Introduction to Statistical Thinking

(With R, Without Calculus)

Benjamin Yakir, The Hebrew University

March, 2011
2
In memory of my father, Moshe Yakir, and the family he lost.
ii
Preface

The target audience for this book is college students who are required to learn
statistics, students with little background in mathematics and often no motiva-
tion to learn more. It is assumed that the students do have basic skills in using
computers and have access to one. Moreover, it is assumed that the students
are willing to actively follow the discussion in the text, to practice, and more
importantly, to think.
Teaching statistics is a challenge. Teaching it to students who are required
to learn the subject as part of their curriculum, is an art mastered by few. In
the past I have tried to master this art and failed. In desperation, I wrote this
book.
This book uses the basic structure of generic introduction to statistics course.
However, in some ways I have chosen to diverge from the traditional approach.
One divergence is the introduction of R as part of the learning process. Many
have used statistical packages or spreadsheets as tools for teaching statistics.
Others have used R in advanced courses. I am not aware of attempts to use
R in introductory level courses. Indeed, mastering R requires much investment
of time and energy that may be distracting and counterproductive for learning
more fundamental issues. Yet, I believe that if one restricts the application of
R to a limited number of commands, the benefits that R provides outweigh the
difficulties that R engenders.
Another departure from the standard approach is the treatment of proba-
bility as part of the course. In this book I do not attempt to teach probability
as a subject matter, but only specific elements of it which I feel are essential
for understanding statistics. Hence, Kolmogorov’s Axioms are out as well as
attempts to prove basic theorems and a Balls and Urns type of discussion. On
the other hand, emphasis is given to the notion of a random variable and, in
that context, the sample space.
The first part of the book deals with descriptive statistics and provides prob-
ability concepts that are required for the interpretation of statistical inference.
Statistical inference is the subject of the second part of the book.
The first chapter is a short introduction to statistics and probability. Stu-
dents are required to have access to R right from the start. Instructions regarding
the installation of R on a PC are provided.
The second chapter deals with data structures and variation. Chapter 3
provides numerical and graphical tools for presenting and summarizing the dis-
tribution of data.
The fundamentals of probability are treated in Chapters 4 to 7. The concept
of a random variable is presented in Chapter 4 and examples of special types of
random variables are discussed in Chapter 5. Chapter 6 deals with the Normal

iii
iv PREFACE

random variable. Chapter 7 introduces sampling distribution and presents the

Central Limit Theorem and the Law of Large Numbers. Chapter 8 summarizes
the material of the first seven chapters and discusses it in the statistical context.
Chapter 9 starts the second part of the book and the discussion of statis-
tical inference. It provides an overview of the topics that are presented in the
subsequent chapter. The material of the first half is revisited.
Chapters 10 to 12 introduce the basic tools of statistical inference, namely
point estimation, estimation with a confidence interval, and the testing of statis-
tical hypothesis. All these concepts are demonstrated in the context of a single
measurements.
Chapters 13 to 15 discuss inference that involve the comparison of two mea-
surements. The context where these comparisons are carried out is that of
regression that relates the distribution of a response to an explanatory variable.
In Chapter 13 the response is numeric and the explanatory variable is a factor
with two levels. In Chapter 14 both the response and the explanatory variable
are numeric and in Chapter 15 the response in a factor with two levels.
Chapter 16 ends the book with the analysis of two case studies. These
analyses require the application of the tools that are presented throughout the
book.
This book was originally written for a pair of courses in the University of the
People. As such, each part was restricted to 8 chapters. Due to lack of space,
some important material, especially the concepts of correlation and statistical
independence were omitted. In future versions of the book I hope to fill this
gap.
Large portions of this book, mainly in the first chapters and some of the
quizzes, are based on material from the online book “Collaborative Statistics”
by Barbara Illowsky and Susan Dean (Connexions, March 2, 2010. http://
cnx.org/content/col10522/1.37/). Most of the material was edited by this
author, who is the only person responsible for any errors that where introduced
in the process of editing.
Case studies that are presented in the second part of the book are taken
from Rice Virtual Lab in Statistics can be found in their Case Studies section.
The responsibility for mistakes in the analysis of the data, if such mistakes are
found, are my own.
I would like to thank my mother Ruth who, apart from giving birth, feeding
and educating me, has also helped to improve the pedagogical structure of this
text.
This book is an open source and may be used by anyone who wishes to do so.
(Under the conditions of the Creative Commons Attribution License (CC-BY
3.0).))

Jerusalem, March 2011 Benjamin Yakir

Contents

Preface iii

I Introduction to Statistics 1
1 Introduction 3
1.1 Student Learning Objectives . . . . . . . . . . . . . . . . . . . . . 3
1.2 Why Learn Statistics? . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Probability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.5 Key Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.6 The R Programming Environment . . . . . . . . . . . . . . . . . 7
1.6.1 Some Basic R Commands . . . . . . . . . . . . . . . . . . 7
1.7 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2 Sampling and Data Structures 15

2.1 Student Learning Objectives . . . . . . . . . . . . . . . . . . . . . 15
2.2 The Sampled Data . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.1 Variation in Data . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.2 Variation in Samples . . . . . . . . . . . . . . . . . . . . . 16
2.2.3 Frequency . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.4 Critical Evaluation . . . . . . . . . . . . . . . . . . . . . . 18
2.3 Reading Data into R . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.1 Saving the File and Setting the Working Directory . . . . 19
2.3.2 Reading a CSV File into R . . . . . . . . . . . . . . . . . . 23
2.3.3 Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.4 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3 Descriptive Statistics 29
3.1 Student Learning Objectives . . . . . . . . . . . . . . . . . . . . . 29
3.2 Displaying Data . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.1 Histograms . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2.2 Box Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3 Measures of the Center of Data . . . . . . . . . . . . . . . . . . . 35
3.3.1 Skewness, the Mean and the Median . . . . . . . . . . . . 36
3.4 Measures of the Spread of Data . . . . . . . . . . . . . . . . . . . 38

v
vi CONTENTS

3.5 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

4 Probability 47
4.1 Student Learning Objective . . . . . . . . . . . . . . . . . . . . . 47
4.2 Different Forms of Variability . . . . . . . . . . . . . . . . . . . . 47
4.3 A Population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.4 Random Variables . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.4.1 Sample Space and Distribution . . . . . . . . . . . . . . . 54
4.4.2 Expectation and Standard Deviation . . . . . . . . . . . . 56
4.5 Probability and Statistics . . . . . . . . . . . . . . . . . . . . . . 59
4.6 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

5 Random Variables 65
5.1 Student Learning Objective . . . . . . . . . . . . . . . . . . . . . 65
5.2 Discrete Random Variables . . . . . . . . . . . . . . . . . . . . . 65
5.2.1 The Binomial Random Variable . . . . . . . . . . . . . . . 66
5.2.2 The Poisson Random Variable . . . . . . . . . . . . . . . 71
5.3 Continuous Random Variable . . . . . . . . . . . . . . . . . . . . 74
5.3.1 The Uniform Random Variable . . . . . . . . . . . . . . . 75
5.3.2 The Exponential Random Variable . . . . . . . . . . . . . 79
5.4 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

6 The Normal Random Variable 87

6.1 Student Learning Objective . . . . . . . . . . . . . . . . . . . . . 87
6.2 The Normal Random Variable . . . . . . . . . . . . . . . . . . . . 87
6.2.1 The Normal Distribution . . . . . . . . . . . . . . . . . . 88
6.2.2 The Standard Normal Distribution . . . . . . . . . . . . . 90
6.2.3 Computing Percentiles . . . . . . . . . . . . . . . . . . . . 92
6.2.4 Outliers and the Normal Distribution . . . . . . . . . . . 94
6.3 Approximation of the Binomial Distribution . . . . . . . . . . . . 96
6.3.1 Approximate Binomial Probabilities and Percentiles . . . 96
6.3.2 Continuity Corrections . . . . . . . . . . . . . . . . . . . . 97
6.4 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

7 The Sampling Distribution 105

7.1 Student Learning Objective . . . . . . . . . . . . . . . . . . . . . 105
7.2 The Sampling Distribution . . . . . . . . . . . . . . . . . . . . . 105
7.2.1 A Random Sample . . . . . . . . . . . . . . . . . . . . . . 106
7.2.2 Sampling From a Population . . . . . . . . . . . . . . . . 107
7.2.3 Theoretical Models . . . . . . . . . . . . . . . . . . . . . . 112
7.3 Law of Large Numbers and Central Limit Theorem . . . . . . . . 115
7.3.1 The Law of Large Numbers . . . . . . . . . . . . . . . . . 115
7.3.2 The Central Limit Theorem (CLT) . . . . . . . . . . . . . 116
7.3.3 Applying the Central Limit Theorem . . . . . . . . . . . . 119
7.4 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
7.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
CONTENTS vii

8 Overview and Integration 125

8.1 Student Learning Objective . . . . . . . . . . . . . . . . . . . . . 125
8.2 An Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
8.3 Integrated Applications . . . . . . . . . . . . . . . . . . . . . . . 127
8.3.1 Example 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
8.3.2 Example 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
8.3.3 Example 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
8.3.4 Example 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
8.3.5 Example 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . 134

II Statistical Inference 137

9 Introduction to Statistical Inference 139
9.1 Student Learning Objectives . . . . . . . . . . . . . . . . . . . . . 139
9.2 Key Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
9.3 The Cars Data Set . . . . . . . . . . . . . . . . . . . . . . . . . . 141
9.4 The Sampling Distribution . . . . . . . . . . . . . . . . . . . . . 144
9.4.1 Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
9.4.2 The Sampling Distribution . . . . . . . . . . . . . . . . . 145
9.4.3 Theoretical Distributions of Observations . . . . . . . . . 146
9.4.4 Sampling Distribution of Statistics . . . . . . . . . . . . . 147
9.4.5 The Normal Approximation . . . . . . . . . . . . . . . . . 148
9.4.6 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . 149
9.5 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
9.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

10 Point Estimation 159

10.1 Student Learning Objectives . . . . . . . . . . . . . . . . . . . . . 159
10.2 Estimating Parameters . . . . . . . . . . . . . . . . . . . . . . . . 159
10.3 Estimation of the Expectation . . . . . . . . . . . . . . . . . . . . 160
10.3.1 The Accuracy of the Sample Average . . . . . . . . . . . 161
10.3.2 Comparing Estimators . . . . . . . . . . . . . . . . . . . . 164
10.4 Variance and Standard Deviation . . . . . . . . . . . . . . . . . . 166
10.5 Estimation of Other Parameters . . . . . . . . . . . . . . . . . . 171
10.6 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
10.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178

11 Confidence Intervals 181

11.1 Student Learning Objectives . . . . . . . . . . . . . . . . . . . . . 181
11.2 Intervals for Mean and Proportion . . . . . . . . . . . . . . . . . 181
11.2.1 Examples of Confidence Intervals . . . . . . . . . . . . . . 182
11.2.2 Confidence Intervals for the Mean . . . . . . . . . . . . . 183
11.2.3 Confidence Intervals for a Proportion . . . . . . . . . . . 187
11.3 Intervals for Normal Measurements . . . . . . . . . . . . . . . . . 188
11.3.1 Confidence Intervals for a Normal Mean . . . . . . . . . . 190
11.3.2 Confidence Intervals for a Normal Variance . . . . . . . . 192
11.4 Choosing the Sample Size . . . . . . . . . . . . . . . . . . . . . . 195
11.5 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
11.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
viii CONTENTS

12 Testing Hypothesis 203

12.1 Student Learning Objectives . . . . . . . . . . . . . . . . . . . . . 203
12.2 The Theory of Hypothesis Testing . . . . . . . . . . . . . . . . . 203
12.2.1 An Example of Hypothesis Testing . . . . . . . . . . . . . 204
12.2.2 The Structure of a Statistical Test of Hypotheses . . . . . 205
12.2.3 Error Types and Error Probabilities . . . . . . . . . . . . 208
12.2.4 p-Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
12.3 Testing Hypothesis on Expectation . . . . . . . . . . . . . . . . . 211
12.4 Testing Hypothesis on Proportion . . . . . . . . . . . . . . . . . . 218
12.5 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
12.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224

13 Comparing Two Samples 227

13.1 Student Learning Objectives . . . . . . . . . . . . . . . . . . . . . 227
13.2 Comparing Two Distributions . . . . . . . . . . . . . . . . . . . . 227
13.3 Comparing the Sample Means . . . . . . . . . . . . . . . . . . . . 229
13.3.1 An Example of a Comparison of Means . . . . . . . . . . 229
13.3.2 Confidence Interval for the Difference . . . . . . . . . . . 232
13.3.3 The t-Test for Two Means . . . . . . . . . . . . . . . . . . 235
13.4 Comparing Sample Variances . . . . . . . . . . . . . . . . . . . . 237
13.5 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
13.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245

14 Linear Regression 247

14.1 Student Learning Objectives . . . . . . . . . . . . . . . . . . . . . 247
14.2 Points and Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
14.2.1 The Scatter Plot . . . . . . . . . . . . . . . . . . . . . . . 248
14.2.2 Linear Equation . . . . . . . . . . . . . . . . . . . . . . . 251
14.3 Linear Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
14.3.1 Fitting the Regression Line . . . . . . . . . . . . . . . . . 253
14.3.2 Inference . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
14.4 R-squared and the Variance of Residuals . . . . . . . . . . . . . . 260
14.5 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
14.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278

15 A Bernoulli Response 281

15.1 Student Learning Objectives . . . . . . . . . . . . . . . . . . . . . 281
15.2 Comparing Sample Proportions . . . . . . . . . . . . . . . . . . . 282
15.3 Logistic Regression . . . . . . . . . . . . . . . . . . . . . . . . . . 285
15.4 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 289

16 Case Studies 299

16.1 Student Learning Objective . . . . . . . . . . . . . . . . . . . . . 299
16.2 A Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
16.3 Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
16.3.1 Physicians’ Reactions to the Size of a Patient . . . . . . . 300
16.3.2 Physical Strength and Job Performance . . . . . . . . . . 306
16.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
16.4.1 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . 313
16.4.2 Discussion in the Forum . . . . . . . . . . . . . . . . . . . 314
 Part I

Introduction to Statistics

1
Chapter 1

Introduction

1.1 Student Learning Objectives

This chapter introduces the basic concepts of statistics. Special attention is
given to concepts that are used in the first part of this book, the part that
deals with graphical and numeric statistical ways to describe data (descriptive
statistics) as well as mathematical theory of probability that enables statisticians
to draw conclusions from data.
The course applies the widely used freeware programming environment for
statistical analysis, known as R. In this chapter we will discuss the installation
of the program and present very basic features of that system.
By the end of this chapter, the student should be able to:

• Recognize key terms in statistics and probability.

• Install the R program on an accessible computer.

• Learn and apply a few basic operations of the computational system R.

1.2 Why Learn Statistics?

You are probably asking yourself the question, “When and where will I use
statistics?”. If you read any newspaper or watch television, or use the Internet,
you will see statistical information. There are statistics about crime, sports,
education, politics, and real estate. Typically, when you read a newspaper
article or watch a news program on television, you are given sample information.
With this information, you may make a decision about the correctness of a
statement, claim, or “fact”. Statistical methods can help you make the “best
educated guess”.
Since you will undoubtedly be given statistical information at some point in
your life, you need to know some techniques to analyze the information thought-
fully. Think about buying a house or managing a budget. Think about your
chosen profession. The fields of economics, business, psychology, education, bi-
ology, law, computer science, police science, and early childhood development
require at least one course in statistics.

3
4 CHAPTER 1. INTRODUCTION

4
3
y = Frequency

2
1
0

5 5.5 6 6.5 7 8 9

x = Time

Figure 1.1: Frequency of Average Time (in Hours) Spent Sleeping per Night

Included in this chapter are the basic ideas and words of probability and
statistics. In the process of learning the first part of the book, and more so in
the second part of the book, you will understand that statistics and probability
work together.

1.3 Statistics
The science of statistics deals with the collection, analysis, interpretation, and
presentation of data. We see and use data in our everyday lives. To be able
to use data correctly is essential to many professions and is in your own best
self-interest.
For example, assume the average time (in hours, to the nearest half-hour) a
group of people sleep per night has been recorded. Consider the following data:

5, 5.5, 6, 6, 6, 6.5, 6.5, 6.5, 6.5, 7, 7, 8, 8, 9 .

In Figure 1.1 this data is presented in a graphical form (called a bar plot). A bar
plot consists of a number axis (the x-axis) and bars (vertical lines) positioned
1.4. PROBABILITY 5

above the number axis. The length of each bar corresponds to the number
of data points that obtain the given numerical value. In the given plot the
frequency of average time (in hours) spent sleeping per night is presented with
hours of sleep on the horizontal x-axis and frequency on vertical y-axis.
Think of the following questions:
• Would the bar plot constructed from data collected from a different group
of people look the same as or different from the example? Why?
• If one would have carried the same example in a different group with the
same size and age as the one used for the example, do you think the results
would be the same? Why or why not?
• Where does the data appear to cluster? How could you interpret the
clustering?
The questions above ask you to analyze and interpret your data. With this
example, you have begun your study of statistics.
In this course, you will learn how to organize and summarize data. Or-
ganizing and summarizing data is called descriptive statistics. Two ways to
summarize data are by graphing and by numbers (for example, finding an av-
erage). In the second part of the book you will also learn how to use formal
methods for drawing conclusions from “good” data. The formal methods are
called inferential statistics. Statistical inference uses probabilistic concepts to
determine if conclusions drawn are reliable or not.
Effective interpretation of data is based on good procedures for producing
data and thoughtful examination of the data. In the process of learning how
to interpret data you will probably encounter what may seem to be too many
mathematical formulae that describe these procedures. However, you should
always remember that the goal of statistics is not to perform numerous calcu-
lations using the formulae, but to gain an understanding of your data. The
calculations can be done using a calculator or a computer. The understanding
must come from you. If you can thoroughly grasp the basics of statistics, you
can be more confident in the decisions you make in life.

