Chapter 7

Random Variables
Working with Uncertain Numbers

Chapter Outline
7.1 Discrete Random Variables 156 7.4 The Normal Approximation to the Binomial 173
Finding the Mean and Standard Deviation 156 7.5 Two Other Distributions: The Poisson and the Exponential 176
7.2 The Binomial Distribution 158 The Poisson Distribution 176
Definition of Binomial Distribution and Proportion 159 The Exponential Distribution 177
Finding the Mean and Standard Deviation the Easy Way 161 7.6 End-of-Chapter Materials 178
Finding the Probabilities 162 Summary 178
7.3 The Normal Distribution 164 Key Words 180
Visualize Probabilities as the Area under the Curve 165 Questions 180
The Standard Normal Distribution Z and Its Probabilities 166 Problems 180
Solving Word Problems for Normal Probabilities 166 Database Exercises 185
The Four Different Probability Calculations 172 Projects 185
Be Careful: Things Need Not Be Normal! 173 Case 186

Many business situations involve random variables, such as Three: The number of qualified people who will
waiting to find out your investment portfolio performance or respond to your “help wanted” advertisement for a
asking customers in a marketing survey how much they new full-time employee.
would spend. Whenever a random experiment produces a num- Four: The price per barrel of oil next year.
ber (or several numbers) as part of its outcome, you can be sure Five: The reported income of the next family to respond
that random variables are involved. Naturally, you will want to to your information poll.
be able to compute and interpret summary measures (such as
A random variable may be defined as a specification
typical value and risk) as well as probabilities of events that
or description of a numerical result from a random experi-
depend on the observed random quantity—for example, the
ment. The value itself is called an observation. For exam-
probability that your portfolio grows by 10% or more.
ple, “next quarter’s sales” is a random variable because it
You can also think about random variables as being where
specifies and describes the number that will be produced
data sets come from. That is, many of the data sets you worked
by the random experiment of waiting until next quarter’s
with in Chapters 2–5 were obtained as observations of ran-
numbers are in and computing the sales. The actual future
dom variables. In this sense, the random variable itself repre-
value, $3,955,846, is an observation of this random vari-
sents the population (or the process of sampling from the
able. Note the distinction between a random variable
population), whereas the observed values of the random vari-
(which refers to the random process involved) and an obser-
able represent the sample data. Much more on population and
vation (which is a fixed number, once it has been observed).
samples is coming in Chapter 8 and beyond, but the funda-
The pattern of probabilities for a random variable is called
mentals of random numbers are covered here in this chapter.
its probability distribution.
Here are some examples of random variables. Note that
Many random variables have a mean and a standard
each one is random until its value is observed:
deviation. 1 In addition, there is a probability for each
One: Next quarter’s sales—a number that is currently event based on a random variable. We will consider two
unknown and that can take on one of a number of
different values. 1. All of the random variables considered in this chapter have a mean and
Two: The number of defective machines produced next a standard deviation, although in theory there do exist random variables
week. that have neither a mean nor a standard deviation.

Practical Business Statistics, Sixth Edition. DOI: 10.1016/B978-0-12-385208-3.00007-9 155

© 2012 Andrew F. Siegel. Published by Elsevier Inc. All rights reserved.
156 PART | II Probability

types of random variables: discrete and continuous. It is

easier to work with a discrete random variable because Example
you can make a list of all of its possible values. You Profit under Various Economic Scenarios
will learn about two particular distributions that are espe- During a brainstorming session devoted to evaluation of your
cially useful: the binomial distribution (which is discrete) firm’s future prospects, there was a general discussion of
and the normal distribution (which is continuous). what might happen in the future. It was agreed to simplify
the situation by considering a best-case scenario, a worst-
Furthermore, in many cases, you may use a (much sim-
case scenario, and two intermediate possibilities. For each of
pler) normal probability calculation as a close approxima-
these four scenarios, after considerable discussion, there was
tion to a binomial probability. general agreement on the approximate profit that might occur
Since there are so many different types of situations in and its likelihood. Note that this defines the probability distribu-
which data values can arise, there are many types of random tion for the random variable “profits” because we have a list of
variables. The exponential and the Poisson distributions values and probabilities: one column shows the values (in this
provide a look at the tip of this iceberg. case, profit) and another column shows the probabilities.
A random variable is discrete if you can list all possible
Economic Scenario Profit ($ millions) Probability
values it can take on when it is observed. A random variable
is continuous if it can take on any value in a range (for Great 10 0.20
example, any positive number). For some random variables, Good 5 0.40
OK 1 0.25
it is unclear whether they are discrete or continuous. For
Lousy –4 0.15
example, next quarter’s sales might be $385,298.61, or
$385,298.62, or $385,298.63, or a similar amount up to This probability distribution can be easily used to find
some very large number such as $4,000,000.00. Literally probabilities of all events concerning profit. The probability
speaking, this is discrete (since you can list all the possible that the profit is $10 million, for example, is 0.20. The prob-
outcomes). However, from a practical viewpoint, since the ability of making $3 million or more is found as follows:
dividing points are so close together and no single outcome 0.20 + 0.40 = 0.60—because there are two outcomes
is very likely, you may work with it as if it were continuous. (“Great” and “Good”) that correspond to this event by having
profit of $3 million or more.

7.1 DISCRETE RANDOM VARIABLES

When you have the list of values and probabilities (which Finding the Mean and Standard Deviation
defines the probability distribution) for a discrete random vari- The mean or expected value of a discrete random variable
able, you know everything possible about the process that will is an exact number that summarizes it in terms of a typical
produce a random and uncertain number. Using this list, you value, in much the same way that the average summarizes a
will be able to calculate any summary measure (e.g., of typi- list of data.2 The mean is denoted by the lowercase Greek
cal value or of risk) or probability (of any event determined letter μ (mu) or by E(X) (read as “expected value of X ”)
by the observed value) that might be of interest to you. for a random variable X. The formula is
Here are some examples of discrete random variables:
1. The number of defective machines produced next week. Mean or Expected Value of a Discrete Random
The list of possible values is 0, 1, 2, …. Variable X
2. The number of qualified people who will respond to your
μ = EðXÞ = Sum of ðvalue times probabilityÞ
“help wanted” advertisement for a new full-time
employee. Again, the list of possible values is 0, 1, 2, …. = ∑XPðXÞ
3. The resulting budget when a project is selected from
four possibilities with costs of $26,000, $43,000,
If the probabilities were all equal, this would be the
$54,000, and $83,000. The list of possible values is
average of the values. In general, the mean of a random
(in thousands of dollars) 26, 43, 54, and 83.
variable is a weighted average of the values using the prob-
Such a list of possible values, together with the prob- abilities as weights.
ability of each happening, is the probability distribution This mean profit in the preceding example is
of the discrete random variable. These probabilities must Expected profit = ð10 × 0:20Þ + ð5 × 0:40Þ + ð1 × 0:25Þ
be positive numbers (or 0) and must add up to 1. From
this distribution, you can find the mean and standard devia- + ð−4 × 0:15Þ = 3:65
tion of the random variable. You can also find the probabil-
ity of any event, simply by adding up the probabilities in 2. In fact, the mean of a random variable is also called its average; however,
the table that correspond to the event. we will often use mean for random variables and average for data.
Chapter | 7 Random Variables 157

Thus, the expected profit is $3.65 million. This number the Defined names section of the Formulas Ribbon. The
summarizes the various possible outcomes (10, 5, 1, –4) mean (3.65) is the sum of the products of value times
using a single number that reflects their likelihoods. probability; hence, the formula is “=SUMPRODUCT
The standard deviation of a discrete random variable (Profit,Probability).” Give this cell (which now contains
indicates approximately how far you expect it to be from the mean) the name “Mean.” The standard deviation
its mean. In many business situations, the standard deviation (4.40) is the square root (SQRT) of the sum of the pro-
indicates the risk by showing just how uncertain the situation ducts of the square of value minus mean times probability.
is. The standard deviation is denoted by σ, which matches our Hence, the formula is
use of σ as the population standard deviation. The formula is
= SQRTðSUMPRODUCTððProfit − MeanÞ^2,

Standard Deviation of a Discrete Random Variable X ProbabilityÞÞ

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi These formulas give us 3.65 for the mean and 4.40 for
σ = Sum of ðsquared deviation times probabilityÞ the standard deviation, as before.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
= ∑ðX − μÞ2 PðXÞ

Note that you would not get the correct answer by sim- This
cell is
ply using the Σ key on your calculator to accumulate only named
“Mean”
the single column of values, since this would not make
proper use of the probabilities.
The standard deviation of profit for our example is

pffi
σ= f½ð10 − 3:65Þ2 0:20 + ½ð5 − 3:65Þ2 0:40 Figure 7.1.1 shows the probability distribution, with the
heights of the lines indicating the probability and the loca-
+ ½ð1 − 3:65Þ 0:25 + ½ð−4 − 3:65Þ 0:15g
2 2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi tion of the lines indicating the amount of profit in each case.
= 8:064500 + 0:729000 + 1:755625 + 8:778375 Also indicated is the expected value, $3.65 million, and the
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi standard deviation, $4.40 million.
= 19:3275 = 4:40

The standard deviation of $4.40 million shows that there Example

Evaluating Risk and Return
is considerable risk involved here. Profit might reasonably
be about $4.40 million above or below its mean value of Your job is to evaluate three different projects (X, Y, and Z) and
$3.65 million. Table 7.1.1 shows the details of the compu- make a recommendation to upper management. Each project
requires an investment of $12,000 and pays off next year.
tations involved in finding the standard deviation.
Project X pays $14,000 for sure. Project Y pays either
To use Excel ® to compute the mean and standard
$10,000 or $20,000 with probability 0.5 in each case.
deviation of a discrete random variable, you might pro- Project Z pays nothing with probability 0.98 and $1,000,000
ceed as follows. Using Excel’s menu commands, give with probability 0.02. A summary is shown in Table 7.1.2.
names to these columns by selecting the numbers with (Continued)
the titles, then choosing Excel’s Create from Selection in

TABLE 7.1.1 Finding the Standard Deviation for a Discrete Random Variable
Squared Deviation
Profit Probability Deviation from Mean Squared Deviation Times Probability
10 0.20 6.35 40.3225 8.064500

5 0.40 1.35 1.8225 0.729000

1 0.25 −2.65 7.0225 1.755625

−4 0.15 −7.65 58.5225 8.778375

Sum: 19.3275

Square root: 4.40

158 PART | II Probability

Probability
Good
These standard deviations confirm your suspicions.
0.40 Project Z is indeed the riskiest—far more so than either of
the others. Project X is the safest—a sure thing with no risk
0.30
at all. Project Y involves a risk of $5,000.
OK Which project should be chosen? This question cannot
Great be answered by statistical analysis alone. Although the
0.20 expected value and the standard deviation provide helpful
Lousy
summaries to guide you in choosing a project, they do not
0.10 finish the task. Generally, people prefer larger expected pay-
offs and lower risk. However, with the choices presented
here, to achieve a larger expected payoff, you must take a
–5 0 5 10 greater risk. The ultimate choice of project will involve
your (and your firm’s) “risk versus return” preference to deter-
Standard deviation: Standard deviation: mine whether or not the increased expected payoff justifies
$4.40 million $4.40 million the increased risk.3
Expected profit: $3.65 million What if you measure projects in terms of profit instead of
Profit (millions) payoff? Since each project involves an initial investment of
$12,000, you can convert from payoff to profit by subtracting
FIGURE 7.1.1 The probability distribution of future profits, with the $12,000 from each payoff value in the probability distribu-
mean (expected profits) and standard deviation (risk) indicated. tion table:
Profit = Payoff − $12,000
TABLE 7.1.2 Payoffs and Probabilities for Three Projects
Using the rules from Section 5.4, which apply to summa-
Project Payoff Probability ries of random variables as well as to data, subtract $12,000
from each mean value and leave the standard deviation
X 14,000 1.00
alone. Thus, without doing any detailed calculations, you
Y 10,000 0.50 come up with the following expected profits:

20,000 0.50 X: $2,000

Z 0 0.98 Y: $3,000
Z: $8,000
1,000,000 0.02
The standard deviations of profits are the same as for pay-
offs, namely:

