Bayesian Analysis 1

Bayesian Analysis
Certain types of marketing decisions represent one type of example in which Bayes’ theo-
rem can be applied. Consider a scenario in which an electronics manufacturer is considering
marketing a new model of televisions. In the past, 40% of the new-model televisions have
been successful, and 60% have been unsuccessful. Before introducing the new-model televi-
sion, the marketing research department conducts an extensive study and r­ eleases a r­ eport,
either ­favorable or unfavorable. In the past, 80% of the successful new-model television(s)
had ­received favorable ­market research reports, and 30% of the unsuccessful new-model
television(s) had received favorable reports. For the new model of television under consider-
ation, the marketing research department has issued a favorable report. What is the probabil-
ity that the television will be successful?
Bayes’ theorem is developed from the definition of conditional probability. To find the
conditional probability of B given A, consider Equation (4.4b):

P1A and B2 P1A  B2P1B2

P1B  A2 = =
P1A2 P1A2

Bayes’ theorem is derived by substituting Equation (4.8) on page 156 for P(A) in the d­ enominator of
Equation (4.4b).

Bayes’ Theorem
P(A  Bi)P(Bi)
P(Bi  A) = (4.9)
P(A  B1)P(B1) + P(A  B2)P(B2) + g + P(A  Bk)P(Bk)

where Bi is the ith event out of k mutually exclusive and collectively exhaustive events.

To use Equation (4.9) for the television-marketing example, let

event S = successful television event F = favorable report

event S′ = unsuccessful television event F′ = unfavorable report

and

P1S2 = 0.40 P1F  S2 = 0.80

P1S′2 = 0.60 P1F  S′2 = 0.30

Then, using Equation (4.9),

P1F  S2P1S2
P1S  F2 =
P1F  S2P1S2 + P1F  S′2P1S′2
10.80210.402
=
10.80210.402 + 10.30210.602
0.32 0.32
= =
0.32 + 0.18 0.50
= 0.64

The probability of a successful television, given that a favorable report was received, is
0.64. Thus, the probability of an unsuccessful television, given that a favorable report was
­received, is 1 - 0.64 = 0.36.
Table 4.4 summarizes the computation of the probabilities, and Figure 4.3 presents the
decision tree.
Copyright © 2017 Pearson Education, Inc.
2 chapter 4 | Basic Probability

Table 4.4
Bayes’ Theorem Prior Conditional Joint Revised
Computations for the Probability Probability Probability Probability
Television-Marketing Event Si P1 Si 2 P1 F ∣ Si 2 P1 F ∣ Si 2 P1 Si 2 P1 Si ∣ F2
Example S = s uccessful 0.40 0.80 0.32 P1S  F2 = 0.32>0.50
television = 0.64
S′ = u
 nsuccessful 0.60 0.30 0.18 P1S′  F2 = 0.18>0.50
television 0.50 = 0.36

Figure 4.3
Decision tree for marketing P(S and F) = P(F|S) P(S)
= (0.80) (0.40) = 0.32
a new television
P(S) = 0.40

P(S and F′) = P(F′|S) P(S)

= (0.20) (0.40) = 0.08

P(S′ and F) = P(F|S′) P(S′)

= (0.30) (0.60) = 0.18

P(S′) = 0.60

P(S′ and F′) = P(F′|S′) P(S′)

= (0.70) (0.60) = 0.42

Example 4.10 applies Bayes’ theorem to a medical diagnosis problem.

Example 4.10 The probability that a person has a certain disease is 0.03. Medical diagnostic tests are avail-
Using Bayes’ Theo- able to determine whether the person actually has the disease. If the disease is actually present,
rem in a Medical the probability that the medical diagnostic test will give a positive result (indicating that the
Diagnosis Problem disease is present) is 0.90. If the disease is not actually present, the probability of a positive test
result (indicating that the disease is present) is 0.02. Suppose that the medical diagnostic test
has given a positive result (indicating that the disease is present). What is the probability that
the disease is actually present? What is the probability of a positive test result?

Solution Let

event D = has disease event T = test is positive

event D′ = does not have disease event T′ = test is negative

and

P1D2 = 0.03 P1T  D2 = 0.90

P1D′2 = 0.97 P1T  D′2 = 0.02

Using Equation (4.9),

P1T  D2P1D2
P1D  T2 =
P1T  D2P1D2 + P1T  D′2P1D′2
10.90210.032
=
10.90210.032 + 10.02210.972
0.0270 0.0270
= =
(continued) 0.0270 + 0.0194 0.0464
= 0.582
Copyright © 2017 Pearson Education, Inc.
 Bayesian Analysis 3

The probability that the disease is actually present, given that a positive result has occurred
(indicating that the disease is present), is 0.582. Table 4.5 summarizes the computation of the
probabilities, and Figure 4.4 presents the decision tree. The denominator in Bayes’ theorem
represents P(T), the probability of a positive test result, which in this case is 0.0464, or 4.64%.
Table 4.5
Bayes’ Theorem Prior Conditional Joint Revised
Computations for the Probability Probability Probability Probability
Medical Diagnosis Event Di P1 Di 2 P1 T ∣ Di 2 P1 T ∣ Di 2 P1 Di 2 P1 Di ∣ T2
Problem D = has disease 0.03 0.90 0.0270 P1D  T2 = 0.0270>0.0464
= 0.582
D′ = d
 oes not 0.97 0.02 0.0194 P1D′  T 2 = 0.0194>0.0464
have disease 0.0464 = 0.418