1.4 Probability
Probability is the mathematical theory used to study uncertainty. It provides
tools for the formalization and quantification of the notion of uncertainty. In
particular, it deals with the chance of an event occurring. For example, if the
different potential outcomes of an experiment are equally likely to occur then
the probability of each outcome is taken to be the reciprocal of the number of
potential outcomes. As an illustration, consider tossing a fair coin. There are
two possible outcomes – a head or a tail – and the probability of each outcome
is 1/2.
If you toss a fair coin 4 times, the outcomes may not necessarily be 2 heads
and 2 tails. However, if you toss the same coin 4,000 times, the outcomes will
be close to 2,000 heads and 2,000 tails. It is very unlikely to obtain more than
2,060 tails and it is similarly unlikely to obtain less than 1,940 tails. This is
consistent with the expected theoretical probability of heads in any one toss.
Even though the outcomes of a few repetitions are uncertain, there is a regular
6 CHAPTER 1. INTRODUCTION

pattern of outcomes when the number of repetitions is large. Statistics exploits

this pattern regularity in order to make extrapolations from the observed sample
to the entire population.
The theory of probability began with the study of games of chance such as
poker. Today, probability is used to predict the likelihood of an earthquake, of
rain, or whether you will get an “A” in this course. Doctors use probability
to determine the chance of a vaccination causing the disease the vaccination is
supposed to prevent. A stockbroker uses probability to determine the rate of
return on a client’s investments. You might use probability to decide to buy a
lottery ticket or not.
Although probability is instrumental for the development of the theory of
statistics, in this introductory course we will not develop the mathematical the-
ory of probability. Instead, we will concentrate on the philosophical aspects of
the theory and use computerized simulations in order to demonstrate proba-
bilistic computations that are applied in statistical inference.

1.5 Key Terms

In statistics, we generally want to study a population. You can think of a
population as an entire collection of persons, things, or objects under study.
To study the larger population, we select a sample. The idea of sampling is
to select a portion (or subset) of the larger population and study that portion
(the sample) to gain information about the population. Data are the result of
sampling from a population.
Because it takes a lot of time and money to examine an entire population,
sampling is a very practical technique. If you wished to compute the overall
grade point average at your school, it would make sense to select a sample of
students who attend the school. The data collected from the sample would
be the students’ grade point averages. In presidential elections, opinion poll
samples of 1,000 to 2,000 people are taken. The opinion poll is supposed to
represent the views of the people in the entire country. Manufacturers of canned
carbonated drinks take samples to determine if the manufactured 16 ounce
containers does indeed contain 16 ounces of the drink.
From the sample data, we can calculate a statistic. A statistic is a number
that is a property of the sample. For example, if we consider one math class to
be a sample of the population of all math classes, then the average number of
points earned by students in that one math class at the end of the term is an
example of a statistic. The statistic can be used as an estimate of a population
parameter. A parameter is a number that is a property of the population. Since
we considered all math classes to be the population, then the average number of
points earned per student over all the math classes is an example of a parameter.
One of the main concerns in the field of statistics is how accurately a statistic
estimates a parameter. The accuracy really depends on how well the sample
represents the population. The sample must contain the characteristics of the
population in order to be a representative sample.
Two words that come up often in statistics are average and proportion. If
you were to take three exams in your math classes and obtained scores of 86, 75,
and 92, you calculate your average score by adding the three exam scores and
dividing by three (your average score would be 84.3 to one decimal place). If, in
1.6. THE R PROGRAMMING ENVIRONMENT 7

your math class, there are 40 students and 22 are men and 18 are women, then
the proportion of men students is 22/40 and the proportion of women students
is 18/40. Average and proportion are discussed in more detail in later chapters.

1.6 The R Programming Environment

The R Programming Environment is a widely used open source system for sta-
tistical analysis and statistical programming. It includes thousands of functions
for the implementation of both standard and exotic statistical methods and it
is probably the most popular system in the academic world for the development
of new statistical tools. We will use R in order to apply the statistical meth-
ods that will be discussed in the book to some example data sets and in order
to demonstrate, via simulations, concepts associated with probability and its
application in statistics.
The demonstrations in the book involve very basic R programming skills and
the applications are implemented using, in most cases, simple and natural code.
A detailed explanation will accompany the code that is used.
Learning R, like the learning of any other programming language, can be
achieved only through practice. Hence, we strongly recommend that you not
only read the code presented in the book but also run it yourself, in parallel to
the reading of the provided explanations. Moreover, you are encouraged to play
with the code: introduce changes in the code and in the data and see how the
output changes as a result. One should not be afraid to experiment. At worst,
the computer may crash or freeze. In both cases, restarting the computer will
solve the problem . . .
You may download R from the R project home page http://www.r-project.
org and install it on the computer that you are using1 .

1.6.1 Some Basic R Commands

R is an object-oriented programming system. During the session you may cre-
ate and manipulate objects by the use of functions that are part of the basic
installation. You may also use the R programming language. Most of the func-
tions that are part of the system are themselves written in the R language and
one may easily write new functions or modify existing functions to suit specific
needs.
Let us start by opening the R Console window by double-clicking on the
R icon. Type in the R Console window, immediately after the “>” prompt,
the expression “1+2” and then hit the Return key. (Do not include the double
quotation in the expression that you type!):

> 1+2
[1] 3
>

The prompt “>” indicates that the system is ready to receive commands. Writ-
ing an expression, such as “1+2”, and hitting the Return key sends the expression
1 Detailed explanation of how to install the system on an XP Windows Operating System

may be found here: http://pluto.huji.ac.il/~msby/StatThink/install_R_WinXP.html.

8 CHAPTER 1. INTRODUCTION

to be executed. The execution of the expression may produce an object, in this

case an object that is composed of a single number, the number “3”.
Whenever required, the R system takes an action. If no other specifications
are given regarding the required action then the system will apply the pre-
programmed action. This action is called the default action. In the case of
hitting the Return key after the expression that we wrote the default is to
display the produced object on the screen.
Next, let us demonstrate R in a more meaningful way by using it in order
to produce the bar-plot of Figure 1.1. First we have to input the data. We
will produce a sequence of numbers that form the data2 . For that we will use
the function “c” that combines its arguments and produces a sequence with the
arguments as the components of the sequence. Write the expression:
> c(5,5.5,6,6,6,6.5,6.5,6.5,6.5,7,7,8,8,9)
at the prompt and hit return. The result should look like this:
> c(5,5.5,6,6,6,6.5,6.5,6.5,6.5,7,7,8,8,9)
[1] 5.0 5.5 6.0 6.0 6.0 6.5 6.5 6.5 6.5 7.0 7.0 8.0 8.0 9.0
>
The function “c” is an example of an R function. A function has a name, “c”
in this case, that is followed by brackets that include the input to the function.
We call the components of the input the arguments of the function. Arguments
are separated by commas. A function produces an output, which is typically
an R object. In the current example an object of the form of a sequence was
created and, according to the default application of the system, was sent to the
screen and not saved.
If we want to create an object for further manipulation then we should save
it and give it a name. For example, it we want to save the vector of data under
the name “X” we may write the following expression at the prompt (and then
hit return):
> X <- c(5,5.5,6,6,6,6.5,6.5,6.5,6.5,7,7,8,8,9)
>
The arrow that appears after the “X” is produced by typing the less than key
“<” followed by the minus key “-”. This arrow is the assignment operator.
Observe that you may save typing by calling and editing lines of code that
were processes in an earlier part of the session. One may browse through the
lines using the up and down arrows on the right-hand side of the keyboard and
use the right and left arrows to move along the line presented at the prompt.
For example, the last expression may be produced by finding first the line that
used the function “c” with the up and down arrow and then moving to the
beginning of the line with the left arrow. At the beginning of the line all one
has to do is type “X <- ” and hit the Return key.
Notice that no output was sent to the screen. Instead, the output from the
“c” function was assigned to an object that has the name “X”. A new object
by the given name was formed and it is now available for further analysis. In
order to verify this you may write “X” at the prompt and hit return:
2 In R, a sequence of numbers is called a vector. However, we will use the term sequence to

refer to vectors.
1.6. THE R PROGRAMMING ENVIRONMENT 9

Figure 1.2: Save Workspace Dialog

> X
[1] 5.0 5.5 6.0 6.0 6.0 6.5 6.5 6.5 6.5 7.0 7.0 8.0 8.0 9.0
The content of the object “X” is sent to the screen, which is the default output.
Notice that we have not changed the given object, which is still in the memory.
The object “X” is in the memory, but it is not saved on the hard disk.
With the end of the session the objects created in the session are erased unless
specifically saved. The saving of all the objects that were created during the
session can be done when the session is finished. Hence, when you close the
R Console window a dialog box will open (See the screenshot in Figure 1.2).
Via this dialog box you can choose to save the objects that were created in the
session by selecting “Yes”, not to save by selecting the option “No”, or you may
decide to abort the process of shutting down the session by selecting “Cancel”.
If you save the objects then they will be uploaded to the memory the next time
that the R Console is opened.
We used a capital letter to name the object. We could have used a small
letter just as well or practically any combination of letters. However, you should
note that R distinguishes between capital and small letter. Hence, typing “x”
in the console window and hitting return will produce an error message:
> x
Error: object "x" not found

An object named “x” does not exist in the R system and we have not created
such object. The object “X”, on the other hand, does exist.
Names of functions that are part of the system are fixed but you are free to
choose a name to objects that you create. For example, if one wants to create
10 CHAPTER 1. INTRODUCTION

an object by the name “my.vector” that contains the numbers 3, 7, 3, 3, and

-5 then one may write the expression “my.vector <- c(3,7,3,3,-5)” at the
prompt and hit the Return key.
If we want to produce a table that contains a count of the frequency of the
different values in our data we can apply the function “table” to the object
“X” (which is the object that contains our data):
> table(X)
X
5 5.5 6 6.5 7 8 9
1 1 3 4 2 2 1
Notice that the output of the function “table” is a table of the different levels
of the input vector and the frequency of each level. This output is yet another
type of an object.
The bar-plot of Figure 1.1 can be produced by the application of the function
“plot” to the object that is produced as an output of the function “table”:
> plot(table(X))
Observe that a graphical window was opened with the target plot. The plot that
appears in the graphical window should coincide with the plot in Figure 1.3.
This plot is practically identical to the plot in Figure 1.1. The only difference is
in the names given to the access. These names were changed in Figure 1.1 for
clarity.
Clearly, if one wants to produce a bar-plot to other numerical data all one has
to do is replace in the expression “plot(table(X))” the object “X” by an object
that contains the other data. For example, to plot the data in “my.vector” you
may use “plot(table(my.vector))”.

1.7 Solved Exercises

Question 1.1. A potential candidate for a political position in some state is
interested to know what are her chances to win the primaries of her party and be
selected as parties candidate for the position. In order to examine the opinions
of her party voters she hires the services of a polling agency. The polling is
conducted among 500 registered voters of the party. One of the questions that
the pollsters refers to the willingness of the voters to vote for a female candidate
for the job. Forty two percent of the people asked said that they prefer to have
a women running for the job. Thirty eight percent said that the candidate’s
gender is irrelevant. The rest prefers a male candidate. Which of the following
is (i) a population (ii) a sample (iii) a parameter and (iv) a statistic:
1. The 500 registered voters.
2. The percentage, among all registered voters of the given party, of those
that prefer a male candidate.
3. The number 42% that corresponds to the percentage of those that prefer
a female candidate.
4. The voters in the state that are registered to the given party.
1.7. SOLVED EXERCISES 11

4
3
table(X)

2
1
0

5 5.5 6 6.5 7 8 9

Figure 1.3: The Plot Produced by the Expression “plot(table(X))”

Solution (to Question 1.1.1): According to the information in the question

the polling was conducted among 500 registered voters. The 500 registered
voters corresponds to the sample.

Solution (to Question 1.1.2): The percentage, among all registered voters
of the given party, of those that prefer a male candidate is a parameter. This
quantity is a characteristic of the population.

Solution (to Question 1.1.3): It is given that 42% of the sample prefer a
female candidate. This quantity is a numerical characteristic of the data, of the
sample. Hence, it is a statistic.

Solution (to Question 1.1.4): The voters in the state that are registered to
the given party is the target population.

Question 1.2. The number of customers that wait in front of a coffee shop at
the opening was reported during 25 days. The results were:
4, 2, 1, 1, 0, 2, 1, 2, 4, 2, 5, 3, 1, 5, 1, 5, 1, 2, 1, 1, 3, 4, 2, 4, 3 .
12 CHAPTER 1. INTRODUCTION

8
6
table(n.cost)

4
2
0

0 1 2 3 4 5

n.cost

Figure 1.4: The Plot Produced by the Expression “plot(table(n.cost))”

1. Identify the number of days in which 5 costumers where waiting.

2. The number of waiting costumers that occurred the largest number of

times.

3. The number of waiting costumers that occurred the least number of times.

Solution (to Question 1.2): One may read the data into R and create a table
using the code:

> n.cost <- c(4,2,1,1,0,2,1,2,4,2,5,3,1,5,1,5,1,2,1,1,3,4,2,4,3)

> table(n.cost)
n.cost
0 1 2 3 4 5
1 8 6 3 4 3

For convenience, one may also create the bar plot of the data using the code:

> plot(table(n.cost))
1.8. SUMMARY 13

The bar plot is presented in Figure 1.4.

Solution (to Question 1.2.1): The number of days in which 5 costumers

where waiting is 3, since the frequency of the value “5” in the data is 3. That
can be seen from the table by noticing the number below value “5” is 3. It can
also be seen from the bar plot by observing that the hight of the bar above the
value “5” is equal to 3.

Solution (to Question 1.2.2): The number of waiting costumers that oc-
curred the largest number of times is 1. The value ”1” occurred 8 times, more
than any other value. Notice that the bar above this value is the highest.

Solution (to Question 1.2.3): The value ”0”, which occurred only once,
occurred the least number of times.

1.8 Summary
Glossary
Data: A set of observations taken on a sample from a population.
Statistic: A numerical characteristic of the data. A statistic estimates the
corresponding population parameter. For example, the average number
of contribution to the course’s forum for this term is an estimate for the
average number of contributions in all future terms (parameter).
Statistics The science that deals with processing, presentation and inference
from data.

Probability: A mathematical field that models and investigates the notion of

randomness.

Discuss in the forum

A sample is a subgroup of the population that is supposed to represent the
entire population. In your opinion, is it appropriate to attempt to represent the
entire population only by a sample?
When you formulate your answer to this question it may be useful to come
up with an example of a question from you own field of interest one may want to
investigate. In the context of this example you may identify a target population
which you think is suited for the investigation of the given question. The ap-
propriateness of using a sample can be discussed in the context of the example
question and the population you have identified.
14 CHAPTER 1. INTRODUCTION
Chapter 2

Sampling and Data

Structures

2.1 Student Learning Objectives

In this chapter we deal with issues associated with the data that is obtained from
a sample. The variability associated with this data is emphasized and critical
thinking about validity of the data encouraged. A method for the introduction
of data from an external source into R is proposed and the data types used by
R for storage are described. By the end of this chapter, the student should be
able to:
• Recognize potential difficulties with sampled data.
• Read an external data file into R.
• Create and interpret frequency tables.

2.2 The Sampled Data

The aim in statistics is to learn the characteristics of a population on the basis
of a sample selected from the population. An essential part of this analysis
involves consideration of variation in the data.

2.2.1 Variation in Data

Variation is given a central role in statistics. To some extent the assessment of
variation and the quantification of its contribution to uncertainties in making
inference is the statistician’s main concern.
Variation is present in any set of data. For example, 16-ounce cans of bev-
erage may contain more or less than 16 ounces of liquid. In one study, eight 16
ounce cans were measured and produced the following amount (in ounces) of
beverage:
15.8, 16.1, 15.2, 14.8, 15.8, 15.9, 16.0, 15.5 .
Measurements of the amount of beverage in a 16-ounce may vary because the
conditions of measurement varied or because the exact amount, 16 ounces of

15
16 CHAPTER 2. SAMPLING AND DATA STRUCTURES

liquid, was not put into the cans. Manufacturers regularly run tests to determine
if the amount of beverage in a 16-ounce can falls within the desired range.
Be aware that if an investigator collects data, the data may vary somewhat
from the data someone else is taking for the same purpose. This is completely
natural. However, if two investigators or more, are taking data from the same
source and get very different results, it is time for them to reevaluate their
data-collection methods and data recording accuracy.

2.2.2 Variation in Samples

Two or more samples from the same population, all having the same character-
istics as the population, may nonetheless be different from each other. Suppose
Doreen and Jung both decide to study the average amount of time students
sleep each night and use all students at their college as the population. Doreen
may decide to sample randomly a given number of students from the entire
body of collage students. Jung, on the other hand, may decide to sample ran-
domly a given number of classes and survey all students in the selected classes.
Doreen’s method is called random sampling whereas Jung’s method is called
cluster sampling. Doreen’s sample will be different from Jung’s sample even
though both samples have the characteristics of the population. Even if Doreen
and Jung used the same sampling method, in all likelihood their samples would
be different. Neither would be wrong, however.
If Doreen and Jung took larger samples (i.e. the number of data values
is increased), their sample results (say, the average amount of time a student
sleeps) would be closer to the actual population average. But still, their samples
would be, most probably, different from each other.
The size of a sample (often called the number of observations) is important.
The examples you have seen in this book so far have been small. Samples of only
a few hundred observations, or even smaller, are sufficient for many purposes.
In polling, samples that are from 1200 to 1500 observations are considered large
enough and good enough if the survey is random and is well done. The theory of
statistical inference, that is the subject matter of the second part of this book,
provides justification for these claims.