Example—cont’d X: $0
The means are easily found: $14,000 for X, 10,000 × 0.50 + Y: $5,000
20,000 × 0.50 = $15,000 for Y, and 0 × 0.98 + 1,000,000 × Z: $140,000
0.02 = $20,000 for Z. We could write these as follows:
EðXÞ = μX = $14,000 3. In your finance courses, you may learn about another factor that is
EðYÞ = μY = $15,000 often used in valuing projects, namely, the correlation (if any) between
EðZÞ = μZ = $20,000 the random payoffs and the payoffs of a market portfolio. This helps mea-
sure the diversifiable and nondiversifiable risk of a project. Correlation
Based only on these expected values, it would appear that Z (a statistical measure of association) will be presented in Chapter 11.
The nondiversifiable component of risk is also known as systematic or sys-
is best and X is worst. However, these mean values don’t tell temic risk because it is part of the entire economic system and cannot be
the whole story. For example, although project Z has the high- diversified away.
est expected payoff, it also involves considerable risk: 98% of
the time there would be no payoff at all! The risks involved
here are summarized by the standard deviations:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 7.2 THE BINOMIAL DISTRIBUTION
σ X = ð14,000 − 14,000Þ2 × 1:00 = $0
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Percentages play a key role in business. When a percentage
σ Y = ð10,000 − 15,000Þ2 × 0:50 + ð20,000 − 15,000Þ2 × 0:50 is arrived at by counting the number of times something
= $5,000 happens out of the total number of possibilities, the number
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi of occurrences might follow a binomial distribution. If so,
σ Z = ð0 − 20,000Þ2 × 0:98 + ð1,000,000 − 20,000Þ2 × 0:02
there are a number of time-saving shortcuts available for
= $140,000 finding the expected value, standard deviation, and prob-
abilities of various events. Sometimes you will be interested
Chapter | 7 Random Variables 159

in the percentage; at other times the number of occurrences The binomial proportion p is also called a binomial frac-
will be more relevant. The binomial distribution can give tion. You may have recognized it as a relative frequency,
answers in either case. Here are some examples of random which was defined in Chapter 6.
variables that follow a binomial distribution:
1. The number of orders placed, out of the next three Example
telephone calls to your catalog order desk. How Many Orders Are Placed? The Hard Way to Compute
2. The number of defective products out of 10 items This example shows the hard way to analyze a binomial
produced. random variable. Although it is rarely necessary to draw
3. The number of people who said they would buy your the probability tree, since it is usually quite large, seeing
product, out of 200 interviewed. it once will help you understand what is really going on
with the binomial distribution. Furthermore, when the
4. The number of stocks that went up yesterday, out of all
shortcut computations are presented (the easy way) you
issues traded on major exchanges. will appreciate the time they save!
5. The number of female employees in a division of 75 Suppose you are interested in the next n = 3 telephone
people. calls to the catalog order desk, and you know from experi-
6. The number of Republican (or Democratic) votes cast in ence (or are willing to assume4) that π = 0.6, so that 60%
the next election. of calls will result in an order (the others are primarily calls
for information, or misdirected). What can we say about
the number of calls that will result in an order? Certainly,
Definition of Binomial Distribution this number will be either 0, 1, 2, or 3 calls. Since a call is
and Proportion more likely to result in an order than not, we should probably
expect the probability of getting three orders to be larger than
Focus attention on a particular event. Each time the random
the probability of getting none at all. But how can we find
experiment is run, either the event happens or it doesn’t.
these probabilities? The probability tree provides a complete
These two possible outcomes give us the bi in binomial. analysis, as shown in Figure 7.2.1a, indicating the result of
A random variable X, defined as the number of occurrences each of the three phone calls.
of a particular event out of n trials, has a binomial Note that the conditional probabilities along the branches
distribution if are always 0.60 and 0.40 (the individual probabilities for
each call) since we assume orders occur independently
1. For each of the n trials, the event always has the same
and do not influence each other. The number of orders is
probability π of happening. listed at the far right in Figure 7.2.1a; for example, the second
2. The trials are independent of one another. number from the top, 2, reports the fact that the first and sec-
The independence requirement rules out “peeking,” as in ond (but not the third) callers placed an order, resulting in
two orders placed. Note that there are three ways in which
the case of the distribution of people who order the special at
two orders could be placed. To construct the probability dis-
a restaurant. If some people order the special because they tribution of the number of orders placed, you could add up
see other customers obviously enjoying the rich, delicious the probabilities for the different ways that each number
combination of special aromatic ingredients, and say, could happen:
“WOW! I’ll have that too!” the number who order the special
would not follow a binomial distribution. Choices have to be Number of Percentage Probability
made independently in order to get a binomial distribution. Callers Who Who Ordered,
The binomial proportion p is the binomial random Ordered, X p = X/n
variable X expressed as a fraction of n: 0 0.0% 0.064
1 33.3% 0.288 (= 0.096 + 0.096 + 0.096)
2 66.7% 0.432 (= 0.144 + 0.144 + 0.144)
Binomial Proportion 3 100.0% 0.216

p= X Number of occurrences
n = Number of trials This probability distribution is displayed in Figure 7.2.1b.
Now that you have the probability distribution, you can
find all of the probabilities by adding the appropriate ones.
(Note that π is a fixed number, the probability of occur- For example, the probability of at least two orders is 0.432 +
rence, whereas p is a random quantity based on the data.) 0.216 = 0.648. You can also use the formulas for the mean
For example, if you interviewed n = 600 shoppers and and standard deviation from Section 7.1 to find the mean
found that X = 38 plan to buy your product, then the bino- value (1.80 orders) and the standard deviation (0.849 orders).
mial proportion would be However, this would be too much work! There is a much
quicker formula for finding the mean, standard deviation,
p = X = 38 = 0:063, or 6.3% (Continued)
n 600
160 PART | II Probability

Number
Caller 1 Caller 2 Caller 3 of callers
placed order? placed order? placed order? placing
orders

0.60 0.216 3
Yes
0.36 No
0.60 0.40
Yes 0.144 2

0.60 No
0.40 0.60 0.144 2
Yes
0.24 No
0.40
Yes 0.096 1

No
0.60 0.144 2
Yes
0.24 No
0.60 0.40
Yes 0.096 1

0.40 No
0.40 0.60 0.096 1
Yes
0.16 No
0.40
0.064 0

(a)

0.4
Binomial probability

0.3

0.2

0.1

0
0 1 2 3
Number of callers placing orders
(b)

FIGURE 7.2.1 a. The probability tree for three successive telephone calls, each of which either does or does not result in an order being placed. There are
eight combinations (the circles at the far right). In particular, there are three ways in which exactly two calls could result in an order: the second, third, and
fifth circles from the top, giving a probability of 3 × 0.144 = 0.432. b. The binomial probability distribution of the number of calls that result in an order
being placed.
Chapter | 7 Random Variables 161

Example—cont’d Mean and Standard Deviation for a Binomial Distribution

and probabilities. Although it was possible to compute Number of Proportion or
directly in this small example, you will not usually be so Occurrences, X Percentage, p = X/n
lucky. For example, had you considered 10 successive
calls instead of 3, there would have been 1,024 probabilities Mean EðXÞ = μX = nπ EðpÞ = μp = π
at the right of the probability tree instead of the 8 in qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi π ð1 − πÞ
Figure 7.2.1. Standard deviation σX = nπð1 − πÞ σp = n

4. The probability π is usually given in textbook problems involving a

binomial distribution. In real life, they arise just as other probabilities
do: from relative frequency, theoretical probability, or subjective
probability. For the “telephone orders” example, we have n = 3 and
p = 0.60. Using the formulas, the mean and standard devia-
tion are
Think of this example as a way of seeing the under-
lying situation and all combinations and then simplifying Number of Proportion or
to a probability distribution of the number of occur- Occurrences, X Percentage, p = X/n
rences. Conceptually, this is the right way to view the Mean EðXÞ = nπ EðXÞ = π
situation. Now let’s learn the easy way to compute the = 3 × 0:60 = 0:60 or 60%
answers.
= 1:80 calls
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi πð1 − πÞ
Finding the Mean and Standard Deviation Standard σ X = nπð1 − πÞ σp = n
deviation pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
the Easy Way = 3 × 0:60ð1 − 0:60Þ 0:60ð1 − 0:60Þ
=
= 0:849 calls 3
The mean number of occurrences in a binomial situation is
= 0:283 or 28:3%
EðXÞ = nπ, the number of possibilities times the probability
of occurrence. The mean proportion is Thus, we expect 1.80 of these 3 telephone calls to result in
  an order. Sometimes more (i.e., 2 or 3) and sometimes fewer
X
E = EðpÞ = π (i.e., 0 or 1) calls will result in an order. The extent of this
n
uncertainty is measured (as usual) by the standard deviation,
which is the same as the individual probability of 0.849 calls. Similarly, we expect 60% of these 3 calls to result
occurrence.5 in an order. The last number, 28.3%, representing the standard
This is what you would expect. For example, in a poll deviation of the percentage, is interpreted as percentage
of a sample of 200 voters, if each has a 58% chance of points rather than as a percentage of some number. That is,
being in favor of your candidate, on average, you would while the expected percentage is 60%, the actual observed
expect that percentage is typically about 28.3 percentage points above
  this value (at 60 + 28.3 = 88.3%) or below (at 60 – 28.3 =
X
E = EðpÞ = π = 0:58 31.7%). This is natural if you remember that a standard
n deviation is stated in the same units as the data, which are
or 58% of the sample will be in your favor. In terms percentage points in the case of p.
of the number of people, you would expect EðXÞ =
nπ = 200 × 0:58 = 116 people out of the 200 in the sample
Example
to be in your favor. Of course, the actually observed num-
Recalling Advertisements
ber and percentage will probably randomly differ from
Your company is negotiating with a marketing research firm
these expected values.
to provide information on how your advertisements are doing
There are formulas for the standard deviation of the
with the American consumer. Selected people are to come in
binomial number and percentage, summarized along with one day to watch TV programs and ads (for many products
the expected values in the following table: from many companies) and return the next day to answer
questions. In particular, you plan to measure the rate of
recall, which is the percentage of people who remember
your ad the day after seeing it.
5. You might have recognized X/n as the relative frequency of the event. Before you contract with the firm to do the work, you
The fact that E(X/n) is equal to π says that, on average, the relative are curious about how reliable and accurate the results
frequency of an event is equal to its probability. In Chapter 8 we will (Continued)
learn that this property says that p is an unbiased estimator of π.
162 PART | II Probability

and π = 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, and
Example—cont’d 0.95. Here is the exact formula:6
are likely to be. Your budget allows 50 people to be tested.
From your discussions with the research firm, it seems rea-
sonable initially to assume that 35% of people will recall Binomial Probability That X Equals a
the ad, although you really don’t know the exact propor-
!
tion. Based on the assumption that it really is 35%, how n
accurate will the results be? That is, about how far will PðX = aÞ = π a ð1−πÞn−a
a
the measured recall percentage be from the assumed
value π = 0.35 with n = 50 for a binomial distribution? = n! π a ð1−πÞn−a
The answer is a!ðn−aÞ!

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi = 1×2×3××n π a ð1−πÞn−a

πð1 − πÞ ð1×2×3××aÞ½1×2×3××ðn−aÞ
σp =
n
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0:35ð1 − 0:35Þ
=
50 By using this formula with each value of a from 0 to n
= 0:0675 or 6:75% (sometimes a lot of work), you (or a computer) can generate
the entire probability distribution. From these values, you
This says that the standard deviation of the result of the can find any probability you want involving X by adding
recall test (namely, the percentage of people tested who together the appropriate probabilities from this formula.
remembered the ad) is likely to differ from the true percen-
To see how to use the formula, suppose there are n = 5
tage for the entire population typically by about 7 percentage
possibilities with a success probability p = 0.8 for each one,
points in either direction (above or below).
You decide that the results need to be more precise than
and you want to find the probability of exactly a = 3 suc-
that. The way to improve the precision of the results is to cesses. The answer is
gather more information by increasing the sample size, n. !
Checking the budget and negotiating over the rates, you 5
find that n = 150 is a possibility. With this larger sample, PðX = 3Þ = 0:83 ð1 − 0:8Þ5−3
the standard deviation decreases to reflect the extra 3
information: 5!
= 0:83 × 0:22
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3!ð5 − 3Þ!
πð1 − πÞ
σp = 1×2×3×4×5
n = 0:512 × 0:040
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð1 × 2 × 3Þð1 × 2Þ
0:35ð1 − 0:35Þ
= = 10 × 0:02048 = 0:2048
150
= 0:0389 or 3:89%

You are disappointed that the extra cost didn’t bring a

greater improvement in the results. When the size of the 6. The notation n! is read as “n factorial” and is the product of the numbers
from 1 to n. For example, 4! = 1 × 2 × 3 × 4 = 24. (By convention, to get
study was tripled, the precision didn’t even double! This is
the correct answers, we define 0! to be 1.) Many calculators have a factorial
due, technically, to the fact that it is the square root of n,
key that works for values of n from 0 through 69. The notation
rather than n itself, that is involved. Nevertheless, you decide  
that the extra accuracy is worth the cost. n n!
=
a a!ðn − aÞ!
is the binomial coefficient, read aloud as “n choose a,” and also represents
the number of combinations you can make by choosing a items from n
Finding the Probabilities items (where the order of selection does not matter). Thus, it represents
the number of different ways in which you could assign exactly a
Suppose you have a binomial distribution, you know the occurrences to the n possibilities. For example, with n = 5 and a = 3,
values of n and π, and you want to know the probability the binomial coefficient is
that X will be exactly equal to some number a. There is a  
5 5!
formula for this probability that is useful for small to mod- = = 120 = 10
3 3!ð5 − 3Þ! 6×2
erate n. (When n is large, an approximation based on
the normal distribution, to be covered in Section 7.4, will Thus, there are 10 different ways (combinations) in which three out of five
people could buy our product: the first three people could, or the first two
be much easier than the exact method presented here.) In and the fourth might, and so forth. The full list of the 10 combinations is
addition, Table D–3 in Appendix D gives exact binomial (1,2,3), (1,2,4), (1,2,5), (1,3,4), (1,3,5), (1,4,5), (2,3,4), (2,3,5), (2,4,5), and
probabilities and cumulative probabilities for n = 1 to 20 (3,4,5).
Chapter | 7 Random Variables 163

