PROBABILITY AND STATISTICS

WITH APPLICATIONS:
A PROBLEM SOLVING TEXT
SECOND EDITION

For Marilyn, - L.A.

      

ACTEX Learning
Greenland, NH
Copyright  2015, 2023 by ACTEX Learning

All rights reserved. No portion of this book May be reproduced

in any form or by any means Without the prior written
permission of the Copyright owner.

Requests for permission should be addressed to

ACTEX Learning
PO Box 69
Greenland, NH 03840

Manufactured in the United States of America

10 9 8 7 6 5 4 3 2 1

Cover design by Jeff Melaragno

Library of Congress Cataloging-in-Publication Data

Asimow, Leonard A., 1939-

Probability and statistics with applications : a problem solving text / Leonard A. Asimow,
Mark M. Maxwell. -- Second Edition.
pages cm
Revised edition of the authors' Probability and statistics with applications, 2010.
Includes bibliographical references and index.
Summary: "Probability and Statistics with Applications is an introductory textbook designed
to make the subject accessible to college freshmen and sophomores concurrent with their
study of calculus"-- Provided by publisher.
ISBN 978-1-62542-472-3 (alk. paper)
1. Casualty Actuarial Society--Examinations--Study guides. 2. Mathematical statistics--
Examinations, questions, etc. 3. Insurance--Statistics. I. Maxwell, Mark M., 1967- II. Title.
HG8045.A85 2015
519.5024'368--dc23
2015010222

ISBN: 978-1-62542-472-3
 PREFACE

Preface to the Second Edition

This edition contains a number of new topics, primarily in the mathematical statistics portion
of the text. We continue to cover all of the material necessary for the SOA Exam P
(Probability) in Chapters 1-8. We have expanded the second part of the book, now Chapters
9-12, to provide full coverage of the syllabus for the Casualty Actuarial Society Exam ST,
and the statistics portion of the new CAS Exam S.

New sections include explanations and worked examples on the following topics:

 Mixture distributions,
 Non-homogeneous Poisson processes,
 Sufficient statistical estimators and the linear exponential family,
 Bayesian analysis and conjugate prior distributions,
 Nonparametric statistical methods,
 Graphical methods.

As in the first edition, our aim is to provide the best qualities of both a standard textbook and
an actuarial exam study manual. We attempt to provide full explanations and derivations or
proofs wherever feasible for an introductory text. We also provide a multitude of examples
and problems, including many from earlier actuarial exams. Space limitations prevent us
from providing complete solutions to all of the exercises in the body of the text. However, a
complete solutions manual is available as a separate volume.

We have benefited from much assistance from reviewers and editors for this edition. In
particular we express our appreciation to Ali Ishaq, Tom Lonergan, Rajesh Sahasrabuddhe
and Emiliano Valdez, who read all or portions of the manuscript and provided valuable
suggestions.

We are indebted to David Hudak for his assistance with the new material on non-
homogeneous Poisson processes.

We also gratefully acknowledge the detailed editorial review provided by Geoff Tims.
Finally, we extend heartfelt thanks to Stephen Camilli and Garrett Doherty of ACTEX, the
former for his editorial guidance and encouragement, the latter for his formidable technical
skills in producing the final manuscript pages.

L.A.

iii
iv  PREFACE

Preface to the First Edition

A cursory search of Google Books reveals thousands of titles with the words Probability and
Statistics, or Mathematical Statistics, thus prompting the inevitable question, why yet
another? In view of this superabundance of choices we feel impelled to offer a few words of
explanation in our defense for adding to this already crowded marketplace.

The idea for this textbook evolved out of a two-semester course in probability and statistics
we have been offering for many years to sophomore and qualified freshman students,
predominantly actuarial science majors. The goals of our course are twofold: to lay the
foundations of calculus-based probability theory for our students, and to prepare them to pass
the actuarial exam covering this material as early as possible in their collegiate careers.

The primary market segment we are targeting consists of students like ours – freshmen and
sophomores – who are studying calculus-based probability while simultaneously learning the
selfsame calculus it is based upon. We desire that our actuarial students be prepared to pass
Exam P/1 (jointly offered by the Society of Actuaries of Actuaries (SOA) and the Casualty
Actuarial Society (CAS)) no later than the end of their sophomore year. Consequently, the
probability and statistics component of their education tends to overlap with the typical 3-
semester calculus sequence they all take.

Our experience has been that the myriad existing textbooks in probability and statistics fall
into two types. They are either designed to support a “calculus-free” environment suitable for
general business school statistics courses, or they are intended for more advanced, calculus-
based mathematical statistics courses for juniors and seniors with the requisite technical
background. The first category does not provide the depth of understanding required for the
actuarial exam, and the second type of book tends to be too formal and advanced for our
students.

For these reasons we decided to produce an introductory text in calculus-based probability

and statistics whose level is comparable to a modern-day calculus book, and that could
reasonably be used by freshman or sophomore students studying the material concurrently
with their calculus classes. We consciously strive to pace the material in a way that makes it
accessible to a student whose background consists of just one semester of college-level
calculus. This might be, for example, either entering freshmen with AP or high-
school/college credit concurrently taking Calculus II, or sophomores with one or two
semesters of calculus already under their belts.

Chapters 1-4 present the rudiments of probability theory for discrete distributions with little
or no reference to calculus topics, save for the basic knowledge of infinite series required for
understanding the geometric and the Poisson distributions. Chapter 5, entitled Calculus and
Probability, introduces continuous random variables and is the first chapter heavily
dependent on derivatives and integrals. The material on continuous, jointly distributed
random variables comes in Chapter 7, by which time our students will have been introduced
to double integrals in their Calculus III class. The second part of the book, comprising
Chapters 9-11, covers all of the syllabus topics for the statistics portion of CAS Exam 3L –
Life Contingencies and Statistics Segment. Taken as a whole, this book provides ample
 PREFACE  V

content to serve as the text for the standard two-semester introductory sequence in
mathematical statistics and probability.

The text contains nearly 800 exercises, many with multiple parts. Numerical answers are
given in the back of the book and a supplementary manual with complete solutions is
available separately. Many of the exercises, and some examples, are based on previous
actuarial exam questions released by the SOA and the CAS. All of the SOA/CAS Exam P/1
Sample Exam Questions (142 in number at press time in late 2009) have been incorporated
into the text. In addition, we have used many statistics questions from CAS Exam 3 and 3L,
as well as questions from the earlier Exam 110 (Probability and Statistics), in Chapters 8-11.
We are grateful to the Society of Actuaries and the Casualty Actuarial Society for their
permission to use these materials.

While designed primarily with the actuarial audience in mind, we hope this text will have
appeal for a broader audience of mathematics and statistics students. We have sought to
engage students with a light, informal style, emphasizing detailed explanations and providing
a multitude of examples. At the same time, we have sought at each stage to present a
sufficient glimpse into the theoretical underpinnings to make the text suitable for more
advanced students. The overarching leitmotif, however, is problem solving, and we
emphasize the requisite skills throughout. We hope to have a struck a balance that will allow
students at all levels to benefit from a close reading of the text.

We have benefited from the many helpful comments and valuable insights provided by our
reviewers: Carolyn Cuff, Westminster College; Thomas Herzog, Department of Housing and
Urban Development; Thomas Lonergan, CIGNA; Jeffrey Mitchell, Robert Morris University;
Emiliano Valdez, University of Connecticut; Charles Vinsonhaler, University of Connecticut.
We would also like to acknowledge the invaluable editorial assistance provided by Gail Hall
of ACTEX, whose firm but gentle hand managed to guide this project to completion. We are
also indebted to Marilyn Baleshiski, whose patience and skill rendered a lumpy manuscript
into a finished text. Our thanks to you all. Needless to say, for all remaining errors and
indiscretions, the authors have only each other to blame.

Finally, we note that personages (both named and unnamed) who appear in various exercises
and examples are completely fictitious and any resemblance to real people (living or
departed) is purely coincidental.

Len Asimow, Moon Township, PA

Mark Maxwell, Austin, TX
 TABLE OF CONTENTS

CHAPTER 1 COMBINATORIAL PROBABILITY 1

1.1 The Probability Model ....................................................................................................1

1.2 Finite Discrete Models with Equally Likely Outcomes ...................................................5
1.2.1 Tree Diagrams ......................................................................................................6
1.2.2 The Multiplication Principle ................................................................................8
1.2.3 Permutations .........................................................................................................9
1.2.4 The Birthday Problem and its Genre ..................................................................12
1.2.5 Combinations ......................................................................................................13
1.2.6 Partitions .............................................................................................................16
1.3 Sampling and Distribution .............................................................................................23
1.3.1 Sampling .............................................................................................................23
1.3.2 Distributions........................................................................................................25
1.3.3 Sampling and Occupancy United........................................................................29
1.4 More Applications ..........................................................................................................29
1.4.1 The Binomial and Multinomial Theorems ..........................................................29
1.4.2 Poker Hands ........................................................................................................33
1.4.3 The Powerball® Lottery .....................................................................................37
1.5 Chapter 1 Sample Examination ......................................................................................41

CHAPTER 2 GENERAL RULES OF PROBABILITY 43

2.1 Sets, Sample Spaces, and Events ...................................................................................43

2.2 Set Theory Survival Kit..................................................................................................44
2.2.1 The Venn Diagram..............................................................................................46
2.2.2 Basic Rules of Probability ..................................................................................47
2.2.3 DeMorgan’s Laws...............................................................................................49
2.2.4 The Venn Box Diagram ......................................................................................51
2.3 Conditional Probability ..................................................................................................56
2.3.1 Conditional Probability and Tree Diagrams .......................................................58
2.3.2 Bayesian “Cause and Effect” Reasoning ............................................................60
2.4 Independence..................................................................................................................67
2.5 Bayes’ Theorem .............................................................................................................75
2.6 Credibility ......................................................................................................................77
2.7 Chapter 2 Sample Examination ......................................................................................81
vii
viii  TABLE OF CONTENTS

CHAPTER 3 DISCRETE RANDOM VARIABLES 89

3.1 Definition and Examples of Discrete Random Variables...............................................89

3.2 Cumulative Probability Distribution ..............................................................................92
3.3 Measures of Central Tendency .......................................................................................94
3.3.1 Expected Value (Mean) ......................................................................................95
3.3.2 Median of a Data Set ..........................................................................................99
3.3.3 Midrange of a Data Set .....................................................................................100
3.3.4 Mode of a Data Set ...........................................................................................100
3.3.5 Quartiles and Percentiles ..................................................................................101
3.3.6 Random Variables and Percentiles ...................................................................102
3.4 Measures of Dispersion ................................................................................................104
3.4.1 Range, Inter-Quartile Range .............................................................................105
3.4.2 Variance ............................................................................................................106
3.4.3 Standard Deviation ...........................................................................................109
3.4.4 Standardized Random Variables and Z-Scores.................................................110
3.4.5 Chebychev’s Theorem ......................................................................................113
3.4.6 Coefficient of Variation ....................................................................................115
3.5 Conditional Expectation and Variance .........................................................................116
3.6 Jointly Distributed Random Variables (Round 1) ........................................................119
3.7 The Probability Generating Function ...........................................................................126
3.8 Chapter 3 Sample Examination ....................................................................................128