2.2.3 Frequency
The primary way of summarizing the variability of data is via the frequency
distribution. Consider an example. Twenty students were asked how many
hours they worked per day. Their responses, in hours, are listed below:
5, 6, 3, 3, 2, 4, 7, 5, 2, 3, 5, 6, 5, 4, 4, 3, 5, 2, 5, 3 .
Let us create an R object by the name “work.hours” that contains these data:
> work.hours <- c(5,6,3,3,2,4,7,5,2,3,5,6,5,4,4,3,5,2,5,3)
Next, let us create a table that summarizes the different values of working hours
and the frequency in which these values appear in the data:
> table(work.hours)
work.hours
2 3 4 5 6 7
3 5 3 6 2 1
2.2. THE SAMPLED DATA 17

Recall that the function “table” takes as input a sequence of data and produces
as output the frequencies of the different values.
We may have a clearer understanding of the meaning of the output of the
function “table” if we presented outcome as a frequency listing the different
data values in ascending order and their frequencies. For that end we may apply
the function “data.frame” to the output of the “table” function and obtain:
> data.frame(table(work.hours))
work.hours Freq
2 2 3
3 3 5
4 4 3
5 5 6
6 6 2
7 7 1
A frequency is the number of times a given datum occurs in a data set.
According to the table above, there are three students who work 2 hours, five
students who work 3 hours, etc. The total of the frequency column, 20, repre-
sents the total number of students included in the sample.
The function “data.frame” transforms its input into a data frame, which is
the standard way of storing statistical data. We will introduce data frames in
more detail in Section 2.3 below.
A relative frequency is the fraction of times a value occurs. To find the
relative frequencies, divide each frequency by the total number of students in
the sample – 20 in this case. Relative frequencies can be written as fractions,
percents, or decimals.
As an illustration let us compute the relative frequencies in our data:
> freq <- table(work.hours)
> freq
work.hours
2 3 4 5 6 7
3 5 3 6 2 1
> sum(freq)
[1] 20
> freq/sum(freq)
work.hours
2 3 4 5 6 7
0.15 0.25 0.15 0.30 0.10 0.05
We stored the frequencies in an object called “freq”. The content of the object
are the frequencies 3, 5, 3, 6, 2 and 1. The function “sum” sums the components
of its input. The sum of the frequencies is the sample size , the total number of
students that responded to the survey, which is 20. Hence, when we apply the
function “sum” to the object “freq” we get 20 as an output.
The outcome of dividing an object by a number is a division of each ele-
ment in the object by the given number. Therefore, when we divide “freq” by
“sum(freq)” (the number 20) we get a sequence of relative frequencies. The
first entry to this sequence is 3/20 = 0.15, the second entry is 5/20 = 0.25, and
the last entry is 1/20 = 0.05. The sum of the relative frequencies should always
be equal to 1:
18 CHAPTER 2. SAMPLING AND DATA STRUCTURES

> sum(freq/sum(freq))
[1] 1
The cumulative relative frequency is the accumulation of previous relative
frequencies. To find the cumulative relative frequencies, add all the previous
relative frequencies to the relative frequency of the current value. Alternatively,
we may apply the function “cumsum” to the sequence of relative frequencies:
> cumsum(freq/sum(freq))
2 3 4 5 6 7
0.15 0.40 0.55 0.85 0.95 1.00
Observe that the cumulative relative frequency of the smallest value 2 is the
frequency of that value (0.15). The cumulative relative frequency of the second
value 3 is the sum of the relative frequency of the smaller value (0.15) and
the relative frequency of the current value (0.25), which produces a total of
0.15 + 0.25 = 0.40. Likewise, for the third value 4 we get a cumulative relative
frequency of 0.15 + 0.25 + 0.15 = 0.55. The last entry of the cumulative relative
frequency column is one, indicating that one hundred percent of the data has
been accumulated.
The computation of the cumulative relative frequency was carried out with
the aid of the function “cumsum”. This function takes as an input argument a
numerical sequence and produces as output a numerical sequence of the same
length with the cumulative sums of the components of the input sequence.

2.2.4 Critical Evaluation

Inappropriate methods of sampling and data collection may produce samples
that do not represent the target population. A naı̈ve application of statistical
analysis to such data may produce misleading conclusions.
Consequently, it is important to evaluate critically the statistical analyses
we encounter before accepting the conclusions that are obtained as a result of
these analyses. Common problems that occurs in data that one should be aware
of include:
Problems with Samples: A sample should be representative of the popula-
tion. A sample that is not representative of the population is biased.
Biased samples may produce results that are inaccurate and not valid.
Data Quality: Avoidable errors may be introduced to the data via inaccurate
handling of forms, mistakes in the input of data, etc. Data should be
cleaned from such errors as much as possible.
Self-Selected Samples: Responses only by people who choose to respond,
such as call-in surveys, that are often biased.
Sample Size Issues: Samples that are too small may be unreliable. Larger
samples, when possible, are better. In some situations, small samples are
unavoidable and can still be used to draw conclusions. Examples: Crash
testing cars, medical testing for rare conditions.
Undue Influence: Collecting data or asking questions in a way that influences
the response.
2.3. READING DATA INTO R 19

Causality: A relationship between two variables does not mean that one causes
the other to occur. They may both be related (correlated) because of their
relationship to a third variable.

Self-Funded or Self-Interest Studies: A study performed by a person or

organization in order to support their claim. Is the study impartial? Read
the study carefully to evaluate the work. Do not automatically assume
that the study is good but do not automatically assume the study is bad
either. Evaluate it on its merits and the work done.

Misleading Use of Data: Improperly displayed graphs and incomplete data.

Confounding: Confounding in this context means confusing. When the effects

of multiple factors on a response cannot be separated. Confounding makes
it difficult or impossible to draw valid conclusions about the effect of each
factor.

2.3 Reading Data into R

In the examples so far the size of the data set was very small and we were able
to input the data directly into R with the use of the function “c”. In more
practical settings the data sets to be analyzed are much larger and it is very
inefficient to enter them manually. In this section we learn how to upload data
from a file in the Comma Separated Values (CSV) format.
The file “ex1.csv” contains data on the sex and height of 100 individuals.
This file is given in the CSV format. The file can be found on the internet
at http://pluto.huji.ac.il/~msby/StatThink/Datasets/ex1.csv. We will
discuss the process of reading data from a file into R and use this file as an
illustration.

2.3.1 Saving the File and Setting the Working Directory

Before the file is read into R you may find it convenient to obtain a copy of the
file and store it in some directory on the computer and read the file from that
directory. We recommend that you create a special directory in which you keep
all the material associated with this course. In the explanations provided below
we assume that the directory to which the file is stored in called “IntroStat”.
(See Figure 2.1)
Files in the CSV format are ordinary text files. They can be created manually
or as a result of converting data stored in a different format into this particular
format. A convenient way to produce, browse and edit CSV files is by the use
of a standard electronic spreadsheet programs such as Excel or Calc. The Excel
spreadsheet is part of the Microsoft’s Office suite. The Calc spreadsheet is part
of OpenOffice suite that is freely distributed by the OpenOffice Organization.
Opening a CSV file by a spreadsheet program displays a spreadsheet with
the content of the file. Values in the cells of the spreadsheet may be modified
directly. (However, when saving, one should pay attention to save the file in
the CVS format.) Similarly, new CSV files may be created by the entering of
the data in an empty spreadsheet. The first row should include the name of
the variable, preferably as a single character string with no empty spaces. The
20 CHAPTER 2. SAMPLING AND DATA STRUCTURES

Figure 2.1: The File “read.csv”

following rows may contain the data values associated with this variable. When
saving, the spreadsheet should be saved in the CSV format by the use of the
“Save by name” dialog and choosing there the option of CSV in the “Save by
Type” selection.
After saving a file with the data in a directory, R should be notified where
the file is located in order to be able to read it. A simple way of doing so is
by setting the directory with the file as R’s working directory. The working
directory is the first place R is searching for files. Files produced by R are saved
in that directory. In Windows, during an active R session, one may set the
working directory to be some target directory with the “File/Change Dir...”
dialog. This dialog is opened by selecting the option “File” on the left hand
side of the ruler on the top of the R Console window. Selecting the option of
“Change Dir...” in the ruler that opens will start the dialog. (See Figure 2.2.)
Browsing via this dialog window to the directory of choice, selecting it, and
approving the selection by clicking the “OK” bottom in the dialog window will
set the directory of choice as the working directory of R.
Rather than changing the working directory every time that R is opened one
may set a selected directory to be R’s working directory on opening. Again, we
demonstrate how to do this on the XP Windows operating system.
The R icon was added to the Desktop when the R system was installed.
The R Console is opened by double-clicking on this icon. One may change
the properties of the icon so that it sets a directory of choice as R’s working
directory.
In order to do so click on the icon with the mouse’s right bottom. A menu
2.3. READING DATA INTO R 21

Figure 2.2: Changing The Working Directory

opens in which you should select the option “Properties”. As a result, a dialog
window opens. (See Figure 2.3.) Look at the line that starts with the words
“Start in” and continues with a name of a directory that is the current working
directory. The name of this directory is enclosed in double quotes and is given
with it’s full path, i.e. its address on the computer. This name and path should
be changed to the name and path of the directory that you want to fix as the
new working directory.
Consider again Figure 2.1. Imagine that one wants to fix the directory that
contains the file “ex1.csv” as the permanent working directory. Notice that
the full address of the directory appears at the “Address” bar on the top of
the window. One may copy the address and paste it instead of the name of the
current working directory that is specified in the “Properties” dialog of the
R icon. One should make sure that the address to the new directory is, again,
placed between double-quotes. (See in Figure 2.4 the dialog window after the
changing the address of the working directory. Compare this to Figure 2.3 of
the window before the change.) After approving the change by clicking the
“OK” bottom the new working directory is set. Henceforth, each time that the
R Console is opened by double-clicking the icon it will have the designated
directory as its working directory.
In the rest of this book we assume that a designated directory is set as R’s
working directory and that all external files that need to be read into R, such
as “ex1.csv” for example, are saved in that working directory. Once a working
directory has been set then the history of subsequent R sessions is stored in that
directory. Hence, if you choose to save the image of the session when you end
the session then objects created in the session will be uploaded the next time
22 CHAPTER 2. SAMPLING AND DATA STRUCTURES

Figure 2.3: Setting the Working Directory (Before the Change)

Figure 2.4: Setting the Working Directory (After the Change)

2.3. READING DATA INTO R 23

the R Console is opened.

2.3.2 Reading a CSV File into R

Now that a copy of the file “ex1.csv” is placed in the working directory we
would like to read its content into R. Reading of files in the CSV format can be
carried out with the R function “read.csv”. To read the file of the example we
run the following line of code in the R Console window:
> ex.1 <- read.csv("ex1.csv")
The function “read.csv” takes as an input argument the address of a CSV file
and produces as output a data frame object with the content of the file. Notice
that the address is placed between double-quotes. If the file is located in the
working directory then giving the name of the file as an address is sufficient1 .
Consider the content of that R object “ex.1” that was created by the previous
expression:
> ex.1
id sex height
1 5696379 FEMALE 182
2 3019088 MALE 168
3 2038883 MALE 172
4 1920587 FEMALE 154
5 6006813 MALE 174
6 4055945 FEMALE 176
. . . .
. . . .
. . . .
98 9383288 MALE 195
99 1582961 FEMALE 129
100 9805356 MALE 172
>
(Noticed that we have erased the middle rows. In the R Console window you
should obtain the full table. However, in order to see the upper part of the
output you may need to scroll up the window.)
The object “ex.1”, the output of the function “read.csv” is a data frame.
Data frames are the standard tabular format of storing statistical data. The
columns of the table are called variables and correspond to measurements. In
this example the three variables are:
id: A 7 digits number that serves as a unique identifier of the subject.
sex: The sex of each subject. The values are either “MALE” or “FEMALE”.
height: The height (in centimeter) of each subject. A numerical value.
1 If the file is located in a different directory then the complete address, including the path

to the file, should be provided. The file need not reside on the computer. One may provide,
for example, a URL (an internet address) as the address. Thus, instead of saving the file of the
example on the computer one may read its content into an R object by using the line of code
“ex.1 <- read.csv("http://pluto.huji.ac.il/~msby/StatThink/Datasets/ex1.csv")” in-
stead of the code that we provide and the working method that we recommend to follow.
24 CHAPTER 2. SAMPLING AND DATA STRUCTURES

When the values of the variable are numerical we say that it is a quantitative
variable or a numeric variable. On the other hand, if the variable has qualitative
or level values we say that it is a factor. In the given example, sex is a factor
and height is a numeric variable.
The rows of the table are called observations and correspond to the subjects.
In this data set there are 100 subjects, with subject number 1, for example, being
a female of height 182 cm and identifying number 5696379. Subject number 98,
on the other hand, is a male of height 195 cm and identifying number 9383288.

2.3.3 Data Types

The columns of R data frames represent variables, i.e. measurements recorded
for each of the subjects in the sample. R associates with each variable a type
that characterizes the content of the variable. The two major types are
• Factors, or Qualitative Data. The type is “factor”.
• Quantitative Data. The type is “numeric”.
Factors are the result of categorizing or describing attributes of a population.
Hair color, blood type, ethnic group, the car a person drives, and the street a
person lives on are examples of qualitative data. Qualitative data are generally
described by words or letters. For instance, hair color might be black, dark
brown, light brown, blonde, gray, or red. Blood type might be AB+, O-, or
B+. Qualitative data are not as widely used as quantitative data because many
numerical techniques do not apply to the qualitative data. For example, it does
not make sense to find an average hair color or blood type.
Quantitative data are always numbers and are usually the data of choice
because there are many methods available for analyzing such data. Quantitative
data are the result of counting or measuring attributes of a population. Amount
of money, pulse rate, weight, number of people living in your town, and the
number of students who take statistics are examples of quantitative data.
Quantitative data may be either discrete or continuous. All data that are
the result of counting are called quantitative discrete data. These data take
on only certain numerical values. If you count the number of phone calls you
receive for each day of the week, you may get results such as 0, 1, 2, 3, etc.
On the other hand, data that are the result of measuring on a continuous scale
are quantitative continuous data, assuming that we can measure accurately.
Measuring angles in radians may result in the numbers π6 , π3 , π2 , π, 3π4 , etc. If
you and your friends carry backpacks with books in them to school, the numbers
of books in the backpacks are discrete data and the weights of the backpacks
are continuous data.
Example 2.1 (Data Sample of Quantitative Discrete Data). The data are the
number of books students carry in their backpacks. You sample five students.
Two students carry 3 books, one student carries 4 books, one student carries 2
books, and one student carries 1 book. The numbers of books (3, 4, 2, and 1)
are the quantitative discrete data.
Example 2.2 (Data Sample of Quantitative Continuous Data). The data are
the weights of the backpacks with the books in it. You sample the same five
students. The weights (in pounds) of their backpacks are 6.2, 7, 6.8, 9.1, 4.3.
2.4. SOLVED EXERCISES 25

Notice that backpacks carrying three books can have different weights. Weights
are quantitative continuous data because weights are measured.
Example 2.3 (Data Sample of Qualitative Data). The data are the colors of
backpacks. Again, you sample the same five students. One student has a red
backpack, two students have black backpacks, one student has a green backpack,
and one student has a gray backpack. The colors red, black, black, green, and
gray are qualitative data.
The distinction between continuous and discrete numeric data is not reflected
usually in the statistical method that are used in order to analyze the data.
Indeed, R does not distinguish between these two types of numeric data and
store them both as “numeric”. Consequently, we will also not worry about
the specific categorization of numeric data and treat them as one. On the other
hand, emphasis will be given to the difference between numeric and factors data.
One may collect data as numbers and report it categorically. For example,
the quiz scores for each student are recorded throughout the term. At the end
of the term, the quiz scores are reported as A, B, C, D, or F. On the other hand,
one may code categories of qualitative data with numerical values and report
the values. The resulting data should nonetheless be treated as a factor.
As default, R saves variables that contain non-numeric values as factors.
Otherwise, the variables are saved as numeric. The variable type is important
because different statistical methods are applied to different data types. Hence,
one should make sure that the variables that are analyzed have the appropriate
type. Especially that factors using numbers to denote the levels are labeled as
factors. Otherwise R will treat them as quantitative data.

2.4 Solved Exercises

Question 2.1. Consider the following relative frequency table on hurricanes
that have made direct hits on the U.S. between 1851 and 2004 (http://www.
nhc.noaa.gov/gifs/table5.gif). Hurricanes are given a strength category
rating based on the minimum wind speed generated by the storm. Some of the
entries to the table are missing.

Category # Direct Hits Relative Freq. Cum. Relative Freq.

1 109
2 72 0.2637 0.6630
3 0.2601
4 18 0.9890
5 3 0.0110 1.0000

Table 2.1: Frequency of Hurricane Direct Hits

1. What is the relative frequency of direct hits of category 1?

2. What is the relative frequency of direct hits of category 4 or more?

Solution (to Question 2.1.1): The relative frequency of direct hits of cat-
egory 1 is 0.3993. Notice that the cumulative relative frequency of category
26 CHAPTER 2. SAMPLING AND DATA STRUCTURES

1 and 2 hits, the sum of the relative frequency of both categories, is 0.6630.
The relative frequency of category 2 hits is 0.2637. Consequently, the relative
frequency of direct hits of category 1 is 0.6630 - 0.2637 = 0.3993.