This is the probability of exactly three successes. If

you want the probability of three or more successes, Example
you could compute the formula twice more: once for How Many Major Customers Will Call Tomorrow?
a = 4 and once for a = 5; the probability of three or How many of your n = 6 major customers will call tomor-
more successes would be the total of these numbers. row? You are willing to assume that each one has a probabil-
Alternatively, you could use a computer to obtain the ity π = 0.25 of calling and that they call independently of one
another. Thus, the number of major customers that will call
probabilities, for example:
tomorrow, X, follows a binomial distribution.
How many do you expect will call? That is, what
Probability Density Function and Cumulative is the expected value of X? The answer is E(X) = n × π =
Distribution Function 1.5 major customers. The standard deviation is σX =
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Binomial with n = 5 and p = 0.800000 6 × 0:25 × ð1 − 0:25Þ = 1.060660, indicating that you can
reasonably anticipate 1 or 2 more or less than the 1.5 you
expect. Although this gives you an idea of what to expect, it
a P(X = a) P(X <= a)
doesn’t tell you the chances that a given number will call.
0 0.0003 0.0003 Let’s compute the probabilities for this.
1 0.0064 0.0067 What is the probability that exactly a = 2 out of your n = 6
2 0.0512 0.0579 major customers will call? The answer is
3 0.2048 0.2627 !
4 0.4096 0.6723 6
5 0.3277 1.0000 PðX = 2Þ = 0:252 ð1 − 0:25Þð6−2Þ
2
= 15 × 0:0625 × 0:316406 = 0:297
In either case, once you have the individual probabilities
(for 3, 4, and 5 successes), the answer is Here is the entire probability distribution of the number of
major customers who will call you tomorrow, including all
PðX ≥ 3Þ = PðX = 3Þ + PðX = 4Þ + PðX = 5Þ possibilities for the number a from 0 through n = 6:
= 0:2048 + 0:4096 + 0:3277 Probability Density Function and Cumulative
= 0:9421 Distribution Function
Binomial with n = 6 and p = 0.250000
Thus, you have a 94.2% chance of achieving three or
a P(X = a) P(X <= a)
more successes out of these five. Alternatively, using the
complement rule, the probability of three or more must 0 0.1780 0.1780
1 0.3560 0.5339
be one minus the probability of two or less, which is listed
2 0.2966 0.8306
as 0.0579 in the computer output. The answer would then 3 0.1318 0.9624
be found as 1 – 0.0579 = 0.9421. 4 0.0330 0.9954
To use Excel® to compute binomial probabilities, use the 5 0.0044 0.9998
formula “=BINOMDIST(a,n,π, FALSE)” to find the prob- 6 0.0002 1.0000
ability P(X = a) of being equal to a, and use the formula
Note that the most likely outcomes are 1 or 2 calls, just as
“=BINOMDIST(a,n,π,TRUE)” to find the probability
you suspected based on the mean value of 1.5 calls.
P(X ≤ a) of being less than or equal to a, as follows:7
From this probability distribution, you can compute any
probability about the number of major customers who will
call you tomorrow. It is highly unlikely that all 6 will call
(0.0002 or 0.02%, much less than a 1% chance). The prob-
ability that 4 or more will call is 0.0330 + 0.0044 + 0.0002 =
0.0376. From the second column, you can see that the prob-
ability that 3 or fewer will call is 0.9624. Your chances of
spending a quiet day with no calls is 0.178. This probability
distribution is shown in Figure 7.2.2.

Example
How Many Logic Analyzers to Schedule for Manufacturing?
7. The “FALSE” and “TRUE” in Excel’s binomial distribution formula You pay close attention to quality in your production facil-
refer to whether or not the probability distribution is cumulative, i.e., ities, but the logic analyzers you make are so complex that
whether or not it accumulates probabilities for all of the previous (Continued)
(smaller) values of a as well.
164 PART | II Probability

0.4 0.7

0.6
0.3
0.5
Probability

Probability
0.2 0.4

0.3
0.1
0.2

0.0 0.1
0 1 2 3 4 5 6
Number of customers who call 0
10 11 12 13 14 15 16 17 18
FIGURE 7.2.2 The probability distribution of the number of major cus- Number of working logic analyzers (17 scheduled)
tomers who will call you tomorrow. These are binomial probabilities, with
each vertical bar found using the formula based on n = 6 and p = 0.25. The FIGURE 7.2.3 The probability distribution of the number of working
number a is found along the horizontal axis. logic analyzers produced if you plan to produce only 17. This is binomial,
with n = 17 and π = 0.97.

Example—cont’d 0.7
there are still some failures. In fact, based on past experience,
0.6
about 97% of the finished products are in good working
order. Today you will have to ship 17 of these machines. 0.5
The question is: How many should you schedule for produc- Probability
0.4
tion to be reasonably certain that 17 working logic analyzers
will be shipped? 0.3
It is reasonable to assume a binomial distribution for the 0.2
number of working machines produced, with n being the
number that you schedule and π being each one’s probability 0.1
(0.97) of working. Then you can compute the probability that 0
17 or more of the scheduled machines will work. 10 11 12 13 14 15 16 17 18
What happens if you schedule 17 machines, with no mar- Number of working logic analyzers (18 scheduled)
gin for error? You might think that the high (97%) rate would
help you, but, in fact, the probability that all 17 machines FIGURE 7.2.4 The probability distribution of the number of working
will work (using n = 17 and a = 17) is just 0.596: logic analyzers produced if you plan to produce 18. This is binomial,
with n = 18 and π = 0.97.
!
17
PðX = 17 working machinesÞ = 0:9717 0:030
17
= 1 × 0:595826 × 1 = 0:596 So if you schedule 18 for production, you have a 90%
chance of shipping 17 good machines. It looks likely, but
Thus, if you schedule the same number, 17, that you need you would still be taking a 10% chance of failure. This prob-
to ship, you will be taking a big chance! There is only a 59.6% ability distribution is shown in Figure 7.2.4.
chance that you will meet the order, and a 40.4% chance that Similar tedious calculations reveal that if you schedule 19
you will fail to ship the entire order in working condition. The machines for production, you have a 98.2% chance of ship-
probability distribution is shown in Figure 7.2.3. ping 17 good machines (9.2% + 32.9% + 56.1%). So, to be
It looks as though you’d better schedule more than 17. reasonably sure of success, you’d better schedule at least 19
What if you schedule n = 18 units for production? To find machines to get 17 good ones!
the probability that at least 17 working analyzers will be
shipped, you’ll need to find the probabilities for a = 17 and
a = 18 and add them up:
PðX ≥ 17Þ = PðX = 17Þ + PðX = 18Þ 7.3 THE NORMAL DISTRIBUTION
! !
18 18 You already know from Chapter 3 how to tell if a data set
= 0:9717 0:031 + 0:9718 0:030
17 18 is approximately normally distributed. Now it’s time to
learn how to compute probabilities for this familiar bell-
= 18 × 0:595826 × 0:03 + 1 × 0:577951 × 1
shaped distribution. One reason the normal distribution
= 0:322 + 0:578 = 0:900 is particularly useful is the fact that, given only a
mean and a standard deviation, you can compute any
Chapter | 7 Random Variables 165

0 10 20 30 40 50 60 70 80
σ σ
μ 5 5 10 10
20 40
(a) (b)

FIGURE 7.3.1 a. The normal distribution, with mean value μ and standard deviation σ. Note that the mean can be any number, and the standard deviation
can be any positive number. b. Two different normal distributions. The one on the left has a smaller mean value (20) and a smaller standard deviation (5)
than the other. The one on the right has mean 40 and standard deviation 10.

probability of interest (provided that the distribution

really is normal).
The normal distribution, a continuous distribution, is
represented by the familiar bell-shaped curve shown in
Figure 7.3.1a. Note that there is a normal distribution for
each combination of a mean value and a positive standard
deviation value.8 Just slide the curve to the right or left
until the peak is centered above the mean value; then stretch
it wider or narrower until the scale matches the standard
deviation. Two different normal distributions are shown
in Figure 7.3.1b.

Visualize Probabilities as the Area under the The shaded area gives the probability
Curve of being between here and here

The bell-shaped curve gives you a guide for visualizing the FIGURE 7.3.2 The probability that a normally distributed random vari-
able is between any two values is equal to the area under the normal curve
probabilities for a normal distribution. You are more likely between these two values. You are more likely to see values in regions
to see values occurring near the middle, where the curve is close to the mean.
high. At the edges, where the curve is lower, values are not
as likely to occur. Formally, it is the area under the curve
that gives you the probability of being within a region, as
illustrated in Figure 7.3.2.
Note that a shaded strip near the middle of the curve will
have a larger area than a strip of the same width located
nearer to the edge. Compare Figure 7.3.2 to Figure 7.3.3
to see this.

8. The formula for the normal probability distribution with mean μ and
standard deviation σ is
FIGURE 7.3.3 The probability of falling within a region that is farther
1 2
pffiffiffiffiffi e−½ðx−μÞ/σ /2 : from the middle of the curve. Since the normal curve is lower here, the
2π σ probability is smaller than that shown in Figure 7.3.2.
166 PART | II Probability

The Standard Normal Distribution Z and Probability 0.9162

Its Probabilities of being less than 1.38
is represented by
The standard normal distribution is a normal distribution the shaded area
with mean μ = 0 and standard deviation σ = 1. The letter Z being 91.62%
of the total area
is often used to denote a random variable that follows this
standard normal distribution. One way to compute probabil-
ities for a normal distribution is to use tables that give prob-
abilities for the standard one, since it would be impossible
to keep different tables for each combination of mean and
standard deviation. The standard normal distribution can
represent any normal distribution, provided you think in −4 −3 −2 −1 0 1 2 3 4
terms of the number of standard deviations above or
1.38
below the mean instead of the actual units (e.g., dollars) (z value)
of the situation. The standard normal distribution is Using the standard normal probability table
shown in Figure 7.3.4.
The standard normal probability table, shown in FIGURE 7.3.5 The probability that a standard normal random variable is
Table 7.3.1, gives the probability that a standard normal less than z = 1.38 is 0.9162, as found in the standard normal probability
table. This corresponds to the shaded region to the left of 1.38, which is
random variable Z is less than any given number z. For 91.62% of the total area under the curve.
example, the probability of being less than 1.38 is
0.9162, illustrated as an area in Figure 7.3.5. Doesn’t it
usually wrong. Last quarter’s sales were forecast as $18
look like about 90% of the area? To find this number
million but came in at $21.3 million. Sales for the next
(0.9162), look up the value z = 1.38 in the standard normal
quarter are forecast as $20 million, with a standard devia-
probability table. While you’re at it, look up 2.35 (to find
tion (based on previous experience) of $3 million. Assum-
0.9906), 0 (to find 0.5000), and –0.82 (to find 0.2061).
ing a normal distribution centered at the forecast value, find
What is the probability corresponding to the value z = 0.36?
the probability of a “really bad quarter,” which is defined as
sales lower than $15 million.
Solving Word Problems for Normal The beginning part sets the stage. The first numbers (18
Probabilities and 21.3) describe past events but play no further role here.
Instead, you should focus on the following facts:
A typical word problem involving a normal distribution is a
story involving some application to business that gives you ● There is a normal distribution involved here.
a value for the mean and one for the standard deviation. ● Its mean is μ = $20 million.
Then you are asked to find one or more probabilities of ● Its standard deviation is σ = $3 million.
interest. Here is an example of such a word problem: ● You are asked to find the probability that sales will be
The upper-management people at Simplified Technolo- lower than $15 million.
gies, Inc., have finally noticed that sales forecasts are The next step is to convert all of these numbers (except
for the mean and standard deviation) into standardized
numbers; this has to be done before you can look up the
answer in the standard normal probability table. A stan-
dardized number (often written as z) is the number of stan-
dard deviations above the mean (or below the mean, if the
standardized number is negative). This conversion is done
as follows:
Number − Mean
z = Standardized number =
Standard deviation
Number − μ
=
σ
−4 −3 −2 −1 0 1 2 3 4 In this example, the number $15 million is standardized
FIGURE 7.3.4 The standard normal distribution Z with mean value μ = 0 as follows:
and standard deviation σ = 1. The standard normal distribution may be used
15 − μ 15 − 20 = −1:67
to represent any normal distribution, provided you think in terms of the z= =
number of standard deviations above or below the mean. σ 3
 Chapter | 7
TABLE 7.3.1 Standard Normal Probability Table (See Figure 7.3.5)
z Value Probability z Value Probability z Value Probability z Value Probability z Value Probability z Value Probability
–2.00 0.0228 –1.00 0.1587 0.00 0.5000 0.00 0.5000 1.00 0.8413 2.00 0.9772

–2.01 –1.01 –0.01

Random Variables
0.0222 0.1562 0.4960 0.01 0.5040 1.01 0.8438 2.01 0.9778

–2.02 0.0217 –1.02 0.1539 –0.02 0.4920 0.02 0.5080 1.02 0.8461 2.02 0.9783

–2.03 0.0212 –1.03 0.1515 –0.03 0.4880 0.03 0.5120 1.03 0.8485 2.03 0.9788

–2.04 0.0207 –1.04 0.1492 –0.04 0.4840 0.04 0.5160 1.04 0.8508 2.04 0.9793

–2.05 0.0202 –1.05 0.1469 –0.05 0.4801 0.05 0.5199 1.05 0.8531 2.05 0.9798

–2.06 0.0197 –1.06 0.1446 –0.06 0.4761 0.06 0.5239 1.06 0.8554 2.06 0.9803

–2.07 0.0192 –1.07 0.1423 –0.07 0.4721 0.07 0.5279 1.07 0.8577 2.07 0.9808

–2.08 0.0188 –1.08 0.1401 –0.08 0.4681 0.08 0.5319 1.08 0.8599 2.08 0.9812

–2.09 0.0183 –1.09 0.1379 –0.09 0.4641 0.09 0.5359 1.09 0.8621 2.09 0.9817

–2.10 0.0179 –1.10 0.1357 –0.10 0.4602 0.10 0.5398 1.10 0.8643 2.10 0.9821

–2.11 0.0174 –1.11 0.1335 –0.11 0.4562 0.11 0.5438 1.11 0.8665 2.11 0.9826

–2.12 0.0170 –1.12 0.1314 –0.12 0.4522 0.12 0.5478 1.12 0.8686 2.12 0.9830

–2.13 0.0166 –1.13 0.1292 –0.13 0.4483 0.13 0.5517 1.13 0.8708 2.13 0.9834

–2.14 0.0162 –1.14 0.1271 –0.14 0.4443 0.14 0.5557 1.14 0.8729 2.14 0.9838

–2.15 0.0158 –1.15 0.1251 –0.15 0.4404 0.15 0.5596 1.15 0.8749 2.15 0.9842

–2.16 0.0154 –1.16 0.1230 –0.16 0.4364 0.16 0.5636 1.16 0.8770 2.16 0.9846

–2.17 0.0150 –1.17 0.1210 –0.17 0.4325 0.17 0.5675 1.17 0.8790 2.17 0.9850

–2.18 0.0146 –1.18 0.1190 –0.18 0.4286 0.18 0.5714 1.18 0.8810 2.18 0.9854

–2.19 0.0143 –1.19 0.1170 –0.19 0.4247 0.19 0.5753 1.19 0.8830 2.19 0.9857

–2.20 0.0139 –1.20 0.1151 –0.20 0.4207 0.20 0.5793 1.20 0.8849 2.20 0.9861

–2.21 0.0136 –1.21 0.1131 –0.21 0.4168 0.21 0.5832 1.21 0.8869 2.21 0.9864

–2.22 0.0132 –1.22 0.1112 –0.22 0.4129 0.22 0.5871 1.22 0.8888 2.22 0.9868

–2.23 0.0129 –1.23 0.1093 –0.23 0.4090 0.23 0.5910 1.23 0.8907 2.23 0.9871

–2.24 0.0125 –1.24 0.1075 –0.24 0.4052 0.24 0.5948 1.24 0.8925 2.24 0.9875

–2.25 0.0122 –1.25 0.1056 –0.25 0.4013 0.25 0.5987 1.25 0.8944 2.25 0.9878

167
 168
TABLE 7.3.1 Standard Normal Probability Table (See Figure 7.3.5)—cont’d
z Value Probability z Value Probability z Value Probability z Value Probability z Value Probability z Value Probability
–2.26 0.0119 –1.26 0.1038 –0.26 0.3974 0.26 0.6026 1.26 0.8962 2.26 0.9881