CHAPTER 4 SOME DISCRETE DISTRIBUTIONS 133

4.1 The Discrete Uniform Distribution ..............................................................................133

4.2 Bernoulli Trials and the Binomial Distribution ............................................................137
4.3 The Geometric Distribution .........................................................................................143
4.4 The Negative Binomial Distribution ............................................................................149
4.5 The Hyper-geometric Distribution ...............................................................................152
4.6 The Poisson Distribution ..............................................................................................156
4.6.1 The Poisson Probability Function .....................................................................156
4.6.2 Poisson Processes .............................................................................................159
4.6.3 Poisson Process Data Sets.................................................................................160
4.6.4 Sums of Poisson Random Variables .................................................................167
4.6.5 Poisson Approximation to the Binomial Distribution.......................................168
4.7 Summary Comparison of Discrete Distributions .........................................................169
4.8 Chapter 4 Sample Examination ....................................................................................170
 TABLE OF CONTENTS  ix

CHAPTER 5 CALCULUS, PROBABILITY, AND CONTINUOUS DISTRIBUTIONS 175

5.1 Cumulative Distribution Functions ..............................................................................175

5.2 Density Functions.........................................................................................................179
5.3 Great Expectations .......................................................................................................188
5.3.1 The Variance Formula ......................................................................................191
5.3.2 The Mode of a Continuous Distribution ...........................................................193
5.3.3 Medians and Percentiles ...................................................................................195
5.3.4 Calculating the Expected Value of X with the CDF .........................................197
5.4 Mixed Distributions......................................................................................................201
5.5 Applications to Insurance: Deductibles and Caps ........................................................206
5.5.1 Deductible Insurance ........................................................................................206
5.5.2 Capped Insurance ..............................................................................................208
5.5.3 The CDF Method for Deductibles and Caps.....................................................211
5.6 The Moment Generating Function ...............................................................................219
5.6.1 The MGF for the Binomial Distribution ...........................................................226
5.6.2 The MGF for the Geometric and Negative Binomial .......................................227
5.6.3 The MGF for the Poisson Distribution .............................................................228
5.7 Chapter 5 Sample Examination ....................................................................................229

CHAPTER 6 SOME CONTINUOUS DISTRIBUTIONS 233

6.1 Uniform Random Variables .........................................................................................233

6.2 The Exponential Distribution .......................................................................................238
6.2.1 Integration Review ............................................................................................238
6.2.2 Exponential and Poisson Relationship ..............................................................239
6.2.3 Properties of the Exponential Random Variable ..............................................240
6.3 The Normal Distribution ..............................................................................................246
6.3.1 Integration Review ............................................................................................247
6.3.2 The Standard Normal ........................................................................................248
6.3.3 The General Normal Distribution .....................................................................253
6.3.4 The Lognormal Distribution .............................................................................257
6.4 The Law of Averages and the Central Limit Theorem.................................................260
6.4.1 Random Samples ..............................................................................................260
6.4.2 Chebyshev and the Law of Averages................................................................262
6.4.3 Central Limit Theorem (Sum Version) .............................................................265
6.4.4 The Continuity Correction ................................................................................267
6.4.5 Central Limit Theorem (Sample Mean Version) ..............................................270
6.4.6 Outline for a Proof of the Central Limit Theorem ............................................274
x  TABLE OF CONTENTS

6.5 The Gamma Distribution ..............................................................................................277

6.5.1 The Gamma Function .......................................................................................277
6.5.2 Definition and Properties for the Gamma Family .............................................278
6.5.3 Comparing the Gamma and Lognormal ...........................................................282
6.5.4 Gamma and Poisson Connections.....................................................................283
6.5.5 The Distribution of Z 2 .....................................................................................286
6.6 The Beta Family of Distributions .................................................................................288
6.6.1 The Beta Function .............................................................................................288
6.6.2 The Beta Family of Distributions .....................................................................290
6.7 More Continuous Distributions ....................................................................................293
6.7.1 The Hazard Rate ...............................................................................................293
6.7.2 The Pareto Distribution .....................................................................................296
6.7.3 The Weibull Distribution ..................................................................................300
6.8 Chapter 6 Sample Examination ...................................................................................301

CHAPTER 7 MULTIVARIATE DISTRIBUTIONS 305

7.1 Joint Distributions for Discrete Random Variables .....................................................305

7.2 Conditional Distributions – The Discrete Case ............................................................314
7.3 Independence – Discrete Case......................................................................................318
7.4 Covariance and Correlation ..........................................................................................321
7.5 Joint Distributions for Continuous Random Variables ................................................328
7.6 Conditional Distributions – The Continuous Case .......................................................346
7.7 Independence and Covariance in the Continuous Case ................................................355
7.8 The Multinomial Distribution ......................................................................................362
7.9 Bivariate Normal Distributions ....................................................................................363
7.10 Moment Generating Function for a Joint Distribution .................................................368
7.11 Chapter 7 Sample Examination ....................................................................................371

CHAPTER 8 A PROBABILITY POTPOURRI 377

8.1 The Distribution of a Transformed Random Variable .................................................377

8.1.1 The Transformation Formula ............................................................................379
8.1.2 The Distribution of the Sum of Random Variables ..........................................385
8.1.3 The Convolution Integral ..................................................................................389
8.1.4 Simulating Values of the Random Variable X ..................................................393
8.1.5 Order Statistics..................................................................................................399
8.2 The Moment-Generating Function Method .................................................................409
 TABLE OF CONTENTS  xi

8.3 Covariance Formulas....................................................................................................411

8.3.1 Covariance and the Multinomial Distribution ..................................................413
8.3.2 Variance Formula for the Hyper-geometric ......................................................415
8.4 The Conditioning Formulas and Mixtures ...................................................................417
8.5 Poisson Processes Revisited .........................................................................................431
8.5.1 Non-homogenous Poisson Process ...................................................................431
8.5.2 The Poisson Thinning Process ..........................................................................436
8.6 Chapter 8 Sample Examination ...................................................................................442

CHAPTER 9 SAMPLING DISTRIBUTIONS AND ESTIMATION 447

9.1 The Sample Mean as an Estimator ...............................................................................448

9.2 Estimating the Population Variance .............................................................................453
9.2.1 The Sample Variance ........................................................................................453
9.2.2 Sample Variance from a Normal Population ....................................................456
9.2.3 Confidence Interval for Population Variance ...................................................460
9.3 The Student t-Distribution ............................................................................................464
9.4 The F-Distribution........................................................................................................469
9.5 Estimating Proportions .................................................................................................474
9.6 Estimating the Difference between Means ...................................................................476
9.6.1 Large Samples...................................................................................................477
9.6.2 Small Samples...................................................................................................478
9.7 Estimating the Sample Size ..........................................................................................480
9.8 Chapter 9 Sample Examination ....................................................................................482

CHAPTER 10 HYPOTHESIS TESTING 487

10.1 Hypothesis Testing Framework ...................................................................................487

10.1.1 Finding the Likelihood of Type I and Type II Errors......................................489
10.1.2 The Significance Level of a Test.....................................................................495
10.1.3 The p-value of a Hypothesis Test ....................................................................500
10.2 Hypothesis Testing for Population Means ...................................................................502
10.2.1 Standard Normal Tests for the Population Mean ............................................503
10.2.2 Student t Tests for the Population Mean .........................................................507
10.3 Hypothesis Testing for Population Variance................................................................509
10.4 Hypothesis Testing for Proportions..............................................................................512
xii  TABLE OF CONTENTS

10.5 Hypothesis Testing for Differences in Population Means............................................514

10.5.1 Large Samples .................................................................................................514
10.5.2 Small Samples .................................................................................................516
10.5.3 The F Test for Equal Population Variances ....................................................520
10.6 Chi-Square Tests ..........................................................................................................522
10.6.1 Contingency Tables .........................................................................................523
10.6.2 Goodness of Fit Tests ......................................................................................527
10.7 Chapter 10 Sample Examination ..................................................................................532

CHAPTER 11 THEORY OF ESTIMATION AND HYPOTHESIS TESTING 537

11.1 The Bias of an Estimator ..............................................................................................538

11.2 Building Estimators ......................................................................................................541
11.2.1 Method of Moments..........................................................................................541
11.2.2 The Maximum Likelihood Estimator................................................................546
11.3 Properties of Estimators ...............................................................................................557
11.3.1 Consistent Estimators .......................................................................................558
11.3.2 Efficient Estimators ..........................................................................................560
11.3.3 Efficiency and the Cramer-Rao Inequality .......................................................561
11.3.4 Sufficient Statistics ...........................................................................................570
11.4 Hypothesis Testing Theory ..........................................................................................580
11.5 More General Likelihood Ratio Tests ..........................................................................594
11.6 Bayesian Estimation .....................................................................................................601
11.6.1 The Bayesian Recipe ........................................................................................602
11.6.2 The Loss Function and Mean Square Error ......................................................608
11.6.3 Conjugate Prior Distributions and Credible Intervals .......................................614
11.7 Chapter 11 Sample Examination ..................................................................................624

CHAPTER 12 A STATISTICS POTPOURRI 631

12.1 Simple Linear Regression: Basic Formulas .................................................................631

12.1.1 The Least Squares Model .................................................................................632
12.1.2 The Analysis of Variance Perspective ..............................................................644
12.2 Simple Linear Regression: Estimation of Parameters ..................................................650
12.3 Comparing Means Using ANOVA ..............................................................................665
12.4 Non-parametric Tests ...................................................................................................673
12.4.1 Sign Tests..........................................................................................................673
12.4.2 Runs Tests .........................................................................................................675
12.4.3 Signed Rank Tests ............................................................................................680
12.4.4 Mann-Whitney-Wilcoxon U Statistic ...............................................................683
12.4.5 Rank Correlation ...............................................................................................685
 TABLE OF CONTENTS  xiii

12.5 Goodness of Fit: Several Eyeball Tests ........................................................................690

12.6 Chapter 12 Sample Examination ..................................................................................698

APPENDIX I STATISTICAL TABLES 707

ANSWERS TO TEXT EXERCISES 713

INDEX 743
 CHAPTER 1

COMBINATORIAL PROBABILITY

1.1 THE PROBABILITY MODEL

The fundamental object of interest in this text is the probability model. Probability models
are based on experiments, such as flipping a coin or tossing a pair of dice, for which there
are multiple possible outcomes. It is impossible to say in advance which outcome will occur.
In this sense a probability model constitutes an apt metaphor for life, where we rarely know
the ultimate results, or future outcomes, of our actions. This is in contrast to deterministic
models, such as a well-designed chemistry experiment, where one knows in advance the
outcome with a high degree of certainty (a bad odor will arise, a small explosion will occur,
a penny will completely dissolve in 60 seconds, etc.).

A probability model has three essential elements. These three elements can be defined as
follows:

The Elements of a Probability Model

i. The description of the underlying probability experiment,

ii. A list, or a procedure for listing, all possible outcomes of the experiment,
iii. A rule for assigning numbers between 0 (impossible) and 1 (certain) to
subsets of outcomes.

The set of all outcomes on the list in (ii) is called the sample space. Subsets of the sample
space are called events. The numbers in (iii) are called probabilities, and they measure the
likelihood of an outcome along a scale from 0% to 100%. These numbers must satisfy one
other property, denoted below as the additive property, which we will consider in more
detail as we continue.

The Additive Property

iii . The “sum” of the probabilities of all the outcomes must equal one.

