad de X es
para a ≤ x ≤ b. La función de distribución acumulativa de X es P(X ≤ x) =
. X es continua.
La probabilidad de P(c < X < d) se puede hallar calculando el área bajo f(x), entre c y d. Dado que el área correspondiente es un rectángulo, el área se puede hallar simplemente multiplicando el ancho y la altura.
### Revisión de la fórmula
X = un número real entre a y b (en algunos casos, X puede tomar los valores a y b). a = X más pequeño; b = X más grande
X ~ U (a, b)
La media es
La desviación típica es
Función de densidad de probabilidad:
para
Área a la izquierda de P(X < x) = (x – a)
Área a la derecha de P(X > x) = (b – x)
Área entre P(c < x < d) = (base)(altura) = (d – c)
1. pdf:
para a ≤ x ≤ b
2. cdf: P(X ≤ x) =
3. media µ =
4. desviación típica σ
5. P(c < X < d) = (d – c)
### Referencias
McDougall, John A. The McDougall Program for Maximum Weight Loss. Plume, 1995.
Use la siguiente información para responder las próximas diez preguntas. Los datos que siguen son los pies cuadrados (en 1.000 pies cuadrados) de 28 viviendas.
La media muestral = 2,50 y la desviación típica de la muestra = 0,8302.
La distribución se puede escribir como X ~ U(1,5, 4,5).
Use la siguiente información para responder los próximos ocho ejercicios. Una distribución está dada como X ~ U(0, 12).
Use la siguiente información para responder los próximos once ejercicios. La edad de los automóviles en el estacionamiento del personal de un instituto universitario suburbano se distribuye uniformemente desde los seis meses (0,5 años) hasta los 9,5 años.
### Tarea para la casa
Para cada problema de probabilidad y percentil, haga el dibujo.
Use la siguiente información para responder los próximos tres ejercicios. Se supone que el Sky Train llega cada ocho minutos desde la terminal hasta el centro de alquiler de automóviles y el estacionamiento de larga duración. Se sabe que los tiempos de espera del tren siguen una distribución uniforme. |
# Variables aleatorias continuas
## La distribución exponencial
La distribución exponencial suele referirse a la cantidad de tiempo que transcurre hasta que se produce algún evento específico. Por ejemplo, la cantidad de tiempo (que comienza ahora) hasta que se produzca un terremoto tiene una distribución exponencial. Otros ejemplos son la duración, en minutos, de las llamadas telefónicas de larga distancia comerciales y la cantidad de tiempo, en meses, que dura la batería de un automóvil. También se puede demostrar que el valor del cambio que se tiene en el bolsillo o en el monedero sigue una distribución exponencial aproximadamente.
Los valores de una variable aleatoria exponencial se producen de la siguiente manera. Hay menos valores grandes y más valores pequeños. Por ejemplo, los estudios de marketing han demostrado que la cantidad de dinero que los clientes gastan en una visita al supermercado sigue una distribución exponencial. Hay más gente que gasta pequeñas cantidades de dinero y menos gente que gasta grandes cantidades de dinero.
Las distribuciones exponenciales se utilizan habitualmente en cálculos de fiabilidad de productos, es decir, el tiempo que dura un producto.
La variable aleatoria de la distribución exponencial es continua y suele medir el paso del tiempo, aunque puede utilizarse en otras aplicaciones. Las preguntas típicas pueden ser: "¿cuál es la probabilidad de que algún evento ocurra en los próximos horas o días, o cuál es la probabilidad de que algún evento ocurra entre horas y horas, o cuál es la probabilidad de que el evento dure más de horas para llevarse a cabo" En resumen, la variable aleatoria X es iguala (a) el tiempo entre eventos o(b) el paso del tiempo para completar una acción, por ejemplo, esperar a un cliente. La función de densidad de probabilidad viene dada por:
donde μ es el tiempo promedio de espera histórico.
y tiene una media y una desviación típica de 1/μ.
Una forma alternativa de la fórmula de la distribución exponencial reconoce lo que suele llamarse el factor de decaimiento. El factor de decaimiento simplemente mide la rapidez con la que la probabilidad de un evento disminuye a medida que la variable aleatoria X aumenta. Cuando se utiliza la notación con el parámetro de decaimiento m, la función de densidad de probabilidad se presenta como
donde
Para calcular las probabilidades de determinadas funciones de densidad de probabilidad, se utiliza la función de densidad acumulada. La función de densidad acumulativa (cdf) es simplemente la integral de la pdf y es:
### La falta de memoria de la distribución exponencial
Recordemos que la cantidad de tiempo entre clientes para el empleado de correos comentado anteriormente se distribuye exponencialmente con una media de dos minutos. Supongamos que han pasado cinco minutos desde que llegó el último cliente. Dado que ha transcurrido un tiempo inusualmente largo, parece más probable que un cliente llegue durante el próximo minuto. Con la distribución exponencial, esto no es así: el tiempo adicional de espera del siguiente cliente no depende del tiempo que haya transcurrido desde el último cliente. Esto se conoce como la propiedad de falta de memoria. Las funciones de densidad de probabilidad exponencial y geométrica son las únicas funciones de probabilidad que tienen la propiedad de falta de memoria. Específicamente, la propiedad de falta de memoria dice que
P (X > r + t | X > r) = P (X > t) para todo r ≥ 0 y t ≥ 0
Por ejemplo, si han transcurrido cinco minutos desde la llegada del último cliente, la probabilidad de que transcurra más de un minuto antes de que llegue el siguiente cliente se calcula utilizando r = 5 y t = 1 en la ecuación anterior.
P(X > 5 + 1 | X > 5) = P(X > 1) =
= 0,6065.
Es la misma probabilidad que la de esperar más de un minuto a que llegue un cliente después de la llegada anterior.
La distribución exponencial se utiliza a menudo para modelar la longevidad de un dispositivo eléctrico o mecánico. En el , la vida útil de una determinada pieza de computadora tiene la distribución exponencial con una media de diez años. La propiedad de falta de memoria dice que el conocimiento de lo que ha ocurrido en el pasado no tiene ningún efecto sobre probabilidades futuras. En este caso, significa que una pieza usada no tiene más probabilidades de estropearse en un momento determinado que una pieza nueva. En otras palabras, la pieza se mantiene como nueva hasta que se rompe de repente. Por ejemplo, si la pieza ya ha durado diez años, la probabilidad de que dure otros siete es P(X > 17|X > 10) = P(X > 7) = 0,4966, donde la línea vertical se lee como “dada".
### Relación entre la distribución de Poisson y la distribución exponencial
Existe una relación interesante entre la distribución exponencial y la distribución de Poisson. Supongamos que el tiempo que transcurre entre dos eventos sucesivos sigue la distribución exponencial con una media de μ unidades de tiempo. También se supone que estos tiempos son independientes, lo que significa que el tiempo entre eventos no se ve afectado por los tiempos entre eventos anteriores. Si se cumplen estos supuestos, el número de eventos por unidad de tiempo sigue una distribución de Poisson con una media μ. Recordemos que si X tiene la distribución de Poisson con una media μ, entonces
.
La fórmula de la distribución exponencial Donde m = el parámetro de la tasa, o μ = tiempo promedio entre ocurrencias.
Vemos que la exponencial es la pariente de la distribución de Poisson y se relacionan a través de esta fórmula. Existen importantes diferencias que hacen que cada distribución sea relevante para diferentes tipos de problemas de probabilidad.
En primer lugar, la Poisson tiene una variable aleatoria discreta, x, en la que el tiempo; una variable continua se divide artificialmente en trozos discretos. Vimos que el número de ocurrencias de un evento en un intervalo de tiempo dado, x, sigue la distribución de Poisson.
Por ejemplo, el número de veces que suena el teléfono por hora. En cambio, el tiempo entre ocurrencias sigue la distribución exponencial. Por ejemplo. El teléfono acaba de sonar, ¿cuánto tiempo pasará hasta que vuelva a sonar? Estamos midiendo la duración del intervalo, una variable aleatoria continua, exponencial, no los eventos durante un intervalo, Poisson.
### La distribución exponencial frente a la distribución de Poisson
Una forma visual de mostrar tanto las similitudes como las diferencias entre estas dos distribuciones es con una línea del tiempo.
La variable aleatoria de la distribución de Poisson es discreta y, por tanto, cuenta los eventos durante un periodo determinado, de t1 a t2 en la , y calcula la probabilidad de que se produzca ese número. El número de eventos, cuatro en el gráfico, se mide en números que se pueden contar; por lo tanto, la variable aleatoria de Poisson es una variable aleatoria discreta.
La distribución de probabilidad exponencial calcula las probabilidades del paso del tiempo, una variable aleatoria continua. En la esto se muestra como el paréntesis desde t1 hasta la siguiente ocurrencia del evento marcada con un triángulo.
Las preguntas clásicas de la distribución de Poisson son "¿cuántas personas llegarán a mi caja en la próxima hora?".
Las preguntas clásicas de la distribución exponencial son "¿cuánto tiempo pasará hasta que llegue la siguiente persona?", o una variante, "¿cuánto tiempo permanecerá la persona una vez que haya llegado?".
De nuevo, la fórmula de la distribución exponencial es:
Vemos inmediatamente la similitud entre la fórmula exponencial y la fórmula de Poisson.
Ambas funciones de densidad de probabilidad se basan en la relación entre el tiempo y el crecimiento o decaimiento exponencial. La "e" de la fórmula es una constante con el valor aproximado de 2,71828 y es la base de la fórmula del crecimiento exponencial logarítmica natural. Cuando la gente dice que algo ha crecido exponencialmente, se refiere a esto.
Un ejemplo de la exponencial y la Poisson dejará claras las diferencias entre ambas. También mostrará las interesantes aplicaciones que tienen.
Distribución de PoissonSupongamos que históricamente llegan 10 clientes a las filas de espera en las cajas registradoras cada hora. Recuerde que esto es todavía una probabilidad, por lo que nos tienen que decir estos valores históricos. Vemos que se trata de un problema de probabilidad de Poisson.
Podemos introducir esta información en la función de densidad de probabilidad de Poisson y obtener una fórmula general que calculará la probabilidad de que llegue algún número determinado de clientes en la próxima hora.
La fórmula es para cualquier valor de la variable aleatoria que hayamos elegido y, por tanto, la x se pone en la fórmula. Esta es la fórmula:
Como ejemplo, la probabilidad de que lleguen 15 personas a la caja registradora en la próxima hora sería
Aquí insertamos x = 15 y calculamos que la probabilidad de que en la próxima hora lleguen 15 personas es de 0,061.
Distribución exponencialSi mantenemos los mismos hechos históricos de que llegan 10 clientes cada hora, pero ahora nos interesa el tiempo de servicio que pasa una persona en el mostrador, entonces utilizaríamos la distribución exponencial. La función de probabilidad exponencial para cualquier valor de x, la variable aleatoria, para estos datos históricos de la caja registradora es:
Para calcular µ, el tiempo promedio de servicio histórico, simplemente dividimos el número de personas que llegan por hora, 10, entre el periodo, una hora, y tenemos µ = 0,1. Históricamente, la gente pasa el 0,1 de una hora en la caja registradora, es decir, 6 minutos. Esto explica el 0,1 de la fórmula.
Existe una confusión natural con µ tanto en las fórmulas de Poisson como en las exponenciales. Tienen significados diferentes, aunque tengan el mismo símbolo. La media de la exponencial es uno dividido entre la media de Poisson. Si se da el número histórico de llegadas se tiene la media de Poisson. Si se da una duración histórica entre eventos, se tiene la media de una exponencial.
Siguiendo con nuestro ejemplo de la caja, si quisiéramos saber la probabilidad de que una persona tarde 9 minutos o menos en pasar por caja registradora, utilizaríamos esta fórmula. En primer lugar, convertimos a las mismas unidades de tiempo que son partes de una hora. Nueve minutos son 0,15 de una hora. A continuación, observamos que estamos pidiendo un rango de valores. Este es siempre el caso de una variable aleatoria continua. Escribimos la pregunta de probabilidad como
Ahora podemos poner los números en la fórmula y obtenemos nuestro resultado.
La probabilidad de que un cliente emplee 9 minutos o menos en pasar por caja es de 0,7769.
Vemos que tenemos una alta probabilidad de salir en menos de nueve minutos y una mínima probabilidad de que lleguen 15 clientes en la próxima hora.
### Repaso del capítulo
Si X tiene una distribución exponencial con media μ, entonces el parámetro de decaimiento es m =
. La función de densidad de probabilidad de X es f(x) = me (o equivalentemente
. La función de distribución acumulativa de X es P(X ≤ x) = 1 – e–.
### Revisión de la fórmula
1. pdf: f(x) = me(– donde x ≥ 0 y m > 0
2. cdf: P(X ≤ x) = 1 – e(–
3. media µ =
4. desviación típica σ = µ
5. Además
6. Probabilidad de Poisson:
con una media y una varianza de μ
### Referencias
Datos de la Oficina del Censo de Estados Unidos.
Datos de World Earthquakes, 2013. Disponible en línea en http://www.world-earthquakes.com/ (consultado el 11 de junio de 2013).
“No-hitter”. Baseball-Reference.com, 2013. Disponible en línea en http://www.baseball-reference.com/bullpen/No-hitter (consultado el 11 de junio de 2013).
Zhou, Rick. “Exponential Distribution lecture slides”. Disponible en línea en www.public.iastate.edu/~riczw/stat330s11/lecture/lec13.pdf (consultado el 11 de junio de 2013).
Use la siguiente información para responder los próximos diez ejercicios. Un representante del servicio de atención al cliente debe dedicar diferentes cantidades de tiempo a cada cliente para resolver varias preocupaciones. La cantidad de tiempo dedicado a cada cliente se puede modelar mediante la siguiente distribución: X ~ Exp(0,2)
Use la siguiente información para responder los próximos siete ejercicios. Una distribución está dada como X ~ Exp(0,75).
Use la siguiente información para responder los próximos 16 ejercicios. El carbono-14 es un elemento radiactivo con una semivida de unos 5.730 años. Se dice que el carbono-14 se descompone exponencialmente. La tasa de descomposición es de 0,000121. Empezamos con un gramo de carbono-14. Nos interesa el tiempo (años) que tarda en descomponerse el carbono-14.
### Tarea para la casa
Use la siguiente información para responder los próximos tres ejercicios. La vida promedio de un determinado teléfono móvil nuevo es de tres años. El fabricante sustituirá cualquier teléfono móvil que falle durante los dos años siguientes a la fecha de compra. Se sabe que la vida útil de estos teléfonos móviles sigue una distribución exponencial. |
# La distribución normal
## Introducción
La función de densidad de probabilidad normal, una distribución continua, es la más importante de todas las distribuciones. Su uso está muy extendido y su abuso aun más. Su gráfico tiene forma de campana. La curva de campana se ve en casi todas las disciplinas. Algunas de ellas son Psicología, Negocios, Economía, Ciencias, Enfermería y, por supuesto, Matemáticas. Algunos de sus instructores pueden utilizar la distribución normal para ayudar a determinar su calificación. La mayoría de las calificaciones de coeficiente intelectual (Intelligence Quotient, IQ) se distribuyen normalmente. A menudo, los precios de los inmuebles se ajustan a una distribución normal.
La distribución normal es muy importante, pero no se puede aplicar a todo en el mundo real. Recuerde que todavía estamos hablando de la distribución de los datos de la población. Se trata de una discusión sobre la probabilidad y, por tanto, son los datos de la población los que pueden estar distribuidos normalmente, y si lo están, entonces es así como podemos calcular las probabilidades de eventos específicos al igual que hicimos con los datos de la población que pueden tener una distribución binomial o de Poisson. Esta precaución se debe a que en el próximo capítulo veremos que la distribución normal describe algo muy diferente de los datos brutos y constituye la base de la estadística inferencial.
La distribución normal tiene dos parámetros (dos medidas numéricas descriptivas): la media (μ) y la desviación típica (σ). Si X es una cantidad que se va a medir que tiene una distribución normal con media(μ) y desviación típica(σ), la designamos escribiendo la siguiente fórmula de la función de densidad de probabilidad normal:
La función de densidad de probabilidad es una función bastante complicada. No la memorice. No es necesario.
La curva es simétrica respecto a una línea vertical que pasa por la media, μ. La media es la misma que la mediana, que es la misma que la moda, porque el gráfico es simétrico respecto a μ. Como indica la notación, la distribución normal solo depende de la media y de la desviación típica. Observe que esto es diferente a varias funciones de densidad de probabilidad que ya hemos estudiado, como la de Poisson, donde la media es igual a y la desviación típica simplemente la raíz cuadrada de la media, o la binomial, donde p se utiliza para determinar tanto la media como la desviación típica. Dado que el área bajo la curva debe ser igual a uno, un cambio en la desviación típica, σ, provoca un cambio en la forma de la curva normal; la curva se vuelve más abultada y ancha o más delgada y alta dependiendo de σ. Un cambio en μ hace que el gráfico se desplace a la izquierda o a la derecha. Esto significa que hay un número infinito de distribuciones de probabilidad normales. Una de las más interesantes es la llamada distribución normal estándar.
### Revisión de la fórmula
X ∼ N(μ, σ)
|
# La distribución normal
## La distribución normal estándar
La distribución normal estándar es una distribución normal de valores estandarizados llamados puntuaciones . Una puntuación
La media de la distribución normal estándar es cero y la desviación típica es uno. Lo que hace esto es simplificar drásticamente el cálculo matemático de las probabilidades. Tómese un momento y sustituya el cero y el uno en los lugares apropiados de la fórmula anterior y podrá ver que la ecuación se reduce a una que puede resolverse mucho más fácilmente utilizando el cálculo integral. La transformación z =
produce la distribución Z ~ N(0, 1). El valor x en la ecuación dada proviene de una distribución normal conocida con media conocida μ y desviación típica conocida σ. La puntuación zindica cuántas desviaciones típicas se aleja una determinada x de la media.
### Puntuaciones z
Si X es una variable aleatoria normalmente distribuida y X ~ N(μ, σ), entonces la puntuación z para una determinada x es:
La puntuación Los valores de x que son mayores que la media tienen puntuaciones z positivas, y los valores de x que son menores que la media tienen puntuaciones z negativas. Si x es igual a la media, entonces x tiene una puntuación z de cero.
La regla empíricaSi X es una variable aleatoria y tiene una distribución normal con media µ y desviación típica σ, la regla empírica dice lo siguiente:
1. Aproximadamente el 68 % de los valores de x se sitúan entre –1σ y +1σ de la media µ (dentro de una desviación típica de la media).
2. Aproximadamente el 95 % de los valores de x se sitúan entre –2σ y +2σ de la media µ (dentro de dos desviaciones típicas de la media).
3. Aproximadamente el 99,7 % de los valores de x se sitúan entre –3σ y +3σ de la media µ (dentro de las tres desviaciones típicas de la media). Observe que casi todos los valores de x están dentro de las tres desviaciones típicas de la media.
4. Las puntuaciones z para +1σ y –1σ son +1 y –1, respectivamente.
5. Las puntuaciones z para +2σ y –2σ son +2 y –2, respectivamente.
6. Las puntuaciones z para +3σ y –3σ son +3 y –3, respectivamente.
### Referencias
“Blood Pressure of Males and Females”. StatCruch, 2013. Disponible en línea en http://www.statcrunch.com/5.0/viewreport.php?reportid=11960 (consultado el 14 de mayo de 2013).
“The Use of Epidemiological Tools in Conflict-affected populations: Open-access educational resources for policy-makers: Calculation of z-scores”. London School of Hygiene and Tropical Medicine, 2009. Disponible en línea en http://conflict.lshtm.ac.uk/page_125.htm (consultado el 14 de mayo de 2013).
“2012 College-Bound Seniors Total Group Profile Report”. CollegeBoard, 2012. Disponible en línea en http://media.collegeboard.com/digitalServices/pdf/research/TotalGroup-2012.pdf (consultado el 14 de mayo de 2013).
“Digest of Education Statistics: ACT score average and standard deviations by sex and race/ethnicity and percentage of ACT test takers, by selected composite score ranges and planned fields of study: Selected years, 1995 through 2009”. National Center for Education Statistics. Disponible en línea en http://nces.ed.gov/programs/digest/d09/tables/dt09_147.asp (consultado el 14 de mayo de 2013).
Datos de The Mercury News de San José.
Datos de The World Almanac and Book of Facts.
“List of stadiums by capacity”. Wikipedia. Disponible en línea en https://en.wikipedia.org/wiki/List_of_stadiums_by_capacity (consultado el 14 de mayo de 2013).
Datos de la Asociación Nacional de Baloncesto. Disponible en línea en www.nba.com (consultado el 14 de mayo de 2013).
### Repaso del capítulo
Una puntuación z es un valor estandarizado. Su distribución es la normal estándar, Z ~ N(0, 1). La media de las puntuaciones z es cero y la desviación típica es uno. Si z es la puntuación z para un valor x de la distribución normal N(µ, σ), entonces z indica cuántas desviaciones típicas está x por encima (mayor que) o por debajo (menor que) de µ.