–2.27 0.0116 –1.27 0.1020 –0.27 0.3936 0.27 0.6064 1.27 0.8980 2.27 0.9884

–2.28 0.0113 –1.28 0.1003 –0.28 0.3897 0.28 0.6103 1.28 0.8997 2.28 0.9887

–2.29 0.0110 –1.29 0.0985 –0.29 0.3859 0.29 0.6141 1.29 0.9015 2.29 0.9890

–2.30 0.0107 –1.30 0.0968 –0.30 0.3821 0.30 0.6179 1.30 0.9032 2.30 0.9893

–2.31 0.0104 –1.31 0.0951 –0.31 0.3783 0.31 0.6217 1.31 0.9049 2.31 0.9896

–2.32 0.0102 –1.32 0.0934 –0.32 0.3745 0.32 0.6255 1.32 0.9066 2.32 0.9898

–2.33 0.0099 –1.33 0.0918 –0.33 0.3707 0.33 0.6293 1.33 0.9082 2.33 0.9901

–2.34 0.0096 –1.34 0.0901 –0.34 0.3669 0.34 0.6331 1.34 0.9099 2.34 0.9904

–2.35 0.0094 –1.35 0.0885 –0.35 0.3632 0.35 0.6368 1.35 0.9115 2.35 0.9906

–2.36 0.0091 –1.36 0.0869 –0.36 0.3594 0.36 0.6406 1.36 0.9131 2.36 0.9909

–2.37 0.0089 –1.37 0.0853 –0.37 0.3557 0.37 0.6443 1.37 0.9147 2.37 0.9911

–2.38 0.0087 –1.38 0.0838 –0.38 0.3520 0.38 0.6480 1.38 0.9162 2.38 0.9913

–2.39 0.0084 –1.39 0.0823 –0.39 0.3483 0.39 0.6517 1.39 0.9177 2.39 0.9916

–2.40 0.0082 –1.40 0.0808 –0.40 0.3446 0.40 0.6554 1.40 0.9192 2.40 0.9918

–2.41 0.0080 –1.41 0.0793 –0.41 0.3409 0.41 0.6591 1.41 0.9207 2.41 0.9920

–2.42 0.0078 –1.42 0.0778 –0.42 0.3372 0.42 0.6628 1.42 0.9222 2.42 0.9922

–2.43 0.0075 –1.43 0.0764 –0.43 0.3336 0.43 0.6664 1.43 0.9236 2.43 0.9925

–2.44 0.0073 –1.44 0.0749 –0.44 0.3300 0.44 0.6700 1.44 0.9251 2.44 0.9927

–2.45 0.0071 –1.45 0.0735 –0.45 0.3264 0.45 0.6736 1.45 0.9265 2.45 0.9929

–2.46 0.0069 –1.46 0.0721 –0.46 0.3228 0.46 0.6772 1.46 0.9279 2.46 0.9931

–2.47 0.0068 –1.47 0.0708 –0.47 0.3192 0.47 0.6808 1.47 0.9292 2.47 0.9932

PART | II
–2.48 0.0066 –1.48 0.0694 –0.48 0.3156 0.48 0.6844 1.48 0.9306 2.48 0.9934

–2.49 0.0064 –1.49 0.0681 –0.49 0.3121 0.49 0.6879 1.49 0.9319 2.49 0.9936

–2.50 0.0062 –1.50 0.0668 –0.50 0.3085 0.50 0.6915 1.50 0.9332 2.50 0.9938

Probability
–2.51 0.0060 –1.51 0.0655 –0.51 0.3050 0.51 0.6950 1.51 0.9345 2.51 0.9940
 Chapter | 7
TABLE 7.3.1 Standard Normal Probability Table (See Figure 7.3.5)—cont’d
z Value Probability z Value Probability z Value Probability z Value Probability z Value Probability z Value Probability
–2.52 0.0059 –1.52 0.0643 –0.52 0.3015 0.52 0.6985 1.52 0.9357 2.52 0.9941

–2.53 –1.53 –0.53

Random Variables
0.0057 0.0630 0.2981 0.53 0.7019 1.53 0.9370 2.53 0.9943

–2.54 0.0055 –1.54 0.0618 –0.54 0.2946 0.54 0.7054 1.54 0.9382 2.54 0.9945

–2.55 0.0054 –1.55 0.0606 –0.55 0.2912 0.55 0.7088 1.55 0.9394 2.55 0.9946

–2.56 0.0052 –1.56 0.0594 –0.56 0.2877 0.56 0.7123 1.56 0.9406 2.56 0.9948

–2.57 0.0051 –1.57 0.0582 –0.57 0.2843 0.57 0.7157 1.57 0.9418 2.57 0.9949

–2.58 0.0049 –1.58 0.0571 –0.58 0.2810 0.58 0.7190 1.58 0.9429 2.58 0.9951

–2.59 0.0048 –1.59 0.0559 –0.59 0.2776 0.59 0.7224 1.59 0.9441 2.59 0.9952

–2.60 0.0047 –1.60 0.0548 –0.60 0.2743 0.60 0.7257 1.60 0.9452 2.60 0.9953

–2.61 0.0045 –1.61 0.0537 –0.61 0.2709 0.61 0.7291 1.61 0.9463 2.61 0.9955

–2.62 0.0044 –1.62 0.0526 –0.62 0.2676 0.62 0.7324 1.62 0.9474 2.62 0.9956

–2.63 0.0043 –1.63 0.0516 –0.63 0.2643 0.63 0.7357 1.63 0.9484 2.63 0.9957

–2.64 0.0041 –1.64 0.0505 –0.64 0.2611 0.64 0.7389 1.64 0.9495 2.64 0.9959

–2.65 0.0040 –1.65 0.0495 –0.65 0.2578 0.65 0.7422 1.65 0.9505 2.65 0.9960

–2.66 0.0039 –1.66 0.0485 –0.66 0.2546 0.66 0.7454 1.66 0.9515 2.66 0.9961

–2.67 0.0038 –1.67 0.0475 –0.67 0.2514 0.67 0.7486 1.67 0.9525 2.67 0.9962

–2.68 0.0037 –1.68 0.0465 –0.68 0.2483 0.68 0.7517 1.68 0.9535 2.68 0.9963

–2.69 0.0036 –1.69 0.0455 –0.69 0.2451 0.69 0.7549 1.69 0.9545 2.69 0.9964

–2.70 0.0035 –1.70 0.0446 –0.70 0.2420 0.70 0.7580 1.70 0.9554 2.70 0.9965

–2.71 0.0034 –1.71 0.0436 –0.71 0.2389 0.71 0.7611 1.71 0.9564 2.71 0.9966

–2.72 0.0033 –1.72 0.0427 –0.72 0.2358 0.72 0.7642 1.72 0.9573 2.72 0.9967

–2.73 0.0032 –1.73 0.0418 –0.73 0.2327 0.73 0.7673 1.73 0.9582 2.73 0.9968

–2.74 0.0031 –1.74 0.0409 –0.74 0.2296 0.74 0.7704 1.74 0.9591 2.74 0.9969

–2.75 0.0030 –1.75 0.0401 –0.75 0.2266 0.75 0.7734 1.75 0.9599 2.75 0.9970

–2.76 0.0029 –1.76 0.0392 –0.76 0.2236 0.76 0.7764 1.76 0.9608 2.76 0.9971

–2.77 0.0028 –1.77 0.0384 –0.77 0.2206 0.77 0.7794 1.77 0.9616 2.77 0.9972

169
 170
TABLE 7.3.1 Standard Normal Probability Table (See Figure 7.3.5)—cont’d
z Value Probability z Value Probability z Value Probability z Value Probability z Value Probability z Value Probability
–2.78 0.0027 –1.78 0.0375 –0.78 0.2177 0.78 0.7823 1.78 0.9625 2.78 0.9973

–2.79 0.0026 –1.79 0.0367 –0.79 0.2148 0.79 0.7852 1.79 0.9633 2.79 0.9974

–2.80 0.0026 –1.80 0.0359 –0.80 0.2119 0.80 0.7881 1.80 0.9641 2.80 0.9974

–2.81 0.0025 –1.81 0.0351 –0.81 0.2090 0.81 0.7910 1.81 0.9649 2.81 0.9975

–2.82 0.0024 –1.82 0.0344 –0.82 0.2061 0.82 0.7939 1.82 0.9656 2.82 0.9976

–2.83 0.0023 –1.83 0.0336 –0.83 0.2033 0.83 0.7967 1.83 0.9664 2.83 0.9977

–2.84 0.0023 –1.84 0.0329 –0.84 0.2005 0.84 0.7995 1.84 0.9671 2.84 0.9977

–2.85 0.0022 –1.85 0.0322 –0.85 0.1977 0.85 0.8023 1.85 0.9678 2.85 0.9978

–2.86 0.0021 –1.86 0.0314 –0.86 0.1949 0.86 0.8051 1.86 0.9686 2.86 0.9979

–2.87 0.0021 –1.87 0.0307 –0.87 0.1922 0.87 0.8078 1.87 0.9693 2.87 0.9979

–2.88 0.0020 –1.88 0.0301 –0.88 0.1894 0.88 0.8106 1.88 0.9699 2.88 0.9980

–2.89 0.0019 –1.89 0.0294 –0.89 0.1867 0.89 0.8133 1.89 0.9706 2.89 0.9981

–2.90 0.0019 –1.90 0.0287 –0.90 0.1841 0.90 0.8159 1.90 0.9713 2.90 0.9981

–2.91 0.0018 –1.91 0.0281 –0.91 0.1814 0.91 0.8186 1.91 0.9719 2.91 0.9982

–2.92 0.0018 –1.92 0.0274 –0.92 0.1788 0.92 0.8212 1.92 0.9726 2.92 0.9982

–2.93 0.0017 –1.93 0.0268 –0.93 0.1762 0.93 0.8238 1.93 0.9732 2.93 0.9983

–2.94 0.0016 –1.94 0.0262 –0.94 0.1736 0.94 0.8264 1.94 0.9738 2.94 0.9984

–2.95 0.0016 –1.95 0.0256 –0.95 0.1711 0.95 0.8289 1.95 0.9744 2.95 0.9984

–2.96 0.0015 –1.96 0.0250 –0.96 0.1685 0.96 0.8315 1.96 0.9750 2.96 0.9985

–2.97 0.0015 –1.97 0.0244 –0.97 0.1660 0.97 0.8340 1.97 0.9756 2.97 0.9985

PART | II
–2.98 0.0014 –1.98 0.0239 –0.98 0.1635 0.98 0.8365 1.98 0.9761 2.98 0.9986

–2.99 0.0014 –1.99 0.0233 –0.99 0.1611 0.99 0.8389 1.99 0.9767 2.99 0.9986

–3.00 0.0013 –2.00 0.0228 –1.00 0.1587 1.00 0.8413 2.00 0.9772 3.00 0.9987

Probability
Chapter | 7 Random Variables 171

This (z = –1.67) tells you that $15 million is 1.67 stan-

dard deviations below the mean (the forecast value).9 Your
problem has now been reduced to finding a standard normal
probability:
Find the probability that a standard normal variable is
less than z = –1.67.
From the table, you find the answer:
The probability of a really bad quarter is 0.0475, or
about 5%. $10 $15 $20 $25 $30
Whew! It seems that a really bad quarter is not very Sales (millions)
likely. However, a 5% chance is an outside possibility FIGURE 7.3.6 The probability of a really bad quarter (sales less than
that should not be disregarded altogether. $15 million) is represented by the shaded area under the curve. This is
Figures 7.3.6 and 7.3.7 show this probability calculation, based on the forecast of $20 million and the standard deviation of $3 million.
both in terms of sales dollars and in terms of standardized The answer is found by standardizing and then using the standard normal
probability table.
sales numbers (standard deviations above or below the mean).
This was an easy problem, since the answer was found
directly from the standard normal probability table. Here is
a question that requires a little more care:
Continuing with the sales-forecasting problem, find the
probability of a “really good quarter,” which is defined
as sales in excess of $24 million.
The first step is to standardize the sales number: $24 mil-
lion is z = (24 – 20)/3 = 1.33 standard deviations above the
mean. Thus, you are asked to solve the following problem:
Find the probability that a standard normal variable
exceeds z = 1.33. −4 −3 −2 −1 0 1 2 3 4
z = − 1.67
Using the complement rule, we know that this probabil-
Standardized sales numbers (z)
ity is 1 minus the probability of being less than z = 1.33.
Looking up 1.33 in the table, you find the answer: FIGURE 7.3.7 The probability of a really bad quarter, in terms of stan-
dardized sales numbers. This is the probability that sales will be more than
Probability of a really good quarter = 1 − 0:9082 z = –1.67 standard deviations below the mean. The answer is 0.0475.
= 0.0918, or about 9%
This probability is illustrated, in standardized numbers,
in Figure 7.3.8.
Here’s another kind of problem:
Continuing with the sales-forecasting problem, find the
probability of a “typical quarter,” which is defined as
sales between $16 million and $23 million.
Begin by standardizing both of these numbers, to see
that your task is to solve the following problem:
Find the probability that a standard normal is between
z1 = –1.33 and z2 = 1.00. −4 −3 −2 −1 0 1 2 3 4
z = 1.33
Standardized sales numbers (z)

