6 Independence, Conditional Probability, and the Multiplication Rule
Tải bản đầy đủ - 0trang
150
❍
CHAPTER 4 PROBABILITY AND PROBABILITY DISTRIBUTIONS
EXAMPLE
4.19
In a color preference experiment, eight toys are placed in a container. The toys
are identical except for color—two are red, and six are green. A child is asked to
choose two toys at random. What is the probability that the child chooses the two red
toys?
You can visualize the experiment using a tree diagram as shown in
Figure 4.13. Deﬁne the following events:
Solution
R: Red toy is chosen
G: Green toy is chosen
FI GU R E 4 .1 3
Tree diagram for
Example 4.19
●
First choice
Second choice
Red (1/7)
Simple event
RR
Red (2/8)
Green (6/7)
Red (2/7)
RG
GR
Green (6/8)
Green (5/7)
GG
The event A (both toys are red) can be constructed as the intersection of two events:
A ϭ (R on ﬁrst choice) ʝ (R on second choice)
Since there are only two red toys in the container, the probability of choosing red on
the ﬁrst choise is 2/8. However, once this red toy has been chosen, the probability of
red on the second choice is dependent on the outcome of the ﬁrst choice (see Figure 4.13). If the ﬁrst choice was red, the probability of choosing a second red toy is
only 1/7 because there is only one red toy among the seven remaining. If the ﬁrst
choice was green, the probability of choosing red on the second choice is 2/7 because
there are two red toys among the seven remaining. Using this information and the
Multiplication Rule, you can ﬁnd the probability of event A.
P(A) ϭ P(R on ﬁrst choice ʝ R on second choice)
ϭ P(R on ﬁrst choice) P(R on second choice)͉R on ﬁrst)
2 1
2
1
ϭ ᎏᎏ ᎏᎏ ϭ ᎏᎏ ϭ ᎏᎏ
8 7
56
28
Sometimes you may need to use the Multiplication Rule in a slightly different form,
so that you can calculate the conditional probability, P(AͦB). Just rearrange the terms
in the Multiplication Rule.
4.6 INDEPENDENCE, CONDITIONAL PROBABILITY, AND THE MULTIPLICATION RULE
❍
151
CONDITIONAL PROBABILITIES
The conditional probability of event A, given that event B has occurred is
P(A ʝ B)
P(A͉B) ϭ ᎏᎏ if P(B) 0
P(B)
The conditional probability of event B, given that event A has occurred is
P(A ʝ B)
P(B͉A) ϭ ᎏᎏ
P(A)
if
P(A)
0
Suppose that in the general population, there are
51% men and 49% women, and that the proportions of colorblind men and women
are shown in the probability table below:
Colorblindness, continued
Men(B)
Women (BC)
Total
Colorblind (A)
Not Colorblind (AC)
.04
.47
.002
.488
.042
.958
Total
.51
.49
1.00
If a person is drawn at random from this population and is found to be a man (event
B), what is the probability that the man is colorblind (event A)? If we know that the
event B has occurred, we must restrict our focus to only the 51% of the population
that is male. The probability of being colorblind, given that the person is male, is 4%
of the 51%, or
P(A ʝ B)
.04
P(A͉B) ϭ ᎏᎏ ϭ ᎏᎏ ϭ .078
P(B)
.51
What is the probability of being colorblind, given that the person is female? Now we
are restricted to only the 49% of the population that is female, and
P(A ʝ BC)
.002
P(A͉BC) ϭ ᎏᎏ
ϭ ᎏᎏ ϭ .004
P(BC)
.49
Notice that the probability of event A changed, depending on whether event B occured.
This indicates that these two events are dependent.
When two events are independent—that is, if the probability of event B is the same,
whether or not event A has occurred, then event A does not affect event B and
P(B͉A) ϭ P(B)
The Multiplication Rule can now be simpliﬁed.
THE MULTIPLICATION RULE FOR INDEPENDENT
EVENTS
If two events A and B are independent, the probability that both A and B occur is
P(A ʝ B) ϭ P(A)P(B)
Similarly, if A, B, and C are mutually independent events (all pairs of events are
independent), then the probability that A, B, and C all occur is
P(A ʝ B ʝ C) ϭ P(A)P(B)P(C)
152
❍
CHAPTER 4 PROBABILITY AND PROBABILITY DISTRIBUTIONS
Coin Tosses at Football Games A football team is involved in two overtime periods during a given game, so that there are three coins tosses. If the coin is
fair, what is the probability that they lose all three tosses?