### Revisión de la fórmula
Z ~ N(0, 1)
z = un valor estandarizado (puntuación z)
media = 0; desviación típica = 1
Para hallar el percentil K de X , x, cuando se conocen las puntuaciones z:x = μ + (z)σ
puntuación z: k =
o z =
Z = la variable aleatoria de las puntuaciones z
Z ~ N(0, 1)
Use la siguiente información para responder los dos próximos ejercicios: la vida de los reproductores de CD de Sunshine se distribuye normalmente, con una media de 4,1 años y una desviación típica de 1,3 años. El reproductor de CD tiene una garantía de tres años. Nos interesa la duración de un reproductor de CD.
### Tarea para la casa
Use la siguiente información para responder los dos próximos ejercicios: el tiempo de recuperación del paciente de un procedimiento quirúrgico en particular se distribuye normalmente, con una media de 5,3 días y una desviación típica de 2,1 días. |
# La distribución normal
## Uso de la distribución normal
El área sombreada en el siguiente gráfico indica el área a la derecha de x. Esta zona está representada por la probabilidad P(X > x). Las tablas normales proporcionan la probabilidad entre la media, cero para la distribución normal estándar y un valor específico como . Esta es la parte no sombreada del gráfico desde la media hasta .
Como la distribución normal es simétrica, si estuviera a la misma distancia a la izquierda de la media, el área (la probabilidad) en la cola izquierda, sería la misma que el área sombreada en la cola derecha. Además, hay que tener en cuenta que, debido a la simetría de esta distribución, la mitad de la probabilidad está a la derecha de la media y la otra mitad a la izquierda.
### Cálculo de probabilidades
Para hallar la probabilidad de las funciones de densidad de probabilidad con una variable aleatoria continua necesitamos calcular el área bajo la función a través de los valores de X que nos interesan. Para la distribución normal esto parece una tarea difícil dada la complejidad de la fórmula. Sin embargo, hay una forma sencilla de conseguir lo que queremos. Aquí tenemos de nuevo la fórmula de la distribución normal:
Al observar la fórmula de la distribución normal no está claro cómo vamos a resolver la probabilidad haciéndolo de la misma manera que lo hicimos con las funciones de probabilidad anteriores. Allí pusimos los datos en la fórmula e hicimos las cuentas.
Para resolver este rompecabezas, desde el principio sabemos que el área bajo una función de densidad de probabilidad es la probabilidad.
Esto demuestra que el área entre X1 y X2 es la probabilidad que se indica en la fórmula: P (X1 ≤ x ≤ X2)
La herramienta matemática necesaria para calcular el área bajo una curva es el cálculo integral. La integral de la función de densidad de probabilidad normal entre los dos puntos x1 y x2 es el área bajo la curva entre estos dos puntos y es la probabilidad entre estos dos puntos.
Hacer estas integrales no es divertido y puede llevar mucho tiempo. Pero ahora, recordando que hay un número infinito de distribuciones normales, podemos considerar la que tiene una media de cero y una desviación típica de 1. Esta particular distribución normal recibe el nombre de distribución normal estándar. Al poner estos valores en la fórmula se reduce a una ecuación muy sencilla. Ahora podemos calcular fácilmente todas las probabilidades para cualquier valor de x en esta distribución normal particular, que tiene una media de cero y una desviación típica de 1. Se han elaborado y están disponibles aquí en el apéndice del texto o en cualquier lugar de la web. Se presentan de varias maneras. La tabla de este texto es la presentación más habitual y se establece con las probabilidades de la mitad de la distribución, que comienza por el cero, la media, y moviéndose hacia fuera. El área sombreada en el gráfico de la parte superior de la tabla en Tablas estadísticas representa la probabilidad desde cero hasta el valor Z específico anotado en el eje horizontal, Z.
El único problema es que, incluso con esta tabla, sería una ridícula coincidencia que nuestros datos tuvieran una media de cero y una desviación típica de uno. La solución es convertir la distribución que tenemos con su media y desviación típica a esta nueva distribución normal estándar. La normal estándar tiene una variable aleatoria llamada Z.
Al usar la tabla normal estándar, que por lo general se llama tabla normal, para hallar la probabilidad de una desviación típica, vaya a la columna Z, lea hasta 1,0 y luego lea en la columna 0. Ese número, 0,3413 es la probabilidad de cero a 1 desviación típica. En la parte superior de la tabla se encuentra la zona sombreada de la distribución que es la probabilidad para una desviación típica. La tabla ha resuelto nuestro problema de cálculo integral, pero solo si nuestros datos tienen una media de cero y una desviación típica de 1.
Sin embargo, el punto esencial aquí es que la probabilidad de una desviación típica en una distribución normal es la misma en todas las distribuciones normales. Si el conjunto de datos de la población tiene una media de 10 y una desviación típica de 5, entonces la probabilidad de 10 a 15, una desviación típica, es la misma que de cero a 1, una desviación típica en la distribución normal estándar. Para calcular las probabilidades, las áreas, para cualquier distribución normal, solo tenemos que convertir la distribución normal particular a la distribución normal estándar y buscar la respuesta en las tablas. Como revisión, aquí está de nuevo la fórmula de normalización:
donde Z es el valor de la distribución normal estándar, X es el valor de una distribución normal que se desea convertir a la normal estándar, μ y σ son, respectivamente, la media y la desviación típica de esa población. Tenga en cuenta que la ecuación utiliza μ y σ lo que denota parámetros poblacionales. Esto sigue tratando con la probabilidad, por lo que siempre estamos tratando con la población, con valores de parámetros conocidos y una distribución conocida. También es importante tener en cuenta que, como la distribución normal es simétrica, no importa si la puntuación z es positiva o negativa a la hora de calcular una probabilidad. Una desviación típica a la izquierda (puntuación Z negativa) cubre la misma área que una desviación típica a la derecha (puntuación Z positiva). Este hecho es la razón por la que las tablas de la normal estándar no proporcionan áreas para el lado izquierdo de la distribución. Debido a esta simetría, la fórmula de la puntuación Z se escribe a veces como
Las líneas verticales de la ecuación significan el valor absoluto del número.
Lo que realmente hace la fórmula de estandarización es calcular el número de desviaciones típicas que tiene X respecto a la media de su propia distribución. La fórmula de estandarización y el concepto de contar desviaciones típicas de la media es el secreto de todo lo que haremos en esta clase de Estadística. La razón de esto es que toda la estadística se reduce a la variación, y el recuento de las desviaciones típicas es una medida de variación.
Esta fórmula, con muchas apariencias, reaparecerá una y otra vez a lo largo de este curso.
### Referencias
“Naegele’s rule”. Wikipedia. Disponible en línea en http://en.wikipedia.org/wiki/Naegele's_rule (consultado el 14 de mayo de 2013).
“403: NUMMI”. Chicago Public Media & Ira Glass, 2013. Disponible en línea en http://www.thisamericanlife.org/radio-archives/episode/403/nummi (consultado el 14 de mayo de 2013).
“Scratch-Off Lottery Ticket Playing Tips”. WinAtTheLottery.com, 2013. Disponible en línea en http://www.winatthelottery.com/public/department40.cfm (consultado el 14 de mayo de 2013).
“Smart Phone Users, By The Numbers”. Visual.ly, 2013. Disponible en línea en http://visual.ly/smart-phone-users-numbers (consultado el 14 de mayo de 2013).
“Facebook Statistics”. Statistics Brain. Disponible en línea en http://www.statisticbrain.com/facebook-statistics/ (consultado el 14 de mayo de 2013). |
# La distribución normal
## Estimación de la binomial con la distribución normal
Ya hemos visto que varias funciones de densidad de probabilidad son las distribuciones límite de otras; por tanto, podemos estimar una con otra en determinadas circunstancias. Aquí veremos que la distribución normal puede utilizarse para estimar un proceso binomial. La Poisson se utilizó para estimar la binomial previamente, y la binomial se utilizó para estimar la distribución hipergeométrica.
En el caso de la relación entre la distribución hipergeométrica y la binomial, tuvimos que reconocer que un proceso binomial asume que la probabilidad de un éxito permanece constante de un ensayo a otro: una cara en el último lanzamiento no puede tener un efecto en la probabilidad de una cara en el siguiente lanzamiento. En la distribución hipergeométrica esta es la esencia de la cuestión porque el experimento asume que cualquier "extracción" es sin reemplazo. Si se extrae sin reemplazo, todos las "extracciones" posteriores son probabilidades condicionales. Descubrimos que si el experimento hipergeométrico saca extrae un pequeño porcentaje del total de objetos, entonces podemos ignorar el impacto en la probabilidad de una extracción a otra.
Imagine que hay 312 cartas en una baraja compuesta por 6 mazos normales. Si el experimento exigía extraer solo 10 cartas, menos del 5% del total, entonces aceptaremos la estimación binomial de la probabilidad, aunque en realidad se trata de una distribución hipergeométrica porque las cartas se extraen presumiblemente sin reemplazo.
La Poisson también se consideró una estimación adecuada de la binomial en determinadas circunstancias. En el capítulo 4 encontramos que si el número de ensayos de interés es grande y la probabilidad de éxito es pequeña, tal que < , la Poisson puede utilizarse para estimar la binomial con buenos resultados. Una vez más, estas reglas empíricas no pretenden en modo alguno que la probabilidad real sea la que determina la estimación, sino que la diferencia está en el tercer o cuarto decimal y, por tanto, es de minimus.
Aquí, de nuevo, encontramos que la distribución normal hace estimaciones particularmente precisas de un proceso binomial bajo ciertas circunstancias. La es una distribución de frecuencia de un proceso binomial para el experimento de lanzar tres monedas donde la variable aleatoria es el número de caras. El espacio muestral se encuentra debajo de la distribución. En el experimento se ha asumido que la probabilidad de éxito es de 0,5; por tanto, la probabilidad de fracaso, de cola, es también de 0,5. Al observar la nos llama la atención que la distribución sea simétrica. La raíz de este resultado es que las probabilidades de éxito y de fracaso son las mismas, 0,5. Si la probabilidad de éxito fuera inferior a 0,5, la distribución se vuelve sesgada hacia la derecha. De hecho, a medida que la probabilidad de éxito disminuye, el grado de asimetría aumenta. Si la probabilidad de éxito aumenta a partir de 0,5, la asimetría aumenta en la cola inferior, lo que da lugar a una distribución sesgada a la izquierda.
La razón por la que la asimetría de la distribución binomial es importante es porque si se va a estimar con una distribución normal, entonces tenemos que reconocer que la distribución normal es simétrica. Cuanto más se acerque la distribución binomial subyacente a ser simétrica, mejor será la estimación producida por la distribución normal. La muestra una distribución normal simétrica transpuesta en un gráfico de una distribución binomial donde p = 0,2 y n = 5. La discrepancia entre la probabilidad estimada utilizando una distribución normal y la probabilidad de la distribución binomial original es evidente. El criterio para utilizar una distribución normal para estimar una binomial aborda, pues, este problema al exigir que TANTO np COMO n(1 - p) sean mayores que cinco. De nuevo, se trata de una regla general, pero es eficaz y da lugar a estimaciones aceptables de la probabilidad binomial.
### Repaso del capítulo
La distribución normal, que es continua, es la más importante de todas las distribuciones de probabilidad. Su gráfico tiene forma de campana. Esta curva en forma de campana se utiliza en casi todas las disciplinas. Al tratarse de una distribución continua, el área total debajo de la curva es uno. Los parámetros de la normal son la media µ y la desviación típica σ. Una distribución normal especial, llamada distribución normal estándar, es la distribución de las puntuaciones z. Su media es cero y su desviación típica es uno.
### Revisión de la fórmula
Distribución normal: X ~ N(µ, σ) donde µ es la media y σ es la desviación típica.
Distribución normal estándar: Z ~ N(0, 1).
Use la siguiente información para responder los próximos cuatro ejercicios:
X ~ N(54, 8)
### Tarea para la casa
Use la siguiente información para responder los dos próximos ejercicios: el tiempo de recuperación del paciente de un procedimiento quirúrgico en particular se distribuye normalmente, con una media de 5,3 días y una desviación típica de 2,1 días.
Use la siguiente información para responder los próximos tres ejercicios: El tiempo que se tarda en encontrar un puesto de estacionamiento a las 9 a. m. sigue una distribución normal con una media de cinco minutos y una desviación típica de dos minutos. |
# El teorema del límite central
## Introducción
¿Por qué nos preocupan tanto las medias? Hay dos razones: nos dan un punto medio de comparación y son fáciles de calcular. En este capítulo, estudiará las medias y el teorema del límite central.
El teorema del límite central es una de las ideas más poderosas y útiles de toda la estadística. El teorema central del límite es un teorema, lo que significa que NO es una teoría o simplemente la idea de alguien sobre cómo funcionan las cosas. Como teorema, está a la altura del teorema de Pitágoras, o el teorema que nos dice que la suma de los ángulos de un triángulo debe sumar 180. Son hechos de las formas del mundo rigurosamente demostrados con precisión matemática y lógica. Como veremos, este poderoso teorema determinará lo que podemos y no podemos decir en la estadística inferencial. El teorema del límite central se ocupa de extraer muestras finitas de tamaño n de una población con una media conocida, μ, y una desviación típica conocida, σ. La conclusión es que si recogemos muestras de tamaño n con un "n suficientemente grande", calculamos la media de cada muestra y creamos un histograma (distribución) de esas medias, la distribución resultante tenderá a tener una distribución normal aproximada.
El resultado asombroso es que no importa cuál es la distribución de la población original, ni siquiera es necesario conocerla. El hecho importante es que la distribución de las medias muestrales tiende a seguir la distribución normal.
El tamaño de la muestra, n, que se requiere para ser “suficientemente grande” depende de la población original de la que se extraen las muestras (el tamaño de la muestra debe ser, al menos, 30 o los datos deben proceder de una distribución normal). Si la población original está lejos de la normal, se necesitan más observaciones para las medias de la muestra. El muestreo se realiza de forma aleatoria y con reemplazo en el modelo teórico. |
# El teorema del límite central
## Teorema del límite central de las medias muestrales
La distribución muestral es una distribución teórica. Se crea tomando muchas muestras de tamaño n de una población. Cada media muestral se trata entonces como una única observación de esta nueva distribución, la distribución muestral. La genialidad de pensar de esta manera es que reconoce que cuando tomamos una muestra estamos creando una observación y esa observación debe provenir de alguna distribución particular. El teorema del límite central responde a la pregunta: ¿de qué distribución procede la media de una muestra? Si se descubre esto, entonces podemos tratar una media muestral como cualquier otra observación y calcular probabilidades sobre los valores que puede tomar. En efecto, hemos pasado del mundo de la estadística, en el que solo conocemos lo que tenemos de la muestra, al mundo de la probabilidad, en el que conocemos la distribución de la que procede la media muestral y los parámetros de esa distribución.
Las razones por las que se toma una muestra de una población son obvias. El tiempo y el gasto de comprobar cada factura para determinar su validez o cada envío para ver si contiene todos los artículos puede superar con creces el coste de los errores de facturación o envío. En el caso de algunos productos, el muestreo requeriría su destrucción, lo que se denomina muestreo destructivo. Un ejemplo de ello es la medición de la capacidad de un metal para resistir la corrosión del agua salada en las piezas de los buques oceánicos.
El muestreo plantea, pues, una cuestión importante: qué muestra se ha extraído. Incluso si la muestra se extrajera al azar, en teoría hay un número casi infinito de muestras. Con solo 100 artículos, se pueden extraer más de 75 millones de muestras únicas de tamaño cinco. Si hay seis en la muestra, el número de muestras posibles aumenta a algo más de mil millones. Entonces, de los 75 millones de muestras posibles, ¿cuál obtuvo? Si hay variación en los artículos que se van a muestrear, habrá variación en las muestras. Se podría extraer una muestra "desafortunada" y sacar conclusiones muy erróneas sobre la población. Este reconocimiento de que cualquier muestra que extraigamos es en realidad solo una de una distribución de muestras nos proporciona el que probablemente sea el teorema más importante de la estadística: el teorema del límite central. Sin el teorema del límite central sería imposible pasar a la estadística inferencial a partir de la teoría de la probabilidad simple. En su forma más básica, el teorema del límite central establece que, independientemente de la función de densidad de probabilidad subyacente de los datos de la población, la distribución teórica de las medias de las muestras de la población se distribuirá normalmente. En esencia, esto dice que la media de una muestra debe tratarse como una observación extraída de una distribución normal. El Teorema del Límite Central solo se cumple si el tamaño de la muestra es "suficientemente grande", lo que se ha demostrado que es de solo 30 observaciones o más.
La muestra gráficamente esta importante propuesta.
Observe que el eje horizontal del panel superior está etiquetado como X. Se trata de las observaciones individuales de la población. Esta es la distribución desconocida de los valores de la población. El gráfico está dibujado a propósito de forma cuadriculada para mostrar que no importa lo extraña que sea en realidad. Recuerde que nunca sabremos cómo es esta distribución, ni su media ni su desviación típica.
El eje horizontal del panel inferior está etiquetado como
's. Se trata de la distribución teórica denominada distribución muestral de las medias. Cada observación de esta distribución es una media muestral. Todas estas medias muestrales se calcularon a partir de muestras individuales con el mismo tamaño de muestra. La distribución muestral teórica contiene todos los valores medios muestrales de todas las muestras posibles que podrían haberse tomado de la población. Por supuesto, nadie tomaría realmente todas estas muestras, pero si lo hicieran, este es el aspecto que tendrían. Y el teorema del límite central dice que se distribuirán normalmente.
El teorema del límite central va más allá y nos indica la media y la desviación típica de esta distribución teórica.
La importancia práctica del teorema del límite central es que ahora podemos calcular las probabilidades de obtener una media muestral,
, de la misma manera que lo hicimos para extraer observaciones específicas, X's, cuando conocíamos la media y la desviación típica de la población y que los datos de la población estaban distribuidos normalmente. La fórmula de estandarización tiene que modificarse para reconocer que la media y la desviación típica de la distribución muestral, a veces llamada error estándar de la media, son diferentes de las de la distribución de la población, pero por lo demás no ha cambiado nada. La nueva fórmula de estandarización es
Observe que en la primera fórmula se ha cambiado por simplemente µ en la segunda versión. La razón es que matemáticamente se puede demostrar que el valor esperado de es igual a µ. Esto se ha indicado en la anteriormente. Matemáticamente, el símbolo E(x) indica el "valor esperado de x". Esta fórmula se utilizará en la siguiente unidad para proporcionar estimaciones del parámetro poblacional desconocido μ.
### Referencias
Baran, Daya. “20 Percent of Americans Have Never Used Email.”WebGuild, 2010. Disponible en línea en http://www.webguild.org/20080519/20-percent-of-americans-have-never-used-email (consultado el 17 de mayo de 2013).
Datos de The Flurry Blog, 2013. Disponible en línea en http://blog.flurry.com (consultado el 17 de mayo de 2013).
Datos del Departamento de Agricultura de Estados Unidos.
### Repaso del capítulo
En una población cuya distribución puede ser conocida o desconocida, si el tamaño (n) de las muestras es suficientemente grande, la distribución de las medias muestrales será aproximadamente normal. La media de las medias muestrales será igual a la media poblacional. La desviación típica de la distribución de las medias muestrales, denominada error estándar de la media, es igual a la desviación típica de la población dividida entre la raíz cuadrada del tamaño de la muestra (n).
### Revisión de la fórmula
El teorema del límite central para las medias muestrales:
~ N
La media
Teorema del límite central de las medias muestrales de puntuación z
Error estándar de la media (desviación típica (
)):
Factor de corrección de la población finita para la distribución muestral de las medias:
Factor de corrección de la población finita para la distribución muestral de las proporciones:
### Tarea para la casa
|
# El teorema del límite central
## Uso del teorema del límite central
### Ejemplos del teorema del límite central
### Ley de los grandes números
La ley de los grandes números dice que si se toman muestras cada vez más grandes de cualquier población, entonces la media de la distribución muestral, tiende a acercarse cada vez más a la verdadera media de la población, μ. A partir del teorema del límite central, sabemos que a medida que n se hace más grande, las medias muestrales siguen una distribución normal. Cuanto mayor sea n, menor será la desviación típica de la distribución muestral. (Recuerde que la desviación típica de la distribución muestral de
es
). Esto significa que la media muestral
debe estar más cerca de la media poblacional μ a medida que n aumenta. Podemos decir que μ es el valor al que se acercan las medias muestrales a medida que n es mayor. El teorema del límite central ilustra la ley de los grandes números.
Este concepto es tan importante y desempeña un papel tan decisivo en lo que sigue que merece un mayor desarrollo. De hecho, hay dos cuestiones críticas que se derivan del teorema del límite central y de la aplicación de la ley de los grandes números a este. Estos son
1. La función de densidad de probabilidad de la distribución muestral de las medias se distribuye normalmente independientemente de la distribución subyacente de las observaciones de la población y
2. la desviación típica de la distribución muestral disminuye a medida que aumenta el tamaño de las muestras que se utilizaron para calcular las medias de la distribución muestral.
Tomando estos en orden. Parece contradictorio que la población pueda tener cualquier distribución y que la distribución de las medias procedentes de ella se distribuya normalmente. Con el uso de computadoras, se pueden simular experimentos que muestren el proceso por el cual la distribución de muestreo cambia a medida que aumenta el tamaño de la muestra. Estas simulaciones muestran visualmente los resultados de la demostración matemática del teorema del límite central.