Solution (to Question 2.1.2): The relative frequency of direct hits of cate-
gory 4 or more is 0.0769. Observe that the cumulative relative of the value “3”
is 0.6630 + 0.2601 = 0.9231. This follows from the fact that the cumulative
relative frequency of the value “2” is 0.6630 and the relative frequency of the
value “3” is 0.2601. The total cumulative relative frequency is 1.0000. The
relative frequency of direct hits of category 4 or more is the difference between
the total cumulative relative frequency and cumulative relative frequency of 3
hits: 1.0000 - 0.9231 = 0.0769.

Question 2.2. The number of calves that were born to some cows during their
productive years was recorded. The data was entered into an R object by the
name “calves”. Refer to the following R code:

> freq <- table(calves)

> cumsum(freq)
1 2 3 4 5 6 7
4 7 18 28 32 38 45

1. How many cows were involved in this study?

2. How many cows gave birth to a total of 4 calves?

3. What is the relative frequency of cows that gave birth to at least 4 calves?

Solution (to Question 2.2.1): The total number of cows that were involved
in this study is 45. The object “freq” contain the table of frequency of the
cows, divided according to the number of calves that they had. The cumulative
frequency of all the cows that had 7 calves or less, which includes all cows in
the study, is reported under the number “7” in the output of the expression
“cumsum(freq)”. This number is 45.

Solution (to Question 2.2.2): The number of cows that gave birth to a total
of 4 calves is 10. Indeed, the cumulative frequency of cows that gave birth to
4 calves or less is 28. The cumulative frequency of cows that gave birth to 3
calves or less is 18. The frequency of cows that gave birth to exactly 4 calves is
the difference between these two numbers: 28 - 18 = 10.

Solution (to Question 2.2.3): The relative frequency of cows that gave birth
to at least 4 calves is 27/45 = 0.6. Notice that the cumulative frequency of
cows that gave at most 3 calves is 18. The total number of cows is 45. Hence,
the number of cows with 4 or more calves is the difference between these two
numbers: 45 - 18 = 27. The relative frequency of such cows is the ratio between
this number and the total number of cows: 27/45 = 0.6.
2.5. SUMMARY 27

2.5 Summary
Glossary
Population: The collection, or set, of all individuals, objects, or measurements
whose properties are being studied.
Sample: A portion of the population understudy. A sample is representative
if it characterizes the population being studied.
Frequency: The number of times a value occurs in the data.
Relative Frequency: The ratio between the frequency and the size of data.
Cumulative Relative Frequency: The term applies to an ordered set of data
values from smallest to largest. The cumulative relative frequency is the
sum of the relative frequencies for all values that are less than or equal to
the given value.
Data Frame: A tabular format for storing statistical data. Columns corre-
spond to variables and rows correspond to observations.

Variable: A measurement that may be carried out over a collection of subjects.

The outcome of the measurement may be numerical, which produces a
quantitative variable; or it may be non-numeric, in which case a factor is
produced.
Observation: The evaluation of a variable (or variables) for a given subject.

CSV Files: A digital format for storing data frames.

Factor: Qualitative data that is associated with categorization or the descrip-
tion of an attribute.
Quantitative: Data generated by numerical measurements.

Discuss in the forum

Factors are qualitative data that are associated with categorization or the de-
scription of an attribute. On the other hand, numeric data are generated by
numerical measurements. A common practice is to code the levels of factors
using numerical values. What do you think of this practice?
In the formulation of your answer to the question you may think of an
example of factor variable from your own field of interest. You may describe a
benefit or a disadvantage that results from the use of a numerical values to code
the level of this factor.
28 CHAPTER 2. SAMPLING AND DATA STRUCTURES
Chapter 3

Descriptive Statistics

3.1 Student Learning Objectives

This chapter deals with numerical and graphical ways to describe and display
data. This area of statistics is called descriptive statistics. You will learn to
calculate and interpret these measures and graphs. By the end of this chapter,
you should be able to:
• Use histograms and box plots in order to display data graphically.
• Calculate measures of central location: mean and median.
• Calculate measures of the spread: variance, standard deviation, and inter-
quartile range.
• Identify outliers, which are values that do not fit the rest of the distribu-
tion.

3.2 Displaying Data

Once you have collected data, what will you do with it? Data can be described
and presented in many different formats. For example, suppose you are inter-
ested in buying a house in a particular area. You may have no clue about the
house prices, so you may ask your real estate agent to give you a sample data
set of prices. Looking at all the prices in the sample is often overwhelming. A
better way may be to look at the median price and the variation of prices. The
median and variation are just two ways that you will learn to describe data.
Your agent might also provide you with a graph of the data.
A statistical graph is a tool that helps you learn about the shape of the
distribution of a sample. The graph can be a more effective way of presenting
data than a mass of numbers because we can see where data clusters and where
there are only a few data values. Newspapers and the Internet use graphs to
show trends and to enable readers to compare facts and figures quickly.
Statisticians often start the analysis by graphing the data in order to get an
overall picture of it. Afterwards, more formal tools may be applied.
In the previous chapters we used the bar plot, where bars that indicate the
frequencies in the data of values are placed over these values. In this chapter

29
30 CHAPTER 3. DESCRIPTIVE STATISTICS

Histogram of ex.1$height

25
20
Frequency

15
10
5
0

120 140 160 180 200

ex.1$height

Figure 3.1: Histogram of Height

our emphasis will be on histograms and box plots, which are other types of
plots. Some of the other types of graphs that are frequently used, but will not
be discussed in this book, are the stem-and-leaf plot, the frequency polygon
(a type of broken line graph) and the pie charts. The types of plots that will
be discussed and the types that will not are all tightly linked to the notion of
frequency of the data that was introduced in Chapter 2 and intend to give a
graphical representation of this notion.

3.2.1 Histograms
The histogram is a frequently used method for displaying the distribution of
continuous numerical data. An advantage of a histogram is that it can readily
display large data sets. A rule of thumb is to use a histogram when the data
set consists of 100 values or more.
One may produce a histogram in R by the application of the function “hist”
to a sequence of numerical data. Let us read into R the data frame “ex.1” that
contains data on the sex and height and create a histogram of the heights:
> ex.1 <- read.csv("ex1.csv")
3.2. DISPLAYING DATA 31

> hist(ex.1$height)
The outcome of the function is a plot that apears in the graphical window and
is presented in Figure 3.1.
The data set, which is the content of the CSV file “ex1.csv”, was used in
Chapter 2 in order to demonstrate the reading of data that is stored in a external
file into R. The first line of the above script reads in the data from “ex1.csv”
into a data frame object named “ex.1” that maintains the data internally in R.
The second line of the script produces the histogram. We will discuss below the
code associated with this second line.
A histogram consists of contiguous boxes. It has both a horizontal axis and
a vertical axis. The horizontal axis is labeled with what the data represents (the
height, in this example). The vertical axis presents frequencies and is labeled
“Frequency”. By the examination of the histogram one can appreciate the shape
of the data, the center, and the spread of the data.
The histogram is constructed by dividing the range of the data (the x-axis)
into equal intervals, which are the bases for the boxes. The height of each box
represents the count of the number of observations that fall within the interval.
For example, consider the box with the base between 160 and 170. There is a
total of 19 subjects with height larger that 160 but no more than 170 (that is,
160 < height ≤ 170). Consequently, the height of that box1 is 19.
The input to the function “hist” should be a sequence of numerical values.
In principle, one may use the function “c” to produce a sequence of data and
apply the histogram plotting function to the output of the sequence producing
function. However, in the current case we have already the data stored in the
data frame “ex.1”, all we need to learn is how to extract that data so it can be
used as input to the function “hist” that plots the histogram.
Notice the structure of the input that we have used in order to construct
the histogram of the variable “height” in the “ex.1” data frame. One may
address the variable “variable.name” in the data frame “dataframe.name”
using the format: “dataframe.name$variable.name”. Indeed, when we type
the expression “ex.1$height” we get as an output the values of the variable
“height” from the given data frame:
> ex.1$height
[1] 182 168 172 154 174 176 193 156 157 186 143 182 194 187 171
[16] 178 157 156 172 157 171 164 142 140 202 176 165 176 175 170
[31] 169 153 169 158 208 185 157 147 160 173 164 182 175 165 194
[46] 178 178 186 165 180 174 169 173 199 163 160 172 177 165 205
[61] 193 158 180 167 165 183 171 191 191 152 148 176 155 156 177
[76] 180 186 167 174 171 148 153 136 199 161 150 181 166 147 168
[91] 188 170 189 117 174 187 141 195 129 172
This is a numeric sequence and can serve as the input to a function that expects a
numeric sequence as input, a function such as “hist”. (But also other functions,
for example, “sum” and “cumsum”.)
1 In some books an histogram is introduced as a form of a density. In densities the area of

the box represents the frequency or the relative frequency. In the current example the height
would have been 19/10 = 1.9 if the area of the box would have represented the frequency
and it would have been (19/100)/10 = 0.019 if the area of the box would have represented
the relative frequency. However, in this book we follow the default of R in which the height
represents the frequency.
32 CHAPTER 3. DESCRIPTIVE STATISTICS

There are 100 observations in the variable “ex.1$height”. So many ob-

servations cannot be displayed on the screen on one line. Consequently, the
sequence of the data is wrapped and displayed over several lines. Notice that
the square brackets on the left hand side of each line indicate the position in
the sequence of the first value on that line. Hence, the number on the first
line is “[1]”. The number on the second line is “[16]”, since the second line
starts with the 16th observation in the display given in the book. Notice, that
numbers in the square brackets on your R Console window may be different,
depending on the setting of the display on your computer.

3.2.2 Box Plots

The box plot, or box-whisker plot, gives a good graphical overall impression of
the concentration of the data. It also shows how far from most of the data the
extreme values are. In principle, the box plot is constructed from five values: the
smallest value, the first quartile, the median, the third quartile, and the largest
value. The median, the first quartile, and the third quartile will be discussed
here, and then once more in the next section.
The median, a number, is a way of measuring the “center” of the data. You
can think of the median as the “middle value,” although it does not actually
have to be one of the observed values. It is a number that separates ordered
data into halves. Half the values are the same size or smaller than the median
and half the values are the same size or larger than it. For example, consider
the following data that contains 14 values:
1, 11.5, 6, 7.2, 4, 8, 9, 10, 6.8, 8.3, 2, 2, 10, 1 .
Ordered, from smallest to largest, we get:
1, 1, 2, 2, 4, 6, 6.8, 7.2, 8, 8.3, 9, 10, 10, 11.5 .
The median is between the 7th value, 6.8, and the 8th value 7.2. To find the
median, add the two values together and divide by 2:
6.8 + 7.2
=7
2
The median is 7. Half of the values are smaller than 7 and half of the values
are larger than 7.
Quartiles are numbers that separate the data into quarters. Quartiles may
or may not be part of the data. To find the quartiles, first find the median or
second quartile. The first quartile is the middle value of the lower half of the
data and the third quartile is the middle value of the upper half of the data.
For illustration consider the same data set from above:
1, 1, 2, 2, 4, 6, 6.8, 7.2, 8, 8.3, 9, 10, 10, 11.5 .
The median or second quartile is 7. The lower half of the data is:
1, 1, 2, 2, 4, 6, 6.8 .
The middle value of the lower half is 2. The number 2, which is part of the data
in this case, is the first quartile which is denoted Q1. One-fourth of the values
are the same or less than 2 and three-fourths of the values are more than 2.
3.2. DISPLAYING DATA 33

10
8
6
4
2

Figure 3.2: Box Plot of the Example

The upper half of the data is:

7.2, 8, 8.3, 9, 10, 10, 11.5

The middle value of the upper half is 9. The number 9 is the third quartile
which is denoted Q3. Three-fourths of the values are less than 9 and one-fourth
of the values2 are more than 9.
Outliers are values that do not fit with the rest of the data and lie outside of
the normal range. Data points with values that are much too large or much too
small in comparison to the vast majority of the observations will be identified
as outliers. In the context of the construction of a box plot we identify potential
outliers with the help of the inter-quartile range (IQR). The inter-quartile range
is the distance between the third quartile (Q3) and the first quartile (Q1), i.e.,
IQR = Q3 − Q1. A data point that is larger than the third quartile plus 1.5
times the inter-quartile range will be marked as a potential outlier. Likewise,
a data point smaller than the first quartile minus 1.5 times the inter-quartile
2 The actual computation in R of the first quartile and the third quartile may vary slightly

from the description given here, depending on the exact structure of the data.
34 CHAPTER 3. DESCRIPTIVE STATISTICS

range will also be so marked. Outliers may have a substantial effect on the
outcome of statistical analysis, therefore it is important that one is alerted to
the presence of outliers.
In the running example we obtained an inter-quartile range of size 9-2=7.
The upper threshold for defining an outlier is 9 + 1.5 × 7 = 19.5 and the lower
threshold is 2 − 1.5 × 7 = −8.5. All data points are within the two thresholds,
hence there are no outliers in this data.
In the construction of a box plot one uses a vertical rectangular box and two
vertical “whiskers” that extend from the ends of the box to the smallest and
largest data values that are not outliers. Outlier values, if any exist, are marked
as points above or blow the endpoints of the whiskers. The smallest and largest
non-outlier data values label the endpoints of the axis. The first quartile marks
one end of the box and the third quartile marks the other end of the box. The
central 50% of the data fall within the box.
One may produce a box plot with the aid of the function “boxplot”. The
input to the function is a sequence of numerical values and the output is a plot.
As an example, let us produce the box plot of the 14 data points that were used
as an illustration:

> boxplot(c(1,11.5,6,7.2,4,8,9,10,6.8,8.3,2,2,10,1))

The resulting box plot is presented in Figure 3.2. Observe that the end
points of the whiskers are 1, for the minimal value, and 11.5 for the largest
value. The end values of the box are 9 for the third quartile and 2 for the first
quartile. The median 7 is marked inside the box.
Next, let us examine the box plot for the height data:

> boxplot(ex.1$height)

The resulting box plot is presented in Figure 3.3. In order to assess the plot let
us compute quartiles of the variable:

> summary(ex.1$height)
Min. 1st Qu. Median Mean 3rd Qu. Max.
117.0 158.0 171.0 170.1 180.2 208.0

The function “summary”, when applied to a numerical sequence, produce the

minimal and maximal entries, as well the first, second and third quartiles (the
second is the Median). It also computes the average of the numbers (the Mean),
which will be discussed in the next section.
Let us compare the results with the plot in Figure 3.3. Observe that the
median 171 coincides with the thick horizontal line inside the box and that the
lower end of the box coincides with first quartile 158.0 and the upper end with
180.2, which is the third quartile. The inter-quartile range is 180.2 − 158.0 =
22.2. The upper threshold is 180.2 + 1.5 × 22.2 = 213.5. This threshold is
larger than the largest observation (208.0). Hence, the largest observation is
not an outlier and it marks the end of the upper whisker. The lower threshold
is 158.0 − 1.5 × 22.2 = 124.7. The minimal observation (117.0) is less than this
threshold. Hence it is an outlier and it is marked as a point below the end of the
lower whisker. The second smallest observation is 129. It lies above the lower
threshold and it marks the end point of the lower whisker.
3.3. MEASURES OF THE CENTER OF DATA 35

200
180
160
140
120

Figure 3.3: Box Plot of Height

3.3 Measures of the Center of Data

The two most widely used measures of the central location of the data are the
mean (average) and the median. To calculate the average weight of 50 people
one should add together the 50 weights and divide the result by 50. To find
the median weight of the same 50 people, one may order the data and locate
a number that splits the data into two equal parts. The median is generally a
better measure of the center when there are extreme values or outliers because
it is not affected by the precise numerical values of the outliers. Nonetheless,
the mean is the most commonly used measure of the center.
We shall use small Latin letters such as x to mark the sequence of data.
In such a case we may mark the sample mean by placing a bar over the x: x̄
(pronounced “x bar”).
The mean can be calculated by averaging the data points or it also can be
calculated with the relative frequencies of the values that are present in the data.
In the latter case one multiplies each distinct value by its relative frequency and
then sum the products across all values. To see that both ways of calculating
36 CHAPTER 3. DESCRIPTIVE STATISTICS

Histogram of x

6
Frequency

4
2
0

4 5 6 7 8 9 10

Histogram of x
6
Frequency

4
2
0

4 5 6 7 8 9 10

Histogram of x
6
Frequency

4
2
0

4 5 6 7 8 9 10

Figure 3.4: Three Histograms

the mean are the same, consider the data:

1, 1, 1, 2, 2, 3, 4, 4, 4, 4, 4 .