Solution
If the coin is fair, the event can be described in three steps:
A: lose the ﬁrst toss
B: lose the second toss
C: lose the third toss
Since the tosses are independent, and since P(win) ϭ P(lose) ϭ .5 for any of the three
tosses,
P(A ʝ B ʝ C) ϭ P(A)P(B)P(C) ϭ (.5)(.5)(.5) ϭ .125
How can you check to see if two events are independent or dependent? The easiest
solution is to redeﬁne the concept of independence in a more formal way.
CHECKING FOR INDEPENDENCE
Two events A and B are said to be independent if and only if either
P(A ʝ B) ϭ P(A)P(B)
or
P(B͉A) ϭ P(B)
Otherwise, the events are said to be dependent.
EXAMPLE
4.20
Toss two coins and observe the outcome. Deﬁne these events:
A: Head on the ﬁrst coin
B: Tail on the second coin
Are events A and B independent?
From previous examples, you know that S ϭ {HH, HT, TH, TT}. Use
these four simple events to ﬁnd
Solution
Remember,
independence ⇔
P(A ʝ B) ϭ P (A)P(B).
1
1
1
P(A) ϭ ᎏᎏ, P(B) ϭ ᎏᎏ, and P(A ʝ B) ϭ ᎏᎏ.
2
2
4
1 1
1
1
Since P(A)P(B) ϭ ᎏᎏ ᎏᎏ ϭ ᎏᎏ and P(A ʝ B) ϭ ᎏᎏ, we have P(A)P(B) ϭ P(A ʝ B)
2 2
4
4
and the two events must be independent.
EXAMPLE
4.21
Refer to the probability table in Example 4.18, which is reproduced below.
Child in College (D)
No Child in College (E )
Too High
(A)
Right Amount
(B)
Too Little
(C )
.35
.25
.08
.20
.01
.11
Are events D and A independent? Explain.
4.6 INDEPENDENCE, CONDITIONAL PROBABILITY, AND THE MULTIPLICATION RULE
❍
153
Solution
1. Use the probability table to ﬁnd P(A ʝ D) ϭ .35, P(A) ϭ .60, and
P(D) ϭ .44. Then
P(A)P(D) ϭ (.60)(.44) ϭ .264 and P(A ʝ D) ϭ .35
Since these two probabilities are not the same, events A and D are dependent.
2. Alternately, calculate
P(A ʝ D)
.35
P(A͉ D) ϭ ᎏᎏ ϭ ᎏᎏ ϭ .80
P(D)
.44
Since P(A͉ D) ϭ .80 and P(A) ϭ .60, we are again led to the conclusion that
events A and D are dependent.
What’s the Difference between Mutually Exclusive
and Independent Events?
Many students ﬁnd it hard to tell the difference between mutually exclusive and
independent events.
• When two events are mutually exclusive or disjoint, they cannot both happen when the experiment is performed. Once the event B has occurred,
event A cannot occur, so that P(A͉B) ϭ 0, or vice versa. The occurrence of
event B certainly affects the probability that event A can occur.
•
Therefore, mutually exclusive events must be dependent.
•
When two events are mutually exclusive or disjoint,
P(A ʝ B) ϭ 0 and P(A ʜ B) ϭ P(A) ϩ P(B).
•
When two events are independent,
P(A ʝ B) ϭ P(A)P(B), and P(A ʜ B) ϭ P(A) ϩ P(B) Ϫ P(A)P(B).
Exercise Reps
Use the relationships above to ﬁll in the blanks in the table below.
P(A)
P(B)
.3
.4
Conditions for Events A and B
Mutually exclusive
.3
.4
Independent
.1
.5
.2
.5
P(A 8 B)
P(A 9 B)
P(AͦB)
.6
.10
Answers are located on the perforated card at the back of this book.
Using probability rules to calculate the probability of an event requires some experience and ingenuity. You need to express the event of interest as a union or intersection (or the combination of both) of two or more events whose probabilities are
known or easily calculated. Often you can do this in different ways; the key is to ﬁnd
the right combination.