“Sum” is written in quotation marks because the summing process may be more complicated
than simple addition, possibly involving an infinite series or a definite integral. As a
consequence of the additive property, it seems sensible (and it is correct) to define the
1
2  CHAPTER 1

probability of an event as the “sum” of the probabilities of all the outcomes contained in the
event. The additive property assures that the event consisting of the entire sample space has
probability one. That is, it is certain that one of the outcomes listed in (ii) will occur.

Example 1.1-1 Elements of a Probability Model

The description of an experiment could be as simple as “toss a coin once.” A complete list of
outcomes would be the set consisting of heads or tails. A simple rule for assigning
probabilities is 12 for heads and 12 for tails. Notation that summarizes these statements is
sample space  {heads, tails}, Pr( heads )  12 , and Pr(tails )  12 .

In this case the experiment, the outcomes, and the probability assignments constitute an
accurate model of the real-world activity of tossing a balanced (fair) coin. In life there is no
such thing as a perfectly balanced coin, and there is always the remote possibility of a third
outcome (e.g., coin stands on edge). But since we are constructing an idealized mathematical
model of an experiment, we keep things as simple as possible while preserving the essential
structure of the real world phenomena we are modeling.

Example 1.1-2 Sample Space

Suppose that a couple conducts the experiment of having three children. Their concern is the
gender of the children.

a. List the sample space of all possible outcomes.

b. List the event that a boy is the first born child.

Solution
The sample space can be represented by the set {BBB, BBG , BGB, BGG, GBB, GBG, GGB, GGG},
where B denotes the birth of a boy and G denotes the birth of a girl. Therefore the outcome
BGG denotes a boy being the first born, followed by the births of two girls. The event of a
boy being the first born consists of the subset {BBB, BBG , BGB, BGG}.

Here, the set of outcomes was identified as a list of words. By words in this text we mean
any systematic grouping of letters or numbers used to describe outcomes. In this example,
the outcomes consist of all 3 letter words (not just dictionary words) consisting of the letters
B and G exclusively. Note that the words are systematically listed in alphabetical order,
which is not only convenient, but helps to assure that we don’t inadvertently omit an
outcome. 

If all experiments were as simple as tossing a single coin, or having three children (which is
conceptually simple, although more complicated in actual practice1), then you wouldn’t need
this text. Needless to say, probability models can be quite complex, with a very large (or
even infinite) number of outcomes.

1 This comment was written by Len, father of two.

 COMBINATORIAL PROBABILITY  3

There are a number of different ways of classifying probability models, but the most basic is
the distinction between discrete probability models and continuous probability models. In
discrete probability models, the outcomes can be enumerated in a list, as outcome(1),
outcome(2), etc. This list can be finite or countably infinite. An example of the latter would
be the experiment of tossing a coin until the first head appeared. One way of listing this
sample space would be,
H , TH , TTH , TTTH ,
Alternatively and equivalently, we could describe the sample space as {1, 2,3, 4,...}, where
the natural number stands for the coin toss on which the first head occurs. If this were a fair
coin, then an appropriate assignment of probabilities would be Pr( H )  Pr(1)  .5,
Pr(TH )  Pr(2)  .25, Pr(TTH )  Pr(3)  .125, and so forth. In Chapter 4 we will verify that
the infinite sum of these probabilities equals one.

A continuous model has outcomes along a continuum. A description of such an experiment

might be, “toss a dart at the unit interval [0,1]. ” The set of outcomes consists of all real
numbers between 0 and 1. For those readers impertinently protesting that one can’t toss a
dart at a one-dimensional line in the real world, we can reformulate the experiment as “spin
a wheel-of-fortune” whose edge is calibrated with the unit interval. Here, the fixed pointer
plays the role of the dart, and we are now “throwing the unit interval at the dart,” so to
speak. In either case, the mathematical model is the same. The real difficulty here is in
assigning the probability numbers. Since the model is continuous, it should not be surprising
that we require the tools of calculus to do this properly. Essentially, probabilities are
assigned to intervals, not single points, and the summing of the probabilities referred to in
additive property ( iii ) requires integration. In this example, the probability of the pointer
landing in the interval [a, b] is its length, b  a. We will leave the details until Chapter 5.

Finally, it must be emphasized that the way in which the list in (ii) is constructed can be
extremely important.

Example 1.1-3 Sample Space and Assigning Probabilities

Consider the experiment of rolling a pair of fair dice.

(a) List all outcomes in the sample space.

(b) Assign probabilities to the outcomes.
(c) Find the probability that the sum of the dice is 8.

Solution
Generally, when we roll a pair of dice the outcome of real interest is the sum. In particular,
we would like to model the real-life probability of the outcome sum equals 8. Our list of all
outcomes could therefore be the eleven integers from 2 to 12. That leaves the issue of
assigning probabilities. Since this is a mathematical model we can assign probabilities any
way we like so long as the probabilities sum to one. In particular, we could arbitrarily assign
equally likely outcomes so that the probability of a sum equal to 8 would be the same as the
4  CHAPTER 1

probability of a sum equal to 12 or any other outcome, namely 1/11. However, this model
will not conform to our experience gained over the many years of playing with dice.

We will therefore take a different approach to listing the outcomes. We first imagine in our
mathematical model that we can distinguish between the two dice (assume that one is green
and the other is grey). We then list all outcomes as two-letter words with the letters being the
integers from 1 to 6), with the green die result followed by the grey die:

For example, the word (14) represents the green die (left) showing one dot and the
grey die (right) showing four dots.

11 21 31 41 51 61
12 22 32 42 52 62
13 23 33 43 53 63
14 24 34 44 54 64
15 25 35 45 55 65
16 26 36 46 56 66

In this second method of listing, it is reasonable to assume that the outcomes are all equally
likely. This is because we assume that the individual die gives equally likely outcomes from
1 to 6 (such dice are called fair and balanced - like the ideal newscast), and therefore the
pairs are also equally likely to occur. In this second method, there are a total of 5 outcomes
out of 36 with a sum equal to 8. Thus, “sum of 8” is no longer an outcome, but an event. If
the 36 words are all assigned the probability of 1/36, then the probability of the event sum of
5 , which conforms to actual experience. Even though there are more outcomes (36
8 is 36
versus 11) to deal with, the individual outcomes have the advantage of being naturally
equally likely. This provides a mathematical model that conforms more readily to reality.

In general it is better to list outcomes in a way that preserves as much information as

possible about the underlying experiment, and which takes advantage of an intuitively
natural assignment of equally likely outcomes. 

Exercise 1-1 At a picnic, you select a main course from a hamburger or a chicken
sandwich. You select a side from potato chips or coleslaw, and you select a beverage
from soda, water, or milk. These 3 selections result in a meal.
(a) Write out the sample space of all meals.
(b) Write out the event (subset) consisting of all meals with soda as the beverage.
(c) Assuming all meals are equally likely, calculate the probability of getting soda with
your meal.

Exercise 1-2 Roll a pair of fair six-sided dice.

(a) How many ways can doubles (both die have the same number of dots) be rolled?
(b) What is the probability that the sum of the two dice is a prime number?
 COMBINATORIAL PROBABILITY  5

Exercise 1-3 A convenience store has three packages of plain M&M’S®,2 two packages
of peanut M&M’S, one package of dark chocolate M&M’S, one package of peanut
butter M&M’S, and one package of almond M&M’S. Select two packages of candy.

(a) Write out the sample space of all possible outcomes.

(b) Assume the outcomes are equally likely. Calculate the probability you select exactly
one package of plain M&M’S, (and the other package is not plain – either peanut,
dark chocolate, peanut butter, or almond).
(c) Calculate the probability of getting at least one package of plain M&M’S.

1.2 FINITE DISCRETE MODELS WITH EQUALLY LIKELY OUTCOMES

In this section we consider in more detail the wide class of probability models in which there
is only a finite number of possible outcomes. Probabilities are assigned, as in the dice
example in the previous section, so that all outcomes are equally likely. This situation of
equally likely outcomes governs many of the typical probability problems that arise in
modeling games (drawing cards, tossing dice, flipping coins, etc.).

In the case of equally likely outcomes, the assignment of probabilities is straightforward.

Since the sum of the probabilities of outcomes must equal one, and the outcomes are equally
likely, each outcome has probability 1n , where n is the total number of possible outcomes.

Probability with Equally Likely Outcomes

Let S be a sample space consisting of n equally likely outcomes. Let E be an event

in S consisting of r outcomes. Then the probability of event E, denoted Pr( E ), is

Pr( E )  number of outcomes in event E  r  N ( E ) ,

total number of possible outcomes n N ( S )

where N ( E ) denotes the number of elements in the set E.

Determining the value of n or r is often the most complicated part. This leads us to a
discussion of various counting techniques (called combinatorics) such as permutations and
combinations.

2 ®/™ M&M’S is a registered trademark of Mars, Incorporated and its affiliates. Mars, Incorporated is not

associated with ACTEX Publications, Inc. The M&M’S® trademark is used with permission of Mars,
Incorporated. ©Mars, Inc. 2010.
6  CHAPTER 1

1.2.1 TREE DIAGRAMS

A tree diagram is a fundamental conceptual tool for constructing lists of outcomes.

It is also a valuable approach for tracking probabilities in a multistage experiment.

Example 1.2-1 Tree Diagram

A family has two children. Construct a tree diagram showing all possible combinations of
boys and girls.

Solution
boy

girl
boy

girl

first child boy

girl

second child

There are four distinct possible outcomes, each corresponding to exactly one path through
the diagram. Traveling across the top path, for example, denotes the birth of two boys. Using
the diagram, we can write the sample space as {BB, BG , GB, GG} where B denotes the birth
of a boy and G denotes the birth of a girl.

Example 1.2-2 Tree Diagram of Your Wardrobe

Suppose that your wardrobe consists of two types of shoes (Birkenstocks and Nikes), three
different shirts (baseball jersey, polo, tee-shirt), and two types of pants (shorts and Levi
button-fly 501 jeans).

(a) Construct a tree diagram showing all possible outfits that you could wear.
(b) Assuming all outfits are equally likely, calculate the probability of choosing an outfit
that includes a polo shirt.
 COMBINATORIAL PROBABILITY  7

Solution

(a) shorts
Levis
baseball jersey
shorts
polo
Levis Outfit is: Birkenstock
footwear, polo shirt, and
t-shirt
Birkenstocks shorts Levis jeans.

Levis

shorts
Levis
baseball jersey
Nikes
polo shorts
Levis
t-shirt

shorts
Levis

(b) The diagram consists of twelve distinct branches, each corresponding to a distinct outfit.
Four of these branches include a polo shirt, so that the probability is 4/12 = 1/3. This is,
of course, the same as the probability of choosing a polo shirt from among the 3 possible
shirts available. 

In these examples, each outcome can be represented by a particular path through the tree. In
the first example there are two possible results (boy or girl) for the gender of the first child,
and each of these leads to two possible results for the gender of the second child. This leads
to a total of 2  2  4 paths (outcomes) through the tree. In the second example we have 2
possible pairs of shoes, and for each of these there are 3 possible shirts. For each choice of
shoes and shirt, there are 2 choices for pants. The total number of paths through the tree is
2  3 2  12.

The general rule at work here is called the multiplication principle.