In the first way of calculating the mean we get:

1+1+1+2+2+3+4+4+4+4+4
x̄ = = 2.7 .
11
Alternatively, we may note that the distinct values in the sample are 1, 2, 3,
and 4 with relative frequencies of 3/11, 2/11, 1/11 and 5/11, respectively. The
alternative method of computation produces:
3 2 1 5
x̄ = 1 × +2× +3× +4× = 2.7 .
11 11 11 11

3.3.1 Skewness, the Mean and the Median

Consider the following data set:

4, 5, 6, 6, 6, 7, 7, 7, 7, 7, 7, 8, 8, 8, 9, 10
3.3. MEASURES OF THE CENTER OF DATA 37

This data produces the upper most histogram in Figure 3.4. Each interval has
width one and each value is located at the middle of an interval. The histogram
displays a symmetrical distribution of data. A distribution is symmetrical if a
vertical line can be drawn at some point in the histogram such that the shape
to the left and to the right of the vertical line are mirror images of each other.
Let us compute the mean and the median of this data:

> x <- c(4,5,6,6,6,7,7,7,7,7,7,8,8,8,9,10)

> mean(x)
[1] 7
> median(x)
[1] 7

The mean and the median are each 7 for these data. In a perfectly symmetrical
distribution, the mean and the median are the same3 .
The functions “mean” and “median” were used in order to compute the mean
and median. Both functions expect a numeric sequence as an input and produce
the appropriate measure of centrality of the sequence as an output.
The histogram for the data:

4, 5, 6, 6, 6, 7, 7, 7, 7, 7, 7, 8

is not symmetrical and is displayed in the middle of Figure 3.4. The right-hand
side seems “chopped off” compared to the left side. The shape of the distribution
is called skewed to the left because it is pulled out towards the left.
Let us compute the mean and the median for this data:

> x <- c(4,5,6,6,6,7,7,7,7,7,7,8)

> mean(x)
[1] 6.416667
> median(x)
[1] 7

(Notice that the original data is replaced by the new data when object x is
reassigned.) The median is still 7, but the mean is less than 7. The relation
between the mean and the median reflects the skewing.
Consider yet another set of data:

6, 7, 7, 7, 7, 7, 7, 8, 8, 8, 9, 10

The histogram for the data is also not symmetrical and is displayed at the
bottom of Figure 3.4. Notice that it is skewed to the right. Compute the mean
and the median:

> x <- c(6,7,7,7,7,7,7,8,8,8,9,10)

> mean(x)
[1] 7.583333
> median(x)
[1] 7
3 In the case of a symmetric distribution the vertical line of symmetry is located at the

mean, which is also equal to the median.

38 CHAPTER 3. DESCRIPTIVE STATISTICS

The median is yet again equal to 7, but this time the mean is greater than 7.
Again, the mean reflects the skewing.
In summary, if the distribution of data is skewed to the left then the mean
is less than the median. If the distribution of data is skewed to the right then
the median is less than the mean.
Examine the data on the height in “ex.1”:
> mean(ex.1$height)
[1] 170.11
> median(ex.1$height)
[1] 171
Observe that the histogram of the height (Figure 3.1) is skewed to the left. This
is consistent with the fact that the mean is less than the median.

3.4 Measures of the Spread of Data

One measure of the spread of the data is the inter-quartile range that was
introduced in the context of the box plot. However, the most important measure
of spread is the standard deviation.
Before dealing with the standard deviation let us discuss the calculation of
the variance. If xi is a data value for subject i and x̄ is the sample mean,
then xi − x̄ is called the deviation of subject i from the mean, or simply the
deviation. In a data set, there are as many deviations as there are data values.
The variance is in principle the average of the squares of the deviations.
Consider the following example: In a fifth grade class, the teacher was inter-
ested in the average age and the standard deviation of the ages of her students.
Here are the ages of her students to the nearest half a year:
9, 9.5, 9.5, 10, 10, 10, 10, 10.5, 10.5, 10.5, 10.5, 11, 11, 11, 11, 11, 11,
11.5, 11.5, 11.5 .
In order to explain the computation of the variance of these data let us create
an object x that contains the data:
> x <- c(9,9.5,9.5,10,10,10,10,10.5,10.5,10.5,10.5,11,11,11,11,11,
+ 11,11.5,11.5,11.5)
> length(x)
[1] 20
Pay attention to the fact that we did not write the “+” at the beginning of the
second line. That symbol was produced by R when moving to the next line to
indicate that the expression is not complete yet and will not be executed. Only
after inputting the right bracket and the hitting of the Return key does R carry
out the command and creates the object “x”. When you execute this example
yourself on your own computer make sure not to copy the “+” sign. Instead, if
you hit the return key after the last comma on the first line, the plus sign will
be produced by R as a new prompt and you can go on typing in the rest of the
numbers.
The function “length” returns the length of the input sequence. Notice that
we have a total of 20 data points.
The next step involves the computation of the deviations:
3.4. MEASURES OF THE SPREAD OF DATA 39

> x.bar <- mean(x)

> x.bar
[1] 10.525
> x - x.bar
[1] -1.525 -1.025 -1.025 -0.525 -0.525 -0.525 -0.525 -0.025
[9] -0.025 -0.025 -0.025 0.475 0.475 0.475 0.475 0.475
[17] 0.475 0.975 0.975 0.975
The average of the observations is equal to 10.525 and when we delete this
number from each of the components of the sequence x we obtain the deviations.
For example, the first deviation is obtained as 9 - 10.525 = -1.525, the second
deviation is 9.5 - 10.525 = -1.025, and so forth. The 20th deviation is 11.5 -
10.525 = 0.975, and this is the last number that is presented in the output.
From a more technical point of view observe that the expression that com-
puted the deviations, “x - x.bar”, involved the deletion of a single value
(x.bar) from a sequence with 20 values (x). The expression resulted in the
deletion of the value from each component of the sequence. This is an example
of the general way by which R operates on sequences. The typical behavior of
R is to apply an operation to each component of the sequence.
As yet another illustration of this property consider the computation of the
squares of the deviations:
> (x - x.bar)^2
[1] 2.325625 1.050625 1.050625 0.275625 0.275625 0.275625
[7] 0.275625 0.000625 0.000625 0.000625 0.000625 0.225625
[13] 0.225625 0.225625 0.225625 0.225625 0.225625 0.950625
[19] 0.950625 0.950625

Recall that “x - x.bar” is a sequence of length 20. We apply the square func-
tion to this sequence. This function is applied to each of the components of the
sequence. Indeed, for the first component we have that (−1.525)2 = 2.325625,
for the second component (−1.025)2 = 1.050625, and for the last component
(0.975)2 = 0.950625.
For the variance we sum the square of the deviations and divide by the total
number of data values minus one (n − 1). The standard deviation is obtained
by taking the square root of the variance:
> sum((x - x.bar)^2)/(length(x)-1)
[1] 0.5125
> sqrt(sum((x - x.bar)^2)/(length(x)-1))
[1] 0.715891
If the variance is produced as a result of dividing the sum of squares by the
number of observations minus one then the variance is called the sample vari-
ance.
The function “var” computes the sample variance and the function “sd”
computes the standard deviations. The input to both functions is the sequence
of data values and the outputs are the sample variance and the standard devi-
ation, respectively:
> var(x)
[1] 0.5125
40 CHAPTER 3. DESCRIPTIVE STATISTICS

> sd(x)
[1] 0.715891

In the computation of the variance we divide the sum of squared deviations

by the number of deviations minus one and not by the number of deviations.
The reason for that stems from the theory of statistical inference that will be
discussed in Part II of this book. Unless the size of the data is small, dividing
by n or by n − 1 does not introduce much of a difference.
The variance is a squared measure and does not have the same units as
the data. Taking the square root solves the problem. The standard deviation
measures the spread in the same units as the data.
The sample standard deviation, s, is either zero or is larger than zero. When
s = 0, there is no spread and the data values are equal to each other. When s
is a lot larger than zero, the data values are very spread out about the mean.
Outliers can make s very large.
The standard deviation is a number that measures how far data values are
from their mean. For example, if the data contains the value 7 and if the mean
of the data is 5 and the standard deviation is 2, then the value 7 is one standard
deviation from its mean because 5 + 1 × 2 = 7. We say, then, that 7 is one
standard deviation larger than the mean 5 (or also say “to the right of 5”). If
the value 1 was also part of the data set, then 1 is two standard deviations
smaller than the mean (or two standard deviations to the left of 5) because
5 − 2 × 2 = 1.
The standard deviation, when first presented, may not be too simple to
interpret. By graphing your data, you can get a better “feel” for the deviations
and the standard deviation. You will find that in symmetrical distributions, the
standard deviation can be very helpful but in skewed distributions, the standard
deviation is less so. The reason is that the two sides of a skewed distribution
have different spreads. In a skewed distribution, it is better to look at the first
quartile, the median, the third quartile, the smallest value, and the largest value.

3.5 Solved Exercises

Question 3.1. Three sequences of data were saved in 3 R objects named “x1”,
“x2” and “x3”, respectively. The application of the function “summary” to each
of these objects is presented below:

> summary(x1)
Min. 1st Qu. Median Mean 3rd Qu. Max.
0.000 2.498 3.218 3.081 3.840 4.871
> summary(x2)
Min. 1st Qu. Median Mean 3rd Qu. Max.
0.0001083 0.5772000 1.5070000 1.8420000 2.9050000 4.9880000
> summary(x3)
Min. 1st Qu. Median Mean 3rd Qu. Max.
2.200 3.391 4.020 4.077 4.690 6.414

In Figure 3.5 one may find the histograms of these three data sequences, given
in a random order. In Figure 3.6 one may find the box plots of the same data,
given in yet a different order.
3.5. SOLVED EXERCISES 41

0 1 2 3 4 5

Histogram 1

2 3 4 5 6

Histogram 2

0 1 2 3 4 5

Histogram 3

Figure 3.5: Three Histograms

1. Match the summary result with the appropriate histogram and the appro-
priate box plot.

2. Is the value 0.000 in the sequence “x1” an outlier?

3. Is the value 6.414 in the sequence “x3” an outlier?

Solution (to Question 3.1.1): Consider the data “x1”. From the summary
we see that it is distributed in the range between 0 and slightly below 5. The
central 50% of the distribution are located between 2.5 and 3.8. The mean and
median are approximately equal to each other, which suggests an approximately
symmetric distribution. Consider the histograms in Figure 3.5. Histograms 1
and 3 correspond to a distributions in the appropriate range. However, the
distribution in Histogram 3 is concentrated in lower values than suggested by
the given first and third quartiles. Consequently, we match the summary of
“x1” with Histograms 1.
Consider the data “x2”. Again, the distribution is in the range between 0 and
slightly below 5. The central 50% of the distribution are located between 0.6 and
1.8. The mean is larger than the median, which suggests a distribution skewed
42 CHAPTER 3. DESCRIPTIVE STATISTICS

6
5
4
3
2
1

●
●
●
0

Boxplot 1 Boxplot 2 Boxplot 3

Figure 3.6: Three Box Plots

to the right. Therefore, we match the summary of “x2” with Histograms 3.

For the data in “x3” we may note that the distribution is in the range
between 2 and 6. The histogram that fits this description is Histograms 2.
The box plot is essentially a graphical representation of the information pre-
sented by the function “summary”. Following the rational of matching the sum-
mary with the histograms we may obtain that Histogram 1 should be matched
with Box-plot 2 in Figure 3.6, Histogram 2 matches Box-plot 3, and Histogram 3
matches Box-plot 1. Indeed, it is easier to match the box plots with the sum-
maries. However, it is a good idea to practice the direct matching of histograms
with box plots.

Solution (to Question 3.1.2): The data in “x1” fits Box-plot 2 in Figure 3.6.
The value 0.000 is the smallest value in the data and it corresponds to the
smallest point in the box plot. Since this point is below the bottom whisker it
follows that it is an outlier. More directly, we may note that the inter-quartile
range is equal to IQR = 3.840 − 2.498 = 1.342. The lower threshold is equal to
2.498 − 1.5 × 1.342 = 0.485, which is larger that the given value. Consequently,
the given value 0.000 is an outlier.
3.5. SOLVED EXERCISES 43

Solution (to Question 3.1.3): Observe that the data in “x3” fits Box-plot 3
in Figure 3.6. The vale 6.414 is the largest value in the data and it corresponds
to the endpoint of the upper whisker in the box plot and is not an outlier.
Alternatively, we may note that the inter-quartile range is equal to IQR =
4.690 − 3.391 = 1.299. The upper threshold is equal to 4.690 + 1.5.299 = 6.6385,
which is larger that the given value. Consequently, the given value 6.414 is not
an outlier.

Question 3.2. The number of toilet facilities in 30 buildings were counted.

The results are recorded in an R object by the name “x”. The frequency table
of the data “x” is:

> table(x)
x
2 4 6 8 10
10 6 10 2 2

1. What is the mean (x̄) of the data?

2. What is the sample standard deviation of the data?

3. What is the median of the data?

4. What is the inter-quartile range (IQR) of the data?

5. How many standard deviations away from the mean is the value 10?

Solution (to Question 3.2.1): In order to compute the mean of the data we
may write the following simple R code:

> x.val <- c(2,4,6,8,10)

> freq <- c(10,6,10,2,2)
> rel.freq <- freq/sum(freq)
> x.bar <- sum(x.val*rel.freq)
> x.bar
[1] 4.666667

We created an object “x.val” that contains the unique values of the data
and an object “freq” that contains the frequencies of the values. The object
“rel.freq” contains the relative frequencies, the ratios between the frequencies
and the number of observations. The average is computed as the sum of the
products of the values with their relative frequencies. It is stored in the objects
“x.bar” and obtains the value 4.666667.
An alternative approach is to reconstruct the original data from the fre-
quency table. A simple trick that will do the job is to use the function “rep”.
The first argument to this function is a sequence of values. If the second argu-
ment is a sequence of the same length that contains integers then the output
will be composed of a sequence that contains the values of the first sequence,
each repeated a number of times indicated by the second argument. Specifically,
if we enter to this function the unique value “x.val” and the frequency of the
values “freq” then the output will be the sequence of values of the original
sequence “x”:
44 CHAPTER 3. DESCRIPTIVE STATISTICS

> x <- rep(x.val,freq)

> x
[1] 2 2 2 2 2 2 2 2 2 2 4 4 4 4 4 4 6 6 6
[20] 6 6 6 6 6 6 6 8 8 10 10
> mean(x)
[1] 4.666667

Observe that when we apply the function “mean” to “x” we get again the value
4.666667.

Solution (to Question 3.2.2): In order to compute the sample standard

deviation we may compute first the sample variance and then take the square
root of the result:

> var.x <- sum((x.val-x.bar)^2*freq)/(sum(freq)-1)

> sqrt(var.x)
[1] 2.425914

Notice that the expression “sum((x.val-x.bar)^2*freq)” compute the sum

of square deviations. The expression “(sum(freq)-1)” produces the number of
observations minus 1 (n − 1). The ratio of the two gives the sample variance.
Alternatively, had we produced the object “x” that contains the data, we
may apply the function “sd” to get the sample standard deviation:

> sd(x)
[1] 2.425914

Observe that in both forms of computation we obtain the same result: 2.425914.

Solution (to Question 3.2.3): In order to compute the median one may
produce the table of cumulative relative frequencies of “x”:

> data.frame(x.val,cumsum(rel.freq))
x.val cumsum.rel.freq.
1 2 0.3333333
2 4 0.5333333
3 6 0.8666667
4 8 0.9333333
5 10 1.0000000

Recall that the object “x.val” contains the unique values of the data. The
expression “cumsum(rel.freq)” produces the cumulative relative frequencies.
The function “data.frame” puts these two variables into a single data frame
and provides a clearer representation of the results.
Notice that more that 50% of the observations have value 4 or less. However,
strictly less than 50% of the observations have value 2 or less. Consequently,
the median is 4. (If the value of the cumulative relative frequency at 4 would
have been exactly 50% then the median would have been the average between
4 and the value larger than 4.)
In the case that we produce the values of the data “x” then we may apply
the function “summary” to it and obtain the median this way
3.6. SUMMARY 45

> summary(x)
Min. 1st Qu. Median Mean 3rd Qu. Max.
2.000 2.000 4.000 4.667 6.000 10.000

Solution (to Question 3.2.4): As for the inter-quartile range (IQR) notice
that the first quartile is 2 and the third quartile is 6. Hence, the inter-quartile
range is equal to 6 - 2 = 4. The quartiles can be read directly from the output
of the function “summary” or can be obtained from the data frame of the cu-
mulative relative frequencies. For the later observe that more than 25% of the
data are less or equal to 2 and more 75% of the data are less or equal to 6 (with
strictly less than 75% less or equal to 4).

Solution (to Question 3.2.5): In order to answer the last question we conduct
the computation: (10 − 4.666667)/2.425914 = 2.198484. We conclude that the
value 10 is approximately 2.1985 standard deviations above the mean.

3.6 Summary
Glossary
Median: A number that separates ordered data into halves: half the values are
the same number or smaller than the median and half the values are the
same number or larger than the median. The median may or may not be
part of the data.
Quartiles: The numbers that separate the data into quarters. Quartiles may
or may not be part of the data. The second quartile is the median of the
data.
Outlier: An observation that does not fit the rest of the data.
Interquartile Range (IQR) : The distance between the third quartile (Q3)
and the first quartile (Q1). IQR = Q3 - Q1.
Mean: A number that measures the central tendency. A common name for
mean is ‘average.’ The term ‘mean’ is a shortened form of ‘arithmetic
mean.’ By definition, the mean for a sample (denoted by x̄) is
Sum of all values in the sample
x̄ = .
Number of values in the sample
(Sample) Variance: Mean of the squared deviations from the mean. Square
of the standard deviation. For a set of data, a deviation can be represented
as x − x̄ where x is a value of the data and x̄ is the sample mean. The
sample variance is equal to the sum of the squares of the deviations divided
by the difference of the sample size and 1:
Sum of the squares of the deviations
s2 = .
Number of values in the sample − 1
(Sample) Standard Deviation: A number that is equal to the square root
of the
√ variance and measures how far data values are from their mean.
s = s2 .
46 CHAPTER 3. DESCRIPTIVE STATISTICS

Discuss in the forum

An important practice is to check the validity of any data set that you are
supposed to analyze in order to detect errors in the data and outlier observations.
Recall that outliers are observations with values outside the normal range of
values of the rest of the observations.
It is said by some that outliers can help us understand our data better. What
is your opinion?
When forming your answer to this question you may give an example of how
outliers may provide insight or, else, how they may abstract our understanding.
For example, consider the price of a stock that tend to go up or go down at
most 2% within each trading day. A sudden 5% drop in the price of the stock
may be an indication to reconsidering our position with respect to this stock.