154
❍
EXAMPLE
CHAPTER 4 PROBABILITY AND PROBABILITY DISTRIBUTIONS
4.22
Two cards are drawn from a deck of 52 cards. Calculate the probability that the draw
includes an ace and a ten.
Solution
Consider the event of interest:
A: Draw an ace and a ten
Then A ϭ B ʜ C, where
B: Draw the ace on the ﬁrst draw and the ten on the second
C: Draw the ten on the ﬁrst draw and the ace on the second
Events B and C were chosen to be mutually exclusive and also to be intersections of
events with known probabilities; that is,
B ϭ B1 ʝ B2 and C ϭ C1 ʝ C2
where
B1: Draw an ace on the ﬁrst draw
B2: Draw a ten on the second draw
C1: Draw a ten on the ﬁrst draw
C2: Draw an ace on the second draw
Applying the Multiplication Rule, you get
P(B1 ʝ B2) ϭ P(B1)P(B2͉B1)
4
4
ϭ ᎏᎏ ᎏᎏ
52 51
and
4
4
P(C1 ʝ C2) ϭ ᎏᎏ ᎏᎏ
52 51
Then, applying the Addition Rule,
P(A) ϭ P(B) ϩ P(C)
4
4
4
4
8
ϭ ᎏᎏ ᎏᎏ ϩ ᎏᎏ ᎏᎏ ϭ ᎏᎏ
52 51
52 51
663
Check each composition carefully to be certain that it is actually equal to the event of
interest.
4.6
EXERCISES
EXERCISE REPS
These exercises refer to the MyPersonal Trainer section on page 153.
4.40 Use event relationships to ﬁll in the blanks in the table below.
P(A)
P(B)
.3
.4
Conditions for Events A and B
.3
.4
.1
.5
Mutually exclusive
.2
.5
Independent
P(A 8 B)
P(A 9 B)
.12
.7
P(AͦB)
4.6 INDEPENDENCE, CONDITIONAL PROBABILITY, AND THE MULTIPLICATION RULE
❍
155
4.41 Use event relationships to ﬁll in the blanks in the table below.
P(A)
P(B)
Conditions for Events A and B
.3
.4
Mutually exclusive
.3
.4
Independent
.1
.5
.2
.5
P(A 8 B)
P(A 9 B)
P(AͦB)
.1
0
BASIC TECHNIQUES
A: Observe a number less than 4
4.42 An experiment can result in one of ﬁve equally
B: Observe a number less than or equal to 2
C: Observe a number greater than 3
likely simple events, E1, E2, . . . , E5. Events A, B, and
C are deﬁned as follows:
A: E1, E3
P(A) ϭ .4
B: E1, E2, E4, E5
C: E3, E4
P(B) ϭ .8
P(C) ϭ .4
Find the probabilities associated with these compound
events by listing the simple events in each.
b. A ʝ B
c. B ʝ C
a. Ac
d. A ʜ B
e. B͉C
f. A͉B
g. A ʜ B ʜ C
h. (A ʝ B)c
4.43 Refer to Exercise 4.42. Use the deﬁnition of a
complementary event to ﬁnd these probabilities:
a. P(Ac )
b. P((A ʝ B)c )
Do the results agree with those obtained in Exercise 4.42?
4.44 Refer to Exercise 4.42. Use the deﬁnition of
conditional probability to ﬁnd these probabilities:
a. P(A͉B)
b. P(B͉C)
Do the results agree with those obtained in Exercise 4.42?
4.45 Refer to Exercise 4.42. Use the Addition and
Multiplication Rules to ﬁnd these probabilities:
a. P(A ʜ B)
b. P(A ʝ B)
c. P(B ʝ C)
Do the results agree with those obtained in Exercise 4.42?
Find the probabilities associated with the events below
using either the simple event approach or the rules and
deﬁnitions from this section.
a. S
b. A͉B
c. B
d. A ʝ B ʝ C e. A ʝ B
f. A ʝ C
g. B ʝ C
h. A ʜ C
i. B ʜ C
4.48 Refer to Exercise 4.47.
a. Are events A and B independent? Mutually exclusive?
b. Are events A and C independent? Mutually exclusive?