He aquí tres ejemplos de distribuciones poblacionales muy diferentes y la evolución de la distribución muestral hacia una distribución normal a medida que aumenta el tamaño de la muestra. El panel superior en estos casos representa el histograma de los datos originales. Los tres paneles muestran los histogramas de 1000 muestras extraídas al azar para diferentes tamaños de muestra: n=10, n= 25 y n=50. A medida que aumenta el tamaño de la muestra, y el número de muestras tomadas se mantiene constante, la distribución de las medias de 1000 muestras se acerca más a la línea suave que representa la distribución normal.
La es para una distribución normal de las observaciones individuales y esperaríamos que la distribución de muestreo convergiera en la normal rápidamente. Los resultados lo demuestran y muestran que, incluso con un tamaño de muestra muy pequeño, la distribución se aproxima a la distribución normal.
La es una distribución uniforme que, de forma un poco sorprendente, se acerca rápidamente a la distribución normal incluso con solo una muestra de 10.
La es una distribución sesgada. Esta última podría ser una exponencial, geométrica o binomial con una pequeña probabilidad de éxito creando el sesgo en la distribución. En el caso de las distribuciones asimétricas, nuestra intuición nos dice que se necesitarán tamaños de muestra mayores para pasar a una distribución normal y de hecho, eso es lo que observamos en la simulación. Sin embargo, con un tamaño de muestra de 50, que no se considera muy grande, la distribución de las medias muestrales ha adquirido muy decididamente la forma de la distribución normal.
El teorema del límite central proporciona algo más que la prueba de que la distribución muestral de las medias se distribuye normalmente. También nos proporciona la media y la desviación típica de esta distribución. Además, como se ha comentado anteriormente, el valor esperado de la media, , es igual a la media de la población de los datos originales que es lo que nos interesa estimar a partir de la muestra que tomamos. Ya hemos insertado esta conclusión del teorema del límite central en la fórmula que utilizamos para estandarizar desde la distribución muestral a la distribución normal estándar. Y, por último, el teorema del límite central también ha proporcionado la desviación típica de la distribución muestral, , y esto es crítico para poder calcular las probabilidades de los valores de la nueva variable aleatoria, .
La muestra una distribución de muestreo. La media se ha marcado en el eje horizontal de las y la desviación típica se ha escrito a la derecha sobre la distribución. Observe que la desviación típica de la distribución muestral es la desviación típica original de la población, dividida entre el tamaño de la muestra. Ya hemos visto que, a medida que aumenta el tamaño de la muestra, la distribución muestral se acerca cada vez más a la distribución normal. Como esto ocurre, la desviación típica de la distribución muestral cambia de otra manera; la desviación típica disminuye a medida que n aumenta. Cuando n es muy grande, la desviación típica de la distribución muestral se hace muy pequeña y en el infinito colapsa sobre la media de la población. Esto es lo que significa que el valor esperado de es la media de la población, µ.
En valores no extremos de n, esta relación entre la desviación típica de la distribución muestral y el tamaño de la muestra desempeña un papel muy importante en nuestra capacidad para estimar los parámetros que nos interesan.
La muestra tres distribuciones de muestreo. El único cambio que se ha realizado es el tamaño de la muestra que se utilizó para obtener las medias muestrales de cada distribución. A medida que aumenta el tamaño de la muestra, n pasa de 10 a 30 a 50, las desviaciones típicas de las respectivas distribuciones muestrales disminuyen porque el tamaño de la muestra está en el denominador de las desviaciones típicas de las distribuciones muestrales.
Las implicaciones de esto son muy importantes. La muestra el efecto del tamaño de la muestra en la confianza que tendremos en nuestras estimaciones. Se trata de dos distribuciones muestrales de la misma población. Una distribución de muestreo se creó con muestras de tamaño 10 y la otra con muestras de tamaño 50. Si todo lo demás es constante, la distribución de muestreo con un tamaño de muestra de 50 tiene una desviación típica menor que hace que el gráfico sea más alto y estrecho. El efecto importante de esto es que para la misma probabilidad de una desviación típica de la media, esta distribución cubre mucho menos rango de valores posibles que la otra distribución. Una desviación típica está marcada en el eje para cada distribución. Esto se muestra con las dos flechas que son más o menos una desviación típica para cada distribución. Si la probabilidad de que la verdadera media esté a una desviación típica de la media, entonces para la distribución de muestreo con el tamaño de muestra más pequeño, el rango posible de valores es mucho mayor. Una pregunta sencilla es: ¿preferiría tener una media muestral de la distribución estrecha y ajustada o de la distribución plana y amplia como estimación de la media de la población? Su respuesta nos dice por qué la gente intuitivamente siempre elegirá datos de una muestra grande en lugar de una muestra pequeña. La media muestral que obtienen procede de una distribución más compacta. Este concepto será la base de lo que se llamará nivel de confianza en la siguiente unidad.
### Repaso del capítulo
El teorema del límite central puede utilizarse para ilustrar la ley de los grandes números. La ley de los grandes números establece que cuanto mayor sea el tamaño de la muestra que se tome de una población, más se acercará la media muestral
llega a μ.
Use la siguiente información para responder los próximos diez ejercicios: un fabricante produce pesas de 25 libras. El peso real más bajo es de 24 libras, y el más alto de 26 libras. Cada pesa tiene la misma probabilidad, por lo que la distribución de los pesos es uniforme. Se toma una muestra de 100 pesas.
La duración de la batería de un determinado teléfono inteligente sigue una distribución exponencial con una media de diez meses. Se toma una muestra de 64 de estos teléfonos inteligentes.
una distribución uniforme tiene un mínimo de seis y un máximo de diez. Se toma una muestra de 50 personas.
### Tarea para la casa
|
# El teorema del límite central
## Teorema del límite central de las proporciones
El teorema del límite central nos dice que la estimación puntual de la media muestral, , proviene de una distribución normal de 's. Esta distribución teórica se denomina distribución muestral de 's. Ahora investigamos la distribución de muestreo para otro parámetro importante que deseamos estimar; p de la función de densidad de probabilidad binomial.
Si la variable aleatoria es discreta, como en el caso de los datos categóricos, el parámetro que deseamos estimar es la proporción de la población. Esta es, por supuesto, la probabilidad de obtener un éxito en cualquier sorteo aleatorio. A diferencia del caso que acabamos de discutir para una variable aleatoria continua en la que no conocíamos la distribución poblacional de las X, aquí sí conocemos la función de densidad de probabilidad subyacente para estos datos; es la binomial. La variable aleatoria es X = el número de aciertos y el parámetro que deseamos conocer es p, la probabilidad de sacar un acierto que es, por supuesto, la proporción de aciertos en la población. La pregunta que se plantea es: ¿a partir de qué distribución se obtuvo la proporción de la muestra, extraída? El tamaño de la muestra es n y X es el número de aciertos encontrados en esa muestra. Se trata de una pregunta paralela a la que acaba de responder el teorema del límite central: ¿de qué distribución era la media de la muestra, , extraída? Vimos que una vez que supimos que la distribución era la normal, pudimos crear intervalos de confianza para el parámetro poblacional: µ. También utilizaremos esta misma información para comprobar las hipótesis sobre la media de la población más adelante. Ahora queremos ser capaces de desarrollar intervalos de confianza para el parámetro poblacional "p" a partir de la función de densidad de probabilidad binomial.
Para hallar la distribución de la que proceden las proporciones muestrales, necesitamos desarrollar la distribución muestral de las proporciones muestrales, al igual que hicimos con las medias muestrales. Imaginemos de nuevo que tomamos una muestra aleatoria de, por ejemplo, 50 personas y les preguntamos si apoyan la nueva emisión de bonos escolares. A partir de esto encontramos una proporción muestral, p', y la graficamos en el eje de las p'. Hacemos esto una y otra vez, etc., hasta que tengamos la distribución teórica de las p'. Algunas proporciones de la muestra presentarán una alta favorabilidad hacia la emisión de bonos y otras presentarán una baja favorabilidad porque el muestreo aleatorio reflejará la variación de opiniones dentro de la población. Lo que hemos hecho puede verse en la . El panel superior es la distribución poblacional de probabilidades para cada valor posible de la variable aleatoria X. Aunque no sabemos cómo es la distribución específica porque no conocemos p, el parámetro poblacional, sí sabemos que debe ser algo así. En realidad, no conocemos ni la media ni la desviación típica de esta distribución de la población, la misma dificultad a la que nos enfrentamos al analizar las X anteriormente.
La sitúa la media en la distribución de probabilidades de la población como pero, por supuesto, no conocemos realmente la media de la población porque no conocemos la probabilidad de éxito de la población, . Debajo de la distribución de los valores de la población se encuentra la distribución muestral de 's. De nuevo, el teorema del límite central nos dice que esta distribución se distribuye normalmente al igual que el caso de la distribución muestral para 's. Esta distribución muestral también tiene una media, la media de ', y una desviación típica, .
Es importante destacar que, en el caso del análisis de la distribución de las medias muestrales, el teorema del límite central nos indicó el valor esperado de la media de las medias muestrales en la distribución muestral, y la desviación típica de la distribución muestral. De nuevo, el teorema del límite central proporciona esta información para la distribución de muestreo de las proporciones. Las respuestas son
Estas dos conclusiones son las mismas que hemos encontrado para la distribución de muestreo de las medias de las muestras. Sin embargo, en este caso, como la media y la desviación típica de la distribución binomial dependen de , la fórmula de la desviación típica de la distribución muestral requiere una manipulación algebraica para ser útil. Lo abordaremos en el próximo capítulo. A continuación, se ofrece la demostración de estas importantes conclusiones del teorema del límite central.
(El valor esperado de X, E(x), es simplemente la media de la distribución binomial que sabemos que es np).
La desviación típica de la distribución muestral de las proporciones es, por tanto, la siguiente
La resume estos resultados y muestra la relación entre la población, la muestra y la distribución muestral. Nótese el paralelismo entre esta Tabla y la Tabla 7.1 para el caso en que la variable aleatoria es continua y estábamos desarrollando la distribución muestral para las medias.
Repasando la fórmula de la desviación típica de la distribución muestral para las proporciones vemos que a medida que n aumenta la desviación típica disminuye. Esta es la misma observación que hicimos para la desviación típica de la distribución de muestreo para las medias. De nuevo, a medida que aumenta el tamaño de la muestra, se observa que la estimación puntual de µ o p procede de una distribución cada vez más estrecha. Llegamos a la conclusión de que, con un nivel de probabilidad determinado, el rango del que procede la estimación puntual es menor a medida que aumenta el tamaño de la muestra, n. La figura 7.8 muestra este resultado para el caso de las medias muestrales. Simplemente sustituya por y podemos ver el impacto del tamaño de la muestra en la estimación de la proporción de la muestra.
### Repaso del capítulo
El teorema del límite central también puede utilizarse para ilustrar que la distribución muestral de las proporciones de la muestra se distribuye normalmente con el valor esperado de p y una desviación típica de
### Tarea para la casa
|
# El teorema del límite central
## Factor de corrección de población finita
Hemos visto que el tamaño de la muestra tiene un efecto importante en la varianza y, por tanto, en la desviación típica de la distribución muestral. También es interesante la proporción de la población total que ha sido muestreada. Hemos asumido que la población es extremadamente grande y que hemos muestreado una pequeña parte de ella. A medida que la población se hace más pequeña y muestreamos un mayor número de observaciones, las observaciones de la muestra no son independientes entre sí. Para corregir el impacto de esto, se puede utilizar el factor de corrección finito para ajustar la varianza de la distribución de muestreo. Es apropiado cuando se muestrea más del 5 % de la población y esta tiene un tamaño poblacional conocido. Hay casos en los que se conoce la población, por lo que hay que aplicar el factor de corrección. El problema se plantea tanto para la distribución muestral de las medias como para la distribución muestral de las proporciones. El factor de corrección de la población finita para la varianza de las medias que aparece en la fórmula de normalización es:
y para la varianza de las proporciones es:
Los siguientes ejemplos muestran cómo aplicar el factor. Las varianzas muestrales se ajustan mediante la fórmula anterior.
### Tarea para la casa
|
# Intervalos de confianza
## Introducción
Supongamos que intenta determinar el alquiler medio de un apartamento de dos habitaciones en su ciudad. Puede buscar en la sección de anuncios del periódico, anotar varios alquileres que aparezcan y hacer un promedio entre ellos. Habría obtenido una estimación puntual de la media real. Si intenta determinar el porcentaje de veces que encesta cuando lanza una pelota de baloncesto, puede contar el número de tiros que lo logra y dividirlo entre el número de tiros que intenta. En este caso, se habría obtenido una estimación puntual de la proporción verdadera del parámetro p en la función de densidad de probabilidad binomial.
Utilizamos los datos de la muestra para hacer generalizaciones sobre una población desconocida. Esta parte de la Estadística se llama Estadística Inferencial. Los datos de la muestra nos ayudan a hacer una estimación de un de la población. Nos damos cuenta de que lo más probable es que la estimación puntual no sea el valor exacto del parámetro poblacional, sino que se acerque a él. Después de calcular las estimaciones puntuales, construimos las estimaciones de intervalo, llamadas intervalos de confianza. Lo que la estadística nos proporciona, más allá de un simple promedio o estimación puntual, es una estimación a la que podemos atribuir una probabilidad de exactitud, lo que llamaremos un nivel de confianza. Hacemos inferencias con un nivel de probabilidad conocido.
En este capítulo aprenderá a construir e interpretar intervalos de confianza. También aprenderá una nueva distribución, la t de Student, y cómo se utiliza con estos intervalos. A lo largo del capítulo es importante tener en cuenta que el intervalo de confianza es una variable aleatoria. Es el parámetro poblacional que se fija.
Si usted trabajara en el departamento de mercadeo de una compañía de entretenimiento, podría interesarse por el número medio de canciones que un consumidor descarga al mes de iTunes. Si es así, puede hacer una encuesta y calcular la media muestral,
, y la desviación típica de la muestra, s. Usaría
para estimar la media de la población y s para estimar la desviación típica de la población. La media muestral,
, es la estimación puntual de la media de la población, μ. La desviación típica de la muestra, s, es la estimación puntual de la desviación típica de la población, σ.
y s se denominan cada uno una estadística.
Un intervalo de confianza es otro tipo de estimación pero, en vez de ser un solo número, es un intervalo de números. El intervalo de números es un rango de valores calculado a partir de un conjunto determinado de datos de muestra. Es probable que el intervalo de confianza incluya el parámetro poblacional desconocido.
Supongamos, para el ejemplo de iTunes, que no conocemos la media poblacional μ, pero sí sabemos que la desviación típica de la población es σ = 1 y que nuestro tamaño de muestra es 100. Entonces, por el teorema del límite central, la desviación típica de la distribución muestral de las medias de la muestra es
.
La regla empírica, que se aplica a la distribución normal, dice que en aproximadamente el 95 % de las muestras, la media muestral,
, estará dentro de las dos desviaciones típicas de la media poblacional μ. Para nuestro ejemplo de iTunes, dos desviaciones típicas son (2)(0,1) = 0,2. La media muestral
es probable que esté dentro de 0,2 unidades de μ.
Dado que
está dentro de 0,2 unidades de μ, que es desconocido, entonces es probable que μ esté dentro de 0,2 unidades de
con un 95 % de probabilidad. La media poblacional μ está contenida en un intervalo cuyo número inferior se calcula tomando la media muestral y restando dos desviaciones típicas (2)(0,1) y cuyo número superior se calcula tomando la media muestral y sumando dos desviaciones típicas. En otras palabras, μ está entre
y
en el 95 % de las muestras.
Para el ejemplo de iTunes, supongamos que una muestra produce una media muestral
. Entonces con un 95 % de probabilidad la media poblacional desconocida μ está entre
y
Decimos que tenemos un 95 % de confianza en que la media de la población desconocida de canciones descargadas de iTunes al mes está entre 1,8 y 2,2. El intervalo de confianza del 95 % es (1,8; 2,2). Tenga en cuenta que hablamos en términos de confianza del 95 % utilizando la regla empírica. La regla empírica para dos desviaciones típicas es solo aproximadamente el 95 % de la probabilidad bajo la distribución normal. Para ser precisos, dos desviaciones típicas en una distribución normal son en realidad el 95,44 % de la probabilidad. Para calcular el nivel de confianza exacto del 95 % utilizaríamos 1,96 desviaciones típicas.
El intervalo de confianza del 95 % implica dos posibilidades. O bien el intervalo (1,8, 2,2) contiene la verdadera media μ, o bien nuestra muestra produjo un
que no esté a menos de 0,2 unidades de la media verdadera μ. La segunda posibilidad solo se da en el 5 % de todas las muestras (95 % menos 100 % = 5 %).
Recuerde que un intervalo de confianza se crea para un parámetro poblacional desconocido como la media poblacional, μ.
Para el intervalo de confianza de una media la fórmula sería
O escrito de otra manera como:
Donde es la media de la muestra. se determina por el nivel de confianza deseado por el analista, y es la desviación típica de la distribución muestral para las medias que nos da el teorema del límite central. |
# Intervalos de confianza
## Un intervalo de confianza para una desviación típica de la población, con un tamaño de muestra conocido o grande
Un intervalo de confianza para una media poblacional con una desviación típica poblacional conocida se basa en la conclusión del teorema del límite central de que la distribución muestral de las medias muestrales sigue una distribución aproximadamente normal.
### Cálculo del intervalo de confianza
Considere la fórmula de estandarización para la distribución de muestreo desarrollada en la discusión del Teorema del Límite Central:
Observe que µ se sustituye por porque sabemos que el valor esperado de es µ del teorema del límite central y se sustituye por , también del teorema del límite central.
En esta fórmula sabemos , y n, el tamaño de la muestra. (En realidad, no conocemos la desviación típica de la población, pero tenemos una estimación puntual de la misma, s, a partir de la muestra que hemos tomado. Más adelante se hablará de esto). Lo que no sabemos es μ o Z1. Podemos resolver cualquiera de ellas en términos de la otra. Resolviendo para μ en términos de Z1 se obtiene:
Recordando que el teorema del límite central nos dice que la distribución de 's, la distribución muestral para las medias, es normal, y que la distribución normal es simétrica, podemos reordenar los términos así:
Esta es la fórmula de un intervalo de confianza para la media de una población.
Observe que Zα ha sido sustituido por Z1 en esta ecuación. Aquí es donde el estadístico debe hacer una elección. El analista debe decidir el nivel de confianza que desea imponer al intervalo de confianza. α es la probabilidad de que el intervalo no contenga la verdadera media de la población. El nivel de confianza se define como (1-α). Zα es el número de desviaciones típicas que se aleja de la media con una cierta probabilidad. Si elegimos Zα = 1,96, estamos pidiendo el intervalo de confianza del 95 % porque estamos fijando en 0,95 la probabilidad de que la verdadera media se encuentre dentro del rango. Si fijamos Zα en 1,64, estamos pidiendo el intervalo de confianza del 90% porque hemos fijado la probabilidad en 0,90. Estos números pueden verificarse consultando la tabla estandarizada. Divida 0,95 o 0,90 por la mitad y encuentra esa probabilidad dentro del cuerpo de la tabla. A continuación, lea en los márgenes superior e izquierdo el número de desviaciones típicas que se necesitan para obtener este nivel de probabilidad.
En realidad, podemos establecer cualquier nivel de confianza que deseemos simplemente cambiando el valor Zα en la fórmula. Es la elección del analista. La convención común en economía y en la mayoría de las ciencias sociales establece los intervalos de confianza en niveles del 90, 95 o 99 por ciento. Los niveles inferiores al 90% se consideran de poco valor. El nivel de confianza de una determinada estimación de intervalo se denomina (1-α).
Una buena forma de ver el desarrollo de un intervalo de confianza es representar gráficamente la solución de un problema solicitando un intervalo de confianza. Esto se presenta en la para el ejemplo de la introducción relativo al número de descargas de iTunes. Ese caso era para un intervalo de confianza del 95%, pero se podrían haber elegido otros niveles de confianza con la misma facilidad, según la necesidad del analista. Sin embargo, el nivel de confianza DEBE estar preestablecido y no estar sujeto a revisión como resultado de los cálculos.
Para este ejemplo, digamos que sabemos que el número de la media poblacional real de descargas de iTunes es de 2,1. La verdadera media de la población se encuentra dentro del rango del intervalo de confianza del 95%. No hay absolutamente nada que garantice que esto ocurra. Además, si la verdadera media queda fuera del intervalo, nunca la conoceremos. Debemos recordar siempre que nunca conoceremos la verdadera media. La estadística simplemente nos permite, con un determinado nivel de probabilidad (confianza), decir que la verdadera media está dentro del rango calculado. Esto es lo que se llamó en la introducción, el "nivel de ignorancia admitido".
### Modificación del nivel de confianza o del tamaño de la muestra
Aquí está de nuevo la fórmula para un intervalo de confianza para una media poblacional desconocida asumiendo que conocemos la desviación típica de la población:
Está claro que el intervalo de confianza se rige por dos cosas, el nivel de confianza elegido, , y la desviación típica de la distribución muestral. La desviación típica de la distribución muestral se ve afectada además por dos cosas, la desviación típica de la población y el tamaño de la muestra que hemos elegido para nuestros datos. Aquí queremos examinar los efectos de cada una de las elecciones que hemos hecho sobre el intervalo de confianza calculado, el nivel de confianza y el tamaño de la muestra.