Commonly Used Symbols

P
• The symbol means to add or to find the sum.
• n = the number of data values in a sample.

• x̄ = the sample mean.

• s = the sample standard deviation.
• f = frequency.
• f /n = relative frequency.

• x = numerical value.

Commonly Used Expressions

• x × (fx /n) = A value multiplied by its respective relative frequency.
Pn
• i=1 xi = The sum of the data values.
P
• x (x × fx /n)= The sum of values multiplied by their respective relative
frequencies.
• x − x̄ = Deviations from the mean (how far a value is from the mean).

• (x − x̄)2 = Deviations squared.

Formulas:
1
Pn P

• Mean: x̄ = n i=1 xi = x x × (fx /n)
1
Pn n


• Variance: s2 = − x̄)2 = (x − x̄)2 × (fx /n)
P
n−1 i=1 (xi n−1 x
√
• Standard Deviation: s = s2
Chapter 4

Probability

4.1 Student Learning Objective

This section extends the notion of variability that was introduced in the con-
text of data to other situations. The variability of the entire population and
the concept of a random variable is discussed. These concepts are central for
the development and interpretation of statistical inference. By the end of the
chapter the student should:
• Consider the distribution of a variable in a population and compute pa-
rameters of this distribution, such as the mean and the standard deviation.
• Become familiar with the concept of a random variable.
• Understand the relation between the distribution of the population and
the distribution of a random variable produced by sampling a random
subject from the population.
• Identify the distribution of the random variable in simple settings and
compute its expectation and variance.

4.2 Different Forms of Variability

In the previous chapters we examined the variability in data. In the statisti-
cal context, data is obtained by selecting a sample from the target population
and measuring the quantities of interest for the subjects that belong to the
sample. Different subjects in the sample may obtain different values for the
measurement, leading to variability in the data.
This variability may be summarized with the aid of a frequency table, a
table of relative frequency, or via the cumulative relative frequency. A graphical
display of the variability in the data may be obtained with the aid of the bar
plot, the histogram, or the box plot.
Numerical summaries may be computed in order to characterize the main
features of the variability. We used the mean and the median in order to identify
the location of the distribution. The sample variance, or better yet the sample
standard deviation, as well as the inter-quartile range were all described as tools
to quantify the overall spread of the data.

47
48 CHAPTER 4. PROBABILITY

The aim of all these graphical representations and numerical summaries is

to investigate the variability of the data.
The subject of this chapter is to introduce two other forms of variability,
variability that is not associated, at least not directly, with the data that we
observe. The first type of variability is the population variability. The other
type of variability is the variability of a random variable.
The notions of variability that will be presented are abstract, they are not
given in terms of the data that we observe, and they have a mathematical-
theoretical flavor to them. At first, these abstract notions may look to you
as a waste of your time and may seem to be unrelated to the subject matter
of the course. The opposite is true. The very core of statistical thinking is
relating observed data to theoretical and abstract models of a phenomena. Via
this comparison, and using the tools of statistical inference that are presented
in the second half of the book, statisticians can extrapolate insights or make
statements regarding the phenomena on the basis of the observed data. Thereby,
the abstract notions of variability that are introduced in this chapter, and are
extended in the subsequent chapters up to the end of this part of the book, are
the essential foundations for the practice of statistics.
The first notion of variability is the variability that is associated with the
population. It is similar in its nature to the variability of the data. The dif-
ference between these two types of variability is that the former corresponds to
the variability of the quantity of interest across all members of the population
and not only for those that were selected to the sample.
In Chapters 2 and 3 we examined the data set “ex.1” which contained data
on the sex and height of a sample of 100 observations. In this chapter we will
consider the sex and height of all the members of the population from which the
sample was selected. The size of the relevant population is 100,000, including
the 100 subjects that composed the sample. When we examine the values of the
height across the entire population we can see that different people may have
different heights. This variability of the heights is the population variability.
The other abstract type of variability, the variability of a random variable,
is a mathematical concept. The aim of this concept is to model the notion
of randomness in measurements or the uncertainty regarding the outcome of a
measurement. In particular we will initially consider the variability of a random
variable in the context of selecting one subject at random from the population.
Imagine we have a population of size 100,000 and we are about to select at
random one subject from this population. We intend to measure the height of
the subject that will be selected. Prior to the selection and measurement we are
not certain what value of height will be obtained. One may associate the notion
of variability with uncertainty — different subjects to be selected may obtain
different evaluations of the measurement and we do not know before hand which
subject will be selected. The resulting variability is the variability of a random
variable.
Random variables can be defined for more abstract settings. Their aim is to
provide models for randomness and uncertainty in measurements. Simple ex-
amples of such abstract random variables will be provided in this chapter. More
examples will be introduced in the subsequent chapters. The more abstract ex-
amples of random variables need not be associated with a specific population.
Still, the same definitions that are used for the example of a random variable
that emerges as a result of sampling a single subject form a population will
4.3. A POPULATION 49

apply to the more abstract constructions.

All types of variability, the variability of the data we dealt with before as
well as the other two types of variability, can be displayed using graphical tools
and characterized with numerical summaries. Essentially the same type of plots
and numerical summaries, possibly with some modifications, may and will be
applied.
A point to remember is that the variability of the data relates to a concrete
list of data values that is presented to us. In contrary to the case of the variabil-
ity of the data, the other types of variability are not associated with quantities
we actually get to observe. The data for the sample we get to see but not the
data for the rest of the population. Yet, we can still discuss the variability of
a population that is out there, even though we do not observe the list of mea-
surements for the entire population. (The example that we give in this chapter
of a population was artificially constructed and serves for illustration only. In
the actual statistical context one does not obtain measurements from the entire
population, only from the subjects that went into the sample.) The discussion
of the variability in this context is theoretical in its nature. Still, this theoretical
discussion is instrumental for understanding statistics.

4.3 A Population
In this section we introduce the variability of a population and present some
numerical summaries that characterizes this variability. Before doing so, let us
review with the aid of an example some of the numerical summaries that were
used for the characterization of the variability of data.
Recall the file “ex1.csv” that contains data on the height and sex of 100
subjects. (The data file can be obtained from http://pluto.huji.ac.il/
~msby/StatThink/Datasets/ex1.csv.) We read the content of the file into a
data frame by the name “ex.1” and apply the function “summary” to the data
frame:

> ex.1 <- read.csv("ex1.csv")

> summary(ex.1)
id sex height
Min. :1538611 FEMALE:54 Min. :117.0
1st Qu.:3339583 MALE :46 1st Qu.:158.0
Median :5105620 Median :171.0
Mean :5412367 Mean :170.1
3rd Qu.:7622236 3rd Qu.:180.2
Max. :9878130 Max. :208.0

We saw in the previous chapter that, when applied to a numeric sequence, the
function “summary” produces the smallest and largest values in the sequence,
the three quartiles (including the median) and the mean. If the input of the
same function is a factor then the outcome is the frequency in the data of each
of the levels of the factor. Here “sex” is a factor with two levels. From the
summary we can see that 54 of the subjects in the sample are female and 46 are
male.
Notice that when the input to the function “summary” is a data frame, as is
the case in this example, then the output is a summary of each of the variables
50 CHAPTER 4. PROBABILITY

3500
3000
2500
2000
Frequency

1500
1000
500
0

117 127 137 147 157 167 177 187 197 207 217

height

Figure 4.1: Bar Plot of Height

of the data frame. In this example two of the variables are numeric (“id” and
“height”) and one variable is a factor (“sex”).
Recall that the mean is the arithmetic average of the data which is computed
by summing all the values of the variable and dividing the result by the number
of observations. Hence, if n is the number of observations (n = 100 in this
example) and xi is the value of the variable for subject i, then one may write
the mean in a formula form as
Pn
Sum of all values in the data xi
x̄ = = i=1 ,
Number of values in the data n
Pn
where x̄ corresponds to the mean of the data and the symbol “ i=1 xi ” corre-
sponds to the sum of all values in the data.
The median is computed by ordering the data values and selecting a value
that splits the ordered data into two equal parts. The first and third quartile
are obtained by further splitting each of the halves into two quarters.
Let us discuss the variability associated with an entire target population.
The file “pop1.csv” that contains the population data can be found on the inter-
net (http://pluto.huji.ac.il/~msby/StatThink/Datasets/pop1.csv). It
4.3. A POPULATION 51

is a CSV file that contains the information on sex and height of an entire adult
population of some imaginary city. (The data in “ex.1” corresponds to a sample
from this city.) Read the population data into R and examine it:
> pop.1 <- read.csv(file="pop1.csv")
> summary(pop.1)
id sex height
Min. : 1000082 FEMALE:48888 Min. :117.0
1st Qu.: 3254220 MALE :51112 1st Qu.:162.0
Median : 5502618 Median :170.0
Mean : 5502428 Mean :170.0
3rd Qu.: 7757518 3rd Qu.:178.0
Max. : 9999937 Max. :217.0

The object “pop.1” is a data frame of the same structure as the data frame
“ex.1”. It contains three variables: a unique identifier of each subject (id),
the sex of the subject (sex), and its height (height). Applying the function
“summary” to the data frame produces the summary of the variables that it
contains. In particular, for the variable “sex”, which is a factor, it produces the
frequency of its two categories – 48,888 female and 51,112 – a total of 100,000
subjects. For the variable “height”, which is a numeric variable, it produces
the extreme values, the quartiles, and the mean.
Let us concentrate on the variable “height”. A bar plot of the distribution of
the heights in the entire population is given in Figure 4.1 1 . Recall that a vertical
bar is placed above each value of height that appears in the population, with
the height of the bar representing the frequency of the value in the population.
One may read out of the graph or obtain from the numerical summaries that
the variable takes integer values in the range between 117 and 217 (heights
are rounded to the nearest centimeter). The distribution is centered at 170
centimeter, with the central 50% of the values spreading between 162 and 178
centimeters.
The mean of the height in the entire population is equal to 170 centimeter.
This mean, just like the mean for the distribution of data, is obtained by the
summation of all the heights in the population divided by the population size.
Let us denote the size of the entire population by N . In this example N =
100, 000. (The size of the sample for the data was called n and was equal to
n = 100 in the parallel example that deals with the data of a sample.) The mean
of an entire population is denoted by the Greek letter µ and is read “mew ”. (The
average for the data was denoted x̄). The formula of the population mean is:
PN
Sum of all values in the population xi
µ= = i=1 .
Number of values in the population N
Observe the similarity between the definition of the mean for the data and
the definition of the mean for the population. In both cases the arithmetic
average is computed. The only difference is that in the case of the mean of the
data the computation is with respect to the values that appear in the sample
whereas for the population all the values in the population participate in the
computation.
1 Such a bar plot can be produced with the expression “plot(table(pop.1$height))”.
52 CHAPTER 4. PROBABILITY

In actual life, we will not have all the values of a variable in the entire
population. Hence, we will not be able to compute the actual value of the
population mean. However, it is still meaningful to talk about the population
mean because this number exists, even though we do not know what its value
is. As a matter of fact, one of the issues in statistics is to try to estimate this
unknown quantity on the basis of the data we do have in the sample.
A characteristic of the distribution of an entire population is called a pa-
rameter. Hence, µ, the population average, is a parameter. Other examples
of parameters are the population median and the population quartiles. These
parameters are defined exactly like their data counterparts, but with respect
to the values of the entire population instead of the observations in the sample
alone.
Another example of a parameter is the population variance. Recall that the
sample variance was defined with the aid of the deviations xi − x̄, where xi is the
value of the measurement for the ith subject and x̄ is the mean for the data. In
order to compute the sample variance these deviations were squared to produce
the squared deviations. The squares were summed up and then divided by the
sample size minus one (n − 1). The sample variance, computed from the data,
was denoted s2 .
The population variance is defined in a similar way. First, the deviations
from the population mean xi − µ are considered for each of the members of
the population. These deviations are squared and the average of the squares is
computed. We denote this parameter by σ 2 (read “sigma square”). A minor
difference between the sample variance and the population variance is that for
the latter we should divide the sum of squared deviations by the population size
(N ) and not by the population size minus one (N − 1):

σ 2 =The average square deviation in the population

Sum of the squares of the deviations in the population
=
Number of values in the population
PN
(xi − µ)2
= i=1 .
N
The standard deviation of the population, yet another parameter, is denoted
by σ and is equal to the square root of the variance. The standard deviation
summarizes the overall variability of the measurement across the population.
Again, the typical situation is that we do not know what the actual value of the
standard deviation of the population is. Yet, we may refer to it as a quantity
and we may try to estimate its value based on the data we do have from the
sample.
For the height of the subjects in our imaginary city we get that√the variance
is equal to σ 2 = 126.1576. The standard deviation is equal to σ = 126.1576 =
11.23199. These quantities can be computed in this example from the data
frame “pop.1” with the aid of the functions “var” and “sd”, respectively2 .
2 Observe that the function “var” computes the sample variance. Consequently, the sum

of squares is divided by N − 1. We can correct that when computing the population variance
by multiplying the result by N − 1 and dividing by N . Notice that the difference between
the two quantities is negligible for a large population. Henceforth we will use the functions
“var” and “sd” to compute the variance and standard deviations of populations without the
application of the correction.
4.4. RANDOM VARIABLES 53

4.4 Random Variables

In the previous section we dealt with the variability of the population. Next we
consider the variability of a random variable. As an example, consider taking
a sample of size n = 1 from the population (a single person) and measuring
his/her height.
The object pop.1$height is a sequence with 100,000 entries. Think of it as
a population. We will apply the function “sample” to this sequence:

> sample(pop.1$height,1)
[1] 162

The first entry to the function is the given sequence of heights. When we
set the second argument to 1 then the function selects one of the entries of
the sequence at random, with each entry having the same likelihood of being
selected. Specifically, in this example an entry that contains the value 162 was
selected. Let us run the function again:

> sample(pop.1$height,1)
[1] 192

In this instance an entry with a different value was selected. Try to run the
command several times yourself and see what you get. Would you necessarily
obtain a different value in each run?
Now let us enter the same command without pressing the return key:

> sample(pop.1$height,1)

Can you tell, before pressing the key, what value will you get?
The answer to this question is of course “No”. There are 100,000 entries
with a total of 94 distinct values. In principle, any of the values may be selected
and there is no way of telling in advance which of the values will turn out as an
outcome.
A random variable is the future outcome of a measurement, before the
measurement is taken. It does not have a specific value, but rather a collection
of potential values with a distribution over these values. After the measurement
is taken and the specific value is revealed then the random variable ceases to be
a random variable! Instead, it becomes data.
Although one is not able to say what the outcome of a random variable
will turn out to be. Still, one may identify patterns in this potential outcome.
For example, knowing that the distribution of heights in the population ranges
between 117 and 217 centimeter one may say in advance that the outcome of
the measurement must also be in that interval. Moreover, since there is a total
of 3,476 subjects with height equal to 168 centimeter and since the likelihood
of each subject to be selected is equal then the likelihood of selecting a subject
of this height is 3,476/100,000 = 0.03476. In the context of random variables
we call this likelihood probability. In the same vain, the frequency of subjects
with hight 192 centimeter is 488, and therefore the probability of measuring
such a height is 0.00488. The frequency of subjects with height 200 centimeter
or above is 393, hence the probability of obtaining a measurement in the range
between 200 and 217 centimeter is 0.00393.
54 CHAPTER 4. PROBABILITY

4.4.1 Sample Space and Distribution

Let us turn to the formal definition of a random variable: A random variable re-
fer to numerical values, typically the outcome of an observation, a measurement,
or a function thereof.
A random variable is characterized via the collection of potential values it
may obtain, known as the sample space and the likelihood of obtaining each of
the values in the sample space (namely, the probability of the value). In the
given example, the sample space contains the 94 integer values that are marked
in Figure 4.1. The probability of each value is the height of the bar above the
value, divided by the total frequency of 100,000 (namely, the relative frequency
in the population).
We will denote random variables with capital Latin letters such as X, Y ,
and Z. Values they may obtain will be marked by small Latin letters such as
x, y, z. For the probability of values we will use the letter “P”. Hence, if we
denote by X the measurement of height of a random individual that is sampled
from the given population then:

P(X = 168) = 0.03476

and
P(X ≥ 200) = 0.00393 .
Consider, as yet another example, the probability that the height of a random
person sampled from the population differs from 170 centimeter by no more
than 10 centimeters. (In other words, that the height is between 160 and 180
centimeters.) Denote by X the height of that random person. We are interested
in the probability P(|X − 170| ≤ 10).3
The random person can be any of the subjects of the population with equal
probability. Thus, the sequence of the heights of the 100,000 subjects represents
the distribution of the random variable X:

> pop.1 <- read.csv(file="pop1.csv")

> X <- pop.1$height

Notice that the object “X” is a sequence of lenght 100,000 that stores all the
heights of the population. The probability we seek is the relative frequency in
this sequency of values between 160 and 180. First we compute the probability
and then explain the method of computation:

> mean(abs(X-170) <= 10)

[1] 0.64541

We get that the height of a person randomly sampled from the population is
between 160 and 180 centimeters with probability 0.64541.
Let us produce a small example that will help us explain the computation
of the probability. We start by forming a sequence with 10 numbers:

> Y <- c(6.3, 6.9, 6.6, 3.4, 5.5, 4.3, 6.5, 4.7, 6.1, 5.3)
3 The expression {|X − 170| ≤ 10} reads as “the absolute value of the difference between X

and 170 is no more that 10”. In other words, {−10 ≤ X − 170 ≤ 10}, which is equivalent to
the statement that {160 ≤ X ≤ 180}. It follows that P(|X − 170| ≤ 10) = P(160 ≤ X ≤ 180).
4.4. RANDOM VARIABLES 55