4.49 Suppose that P(A) ϭ .4 and P(B) ϭ .2. If events
A and B are independent, ﬁnd these probabilities:
a. P(A ʝ B)
b. P(A ʜ B)
4.50 Suppose that P(A) ϭ .3 and P(B) ϭ .5. If
events A and B are mutually exclusive, ﬁnd these
probabilities:
a. P(A ʝ B)
b. P(A ʜ B)
4.51 Suppose that P(A) ϭ .4 and P(A ʝ B) ϭ .12.
a. Find P(B͉A).
b. Are events A and B mutually exclusive?
c. If P(B) ϭ .3, are events A and B independent?
4.52 An experiment can result in one or both of
events A and B with the probabilities shown in this
probability table:
4.46 Refer to Exercise 4.42.
a. Are events A and B independent?
b. Are events A and B mutually exclusive?
4.47 Dice An experiment consists of tossing a single
die and observing the number of dots that show on the
upper face. Events A, B, and C are deﬁned as follows:
B
Bc
A
Ac
.34
.15
.46
.05
Find the following probabilities:
a. P(A)
d. P(A ʜ B)
b. P(B)
e. P(A͉B)
c. P(A ʝ B)
f. P(B͉A)
156
❍
CHAPTER 4 PROBABILITY AND PROBABILITY DISTRIBUTIONS
4.53 Refer to Exercise 4.52.
a. Are events A and B mutually exclusive? Explain.
b. Are events A and B independent? Explain.
A: The offender has 10 or more years of education
B: The offender is convicted within 2 years after
completion of treatment
APPLICATIONS
4.54 Drug Testing Many companies are testing
prospective employees for drug use, with the intent of
improving efficiency and reducing absenteeism, accidents, and theft. Opponents claim that this procedure is
creating a class of unhirables and that some persons
may be placed in this class because the tests themselves are not 100% reliable. Suppose a company uses
a test that is 98% accurate—that is, it correctly identiﬁes a person as a drug user or nonuser with probability
.98—and to reduce the chance of error, each job applicant is required to take two tests. If the outcomes of
the two tests on the same person are independent
events, what are the probabilities of these events?
a. A nonuser fails both tests.
b. A drug user is detected (i.e., he or she fails at least
one test).
c. A drug user passes both tests.
4.55 Grant Funding Whether a grant proposal is
funded quite often depends on the reviewers. Suppose
a group of research proposals was evaluated by a
group of experts as to whether the proposals were worthy of funding. When these same proposals were submitted to a second independent group of experts, the
decision to fund was reversed in 30% of the cases. If
the probability that a proposal is judged worthy of
funding by the ﬁrst peer review group is .2, what are
the probabilities of these events?
a. A worthy proposal is approved by both groups.
b. A worthy proposal is disapproved by both groups.
c. A worthy proposal is approved by one group.
4.56 Drug Offenders A study of the behavior of a
large number of drug offenders after treatment for drug
abuse suggests that the likelihood of conviction within
a 2-year period after treatment may depend on the
offender’s education. The proportions of the total
number of cases that fall into four education/conviction
categories are shown in the table below.
Status Within 2
Years After Treatment
Convicted
Not
Convicted
Totals
10 Years or More
9 Years or Less
.10
.27
.30
.33
.40
.60
Totals
.37
.63
1.00
Education
Suppose a single offender is selected from the treatment program. Here are the events of interest:
Find the appropriate probabilities for these events:
a. A
b. B
c. A ʝ B
d. A ʜ B
e. Ac
f. (A ʜ B)c
g. (A ʝ B)c
h. A given that B has occurred
i. B given that A has occurred
4.57 Use the probabilities of Exercise 4.56 to show
that these equalities are true:
a. P(A ʝ B) ϭ P(A)P(B͉A)
b. P(A ʝ B) ϭ P(B)P(A͉B)
c. P(A ʜ B) ϭ P(A) ϩ P(B) Ϫ P(A ʝ B)
4.58 The Birthday Problem Two people enter a
room and their birthdays (ignoring years) are recorded.
a. Identify the nature of the simple events in S.
b. What is the probability that the two people have a
speciﬁc pair of birthdates?
c. Identify the simple events in event A: Both people
have the same birthday.
d. Find P(A).
e. Find P(Ac).