9. You know that this is below the mean because the standardized number FIGURE 7.3.8 The probability of a really good quarter, in terms of stan-
z = –1.67 is negative. The standardized number z will be positive for any dardized sales numbers. The shaded area is 1 minus the unshaded area
number above the mean. The standardized number z for the mean itself under the curve, which may be looked up in the table. The answer is
is 0. 0.0918.
172 PART | II Probability

To solve this kind of problem, look up each stan-

dardized number in the table and find the difference
between the probabilities for the answer. Be sure to subtract
the smaller from the larger so that your answer is a positive
number and therefore a “legal” probability!
Probability of a typical quarter = 0:8413 − 0:0918
= 0.7495, or about 75%
This probability is illustrated, in standardized numbers,
in Figure 7.3.9. −4 −3 −2 −1 0 1 2 3 4
Finally, here’s yet another kind of problem:
z1 = −1.33 z2 = 1
Continuing with the sales-forecasting problem, find the
Standardized sales numbers
probability of a “surprising quarter,” which is defined
as sales either less than $16 million or more than FIGURE 7.3.10 The probability of a surprising quarter, in terms of stan-
$23 million. dardized sales numbers. The shaded area is found by looking up each stan-
dardized number in the table, finding the difference between the
This asks for the probability of not being between two probabilities, and subtracting the result from 1. The answer is 0.2505.
numbers. Using the complement rule, you may simply
take 1 minus the probability found in the preceding exam- NORMDIST calculations because it is the probability of
ple, which was the probability of being between these two being between two values. Here are the results:
values. The answer is therefore as follows:
Probability of a surprising quarter = 1 − 0:7495
= 0.2505, or about 25%
This probability is illustrated, in standardized numbers, This cell is
named “Mean”
in Figure 7.3.10.
To use Excel® to compute these first three probabilities, This cell is
named “stdDev”
we use the function “NORMDIST(value,mean,standardDe-
viation,TRUE)” to find the probability that a normal distri-
bution with specified mean and standard deviation is less
than some value. There is no need to standardize because
Excel will do this for you as part of the calculation. The
first calculation is straightforward because it is a probability
of being less. The second calculation is one minus the
NORMDIST function because it is a probability of being
greater. The third calculation is the difference of two The Four Different Probability Calculations
Here is a summary table of the four types of problems and
how to solve them. The values z, z1, and z2 represent stan-
dardized numbers from the problem, found by subtracting
the mean and dividing by the standard deviation. The
table referred to is the standard normal probability table.

Computing Probabilities for a Normal Distribution

To Find the Probability Procedure

−4 −3 −2 −1 0 1 2 3 4 of Being
Less than z Look up z in the table
z1 = − 1.33 z2 = 1 More than z Subtract above answer from 1
Standardized sales numbers Between z1 and z2 Look up z1 and z2 in the
table, and subtract smaller
FIGURE 7.3.9 The probability of a typical quarter in terms of stan- probability from larger
dardized sales numbers. The shaded area is found by looking up each stan- Not between z1 and z2 Subtract above answer (for
dardized number in the table and then subtracting. Subtracting eliminates “between z1 and z2”) from 1
the unshaded area at the far left. The answer is 0.7495.
Chapter | 7 Random Variables 173

You may be wondering if there is a difference between

the two events “sales exceeded $22 million” and “sales anything at all is to win the full amount, $500, which is at
least $50.
were at least $22 million.” The term exceeded means
What if you assumed a normal distribution with this same
more than, whereas the term at least means more than or
mean ($50) and standard deviation ($150)? How far from the
equal to. In fact, for a normal distribution, there is no differ- correct answer (10%) would you be? Very far away, because
ence between the probabilities of these two events; the dif- the probability that a normally distributed random variable
ference between the probabilities is just a geometric line, exceeds its mean is 0.5 or 50%.
which represents no area under the normal curve. This is a big difference: 10% (the correct answer) versus
50% (computed by wrongly assuming a normal distribution).
Figure 7.3.11 shows the large difference between the actual
Be Careful: Things Need Not Be Normal! discrete distribution and the normal distribution with the
If you have a normal distribution and you know the mean same mean and standard deviation. Always be careful
and standard deviation, you can find correct probabilities about assuming a normal distribution!
by standardizing and then using the standard normal
probability table. Fortunately, if the distribution is only
approximately normal, your probabilities will still be
approximately correct.
7.4 THE NORMAL APPROXIMATION TO
However, if the distribution is very far from normal, THE BINOMIAL
then any probabilities you might compute based on the Remember the binomial distribution? It is the number of
mean, the standard deviation, and the normal table could times that something happens out of n independent tries.
be very wrong indeed. A binomial distribution can never be exactly normal, for
two reasons. First, any normal distribution is free to pro-
Example duce observations with decimal parts (e.g., 7.11327),
A Lottery (or Risky Project) whereas the binomial number X is restricted to whole num-
Consider a lottery (or a risky project, if you prefer) that pays bers (e.g., 7). Second, a binomial distribution is skewed
back nothing 90% of the time, but pays $500 the remaining whenever π is any number other than 0.5 (becoming more
10% of the time. The expected (mean) payoff is $50, and the and more skewed when π is close to 0 or to 1), whereas nor-
standard deviation is $150 for this discrete random variable. mal distributions are always perfectly symmetric.
Note that this does not represent a normal distribution; it’s However, a binomial distribution can be closely
not even close because it’s so discrete, with only two possi- approximated by a normal distribution whenever the bino-
ble values. mial n is large and the probability π is not too close to
What is the probability of winning at least $50? The
0 or 1.10,11 This fact will help you find probabilities (of
correct answer is 10% because the only way to win
being less, more, between, or not between given numbers)
for a binomial distribution by replacing many complex and
difficult calculations (using the earlier formula for indivi-
1.0
dual binomial probabilities) with a single simpler calcula-
Probability (for discrete distribution)

0.9 tion (using the normal distribution).

0.8 But how to choose a normal distribution that will be
0.7 close to a given binomial distribution? A good choice is
0.6 to use the normal distribution that has the same mean
and standard deviation as the binomial distribution you
0.5
wish to approximate. Since you already know how to
0.4
find the mean and standard deviation for a binomial distri-
0.3 bution (from Section 7.2), and you know how to find
0.2 probabilities for a normal distribution with given mean
0.1
0.0
($500) $0 $500 10. When π is close to 0 or 1, the approach to a normal distribution is
Payoff slower as n increases due to skewness of the binomial with rare or
nearly definite events. The Poisson distribution, covered in a later
FIGURE 7.3.11 The discrete distribution of the payoff and the normal section, is a good approximation to the binomial when n is large and π
distribution having the same mean ($50) and standard deviation ($150). is close to 0.
These distributions and their probabilities are very different. The discrete 11. The central limit theorem, to be covered in Chapter 8, tells how a
distribution gives the correct answers; the assumption of normality is normal distribution emerges when many independent random trials are
wrong in this case. combined by adding or averaging.
174 PART | II Probability

n = 100 π = 0.1 n = 10 π = 0.1

0.5
0.15
0.4

Binomial probability
Binomial probability

0.10 0.3

0.2

0.05
0.1

0
0 0 1 2 3 4 5 6 7 8 9 10
0 10 20
FIGURE 7.4.2 The probability distribution of a binomial with n = 10 and
FIGURE 7.4.1 The probability distribution of a binomial with n = 100 π = 0.10 is not very normal because n is not large enough.
and π = 0.10 is fairly close to normal.

and standard deviation (from Section 7.3), it should not be Using the Normal Approximation to the Binomial
difficult for you to compute these approximate binomial (Whole Numbers a and b):
probabilities.
Here is convincing evidence of the binomial approxima- The Probability Is Approximated by the Probability
tion to the normal. Suppose n is 100 and π is 0.10. The That the Binomial Is: That the Corresponding Normal Is:
probability distribution, computed using the binomial for- Exactly 8 Between 7.5 and 8.5
mula, is shown in Figure 7.4.1. It certainly has the bell Exactly a Between a – 0.5 and a + 0.5
shape of a normal distribution. Although it is still discrete, Between 15 and 23 Between 14.5 and 23.5
with separate, individual bars, there are enough observa- Between a and b Between a – 0.5 and b + 0.5
tions that the discreteness is not a dominant feature.
To approximate a binomial (which is discrete, only
taking on whole-number values) by using a normal ran- Compare Figure 7.4.1 (n = 100) to Figure 7.4.2 (n = 10)
dom variable (which is continuous), we will do better if to see that, with smaller n, the distribution is not as normal.
we extend the limits by one-half in each direction in Furthermore, the discreteness is more important when n is
order to include all numbers that round to the whole num- small.
ber(s).12 For example, to approximate the probability that
a binomial X is equal to 3, we would find the probability Example
that a normal distribution (with the same mean and stan- High- and Low-Speed Microprocessors
dard deviation) is between 2.5 and 3.5. We need to do We often don’t have as much control over a manufacturing
this because the probability is zero that any normal ran- process as we would like. Such is the case with sophisticated
dom variable is equal to 3 and, actually, all values that microprocessor chips, such as some of those used in micro-
the normal produced that were between 2.5 and 3.5 computers, which can have over a billion transistors placed
would round to 3. Similarly, to find the probability that on a chip of silicon smaller than a square inch. Despite care-
a binomial is between 6 and 9, you would find the prob- ful controls, there is variation within the resulting chips:
Some will run at higher speeds than others.
ability that a normal (with same mean and standard devia-
In the spirit of the old software saying “It’s not a bug, it’s a
tion) is between 5.5 and 9.5. The probability of not being
feature!” the chips are sorted according to the speed at which
between two numbers is, as usual, one minus the probabil- they will actually run and priced accordingly (with the faster
ity of being between them. chips commanding a higher price). The catalog lists two
products: 2 gigahertz (slower) and 3 gigahertz (faster).
Your machinery is known to produce the slow chips 80%
of the time, on average, and fast chips the remaining 20% of
12. We assume here that you are looking for probabilities about a binomial
the time, with chips being slow or fast independently of one
number of occurrences X. If, on the other hand, you need probabilities for a another. Today your goal is to ship 1,000 slow chips and 300
binomial proportion or percentage p, you should first convert to X and then fast chips, perhaps with some chips left over. How many
proceed from there. For example, the probability of observing “at least should you schedule for production?
20% of 261” is the same as the probability of observing “at least 53 of If you schedule 1,300 total chips, you expect 80% (1,040
261” since you need at least 0.20 × 261 = 52.2 and can observe only chips) to be slow and 20% (260 chips) to be fast. You would
whole numbers.
Chapter | 7 Random Variables 175

One of the uses of probability is to help you understand

have enough slow ones, but not enough fast ones, on what is happening “behind the scenes” in the real world.
average.
Let’s see what might really be happening in an opinion poll
Since you know that you are limited by the number of fast
by using a What if scenario analysis.
chips, you compute 300/0.20 = 1,500. This tells you that if
you schedule 1,500 chips, you can expect 20% of these
(or 300 chips) to be fast. So on average, you would just Example
meet the goal. Unfortunately, this means that you have Polling the Electorate
only about a 50% chance of meeting the goal for fast chips! Your telephone polling and research firm was hired to con-
Suppose you schedule 1,650 chips for production. What duct an opinion poll to see if a new municipal bond initiative
is the probability that you will be able to meet the goal? To is likely to be approved by the voters in the next election.
solve this, you first state it as a complete probability question: You decided to interview 800 randomly selected representa-
Given a binomial random variable (the number of fast tive people who are likely to vote, and you found that 437
chips produced) with n = 1,650 total chips produced intend to vote in favor. Here’s the What if: If the entire elec-
and π = 0.20 probability that a chip is fast, find the prob- torate were, in fact, evenly divided on the issue, what is the
ability that this random variable is at least 300 but no probability that you would expect to see this many or more of
more than 650.13 your sample in favor?

To solve this problem using the binomial distribution Your coworker: “It looks pretty close: 437 out of 800 is
directly would require that you compute the probability for pretty close to 50–50, which would be 400 out of 800.”
300, for 301, for 302, and so on until you got tired. The nor- You: “But 437 seems lots bigger than 400 to me.
mal approximation to the binomial allows you to solve it Let’s find out if the extra 37 could reasonably be just
much faster using the standard normal probability table. randomness.”
You will need to know the mean and standard deviation of Your coworker: “OK. Let’s assume that each person is as
the number of fast chips produced: likely to be in favor as not. Then we can compute the
chances of seeing 437 or more.”
μðNumber of fast chipsÞ = nπ You: “OK. If the chances are more than 5% or 10%, then
= 1,650 × 0:20 = 330 the extra 37 could reasonably be just randomness. But if
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the chances are really small, say under 5% or under 1%,
σ ðNumber of fast chipsÞ =nπð1 − πÞ then it would seem that more than just randomness
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
= 1,650 × 0:20 × 0:80 = 16:24807 is involved.”