Por un momento debemos preguntarnos qué deseamos en un intervalo de confianza. Nuestro objetivo era estimar la media de la población a partir de una muestra. Hemos abandonado la esperanza de encontrar alguna vez la verdadera media de la población, y la desviación típica de la población, para cualquier caso, excepto cuando tenemos una población extremadamente pequeña y el coste de recopilar los datos de interés es muy pequeño. En todos los demás casos, debemos recurrir a las muestras. Con el teorema del límite central tenemos las herramientas para proporcionar un intervalo de confianza significativo con un nivel de confianza determinado, lo que significa una probabilidad conocida de estar equivocado. Por intervalo de confianza significativo entendemos uno que sea útil. Imagine que le piden un intervalo de confianza para las edades de sus compañeros. Ha tomado una muestra y encuentra una media de 19,8 años. Desea estar muy seguro, por lo que informa de un intervalo entre 9,8 años y 29,8 años. Este intervalo contendría sin duda la verdadera media de la población y tendría un nivel de confianza muy alto. Sin embargo, difícilmente puede calificarse de significativo. El mejor intervalo de confianza es el que es estrecho y a la vez de alta confianza. Existe una tensión natural entre estos dos objetivos. Cuanto más alto sea el nivel de confianza, más amplio será el intervalo de confianza, como en el caso de las edades de los estudiantes. Podemos ver esta tensión en la ecuación del intervalo de confianza.
El intervalo de confianza aumentará el ancho a medida que aumenta, aumenta a medida que aumenta el nivel de confianza. Existe un compromiso entre el nivel de confianza y el ancho del intervalo. Ahora volvamos a ver la fórmula y veremos que el tamaño de la muestra también juega un papel importante en el ancho del intervalo de confianza. El tamaño de la muestra, , aparece en el denominador de la desviación típica de la distribución muestral. A medida que aumenta el tamaño de la muestra, disminuye la desviación típica de la distribución muestral y, por tanto, el ancho del intervalo de confianza, manteniendo constante el nivel de confianza. Esta relación se demostró en la . Una vez más, vemos la importancia de contar con muestras grandes para nuestro análisis, aunque entonces nos enfrentamos a una segunda limitación, el coste de la recopilación de datos.
### Cálculo del intervalo de confianza: un enfoque alternativo
Otra forma de enfocar los intervalos de confianza es mediante el uso de algo llamado límite de error. El límite de error recibe su nombre del reconocimiento de que proporciona el límite del intervalo derivado del error estándar de la distribución muestral. En las ecuaciones anteriores se ve que el intervalo es simplemente la media estimada, la media muestral, más o menos algo. Ese algo es el límite de error y está impulsado por la probabilidad que deseamos mantener en nuestra estimación, , por la desviación típica de la distribución muestral. El límite de error de una media recibe el nombre de media con límite de error (Error Bound Mean, EBM).
Para construir un intervalo de confianza para una única media poblacional desconocida μ, cuando se conoce la desviación típica de la población, necesitamos
como una estimación de μ y necesitamos el margen de error. Aquí, el margen de error (EBM) se denomina límite de error para una media poblacional (abreviado ). La media muestral
es la estimación puntual de la media poblacional desconocida μ.
La estimación del intervalo de confianza tendrá la forma:
(estimación puntual – límite de error, estimación puntual + límite de error) o, en símbolos, (
)
La fórmula matemática de este intervalo de confianza es:
El margen de error (EBM) depende del nivel de confianza (). El nivel de confianza suele considerarse la probabilidad de que la estimación del intervalo de confianza calculado contenga el verdadero parámetro poblacional. Sin embargo, es más preciso afirmar que el nivel de confianza es el porcentaje de intervalos de confianza que contienen el verdadero parámetro de la población cuando se toman muestras repetidas. La mayoría de las veces, la persona que construye el intervalo de confianza elige un nivel de confianza del 90 % o superior porque quiere estar razonablemente segura de sus conclusiones.
Existe otra probabilidad llamada alfa (α). α está relacionada con el nivel de confianza, CL. α es la probabilidad de que el intervalo no contenga el parámetro poblacional desconocido. Matemáticamente, 1 - α = CL.
Un intervalo de confianza para una media poblacional con una desviación típica conocida se basa en que la distribución muestral de las medias de la muestra sigue una distribución aproximadamente normal. Supongamos que nuestra muestra tiene una media de
= 10, y hemos construido el intervalo de confianza del 90 % (5, 15) donde EBM = 5.
Para obtener un intervalo de confianza del 90 %, debemos incluir el 90 % central de la probabilidad de la distribución normal. Si incluimos el 90 % central, dejamos fuera un total de α = 10 % en ambas colas, o 5 % en cada cola, de la distribución normal.
Para captar el 90% central, debemos movernos 1,645 desviaciones típicas a cada lado de la media muestral calculada. El valor 1,645 es la puntuación z de una distribución de probabilidad normal estándar que sitúa un área de 0,90 en el centro, un área de 0,05 en la cola extrema izquierda y un área de 0,05 en la cola extrema derecha.
Es importante que la desviación típica utilizada debe ser la adecuada para el parámetro que estamos estimando, por lo que en este apartado debemos utilizar la desviación típica que se aplica a la distribución muestral para medias que estudiamos con el teorema del límite central y es,
.
### Cálculo del intervalo de confianza con el EMB
Para construir una estimación de intervalo de confianza para una media poblacional desconocida necesitamos datos de una muestra aleatoria. Los pasos para construir e interpretar el intervalo de confianza son:
1. Calcular la media muestral
de los datos de la muestra. Recuerde que en esta sección conocemos la desviación típica de la población σ.
2. Calcule la puntuación z de la tabla estandarizada que corresponde al nivel de confianza deseado.
3. Calcular el límite de error EBM.
4. Construir el intervalo de confianza.
5. Escriba una oración que interprete la estimación en el contexto de la situación del problema.
Primero examinaremos cada paso con más detalle y luego ilustraremos el proceso con algunos ejemplos.
### Calcular la puntuación z para el nivel de confianza declarado
Cuando conocemos la desviación típica de la población σ, utilizamos una distribución normal estándar para calcular el EBM y construir el intervalo de confianza. Necesitamos hallar el valor de z que pone un área igual al nivel de confianza (en forma decimal) en el centro de la distribución normal estándar Z ~ N(0, 1).
El nivel de confianza, CL, es el área en el medio de la distribución normal estándar. CL = 1 – α, por lo que α es el área que se divide por igual entre las dos colas. Cada una de las colas contiene un área igual a
.
La puntuación z que tiene un área a la derecha de
se denota por
.
Por ejemplo, cuando CL = 0,95, α = 0,05 y
= 0,025; escribimos
= .
La zona a la derecha de es 0,025 y el área a la izquierda de es 1 – 0,025 = 0,975.
, utilizando una tabla de probabilidad normal. Más adelante veremos que podemos utilizar una tabla de probabilidad diferente, la distribución t de Student, para encontrar el número de desviaciones típicas de los niveles de confianza más utilizados.
### Cálculo del límite de error (EBM)
La fórmula del límite de error para una media poblacional desconocida μ cuando se conoce la desviación típica poblacional σ es
1. EBM =
### Construcción del intervalo de confianza
1. La estimación del intervalo de confianza tiene el formato
o la fórmula:
El gráfico da una idea de toda la situación.
CL +
+
= CL + α = 1.
Ya hemos visto este efecto cuando revisamos los efectos de cambiar el tamaño de la muestra, n, en el teorema del límite central. Consulte la para ver este efecto. Antes vimos que a medida que aumenta el tamaño de la muestra disminuye la desviación típica de la distribución muestral. Por eso elegimos una media muestral grande en comparación con la de una muestra pequeña, manteniendo el resto constante.
Hasta ahora hemos asumido que conocíamos la desviación típica de la población. Esto prácticamente nunca será así. Sin embargo, tendremos la desviación típica de la muestra, s. Se trata de una estimación puntual de la desviación típica de la población y puede sustituirse en la fórmula de los intervalos de confianza para una media en determinadas circunstancias. Acabamos de ver el efecto que tiene el tamaño de la muestra en el ancho del intervalo de confianza y el impacto en la distribución muestral para nuestra discusión del teorema del límite central. Podemos invocar esto para sustituir la estimación puntual por la desviación típica si el tamaño de la muestra es lo suficientemente grande. Los estudios de simulación indican que 30 observaciones o más serán suficientes para eliminar cualquier sesgo significativo en el intervalo de confianza estimado.
### Referencias
“American Fact Finder”. U.S. Census Bureau. Disponible en línea en http://factfinder2.census.gov/faces/nav/jsf/pages/searchresults.xhtml?refresh=t (consultado el 2 de julio de 2013).
“Disclosure Data Catalog: Candidate Summary Report 2012”. U.S. Federal Election Commission. Disponible en línea en http://www.fec.gov/data/index.jsp (consultado el 2 de julio de 2013).
“Headcount Enrollment Trends by Student Demographics Ten-Year Fall Trends to Most Recently Completed Fall”. Foothill De Anza Community College District. Disponible en línea en http://research.fhda.edu/factbook/FH_Demo_Trends/FoothillDemographicTrends.htm (consultado el 30 de septiembre de 2013).
Kuczmarski, Robert J., Cynthia L. Ogden, Shumei S. Guo, Laurence M. Grummer-Strawn, Katherine M. Flegal, Zuguo Mei, Rong Wei, Lester R. Curtin, Alex F. Roche, Clifford L. Johnson. “2000 CDC Growth Charts for the United States: Methods and Development”. Centers for Disease Control and Prevention. Disponible en línea en http://www.cdc.gov/growthcharts/2000growthchart-us.pdf (consultado el 2 de julio de 2013).
La, Lynn, Kent German. “Cell Phone Radiation Levels”. c|net parte de CBX Interactive Inc. Disponible en línea en http://reviews.cnet.com/cell-phone-radiation-levels/ (consultado el 2 de julio de 2013).
“Mean Income in the Past 12 Months (in 2011 Inflaction-Adjusted Dollars): 2011 American Community Survey 1-Year Estimates”. American Fact Finder, U.S. Census Bureau. Disponible en línea en http://factfinder2.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_11_1YR_S1902&prodType=table (consultado el 2 de julio de 2013).
“Metadata Description of Candidate Summary File”. U.S. Federal Election Commission. Disponible en línea en http://www.fec.gov/finance/disclosure/metadata/metadataforcandidatesummary.shtml (consultado el 2 de julio de 2013).
“National Health and Nutrition Examination Survey”. Centers for Disease Control and Prevention. Disponible en línea en http://www.cdc.gov/nchs/nhanes.htm (consultado el 2 de julio de 2013).
### Revisión de la fórmula
La forma general de un intervalo de confianza para una media poblacional única, desviación típica conocida, distribución normal, viene dada por Esta fórmula se utiliza cuando se conoce la desviación típica de la población.
CL = nivel de confianza, o la proporción de intervalos de confianza creados que se espera que contengan el verdadero parámetro poblacional
α = 1 – CL = la proporción de intervalos de confianza que no contendrán el parámetro poblacional
= la puntuación z con la propiedad de que el área a la derecha de la puntuación z es
esta puntuación z utilizada en el cálculo de “EBM donde α = 1 – CL. |
# Intervalos de confianza
## Un intervalo de confianza para una desviación típica de población desconocida, caso de una muestra pequeña
En la práctica, pocas veces conocemos la desviación típica de la población. En el pasado, cuando el tamaño de la muestra era grande, esto no suponía un problema para los estadísticos. Utilizaron la desviación típica de la muestra s como una estimación de σ y procedieron como antes para calcular un intervalo de confianza con resultados suficientemente cercanos. Esto es lo que hicimos en el arriba. La estimación puntual de la desviación típica, s, se sustituyó en la fórmula del intervalo de confianza para la desviación típica de la población. En este caso hay 80 observaciones muy por encima de las 30 sugeridas para eliminar cualquier sesgo de una muestra pequeña. Sin embargo, los estadísticos se encontraron con problemas cuando el tamaño de la muestra era pequeño. El pequeño tamaño de la muestra provocó imprecisiones en el intervalo de confianza.
William S. Goset (1876-1937), de la fábrica de cerveza Guinness de Dublín (Irlanda), se encontró con este problema. Sus experimentos con lúpulo y cebada produjeron muy pocas muestras. La simple sustitución de σ por s no produjo resultados precisos cuando intentó calcular un intervalo de confianza. Se dio cuenta de que no podía utilizar una distribución normal para el cálculo; descubrió que la distribución real depende del tamaño de la muestra. Este problema lo llevó a “descubrir” lo que se llama la distribución t de Student. El nombre proviene del hecho de que Gosset escribió bajo el seudónimo de "Un estudiante".
Hasta mediados de los años 70, algunos estadísticos utilizaban la aproximación de la distribución normal para tamaños de muestra grandes y utilizaban la distribución t de Student solo para tamaños de muestra de un máximo de 30 observaciones.
Si se extrae una muestra aleatoria simple de tamaño n de una población con media μ y desviación típica poblacional desconocida σ y se calcula la puntuación t t =
, entonces las puntuaciones t siguen una distribución t de Student con . La puntuación t tiene la misma interpretación que la puntuación . Mide la distancia en unidades de desviación típica
es de su media μ. Para cada tamaño de muestra n existe una distribución t de Student diferente.
Los grados de libertad, , proceden del cálculo de la desviación típica de la muestra . Recuerde que cuando calculamos por primera vez una desviación típica de la muestra, dividimos la suma de las desviaciones al cuadrado por n – 1, pero utilizamos n desviaciones
para calcular . Como la suma de las desviaciones es cero, podemos hallar la última desviación una vez que conocemos las otras desviaciones. Las otras desviaciones pueden cambiar o variar libremente. Llamamos al número en reconocimiento de que uno se pierde en los cálculos. El efecto de la pérdida de un grado de libertad es que el valor t aumenta y el intervalo de confianza aumenta su anchura.
2. La gráfica de la distribución t de Student es similar a la curva normal estándar y a infinitos grados de libertad es la distribución normal. Puede confirmarlo leyendo la línea inferior a infinitos grados de libertad para un nivel de confianza conocido, por ejemplo, en la columna 0,05, nivel de confianza del 95 %, encontramos el valor t de 1,96 a infinitos grados de libertad.
3. La media de la distribución t de Student es cero y la distribución es simétrica respecto a cero, de nuevo como la distribución normal estándar.
4. La distribución t de Student tiene más probabilidad en sus colas que la distribución normal estándar porque la dispersión de la distribución t es mayor que la dispersión de la normal estándar. Así, el gráfico de la distribución t de Student será más gruesa en las colas y más corta en el centro que el gráfico de la distribución normal estándar.
5. La forma exacta de la distribución t de Student depende de los grados de libertad. A medida que aumentan los grados de libertad, el gráfico de la distribución t de Student se parece más al gráfico de la distribución normal estándar.
6. Se supone que la población subyacente de observaciones individuales se distribuye normalmente, con una media poblacional desconocida μ y una desviación típica poblacional desconocida σ. Esta suposición proviene del teorema del límite central porque las observaciones individuales en este caso son las de la distribución muestral. El tamaño de la población subyacente no suele ser relevante, a menos que sea muy pequeña. Si es normal, se cumple el supuesto y no es necesario discutirlo.
Se utiliza una tabla de probabilidad para la distribución t de Student para calcular los valores t en varios niveles de confianza comúnmente utilizados. La tabla muestra las puntuaciones t que corresponden al nivel de confianza (columna) y los grados de libertad (fila). Al utilizar una tabla t, tenga en cuenta que algunas tablas están formateadas para mostrar el nivel de confianza en los títulos de las columnas, mientras que los títulos de las columnas de algunas tablas pueden mostrar solo el área correspondiente en una o ambas colas. Observe que en la parte inferior de la tabla aparecerá el valor t para infinitos grados de libertad. Matemáticamente, a medida que aumentan los grados de libertad, la distribución tse aproxima a la distribución normal estándar. Puede encontrar los valores Z conocidos buscando en la columna alfa correspondiente y leyendo el valor en la última fila. Una tabla t de Student (vea el A - CUADROS ESTADÍSTICOS) da las puntuaciones t dados los grados de libertad y la probabilidad de cola derecha.
La distribución t de Student tiene una de las propiedades más deseables de la normal: es simétrica. Lo que hace la distribución t de Student es extender el eje horizontal, de modo que se necesita un mayor número de desviaciones típicas para capturar la misma cantidad de probabilidad. En realidad, hay un número infinito de distribuciones t de Student, una para cada ajuste del tamaño de la muestra. A medida que aumenta el tamaño de la muestra, la distribución t de Student se parece cada vez más a la distribución normal. Cuando el tamaño de la muestra llega a 30, la distribución normal suele sustituirse por la t de Student porque son muy parecidas. Esta relación entre la distribución t de Student y la distribución normal se muestra en la .
Este es otro ejemplo de una distribución que limita a otra, en este caso la distribución normal es la distribución que limita a la t de Student cuando los grados de libertad en la t de Student se acercan a infinito. Esta conclusión proviene directamente de la derivación de la distribución t de Student realizada por el Sr. Gosset. Reconoció que el problema consistía en tener pocas observaciones y no estimar la desviación típica de la población. Sustituía la desviación típica de la muestra y obtenía resultados volátiles. Por lo tanto, creó la distribución t de Student como una relación entre la distribución normal y la distribución chi-cuadrado. La distribución chi-cuadrado es a su vez un cociente de dos varianzas, en este caso la varianza de la muestra y la varianza de la población desconocida. La distribución t de Student, por tanto, está ligada a la distribución normal, pero tiene grados de libertad que provienen de los de la distribución chi-cuadrado. La solución algebraica demuestra este resultado.
2.
donde
3.
4.
Hay que replantear la fórmula de un intervalo de confianza para la media para los casos en que el tamaño de la muestra es inferior a 30 y no conocemos la desviación típica de la población, σ:
Aquí la estimación puntual de la desviación típica de la población, s ha sido sustituida por la desviación típica de la población, σ, y tν,α ha sido sustituida por Zα. La letra griega ν (pronunciada niu) se coloca en la fórmula general en reconocimiento de que hay muchas distribuciones de Student tv, una para cada tamaño de muestra. ν es el símbolo de los grados de libertad de la distribución y depende del tamaño de la muestra. A menudo se utiliza “df” para abreviar los grados de libertad. Para este tipo de problema, los grados de libertad son ν = n-1, donde n es el tamaño de la muestra. Para buscar una probabilidad en la tabla t de Student tenemos que conocer los grados de libertad del problema.
### Referencias
“America’s Best Small Companies”. Forbes, 2013. Disponible en línea en http://www.forbes.com/best-small-companies/list/ (consultado el 2 de julio de 2013).
Datos de Microsoft Bookshelf.
Datos de http://www.businessweek.com/.
Datos de http://www.forbes.com/.
“Disclosure Data Catalog: Leadership PAC and Sponsors Report, 2012”. Federal Election Commission. Disponible en línea en http://www.fec.gov/data/index.jsp (consultado el 2 de julio de 2013).
“Human Toxome Project: Mapping the Pollution in People”. Environmental Working Group. Disponible en línea en http://www.ewg.org/sites/humantoxome/participants/participant-group.php?group=in+utero%2Fnewborn (consultado el 2 de julio de 2013).
“Metadata Description of Leadership PAC List”. Federal Election Commission. Disponible en línea en http://www.fec.gov/finance/disclosure/metadata/metadataLeadershipPacList.shtml (consultado el 2 de julio de 2013).
### Repaso del capítulo
En muchos casos, el investigador no conoce la desviación típica de la población, σ, de la medida estudiada. En estos casos, es habitual utilizar la desviación típica de la muestra, s, como estimación de σ. La distribución normal crea intervalos de confianza precisos cuando se conoce σ, pero no es tan precisa cuando se utiliza s como estimación. En este caso, la distribución t de Student es mucho mejor. Defina una puntuación t mediante la siguiente fórmula:
La puntuación t sigue la distribución t de Student con n – 1 grados de libertad. El intervalo de confianza bajo esta distribución se calcula con donde
es la puntuación t con un área a la derecha igual a
, s es la desviación típica de la muestra y n es el tamaño de la muestra. Utilice una tabla, una calculadora o una computadora para hallar
para una α determinada.
### Revisión de la fórmula
s = la desviación típica de los valores de la muestra.
es la fórmula de la puntuación t que mide la distancia de una medida con respecto a la media de la población en la distribución t de Student
df = n – 1; los grados de libertad para una distribución t de Student donde n representa el tamaño de la muestra
T~t es la variable aleatoria, T, tiene una distribución t de Student con df grados de libertad
La forma general de un intervalo de confianza para una media única, una desviación típica de la población desconocida y un tamaño de muestra inferior a 30 t de Student viene dada por
Use la siguiente información para responder los próximos cinco ejercicios. Un hospital intenta reducir los tiempos de espera en la sala de emergencias. Se interesa por el tiempo que los pacientes deben esperar antes de que los llamen para examinarlos. Un comité de investigación encuestó al azar a 70 pacientes. La media muestral fue de 1,5 horas con una desviación típica de la muestra de 0,5 horas.