4.59 The Birthday Problem, continued If n peo-
ple enter a room, ﬁnd these probabilities:
A: None of the people have the same birthday
B: At least two of the people have the same birthday
Solve for
a. n ϭ 3
b. n ϭ 4
[NOTE: Surprisingly, P(B) increases rapidly as n
increases. For example, for n ϭ 20, P(B) ϭ .411;
for n ϭ 40, P(B) ϭ .891.]
4.60 Starbucks or Peet’s®? A college student fre-
quents one of two coffee houses on campus, choosing
Starbucks 70% of the time and Peet’s 30% of the time.
Regardless of where she goes, she buys a cafe mocha
on 60% of her visits.
a. The next time she goes into a coffee house on campus, what is the probability that she goes to Starbucks and orders a cafe mocha?
b. Are the two events in part a independent? Explain.
c. If she goes into a coffee house and orders a cafe
mocha, what is the probability that she is at Peet’s?
4.6 INDEPENDENCE, CONDITIONAL PROBABILITY, AND THE MULTIPLICATION RULE
d. What is the probability that she goes to Starbucks
or orders a cafe mocha or both?
4.61 Inspection Lines A certain manufactured item
is visually inspected by two different inspectors. When
a defective item comes through the line, the probability that it gets by the ﬁrst inspector is .1. Of those that
get past the ﬁrst inspector, the second inspector will
“miss” 5 out of 10. What fraction of the defective
items get by both inspectors?
4.62 Smoking and Cancer A survey of people in a
given region showed that 20% were smokers. The
probability of death due to lung cancer, given that a
person smoked, was roughly 10 times the probability
of death due to lung cancer, given that a person did not
smoke. If the probability of death due to lung cancer in
the region is .006, what is the probability of death due
to lung cancer given that a person is a smoker?
4.63 Smoke Detectors A smoke-detector system
uses two devices, A and B. If smoke is present, the
probability that it will be detected by device A is .95;
by device B, .98; and by both devices, .94.
a. If smoke is present, ﬁnd the probability that the
smoke will be detected by device A or device B or
both devices.
b. Find the probability that the smoke will not be
detected.
4.64 Plant Genetics Gregor Mendel was a monk
who suggested in 1865 a theory of inheritance based
on the science of genetics. He identiﬁed heterozygous
individuals for ﬂower color that had two alleles (one
r ϭ recessive white color allele and one R ϭ dominant
red color allele). When these individuals were mated,
3/4 of the offspring were observed to have red ﬂowers
and 1/4 had white ﬂowers. The table summarizes this
mating; each parent gives one of its alleles to form the
gene of the offspring.
Parent 2
Parent 1
r
R
r
R
rr
Rr
rR
RR
We assume that each parent is equally likely to give
either of the two alleles and that, if either one or two
of the alleles in a pair is dominant (R), the offspring
will have red ﬂowers.
a. What is the probability that an offspring in this mating has at least one dominant allele?
b. What is the probability that an offspring has at least
one recessive allele?
❍
157
c. What is the probability that an offspring has one
recessive allele, given that the offspring has red
ﬂowers?
4.65 Soccer Injuries During the inaugural season of
Major League Soccer in the United States, the medical
teams documented 256 injuries that caused a loss of
participation time to the player. The results of this
investigation, reported in The American Journal of
Sports Medicine, is shown in the table.3
Severity
Minor (A)
Moderate (B)
Major (C)
Total
Practice (P)
Game (G)
Total
66
23
12
88
44
23
154
67
35
101
155
256
If one individual is drawn at random from this group
of 256 soccer players, ﬁnd the following probabilities:
a. P(A)
b. P(G)
c. P(A ʝ G)
d. P(G͉A)
e. P(G͉B)
f. P(G͉C)
g. P(C͉P)
h. P(Bc )
4.66 Choosing a Mate Men and women often disagree on how they think about selecting a mate. Suppose that a poll of 1000 individuals in their twenties
gave the following responses to the question of
whether it is more important for their future mate to be
able to communicate their feelings (F ) than it is for
that person to make a good living (G).