The goal is to compute the proportion of numbers that are in the range [4, 6]
(or, equivalently, {|Y − 5| ≤ 1}).
The function “abs” computes the absolute number of its input argument.
When the function is applied to the sequence “Y-5” it produces a sequence of
the same length with the distances between the components of “Y” and the
number 5:
> abs(Y-5)
[1] 1.3 1.9 1.6 1.6 0.5 0.7 1.5 0.3 1.1 0.3

Compare the resulting output to the original sequence. The first value in the
input sequence is 6.3. Its distance from 5 is indeed 1.3. The fourth value in
the input sequence is 3.4. The difference 3.4 - 5 is equal to -1.6, and when the
absolute value is taken we get a distance of 1.6.
The function “<=” expects an argument to the right and an argument to the
left. It compares each component to the left with the parallel component to the
right and returns a logical value, “TRUE” or “FALSE”, depending on whether the
relation that is tested holds or not:
> abs(Y - 5) <= 1
[1] FALSE FALSE FALSE FALSE TRUE TRUE FALSE TRUE FALSE TRUE

Observe the in this example the function “<=” produced 10 logical values, one
for each of the elements of the sequence to the left of it. The first input in the
sequence “Y” is 6.3, which is more than one unit away from 5. Hence, the first
output of the logical expression is “FALSE”. On the other hand, the last input
in the sequence “Y” is 5.3, which is within the range. Therefore, the last output
of the logical expression is “TRUE”.
Next, we compute the proportion of “TRUE” values in the sequence:
> mean(abs(Y - 5) <= 1)
[1] 0.4
When a sequence with logical values is entered into the function “mean” then the
function replaces the TRUE’s by 1 and the FALSE’s by 0. The average produces
then the relative frequency of TRUE’s in the sequence as required. Specifically,
in this example there are 4 TRUE’s and 6 FALSE’s. Consequently, the output of
the final expression is 4/10 = 0.4.
The computation of the probability that the sampled height falls within 10
centimeter of 170 is based on the same code. The only differences are that the
input sequence “Y” is replaced by the sequence of population heights “X” as
input. the number “5” is replaced by the number “170” and the number “1”
is replaced by the number “10”. In both cases the result of the computation is
the relative proportion of the times that the values of the input sequence fall
within a given range of the indicated number.
The probability function of a random variable is defined for any value that
the random variable may obtain and produces the distribution of the random
variable. The probability function may emerge as a relative frequency as in
the given example or it may be a result of theoretical modeling. Examples of
theoretical random variables are presented mainly in the next two chapters.
Consider an example of a random variable. The sample space and the prob-
ability function specify the distribution of the random variable. For example,
56 CHAPTER 4. PROBABILITY

assume it is known that a random variable X may obtain the values 0, 1, 2, or

3. Moreover, imagine that it is known that P(X = 1) = 0.25, P(X = 2) = 0.15,
and P(X = 3) = 0.10. What is P(X = 0), the probability that X is equal to 0?
The sample space, the collection of possible values that the random variable
may obtain is the collection {0, 1, 2, 3}. Observe that the sum over the positive
values is:

P(X > 0) = P(X = 1) + P(X = 2) + P(X = 3) = 0.25 + 0.15 + 0.10 = 0.50 .

It follows, since the sum of probabilities over the entire sample space is equal
to 1, that P(X = 0) = 1 − 0.5 = 0.5.

Value Probability Cum. Prob.

0 0.50 0.50
1 0.25 0.75
2 0.15 0.90
3 0.10 1.00

Table 4.1: The Distribution of X

Table 4.1 summarizes the distribution of the random variable X. Observe

the similarity between the probability function and the notion of relative fre-
quency that was discussed in Chapter 2. Both quantities describe distribution.
Both are non-negative and sum to 1. Likewise, notice that one may define the
cumulative probability the same way cumulative relative frequency is defined:
Ordering the values of the random variable from smallest to largest, the cumu-
lative probability at a given value is the sum of probabilities for values less or
equal to the given value.
Knowledge of the probabilities of a random variable (or the cumulative prob-
abilities) enables the computation of other probabilities that are associated with
the random variable. For example, considering the random variable X of Ta-
ble 4.1, we may calculate the probability of X falling in the interval [0.5, 2.3].
Observe that the given range contains two values from the sample space, 1 and
2, therefore:

P(0.5 ≤ X ≤ 2.3) = P(X = 1) + P(X = 2) = 0.25 + 0.15 = 0.40 .

Likewise, we may produce the probability of X obtaining an odd value:

P(X = odd) = P(X = 1) + P(X = 3) = 0.25 + 0.10 = 0.35 .

Observe that both {0.5 ≤ X ≤ 2.3} and {X = odd} refer to subsets of values
of the sample space. Such subsets are denoted events. In both examples the
probability of the event was computed by the summation of the probabilities
associated with values that belong to the event.

4.4.2 Expectation and Standard Deviation

We may characterize the center of the distribution of a random variable and the
spread of the distribution in ways similar to those used for the characterization
of the distribution of data and the distribution of a population.
4.4. RANDOM VARIABLES 57

The expectation marks the center of the distribution of a random variable. It

is equivalent to the data average x̄ and the population average µ, which was used
in order to mark the location of the distribution of the data and the population,
respectively.
Recall from Chapter 3 that the average of the data can be computed as the
weighted average of the values that are present in the data, with weights given
by the relative frequency. Specifically, we saw for the data

1, 1, 1, 2, 2, 3, 4, 4, 4, 4, 4

that
1+1+1+2+2+3+4+4+4+4+4 3 2 1 5
= 1× +2× +3× +4× ,
11 11 11 11 11
producing the value of x̄ = 2.727 in both representations. Using a formula, the
equality between the two ways of computing the mean is given in terms of the
equation: Pn
xi X
x̄ = i=1


= x × (fx /n) .
n x

In the first representation of the arithmetic mean, the average is computed by

the summation of all data points and dividing the sum by the sample size. In the
second representation, that uses a weighted sum, the sum extends over all the
unique values that appear in the data. For each unique value the value is mul-
tiplied by the relative frequency of the value in the data. These multiplications
are summed up to produce the mean.
The expectation of a random variable is computed in the spirit of the second
formulation. The expectation of a random variable is marked with the letter
“E” and is define via the equation:
X

E(X) = x × P(x) .
x

In this definition all the unique values of the sample space are considered. For
each value a product of the value and the probability of the value is taken. The
expectation is obtained by the summation of all these products. In this defi-
nition the probability P(x) replaces the relative frequency fx /n but otherwise,
the definition of the expectation and the second formulation of the mean are
identical to each other.
Consider the random variable X with distribution that is described in Ta-
ble 4.1. In order to obtain its expectation we multiply each value in the sample
space by the probability of the value. Summation of the products produces the
expectation (see Table 4.2):

E(X) = 0 × 0.5 + 1 × 0.25 + 2 × 0.15 + 3 × 0.10 = 0.85 .

In the example of height we get that the expectation is equal to 170.035 cen-
timeter. Notice that this expectation is equal to µ, the mean of the population4 .
This is no accident. The expectation of a potential measurement of a randomly
selected subject from a population is equal to the average of the measurement
across all subjects.
4 The mean of the population can be computed with the expression “mean(pop.1$height)”
58 CHAPTER 4. PROBABILITY

Value Probability x × P(X = x)

0 0.50 0.00
1 0.25 0.25
2 0.15 0.30
3 0.10 0.30
E(X) = 0.85

Table 4.2: The Expectation of X

The sample variance (s2 ) is obtained as the sum of the squared deviations
from the average, divided by the sample size (n) minus 1:
Pn
2 (xi − x̄)2
s = i=1 .
n−1
A second formulation for the computation of the same quantity is via the use
of relative frequencies. The formula for the sample variance takes the form
n X
s2 = (x − x̄)2 × (fx /n) .


n−1 x

In this formulation one considers each of the unique value that are present in
the data. For each value the deviation between the value and the average is
computed. These deviations are then squared and multiplied by the relative
frequency. The products are summed up. Finally, the sum is multiplied by the
ratio between the sample size n and n − 1 in order to correct for the fact that
in the sample variance the sum of squared deviations is divided by the sample
size minus 1 and not by the sample size.
In a similar way, the variance of a random variable may be defined via the
probability of the values that make the sample space. For each such value one
computes the deviation from the expectation. This deviation is then squared
and multiplied by the probability of the value. The multiplications are summed
up in order to produce the variance:
X
(x − E(X))2 × P(x) .


Var(X) =
x

Notice that the formula for the computation of the variance of a random variable
is very similar to the second formulation for the computation of the sample
variance. Essentially, the mean of the data is replaced by the expectation of
the random variable and the relative frequency of a value is replaced by the
probability of the value. Another difference is that the correction factor is not
used for the variance of a random variable.
As an example consider the variance of the random variable X. The com-
putation of the variance of this random variable is carried out in Table 4.3).
The sample space, the values that the random variable may obtain, are given
in the first column and the probabilities of the values are given in the second
column. In the third column the deviation of the value from the expectation
E(X) = 0.85 is computed for each value. The 4th column contains the square of
these deviations and the 5th and last column involves the product of the square
deviations and the probabilities. The variance is obtained by summing up the
4.5. PROBABILITY AND STATISTICS 59

Value Prob. x − E(X) (x − E(X))2 (x − E(X))2 × P(X = x)

0 0.50 -0.85 0.7225 0.361250
1 0.25 0.15 0.0225 0.005625
2 0.15 1.15 1.3225 0.198375
3 0.10 2.15 4.6225 0.462250
Var(X) = 1.027500

Table 4.3: The Variance of X

products in the last column. In the given example:

Var(X) =(0 − 0.85)2 × 0.5 + (1 − 0.85)2 × 0.25

+ (2 − 0.85)2 × 0.15 + (3 − 0.85)2 × 0.10 = 1.0275 .

The standard deviation of a random p variable is √

the square root of the vari-
ance. The standard deviation of X is Var(X) = 1.0275 = 1.013657.
In the example that involves the height of a subject selected from the popula-
tion at random we obtain that the variance is 126.1576, equal to the population
variance, and the standard deviation is 11.23199, the square root of the variance.
Other characterization of the distribution that were computed for data, such
as the median, the quartiles, etc., may also be defined for random variables.

4.5 Probability and Statistics

Modern science may be characterized by a systematic collection of empirical
measurements and the attempt to model laws of nature using mathematical
language. The drive to deliver better measurements led to the development
of more accurate and more sensitive measurement tools. Nonetheless, at some
point it became apparent that measurements may not be perfectly reproducible
and any repeated measurement of presumably the exact same phenomena will
typically produce variability in the outcomes. On the other hand, scientists also
found that there are general laws that govern this variability in repetitions. For
example, it was discovered that the average of several independent repeats of
the measurement is less variable and more reproducible than each of the single
measurements themselves.
Probability was first introduced as a branch of mathematics in the investiga-
tion of uncertainty associated with gambling and games of chance. During the
early 19th century probability began to be used in order to model variability in
measurements. This application of probability turned out to be very successful.
Indeed, one of the major achievements of probability was the development of the
mathematical theory that explains the phenomena of reduced variability that is
observed when averages are used instead of single measurements. In Chapter ??
we discuss the conclusions of this theory.
Statistics study method for inference based on data. Probability serves as
the mathematical foundation for the development of statistical theory. In this
chapter we introduced the probabilistic concept of a random variable. This
concept is key for understanding statistics. In the rest of Part I of this book we
discuss the probability theory that is used for statistical inference. Statistical
inference itself is discussed in Part II of the book.
60 CHAPTER 4. PROBABILITY

Value Probability
0 p
1 2p
2 3p
3 4p
4 5p
5 6p

Table 4.4: The Distribution of Y

4.6 Solved Exercises

Question 4.1. Table 4.6 presents the probabilities of the random variable Y .
These probabilities are a function of the number p, the probability of the value
“0”. Answer the following questions:

1. What is the value of p?

2. P(Y < 3) = ?

3. P(Y = odd) = ?

4. P(1 ≤ Y < 4) = ?

5. P(|Y − 3| < 1.5) = ?

6. E(Y ) = ?

7. Var(Y ) = ?

8. What is the standard deviation of Y .

Solution (to Question 4.1.1): Consult Table 4.6. The probabilities of the
different values of Y are {p, 2p, . . . , 6p}. These probabilities sum to 1, conse-
quently

p + 2p + 3p + 4p + 5p + 6p = (1 + 2 + 3 + 4 + 5 + 6)p = 21p = 1 =⇒ p = 1/21 .

Solution (to Question 4.1.2): The event {Y < 3} contains the values 0, 1
and 2. Therefore,
1 2 3 6
P(Y < 3) = P(Y = 0) + P(Y = 1) + P(Y = 2) = + + = = 0.2857 .
21 21 21 21

Solution (to Question 4.1.3): The event {Y = odd} contains the values 1,
3 and 5. Therefore,
2 4 6 12
P(Y = odd) = P(Y = 1)+P(Y = 3)+P(Y = 5) = + + = = 0.5714 .
21 21 21 21
4.6. SOLVED EXERCISES 61

Solution (to Question 4.1.4): The event {1 ≤ Y < 4} contains the values 1,
2 and 3. Therefore,

2 3 4 9
P(1 ≤ Y < 4) = P(Y = 1)+P(Y = 2)+P(Y = 3) = + + = = 0.4286 .
21 21 21 21

Solution (to Question 4.1.5): The event {|Y − 3| < 1.5} contains the values
2, 3 and 4. Therefore,

3 4 5 12
P(|Y −3| < 1.5) = P(Y = 2)+P(Y = 3)+P(Y = 4) = + + = = 0.5714 .
21 21 21 21

Solution (to Question 4.1.6): The values that the random variable Y ob-
tains are the numbers 0, 1, 2, . . . , 5, with probabilities {1/21, 2/21, . . . , 6/21},
respectively. The expectation is obtained by the multiplication of the values by
their respective probabilities and the summation of the products. Let us carry
out the computation in R:

> Y.val <- c(0,1,2,3,4,5)

> P.val <- c(1,2,3,4,5,6)/21
> E <- sum(Y.val*P.val)
> E
[1] 3.333333

We obtain an expectation E(Y ) = 3.3333.

Solution (to Question 4.1.7): The values that the random variable Y ob-
tains are the numbers 0, 1, 2, . . . , 5, with probabilities {1/21, 2/21, . . . , 6/21},
respectively. The expectation is equal to E(Y ) = 3.333333. The variance is
obtained by the multiplication of the squared deviation from the expectation of
the values by their respective probabilities and the summation of the products.
Let us carry out the computation in R:

> Var <- sum((Y.val-E)^2*P.val)

> Var
[1] 2.222222

We obtain a variance Var(Y ) = 2.2222.

Solution (top Question 4.1.8):

√ The standard deviation is the square root of
the variance: Var(Y ) = 2.2222 = 1.4907.

Question 4.2. One invests $2 to participate in a game of chance. In this game

a coin is tossed three times. If all tosses end up “Head” then the player wins
$10. Otherwise, the player losses the investment.

1. What is the probability of winning the game?

2. What is the probability of loosing the game?

62 CHAPTER 4. PROBABILITY

3. What is the expected gain for the player that plays this game? (Notice
that the expectation can obtain a negative value.)
Solution (to Question 4.2.1): An outcome of the game of chance may be
represented by a sequence of length three composed of the letters “H” and “T”.
For example, the sequence “THH” corresponds to the case where the first toss
produced a ”Tail”, the second a “Head” and the third a “Head”.
With this notation we obtain that the possible outcomes of the game are
{HHH, THH, HTH, TTH, HHT, THT, HTT, TTT}. All outcomes are equally
likely. There are 8 possible outcomes and only one of which corresponds to
winning. Consequently, the probability of winning is 1/8.

Solution (to Question 4.2.2): Consider the previous solution. One looses if
any other of the outcomes occurs. Hence, the probability of loosing is 7/8.

Solution (to Question 4.2.3): Denote the gain of the player by X. The
random variable X may obtain two values: 10-2 = 8 if the player wins and -2 if
the player looses. The probabilities of these values are {1/8, 7/8}, respectively.
Therefore, the expected gain, the expectation of X is:
1 7
E(X) = 8 × + (−2) × = −0.75 .
8 8

4.7 Summary
Glossary
Random Variable: The probabilistic model for the value of a measurement,
before the measurement is taken.
Sample Space: The set of all values a random variable may obtain.
Probability: A number between 0 and 1 which is assigned to a subset of the
sample space. This number indicates the likelihood of the random variable
obtaining a value in that subset.
Expectation: The central value for a random variable. The expectation of the
random variable X is marked by E(X).
Variance: The (squared) spread of a random variable. The variance of the
random variable X is marked by Var(X). The standard deviation is the
square root of the variance.

Discussion in the Forum

Random variables are used to model situations in which the outcome, before
the fact, is uncertain. One component in the model is the sample space. The
sample space is the list of all possible outcomes. It includes the outcome that
took place, but also all other outcomes that could have taken place but never
did materialize. The rationale behind the consideration of the sample space is
4.7. SUMMARY 63

the intention to put the outcome that took place in context. What do you think
of this rationale?
When forming your answer to this question you may give an example of a
situation from you own field of interest for which a random variable can serve
as a model. Identify the sample space for that random variable and discuss the
importance (or lack thereof) of the correct identification of the sample space.
For example, consider a factory that produces car parts that are sold to car
makers. The role of the QA personnel in the factory is to validate the quality
of each batch of parts before the shipment to the client.
To achieve that, a sample of parts may be subject to a battery of quality test.
Say that 20 parts are selected to the sample. The number of those among them
that will not pass the quality testing may be modeled as a random variable.
The sample space for this random variable may be any of the numbers 0, 1, 2,
. . . , 20.
The number 0 corresponds to the situation where all parts in the sample
passed the quality testing. The number 1 corresponds to the case where 1 part
did not pass and the other 19 did. The number 2 describes the case where 2 of
the 20 did not pass and 18 did pass, etc.