You also need to standardize the bounds on the number To do the calculation, let X represent the following bino-
of fast chips needed, 300 and 650 (after extending them by mial random variable: the number of people (out of 800 inter-
a half to 299.5 and 650.5), using the mean and standard viewed) who say they intend to vote in favor. If we assume that
deviation computed just above: people are evenly divided on the issue, then the probability for
each person interviewed is π = 0.50 that they intend to vote in
z1 = Standardized lower number of fast chips = 299:5 − 330 favor. Now let’s find the mean and standard deviation of X
16:24807
using formulas for the binomial distribution:
= −1:88
μX = nπ = ð800Þð0:50Þ = 400

650:5 − 330 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

z2 = Standardized upper number of fast chips = σX = nπð1 − πÞ
16:24807
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
= 19:73 = ð800Þð0:50Þð1 − 0:50Þ = 14:14214

Now, to find the probability that X is at least 437, we extend

Looking up these standardized values in the standard nor-
the limit by one-half to find the probability that X is at least
mal probability table, you find 0.030 for z1 = –1.88 and note
436.5 and then use the fact that X is approximately normal.
that z2 = 19.73 is way off the end of the table, so you use the
That is, we now find the probability that a normally distributed
value 1.14 Subtracting these numbers to find the probability
random variable with mean 400 and standard deviation
of being between the bounds, the answer is 1 – 0.030 =
14.14214 is larger than 436.5 by standardizing as follows:
0.970. You conclude that if 1,650 chips are scheduled, you
have a 97% chance of meeting the goal of shipping 300 fast 436:5 − μX
z = Standardized value =
chips and 1,000 slow ones. σX
436:5 − 400
= = 2:58
13. The reason is that more than 1,650 – 1,000 = 650 fast chips would 14:14214
imply fewer than 1,000 slow chips; thus, you would not be able to
meet the goal for slow chips. Using the normal probability tables, you find that the
14. This makes sense; the probability that a standard normal variable will
probability of seeing this large a margin (or larger) in favor
be less than z2 = 19.73 standard deviations above its mean is essentially 1,
since this nearly always happens. (Continued)
176 PART | II Probability

0.8
Example—cont’d
0.7
in the sample, assuming that the population is evenly
divided, is 1 – 0.995 = 0.005. This is very unlikely: a prob- 0.6

Poisson probability
ability of about a half a percent, or 1 out of 200. 0.5
You asked, what if the population were evenly divided,
0.4
and found the answer: A sample percentage of 54.6% (this
is 437/800) or more is highly unlikely. The conclusion is 0.3
that you have evidence against the What if scenario of evenly
0.2
divided voters. It looks good for the initiative!
0.1
0
0 1 2 3 4 5

7.5 TWO OTHER DISTRIBUTIONS: THE Mean = 0.5

POISSON AND THE EXPONENTIAL Number of occurrences

Many other probability distributions are useful in statistics. FIGURE 7.5.1 The Poisson distribution with 0.5 occurrences expected is
a skewed distribution. There is a high probability, 0.607, that no occur-
This section provides brief descriptions of two such distribu- rences will happen at all.
tions with an indication of how they might fit in with some
general classes of business applications. The Poisson distri- 0.4
bution is often useful as a model of the number of events
that occur during a fixed time, such as arrivals. The expo-
nential distribution can work well as a model of the amount 0.3
Poisson probability

of time, such as that required to complete an operation.

These distributions work well together (as the Poisson pro-
0.2
cess) with the Poisson representing the number of events
and the exponential describing the time between events.
0.1
The Poisson Distribution
The Poisson distribution, like the binomial, is a counted 0
0 1 2 3 4 5 6 7 8 9 10
number of times something happens. The difference is that
there is no specified number n of possible tries. Here is
Mean = 2
one way that it can arise. If an event happens independently
Number of occurrences
and randomly over time, and the mean rate of occurrence is
constant over time, then the number of occurrences in a fixed FIGURE 7.5.2 The Poisson distribution with two occurrences expected.
amount of time will follow the Poisson distribution.15 The The distribution is still somewhat skewed.
Poisson is a discrete distribution and depends only on the
mean number of occurrences expected. 0.10
Here are some random variables that might follow a 0.09
Poisson distribution: 0.08
Poisson probability

0.07
1. The number of orders your firm receives tomorrow. 0.06
2. The number of people who apply for a job tomorrow to 0.05
your human resources division.
0.04
3. The number of defects in a finished product.
0.03
4. The number of calls your firm receives next week for
0.02
help concerning an “easy-to-assemble” toy.
0.01
5. A binomial number X when n is large and π is small.
0
The following figures show what the Poisson probabil- 0 10 20 30 40 50
ities look like for a system expecting a mean of 0.5 occur-
Mean = 20
rence (Figure 7.5.1), 2 occurrences (Figure 7.5.2), and 20
Number of occurrences

FIGURE 7.5.3 The Poisson distribution with 20 occurrences expected.

15. Poisson is a French name, pronounced (more or less) “pwah-soh.” The distribution, although still discrete, is now fairly close to normal.
Chapter | 7 Random Variables 177

occurrences (Figure 7.5.3). Note from the bell shape of

Figure 7.5.3 that the Poisson distribution is approximately
normal when many occurrences are expected.
There are three important facts about a Poisson distri-
bution. These facts, taken together, tell you how to find
probabilities for a Poisson distribution when you know
only its mean.

For a Poisson Distribution

1. The standard deviation is always equal to the square root

pffiffiffi Example
of the mean: σ ¼ .
2. The exact probability that a Poisson random variable X How Many Phone Calls?
with mean  is equal to a is given by the formula Your firm handles 460 calls per day, on average. Assuming a
Poisson distribution, find the probability that you will be
μ −μ
a
PðX = aÞ = e overloaded tomorrow, with 500 or more calls received.
a! The mean, μ = 460, is given. The standard deviation is σ =
where e = 2.71828 . . . is a special number.16 21.44761. You may use the normal approximation because
3. If the mean is large, then the Poisson distribution is the mean (460) is so large. Since the normal distribution is
approximately normal. continuous, any value over 499.5 will round to 500 or
more. The standardized number of calls is
16. This special mathematical number also shows up in continuously
compounded interest formulas. 499:5 − μ 499:5 − 460
z= = = 1:84
σ 21:44761
When you use the standard normal probability table, the
answer is a probability of 1 – 0.967 = 0.033, so you may
Example expect to be overloaded tomorrow with probability only
How Many Warranty Returns? about 3% (not very likely but within possibility).
Because your firm’s quality is so high, you expect only 1.3 of
your products to be returned, on average, each day for war-
ranty repairs. What are the chances that no products will be
returned tomorrow? That one will be returned? How about The Exponential Distribution
two? How about three?
Since the mean (1.3) is so small, exact calculations are The exponential distribution is the very skewed continu-
needed. Here are the details: ous distribution shown in Figure 7.5.4. Its rise is vertical
at 0, on the left, and it descends gradually, with a long
0
PðX = 0Þ = 1:3 e−1:3 = 1 × 0:27253 = 0:27253 tail on the right.
0! 1
The following is a situation in which the exponential
1:31 −1:3 1:3
PðX = 1Þ = e = × 0:27253 = 0:35429 distribution is appropriate. If events happen independently
1! 1
2
PðX = 2Þ = 1:3 e−1:3 = 1:69 × 0:27253 = 0:23029
2! 2 The exponential distribution
3
1:3 −1:3 2:197
PðX = 3Þ = e = × 0:27253 = 0:09979
3! 6
From these basic probabilities, you could add up
the appropriate probabilities for 0, 1, and 2 to also find the
probability that two items or fewer will be returned. The
probability is, then, 0.27253 + 0.35429 + 0.23029 = 0.857,
or 85.7%.
To use Excel to compute these probabilities, you could
use the function “POISSON(value,mean,FALSE)” to find
the probability that a Poisson random variable is exactly
equal to some value, and you could use “POISSON(value,
mean,TRUE)” to find the probability that a Poisson random 0
0
variable is less than or equal to the value. Here are the
results: FIGURE 7.5.4 The exponential distribution is a very skewed distribution
that is often used to represent waiting times between events.
178 PART | II Probability

and randomly with a constant rate over time, the waiting Waiting time
time between successive events follows an exponential between successive events
is exponentially distributed
distribution.17
Here are some examples of random variables that might
follow an exponential distribution: Time

1. Time between customer arrivals at an auto repair shop. Number of events

in a fixed time interval
2. The amount of time your copy machine works between follows a Poisson distribution
visits by the repair people. A fixed time A fixed time
3. The length of time of a typical telephone call.
4. The time until a TV system fails. FIGURE 7.5.5 The relationship between the exponential and the Poisson
5. The time it takes to provide service for one customer. distributions when events happen over time independently and at a con-
stant rate.
The exponential distribution has no memory in the sur-
prising sense that after you have waited awhile without suc-
cess for the next event, your mean waiting time remaining Example
until the next event is no shorter than it was when you Customer Arrivals
started! This makes sense for waiting times, since occur- Suppose customers arrive independently at a constant mean
rences are independent of one another and “don’t know” rate of 40 per hour. To find the probability that at least one cus-
that none have happened recently. tomer arrives in the next five minutes, note that this is the prob-
What does this property say about telephone calls? Sup- ability that the exponential waiting time until the next customer
pose you are responsible for a switching unit for which the arrives is less than five minutes. Since 40 customers arrive each
average call lasts five minutes. Consider all calls received at hour, on average, the mean of this exponential random variable
a given moment. On average, you expect them to last five is μ = 1/40 = 0.025 hours, or 0.025 × 60 = 1.5 minutes. The
minutes, with the individual durations following the expo- probability is then PðX ≤ 5Þ = 1 − e−5/1:5 = 0:964, which may
nential distribution. After one minute passes, some of be computed in Excel® using the formula “=1−EXP(–5/1.5)”.
these calls have ended. However, the calls that remain are So the chances are high (96.4%) that at least one customer
will arrive in the next 5 minutes.
all expected, on average, to last five minutes more. The rea-
son is that the shorter calls have already been eliminated.
While this may be difficult to believe, it has been confirmed
(approximately) using real data. 7.6 END-OF-CHAPTER MATERIALS
Here are the basic facts for an exponential distribution.
Note that there is no “normal approximation” because the Summary
exponential distribution is always very skewed.
A random variable is a specification or description of a
numerical result from a random experiment. A particular
For an Exponential Distribution value taken on by a random variable is called an observa-
tion. The pattern of probabilities for a random variable is
1. The standard deviation is always equal to the mean: σ = μ.
2. The exact probability that an exponential random vari-
called its probability distribution. Random variables are
able X with mean μ is less than a is given by the formula either discrete (if you can list all possible outcomes) or
continuous (if any number in a range is possible). Some
PðX ≤ aÞ = 1 − e−a/μ random variables are actually discrete, but you can work
with them as though they were continuous.
For a discrete random variable, the probability distribu-
There is a relationship between the exponential and the
tion is a list of the possible values together with their prob-
Poisson distributions when events happen independently at
abilities of occurrence. The mean or expected value and the
a constant rate over time. The number of events in any fixed
standard deviation are computed as follows.
time period is Poisson, and the waiting time between events
For a discrete random variable:
is exponential. This is illustrated in Figure 7.5.5. In fact, the
distribution of the waiting time from any fixed time until μ = EðXÞ = Sum of ðvalue times probabilityÞ = ∑ XPðXÞ
the next event is exponential. pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
σ = Sum of ðsquared deviation times probabilityÞ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
= ∑ðX − μÞ2 PðXÞ

17. Note that this implies that the total number of events follows a Poisson The interpretations are familiar. The mean or expected
distribution. value indicates the typical or average value, and the
Chapter | 7 Random Variables 179

standard deviation indicates the risk in terms of approxi- number of standard deviations above the mean, or below
mately how far from the mean you can expect to be. the mean if the standardized number is negative) by sub-
A random variable X has a binomial distribution if it tracting the mean and dividing by the standard deviation:
represents the number of occurrences of an event out of n
Number − Mean
trials, provided (1) for each of the n trials, the event always z = Standardized number =
Standard deviation
has the same probability π of happening, and (2) the trials
Number − μ
are independent of one another. The binomial proportion =
is p = X/n, which also represents a percentage. The mean σ
and standard deviation of a binomial or binomial proportion Finally, look up the standardized number or numbers in
may be found as follows: the standard normal probability table and use the following
summary table (where z, z1, and z2 are standardized num-
Mean and Standard Deviation for a Binomial bers) to find the final answer:
Distribution
Number of Proportion or
Computing Probabilities for a Normal Distribution
Occurrences, X Percentage, p = X/n To Find the Procedure
Mean EðXÞ = μX = nπ EðpÞ = μp = π Probability of Being
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi πð1 − πÞ Less than z Look up z in the table
Standard deviation σ X = nπð1 − πÞ σp =
n More than z Subtract above answer from 1
Between z1 and z2 Look up z1 and z2 in the table, and
The probability that a binomial random variable X is subtract smaller probability from larger
equal to some given number a (from 0 to n) is given by Not between z1 and z2 Subtract above answer
the following formula. Binomial probability that X equals a: (for “between z1 and z2”) from 1