Use la siguiente información para responder los próximos seis ejercicios: se encuestaron ciento ocho estadounidenses para determinar el número de horas que pasan viendo televisión cada mes. Se reveló que veían un promedio de 151 horas al mes con una desviación típica de 32 horas. Supongamos que la distribución de la población subyacente es normal.
Use la siguiente información para responder los próximos 13 ejercicios: los datos que figuran en la son el resultado de una encuesta aleatoria de 39 banderas nacionales (con reemplazo entre selecciones) de varios países. Estamos interesados en hallar un intervalo de confianza para el verdadero número medio de colores en una bandera nacional. Supongamos que X = el número de colores de una bandera nacional.
Construya un intervalo de confianza del 95 % para el número medio real de colores en las banderas nacionales.
### Tarea para la casa
Use la siguiente información para responder los próximos dos ejercicios: un especialista en control de calidad de una cadena de restaurantes toma una muestra aleatoria de tamaño de 12 para comprobar la cantidad de gaseosa que se sirve en la porción de 16 oz. La media muestral es de 13,30 con una desviación típica de la muestra de 1,55. Supongamos que la población subyacente se distribuye normalmente. |
# Intervalos de confianza
## Un intervalo de confianza para una proporción de población
Durante un año electoral vemos artículos en el periódico que indican intervalos de confianza en términos de proporciones o porcentajes. Por ejemplo, un sondeo para un candidato determinado que se presenta a las elecciones presidenciales puede mostrar que el candidato tiene el 40 % de los votos con una diferencia de tres puntos porcentuales (si la muestra es lo suficientemente grande). A menudo, las encuestas electorales se calculan con un 95 % de confianza, por lo que los encuestadores tendrían un 95 % de confianza en que la verdadera proporción de votantes que favorecen al candidato estaría entre el 0,37 y el 0,43.
Los inversores en bolsa se interesan por la proporción real de acciones que suben y bajan cada semana. Las compañías que venden computadoras personales están interesadas en la proporción de hogares de Estados Unidos que tienen computadoras personales. Se pueden calcular intervalos de confianza para la proporción real de acciones que suben o bajan cada semana y para la proporción real de hogares en Estados Unidos que poseen computadoras personales.
El procedimiento para calcular el intervalo de confianza de una proporción poblacional es similar al de la media poblacional, pero las fórmulas son un poco diferentes, aunque conceptualmente idénticas. Aunque las fórmulas son diferentes, se basan en el mismo fundamento matemático que nos proporciona el teorema central del límite. Por ello, veremos el mismo formato básico utilizando los mismos tres datos: el valor muestral del parámetro en cuestión, la desviación típica de la distribución muestral correspondiente y el número de desviaciones típicas que necesitamos para tener la confianza en nuestra estimación que deseamos.
¿Cómo sabe que está ante un problema de proporción? En primer lugar, la distribución subyacente tiene una variable aleatoria binaria y, por tanto, es una distribución binomial. (No se menciona la media o el promedio). Si X es una variable aleatoria binomial, entonces X ~ B(n, p) donde n es el número de ensayos y p es la probabilidad de acierto Para formar una proporción de la muestra, tome X, la variable aleatoria para el número de aciertos y divídala por n, el número de ensayos (o el tamaño de la muestra). La variable aleatoria P′ (léase "P prima") es la proporción de la muestra,
(a veces, la variable aleatoria se denota como
, que se lee “estimador de P”).
p′ = la proporción estimada de éxitos o la proporción muestral de éxitos(p′ es una estimación puntual de p, la verdadera proporción poblacional, y, por tanto, q es la probabilidad de un fracaso en cualquier ensayo).
x = número de aciertos en la muestra
n = el tamaño de la muestra
La fórmula del intervalo de confianza para una proporción de la población sigue el mismo formato que el de la estimación de una media de la población. Recordando la distribución de muestreo para la proporción del Capítulo 7, se encontró que la desviación típica es:
Por lo tanto, el intervalo de confianza para una proporción poblacional se convierte en
se fija en función del grado de confianza que deseemos y es la desviación típica de la distribución muestral.
Las proporciones muestrales . Se utilizan las proporciones estimadas p′ y q′ porque p y q no se conocen.
Recuerde que a medida que p se aleja de 0,5 la distribución binomial se vuelve menos simétrica. Como estamos estimando la binomial con la distribución normal simétrica, cuanto más se aleje de la simetría la binomial, menos confianza tendremos en la estimación.
Esta conclusión puede demostrarse mediante el siguiente análisis. Las proporciones se basan en la distribución de probabilidad binomial. Los posibles resultados son binarios, "éxito" o "fracaso". Esto da lugar a una proporción, es decir, el porcentaje de los resultados que son "éxitos". Se demostró que la distribución binomial podía entenderse completamente si solo conocíamos la probabilidad de éxito en un ensayo cualquiera, llamada p. Se encontró que la media y la desviación típica de la binomial eran:
También se demostró que la binomial podía ser estimada por la distribución normal si TANTO np COMO nq eran mayores que 5. A partir de la discusión anterior, se encontró que la fórmula de estandarización para la distribución binomial es:
que no es más que un replanteamiento de la fórmula general de normalización con las sustituciones adecuadas para μ y σ del binomio. Podemos utilizar la distribución normal estándar, la razón por la que Z está en la ecuación, porque la distribución normal es la distribución limitante de la binomial. Este es otro ejemplo del teorema del límite central. Ya hemos visto que la distribución muestral de las medias se distribuye normalmente. Recordemos la extensa discusión del Capítulo 7 sobre la distribución muestral de las proporciones y las conclusiones del teorema del límite central.
Ahora podemos manipular esta fórmula de la misma manera que hicimos para calcular los intervalos de confianza para una media, pero para calcular el intervalo de confianza para el parámetro poblacional binomial, p.
Donde p′ = x/n, la estimación puntual de p tomada de la muestra. Observe que p′ sustituyó a p en la fórmula. Esto se debe a que no conocemos p, de hecho, esto es justo lo que estamos tratando de estimar.
Lamentablemente, no existe un factor de corrección para los casos en los que el tamaño de la muestra es pequeño, por lo que np′ y nq' deben ser siempre superiores a 5 para desarrollar una estimación de intervalo para p.
### Referencias
Jensen, Tom. “Democrats, Republicans Divided on Opinion of Music Icons”. Public Policy Polling. Disponible en línea en http://www.publicpolicypolling.com/Day2MusicPoll.pdf (consultado el 2 de julio de 2013).
Madden, Mary, Amanda Lenhart, Sandra Coresi, Urs Gasser, Maeve Duggan, Aaron Smith y Meredith Beaton. “Teens, Social Media, and Privacy”. PewInternet, 2013. Disponible en línea en http://www.pewinternet.org/Reports/2013/Teens-Social-Media-And-Privacy.aspx (consultado el 2 de julio de 2013).
Prince Survey Research Associates International. “2013 Teen and Privacy Management Survey”. Pew Research Center: Internet and American Life Project. Disponible en línea en http://www.pewinternet.org/~/media//Files/Questionnaire/2013/Methods%20and%20Questions_Teens%20and%20Social%20Media.pdf (consultado el 2 de julio de 2013).
Saad, Lydia. “Three in Four U.S. Workers Plan to Work Pas Retirement Age: Slightly more say they will do this by choice rather than necessity”. Gallup® Economy, 2013. Disponible en línea en http://www.gallup.com/poll/162758/three-four-workers-plan-work-past-retirement-age.aspx (consultado el 2 de julio de 2013).
The Field Poll. Disponible en línea en http://field.com/fieldpollonline/subscribers/ (consultado el 2 de julio de 2013).
Zogby. “New SUNYIT/Zogby Analytics Poll: Few Americans Worry about Emergency Situations Occurring in Their Community; Only one in three have an Emergency Plan; 70% Support Infrastructure ‘Investment’ for National Security”. Zogby Analytics, 2013. Disponible en línea en http://www.zogbyanalytics.com/news/299-americans-neither-worried-nor-prepared-in-case-of-a-disaster-sunyit-zogby-analytics-poll (consultado el 2 de julio de 2013).
“52% Say Big-Time College Athletics Corrupt Education Process”. Rasmussen Reports, 2013. Disponible en línea en http://www.rasmussenreports.com/public_content/lifestyle/sports/may_2013/52_say_big_time_college_athletics_corrupt_education_process (consultado el 2 de julio de 2013).
### Repaso del capítulo
Algunas medidas estadísticas, como muchas preguntas de las encuestas, miden datos cualitativos en vez de cuantitativos. En este caso, el parámetro poblacional que se estima es una proporción. Es posible crear un intervalo de confianza para la verdadera proporción de la población siguiendo procedimientos similares a los utilizados para crear intervalos de confianza para las medias de la población. Las fórmulas son ligeramente diferentes, pero siguen el mismo razonamiento.
Supongamos que p′ representa la proporción de la muestra, x/n, donde x representa el número de aciertos y n el tamaño de la muestra. Supongamos que q′ = 1 – p′. Entonces el intervalo de confianza para una proporción poblacional viene dado por la siguiente fórmula:
### Revisión de la fórmula
p′= donde x representa el número de aciertos en una muestra y n representa el tamaño de la muestra. La variable p′ es la proporción de la muestra y sirve como estimación puntual de la verdadera proporción de la población.
q′ = 1 – p′
La variable p′ tiene una distribución binomial que se puede aproximar con la distribución normal que se muestra aquí. El intervalo de confianza para la verdadera proporción de la población que viene dado por la fórmula:
proporciona el número de observaciones necesarias en la muestra para estimar la proporción poblacional, p, con confianza 1 - α y margen de error e. Donde e = la diferencia aceptable entre la proporción real de la población y la proporción de la muestra.
Use la siguiente información para responder los dos próximos ejercicios: Compañías de mercadeo están interesadas en conocer el porcentaje de población femenina que toma la mayoría de las decisiones de compra en el hogar.
Use la siguiente información para responder los próximos cinco ejercicios: Supongamos que la compañía de mercadeo hace una encuesta. Encuestaron al azar 200 hogares y hallaron que en 120 de ellos la mujer tomaba la mayoría de las decisiones de compra. Nos interesa la proporción de hogares en los que las mujeres toman la mayoría de las decisiones de compra.
Use la siguiente información para responder los próximos cinco ejercicios: de 1.050 adultos seleccionados al azar, 360 se identificaron como trabajadores manuales, 280 se identificaron como asalariados no manuales, 250 se identificaron como gerentes de nivel medio y 160 se identificaron como ejecutivos. En la encuesta, el 82 % de los trabajadores manuales prefieren camiones, así como el 62 % de los asalariados no manuales, el 54 % de los gerentes de nivel medio y el 26 % de los ejecutivos.
Use la siguiente información para responder los próximos cinco ejercicios: un sondeo realizado a 1.200 votantes preguntaba cuál era el asunto más importante en las próximas elecciones. El sesenta y cinco por ciento respondió que la economía. Nos interesa la proporción de población de los votantes que consideran que la economía es lo más importante.
Use la siguiente información para responder los próximos 16 ejercicios: el Ice Chalet ofrece docenas de clases de patinaje sobre hielo para principiantes. Todos los nombres de las clases se ponen en una cubeta. Se eligió la clase de patinaje sobre hielo para principiantes de 8 a 12 años a las 5 p. m. del lunes. En esa clase había 64 niñas y 16 niños. Supongamos que estamos interesados en la proporción real de niñas, de 8 a 12 años, en todas las clases de patinaje sobre hielo para principiantes en el Ice Chalet. Supongamos que los niños de la clase seleccionada son una muestra aleatoria de la población.
### Tarea para la casa
Use la siguiente información para responder los próximos tres ejercicios: según Field Poll, el 79 % de los adultos de California (los resultados reales son 400 de 506 encuestados) consideran que “la educación y nuestras escuelas” es uno de los principales problemas a los que se enfrenta California. Queremos construir un intervalo de confianza del 90 % para la verdadera proporción de adultos de California que piensan que la educación y las escuelas son uno de los principales problemas a los que se enfrenta el estado.
Use la siguiente información para responder los próximos dos ejercicios: se encuestaron aleatoriamente quinientos once (511) hogares de una determinada comunidad del sur de California para averiguar si cumplen las recomendaciones mínimas de preparación ante un terremoto. Ciento setenta y tres (173) de las viviendas encuestadas cumplían las recomendaciones mínimas de preparación para terremotos y 338 no. |
# Intervalos de confianza
## Cálculo del tamaño de la muestra n: variables aleatorias continuas y binarias
Variables aleatorias continuasNormalmente no tenemos control sobre el tamaño de la muestra de un conjunto de datos. Sin embargo, si podemos fijar el tamaño de la muestra, como en los casos en los que realizamos una encuesta, es muy útil saber cuál debe ser su tamaño para proporcionar la máxima información. El muestreo puede ser muy costoso, tanto en tiempo como en producto. Las simples encuestas telefónicas cuestan aproximadamente 30 dólares cada una, por ejemplo, y algunos muestreos requieren la destrucción del producto.
Si volvemos a nuestra fórmula de normalización de la distribución muestral para las medias, podemos ver que es posible resolverla para n. Si hacemos esto tenemos en el denominador.
Como aún no hemos tomado una muestra, no conocemos ninguna de las variables de la fórmula, excepto que podemos establecer Zα al nivel de confianza que deseamos, tal como hicimos al determinar los intervalos de confianza. Si establecemos un error aceptable predeterminado, o tolerancia, para la diferencia entre y μ, denominado e en la fórmula, estamos mucho más lejos en la resolución del tamaño de la muestra n. Todavía no conocemos la desviación típica de la población, σ. En la práctica, se suele hacer una encuesta previa que permite afinar el cuestionario y que da una desviación típica de la muestra que se puede utilizar. En otros casos, se puede utilizar la información previa de otras encuestas para σ en la fórmula. Aunque es rudimentario, este método para determinar el tamaño de la muestra puede ayudar a reducir los costos de forma significativa. Serán los datos reales recogidos los que determinen las inferencias sobre la población, por lo que conviene ser cauteloso con el tamaño de la muestra exigiendo altos niveles de confianza y pequeños errores de muestreo.
Variables aleatorias binariasLo que se hizo en los casos en los que se buscaba la media de una distribución también se puede hacer cuando se hace un muestreo para determinar el parámetro poblacional p de las proporciones. La manipulación de la fórmula de normalización de las proporciones da como resultado:
donde e = (p′-p), y es el error de muestreo aceptable, o tolerancia, para esta aplicación. Esto se medirá en puntos porcentuales.
En este caso el propio objeto de nuestra búsqueda está en la fórmula, p, y por supuesto q porque q =1-p. Este resultado se produce porque la distribución binomial es una distribución de un parámetro. Si conocemos p entonces conocemos la media y la desviación típica. Por lo tanto, p aparece en la desviación típica de la distribución muestral que es de donde sacamos esta fórmula. Si en un exceso de precaución sustituimos p por 0,5, extraeremos el mayor tamaño de muestra necesario que proporcione el nivel de confianza especificado por Zα y la tolerancia que hemos seleccionado. Esto es cierto porque de todas las combinaciones de dos fracciones que suman uno, el mayor múltiplo es cuando cada una es 0,5. Sin ninguna otra información sobre el parámetro poblacional p, esta es la práctica habitual. Esto puede dar lugar a un sobremuestreo, pero ciertamente no a un submuestreo, por lo que se trata de un enfoque prudente.
Existe un interesante equilibrio entre el nivel de confianza y el tamaño de la muestra que aparece aquí cuando se considera el costo del muestreo. La muestra el tamaño de la muestra apropiado para diferentes niveles de confianza y diferentes niveles de error aceptable, o tolerancia.
Esta tabla está diseñada para mostrar el tamaño máximo de la muestra requerido en diferentes niveles de confianza dado un supuesto p= 0,5 y q=0,5 como se comentó anteriormente.
El error aceptable, denominado tolerancia en la tabla, se mide en valores más o menos de la proporción real. Por ejemplo, un error aceptable del 5 % significa que, si la proporción de la muestra es del 26 %, la conclusión sería que la proporción real de la población está entre el 21 % y el 31 % con un nivel de confianza del 90 % si se hubiera tomado una muestra de 271 personas. Asimismo, si el error aceptable se fijara en el 2 %, la proporción de la población se situaría entre el 24 % y el 28 % con un nivel de confianza del 90 %, pero exigiría aumentar el tamaño de la muestra de 271 a 1691. Si quisiéramos un mayor nivel de confianza, necesitaríamos una muestra de mayor tamaño. Pasar de un nivel de confianza del 90 % a un nivel del 95 % con una tolerancia de más o menos el 5 % requiere cambiar el tamaño de la muestra de 271 a 384. Un tamaño de muestra muy común que suele aparecer en las encuestas políticas es de 384. Con los resultados de las encuestas se suele decir que los resultados son buenos con un nivel de "exactitud" de más o menos el 5 %.
### Repaso del capítulo
A veces, los investigadores saben de antemano que quieren estimar una media poblacional dentro de un margen de error específico para un nivel de confianza dado. En ese caso, resuelva la fórmula del intervalo de confianza correspondiente para n a fin de descubrir el tamaño de la muestra que se necesita para lograr este objetivo:
Si la variable aleatoria es binaria, la fórmula para el tamaño de muestra adecuado a fin de que se mantenga un nivel de confianza determinado con un nivel de tolerancia específico viene dada por
### Revisión de la fórmula
n =
= fórmula utilizada para determinar el tamaño de la muestra(n) necesario para alcanzar un margen de error deseado con un nivel de confianza determinado para una variable aleatoria continua
= la fórmula utilizada para determinar el tamaño de la muestra si la variable aleatoria es binaria
Use la siguiente información para responder los próximos cinco ejercicios: se sabe que la desviación típica del peso de los elefantes es de 15 libras aproximadamente. Queremos construir un intervalo de confianza del 95 % para el peso medio de las crías de elefante recién nacidas. Se pesan cincuenta elefantes recién nacidos. La media muestral es de 244 libras. La desviación típica de la muestra es de 11 libras.
Use la siguiente información para responder los próximos siete ejercicios: la Oficina del Censo de EE. UU. realiza un estudio para determinar el tiempo necesario para rellenar el formulario corto. La oficina encuesta a 200 personas. La media muestral es de 8,2 minutos. Se conoce una desviación típica de 2,2 minutos. Se supone que la distribución de la población es normal.
Use la siguiente información para responder los próximos diez ejercicios: se seleccionó una muestra de 20 cabezas de lechuga. Supongamos que la distribución poblacional del peso de la cabeza es normal. Luego se registró el peso de cada cabeza de lechuga. El peso medio era de 2,2 libras con una desviación típica de 0,1 libras. Se sabe que la desviación típica de la población es de 0,2 libras.
Utilice la siguiente información para responder a los siguientes 14 ejercicios: la edad media de todos los estudiantes del Foothill College en el trimestre de otoño pasado fue de 33,2 años. La desviación típica de la población ha sido bastante constante en 15. Supongamos que se seleccionan al azar veinticinco estudiantes del semestre de invierno. La edad media de la muestra era de 30,4 años. Estamos interesados en la verdadera edad media de los estudiantes del semestre de invierno del Foothill College. Supongamos que X = la edad de un estudiante del semestre de invierno del Foothill College.
Construya un Intervalo de Confianza del 95 % para la edad media real de los estudiantes del semestre de invierno del Foothill College, elabore y responda los siguientes siete ejercicios.
### Tarea para la casa
|
# Pruebas de hipótesis con una muestra
## Introducción
Ahora nos encontramos con el trabajo principal del estadístico: desarrollar y probar hipótesis. Es importante situar este material en un contexto más amplio para que se entienda completamente el método por el que se forma una hipótesis. El uso de ejemplos de libros de texto a menudo nubla el verdadero origen de las hipótesis estadísticas.
Las pruebas estadísticas forman parte de un proceso mucho más amplio conocido como método científico. Este método se desarrolló hace más de dos siglos como la forma aceptada de crear nuevos conocimientos. Hasta entonces, y desgraciadamente aún hoy, entre algunos, el "conocimiento" podía crearse simplemente porque alguna autoridad dijera que algo era así, ipso dicta. La superstición y las teorías de la conspiración eran (¿son?) aceptadas sin crítica.
El método científico, brevemente, establece que solo siguiendo un proceso cuidadoso y específico se puede incluir alguna afirmación en el cuerpo de conocimientos aceptado. Este proceso comienza con un conjunto de supuestos sobre los que se construye una teoría, a veces llamada modelo. Esta teoría, si tiene alguna validez, dará lugar a predicciones; lo que llamamos hipótesis.
Por ejemplo, en Microeconomía la teoría de la elección del consumidor parte de ciertos supuestos relativos al comportamiento humano. A partir de estos supuestos se elabora una teoría de cómo los consumidores toman decisiones utilizando las curvas de indiferencia y la línea presupuestaria. Esta teoría dio lugar a una predicción muy importante, a saber, que existía una relación inversa entre el precio y la cantidad demandada. Esta relación se conoce como curva de demanda. La pendiente negativa de la curva de demanda es en realidad una predicción, o una hipótesis, que puede comprobarse con herramientas estadísticas.