Figure 4.4
P(D and T ) = P(T |D)P(D)
Decision tree for a medical = (0.90)(0.03) = 0.0270
diagnosis problem
P(D) = 0.03

P(D and T ′) = P(T ′|D)P(D)

= (0.10)(0.03) = 0.0030

P(D′ and T ) = P(T |D′)P(D′)

= (0.02)(0.97) = 0.0194
P(D′) = 0.97

P(D′ and T ′) = P(T ′|D′)P(D′)

= (0.98)(0.97) = 0.9506

Problems for Section 4.4

Learning the Basics their primary financial institution, 49% of banking customers have
4.41 If P1B2 = 0.05, P1A  B2 = 0.80, P1B′2 = 0.95, and moderate trust in their primary financial institution, and 7% have
P1A  B′2 = 0.40, find P1B  A2. minimal or no trust in their primary financial institution. Of the
banking customers that have complete trust in their primary finan-
4.42 If P1B2 = 0.30, P1A  B2 = 0.60, P1B′2 = 0.70, and cial institution, 68% are very likely to recommend their primary
and P1A  B′2 = 0.50, find P1B  A2. financial institution; of the banking customers that have moderate
trust in their primary financial institution, 20% are very likely to
Applying the Concepts recommend their primary financial institution; and of the banking
4.43 In Example 4.10, suppose that the probability that a medical customers that have minimal or no trust in their primary financial
diagnostic test will give a positive result if the disease is not pres- institution, 3% are very likely to recommend their primary finan-
ent is reduced from 0.02 to 0.01. cial institution. (Data extracted from “Global Consumer Banking
a. If the medical diagnostic test has given a positive result (indi- Survey-2014,” bit.ly/1gwJJuT.)
cating that the disease is present), what is the probability that a. Compute the probability that if the banking customer indicates
the disease is actually present? he or she is very likely to recommend his or her primary finan-
b. If the medical diagnostic test has given a negative result (indi- cial institution, the banking customer also has complete trust in
cating that the disease is not present), what is the probability his or her primary financial institution.
that the disease is not present? b. Compute the probability that the banking customer is very
likely to recommend his or her primary financial institution.
4.44 A banking executive is studying the role of trust in creat-
ing customer advocates, and how valuable trust is to the overall 4.45 Olive Construction Company is determining whether it
banking relationship. Based on study results, the executive has should submit a bid for a new shopping ­center. In the past, Olive’s
­determined that 44% of banking customers have complete trust in main competitor, Base C­ onstruction Company, has submitted bids

Copyright © 2017 Pearson Education, Inc.

4 chapter 4 | Basic Probability

70% of the time. If Base Construction Company does not bid on a are huge successes, 20% are modest successes, 40% break-even,
job, the probability that Olive Construction Company will get the and 30% are losers. However, before a publishing decision is made,
job is 0.50. If Base Construction Company bids on a job, the prob- the book will be reviewed. In the past, 99% of the huge successes
ability that Olive Construction Company will get the job is 0.25. ­received favorable reviews, 70% of the moderate successes r­ eceived
a. If Olive Construction Company gets the job, what is the prob- favorable reviews, 40% of the break-even books received favorable
ability that Base Construction Company did not bid? reviews, and 20% of the losers received favorable reviews.
b. What is the probability that Olive Construction Company will a. If the proposed textbook receives a favorable review, how
get the job? should the editor revise the probabilities of the various out-
comes to take this information into account?
4.46 According to a report by The Kauffman Foundation and
b. What proportion of textbooks receive favorable reviews?
LegalZoom, 35% of new businesses in a recent year were started
by women and 65% of new businesses were started by men. 4.48 A municipal bond service has three rating categories (A, B,
Twenty percent of new businesses started by women had revenues and C). Suppose that in the past year, of the municipal bonds issued
of $50,000 and above, whereas 24% of new businesses started by throughout the United States, 70% were rated A, 20% were rated
men had revenues of $50,000 and above. B, and 10% were rated C. Of the municipal bonds rated A, 50%
a. If a new business is selected at random and the new business were issued by cities, 40% by suburbs, and 10% by rural areas. Of
had revenues of $50,000 and above, what is the probability that the municipal bonds rated B, 60% were issued by cities, 20% by
the new business was started by a woman? suburbs, and 20% by rural areas. Of the municipal bonds rated C,
b. If a new business is selected at random, what is the probability 90% were issued by cities, 5% by suburbs, and 5% by rural areas.
that the new business had revenues of $50,000 and above? a. If a new municipal bond is to be issued by a city, what is the
probability that it will receive an A rating?
4.47 The editor of a textbook publishing company is trying to
b. What proportion of municipal bonds are issued by cities?
d­ ecide whether to publish a proposed business statistics textbook.
c. What proportion of municipal bonds are issued by suburbs?
Information on previous textbooks published indicates that 10%

BAYESIAN ANALYSIS
Key Technique Use Excel arithmetic formulas.
Example Apply Bayes’ theorem to the television market-
ing example.
Workbook Use the COMPUTE worksheet of the Bayes
workbook as a template.
The worksheet (shown below) already contains the prob-
abilities for the Section 4.3 example. For other problems,
change those probabilities in the cell range B5:C6.
Open to the COMPUTE_FORMULAS worksheet to
examine the arithmetic formulas that compute the probabili-
ties, which are also shown as an inset to the worksheet.

Copyright © 2017 Pearson Education, Inc.