!
n Probabilities for a binomial distribution may be approxi-
PðX = aÞ = π a ð1− πÞn−a mated using the normal distribution with the same mean
a and standard deviation, provided n is large and π is not
n! too close to 0 or 1. Since the normal distribution is contin-
= π a ð1 −πÞn−a
a!ðn −aÞ! uous, extend the limits by one-half in each direction (for
1× 2× 3× . . . ×n example, the probability that a binomial is exactly a
= π a ð1− πÞn−a
ð1 ×2 ×3 × . . . × aÞ½1 ×2 ×3 × . . . × ðn− aÞ whole number a is approximated by the probability that
the corresponding normal is between a – 0.5 and a + 0.5).
If occurrences happen independently and randomly over
The notation n! is n factorial, the product of the numbers time, and the average rate of occurrence is constant
from 1 to n, with 0! = 1 by definition. The notation over time, then the number of occurrences that happen in a
 
n n! fixed amount of time will follow the Poisson distribution,
= a discrete random variable. The standard deviation is the
a a!ðn − aÞ!
square root of the mean. If the mean is large, the Poisson
is the binomial coefficient, read aloud as “n choose a.” distribution is approximately normal and the standard
The normal distribution, a continuous distribution, is normal probability table may be used. Exact Poisson prob-
represented by the familiar bell-shaped curve. The probabil- abilities may be found using the following formula:
ity that a normal random variable will be between any two
μa −μ
values is equal to the area under the normal curve between PðX = aÞ = e
these two values. There is a normal distribution for each a!
combination of a mean μ and a (positive) standard deviation The exponential distribution is a very skewed contin-
σ. The standard normal distribution is a normal distribu- uous distribution useful for understanding such variables as
tion with mean μ = 0 and standard deviation σ = 1. You may waiting times and durations of telephone calls. It has no
think of the standard normal distribution as representing the “memory,” in the sense that after you have waited awhile
number of standard deviations above or below the mean. without success for the next event, your average waiting
The standard normal probability table gives the prob- time until the next event is no shorter than it was when
ability that a standard normal random variable Z is less you started. Its standard deviation is always equal to its
than any given number z. mean. The probability that an exponential random variable
To solve word problems involving normal probabilities, X with mean μ is less than or equal to a is PðX ≤ aÞ =
first identify the mean μ, standard deviation σ, and the prob- 1 − e− a/μ . There is no normal approximation for an expo-
ability asked for. Convert to a standardized number z (the nential random variable.
180 PART | II Probability

Key Words 7. a. What is a normal distribution?

b. Identify all of the different possible normal
binomial distribution, 159 distributions.
binomial proportion, 159 c. What does the area under the normal curve
continuous random variable, 156 represent?
discrete random variable, 156 d. What is the standard normal distribution? What is it
exponential distribution, 177 used for?
mean or expected value, 156 e. What numbers are found in the standard normal
normal distribution, 165 probability table?
observation, 155 f. Find the probability that a standard normal random
Poisson distribution, 176 variable is less than –1.65.
probability distribution, 155 g. How do you standardize a number?
random variable, 155 8. a. What kinds of situations give rise to a Poisson
standard deviation, 157 distribution?
standard normal distribution, 166 b. Is the Poisson a discrete or a continuous
standard normal probability table, 166 distribution?
standardized number, 166 c. What is the standard deviation of a Poisson
distribution?
d. How do you find probabilities for a Poisson distribu-
tion if the mean is large?
Questions e. How do you find exact probabilities for a Poisson
distribution?
1. a. What is a random variable? 9. a. What kinds of situations give rise to an exponential
b. What is the difference between a random variable distribution?
and a number? b. What is meant by the fact that an exponential ran-
2. a. What is a discrete random variable? dom variable has no memory?
b. What is a continuous random variable? c. Can the standard normal probability table be used to
c. Give an example of a discrete random variable that find probabilities for an exponential distribution?
is continuous for practical purposes. Why or why not?
3. a. What is the probability distribution of a discrete ran- d. How do you find probabilities for an exponential
dom variable? distribution?
b. How do you find the mean of a discrete random
variable? How do you interpret the result?
c. How do you find the standard deviation of a dis- Problems
crete random variable? How do you interpret the
Problems marked with an asterisk (*) are solved in the Self
result?
Test in Appendix C
4. a. How do you tell if a random variable has a binomial
distribution? 1. A call option on common stock is being evaluated. If the
b. What is a binomial proportion? stock goes down, the option will expire worthless. If
c. What are n, π, X, and p? the stock goes up, the payoff depends on just how
5. For a binomial distribution: high the stock goes. For simplicity, the payoffs are mod-
a. Why don’t you just construct the probability tree to eled as a discrete distribution with the probability distri-
find the probabilities? bution in Table 7.6.1. Even though options markets, in
b. How do you find the mean and the standard fact, behave more like a continuous random variable,
deviation? this discrete approximation will give useful approximate
c. How do you find the probability that X is equal to results. Answer the following questions based on the
some number? discrete probability distribution given.
d. How do you find the exact probability that X is a.* Find the mean, or expected value, of the option
greater than or equal to some number? payoff.
e. If n is a large number, how do you find the approxi- b.* Describe briefly what this expected value represents.
mate probability that X is greater than or equal to c.* Find the standard deviation of the option payoff.
some number? d.* Describe briefly what this standard deviation
6. a. What is a factorial? represents.
b. Find 3!, 0!, and 15!. e.* Find the probability that the option will pay at
c. What is a binomial coefficient? What does it repre- least $20.
sent in the formula for a binomial probability? f. Find the probability that the option will pay less
d. Find the binomial coefficient “8 choose 5.” than $30.
Chapter | 7 Random Variables 181

of qualified people is limited and Table 7.6.3 shows

TABLE 7.6.1 Probability Distribution of Payoff your subjective probabilities for each outcome.
a. Find the probability of obtaining at least one applicant.
Payoff Probability
b. Find the probability of obtaining two or more
$0 0.50 applicants.
c. Find the mean number of applicants.
10 0.25
d. Find the standard deviation of the number of
20 0.15 applicants and write a sentence interpreting its
meaning.
30 0.10
7. You work for the loan department of a large bank. You
know that one of your customers has been having
trouble with the recession and may not be able to
make the loan payment that is due next week. You
TABLE 7.6.2 Probability Distribution of Downtime believe there is a 60% chance that the payment of
$50,000 will be made in full, a 30% chance that only
Problem Downtime Probability half will be paid, and a 10% chance that no payment
(minutes) will be made at all.
Minor 5 0.60
a. Find the expected loan payment.
b. Find the degree of risk for this situation.
Substantial 30 0.30 8. You are planning to invest in a new high-tech company,
Catastrophic 120 0.10
and figure your rate of return over the coming year as in
Table 7.6.4 (where 100% says that you doubled your
money, –50% says you lost half, etc.).
a. Find the mean rate of return and explain what it
2. The length of time a system is “down” (that is, broken) is represents.
described (approximately) by the probability distribution b. Find the standard deviation of the rate of return and
in Table 7.6.2. Assume that these downtimes are exact. explain what it represents.
That is, there are three types of easily recognized problems c. Find the probability that you will earn more than
that always take this long (5, 30, or 120 minutes) to fix. 40%, according to the table.
a. What kind of probability distribution does this table d. How would you measure the risk of this investment?
represent?
b. Find the mean downtime.
c. Find the standard deviation of the downtime.
TABLE 7.6.3 Probabilities for Qualified Technical
d. What is the probability that the downtime will be
greater than 10 minutes, according to this table? Applicants
e. What is the probability that the downtime is literally Number of Applicants Probability
within one standard deviation of its mean? Is
this about what you would expect for a normal 0 0.30
distribution? 1 0.55
3. An investment will pay $105 with probability 0.7, and
$125 with probability 0.3. Find the risk (as measured 2 0.10
by standard deviation) for this investment. 3 0.05
4. On a given day, assume that there is a 30% chance you
will receive no orders, a 50% chance you will receive
one order, a 15% chance of two orders, and a 5%
chance of three orders. Find the expected number of
orders and the variability in the number of orders.
5. A new project has an uncertain cash flow. A group meet- TABLE 7.6.4 Rates of Return and Probabilities for
ing has resulted in a consensus that a reasonable way to Four Scenarios
view the possible risks and rewards is to say that the
Rate of Return Probability
project will pay $50,000 with probability 0.2, will pay
$100,000 with probability 0.3, will pay $200,000 with 100% 0.20
probability 0.4, and will pay $400,000 with probability
50 0.40
0.1. How much risk is involved here? Please give both
the name and the numerical value of your answer. 0 0.25
6. Your company is hoping to fill a key technical position
–50 0.15
and has advertised in hopes of obtaining qualified appli-
cants. Because of the demanding qualifications, the pool
182 PART | II Probability

9. You can invest in just one of four projects on a lot of 11. Suppose that 8% of the loans you authorize as vice
land you own. For simplicity, you have modeled the president of the consumer loan division of a neighbor-
payoffs (as net present value in today’s dollars) of the hood bank will never be repaid. Assume further that
projects as discrete distributions. By selling the land, you authorized 284 loans last year and that loans go
you can make $60,000 for sure. If you build an apart- sour independently of one another.
ment, you estimate a payoff of $130,000 if things a. How many of these loans, authorized by you, do
go well (with probability 0.60) and $70,000 otherwise. you expect will never be repaid? What percentage
If you build a single-family house, the payoff is do you expect?
$100,000 (with probability 0.60) and $60,000 other- b. Find the usual measure of the level of uncertainty in
wise. Finally, you could build a gambling casino the number of loans you authorized that will never
which would pay very well—$500,000—but with be repaid. Briefly interpret this number.
a probability of just 0.10 since the final government c. Find the usual measure of the level of uncertainty in
permits are not likely to be granted; all will be lost the percentage of loans you authorized that will
otherwise. never be repaid. Briefly interpret this number.
a. Find the expected payoff for each of these four 12. Your company is planning to market a new reading lamp
projects. In terms of just the expected payoff, rank and has segmented the market into three groups—avid
these projects in order from best to worst. readers, regular readers, and occasional readers—and
b. Find the standard deviation for each of these four currently assumes that 25% of avid readers, 15% of
projects. In terms of risk only, rank the projects from regular readers, and 10% of occasional readers will
best to worst. want to buy the new product. As part of a marketing
c. Considering both the expected payoff and the risk survey, 400 individuals will be randomly selected
involved, can any project or projects be eliminated from the population of regular readers. Using the cur-
from consideration entirely? rent assumptions, find the mean and standard deviation
d. How would you decide among the remaining of the percentage among those surveyed who will
projects? In particular, does any single project dom- want to buy the new product.
inate the others completely? 13. A company is conducting a survey of 235 people to
10. Your quality control manager has identified the four measure the level of interest in a new product. Assume
major problems, the extent to which each one occurs that the probability of a randomly selected person’s
(i.e., the probability that this problem occurs per item being “very interested” is 0.88 and that people are
produced), and the cost of reworking to fix each one selected independently of one another.
(see Table 7.6.5). Assume that only one problem can a. Find the standard deviation of the percentage who
occur at a time. will be found by the survey to be very interested.
a. Compute the expected rework cost for each problem b. How much uncertainty is there in the number of
separately. For example, the expected rework cost people who will be found to be very interested?
for “broken case” is 0.04 × 6.88. Compare the c. Find the expected number of people in the sample
results and indicate the most serious problem in who will say that they are very interested.
terms of expected dollar costs. d. Find the expected percentage that the survey will
b. Find the overall expected rework cost due to all four identify as being very interested.
problems together. 14. An election coming up next week promises to be very
c. Find the standard deviation of rework cost (don’t close. In fact, assume that 50% are in favor and 50%
forget the nonreworked items). are against. Suppose you conduct a poll of 791 ran-
d. Write a brief memo, as if to your supervisor, describ- domly selected likely voters. Approximately how differ-
ing and analyzing the situation. ent will the percent in favor (from the poll) be from the
50% in the population you are trying to estimate?
15. Repeat the previous problem, but now assume that 85%
are in favor in the population. Is the uncertainty larger or
smaller than when 50% was assumed? Why?
TABLE 7.6.5 Quality Control Problems: Type, 16. You have just performed a survey interviewing 358 ran-
Extent, and Cost domly selected people. You found that 94 of them are
interested in possibly purchasing a new cable TV ser-
Problem Probability Rework Cost
vice. How much uncertainty is there in this number
Broken case 0.04 $6.88 “94” as compared to the average number you would
expect to find in such a survey? (You may assume that
Faulty electronics 0.02 12.30
exactly 25% of all people you might have interviewed
Missing connector 0.06 0.75 would have been interested.)
17. You are planning to make sales calls at eight firms today.
Blemish 0.01 2.92
As a rough approximation, you figure that each call has
a 20% chance of resulting in a sale and that firms make
Chapter | 7 Random Variables 183