A menos que cientos y cientos de pruebas estadísticas de esta hipótesis no hubieran confirmado esta relación, la llamada ley de la demanda habría sido descartada hace años. Este es el papel de la estadística, poner a prueba las hipótesis de diversas teorías para determinar si deben ser admitidas en el cuerpo de conocimientos aceptado; cómo entendemos nuestro mundo. Sin embargo, una vez admitidas, pueden ser descartadas posteriormente si aparecen nuevas teorías que hagan mejores predicciones.
No hace mucho, dos científicos afirmaron que podían obtener más energía de un proceso que la que se introducía en este. Esto causó un tremendo revuelo por razones obvias. Aparecieron en la portada de Time y se les ofrecieron sumas extravagantes para que llevaran sus trabajos de investigación a la industria privada y a cualquier universidad. No pasó mucho tiempo hasta que su trabajo fue sometido a las rigurosas pruebas del método científico y se descubrió que era un fracaso. Ningún otro laboratorio pudo replicar sus hallazgos. En consecuencia, se han hundido en la oscuridad y su teoría fue descartada. Es posible que vuelva a salir a la luz cuando alguien pueda superar las pruebas de las hipótesis exigidas por el método científico, pero hasta entonces es solo una curiosidad. A lo largo del tiempo se han intentado muchos fraudes auténticos, pero la mayoría se han descubierto aplicando el proceso del método científico.
Este debate pretende mostrar en qué punto de este proceso se encuentra la estadística. La estadística y los estadísticos no se dedican necesariamente a desarrollar teorías, sino a probar las teorías de otros. Las hipótesis proceden de estas teorías basadas en un conjunto explícito de supuestos y una lógica sólida. La hipótesis es lo primero, antes de recopilar los datos. Los datos no crean hipótesis, sino que se utilizan para probarlas. Si tenemos esto en cuenta al estudiar esta sección, el proceso de formación y comprobación de hipótesis tendrá más sentido.
Uno de los trabajos de un estadístico es hacer inferencias estadísticas sobre las poblaciones a partir de muestras tomadas de la población. Los intervalos de confianza son una forma de estimar un parámetro poblacional. Otra forma de hacer una inferencia estadística es tomar una decisión sobre el valor de un parámetro específico. Por ejemplo, un concesionario de automóviles anuncia que su nueva camioneta pequeña recorre un promedio de 35 millas por galón. Un servicio de tutoría afirma que su método de enseñanza ayuda al 90 % de sus estudiantes a obtener una calificación A o B. Una compañía dice que las mujeres administradoras de su compañía ganan un promedio de 60.000 dólares al año.
Un estadístico tomará una decisión sobre estas declaraciones. Este proceso se llama “prueba de hipótesis”. Una prueba de hipótesis consiste en recopilar datos de una muestra y evaluarlos. Luego, el estadístico decide si existen o no pruebas suficientes basándose en el análisis de los datos para rechazar la hipótesis nula.
En este capítulo hará pruebas de hipótesis sobre medias simples y proporciones simples. También conocerá los errores asociados a estas pruebas. |
# Pruebas de hipótesis con una muestra
## Hipótesis nula y alternativa
La prueba real comienza considerando dos hipótesis. Se denominan hipótesis nula e hipótesis alternativa. Estas hipótesis contienen puntos de vista opuestos.
H: La hipótesis nula: Es una afirmación de que no hay diferencia entre las variables: no están relacionadas. A menudo, esto puede considerarse el statu quo y, como resultado, si no se puede aceptar lo nulo, se requiere alguna acción.
H: La hipótesis alternativa: Es una afirmación sobre la población que es contradictoria con H y lo que concluimos cuando no podemos aceptar H. Esto es normalmente lo que el investigador está tratando de probar. La hipótesis alternativa es la contendiente y debe ganar con pruebas significativas para derrocar el statu quo. Este concepto se conoce a veces como la tiranía del statu quo porque, como veremos más adelante, para derribar la hipótesis nula se necesita normalmente un 90 % o más de confianza en que esta es la decisión correcta.
Dado que las hipótesis nula y alternativa son contradictorias, debe examinar las pruebas para decidir si tiene suficiente evidencia para rechazar la hipótesis nula o no. Las pruebas se presentan en forma de datos de muestra.
Una vez que haya determinado qué hipótesis apoya la muestra, tome una decisión. Hay dos opciones para tomar una decisión. Son “no puede aceptar H” si la información de la muestra favorece la hipótesis alternativa o “no se rechaza H” o “se declina rechazar H” si la información de la muestra es insuficiente para rechazar la hipótesis nula. Todas estas conclusiones se basan en un nivel de probabilidad, un nivel de significación, que establece el analista.
La tabla 9.1 presenta las distintas hipótesis en los pares correspondientes. Por ejemplo, si la hipótesis nula es igual a algún valor, la alternativa no puede ser igual a ese valor.
### Repaso del capítulo
En una prueba de hipótesis se evalúan los datos de la muestra para llegar a una decisión sobre algún tipo de afirmación. Si se cumplen determinadas condiciones sobre la muestra, la afirmación se puede evaluar para una población. En una prueba de hipótesis, nosotros:
### Tarea para la casa
### Referencias
Datos del Instituto Nacional de Salud Mental. Disponible en línea en http://www.nimh.nih.gov/publicat/depression.cfm. |
# Pruebas de hipótesis con una muestra
## Resultados y errores de tipo I y II
Cuando se realiza una prueba de hipótesis hay cuatro resultados posibles en según la verdad (o falsedad) de la hipótesis nula H y de la decisión de rechazarla o no. Los resultados se resumen en el siguiente cuadro:
Los cuatro resultados posibles en la tabla son:
Cada uno de los errores se produce con una probabilidad determinada. Las letras griegas α y β representan las probabilidades.
α = probabilidad de un error de tipo I = = probabilidad de rechazar la hipótesis nula cuando la hipótesis nula es verdadera: rechazar un buen nulo.
β = probabilidad de un error tipo II = = probabilidad de no rechazar la hipótesis nula cuando la hipótesis nula es falsa. (1 - β) se denomina la potencia de la prueba.
α y β deben ser lo más pequeños posible porque son probabilidades de error.
La estadística nos permite establecer la probabilidad de que cometamos un error de tipo I. La probabilidad de cometer un error de tipo I es α. Recordemos que los intervalos de confianza en la última unidad se establecían al elegir un valor llamado Zα (o tα) y el valor alfa determinaba el nivel de confianza de la estimación porque era la probabilidad de que el intervalo no captara la verdadera media (o parámetro de proporción p). Esta y aquella alfa son iguales.
La forma más fácil de ver la relación entre el error alfa y el nivel de confianza es con la siguiente figura.
En el centro de la hay una distribución normal de muestreo, marcada H0. Se trata de una distribución de muestreo de y por el teorema del límite central se distribuye normalmente. La distribución del centro se marca H y representa la distribución para la hipótesis nula H: µ = 100. Este es el valor que se está probando. Los enunciados formales de las hipótesis nula y alternativa se enumeran debajo de la figura.
Las distribuciones a ambos lados de la distribución H0 representan las que serían verdaderas si H es falsa, bajo la hipótesis alternativa, indicada como H. No sabemos cuál es la verdad, y nunca lo sabremos. De hecho, hay un número infinito de distribuciones de las que se podrían haber extraído los datos si H es verdadera, pero solo dos de ellas están en la representando a todas las demás.
Para comprobar una hipótesis, tomamos una muestra de la población y determinamos si proviene de la distribución hipotética con un nivel de significación aceptable. Este nivel de significación es el error alfa y está marcado en la como las áreas sombreadas en cada cola de la distribución H. (Cada área es en realidad α/2 porque la distribución es simétrica y la hipótesis alternativa posibilita que el valor sea mayor o menor que el valor hipotético, la llamada prueba de dos colas).
Si la media muestral está marcada como está en la cola de la distribución de H, entonces concluimos que la probabilidad de que provenga de la distribución H es menor que alfa. En consecuencia, afirmamos que “la hipótesis nula no puede aceptarse con un nivel de significación (α)”. La verdad puede ser que este sí provenía de la distribución H, pero del extremo de la cola. Si es así, hemos rechazado falsamente una hipótesis nula verdadera y hemos cometido un error de tipo I. Lo que la estadística ha hecho es proporcionar una estimación sobre lo que sabemos y lo que controlamos, y esa es la probabilidad de que nos equivoquemos, α.
También observamos en la que la media muestral sería realmente de una distribución H, pero dentro del límite establecido por el nivel alfa. Este caso está marcado como . Existe la probabilidad de que en realidad provenga de H pero aparece en el rango de H entre las dos colas. Esta probabilidad es el error beta, la probabilidad de aceptar un falso nulo.
Nuestro problema es que solo podemos fijar el error alfa porque hay un número infinito de distribuciones alternativas de las que podría haber salido la media que no son iguales a H. En consecuencia, el estadístico recae la carga de la prueba en la hipótesis alternativa. Es decir, no rechazaremos una hipótesis nula, a no ser que haya una probabilidad superior al 90 % o al 95 %, e incluso al 99 %, de que la nula sea falsa: la carga de la prueba recae en la hipótesis alternativa. Por eso lo designamos anteriormente como la tiranía del statu quo.
A modo de ejemplo, el sistema judicial estadounidense parte del supuesto de la “presunción de inocencia” del acusado. Este es el statu quo y es la hipótesis nula. El juez dirá al jurado que no puede declarar al acusado culpable, a no ser que las pruebas indiquen la culpabilidad más allá de una “duda razonable”, que se define en los casos penales como un 95 % de certeza de culpabilidad. Si el jurado no puede aceptar la nulidad, la inocencia, entonces se tomarán medidas, tiempo de cárcel. La carga de la prueba siempre recae en la hipótesis alternativa (en los casos civiles, el jurado solo necesita tener más del 50 % de certeza de que se ha cometido un delito para declarar la culpabilidad, lo que se denomina “preponderancia de las pruebas”).
El ejemplo anterior era para una prueba de una media, pero la misma lógica se aplica a las pruebas de hipótesis para todos los parámetros estadísticos que uno quiera probar.
Los siguientes son ejemplos de errores tipo I y tipo II.
### Repaso del capítulo
En toda prueba de hipótesis, los resultados dependen de una interpretación correcta de los datos. Los cálculos incorrectos o el resumen de estadísticas mal entendidos pueden producir errores que afecten los resultados. Un error tipo I se produce cuando se rechaza una hipótesis nula verdadera. Un error tipo II se produce cuando no se rechaza una hipótesis nula falsa.
Las probabilidades de estos errores se indican con las letras griegas α y β, para un error tipo I y el tipo II, respectivamente. La potencia de la prueba, 1 – β, cuantifica la probabilidad de que una prueba arroje el resultado correcto de que se acepte una hipótesis alternativa verdadera. Es deseable una alta potencia.
### Tarea para la casa
|
# Pruebas de hipótesis con una muestra
## Distribución necesaria para la comprobación de la hipótesis
Anteriormente, hemos hablado de las distribuciones muestrales. Las distribuciones particulares se asocian a la comprobación de la hipótesis. Realizaremos comprobaciones de hipótesis de una media poblacional con una distribución normal o una distribución (recuerde, utilice una distribución t de Student cuando la desviación típica de la población se desconozca y el tamaño de la muestra sea pequeño, donde se considera pequeño a menos de 30 observaciones). Realizamos pruebas de una proporción poblacional mediante una distribución normal cuando podemos suponer que lo sea. Consideramos que esto es cierto si la proporción de la muestra, , por el tamaño de la muestra es superior a 5 y 1- por el tamaño de la muestra también es mayor que 5. Se trata de la misma regla empírica que utilizamos al desarrollar la fórmula del intervalo de confianza para una proporción poblacional.
### Comprobación de la hipótesis para la media
Volviendo a la fórmula de estandarización, podemos derivar el estadístico de prueba para comprobar las hipótesis relativas a las medias.
La fórmula de estandarización no se puede resolver tal cual porque no tenemos μ, la media poblacional. Sin embargo, si sustituimos el valor hipotético de la media, μ0 en la fórmula anterior, podemos calcular un valor Z. Este es el estadístico de prueba con respecto a la comprobación de la hipótesis para una media y se presenta en la . Interpretamos este valor Z como la probabilidad asociada de que una muestra con una media muestral de provendría de una distribución con una media poblacional de H y a este valor Z lo llamamos Zc por “calculado”. y muestran este proceso.
En la se presentan dos de los tres resultados posibles. y están en las colas de la distribución hipotética de H. Observe que el eje horizontal del panel superior está etiquetado como 's. Esta es la misma distribución teórica de 's, la distribución muestral, que el teorema del límite central nos indica que se distribuye normalmente. Por eso podemos dibujarlo con esta forma. El eje horizontal del panel inferior está etiquetado como Z y es la distribución normal estándar. y , denominados valores críticos, están marcados en el panel inferior como los valores Z asociados a la probabilidad que el analista haya establecido como nivel de significación en la prueba, (α). Las probabilidades en las colas de ambos paneles son, por tanto, las mismas.
Observe que para cada hay una Zc asociada, llamada Z calculada, que es el resultado de resolver la ecuación anterior. Esta Z calculada no es más que el número de desviaciones típicas que la media hipotética tiene con respecto a la media muestral. Si la media muestral está a “demasiadas” desviaciones típicas de la media hipotética, concluimos que la media muestral no proviene de la distribución con la media hipotética, dado el nivel de significación requerido. Esto podría venir de H, pero se considera demasiado improbable. En la tanto y están en las colas de la distribución. Se considera que están “demasiado lejos” del valor hipotético de la media, dado el nivel de alfa elegido. Si en realidad esta media muestral provenía de H, pero de la cola, hemos cometido un error de tipo I: hemos rechazado un buen nulo. Nuestro único consuelo real es que conocemos la probabilidad de cometer ese error, α, y podemos controlar el tamaño de α.
La muestra la tercera posibilidad para la ubicación de la media muestral, . Aquí la media muestral está dentro de los dos valores críticos. Es decir, dentro de la probabilidad de (1-α) y no podemos rechazar la hipótesis nula.
Esto nos da la regla de decisión para comprobar una hipótesis en una prueba de dos colas:
Esta regla será siempre la misma, sin importar la hipótesis que estemos comprobando o las fórmulas que utilicemos para hacer la prueba. Lo único será cambiar el Zc por el símbolo apropiado para el estadístico de prueba con respecto al parámetro que se está probando. Expresando la regla de decisión de otra manera: si es improbable que la media muestral provenga de la distribución con la media hipotética, no podemos aceptar la hipótesis nula. Aquí definimos “improbable” como una probabilidad de ocurrir menor que alfa.
### Enfoque del valor P
Se puede desarrollar una regla de decisión alternativa al calcular la probabilidad de que se encuentre una media muestral que resulte en un estadístico de prueba mayor que el hallado a partir de los datos de la muestra actual, suponiendo que la hipótesis nula sea verdadera. Aquí, la noción de “probable” e “improbable” se define por la probabilidad de extraer de una población una muestra con una media que hipotéticamente sea mayor o menor que la calculada en los datos de la muestra. En pocas palabras, el enfoque del valor p compara el nivel de significación deseado, α, con el valor p, que es la probabilidad de obtener una media muestral más alejada del valor hipotético que la media muestral real. Un valor p grande calculado a partir de los datos indica que no debemos rechazar la hipótesis nula. Cuanto más pequeño sea el valor p, más improbable es el resultado y más fuerte es la evidencia contra la hipótesis nula. Rechazaremos la hipótesis nula si las pruebas son contundentes en su contra. La relación entre la regla de decisión de comparar los valores calculados del estadístico de prueba, Zc, y el valor crítico, Zα, y utilizar el valor p se aprecia en la .
El valor calculado del estadístico de prueba es Zc en este ejemplo y está marcado en el gráfico inferior de la distribución normal estándar porque es un valor Z. En este caso el valor calculado está en la cola y, por tanto, no podemos aceptar la hipótesis nula, la asociada es demasiado grande para creer que provenga de la distribución con una media de µ0 y un nivel de significación de α.
Si utilizamos la regla de decisión del valor p, necesitamos un paso más. Tenemos que encontrar en la tabla normalizada la probabilidad asociada al valor calculado del estadístico de prueba, Zc. A continuación, lo comparamos con el α asociado a nuestro nivel de confianza seleccionado. En la vemos que el valor p es menor que α, por lo que no podemos aceptar la nulidad. Sabemos que el valor p es menor que α porque el área bajo el valor p es menor que α/2. Cabe destacar que dos investigadores que extraigan al azar de la misma población pueden calcular dos valores P diferentes en sus muestras. Esto ocurre porque el valor P se calcula como la probabilidad en la cola más allá de la media muestral, asumiendo que la hipótesis nula sea correcta. Dado que las medias muestrales serán con toda probabilidad diferentes, esto creará dos valores P distintos. Sin embargo, las conclusiones en cuanto a la hipótesis nula deberían variar únicamente con el nivel de probabilidad de α.
Esta es una forma sistemática de tomar una decisión sobre si se puede aceptar o rechazar una hipótesis nula si se utiliza el valor y un α preestablecido o preconcebido (el “nivel de significación”). Un α preestablecido es la probabilidad de un error de tipo I (rechazar la hipótesis nula cuando la hipótesis nula es verdadera). Puede que se le entregue o no al principio del problema. En cualquier caso, el valor de α es decisión del analista. Cuando tome la decisión de rechazar o no rechazar H, haga lo siguiente:
1. Si α > valor p, no se puede aceptar H. Los resultados de los datos de la muestra son significativos. Hay pruebas suficientes para concluir que H es una creencia incorrecta y que la hipótesis alternativa, H, puede ser correcta.
2. Si α ≤ valor p, no se puede rechazar H. Los resultados de los datos de la muestra son despreciables. No hay pruebas suficientes para concluir que la hipótesis alternativa, H, sea correcta. En este caso se mantiene el statu quo.
3. Cuando “no puede rechazar H”, no significa que deba creer que H es verdadera. Significa simplemente que los datos de la muestra no han aportado pruebas suficientes para poner en duda la veracidad de H. Recuerde que el nulo es el statu quo y se necesita una alta probabilidad para derrocarlo. Este sesgo a favor de la hipótesis nula es lo que da lugar a la afirmación “tiranía del statu quo” cuando se habla de la comprobación de hipótesis y del método científico.
Ambas reglas de decisión darán lugar a la misma decisión y es cuestión de preferencia cuál se utiliza.
### Pruebas de una y dos colas
El estudio de la a la se basó en las hipótesis nula y alternativa, presentadas en la . Se denominó prueba de dos colas porque la hipótesis alternativa permitía que la media proviniera de una población mayor o menor que la media hipotética en la hipótesis nula. Esto podría verse mediante el enunciado de la hipótesis alternativa como μ ≠ 100, en este ejemplo.
Puede ser que al analista no le preocupe que el valor sea "demasiado" alto o "demasiado" bajo con respecto al valor hipotético. Si este es el caso, se convierte en una prueba de una cola y toda la probabilidad alfa se coloca en una sola cola y no se divide entre α/2, como en el caso anterior de una prueba de dos colas. Cualquier prueba de un reclamo será una prueba de una cola. Por ejemplo, un fabricante de automóviles afirma que su modelo 17B ofrece un consumo de gasolina superior a 25 millas por galón. Las hipótesis nula y alternativa serían:
1. H: µ ≤ 25
2. H: µ > 25
La afirmación estaría en la hipótesis alternativa. La carga de la prueba en la comprobación de hipótesis recae en la alternativa. Esto se debe a que el no rechazar el nulo, el statu quo, deberá lograrse con un 90 % o 95 % de confianza en que no se pueda mantener. Dicho de otro modo, queremos tener solo un 5 % o 10 % de probabilidad de cometer un error de tipo I, rechazar un buen nulo y derrocar el statu quo.
Esta es una prueba de una cola, donde toda la probabilidad alfa se coloca en una sola cola y no se divide entre α/2, como en el caso anterior de la prueba de dos colas.
La muestra los dos casos posibles y la forma de las hipótesis nula y alternativa que los origina.
donde μ0 es el valor hipotético de la media poblacional.
### Efectos del tamaño de la muestra en el estadístico de prueba
Al desarrollar los intervalos de confianza para la media de una muestra, encontramos que la mayoría de las veces no tenemos la desviación típica de la población, σ. Si el tamaño de la muestra fuera inferior a 30, podríamos sustituir simplemente la estimación puntual de σ, la desviación típica de la muestra, s, y utilizar la distribución t de Student para corregir esta falta de información.
A la hora de comprobar las hipótesis nos topamos con este mismo problema y la solución es exactamente igual. A saber: Si se desconoce la desviación típica de la población y el tamaño de la muestra es inferior a 30, sustituya s, la estimación puntual de la desviación típica de la población, σ, en la fórmula del estadístico de prueba y utilice la distribución t de Student. Todas las fórmulas y figuras anteriores no cambian, excepto esta sustitución y el cambio de la distribución Z por la distribución t de Student en el gráfico. Recuerde que la distribución t de Student solo puede calcularse si se conocen los grados de libertad adecuados para el problema. En este caso, los grados de libertad se calculan como antes con intervalos de confianza: df = (n-1). El valor t calculado se compara con el valor t asociado al nivel de confianza preestablecido y que se requiere en la prueba, tα, df, que se encuentra en las tablas t de Student. Si no conocemos σ, pero el tamaño de la muestra es de 30 o más, simplemente sustituimos s por σ y utilizamos la distribución normal.