8  CHAPTER 1

1.2.2 THE MULTIPLICATION PRINCIPLE

The Multiplication Principle

Suppose an experiment can be broken down into a first stage A consisting of

N ( A) outcomes and that for each of these outcomes, there is a second stage B
consisting of N ( B ) outcomes. Then the total number of outcomes for the two
stages combined is equal to N ( A)  N ( B ).

Note: The word and is as indicator that one should multiply. The rule can be
extended to three or more stages.

This principle is commonly called the fundamental theorem of counting.

Example 1.2-3 The Multiplication Principle

For the wardrobe described in Example 1.2-2, use the fundamental theorem of counting to
determine how many different outfits you could wear.

Solution
We used a tree diagram to list the sample space of all of the possible outfits. In this problem,
our concern is only the number of possible outfits we have, so we use the fundamental
theorem of counting.

2  3  2
 12.


Example 1.2-4 The Multiplication Principle

Assume standard license plates in the state of Oregon are comprised of three letters followed
by three numbers. How many different license plates could theoretically be fabricated?

Solution
The total number of standard license plates is 26  26  26  10  10  10  263  103  17,576,000.
Using a tree diagram to list all possible outcomes from AAA000 through ZZZ999 would
require more space than is available here.

26 26 26 10 10 10

O R E G O N
 COMBINATORIAL PROBABILITY  9

Exercise 1-4 Your current wardrobe consists of eleven different items: three pairs of
shoes, two pairs of pants, and six different shirts. Your mom offered to buy you a new
shirt or a new pair of shoes. Which item (shoes or shirt) should your mom buy for you to
maximize your total number of outfits (shoes-shirt-pant)?

Exercise 1-5 Roll three fair six-sided dice.

(a) Assuming the dice are distinguishable and always tossed in a particular order (e.g.,
red-white-blue), how many different outcomes are possible?
(b) How many outcomes are there in which the first two dice come up 3 and 4 respec-
tively?
(c) What is the probability that the first two dice come up 3 and 4 respectively?
(d) Would you like to use a tree diagram to list all of the possible outcomes?

Exercise 1-6 Assuming there are no restrictions, how many different 9-digit social
security numbers are possible?

Exercise 1-7 Assume that a standard California license plate consists of a digit, 3
letters, and then 3 more digits (e.g., 1SAM123), except that I, O, and Q are excluded
from the first and third alpha positions. How many possible standard license plates are
there in California?

Exercise 1-8 A thief plans to rob four banks. His town of Podunk, Idaho has six banks.
He could rob a bank twice, but never consecutively. In how many ways may he select
the four banks?

Exercise 1-9 How many ten-digit phone numbers are possible in the United States?
Note: Creating a mathematical model that accurately reflects the real world is extremely
important and frequently difficult. Rules that govern the problem should be understood
and listed. In this case, we assume that the first digit cannot be 0 or 1 and the fourth digit
cannot be 0.

1.2.3 PERMUTATIONS

Permutations

Given a set of n distinguishable objects, an ordered selection of r different

elements of the set is called a permutation of n objects chosen r at a time.
10  CHAPTER 1

Example 1.2-5 Permutations

In May you travel to Louisville and make a trifecta wager on the Kentucky Derby. A trifecta
bet consists of selecting the first three finishers in order. That is, the one that finishes first
(win), the one that finishes second (place), and the one that finishes third (show).. How many
different trifecta wagers are possible if 14 horses go to post? Rephrased mathematically, how
many permutations are there of 14 objects (distinguishable horses) chosen 3 at a time?

Solution
First we note that any of the 14 horses could win. But the winning horse cannot also finish
second, so there remain 13 horses that could be selected for second. Similarly, there remain
12 horses that can be selected for third. Using the multiplication principle directly we see the
number of distinct trifecta wagers is,

14 13 12


Win Place Show

Factorials
Let n be a whole number. Then n! (read as “n factorial”) is defined by

n !  n  (n 1)  (n  2)    3  2  1.

By convention, we define 0!  1.

Alternatively, we can express the solution to Example 1.2-5 using factorial notation.

Using factorial notation we can easily generalize from the horserace example to a formula
for the number of permutations.

Permutation Formula
The number of permutations on n objects chosen r at a time is denoted by n Pr .

n Pr  n( n 1)( n  2) ( n  r 1)  n ! .

   (n  r )!
first r factors of n !

Note: The (n  r )! in the denominator cancels out the “tail” of n!, leaving just the
first r factors.
Note: Other common ways to denote permutations: n Pr  P ( n, r )  Prn .

This formula results from applying the multiplication principle to the r stage process of
picking the r objects, one at a time and in order.
 COMBINATORIAL PROBABILITY  11

Note
 n!  n!  n! is simply the number of distinct ways of rearranging (choosing a
n Pn
(n  n)! 0!
different ordering) all n objects.

The solution to the horserace problem can be rephrased as the number of permutations on 14
horses chosen 3 at a time,

 14!  14!  14  13  12  2,184.

14 P3   
(14 3)! 11!
win place show

We use permutations when the order of selection is important.

Exercise 1-10 Find the numerical values of the following permutations:

(a) 7 P6 (b) 13 P12 (c) 7 P4
(d) 5 P2 (e) 4 P4 (f) 4 P6
(g) 23 P3 (h) 3 P0 (i) 5 P5

7 P4
(j) 10 P1 (k) (l) 0 P0
4!

Exercise 1-11 An engineering faculty member shuffles the quizzes of his 17 students and
redistributes them randomly for peer grading.

(a) In how many different ways may the quizzes be handed out?
(b) In how many ways may Kendra, one of the 17 students, be assigned her own quiz
for grading?
(c) What is the probability that Kendra will receive her own quiz?

Exercise 1-12 Peter Pickemfast spends a day at the races and decides to play the trifecta in
an eight horse field. Knowing nothing about any of the horses, he chooses horses
randomly and buys one ticket. Calculate the probability he has the winning trifecta ticket.

Exercise 1-13 Henry Hoodlum speaks to two of the jockeys in an 8 horse field, politely
requesting they not finish in the top 3. Assuming they comply, how many trifecta tickets
must Henry buy to assure he holds a winning ticket?

Exercise 1-14 A psychology professor has twelve students in class. She insults one
student, throws an eraser at a second student, lowers the grade of a third student, and
sends a fourth student to the dean’s office. In how many different ways can she perform
these motivational techniques?
12  CHAPTER 1

1.2.4 THE BIRTHDAY PROBLEM AND ITS GENRE

There is a classic problem in probability that asks, “What are the chances that two people in
this room share the same birthday (month and day only)?” The answer, of course, depends
on how many people are in the room, and how many possibilities there are for birthdays. To
keep things simple we assume that there are exactly 365 equally likely birthdays. Thus, if
you happen to have been born on February 29, you don’t count. We could let you celebrate
on February 28, but then we would no longer have equally likely outcomes. It is easier to
just assume you do not exist in our simplified model. You still exist in the real world and our
model aspires to be a simple approximation to reality. We are also aware that birthdays are
not equally likely throughout the year, but dealing with this peculiarity would further
complicate our model.

We will illustrate the use of the multiplication rule and permutations to answer the question in a
general format. Assume there are n distinguishable objects (“birthdays”) and we wish to assign
them to r “people.” We will calculate the probability of the event, E, that there is no duplication
of birthdays. The event that there exists a duplication of birthdays is called the complement of E
and is denoted E . Its probability must equal one minus the probability of the event E.

The sample space, S, consists of all possible ways of distributing n “birthdays” to r “people.”
This is an r stage process with n ways to choose at each stage. Thus, the size of the sample
space equals n r (the multiplication principle in action). Now, the event E consists of all
outcomes in which there are no duplications, that is, permutations of the n “birthdays”
chosen r at a time. Therefore, the event, E, consists of n Pr outcomes. Thus, the probability
of no duplications is given by

n Pr
Pr[ E ]  Pr[unique birthdays]  r
,
n
and the probability of at least one duplication is given by,

n Pr
Pr[ E ]  1  r
.
n

Example 1.2-6 Birthday Problem Probability

Calculate the probability that in a class of 25 students at least one pair of students will have
the same birthday.

Solution
From the second formula, the answer is

365 P25
Pr  at least one shared birthday   1  25
 0.57. 
365

Exercise 1-15 Find the smallest number of students for which the probability of a
birthday duplication is at least ½.
 COMBINATORIAL PROBABILITY  13

Exercise 1-16 What is the smallest number of people in a room to assure that the
probability that at least two were born on the same day of the week is at least 40%?

Exercise 1-17 Seven people enter the elevator on the first floor of a 12 story building.
What is the probability that no two will get off on the same floor? You may assume that
all floors are equally likely and no one gets off the elevator before it starts up.

Exercise 1-18 The church of Individual Self-Actualization has a hymnal consisting of 35

songs. Twelve members come in and open their hymnals to a random song and
commence singing (all songs are equally likely). What is the probability that no two
congregants are on the same song?

Exercise 1-19 Each individual in a group of n students is asked to pick an integer at

random between 1 and 10 (inclusive). What is the smallest value of n that assures at least
a 50% chance that at least two students select the same integer?

1.2.5 COMBINATIONS

We now turn to situations in which the order of selection doesn’t matter. It only matters if an
object is selected or not selected. Think of a poker hand. It does not matter the order that you
are dealt all four aces and a fifth card; what matters is that you were dealt all of the aces.
(Advice: Bet big, this happens infrequently).

Combinations
Given a set of n distinguishable objects, an unordered selection of r different
elements of the set is called a combination of n objects chosen r at a time and is
denoted by n C r and read as n choose r.

Combinations apply when order does not matter. In determining whether to use
combinations or permutations, it is helpful to ask, “Is this like the outcomes of a
horserace where order matters?”, or “Is this like choosing pizza toppings where
order does not matter?”

Example 1.2-7 Combinations

Papa Your Mommas Pizza Parlor has the following four pizza toppings: pepperoni,
mushroom, sausage, and olives.

(a) How many different pizzas with two distinct toppings could be ordered?
(b) List the sample space for pizzas with two distinct toppings.
(c) If all two-topping pizzas are equally likely to be ordered, find the probability that
sausage is on the pizza.
14  CHAPTER 1

Solution
The order in which Papa places toppings on your pie does not matter. That is, unless you are
hopelessly obsessive-compulsive (same as compulsive-obsessive), a pepperoni and mushroom
pizza is identical to a mushroom and pepperoni pizza.

(a) Using permutations, we calculate the number of distinct pizzas as 4 P2 and then
43
divide by the 2! possible orderings of the 2 toppings. Thus, 4 C2  21  6.
(b) The six pizzas are pepperoni-mushroom, pepperoni-sausage, pepperoni-olive,
mushroom-sausage, mushroom-olive, and sausage-olive. Shorthand notation for our
sample space is Sample Space  {PM , PS , PO, MS , MO, SO}.
(c) Three pizzas have sausage: pepperoni-sausage, mushroom-sausage, and sausage-
olive. Therefore Pr(sausage on the 2 topping pizza)  3  0.5. 
6

The formula for the number n C r of combinations can be deduced from the multiplication
principle, applied to a two-step process for listing the n Pr ordered outcomes.

(1) Choose r objects without regard to order ( n C r ways),

(2) Choose a particular ordering (permutation) of the r objects ( r Pr  r ! ways).

Then, n Pr is the product resulting from steps (1) and (2). That is, n Pr  n!  C  r !.
(n  r )! n r
Solving for n C r leads to the following formula.