Feelings (F )
Good Living (G)
Totals
Men (M)
Women (W )
.35
.36
.20
.09
.55
.45
Totals
.71
.29
1.00
If an individual is selected at random from this
group of 1000 individuals, calculate the following
probabilities:
a. P(F)
d. P(F͉W)
b. P(G)
e. P(M͉F)
c. P(F͉M)
f. P(W͉G)
4.67 Jason and Shaq The two stars of the Miami
Heat professional basketball team are very different
when it comes to making free throws. ESPN.com
reports that Jason Williams makes about 80% of his
free throws, while Shaquille O’Neal makes only 53%
of his free throws.4 Assume that the free throws are
independent, and that each player takes two free
throws during a particular game.
a. What is the probability that Jason makes both of
his free throws?
158 ❍
CHAPTER 4 PROBABILITY AND PROBABILITY DISTRIBUTIONS
b. What is the probability that Shaq makes exactly
Player A has probability 1/6 of winning the tournament
if player B enters and probability 3/4 of winning if
player B does not enter the tournament. If the probability that player B enters is 1/3, ﬁnd the probability that
player A wins the tournament.
one of his two free throws?
c. What is the probability that Shaq makes both of
his free throws and Jason makes neither of his?
4.68 Golﬁng Player A has entered a golf tournament
but it is not certain whether player B will enter.
BAYES’ RULE (OPTIONAL)
4.7
Let us reconsider the experiment involving colorblindness from
Section 4.6. Notice that the two events
Colorblindness
B: the person selected is a man
BC: the person selected is a woman
taken together make up the sample space S, consisting of both men and women. Since
colorblind people can be either male or female, the event A, which is that a person is
colorblind, consists of both those simple events that are in A and B and those simple
events that are in A and BC. Since these two intersections are mutually exclusive, you
can write the event A as
A ϭ (A ʝ B) ʜ (A ʝ BC)
and
P(A) ϭ P(A ʝ B) ϩ P(A ʝ BC)
ϭ .04 ϩ .002 ϭ .042
Suppose now that the sample space can be partitioned into k subpopulations,
S1, S2, S3, . . . , Sk, that, as in the colorblindness example, are mutually exclusive and
exhaustive; that is, taken together they make up the entire sample space. In a similar
way, you can express an event A as
A ϭ (A ʝ S1) ʜ (A ʝ S2 ) ʜ (A ʝ S3 )ʜ и и и ʜ (A ʝ Sk )
Then
P(A) ϭ P(A ʝ S1) ϩ P(A ʝ S2 ) ϩ P(A ʝ S3 ) ϩ и и и ϩ P(A ʝ Sk )
This is illustrated for k ϭ 3 in Figure 4.14.
F IGU R E 4 .1 4
Decomposition of event A
●
S
A∩S1
S1
A∩S2
S2
A∩S3
S3
4.7 BAYES’ RULE (OPTIONAL)
❍
159
You can go one step further and use the Multiplication Rule to write P(A ʝ Si) as
P(Si)P(A͉Si), for i ϭ 1, 2, . . . , k. The result is known as the Law of Total Probability.
LAW OF TOTAL PROBABILITY
Given a set of events S1, S2, S3, . . . , Sk that are mutually exclusive and exhaustive and an event A, the probability of the event A can be expressed as
P(A) ϭ P(S1)P(A͉S1) ϩ P(S2)P(A͉S2) ϩ P(S3)P(A͉S3) ϩ и и и ϩ P(Sk)P(A͉Sk)
EXAMPLE
TABLE 4.6
4.23
Sneakers are no longer just for the young. In fact, most adults own multiple pairs of
sneakers. Table 4.6 gives the fraction of U.S. adults 20 years of age and older who
own ﬁve or more pairs of wearable sneakers, along with the fraction of the U.S. adult
population 20 years or older in each of ﬁve age groups.5 Use the Law of Total Probability to determine the unconditional probability of an adult 20 years and older owning ﬁve or more pairs of wearable sneakers.
●
Probability Table
Groups and Ages
Fraction with Ն 5 Pairs
Fraction of U.S. Adults 20 and Older
G1
20–24
G2
25–34
G3
35–49
G4
50–64
G5
Ն 65
.26
.09
.20
.20
.13
.31
.18
.23
.14
.17
Let A be the event that a person chosen at random from the U.S. adult
population 18 years of age and older owns ﬁve or more pairs of wearable sneakers.