La resume estas normas.
### Un enfoque sistemático para comprobar una hipótesis
Un enfoque sistemático de las pruebas de hipótesis sigue los siguientes pasos y en este orden. Esta plantilla servirá para todas las hipótesis que se pongan a prueba.
1. Establezca las hipótesis nula y alternativa. Esta suele ser la parte más difícil del proceso. Aquí se revisa la cuestión planteada. Qué parámetro se está probando, una media, una proporción, diferencias de medias, etc. ¿Es una prueba de una cola o de dos colas?
2. Decida el nivel de significación requerido para este caso particular y determine el valor crítico. Estos se pueden encontrar en la tabla estadística correspondiente. Los niveles de confianza típicos de las empresas son 80 %, 90 %, 95 %, 98 % y 99 %. Sin embargo, el nivel de significación es una decisión política y debería basarse en el riesgo de cometer un error de tipo I y rechazar un buen nulo. Considere las consecuencias de cometer un error de tipo I. A continuación, sobre la base de las hipótesis y el tamaño de la muestra, seleccione la estadística adecuada de la prueba y calcule el valor crítico pertinente: Z
3. Tome una o varias muestras y calcule los parámetros pertinentes: media muestral, desviación típica o proporción. Con base en la fórmula del paso 2, calcule ahora el estadístico de prueba para este caso en particular; utilice los parámetros que acaba de calcular.
4. Compare el valor calculado del estadístico de prueba y el valor crítico. Si se marcan en el gráfico, se obtendrá una buena imagen visual de la situación. Ahora solo hay dos situaciones:
5. Llegue a una conclusión. Es mejor articular la conclusión de dos maneras diferentes. En primer lugar, una conclusión estadística formal como: "Con un nivel de significación del 5 %, no podemos aceptar las hipótesis nulas de que la media de la población es igual a XX (unidades de medida)". El segundo enunciado de la conclusión es menos formal y enuncia la acción, o la falta de acción, requerida. Si la conclusión formal era la anterior, la informal podría ser: "La máquina está estropeada y hay que apagarla y mandarla a reparar".
Todas las hipótesis probadas pasarán por este mismo proceso. Los únicos cambios son las fórmulas pertinentes y estas están determinadas por la hipótesis necesaria para responder la pregunta original.
### Repaso del capítulo
Para que los resultados de una prueba de hipótesis se puedan generalizar a una población se deben cumplir ciertos requisitos.
Cuando se hacen pruebas para una única media poblacional:
1. Se debe utilizar una prueba t de Student si los datos proceden de una muestra aleatoria simple y la población se distribuye aproximadamente normal, o el tamaño de la muestra es grande, con una desviación típica desconocida.
2. La prueba normal funcionará si los datos proceden de una muestra simple y aleatoria y la población se distribuye aproximadamente de forma normal o si el tamaño de la muestra es grande.
Al comprobar una proporción poblacional única, utilice una prueba normal para una proporción poblacional única si los datos provienen de una muestra aleatoria simple, cumplen los requisitos de una distribución binomial y el número medio de éxitos y el número medio de fracasos satisfacen las condiciones: np > 5 y nq > 5, donde n es el tamaño de la muestra, p es la probabilidad de un éxito y q es la probabilidad de un fracaso.
### Revisión de la fórmula
### Tarea para la casa
|
# Pruebas de hipótesis con una muestra
## Ejemplos de pruebas de hipótesis completas
### Pruebas sobre las medias
En el paso 4 tenemos que comparar el estadístico de prueba y el valor crítico y marcarlos en el gráfico. Vemos que el estadístico de prueba está en la cola; por ende, pasamos al paso 4 y llegamos a una conclusión. La probabilidad de que el tiempo promedio de 16 minutos proceda de una distribución con una media poblacional de 16,43 minutos es demasiado improbable para que aceptemos la hipótesis nula. No podemos aceptar la hipótesis nula.
El paso 5 nos hace exponer nuestras conclusiones primero de manera formal y luego de manera menos formal. La conclusión formal sería la siguiente: "Con un nivel de significación del 95 %, no podemos aceptar la hipótesis nula de que el tiempo de natación con gafas procede de una distribución con una media poblacional de 16,43 minutos". De manera menos formal: "Con un 95 % de significación, creemos que las gafas mejoran la velocidad de nado".
Si quisiéramos utilizar el sistema de valores p para llegar a una conclusión, calcularíamos la estadística y daríamos el paso adicional de la probabilidad de estar a 2,08 desviaciones típicas de la media en una distribución t. Este valor es de 0,0187. Al compararlo con el nivel α de 0,05, nos damos cuenta de que no podemos aceptar la hipótesis nula. El valor p se ha puesto en el gráfico como el área sombreada más allá de –2,08 y muestra que es menor que el área sombreada, que es el nivel alfa de 0,05. Con ambos métodos se llega a la misma conclusión de que no podemos aceptar la hipótesis nula.
### Prueba de hipótesis para las proporciones
Al igual que existían intervalos de confianza para las proporciones, o más formalmente, el parámetro poblacional p de la distribución binomial, existe la posibilidad de contrastar hipótesis relativas a p.
El parámetro poblacional para la binomial es p. El valor estimado (estimación puntual) para p es p′ donde p′ = x/n, x es el número de aciertos en la muestra y n es el tamaño de la muestra.
Cuando se realiza una prueba de hipótesis de una proporción poblacional p, se toma una muestra aleatoria simple de la población. Deberán cumplirse las condiciones de la distribución binomial, a saber: que haya un cierto número n de ensayos independientes, lo que significa un muestreo aleatorio; que los resultados de cualquier ensayo sean binarios, éxito o fracaso, y que cada ensayo tenga la misma probabilidad de éxito p. La forma de la distribución binomial debe ser similar a la forma de la distribución normal. Para ello, las cantidades np′ y nq′ deben ser ambas mayores que cinco (np′ > 5 y nq′ > 5). En este caso, la distribución binomial de una proporción muestral (estimada) se calcula aproximadamente por la distribución normal con y . Recuerde que . No hay ninguna distribución que corrija este pequeño sesgo de la muestra; por ende, si no se cumplen estas condiciones, simplemente no podemos probar la hipótesis con los datos disponibles en ese momento. Cumplimos esta condición cuando estimamos por primera vez los intervalos de confianza para p.
Nuevamente, comenzamos con la fórmula normalizadora modificada porque se trata de la distribución de una binomial.
Al sustituir , el valor hipotético de p, tenemos:
Es el estadístico de prueba para comprobar los valores hipotéticos de p, cuando las hipótesis nula y alternativa adoptan una de las siguientes formas:
La regla de decisión indicada anteriormente se aplica también en este caso: si el valor calculado de Zc muestra que la proporción de la muestra está a "demasiadas" desviaciones típicas de la proporción hipotética, no se puede aceptar la hipótesis nula. La decisión sobre lo que es "demasiado" está predeterminada por el analista en función del nivel de significación requerido en la prueba.
### Repaso del capítulo
La prueba de hipótesis en sí tiene un proceso establecido. Esto se sintetiza de la siguiente manera
### Revisión de la fórmula
Estadística de una prueba de hipótesis de proporciones:
### Tarea para la casa
Instrucciones: En los diez ejercicios siguientes, Comprobación de hipótesis: Responda cada una de las preguntas de los diez ejercicios siguientes.
1. Indique la hipótesis nula y la alternativa.
2. Indique el valor p.
3. Indique alfa.
4. ¿Cuál es su decisión?
5. Escriba una conclusión.
6. Responde cualquier otra pregunta que se le plantee en el problema.
### Referencias
Datos de Amit Schitai. Director de tecnología educativa y aprendizaje a distancia. LBCC.
Datos de Bloomberg Businessweek. Disponible en línea en http://www.businessweek.com/news/2011- 09-15/nyc-smoking-rate-falls-to-record-low-of-14-bloomberg-says.html.
Datos de energy.gov. Disponible en línea en http://energy.gov (consultado el 27 de junio de 2013).
Datos de Gallup®. Disponible en línea en www.gallup.com (consultado el 27 de junio de 2013).
Datos de Growing by Degrees de Allen y Seaman.
Datos de La Leche League International. Disponible en línea en http://www.lalecheleague.org/Law/BAFeb01.html.
Datos de la Asociación Americana del Automóvil. Disponible en línea en www.aaa.com (consultado el 27 de junio de 2013).
Datos de la Asociación Americana de Bibliotecas. Disponible en línea en www.ala.org (consultado el 27 de junio de 2013).
Datos de la Oficina de Estadísticas Laborales. Disponible en línea en http://www.bls.gov/oes/current/oes291111.htm.
Datos de los Centros para el Control y la Prevención de Enfermedades. Disponible en línea en www.cdc.gov (consultado el 27 de junio de 2013)
Datos de la Oficina del Censo de EE. UU., disponibles en línea en http://quickfacts.census.gov/qfd/states/00000.html (consultado el 27 de junio de 2013).
Datos de la Oficina del Censo de Estados Unidos. Disponible en línea en http://www.census.gov/hhes/socdemo/language/.
Datos de Toastmasters International. Disponible en línea en http://toastmasters.org/artisan/detail.asp?CategoryID=1&SubCategoryID=10&ArticleID=429&Page=1.
Datos de Weather Underground. Disponible en línea en www.wunderground.com (consultado el 27 de junio de 2013).
Oficina Federal de Investigaciones. “Uniform Crime Reports and Index of Crime in Daviess in the State of Kentucky enforced by Daviess County from 1985 to 2005”. Disponible en línea en http://www.disastercenter.com/kentucky/crime/3868.htm (consultado el 27 de junio de 2013).
“Foothill-De Anza Community College District”. De Anza College, invierno de 2006. Disponible en línea en http://research.fhda.edu/factbook/DAdemofs/Fact_sheet_da_2006w.pdf.
Johansen, C., J. Boice, Jr., J. McLaughlin, J. Olsen. “Cellular Telephones and Cancer—a Nationwide Cohort Study in Denmark”. Institute of Cancer Epidemiology and the Danish Cancer Society, 93(3):203-7. Disponible en línea en http://www.ncbi.nlm.nih.gov/pubmed/11158188 (consultado el 27 de junio de 2013).
Rape, Abuse & Incest National Network. “How often does sexual assault occur?”. RAINN, 2009. Disponible en línea en http://www.rainn.org/get-information/statistics/frequency-of-sexual-assault (consultado el 27 de junio de 2013). |
# Pruebas de hipótesis con dos muestras
## Introducción
Los estudios suelen comparar dos grupos. Por ejemplo, los investigadores están interesados en el efecto que tiene la aspirina en la prevención de ataques al corazón. Durante los años recientes, los periódicos y las revistas han informado de varios estudios sobre la aspirina en los que participan dos grupos. Normalmente, un grupo recibe aspirina y el otro un placebo. Luego, se estudia la tasa de infarto durante varios años.
Hay otras situaciones que tratan de la comparación de dos grupos. Por ejemplo, los estudios comparan varios programas de dieta y ejercicio. Los políticos comparan la proporción de personas de diferentes niveles de ingresos que podrían votar por ellos. Los estudiantes se interesan por saber si los cursos de preparación para la SAT o el Examen de Registro de Graduados (Graduate Record Exam, GRE) ayudan realmente a mejorar sus calificaciones. Muchas aplicaciones empresariales requieren la comparación de dos grupos. Puede tratarse de la rentabilidad de dos estrategias distintas de inversión o de las diferencias en la eficiencia de la producción de distintos estilos de gestión.
Para comparar dos medias o dos proporciones, se trabaja con dos grupos. Los grupos se clasifican como independientes o pares coincidentes. Los grupos independientes consisten en dos muestras que son independientes, es decir, los valores de la muestra seleccionados de una población no están relacionados de ninguna manera con los valores de la muestra seleccionados de la otra población. Los pares coincidentes consisten en dos muestras que son dependientes. El parámetro que se comprueba utilizando pares coincidentes es la media de la población. Los parámetros que se prueban con grupos independientes son las medias de la población o las proporciones de la población de cada grupo. |
# Pruebas de hipótesis con dos muestras
## Comparación de las medias de dos poblaciones independientes
La comparación de dos medias poblacionales independientes es muy común y proporciona una forma de probar la hipótesis de que los dos grupos difieren entre sí. ¿Es el turno de noche menos productivo que el de día, las tasas de rendimiento de las inversiones en activos fijos son diferentes a las de las inversiones en acciones ordinarias, etc.? Una diferencia observada entre dos medias muestrales depende tanto de las medias como de las desviaciones típicas de la muestra. Pueden producirse medias muy diferentes por azar si hay una gran variación entre cada una de las muestras. El estadístico de prueba tendrá que tener en cuenta este hecho. La prueba que compara dos medias poblacionales independientes con desviaciones típicas poblacionales desconocidas y posiblemente desiguales se denomina prueba t de Aspin-Welch. Aspin-Welch ideó la fórmula de los grados de libertad que veremos más adelante.
Cuando desarrollamos la prueba de hipótesis para la media y las proporciones, comenzamos con el teorema del límite central. Reconocemos que la media muestral procede de una distribución de medias muestrales, y las proporciones muestrales proceden de la distribución muestral de las proporciones muestrales. Esto convirtió nuestros parámetros, las medias y las proporciones muestrales, en variables aleatorias. Era importante para nosotros conocer la distribución de la que procedían estas variables aleatorias. El teorema del límite central nos dio la respuesta: la distribución normal. Nuestras estadísticas Z y t provienen de este teorema. Esto nos proporcionó la solución a nuestra pregunta de cómo medir la probabilidad de que la media muestral provenga de una distribución con un valor hipotético particular de la media o proporción. En ambos casos esa era la pregunta: ¿Cuál es la probabilidad de que la media (o proporción) de nuestros datos muestrales proceda de una distribución poblacional con el valor hipotético que nos interesa?
Ahora nos interesa saber si dos muestras tienen o no la misma media. Nuestra pregunta no ha cambiado: ¿Proceden estas dos muestras de la misma distribución poblacional? Para abordar este problema creamos una nueva variable aleatoria. Reconocemos que tenemos dos medias muestrales: una de cada conjunto de datos. Así, tenemos dos variables aleatorias, procedentes de dos distribuciones desconocidas. Para resolver el problema creamos una nueva variable aleatoria: la diferencia entre las medias muestrales. Dicha variable también tiene una distribución. Nuevamente, el teorema del límite central nos indica que esta nueva distribución se distribuye normalmente, sin importar las distribuciones subyacentes de los datos originales. Un gráfico despejaría este concepto.
En la imagen aparecen dos distribuciones de datos, X1 y X2, con medias y desviaciones típicas desconocidas. El segundo panel muestra la distribución muestral de la variable aleatoria recién creada (). Esta es la distribución teórica de muchas medias muestrales de la población 1 menos las medias muestrales de la población 2. El teorema del límite central señala que esta distribución muestral teórica de las diferencias de las medias muestrales se distribuye normalmente, sin importar la distribución de los datos reales de la población que se muestran en el panel superior. Dado que la distribución del muestreo se distribuye normalmente, podemos desarrollar una fórmula de estandarización y calcular las probabilidades a partir de la distribución normal estándar del panel inferior, la distribución Z. Ya hemos visto este mismo análisis en la Figura 7.2 del Capítulo 7.
El teorema del límite central, como antes, nos proporciona la desviación típica de la distribución muestral y, además, que el valor previsto de la media de la distribución de las diferencias de las medias muestrales es igual a las diferencias de las medias poblacionales. Matemáticamente, esto se formula de la siguiente manera:
Ya que desconocemos las desviaciones típicas de la población, las calculamos con las dos desviaciones típicas de nuestras muestras independientes. En la prueba de hipótesis, calculamos la desviación típica o el error estándar, de la diferencia de las medias muestrales,
–
.
Recordemos que la sustitución de la varianza de la muestra por la varianza de la población cuando no teníamos la varianza de la población fue la técnica que utilizamos al construir el intervalo de confianza y el estadístico de prueba para comprobar la hipótesis con respecto a una sola media en Intervalos de confianza y Pruebas de hipótesis con una muestra. El estadístico de prueba (puntuación
El número de grados de libertad ( requiere un cálculo algo complicado. Los df no son siempre un número entero. El anterior estadístico de prueba se calcula aproximadamente mediante la distribución t de Student con df de la siguiente manera:
Cuando los tamaños de las muestras n1 y n2 son de 30 o más, la aproximación de la t de Student es muy buena. Si cada muestra tiene más de 30 observaciones, los grados de libertad pueden calcularse como n1 + n2 - 2.
El formato de la distribución muestral, las diferencias de medias muestrales, especifica que el formato de las hipótesis nula y alternativa es:
donde δ0 es la diferencia hipotética entre las dos medias. Si la pregunta es simplemente: "¿Hay alguna diferencia entre las medias?", entonces δ0 = 0 y las hipótesis nula y alternativa pasan a ser:
Un ejemplo de cuándo δ0 puede no ser cero es cuando la comparación de los dos grupos requiere una diferencia específica para que la decisión sea significativa. Imagine que está haciendo una inversión de capital. Piensa en cambiar su modelo de máquina actual por otro. La productividad de sus máquinas se mide por la velocidad a la que producen el producto. Puede ser que un contendiente para sustituir al modelo antiguo sea más rápido en términos de rendimiento del producto, pero también es más caro. La segunda máquina también puede tener más costes de mantenimiento, de instalación, etc. La hipótesis nula se establecería de forma que la nueva máquina tendría que ser mejor que la antigua en la medida suficiente para cubrir estos costes adicionales en términos de velocidad y coste de producción. Esta forma de las hipótesis nula y alternativa muestra lo valiosa que puede ser esta comprobación de la hipótesis en particular. Para la mayor parte de nuestro trabajo, comprobaremos hipótesis simples al indagar si hay alguna diferencia entre las dos medias de distribución.
### Referencias
Datos de las carreras de Ingeniería e Informática. Disponible en línea en http://www.graduatingengineer.com
Datos de Microsoft Bookshelf.
Datos del sitio web del Senado de Estados Unidos, disponibles en línea en www.Senate.gov (consultado el 17 de junio de 2013).
“Lista de los actuales senadores de Estados Unidos por edad”. Wikipedia. Disponible en línea en http://en.wikipedia.org/wiki/List_of_current_United_States_Senators_by_age (consultado el 17 de junio de 2013).
“Sectorización por grupos industriales”. Nasdaq. Disponible en línea en http://www.nasdaq.com/markets/barchart-sectors.aspx?page=sectors&base=industry (consultado el 17 de junio de 2013).
“Clubes de desnudistas: donde se da la prostitución y la trata”. Investigación y educación sobre la prostitución, 2013. Disponible en línea en www.prostitutionresearch.com/ProsViolPosttrauStress.html (consultado el 17 de junio de 2013).
“Historia de las Series Mundiales”. Almanaque de béisbol, 2013. Disponible en línea en http://www.baseball-almanac.com/ws/wsmenu.shtml (consultado el 17 de junio de 2013).
### Repaso del capítulo
Dos medias poblacionales de muestras independientes en las que se desconocen las desviaciones típicas de la población.
1. Variable aleatoria:
= la diferencia de las medias muestrales
2. Distribución: Distribución t de Student con grados de libertad (varianzas sin agrupar).
### Repaso de la fórmula
Error estándar: SE =
Estadístico de prueba (puntuación t): t =
Grados de libertad:
donde:
y son las desviaciones típicas de la muestra, y y son los tamaños de las muestras.
y
son las medias muestrales.
Use la siguiente información para responder los próximos 15 ejercicios: Indique si la prueba de hipótesis es para:
1. medias de grupos independientes, desviaciones típicas de la población o varianzas conocidas
2. medias de grupos independientes, desviaciones típicas de la población o varianzas desconocidas
3. muestras coincidentes o emparejadas
4. media simple
5. dos proporciones
6. proporción única
Use la siguiente información para responder los próximos tres ejercicios: se realiza un experimento para determinar cuál de dos bebidas gaseosas tiene más azúcar. Hay 13 latas de la bebida A en una muestra y seis latas de la bebida B. La cantidad media de azúcar en la bebida A es de 36 gramos con desviación típica de 0,6 gramos. La cantidad media de azúcar en la bebida B es de 38 gramos con desviación típica de 0,8 gramos. Los investigadores creen que la bebida B tiene más azúcar que la bebida A, en promedio. Ambas poblaciones tienen distribuciones normales.
Utilice la siguiente información para responder los siguientes 12 ejercicios: El Centro para el Control y la Prevención de Enfermedades de EE. UU. informa que la esperanza de vida media era de 47,6 años para personas blancas nacidas en 1900 y de 33,0 años para las personas que no son blancas. Supongamos que usted realiza un estudio aleatorio de los registros de defunción de las personas nacidas en 1900 en un determinado condado. De las 124 personas blancas, la media de vida era de 45,3 años, con una desviación típica de 12,7 años. De las 82 personas que no son blancas, la media de vida era de 34,1 años, con una desviación típica de 15,6 años. Realice una prueba de hipótesis para ver si la media de vida en el condado es la misma para las personas blancas y las que no son blancas.