their buying decisions without consulting each other. 21. Find the probability that you will see moderate
Find the probability of having a really terrible day with improvement in productivity, meaning an increase in
no sales at all. productivity between 6 and 13. You may assume that
18. It’s been a bad day for the market, with 80% of securities the productivity increase follows a normal distribution
losing value. You are evaluating a portfolio of 15 secu- with a mean of 10 and a standard deviation of 7.
rities and will assume a binomial distribution for the 22. Under usual conditions, a distillation unit in a refinery
number of securities that lost value. can process a mean of 135,000 barrels per day of
a.* What assumptions are being made when you use a crude petroleum, with a standard deviation of 6,000 bar-
binomial distribution in this way? rels per day. You may assume a normal distribution.
b.* How many securities in your portfolio would you a. Find the probability that more than 135,000 barrels
expect to lose value? will be produced on a given day.
c.* What is the standard deviation of the number of b. Find the probability that more than 130,000 barrels
securities in your portfolio that lose value? will be produced on a given day.
d.* Find the probability that all 15 securities lose value. c. Find the probability that more than 150,000 barrels
e.* Find the probability that exactly 10 securities lose will be produced on a given day.
value. d. Find the probability that less than 125,000 barrels
f. Find the probability that 13 or more securities lose will be produced on a given day.
value. e. Find the probability that less than 100,000 barrels
19. Your firm has decided to interview a random sample of will be produced on a given day.
10 customers in order to determine whether or not to 23. The quality control section of a purchasing contract for
change a consumer product. Your main competitor valves specifies that the diameter must be between
has already done a similar but much larger study and 2.53 and 2.57 centimeters. Assume that the production
has concluded that exactly 86% of consumers approve equipment is set so that the mean diameter is 2.56 cen-
of the change. Unfortunately, your firm does not have timeters and the standard deviation is 0.01 centimeter.
access to this information (but you may use this figure What percent of valves produced, over the long run,
in your computations here). will be within these specifications, assuming a normal
a. What is the name of the probability distribution of distribution?
the number of consumers who will approve of the 24. Assume that the stock market closed at 13,246 points
change in your study? today. Tomorrow you expect the market to rise a mean
b. What is the expected number of people, out of the of 4 points, with a standard deviation of 115 points.
10 you will interview, who will approve of the Assume a normal distribution.
change? a. Find the probability that the stock market goes
c. What is the standard deviation of the number of down tomorrow.
people, out of the 10 you will interview, who will b. Find the probability that the market goes up more
approve of the change? than 50 points tomorrow.
d. What is the expected percentage of people, out of c. Find the probability that the market goes up more
the 10 you will interview, who will approve of the than 100 points tomorrow.
change? d. Find the probability that the market goes down
e. What is the standard deviation of the percentage of more than 150 points tomorrow.
people, out of the 10 you will interview, who will e. Find the probability that the market changes by
approve of the change? more than 200 points in either direction.
f. What is the probability that exactly eight of your 25. Based on recent experience, you expect this Saturday’s
interviewed customers will approve of the change? total receipts to have a mean of $2,353.25 and a
g. What is the probability that eight or more of standard deviation of $291.63 and to be normally
your interviewed customers will approve of the distributed.
change? a. Find the probability of a typical Saturday, defined as
20. Suppose that the number of hits on your company’s total receipts between $2,000 and $2,500.
website, from noon to 1 p.m. on a typical weekday, fol- b. Find the probability of a terrific Saturday, defined as
lows a normal distribution (approximately) with a mean total receipts over $2,500.
of 190 and a standard deviation of 24. c. Find the probability of a mediocre Saturday, defined
a. Find the probability that the number of hits is more as total receipts less than $2,000.
than 160. 26. The amount of ore (in tons) in a segment of a mine is
b. Find the probability that the number of hits is less assumed to follow a normal distribution with mean
than 215. 185 and standard deviation 40. Find the probability
c. Find the probability that the number of hits is that the amount of ore is less than 175 tons.
between 165 and 195. 27. You are a farmer about to harvest your crop. To describe
d. Find the probability that the number of hits is not the uncertainty in the size of the harvest, you feel that it
between 150 and 225. may be described as a normal distribution with a mean
184 PART | II Probability

value of 80,000 bushels and a standard deviation of and you have projected a probability of 0.53 that a
2,500 bushels. Find the probability that your harvest typical individual will vote to strike.
will exceed 84,000 bushels. a. Identify n and π for this binomial random variable.
28. Assume that electronic microchip operating speeds are b. Find the mean and standard deviation of the num-
normally distributed with a mean of 2.5 gigahertz and a ber who will vote to strike.
standard deviation of 0.4 gigahertz. What percentage of c. Find the (approximate) probability that a strike will
your production would you expect to be “superchips” result (i.e., that a majority will vote to strike).
with operating speeds of 3 gigahertz or more? 33. Reconsider the previous problem and answer each part,
29. Although you don’t know the exact total amount of pay- but assume that 1,000 people will vote. (The probability
ments you will receive next month, based on past for each one remains unchanged.)
experience you believe it will be approximately 34. Assume that if you were to interview the entire popula-
$2,500 more or less than $13,000, and will follow a nor- tion of Detroit, exactly 18.6% would say that they are
mal distribution. Find the probability that you will ready to buy your product. You plan to interview a
receive between $10,000 and $15,000 next month. representative random sample of 250 people. Find the
30. A new project will be declared “successful” if you (approximate) probability that your observed sample
achieve a market share of 10% or more in the next percentage is overoptimistic, where this is defined as
two years. Your marketing department has considered the observed percentage exceeding 22.5%.
all possibilities and decided that it expects the product 35. Suppose 15% of the items in a large warehouse are
to attain a market share of 12% in this time. However, defective. You have chosen a random sample of 250
this number is not certain. The standard deviation is items to examine in detail. Find the (approximate)
forecast to be 3%, indicating the uncertainty in the probability that more than 20% of the sample is
12% forecast as 3 percentage points. You may assume defective.
a normal distribution. 36. You are planning to interview 350 consumers randomly
a.* Find the probability that the new project is selected from a large list of likely sales prospects, in
successful. order to assess the value of this list and whether you
b. Find the probability that the new project fails. should assign salespeople the task of contacting them
c. Find the probability that the new project is wildly all. Assuming that 13% of the large list will respond
successful, defined as achieving at least a 15% mar- favorably, find (approximate) probabilities for the
ket share. following:
d. To assess the precision of the marketing projections, a. More than 10% of randomly selected consumers
find the probability that the attained market share will respond favorably.
falls close to the projected value of 12%, that is, b. More than 13% of randomly selected consumers
between 11% and 13%. will respond favorably.
31. A manufacturing process produces semiconductor chips c. More than 15% of randomly selected consumers
with a known failure rate of 6.3%. Assume that chip fail- will respond favorably.
ures are independent of one another. You will be pro- d. Between 10% and 15% of randomly selected con-
ducing 2,000 chips tomorrow. sumers will respond favorably.
a. What is the name of the probability distribution of 37. You have just sent out a test mailing of a catalog to
the number of defective chips produced tomorrow? 1,000 people randomly selected from a database of
b. Find the expected number of defective chips 12,320 addresses. You will go ahead with the mass
produced. mailing to the remaining 11,320 addresses provided
c. Find the standard deviation of the number of defec- you receive orders from 2.7% or more from the test mail-
tive chips. ing within two weeks. Find the (approximate) probability
d. Find the (approximate) probability that you will pro- that you will do the mass mailing under each of the fol-
duce fewer than 130 defects. lowing scenarios:
e. Find the (approximate) probability that you will pro- a. Assume that, in reality, exactly 2% of the population
duce more than 120 defects. would send in an order within two weeks.
f. You just learned that you will need to ship 1,860 b. Assume that, in reality, exactly 3% of the population
working chips out of tomorrow’s production of would send in an order within two weeks.
2,000. What are the chances that you will succeed? c. Assume that, in reality, exactly 4% of the population
Will you need to increase the scheduled number would send in an order within two weeks.
produced? 38. You expect a mean of 1,671 warranty repairs next
g. If you schedule 2,100 chips for production, what is month, with the actual outcome following a Poisson
the probability that you will be able to ship 1,860 distribution.
working ones? a. Find the standard deviation of the number of such
32. A union strike vote is scheduled tomorrow, and it looks repairs.
close. Assume that the number of votes to strike follows b. Find the (approximate) probability of more than
a binomial distribution. You expect 300 people to vote, 1,700 such repairs.
Chapter | 7 Random Variables 185

39. If tomorrow is a typical day, your human resources 45. Assuming the appropriate probability distribution for the
division will expect to receive résumés from 175 job situation described in the preceding problem:
applicants. You may assume that applicants act inde- a. Find the probability that the system will last 100,000
pendently of one another. hours or more (twice the average lifetime).
a. What is the name of the probability distribution of b. The system is guaranteed to last at least 5,000 hours.
the number of résumés received? What percentage of production is expected to fail
b. What is the standard deviation of the number of during the guarantee period?
résumés received? 46. Compare the “probability of being within one standard
c. Find the approximate probability that you will deviation of the mean” for the exponential and normal
receive more than 185 résumés. distributions.
d. Find the approximate probability of a slow day, with
160 or fewer résumés received.
Database Exercises
40. On a typical day, your clothing store takes care of 2.6
“special customers” on average. These customers are Problems marked with an asterisk (*) are solved in the Self
taken directly to a special room in the back, are assigned Test in Appendix C.
a full-time server, are given tea (or espresso) and scones, Refer to the employee database in Appendix A.
and have clothes brought to them. You may assume that
1. View each column as a collection of independent obser-
the number who will arrive tomorrow follows a Poisson
vations of a random variable.
distribution.
a. In each case, what kind of variable is represented,
a. Find the standard deviation of the number of special
continuous or discrete? Why?
customers.
b.* Consider the event “annual salary is above
b. Find the probability that no special customers arrive
$40,000.” Find the value of the binomial random
tomorrow.
variable X that represents the number of times this
c. Find the probability exactly 4 special customers will
event occurred. Also find the binomial proportion
arrive tomorrow.
p and say what it represents.
41. In order to earn enough to pay your firm’s debt this year,
c. What fraction of employees are male? Interpret
you will need to be awarded at least 2 contracts. This is
this number as a binomial proportion. What is n?
not usually a problem, since the yearly average is 5.1
2. You have a position open and are trying to hire a new per-
contracts. You may assume a Poisson distribution.
son. Assume that the new person’s experience will follow
a. Find the probability that you will not earn enough to
a normal distribution with the mean and (sample) stan-
pay your firm’s debt this year.
dard deviation of your current employees.
b. Find the probability that you will be awarded
a. Find the probability that the new person will have
exactly 3 contracts.
more than six years of experience.
42. Customers arrive at random times, with an exponential
b. Find the probability that the new person will have
distribution for the time between arrivals. Currently the
less than three years of experience.
mean time between customers is 6.34 minutes.
c. Find the probability that the new person will have
a. Since the last customer arrived, three minutes have
between four and seven years of experience.
gone by. Find the mean time until the next customer
3. Suppose males and females are equally likely and that the
arrives.
number of each gender follows a binomial distribution.
b. Since the last customer arrived, 10 minutes have gone
(Note that the database contains observations of random
by. Find the mean time until the next customer arrives.
variables, not the random variables themselves.)
43. In the situation described in the previous problem, a cus-
a. Find n and π for the binomial distribution of the
tomer has just arrived.
number of males.
a. Find the probability that the time until the arrival of
b. Find n and π for the binomial distribution of the
the next customer is less than 3 minutes.
number of females.
b. Find the probability that the time until the arrival of
c. Find the observed value of X for the number of
the next customer is more than 10 minutes.
females.
c. Find the probability that the time until the arrival of
d. Use the normal approximation to the binomial dis-
the next customer is between 5 and 6 minutes.
tribution to find the probability of observing this
44. A TV system is expected to last for 50,000 hours before
many females (your answer to part c) or fewer in
failure. Assume an exponential distribution for the time
the database.
until failure.
a. Is the distribution skewed or symmetric?
b. What is the standard deviation of the length of time Projects
until failure?
c. The system has been working continuously for the 1. Choose a continuous random quantity that you might
past 8,500 hours and is still on. What is the deal with in your current or future work as an executive.
expected time from now until failure? (Be careful!) Model it as a normally distributed random variable and
186 PART | II Probability

estimate (i.e., guess) the mean and standard deviation. years inspires you to do a What if scenario analysis, recog-
Identify three events of interest to you relating to this ran- nizing that there is no obligation to extract the oil and that
dom variable and compute their probabilities. Briefly it could be extracted fairly quickly (taking a few months) at
discuss what you have learned. any time during the three-year lease. You are wondering:
2. Choose a discrete random quantity (taking on from 3 to What if the price of oil rises enough during the three years
10 different values) that you might deal with in your cur- for it to be profitable to develop the oil field? If so, then
rent or future work as an executive. Estimate (i.e., guess) you would extract the oil. But if the price of oil didn’t rise
the probability distribution. Compute the mean and stan- enough, you would let the term of the lease expire in three
dard deviation. Identify two events of interest to you years, leaving the oil still in the ground. You would let the
relating to this random variable and compute their prob- future price of oil determine whether or not to exercise the
abilities. Briefly discuss what you have learned. option to extract the oil.
3. Choose a binomial random quantity that you might deal But such a proposition is risky! How much risk? What are
with in your current or future work as an executive. the potential rewards? You have identified the following
Estimate (i.e., guess) the value of n and π. Compute the basic probability structure for the source of uncertainty in
mean and standard deviation. Identify two events of this situation:
interest to you relating to this random variable and com-
pute their probabilities. Briefly discuss what you have Future Price of Oil Probability
learned. 60 0.10
4. On the Internet, find and record observations on at least 70 0.15
five different random variables such as stock market 80 0.20
indices, interest rates, corporate sales, or any business- 90 0.30
related topic of interest to you. 100 0.15
110 0.10

Case
The Option Value of an Oil Lease Discussion Questions
There’s an oil leasing opportunity that looks too good to be 1. How much money would you make if there were no
true, and it probably is too good to be true: An estimated costs of extraction? Would this be enough to retire?
1,500,000 barrels of oil sitting underground that can be 2. Would you indeed lose money if you leased and
leased for three years for just $1,300,000. It looks like a extracted immediately, considering the costs of extrac-
golden opportunity: Pay just over a million, bring the oil to tion? How much money?
the surface, sell it at the current spot price of $76.45 per bar- 3. Continue the scenario analysis by computing the future
rel, and retire. net payoff implied by each of the future prices of oil.
However, upon closer investigation, you come across the To do this, multiply the price of oil by the number of bar-
facts that explain why nobody else has snapped up this rels; then subtract the cost of extraction. If this is nega-
“opportunity.” Evidently, it is difficult to remove the oil tive, you simply won’t develop the field, so change
from the ground due to the geology and the remote location. negative values to zero. (At this point, do not subtract
A careful analysis shows that estimated costs of extracting the lease cost, because we are assuming that it has
the oil are a whopping $120,000,000. You conclude that already been paid.)
by developing this oil field, you would actually lose 4. Find the average future net payoff, less the cost of the
money. Oh well. lease. How much, on average, would you gain (or
During the next week, although you are busy investigat- lose) by leasing this oil field? (You may ignore the time
ing other capital investment opportunities, your thoughts value of money.)
keep returning to this particular project. In particular, the 5. How risky is this proposition?
fact that the lease is so cheap and that it lasts for three 6. Should you lease or not?