Combination Formula
The number of combinations of r objects chosen from a collection of n
distinguishable objects is given by,
first r factors of n ! on top
 
P
n r n! n  ( n  1)  ( n  2) (n  r 1)
n Cr   
r! r !(n  r )! r  (r 1)  (r  2) 3  2  1
  
all r factors of r ! on bottom

Note: Other common ways to denote combinations: n Cr 

r 
n  C (n, r )  C n
r

The form n Cr  n

r is especially common and is referred to as the binomial coefficient. The
reason for this will be explained in Subsection 1.4.1.
 COMBINATORIAL PROBABILITY  15

Exercise 1-20 Find the numerical values of the following combinations:

(a) C6
7 (b) C12
13 (c) C1
13

(d) 5 C2 (e) 4 C4 (f) 2 C3

(g) 5 C3 (h) 4 C0 (i) 5 C5
7 P4
(j) 5867 C1 (k) 4! (l) 0 C0

Exercise 1-21 Verify that 12 C 3  12 C 9 .

Exercise 1-22 Verify the relationship n C r  n C n  r . In terms of pizza toppings, explain

this relationship.

Exercise 1-23 A father of 6 children comes home to a disheveled house. His slovenly
children are to blame. Since he does not know who is responsible, he will select four of
the six children for clean-up duty. In how many ways can he do this?
Note: 6 C 4  15  6 C 2 , since selecting four children to clean up is equivalent to
selecting two children to spare from chores.

Exercise 1-24 Papa Your Mommas Pizza Parlor has 6 meat toppings and 7 vegetable
toppings from which to select. The parlor has three different sizes of pizza (individual,
large, and giant) and two different types of crust (deep-dish and thin).
(a) How many different two-topping pizzas could be ordered?
(b) How many different two-topping pizzas could be ordered with exactly one meat
topping and exactly one vegetable topping?
(c) How many different four-topping vegetarian pizzas could be ordered?

Exercise 1-25 Coach Cramer has 15 basketball players: four centers, five forwards, and
six guards.
(a) How many ways can she select five players to form her starting line-up, regardless
of position?
(b) How many ways can she select five players to form her starting line-up if she needs
one center, two forwards, and two guards?

Exercise 1-26 Your local Ice Cream Shoppe has eleven flavors of ice cream, three types
of cones (waffle, chocolate, and sugar), and five types of toppings (broken peanuts,
whipped cream, hot fudge, maraschino cherries, and sprinkles). How many different
two-scoop cones can be ordered that have two different flavors of ice cream, exactly one
topping, and your choice of cone?
16  CHAPTER 1

Exercise 1-27 A young man has a collection of fourteen earrings, three nose rings, seven
necklaces, and five hemp bracelets. How many different groupings of jewelry may he
wear if he selects three earrings, one nose ring, two necklaces, and one bracelet?

Exercise 1-28 A high school lottery uses two sets of numbered balls. One set consists of
ten white balls numbered 1-10 and the second set contains twenty blue balls numbered
1-20. To play, you select two white balls and two blue balls.
(a) How many different outcomes are possible?
(b) Your lottery ticket consists of four numbers: two white numbers, each between 1
and 10 inclusive, and two blue numbers, each between 1 and 20, inclusive. What is
the probability that your lottery ticket contains exactly one matching white number
and two matching blue numbers?

Exercise 1-29 At a picnic, there was a bowl of chocolate candy that had 10 pieces each
of Milky Way®, Almond Joy®, Butterfinger®, Nestle Crunch®, Snickers®, and Kit
Kat®. Jen grabbed six pieces at random from this bowl of 60 chocolate candies.
(a) What is the probability that she got one of each variety?
(b) What is the probability that Jen grabs exactly five varieties?

1.2.6 PARTITIONS

Whenever we select a combination of r objects from n distinguishable objects, we are in

effect dividing the n objects into two classes, those that are “in” and those that are “out.” For
example, in selecting 5 starters from 15 players the coach probably just reads off the names of
the 5 who are “in.” Alternatively though, she could convey the same information by reading
off the names of the 10 “out” players, telling them to go sit on the bench (admittedly, an
approach that emphasizes the negative and probably wouldn’t do much for morale.)

Selecting a combination of objects in effect partitions the n objects in two sets, one of size r
(the in set), the other of size n  r (the out set). This is another explanation for the
relationship n C r  n C n  r (see Exercise 1-22).

Many combinatorial problems involve partitioning the population (the n distinguishable

objects) into more than two subsets. The sizes of the subsets are always prescribed in advance.

Partitions
Let A be a set consisting of n distinguishable objects. Let whole numbers
{r1 , r2 , , rk } be given such that r1  r2    rk  n. A partition of A into subsets
of sizes {r1 , r2 , , rk } is a particular distribution of the n objects into disjoint
subsets A1 , A2 , , Ak of sizes r1 , r2 , , rk respectively.
 COMBINATORIAL PROBABILITY  17

Notes
k
(1) Since the Ai are disjoint and r1  r2    rk  n, it follows that A   Ai .
i 1
(2) Once objects are assigned to one of the subsets, their order within the subset is
irrelevant.
(3) As we saw above, a combination of r objects is a partition of A into subsets of size r
(the in set) and size n  r (the out set).

Example 1.2-8 Partitions

An intramural basketball team has five highly versatile starters (Amy, Brandy, Chrystal,
Diane, and Erin) each capable of playing any one of the three positions of center, forward,
and guard. How many different starting lineups consisting of one center, two forwards and
two guards are possible? We do not distinguish here between power forward and small
forward, or point guard and shooting guard.

Solution
A lineup consists of one center, two forwards and two guards. Thus, the solution consists of
calculating the number of partitions of 5 objects (the starters) into 3 subsets (center,
forwards, guards) of sizes 1, 2, and 2 respectively. Any of the five players could play at
center, so there are 5 C1  5 choices for center. Of the four remaining players, we must
choose two to play at forward – and there are 4 C 2  6 ways to accomplish this. The
remaining two players must be assigned to play guard – there are 2 C 2  1 ways to do this
(the one way is to have both remaining ladies play as guards). By the multiplication
principle, the total number of line-ups is 5 C1  4 C 2  2 C 2  5  6  1  30.

Since the numbers are manageable, we can try listing all the partitions.

Position Center Forwards Guards

Size 1 2 2
Set #1 #2 #3
Lineup (1) Amy Brandy, Chrystal Diane, Erin
Lineup (2) Amy Brandy, Diane Chrystal, Erin
Lineup (3) Amy Brandy, Erin Chrystal, Diane
Lineup (4) Amy Chrystal, Diane Brandy, Erin
   
Lineup (30) Erin Chrystal, Diane Amy, Brandy

The enterprising student should complete the list of all 30 partitions. The partial listing
above is in alphabetical order. Note that, since the ordering of the forwards and guards
doesn’t matter, each pair is listed in alphabetical order within position for definiteness. 

Systematically listing all possible outcomes is an extremely useful tool. We have already
used systematic listing to find the sample spaces in a variety of instances, including: families
with three children, flipping coins until a head occurs, rolling a pair of fair dice, constructing
wardrobes, and constructing basketball line-ups.
18  CHAPTER 1

Example 1.2-9 Several Partitions of a Deck of Cards

Suppose the set under consideration is a standard deck of playing cards, consisting of 52
distinguishable objects (no jokers, missing cards, or wild cards). Dividing the deck into four
suits (clubs, diamonds, hearts, and spades) is a particular partition into 4 subsets, all of size 13.

In the card game bridge, the deck is dealt (divided randomly) into 4 hands (sets, often
labeled North, East, South and West) each of size 13. Thus, a bridge hand is one of many
possible partitions of the deck into 4 subsets of size 13.

Another special partition of a standard deck would be into the thirteen denominations, or
ranks of cards (ace, king, queen, …, 2), a partition into 13 subsets, each of size 4.

Partitions of cards need not be of the same size – like in Uno®, gin rummy, and Go Fish. If
all 52 cards are dealt to three players, the sets might be partitioned into three hands of size
18, 17, and 17. 

Example 1.2-10 Partitions

How many different bridge games are possible?

In other words, how many distinct partitions of 52 objects into 4 subsets of size 13 are
possible?

Sample calculations of this type will lead us to a general formula for counting the number of
possible partitions. We can break the problem down into a four step process and employ the
multiplication principle.

Solution
Step 1: How many ways to choose the cards for North?
Answer: 52 C13  52!  635,013,559,600.
(13!)(39!)

Step 2: How many ways to choose the cards for South?

Answer (since there are just 39 cards left for South):
39!  8,122,425,444 .
39 C13 
(13!)(26!)

Step 3: How many ways to choose the cards for East?

Answer: 26 C13  26!  10,400,600.
(13!)(13!)

Step 4: How many ways to choose the cards for West?

Answer (note that after Steps 1-3 there are only 13 cards left):

 13!  1.
13 C13
(13!)(0!)
 COMBINATORIAL PROBABILITY  19

Thus, the total number of bridge hands is given by

 39 C13  26 C13  13 C13  52!  39!  26!  13!

52 C13
(13!)(39!) (13!)(26!) (13!)(13!) (13!)(0!)
 52!
(13!)(13!)(13!)(13!)
 (635,013,559,600)  (8,122,425,444)  (10,400,600)  (1)
 53,644,737,765,488,792,839,237,440,000,

which is one heck of a lot of bridge hands. 

Example 1.2-11 Partitions of Students

There are a total of nine available seats in three different sections of MATH101. MATH101
section A (MATH101-A) has 3 available seats, MATH101-B has 2 available seats, and
MATH101-C has 4 available seats. There are 9 students awaiting assignment to the
remaining 9 open slots in MATH101. How many different assignments of students to
sections are possible?

Solution
Since the order of the students selected for a given section does not matter, we can view this
as the number of ways of partitioning 9 distinguishable objects (students) into sets of size 3,
2, and 4. Thus,

Total ways equals 9 C3  6 C2  4 C4  9!  6!  4!  9!  1260. 

(3!)(6!) (4!)(2!) (4!)(0!) (3!)(2!)(4!)

In general the solution to a partitioning problem is called a multinomial coefficient. The

examples above can be generalized to the multinomial formula:

Multinomial Coefficients

The number of partitions of n distinct objects into k subsets of sizes r1 , r2 , , rk ,

where r1  r2    rk  n is called a multinomial coefficient, denoted by

 r , r ,n, r . The resulting formula is:

1 2 k

 r , r ,n, r   r !r n!! r !.
1 2 k 1 2 k

Proof
First, using the multiplication principle as in the preceding examples, we have,

 r , r ,n, r    rn   n r r   n  rr  r  rr  .
1 2 k 1 2
1 1
3
2 k
k
20  CHAPTER 1

Thus,

 r , r ,n, r    r ! nn! r !  r !nnrr !r !  r !nnrr rr !r ! 
1 2 k 1 1 2 1
1

2 3 1
1 2

2 3

  this equals rk

  
 (n  r1    rk 1 )! 
 
 rk !0! 
 
 

n!
  n  r1 !   n  r1  r2 !   rk ! 
      
 r1 !  n  r1 !   r2 !  n  r1  r2    r3 ! n  r1  r2  r3 !   rk ! 0! 
   

 n! , where the first several cancellations are noted above.

r1 !r2 !r3 ! rk !