### Tarea para la casa
Utilice la información del Conjunto de datos del
Utilice la siguiente información para responder los dos ejercicios siguientes. Las conferencias Este y Oeste de la Liga Mayor de Fútbol cuentan con una nueva división de reserva que permite a los nuevos jugadores desarrollar sus habilidades. Los datos de una fecha elegida al azar mostraron los siguientes objetivos anuales.
Realice una prueba de hipótesis para responder los dos ejercicios siguientes. |
# Pruebas de hipótesis con dos muestras
## Criterios de Cohen para efectos de tamaño pequeño, mediano y grande
La es una medida del "tamaño del efecto", basada en las diferencias entre dos medias. La d de Cohen, llamada así por el estadístico estadounidense Jacob Cohen, mide la fuerza relativa de las diferencias entre las medias de dos poblaciones a partir de los datos de la muestra. El valor calculado del tamaño del efecto se compara entonces con los criterios de Cohen de efecto de tamaño pequeño, mediano y grande.
La d de Cohen es la medida de la diferencia entre dos medias dividida entre la desviación típica combinada:
donde
Cabe destacar que la d de Cohen no proporciona un nivel de confianza en cuanto al tamaño del efecto comparable a las otras comprobaciones de hipótesis que hemos estudiado. Los tamaños de los efectos son simplemente indicativos.
### Repaso del capítulo
La d de Cohen es una medida del "tamaño del efecto", basada en las diferencias entre dos medias.
Cabe destacar que la d de Cohen no proporciona un nivel de confianza en cuanto al tamaño del efecto comparable a las otras comprobaciones de hipótesis que hemos estudiado. Los tamaños de los efectos son simplemente indicativos.
### Revisión de la fórmula
La d de Cohen es la medida del tamaño del efecto:
donde
|
# Pruebas de hipótesis con dos muestras
## Prueba de diferencias de medias: suponer varianzas de población iguales
Normalmente, nunca esperamos conocer ninguno de los parámetros de la población, la media, la proporción o la desviación típica. Cuando se comprueban hipótesis relativas a diferencias de medias, nos enfrentamos a la dificultad de dos varianzas desconocidas que desempeñan un papel fundamental en el estadístico de prueba. Hemos sustituido las varianzas de la muestra tal y como hicimos al comprobar las hipótesis para una única media. Tal como lo hicimos anteriormente, utilizamos una t de Student para compensar esta falta de información sobre la varianza de la población. Sin embargo, hay situaciones en las que no conocemos las varianzas de la población, aunque podemos asumir que las dos poblaciones tienen la misma varianza. Si esto es así, entonces la varianza de la muestra conjunta será menor que las varianzas de las muestras individuales. Así se obtienen estimaciones más precisas y se reduce la probabilidad de descartar un buen nulo. Las hipótesis nula y alternativa siguen siendo las mismas, pero el estadístico de prueba cambia a:
donde es la varianza combinada dada por la fórmula:
### Repaso del capítulo
En situaciones en las que desconocemos las varianzas de la población, pero suponemos que son las mismas, la varianza de la muestra conjunta será menor que las varianzas de la muestra individual.
Así se obtienen estimaciones más precisas y se reduce la probabilidad de descartar un buen nulo.
### Revisión de la fórmula
donde es la varianza combinada dada por la fórmula:
|
# Pruebas de hipótesis con dos muestras
## Comparación de dos proporciones de población independientes
Cuando se realiza una prueba de hipótesis que compara dos proporciones de población independientes se deben dar las siguientes características:
1. Las dos muestras independientes son muestras aleatorias que son independientes.
2. El número de aciertos es, al menos, cinco y el número de fallos es, al menos, cinco para cada una de las muestras.
3. La bibliografía, cada vez más extensa, afirma que la población deberá ser, como mínimo, 10 y hasta 20 veces el tamaño de la muestra. Así se evita que cada población sea objeto de un muestreo excesivo y que los resultados sean sesgados.
La comparación de dos proporciones, al igual que la comparación de dos medias, es de uso común. Si dos proporciones estimadas son diferentes, puede deberse a una diferencia en las poblaciones o al azar en el muestreo. La comprobación de la hipótesis permite determinar si una diferencia en las proporciones estimadas refleja una diferencia en las dos proporciones de la población.
Al igual que en el caso de las diferencias de medias muestrales, construimos una distribución muestral para las diferencias de proporciones muestrales: donde y
son las proporciones de la muestra para los dos conjuntos de datos en cuestión. XA y XB son el número de aciertos en cada grupo de la muestra, respectivamente, y nA y nB son los tamaños de muestra respectivos de los dos grupos. De nuevo acudimos al teorema del límite central para hallar la distribución muestral con respecto a las diferencias en las proporciones de la muestra. También nos encontramos con que esta distribución muestral, al igual que las anteriores, se distribuye normalmente, tal y como demuestra el teorema del límite central, como se ve en la .
En general, la hipótesis nula permite probar una diferencia de un valor determinado, 𝛿0, tal como hicimos para el caso de las diferencias de medias.
Sin embargo, lo más común es la prueba de que las dos proporciones son iguales. Esto es,
Para llevar a cabo la prueba utilizamos una proporción combinada, p.
donde δ0 son las diferencias hipotéticas entre las dos proporciones y pc es la varianza agrupada de la fórmula anterior.
### Referencias
Datos de Educational Resources, catálogo de diciembre.
Datos de los Hoteles Hilton. Disponible en línea en http://www.hilton.com (consultado el 17 de junio de 2013).
Datos de los Hoteles Hyatt. Disponible en línea en http://hyatt.com (consultado el 17 de junio de 2013).
Datos de Estadísticas del Departamento de Salud y Servicios Humanos de Estados Unidos.
Datos de la Exposición del Whitney en préstamo al Museo de Arte de San José.
Datos de la Sociedad Americana del Cáncer. Disponible en línea en http://www.cancer.org/index (consultado el 17 de junio de 2013).
Datos de la Chancellor's Office, California Community Colleges, noviembre de 1994.
“State of the States”. Gallup, 2013. Disponible en línea en http://www.gallup.com/poll/125066/State-States.aspx?ref=interactive (consultado el 17 de junio de 2013).
“West Nile Virus”. Centers for Disease Control and Prevention. Disponible en línea en http://www.cdc.gov/ncidod/dvbid/westnile/index.htm (consultado el 17 de junio de 2013).
### Repaso del capítulo
Prueba de dos proporciones poblacionales a partir de muestras independientes
### Revisión de la fórmula
Proporción combinada: =
Estadístico de prueba (puntuación z):
donde
y son las proporciones de la muestra, y son las proporciones de la población,
P es la proporción combinada, y n y n son los tamaños de las muestras.
Use la siguiente información para los próximos cinco ejercicios. Se están probando dos tipos de sistemas operativos (operating system, OS) de teléfonos para determinar si hay una diferencia en las proporciones de fallos del sistema (caídas). Quince de una muestra aleatoria de 150 teléfonos con OS1 tuvieron fallos del sistema en las primeras ocho horas de funcionamiento. Nueve de otra muestra aleatoria de 150 teléfonos con OS2 tuvieron fallos del sistema en las primeras ocho horas de funcionamiento. Se cree que el OS2 es más estable (tiene menos fallos) que el OS1.
Use la siguiente información para responder los próximos doce ejercicios. En el reciente censo el tres por ciento de la población de EE. UU. declaró que era de dos o más razas. Sin embargo, el porcentaje varía enormemente de un estado a otro. Supongamos que se realizan dos encuestas aleatorias. En la primera encuesta aleatoria, de 1.000 habitantes de Dakota del Norte, solo nueve personas declararon que son de dos o más razas. En la segunda encuesta aleatoria, de 500 nevadenses, 17 personas declararon que son de dos o más razas. Realice una prueba de hipótesis para determinar si los porcentajes de población son iguales para los dos estados o si el porcentaje de Nevada es estadísticamente mayor que el de Dakota del Norte.
### Tarea para la casa
Use la siguiente información para responder los próximos tres ejercicios. El virus neuroinvasivo del Nilo Occidental es una enfermedad grave que afecta el sistema nervioso de las personas. Lo transmite la especie de mosquito Culex. En Estados Unidos en 2010 se registraron 629 casos del virus neuroinvasivo del Nilo Occidental de un total de 1.021 casos notificados, y en 2011 se registraron 486 casos neuroinvasivos de un total de 712 casos. ¿La proporción de casos del virus neuroinvasivo del Nilo Occidental en 2011 es mayor que la proporción de casos de 2010? Use un nivel de significación del 1 % y haga una prueba de hipótesis adecuada.
|
# Pruebas de hipótesis con dos muestras
## Dos medias poblacionales con desviaciones típicas conocidas
Aunque difícilmente se dé esta situación (conocer las desviaciones típicas de la población es improbable), el siguiente ejemplo ilustra las pruebas de hipótesis para medias independientes con desviaciones típicas conocidas de la población. La distribución muestral para la diferencia entre las medias es normal de acuerdo con el teorema del límite central. La variable aleatoria es
. La distribución normal tiene el siguiente formato:
### Referencias
Datos de la Oficina del Censo de Estados Unidos. Disponible en línea en http://www.census.gov/prod/cen2010/briefs/c2010br-02.pdf
Hinduja, Sameer. “Sexting Research and Gender Differences”. Cyberbulling Research Center, 2013. Disponible en línea en http://cyberbullying.us/blog/sexting-research-and-gender-differences/ (consultado el 17 de junio de 2013).
“Smart Phone Users, By the Numbers”. Visually, 2013. Disponible en línea en http://visual.ly/smart-phone-users-numbers (consultado el 17 de junio de 2013).
Smith, Aaron. “35% of American adults own a Smartphone”. Pew Internet, 2013. Disponible en línea en http://www.pewinternet.org/~/media/Files/Reports/2011/PIP_Smartphones.pdf (consultado el 17 de junio de 2013).
“State-Specific Prevalence of Obesity AmongAduls—Unites States, 2007”. MMWR, CDC. Disponible en línea en http://www.cdc.gov/mmwr/preview/mmwrhtml/mm5728a1.htm (consultado el 17 de junio de 2013).
“Texas Crime Rates 1960–1012”. FBI, Uniform Crime Reports, 2013. Disponible en línea en: http://www.disastercenter.com/crime/txcrime.htm (consultado el 17 de junio de 2013).
### Repaso del capítulo
Una prueba de hipótesis de dos medias poblacionales de muestras independientes en las que se conocen las desviaciones típicas de la población tendrá estas características:
### Revisión de la fórmula
Estadístico de prueba (puntuaciónz):
donde: y son las desviaciones típicas conocidas de la población. n1 y n2 son los tamaños de las muestras.
y
son las medias muestrales. μ1 y μ2 son las medias poblacionales.
Use la siguiente información para responder los próximos cinco ejercicios. Se van a comparar las velocidades medias de los lanzamientos de pelotas rápidas de dos lanzadores de béisbol diferentes. Se mide una muestra de 14 lanzamientos de pelotas rápidas de cada lanzador. Las poblaciones tienen distribuciones normales. La muestra el resultado. Los cazatalentos creen que Rodríguez lanza una pelota rápida más rápida.
Use la siguiente información para responder los próximos cinco ejercicios. Un investigador está probando los efectos de los alimentos para plantas en su crecimiento. Nueve plantas han recibido el alimento para plantas. Otras nueve plantas no han recibido el alimento para plantas. Las alturas de las plantas se registran después de ocho semanas. Las poblaciones tienen distribuciones normales. El resultado está en la siguiente tabla. El investigador cree que la comida hace que las plantas crezcan más altas.
Use la siguiente información para responder los siguientes cinco ejercicios. Se están considerando dos aleaciones metálicas como material para los rodamientos de pelotas. Hay que comparar el punto de fusión medio de las dos aleaciones. Se están probando 15 piezas de cada metal. Ambas poblaciones tienen distribuciones normales. El resultado está en la siguiente tabla. Se cree que la aleación zeta tiene un punto de fusión diferente.
### Tarea para la casa
|
# Pruebas de hipótesis con dos muestras
## Muestras coincidentes o emparejadas
En la mayoría de los casos de datos económicos o empresariales tenemos poco o ningún control sobre la recopilación de los datos. En este sentido, los datos no son el resultado de un experimento controlado y planificado. Sin embargo, en algunos casos podemos generar datos que forman parte de un experimento controlado. Esto se da con frecuencia en situaciones de control de calidad. Imagine que las tasas de producción de dos máquinas construidas con el mismo diseño, pero en diferentes plantas de fabricación, se prueban para detectar diferencias en algún sistema de medición de la producción, como la velocidad de salida o el cumplimiento con alguna especificación, como la resistencia del producto. La prueba tiene el mismo formato que la que hemos estado probando, pero aquí podemos tener pares emparejados para los que podemos verificar si existen diferencias. Cada observación tiene su par emparejado con el que se calculan las diferencias. En primer lugar, hay que calcular las diferencias del indicador que se va a probar entre las dos listas de observaciones, lo que se suele etiquetar con la letra "d". A continuación, el promedio de estas diferencias emparejadas, se calcula al igual que su desviación típica, Sd. Esperamos que la desviación típica de las diferencias de los pares emparejados sea menor que la de los pares no emparejados porque presumiblemente deberían existir menos diferencias debido a la correlación entre los dos grupos.
Cuando se utiliza una prueba de hipótesis para muestras emparejadas o pareadas, pueden darse las siguientes características:
1. Se utiliza un muestreo aleatorio simple.
2. El tamaño de las muestras suele ser pequeño.
3. Se toman dos medidas (muestras) del mismo par de personas u objetos.
4. Las diferencias se calculan a partir de las muestras coincidentes o emparejadas.
5. Las diferencias forman la muestra que se utiliza para la prueba de hipótesis.
6. O bien los pares coincidentes tienen diferencias que provienen de una población que es normal o el número de diferencias es lo suficientemente grande como para que la distribución de la media muestral de las diferencias sea aproximadamente normal.
En una prueba de hipótesis para muestras coincidentes o emparejadas los sujetos son coincidentes en pares y se calculan las diferencias. Las diferencias son los datos. A continuación, la media poblacional de las diferencias, μ, se verifica con la prueba t de Student para una única media poblacional con n – 1 grados de libertad, donde n es el número de diferencias, es decir, el número de pares, no el número de observaciones.
### Repaso del capítulo
Una prueba de hipótesis para muestras coincidentes o emparejadas (prueba t) tiene estas características:
1. Compruebe las diferencias restando una medida de la otra
2. Variable aleatoria:
= media de las diferencias
3. Distribución: Distribución t de Student con n – 1 grados de libertad
4. Si el número de diferencias es pequeño (menos de 30), las diferencias deben seguir una distribución normal.
5. Se extraen dos muestras del mismo conjunto de objetos.
6. Las muestras son dependientes.
### Revisión de la fórmula
Estadístico de prueba (puntuación t): t =
donde:
es la media de las diferencias de la muestra. μd es la media de las diferencias de la población. s es la desviación típica de la muestra de las diferencias. n es el tamaño de la muestra.
Use la siguiente información para responder los próximos cinco ejercicios. Se realizó un estudio para comprobar la eficacia de un parche de software en la reducción de fallos del sistema durante un periodo de seis meses. Los resultados de instalaciones seleccionadas al azar se muestran en la . El valor “antes” se compara con un valor “después” y se calculan las diferencias. Las diferencias tienen una distribución normal. Prueba al nivel de significación del 1 %.
Use la siguiente información para responder los próximos cinco ejercicios. Se realizó un estudio para comprobar la eficacia de una clase de malabares. Antes de que empezara la clase seis sujetos hicieron malabares con todas las pelotas que pudieron a la vez. Después de la clase, los mismos seis sujetos hicieron todos los malabares que pudieron con las pelotas. Se calculan las diferencias en el número de pelotas. Las diferencias tienen una distribución normal. Prueba al nivel de significación del 1 %.
Use la siguiente información para responder los siguientes cinco ejercicios. Un médico quiere saber si un medicamento para la presión arterial es eficaz. A seis sujetos se les toma la presión arterial y se registra. Después de doce semanas de uso del medicamento, se vuelve a tomar la presión arterial de los mismos seis sujetos. Para esta prueba, solo se considera la presión sistólica. Prueba al nivel de significación del 1 %.
### Tarea para la casa
Use la siguiente información para responder los próximos dos ejercicios. Se probó un nuevo medicamento para la prevención del sida en un grupo de 224 pacientes con VIH positivo. Cuarenta y cinco pacientes desarrollaron sida después de cuatro años. En un grupo de control de 224 pacientes con VIH positivo, 68 desarrollaron sida al cabo de cuatro años. Queremos comprobar si el método de tratamiento reduce la proporción de pacientes que desarrollan sida al cabo de cuatro años o si las proporciones del grupo tratado y del grupo no tratado se mantienen igual.
Supongamos que el subíndice t = paciente tratado y nt = paciente no tratado.
Use la siguiente información para responder los próximos dos ejercicios. Se realiza un experimento para demostrar que la presión arterial se puede reducir conscientemente en personas entrenadas en un “programa de ejercicios de biorrealimentación”. Se seleccionaron seis sujetos al azar y se registraron las mediciones de la presión arterial antes y después del entrenamiento. Se calculó la diferencia entre las presiones sanguíneas (después – antes) lo que arrojó los siguientes resultados
= −10,2 s = 8,4. Use los datos y compruebe la hipótesis de que la presión arterial ha disminuido después del entrenamiento.
### Resúmalo todo
Use la siguiente información para responder los próximos diez ejercicios. Indique cuál de las siguientes opciones identifica mejor la prueba de hipótesis.
1. medias de grupos independientes, desviaciones típicas de la población o varianzas conocidas
2. medias de grupos independientes, desviaciones típicas de la población o varianzas desconocidas
3. muestras coincidentes o emparejadas
4. media simple
5. dos proporciones
6. proporción única |
# La distribución chi-cuadrado
## Introducción
¿Alguna vez se ha preguntado si los números ganadores de la lotería se distribuyen uniformemente o si algunos números se producen con mayor frecuencia? ¿Qué tal si los tipos de películas que prefiere las personas son diferentes en los distintos grupos de edad? ¿Y si una máquina de café dispensara aproximadamente la misma cantidad de café cada vez? Podría responder estas preguntas mediante una prueba de hipótesis.
Ahora estudiará una nueva distribución, la cual se utiliza para determinar las respuestas de estas preguntas. Esta distribución se denomina distribución chi-cuadrado.
En este capítulo aprenderá las tres principales aplicaciones de la distribución chi-cuadrado
|
# La distribución chi-cuadrado
## Datos sobre la distribución chi-cuadrado
La notación para la distribución chi-cuadrado es:
donde df = grados de libertad, lo cual depende de cómo se utilice el chi-cuadrado (si quiere practicar el cálculo de probabilidades chi-cuadrado, utilice df = n – 1. Los grados de libertad para los tres usos principales se calculan cada uno de forma diferente).
Para la distribución χ, la media poblacional es μ = df y la desviación típica poblacional es
.
La variable aleatoria se muestra como χ.
La variable aleatoria para una distribución chi-cuadrado con k grados de libertad es la suma de variables k normales cuadradas independientes.
χ2 = (Z1)2 + (Z2)2 + ... + (Zk)2
1. La curva no es simétrica y es asimétrica hacia la derecha.
2. Hay una curva de chi-cuadrado diferente para cada df.
3. El estadístico de prueba para cualquier prueba es siempre mayor o igual a cero.
4. Cuando df > 90, la curva chi-cuadrado se aproxima a la distribución normal. Para X ~
la media, μ = df = 1.000 y la desviación típica, σ =
= 44,7. Por tanto, X ~ N(1.000, 44,7), aproximadamente.
5. La media, μ, se encuentra justo a la derecha del pico.
### Referencias
Datos de la Revista Parade.
“HIV/AIDS Epidemiology Santa Clara County”, Departamento de Salud Pública del condado de Santa Clara, mayo de 2011.
### Repaso del capítulo
La distribución chi-cuadrado es una herramienta útil para la evaluación en una serie de categorías de problemas. Estas categorías de problemas incluyen principalmente (i) si un conjunto de datos se ajusta a una determinada distribución; (ii) si las distribuciones de dos poblaciones son iguales; (iii) si dos eventos pueden ser independientes; y (iv) si hay una variabilidad diferente a la esperada dentro de una población.
Un parámetro importante en una distribución chi-cuadrado son los grados de libertad df en un problema dado. La variable aleatoria en la distribución chi-cuadrado es la suma de cuadrados de df variables normales estándar, los cuales deben ser independientes. Las características clave de la distribución chi-cuadrado también dependen directamente de los grados de libertad.
La curva de la distribución chi-cuadrado es asimétrica hacia la derecha, y su forma depende de los grados de libertad df. Para df > 90, la curva se aproxima a la distribución normal. Los estadísticos de prueba basados en la distribución chi-cuadrado son siempre mayores o iguales a cero. Estas pruebas de aplicación son casi siempre pruebas de cola derecha.
### Revisión de la fórmula
χ2 = (Z1)2 + (Z2)2 + … (Z)2 variable aleatoria de distribución chi-cuadrado
μ = df distribución chi-cuadrado media de la población
Distribución chi-cuadrado de la desviación típica de la población
### Tarea para la casa
Decida si las siguientes afirmaciones son verdaderas o falsas. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.