Let G1, G2, . . . , G5 represent the event that the person selected belongs to each of
the ﬁve age groups, respectively. Since the ﬁve groups are exhaustive, you can write
the event A as
Solution
A ϭ (A ʝ G1) ʜ (A ʝ G2) ʜ (A ʝ G3) ʜ (A ʝ G4) ʜ (A ʝ G5)
Using the Law of Total Probability, you can ﬁnd the probability of A as
P(A) ϭ P(A ʝ G1) ϩ P(A ʝ G2) ϩ P(A ʝ G3) ϩ P(A ʝ G4) ϩ P(A ʝ G5)
ϭ P(G1)P(A͉G1) ϩ P(G2)P(A͉G2) ϩ P(G3)P(A͉G3)
ϩ P(G4)P(A͉G4) ϩ P(G5)P(A͉G5)
From the probabilities in Table 4.6,
P(A) ϭ (.09)(.26) ϩ (.20)(.20) ϩ (.31)(.13) ϩ (.23)(.18) ϩ (.17)(.14)
ϭ .0234 ϩ .0400 ϩ .0403 ϩ .0414 ϩ .0238 ϭ .1689
The unconditional probability that a person selected at random from the population
of U.S. adults 20 years of age and older owns at least ﬁve pairs of wearable sneakers
is about .17. Notice that the Law of Total Probability is a weighted average of the
probabilities within each group, with weights .09, .20, .31, .23, and .17, which reﬂect
the relative sizes of the groups.
Often you need to ﬁnd the conditional probability of an event B, given that an
event A has occurred. One such situation occurs in screening tests, which used to be
160
❍
CHAPTER 4 PROBABILITY AND PROBABILITY DISTRIBUTIONS
associated primarily with medical diagnostic tests but are now ﬁnding applications in
a variety of ﬁelds. Automatic test equipment is routinely used to inspect parts in highvolume production processes. Steroid testing of athletes, home pregnancy tests,
and AIDS testing are some other applications. Screening tests are evaluated on the
probability of a false negative or a false positive, and both of these are conditional
probabilities.
A false positive is the event that the test is positive for a given condition, given
that the person does not have the condition. A false negative is the event that the test
is negative for a given condition, given that the person has the condition. You can
evaluate these conditional probabilities using a formula derived by the probabilist
Thomas Bayes.
The experiment involves selecting a sample from one of k subpopulations that
are mutually exclusive and exhaustive. Each of these subpopulations, denoted by
S1, S2, . . . , Sk, has a selection probability P(S1), P(S2), P(S3), . . . , P(Sk), called prior
probabilities. An event A is observed in the selection. What is the probability that the
sample came from subpopulation Si, given that A has occurred?
You know from Section 4.6 that P(Si |A) ϭ [P(A ʝ Si)]/P(A), which can be rewritten as P(Si |A) ϭ [P(Si)P(A|Si)]/P(A). Using the Law of Total Probability to rewrite
P(A), you have
P(Si)P(A|Si)
P(Si |A) ϭ ᎏᎏᎏᎏᎏᎏᎏ
P(S1)P(A|S1) ϩ P(S2)P(A|S2) ϩ P(S3)P(A|S3) ϩ и и и ϩ P(Sk)P(A|Sk)
These new probabilities are often referred to as posterior probabilities—that is, probabilities of the subpopulations (also called states of nature) that have been updated
after observing the sample information contained in the event A. Bayes suggested that
if the prior probabilities are unknown, they can be taken to be 1/k, which implies that
each of the events S1 through Sk is equally likely.
BAYES’ RULE
Let S1, S2, . . . , Sk represent k mutually exclusive and exhaustive subpopulations
with prior probabilities P(S1), P(S2), . . . , P(Sk). If an event A occurs, the posterior probability of Si given A is the conditional probability
P(Si)P(A|Si)
P(Si |A) ϭ ᎏᎏ
k
Α P(Sj)P(A|Sj)
jϭ1
for i ϭ 1, 2, . . . , k.
EXAMPLE
4.24
Refer to Example 4.23. Find the probability that the person selected was 65 years of
age or older, given that the person owned at least ﬁve pairs of wearable sneakers.
Solution
You need to ﬁnd the conditional probability given by
P(A ʝ G5)
P(G5|A) ϭ ᎏᎏ
P(A)
You have already calculated P(A) ϭ .1689 using the Law of Total Probability.
Therefore,