Example 1.2-12 Multinomial Coefficients

The women’s basketball team consists of 15 highly versatile players, each capable of playing
all positions. How many different lineups consisting of one center, two forwards, and two
guards are possible? Note: Chrystal playing guard results in a different lineup from Chrystal
playing forward.

Solution 1
We can break this up into a two step process in which the first step is to choose the five
starting players. The second step is to partition the five players into subsets of sizes 1, 2, and
2. Using the multiplication principle, the number of different lineups is

5
15  
1, 2, 2 
5  15!  5!  (3003)(30)  90,090.
5!10! 1!2!2!
ways to among 5 starters,
select 5 starters ways to assign positions

Solution 2
Alternatively, we can select one student to start at center. There are  15!  15 ways 15 C1
14! 1!
to do this. Then we select two of the remaining 14 players to play guard, 14 C2  14!  91
12! 2!
and two women to play forward, 12 C2  12!  66. By the multiplication principle, there
10! 2!
are a total of 15 C1  14 C 2  12 C 2  90,090 ways.
 COMBINATORIAL PROBABILITY  21

Solution 3
Using the multinomial coefficient directly, we have that 15 players need to be partitioned
into a group of 1 starting center, 2 starting guards, 2 starting forwards, and 10 players
starting on the bench. There are

1, 2,2,10
15
  1! 2!15! 2!10!  15 14 134 12 11  90,090
ways to do this. 

Note
Does the answer change if the coach selects two starting guards first, then 10 bench players,
then two forwards, and finally the center?

Exercise 1-30 Coach Cramer has 15 basketball players – four centers, 5 forwards, and 6
guards. She starts one center, two guards, and two forwards. How many different groups
of bench-warmers are possible?

Exercise 1-31 Due to budgetary constraints, Coach Cramer has just eight uniforms for
her 15 girls. Conference rules require all players wear a uniform. In how many ways can
the coach select five ladies to start, three ladies to dress as subs, and seven ladies to
remain in street clothes (she no longer cares about filling guard/forward/center, just
ways to start/sub/not dress)?

Using multinomial coefficients allows us to easily calculate the number of different

arrangements possible when some of the objects are indistinguishable. For example, we
might have a row of colored light-bulbs on a score board consisting of several
indistinguishable red bulbs, several more indistinguishable blue bulbs, and so forth. The
question would be, “How many distinct arrangements of the colored bulbs are possible?”

Example 1.2-13 Multinomial Coefficients and Different Words

How many different “words” of length 11 are possible using 4 I’s, 1 M, 2 P’s, and 4 S’s?

Solution
The technique is to think of the eleven distinguishable objects in slots forming our “word.”

I I I I M P P S S S S
1 2 3 4 5 6 7 8 9 10 11

Then the question reduces to partitioning the 11 slots into subsets of sizes 4,1,2,4, where the
subsets are labeled (and populated) using the letters I, M, P, and S. One such partition is:

Subset Label I M P S
Subset Size 4 1 2 4
Slots Assigned 2,5,8,11 1 9,10 3,4,6,7
22  CHAPTER 1

resulting in the particular partition,

Label M I S S I S S I P P I
Slot 1 2 3 4 5 6 7 8 9 10 11

Thus, another formulation of the problem is, “how many different arrangements are possible
using the letters MISSISSIPPI?”

Answer:  4,1,112, 4  4!1!2!4!

11!  34,650. 

Exercise 1-32 How many “words” can be spelled using two O’s, an H and a P? What is
the probability of choosing the word HOOP assuming arrangements are equally likely?

Exercise 1-33 How many words can be spelled using all of the letters in ABCBA? List
them.

Exercise 1-34 How many words can be spelled using all of the letters in Pennsylvania?

Exercise 1-35 Six friends are playing poker. Each person is dealt five cards. In how
many different ways can this be accomplished? That is, how many ways can five cards
each be dealt to six different players in the initial deal of cards? Write your solution in
terms of (a) the multiplication principle and (b) multinomial coefficients.

Exercise 1-36 Twenty-five girls are bored.

(a) In how many ways may five girls be chosen to go to a party and five different girls
be chosen to volunteer at a soup kitchen (and the remaining 15 girls stay home)?
(b) How many ways may the girls form two teams of 5 to play basketball against each
other?

Exercise 1-37 A fraternity consisting of 30 members wants to play “seven versus seven”
flag football. How many different match-ups are possible?

Exercise 1-38 Nine workers are assigned to nine jobs. Two of the jobs are considered
bad, four are considered average, and three of the jobs are considered good. The nine
workers consist of seven men and two women. If workers are randomly assigned to jobs,
calculate the probability that the two women are both assigned to bad jobs.
 COMBINATORIAL PROBABILITY  23

1.3 SAMPLING AND DISTRIBUTION

The best way to master combinatorial (counting) type problems is to completely understand
a few of the basic building-blocks, and to learn how seemingly unrelated problems are in
fact structurally the same. In this section we provide a structure to formally classify certain
combinatorial problems by their particular type.

One common formulation for these types of problems – as we have seen in examples from
earlier sections – involves choosing subsets from a set of n distinguishable objects. In
statistics, these subsets (or events) are called samples. A different way of formulating
problems involves assigning, or distributing, markers to the n objects. It turns out that these
two types of problems, sampling and distribution, are closely related. In fact every sampling
problem can be recast as a distribution problem, and vice-versa.

Mathematicians often couch sampling problems in terms of removing balls from urns. The
related assignment problem would be posed as distributing balls into urns. In this sense, the
distribution of balls into urns is frequently referred to as occupancy. We have never
understood the fascination with balls and urns, but the terminology pervades in the classical
textbooks and is convenient to use in designing examples.

1.3.1 SAMPLING

We begin with sampling problems. Consider a set consisting of n distinguishable objects

(for example, numbered balls in an urn). The set of n objects is called the population.
Assuming all outcomes are equally likely, the probability model boils down to the question
“How many distinct samples of size r can be drawn from the n objects?”

Before we can answer this big question we need the answers to two related questions:

I. Are the samples taken with or without replacement (that is, can we pick an object at
most once, or can we pick the same object more than once)?
II. Does the order in which we select the items in the sample matter?

Example 1.3-1 The Four Different Types of Sampling

For each of the four types of sampling described below, calculate the number of distinct
outcomes.

(1) A club consisting of 26 members needs to select an executive board consisting of a

president, a vice-president, a secretary and a treasurer. Each position has different duties
and responsibilities. No individual can hold more than one office.
(2) A club consisting of 26 members needs to select a delegation of 4 (distinct) members to
attend a convention. The delegation of 4 wears identical goofy hats.
(3) A club consisting of 26 members requires volunteers to complete 4 distinct chores;
sweeping the clubhouse, printing off raffle tickets for the drawing at the party that night,
picking up the prizes, and picking up the empties after the party. The same person can
volunteer for more than one job, and each job requires only one volunteer.
24  CHAPTER 1

(4) A club consisting of 26 members requires volunteers to make 4 identical recruiting

phone calls, to be chosen later from a long list. A member can volunteer for one, two,
three, or all four calls.

We classify these four situations with respect to replacement/without replacement, and order
matters/order doesn’t matter.

Example n r Replacement? Order Matters?

#1 26 4 No Yes
#2 26 4 No No
#3 26 4 Yes Yes
#4 26 4 Yes No
Or,
Without replacement With replacement
Order matters Example 1 Example 3
Order doesn’t matter Example 2 Example 4

In (1) and (2) the context makes clear we want to select four different people, so in both
cases we are sampling without replacement. In (1) the order matters since if Rajulio is
selected it makes a difference as to which office he holds. In (2), order doesn’t matter. We
are only concerned with the members of the delegation, not the order in which they were
selected. In (3) and (4), since members can volunteer for multiple chores, we are sampling
with replacement. In (3) the chores are different, so we need to keep track of who is
volunteering for which job. In (4) the calls are essentially identical, so we are only concerned
with how many calls a volunteer makes.

Each member of the club can be uniquely identified (conveniently, with a single letter of the
alphabet since there are exactly 26 members). The outcomes for each of these four sampling
experiments can be put into one-to-one correspondence with certain types of four letter
“words.” “Words” (in quotes) means, as previously, any list of four letters, not just
dictionary words.

In sampling without replacement, no letter can be repeated in a word. If order matters then
ABCD is different from DCBA. If order does not matter, then we agree to list the four letters
chosen for the sample just once, in alphabetical order. Thus, we can rephrase our four
questions this way:

(1) Calculate the number of four letter words with no duplication of letters.
(2) Calculate the number of four letter words with no duplicate letters and the letters
arranged in alphabetical order.
(3) Calculate the number of four letter words, duplicate letters allowed.
(4) Calculate the number of four letter words, duplicate letters allowed, and the letters
arranged in alphabetical order.
 COMBINATORIAL PROBABILITY  25

Solution
The first three are straightforward to work out using permutations, combinations and the
basic multiplication principle, respectively.

(1) Answer: 26  25  24  23  26 P4  358,800.

president vice-pres secretary treasurer

(2) Answer: 26 C 4  26 P4
4! 4  
 26  14,950.

(3) Answer: 26  26 26  26  26 4  456,976.

sweep raffle prizes empties

Example (4) is a little more subtle and we will defer the solution to the next section where it
will be more easily understood as a distribution problem. 

1.3.2 DISTRIBUTIONS

Each of the four types of sampling experiments described above has a dual formulation as a
distribution, or occupancy, experiment. For this purpose, we let n be the number of
distinguishable (fixed and labeled) urns, and let r be the number of balls to be distributed
into the urns. How many different distributions are possible? Again, before we can answer
the question, we need to know:

I. Can an urn hold at most one ball (exclusive) or can it hold many balls (non-
exclusive)?
II. Are the balls distinguishable (for example, bearing unique numbers) or are they
indistinguishable (like plain white ping-pong balls)?

We can rephrase each of the four sampling problems above as a corresponding distribution
problem:

(1) In how many ways can 4 executive board positions (distinguishable balls) be
distributed among 26 members (urns) with exclusion (since no member can hold
more than one position)?
(2) In how many ways can 4 delegation slots (indistinguishable balls) be distributed
among 26 members (urns) with exclusion?
(3) In how many ways can 4 different jobs (distinguishable balls) be distributed among
26 members (urns) without exclusion (since one member can do multiple jobs)?
(4) In how many ways can 4 identical jobs (indistinguishable balls) be distributed
among 26 members (urns) without exclusion (since one member can do multiple
jobs)?

Statements (1)-(3) were solved in Section 1.3.1 formulated as equivalent sampling problems.
Here is a paradigm for counting in sampling or distribution problems like (4).
26  CHAPTER 1

Imagine the 26 urns lined up in a row sharing common partitions:

A B C  Z

We now distribute the 4 indistinguishable balls among the 26 urns. A sample outcome might
be:
  
A B C  Z

Now, observe that the top row in the above schematic can be thought of as a word consisting
of (26 + 1) lines and 4 circles. However, the leftmost line and the rightmost line remain fixed
in every word and are therefore superfluous. A moment’s thought will show that there is a
one-to-one correspondence between distinct distributions into the 26 slots, and distinct
words consisting of  26  1 lines and 4 circles. The distribution above appears as

 |  |||  |  .

Thus, the question is reduced to, “How many (26  1  4) letter words are there consisting of
four circles and (26  1) vertical lines?” Therefore, the solution to example (4) is

 26 41  4   294   23,751.

The above discussion leads us to the following result:

Samples with Replacement When Order Does Not Matter

The number of unordered samples of r objects, with replacement, from n
distinguishable objects is

n  r 1 Cr   
 n  r 1  n  r 1 .
r n 1 
This is equivalent to the number of ways to distribute r indistinguishable balls into
n distinguishable urns without exclusion.

The same reasoning works in general, so that n  r 1 C r is the solution to the following
sampling and distribution problems:

Sampling: How many distinct unordered samples of size r, with replacement, are there from
n distinguishable objects?

Distribution: How many distinct ways are there to distribute r indistinguishable balls into n
distinguishable urns, with multiple balls in an urn allowed?

We like to think of these types of problems in terms of distributing r dollar bills to n

children. One needs n  1 partitions to separate the children, and then select the locations to
place the r dollars.
 COMBINATORIAL PROBABILITY  27

Example 1.3-2 Samples with Replacement When Order Does Not Matter

Nicole wishes to select a dozen bagels from Unleavened Bread Company. Her choices
include: Asiago cheese, plain, nine grain, cinnamon crunch, and very-very blueberry.

(a) How many different orders of a dozen bagels can she select?
(b) How many different orders of a dozen bagels can she select in which she has at least
one of each kind?

Solution
Selecting a dozen bagels is equivalent to distributing 12 indistinguishable markers (ordering
a bagel) into 5 (bagel) bins. Our answer to (a) is
(5112) C12  16 C12  1,820.
For part (b), imagine that we first select one of each type of bagel. Our problem reduces to
selecting the remaining seven bagels from any the five types. This can be done in
(51 7) C7  11 C7  330 ways. 

The “Boston Chicken” example, below, illustrates how identifying the particular type of sample
can be very important. Boston Chicken has 16 distinct side-dishes (the population). Each dinner
is served with the customer’s choice of three side-dishes. Read the following article and explain
the advertisement that there were “more than 3000 combinations” possible.
Thurs, Jan. 26, 1995. Rocky Mountain News, Retail & Marketing section
28  CHAPTER 1

Exercise 1-39 Calculate the number of possibilities under ordered samples without
replacement (customers eat their side dishes in the order of selection and must choose
3 different dishes).

Exercise 1-40 Calculate the number of possibilities using unordered samples without
replacement (customers can eat their dishes in any order but still must choose 3
different dishes).

Exercise 1-41 Calculate the number of possibilities using ordered samples with
replacement (customers eat their side dishes in a definite order, but can order, for
example, corn-corn-mashed potato, which is different from corn-mashed potato-corn.

Exercise 1.42 Calculate the number of possibilities using unordered samples with
replacement (customers eat their side dishes in any order, and corn-corn-mashed
potato is possible and is identical to corn-mashed potato-corn.

Exercise 1-43 Which type of sampling gives an answer closest to “more than 3000?”

Exercise 1-44 Which type of sampling provides the most realistic result?
Hint: The correct answer of 816 is given in the side-bar of the article, although it is
worked out differently using multiple steps.

Exercise 1-45 Assume that Boston Chicken customers choose three side-dishes from the
16 possible in such a way that all unordered samples with replacement are equally
likely. What is the probability that a customer will choose all three side-dishes the
same? That is, in poker parlance, what is the probability of three-of-a-kind?

Exercise 1-46
(a) How many ways can a parent distribute five one-dollar bills to her three children?
(b) How many ways can she accomplish this if each child gets at least one dollar?

Exercise 1-47
(a) How many ways can a witch distribute ten candy bars and seven packages of gum
to four trick-or-treaters?
(b) How many ways can she do this if each child receives at least one candy bar and
one package of gum?

Exercise 1-48 How many ways may a parent distribute ten identical pickled beets to his
five children?
 COMBINATORIAL PROBABILITY  29

Exercise 1-49 How many different 13-card bridge hands are possible?

Exercise 1-50 How many 5-card poker hands are there?

Exercise 1-51 At a local fast-food restaurant in Oregon (no sales tax), fries, soda,
hamburgers, cherry pie, and sundaes cost $1 each. Chicken sandwiches cost $2 each.
You have five dollars. How many different meals can you order?

Exercise 1-52 I have fifteen certificates for a free pizza and 24 cans of Coca-Cola®.
How many ways may I distribute the certificates and the cans of coke to 22 students?

1.3.3 SAMPLING AND OCCUPANCY UNITED

The following diagram shows the complete set of correspondences between sampling and
distribution. Although the formulation of a sampling problem may appear to be quite dissimilar
to the corresponding distribution problem, the two are in fact mathematically equivalent.

Samples of size r Without With

from n distinguishable replacement replacement
objects
n Pr nr
Order matters Distinguishable balls
(Example 1) (Example 3)

Order doesn’t matter

n
r 
n  r 1
r  Indistinguishable balls
(Example 2) (Example 4)
Distributions of r balls
Exclusive Non-exclusive
into n distinguishable urns

1.4 MORE APPLICATIONS

In this section we bring together a variety of special applications of the combinatorial
methods previously discussed. We begin with an illustration of some basic results involving
the algebra of polynomials.

1.4.1 THE BINOMIAL AND MULTINOMIAL THEOREMS

Consider the problem of expanding the binomial expression

( x  y)n  ( 
x  y )( 
x  y )( 
x  y) ( 
x  y) .
factor 1 factor 2 factor 3 factor n
30  CHAPTER 1

Algebraically, this amounts to adding together all possible products consisting of n letters, the
first selected from “factor 1,” the second from “factor 2,” and so forth up to “factor n.” In other
words, the expansion consists of the sum of all n letter “words” consisting of the letters x and y.

To simplify our final result, we group together all words that contain the same number of x’s
and y’s. Also, a stickler for algebra would insist on proper exponent notation (and the fact
that multiplication is commutative), to express words like xxyxyyx and yxxxyxy (with 4
x’s and 3y’s) all as x 4 y 3 .

Now, for a given r (r  0,1, 2,, n), the number of distinct n letter words with (r ) x ’s and
(n  r ) y’s will be the coefficient of x r  y n  r in the expansion of ( x  y ) n . As we have
n

previously seen, this is just the combination n Cr  r . This is the reason that the expression

“binomial coefficient” is used to describe the coefficient  nr of the term x  y r nr
of the

binomial expansion for ( x  y ) n .

The formula in the Binomial Theorem now follows from this observation.

The Binomial Theorem

For every non-negative integer n and real numbers x and y, we have

n
( x  y)n   n Cr  x r  y n  r
r 0

 n C0  y n  n C1  x1  y n 1  n C2  x 2  y n 2    n Cn  x n ,
or equivalently,
n
( x  y)n   n Cr  x n  r  y r
r 0

 n C0  x n  n C1  x n 1  y1  n C2  x n 2  y 2    n Cn  y n

Note
The two forms of the theorem given are in fact equivalent since n Cr  n C n  r is the
coefficient of both x r  y n  r and x n  r  y r .

Example 1.4-1 The Binomial Theorem

Use the Binomial Theorem to expand ( x  2 y )3 .

 COMBINATORIAL PROBABILITY  31

Solution
To apply the binomial theorem, we rewrite ( x  2 y ) 3  ( x  2 y) 3 .

( x  2 y )3  3C0  x3  (2 y )0  3C1  x 2  (2 y )1  3C2  x1  ( 2 y ) 2  3C3  x 0  (2 y )3

 1  x3  1  3  x 2 ( 2 y )  3  x  (4 y 2 )  1 1  (8 y 3 )  x3  6 x 2 y  12 xy 2  8 y 3 

Exercise 1-53 Use the Binomial Theorem to expand the following:

(a) ( x  3) 4
(b) (2 x  y ) 5
(c) (4 x  5 y )3

Exercise 1-54 Use the mathematical definition of combinations to verify the identity:

n  n  1  n  1
     .
r   r 1  r 
Note 1
For a non-algebraic proof, imagine that you belong to a sorority with n members, exactly r
of whom get to go to a party at the fraternity next door. How many ways to select the lucky r
members? Break it down into those combinations that contain you, and those combinations
that do not contain you. Are you one of the party-goers? If so, it remains to choose r  1
members from the n  1 members who are not you. If you stay home completing your
probability homework, then we must count the ways r members are chosen from the n  1
members who are not you.

Note 2
This identity is what makes the generation of Pascal’s3 triangle possible. The n th row of
Pascal’s triangle contains the binomial coefficients for expanding ( x  y ) n . Each new row is
generated by adding the adjacent coefficients (as in the identity above) from the previous row:

Pascal’s Triangle

( x  y)0 1 0th row

( x  y )1 1 1 1st row
( x  y)2 1 2 1 2 nd row
( x  y )3 1 3 3 1 3rd row
( x  y)4 1 4 6 4 1 4th row

3Pascal (1623-1662), considered to be, along with his contemporary, Pierre de Fermat (1601-1665), a progenitor
of modern probability theory.
32  CHAPTER 1

n
Exercise 1-55 Use the binomial theorem to verify the identity  n C k  2 n. For
k 0
4
example, if n  4, then  4 Ck  4 C0  4 C1  4 C 2  4 C3  4 C 4  16  2 4.
k 0

Note
n C k is the number of k  topping pizzas that can be made if there are n toppings from which to
select. You may create a zero-topping pizza, or a one-topping pizza, up to an n-topping pizza.
The left-hand side of the equation is the sum of different numbers of k-topping pizzas. On
the other hand, each pizza topping may either be placed on the pizza or not. There are two
ways to do this (put topping on pizza or do not put topping on pizza). There are n toppings
from which to select. By the multiplication principle, there are 2 
2  2  2 n ways to do
n times
this, which is the right-hand side of the identity.

Exercise 1-56 Explain the fundamental identity  nr    n n r  in terms of pizza toppings.

Exercise 1-57 In the expansion of ( x  y ) n the coefficient of x 4 y n 4 is 3,876 and the
coefficient of x 5 y n 5 is 11,628. Find the coefficient of x 5 y n 4 in the expansion of
( x  y ) n 1. What is the value of n?

It is possible to generalize the arguments involving the binomial theorem in order to

demonstrate the multinomial theorem. We wish to expand out the expression

( x1  x2    xr ) n  ( 
x1  x2 
  x1  x2 
xr )(    x1  x2 
xr ) (   
xr ).
factor 1 factor 2 factor n

The expansion now consists of all possible n-letter words consisting of the letters
x1 , x2 , , xr . We again group together all words with the same exponents on the letters.
How many words are there with ( n1 ) x1’s, ( n2 ) x2 ’s, ( n3 ) x3 ’s,  , ( nr ) xr ’s (think
MISSISSIPPI)? The answer is the number of partitions of n objects into subsets of sizes


n1 , n2 , n3 , , nr , that is
n . 
n1 , n2 ,, nr Therefore, the multinomial expansion can be
expressed as follows:

The Multinomial Theorem

 x1  x2    xr  n  
n1  nr  n
 n , n n,, n   x
1 2 r
n1
1  x2n2    xrnr .
 COMBINATORIAL PROBABILITY  33

The sum runs over all possible partitions n1 , n2 , n3 , , nr such that n1  n2  n3    nr  n.

Finding all of these partitions is the challenging part.

Note
The number of such partitions is the number of ways of distributing n indistinguishable balls
into r urns non-exclusively, that is, n 
n  r 1
. 
Example 1.4-2 The Multinomial Theorem

Use the multinomial theorem to expand ( x  y  z ) 4 .

Solution
You should take the time to find all of the ways to combine x, y, and z so that the total
powers sum to 4.

( x  y  z)4   4,0,0  3,1,0

4 x y z  4 x yz
4 0 0 3 1 0
  0,0,4 4 x y z
0 0 4

 x 4  4 x 3 y  4 x 3 z  6 x 2 y 2  6 x 2 z 2  12 x 2 yz  4 xy 3  12 xy 2 z
 12 xyz 2  4 xz 3  y 4  4 y 3 z  6 y 2 z 2  4 yz 3  z 4

Note that the number of terms is 15, the same as  n nr 1   4 43 1. 

Exercise 1-58 Use the multinomial theorem to expand the following:

(a) ( x  2 y  5 z ) 3

(b) ( w  x  y  2 z ) 2 .

1.4.2 POKER HANDS

Our simple version of poker is played with a standard four-suit, 52-card deck. We are
serious players, so there are neither jokers nor wild cards in our deck. A poker hand consists
of 5 cards dealt from a standard deck. In other words, a poker hand is an unordered random
sample of size 5 chosen from a population of size 52, without replacement (you wouldn’t
want to be caught in Dodge City with 2 Queens of Hearts in your hand). The ace can be
played as either high or low, as explained below. We present the definitions of the various
types of poker hands.
Straight flush: Five cards of the same suit in sequence, such as 7♥6♥5♥4♥3♥. The Ace-
King-Queen-Jack-Ten (A♣K♣Q♣J♣T♣) is called a royal flush. The ace
can also play low so that 5♣4♣3♣2♣A♣ is another straight flush.
34  CHAPTER 1

Four-of-a-kind: Four cards of the same denomination accompanied by another card, like
7♣7♦7♥7♠9♥.

Full house
(a.k.a. a boat): Three cards of one denomination accompanied by two of another, such
as Q♣Q♦Q♥4♠4♥.

Flush: Five cards of the same suit, such as K♠Q♠9♠6♠4♠. Straight flushes are
excluded (they form their own category above).

Straight: Five cards in sequence, such as J♥T♦9♣8♦7♠. The ace plays either high
or low, but a collection like 32AKQ is not allowed. That is, there are no
wrap-around straights. Again, straight flushes are excluded, since they
have been counted separately.

Three-of-a-kind: Three cards of the same denomination and two other cards of different
denominations, such as 7♠7♣7♦K♣2♦.

Two Pair: Two cards of one denomination, two cards of another denomination and
a fifth card of a third denomination, such as K♠K♣8♦8♠7♥.

One Pair: Two cards of one denomination accompanied by three cards of different
denominations, such as T♠T♣Q♦8♠7♥.

High Card Any hand that does not qualify as one of the hands above, such as
(a.k.a. Nothing): AQ987.

Various combinatorial techniques are employed to calculate the probabilities of being dealt
these hands on an initial deal from a standard deck of cards. We illustrate several of these
calculations and provide a summary of the probabilities of all types of poker hands.

Since poker hands consist of five cards (any order) selected without replacement from a

 
52
population of size 52 cards, the size of the sample space is 5  2,598,960.

Example 1.4-3 Straight Flush

Compute the probability of being dealt a straight flush.

Solution
The highest ranked straight flush is AKQJT in one of the four suits (clubs, diamonds, hearts,
and spades), the lowest ranked straight flush is 5432A. There are 10 of these rankings. There are
four suit choices. By the multiplication rule, there are 10  4  40 possible straight flushes.
 COMBINATORIAL PROBABILITY  35

number of straight flushes

Pr(straight flush) 
number of five card poker hands
= 10  4  40  .00001539. 
 
52
5
2,598,960

Example 1.4-4 Full House

Compute the probability of being dealt a full house on the initial deal of 5 cards.

Solution
We begin by selecting a card denomination (Ace, King ,..., 2) for the three-of-a-kind. One
has 13 ways to do this. We then select three of the four cards from the selected denomination
to create the three-of-a-kind. For the pair, there are only 12 remaining denominations to
select, since the card value chosen for the three-of-a-kind is no longer available. We need to
choose two cards with this second denomination.

Pr(full house)  number of full houses

number of five card poker hands

 131    43  121    42  13  4 12  6  .00144058. 

525 2,598,960

Exercise 1-59 Compute the probabilities for all nine poker hand types. Do this by
yourself prior to looking at the following summary. It is an important exercise, even if you
struggle.

Poker Probability Summary:

1. Straight Flush

number of straight flushes

Pr(Straight Flush) 
number of five card poker hands
4 suits  10 possible straights 40
   .00001538.
 
52
5
2,598,960

2. Four-of-a-Kind

Pr(Four-of-a-Kind) 
 1   4   1   1  13  1  12  4
13  4  12  4
  624  .00024010.
5
52 2,598,960 2,598,960
36  CHAPTER 1

3. Full House

Pr(Full House) 
 1   3  1   2
13  4  12  4
 13  4  12  6 
3,744
 .00144058.
5
52 2,598,960 2,598,960

4. Flush
5 cards selected
from the suit straight flushes

suits 
  
already accounted

4  13  40
1 5
Pr(Flush) 
525
4  1, 287  40 5,108
   .00196540.
2,598,960 2,598,960

5. Straight

Pr(Straight) 
  
10  4  4  4  4  4  40
1 1 1 1 1

 
52
5
5 10, 200
 10  4  40   .00392465.
2,598,960 2,598,960

6. Three-of-a-Kind

Pr(Three-of-a-Kind) 
 1   3  2  1  1
13  4  12  4  4

525
54,912
 13  4  66  4  4   .02112846.
2,598,960 2,598,960

7. Two Pair

Pr(Two Pair) 
 2   2  2  1  1 
13  4  4  11  4

525
123,552
 78  6  6  11  4   .04753902.
2,598,960 2,598,960
 COMBINATORIAL PROBABILITY  37

8. One Pair

Pr(One Pair) 
 1   2  3   1   1   1 
13  4  12  4  4  4

525
3 1,098, 240
 13  6  220  4   .42256903
2,598,960 2,598,960

9. Nothing
Pr(Nothing)  1  Pr(other possibilities)


 5
52  (40  624  3,744  5,108  10, 200  54,912  123,552  1,098, 240)

525
2,598,960  1, 296, 420 1,302,540
   .50117739.
2,598,960 2,598,960

Exercise 1-60 (Poker Dice): Play poker using 5 fair dice rather than a deck of cards. Roll the
five dice onto the table. The possible hands are: five-of-a-kind, four-of-a-kind, a full
house, three-of-a-kind, two pair, one pair, a straight, and nothing. Find the probabilities of
these events on a single roll of the dice. Where should the straight rank? The less likely
(lower probability) an event is to happen, the higher it should rank.

1.4.3 THE POWERBALL® LOTTERY

The following rules, prizes, and odds were once found at www.musl.com. To play the game,
we draw five balls out of a drum with 53 numbered white balls, and one power ball out of a
drum with 42 numbered green balls.

Powerball® Prizes and Odds

Match Prize Odds
+ Grand Prize 1 in 120,526,770.00
$100,000 1 in 2,939,677.32
+ $5,000 1 in 502,194.88
$100 1 in 12,248.66
+ $100 1 in 10,685.00
$7 1 in 260.61
+ $7 1 in 1696.85
+ $4 1 in 123.88
$3 1 in 70.39

The overall odds of winning a prize are 1 in 36.06.

The odds presented here are based on a $1 play and are rounded to two decimal places.
38  CHAPTER 1

Example 1.4-5 Probability of winning the $5,000 Powerball® lottery prize

Calculate the probability of winning the $5,000 prize (i.e., matching four of the five white
balls and the green power ball).

Note
The amount that you win is $4,999 since it costs you a dollar to purchase the game ticket.

Solution
There are five winning white balls, of which we need to select four. There are forty-eight
 53  5 losing white balls, of which we need to select one. We also need to select the
correct green power ball, of which there is only one.

Pr(4 white and the power ball) 

 4   1  1
5  48  1

535    421 
 5  48  1
2,869,685  42
 240
120,526,770
 1
502,194.88
 0.000001991.

The Powerball® graphic uses the expression “odds” in the third column. They should have
used “probability” and written the expressions as 1 / 502,194.88. There are a number of ways
in which odds are stated, but all of them are variants of probability statements.

In gambling, it is common to give the ratio of expected losses to expected wins in a fixed
number of plays. This ratio is referred to as “the odds against winning.” For example,
suppose that the betting public’s favorite horse in the Santa Anita derby has odds of 2:1
(read as “two to one”). This implies that if three identical races were run, then our horse
would be expected to lose twice and win once (the word “expect” is given a precise meaning
in Chapter 3). We could translate these odds to the equivalent probability of winning the
race, Pr(win)  1 win1win  1.
2 losses 3

 COMBINATORIAL PROBABILITY  39

Consider an event A whose probability of occurring is p. The complementary event (that A

does not occur) is denoted A.

Odds
The odds against the event A are quoted as the ratio,
Pr(A does not occur) : Pr(A does occur)  Pr( A) : Pr( A)  1 p  : p.

Note: Odds are generally quoted using whole numbers. For example, if P ( A)  2
5
3 2
then the odds against A are :  3: 2.
5 5

If you wanted to bet on A occurring, and the odds against A are 3:2, then you would put $2
in the pot and your opponent would put in $3. If the experiment could be repeated 5 times,
then you would expect to win twice and lose three times. You would lose $6 on the three
losses and win $6 on the two wins, breaking even. This would be considered a fair bet.

Converting Odds to Probability

If the odds against the event A are quoted as b : a, then Pr( A)  a .

ab

Example 1.4-6 Probability of winning the $4 Powerball® lottery prize

Calculate the probability of winning the $4 prize (matching one of the five white balls and
the power ball).

Solution
We need to match one of the 5 winning white balls (therefore select 4 losing white balls) and
match the one winning Powerball®.

Pr(one white and the power ball) 

 
5
1

48
4

1
1

5  194,580  1
  53
5

42
1
2,869,685  42

972,900 1
   .0081. 
120,526,770 123.88

Example 1.4-7 Probability of winning nothing

Calculate the probability of losing one dollar (e.g., matching zero or one or two of the five
white balls and not the power ball).

Solution
Losing options include matching 0, 1, or 2 winning white balls and not the power ball.
40  CHAPTER 1

Pr(losing a dollar) 
        
5
0

48
5

41
1

5
1

48
4

41
1

5
2

48
3

41
1

   53
5

42
1

1  1,712,304  41  5  194,580  41  10  17, 296  41 117,184,724

 
2,869,685  42 120,526,770
 .9723 

Exercise 1-61 Check the remaining probabilities for the Powerball® game listed below.
Verify that theses probabilities sum to 1.

Powerball® Probability Summary

Match Prize Odds
1
+ Grand Prize 120,526,770
41
$100,000 120,526,770
+ $5,000
240
120,526,770
9,840
$100 120,526,770
+ 11,280
$100 120,526,770
462,480
$7 120,526,770
+ 172,960
$7 120,526,770
+ 972,900
$4 120,526,770
1,712,304
$3 120,526,770
117,184,724
Nothing $0 120,526,770

Exercise 1-62 Suppose that the Grand Prize is 50 million dollars and paid in cash today
and that there are no taxes. Find the average winning computed as the sum of the
possible winning times its probability.