Summary of Formulas
Population Size: N = the number of people, things, etc. in the population.
PN
Population Average: µ = (1/N ) i=1 xi
P

Expectation of a Random Variable: E(X) = x x × P(x)
PN
Population Variance: σ 2 = (1/N ) i=1 (xi − µ)2


Variance of a Random Variable: Var(X) = x (x − E(X))2 × P(x)
P
64 CHAPTER 4. PROBABILITY
Chapter 5

Random Variables

5.1 Student Learning Objective

This section introduces some important examples of random variables. The
distributions of these random variables emerge as mathematical models of real-
life settings. In two of the examples the sample space is composed of integers.
In the other two examples the sample space is made of continuum of values. For
random variables of the latter type one may use the density, which is a type of
a histogram, in order to describe the distribution.
By the end of the chapter the student should:
• Identify the Binomial, Poisson, Uniform, and Exponential random vari-
ables, relate them to real life situations, and memorize their expectations
and variances.
• Relate the plot of the density/probability function and the cumulative
probability function to the distribution of a random variable.
• Become familiar with the R functions that produce the density/probability
of these random variables and their cumulative probabilities.
• Plot the density and the cumulative probability function of a random
variable and compute probabilities associated with random variables.

5.2 Discrete Random Variables

In the previous chapter we introduced the notion of a random variable. A
random variable corresponds to the outcome of an observation or a measurement
prior to the actual making of the measurement. In this context one can talk
of all the values that the measurement may potentially obtain. This collection
of values is called the sample space. To each value in the sample space one
may associate the probability of obtaining this particular value. Probabilities
are like relative frequencies. All probabilities are positive and the sum of the
probabilities that are associated with all the values in the sample space is equal
to one.
A random variable is defined by the identification of its sample space and
the probabilities that are associated with the values in the sample space. For

65
66 CHAPTER 5. RANDOM VARIABLES

each type of random variable we will identify first the sample space — the values
it may obtain — and then describe the probabilities of the values. Examples
of situations in which each type of random variable may serve as a model of a
measurement will be provided. The R system provides functions for the com-
putation of probabilities associated with specific types of random variables. We
will use these functions in this and in proceeding chapters in order to carry out
computations associated with the random variables and in order to plot their
distributions.
The distribution of a random variable, just like the distribution of data, can
be characterized using numerical summaries. For the latter we used summaries
such as the mean and the sample variance and standard deviation. The mean
is used to describe the central location of the distribution and the variance and
standard deviation are used to characterize the total spread. Parallel summaries
are used for random variable. In the case of a random variable the name ex-
pectation is used for the central location of the distribution and the variance
and the standard deviation (the square root of the variation) are used to sum-
marize the spread. In all the examples of random variables we will identify the
expectation and the variance (and, thereby, also the standard deviation).
Random variables are used as probabilistic models of measurements. Theo-
retical considerations are used in many cases in order to define random variables
and their distribution. A random variable for which the values in the sample
space are separated from each other, say the values are integers, is called a
discrete random variable. In this section we introduce two important integer-
valued random variables: The Binomial and the Poisson random variables.
These random variables may emerge as models in contexts where the measure-
ment involves counting the number of occurrences of some phenomena.
Many other models, apart from the Binomial and Poisson, exist for discrete
random variables. An example of such model, the Negative-Binomial model, will
be considered in Section 5.4. Depending on the specific context that involves
measurements with discrete values, one may select the Binomial, the Poisson,
or one of these other models to serve as a theoretical approximation of the
distribution of the measurement.

5.2.1 The Binomial Random Variable

The Binomial random variable is used in settings in which a trial that has
two possible outcomes is repeated several times. Let us designate one of the
outcomes as “Success” and the other as “Failure”. Assume that the probability
of success in each trial is given by some number p that is larger than 0 and smaller
than 1. Given a number n of repeats of the trial and given the probability
of success, the actual number of trials that will produce “Success” as their
outcome is a random variable. We call such random variable Binomial. The fact
that a random variable X has such a distribution is marked by the expression:
“X ∼ Binomial(n, p)”.
As an example consider tossing 10 coins. Designate “Head” as success and
“Tail” as failure. For fair coins the probability of “Head” is 1/2. Consequently,
if X is the total number of “Heads” then X ∼ Binomial(10, 0.5), where n = 10
is the number of trials and p = 0.5 is the probability of success in each trial.
It may happen that all 10 coins turn up “Tail”. In this case X is equal to 0.
It may also be the case that one of the coins turns up “Head” and the others
5.2. DISCRETE RANDOM VARIABLES 67

0.25
0.20
0.15
Probability

0.10
0.05
0.00

0 2 4 6 8 10

Figure 5.1: The Binomial(10,0.5) Distribution

turn up “Tail”. The random variable X will obtain the value 1 in such a case.
Likewise, for any integer between 0 and 10 it may be the case that the number
of “Heads” that turn up is equal to that integer with the other coins turning up
“Tail”. Hence, the sample space of X is the set of integers {0, 1, 2, . . . , 10}. The
probability of each outcome may be computed by an appropriate mathematical
formula that will not be discussed here1 .
The probabilities of the various possible values of a Binomial random vari-
able may be computed with the aid of the R function “dbinom” (that uses the
mathematical formula for the computation). The input to this function is a se-
quence of values, the value of n, and the value of p. The output is the sequence
of probabilities associated with each of the values in the first input.
For example, let us use the function in order to compute the probability
that the given Binomial obtains an odd value. A sequence that contains the
odd values in the Binomial sample space can be created with the expression
“c(1,3,5,7,9)”. This sequence can serve as the input in the first argument of
the function “dbinom”. The other arguments are “10” and “0.5”, respectively:
1 If n
x
X ∼ Binomial(n, p) then P(X = x) = x
p (1 − p)n−x , for x = 0, 1, . . . , n.
68 CHAPTER 5. RANDOM VARIABLES

> dbinom(c(1,3,5,7,9),10,0.5)
[1] 0.009765625 0.117187500 0.246093750 0.117187500 0.009765625

Observe that the output of the function is a sequence of the same length as the
first argument. This output contains the Binomial probabilities of the values in
the first argument. In order to obtain the probability of the event {X is odd}
we should sum up these probabilities, which we can do by applying the function
“sum” to the output of the function that computes the Binomial probabilities:

> sum(dbinom(c(1,3,5,7,9),10,0.5))
[1] 0.5

Observe that the probability of obtaining an odd value in this specific case is
equal to one half.
Another example is to compute all the probabilities of all the potential values
of a Binomial(10, 0.5) random variable:

> x <- 0:10

> dbinom(x,10,0.5)
[1] 0.0009765625 0.0097656250 0.0439453125 0.1171875000
[5] 0.2050781250 0.2460937500 0.2050781250 0.1171875000
[9] 0.0439453125 0.0097656250 0.0009765625

The expression “start.value:end.value” produces a sequence of numbers

that initiate with the number “start.value” and proceeds in jumps of size
one until reaching the number “end.value”. In this example, “0:10” produces
the sequence of integers between 0 and 10, which is the sample space of the
current Binomial example. Entering this sequence as the first argument to the
function “dbinom” produces the probabilities of all the values in the sample
space.
One may display the distribution of a discrete random variable with a bar
plot similar to the one used to describe the distribution of data. In this plot
a vertical bar representing the probability is placed above each value of the
sample space. The hight of the bar is equal to the probability. A bar plot of
the Binomial(10, 0.5) distribution is provided in Figure 5.1.
Another useful function is “pbinom”, which produces the cumulative proba-
bility of the Binomial:

> pbinom(x,10,0.5)
[1] 0.0009765625 0.0107421875 0.0546875000 0.1718750000
[5] 0.3769531250 0.6230468750 0.8281250000 0.9453125000
[9] 0.9892578125 0.9990234375 1.0000000000
> cumsum(dbinom(x,10,0.5))
[1] 0.0009765625 0.0107421875 0.0546875000 0.1718750000
[5] 0.3769531250 0.6230468750 0.8281250000 0.9453125000
[9] 0.9892578125 0.9990234375 1.0000000000

The output of the function “pbinom” is the cumulative probability P(X ≤ x)

that the random variable is less than or equal to the input value. Observe
that this cumulative probability is obtained by summing all the probabilities
associated with values that are less than or equal to the input value. Specifi-
cally, the cumulative probability at x = 3 is obtained by the summation of the
5.2. DISCRETE RANDOM VARIABLES 69

probabilities at x = 0, x = 1, x = 2, and x = 3:

P(X ≤ 3) = 0.0009765625+0.009765625+0.0439453125+0.1171875 = 0.171875

The numbers in the sum are the first 4 values from the output of the function
“dbinom(x,10,0.5)”, which computes the probabilities of the values of the
sample space.
In principle, the expectation of the Binomial random variable, like the expec-
tation of any other (discrete) random variable is obtained from the application
of the general formulae:
X X
(x − E(X))2 × P(x) .



E(X) = x × P(X = x) , Var(X) =
x x

However, in the specific case of the Binomial random variable, in which the
probability P(X = x) obeys the specific mathematical formula of the Binomial
distribution, the expectation and the variance reduce to the specific formulae:

E(X) = np , Var(X) = np(1 − p) .

Hence, the expectation is the product of the number of trials n with the proba-
bility of success in each trial p. In the variance the number of trials is multiplied
by the product of a probability of success (p) with the probability of a failure
(1 − p).
As illustration, let us compute for the given example the expectation and
the variance according to the general formulae for the computation of the ex-
pectation and variance in random variables and compare the outcome to the
specific formulae for the expectation and variance in the Binomial distribution:
> X.val <- 0:10
> P.val <- dbinom(X.val,10,0.5)
> EX <- sum(X.val*P.val)
> EX
[1] 5
> sum((X.val-EX)^2*P.val)
[1] 2.5
This agrees with the specific formulae for Binomial variables, since 10 × 0.5 = 5
and 10 × 0.5 × (1 − 0.5) = 2.5.
Recall that the general formula for the computation of the expectation calls
for the multiplication of each value in the sample space with the probability of
that value, followed by the summation of all the products. The object “X.val”
contains all the values of the random variable and the object “P.val” contains
the probabilities of these values. Hence, the expression “X.val*P.val” pro-
duces the product of each value of the random variable times the probability
of that value. Summation of these products with the function “sum” gives the
expectation, which is saved in an object that is called “EX”.
The general formula for the computation of the variance of a random variable
involves the product of the squared deviation associated with each value with
the probability of that value, followed by the summation of all products. The
expression “(X.val-EX)^2” produces the sequence of squared deviations from
the expectation for all the values of the random variable. Summation of the
70 CHAPTER 5. RANDOM VARIABLES

product of these squared deviations with the probabilities of the values (the
outcome of “(X.val-EX)^2*P.val”) gives the variance.
When the value of p changes (without changing the number of trials n) then
the probabilities that are assigned to each of the values of the sample space
of the Binomial random variable change, but the sample space itself does not.
For example, consider rolling a die 10 times and counting the number of times
that the face 3 was obtained. Having the face 3 turning up is a “Success”. The
probability p of a success in this example is 1/6, since the given face is one
out of 6 equally likely faces. The resulting random variable that counts the
total number of success in 10 trials has a Binomial(10, 1/6) distribution. The
sample space is yet again equal to the set of integers {0, 1, . . . , 10}. However, the
probabilities of values are different. These probabilities can again be computes
with the aid of the function “dbinom”:

> dbinom(x,10,1/6)
[1] 1.615056e-01 3.230112e-01 2.907100e-01 1.550454e-01
[5] 5.426588e-02 1.302381e-02 2.170635e-03 2.480726e-04
[9] 1.860544e-05 8.269086e-07 1.653817e-08

In this case smaller values of the random variable are assigned higher probabil-
ities and larger values are assigned lower probabilities..
In Figure 5.2 the probabilities for Binomial(10, 1/6), the Binomial(10, 1/2),
and the Binomial(10, 0.6) distributions are plotted side by side. In all these
3 distributions the sample space is the same, the integers between 0 and 10.
However, the probabilities of the different values differ. (Note that all bars
should be placed on top of the integers. For clarity of the presentation, the
bars associated with the Binomial(10, 1/6) are shifted slightly to the left and
the bars associated with the Binomial(10, 0.6) are shifted slightly to the right.)
The expectation of the Binomial(10, 0.5) distribution is equal to 10×0.5 = 5.
Compare this to the expectation of the Binomial(10, 1/6) distribution, which is
10 × (1/6) = 1.666667 and to the expectation of the Binomial(10, 0.6) distribu-
tion which equals 10 × 0.6 = 6.
The variance of the Binomial(10, 0.5) distribution is 10 × 0.5 × 0.5 = 2.5.
The variance when p = 1/6 is 10 × (1/6) × (5/6) = 1.388889 and the variance
when p = 0.6 is 10 × 0.6 × 0.4 = 2.4.

Example 5.1. As an application of the Binomial distribution consider a pre-

election poll. A candidate is running for office and is interested in knowing
the percentage of support in the general population in its candidacy. Denote
the probability of support by p. In order to estimate the percentage a sample of
size 300 is selected from the population. Let X be the count of supporters in
the sample. A natural model for the distribution of X is the Binomial(300, p)
distribution, since each subject in the sample may be a supporter (“Success”) or
may not be a supporter (“Failure”). The probability that a subject supports the
candidate is p and there are n = 300 subjects in the sample.

Example 5.2. As another example consider the procedure for quality control
that is described in Discussion Forum of Chapter 4. According to the procedure
20 items are tested and the number of faulty items is recorded. If p is the
probability that an item is identified as faulty then the distribution of the total
number of faulty items may be modeled by the Binomial(20, p) distribution.
5.2. DISCRETE RANDOM VARIABLES 71

0.30

p = 0.5
p = 1/6
p = 0.6
0.25
0.20
Probability

0.15
0.10
0.05
0.00

0 2 4 6 8 10

Figure 5.2: The Binomial Distribution for Various Probability of “Success” p

In both examples one may be interested in making statements on the prob-

ability p based on the sample. Statistical inference relates the actual count ob-
tained in the sample to the theoretical Binomial distribution in order to make
such statements.

5.2.2 The Poisson Random Variable

The Poisson distribution is used as an approximation of the total number of
occurrences of rare events. Consider, for example, the Binomial setting that
involves n trials with p as the probability of success of each trial. Then, if p
is small but n is large then the number of successes X has, approximately, the
Poisson distribution.
The sample space of the Poisson random variable is the unbounded collection
of integers: {0, 1, 2, . . .}. Any integer value is assigned a positive probability.
Hence, the Poisson random variable is a convenient model when the maximal
number of occurrences of the events in a-priori unknown or is very large. For
example, one may use the Poisson distribution to model the number of phone
calls that enter a switchboard in a given interval of time or the number of
72 CHAPTER 5. RANDOM VARIABLES

0.25
0.20
Probability

0.15
0.10
0.05
0.00

0 2 4 6 8 10

Figure 5.3: The Poisson(2) Distribution

malfunctioning components in a shipment of some product.

The Binomial distribution was specified by the number of trials n and prob-
ability of success in each trial p. The Poisson distribution is specified by its ex-
pectation, which we denote by λ. The expression “X ∼ Poisson(λ)” states that
the random variable X has a Poisson distribution2 with expectation E(X) = λ.
The function “dpois” computes the probability, according to the Poisson dis-
tribution, of values that are entered as the first argument to the function. The
expectation of the distribution is entered in the second argument. The function
“ppois” computes the cumulative probability. Consequently, we can compute
the probabilities and the cumulative probabilities of the values between 0 and
10 for the Poisson(2) distribution via:

> x <- 0:10

> dpois(x,2)
[1] 1.353353e-01 2.706706e-01 2.706706e-01 1.804470e-01
[5] 9.022352e-02 3.608941e-02 1.202980e-02 3.437087e-03
[9] 8.592716e-04 1.909493e-04 3.818985e-05
2 If X ∼ Poisson(λ) then P(X = x) = e−λ λx /x!, for x = 0, 1, 2, . . ..
5.2. DISCRETE RANDOM VARIABLES 73

0.6

lambda = 0.5
lambda = 1
0.5

lambda = 2
0.4
Probability

0.3
0.2
0.1
0.0

0 2 4 6 8 10

Figure 5.4: The Poisson Distribution for Various Values of λ

> ppois(x,2)
[1] 0.1353353 0.4060058 0.6766764 0.8571235 0.9473470 0.9834364
[7] 0.9954662 0.9989033 0.9997626 0.9999535 0.9999917
The probability function of the Poisson distribution with λ = 2, in the range
between 0 and 10, is plotted in Figure 5.3. Observe that in this example prob-
abilities of the values 8 and beyond are very small. As a matter of fact, the
cumulative probability at x = 7 (the 8th value in the output of “ppois(x,2)”)
is approximately 0.999, out of the total cumulative probability of 1.000, leaving
a total probability of about 0.001 to be distributed among all the values larger
than 7.
Let us compute the expectation of the given Poisson distribution:
> X.val <- 0:10
> P.val <- dpois(X.val,2)
> sum(X.val*P.val)
[1] 1.999907
Observe that the outcome is almost, but not quite, equal to 2.00, which is the
actual value of the expectation. The reason for the inaccuracy is the fact that
74 CHAPTER 5. RANDOM VARIABLES

we have based the computation in R on the first 11 values of the distribution

only, instead of the infinite sequence of values. A more accurate result may be
obtained by the consideration of the first 101 values:

> X.val <- 0:100

> P.val <- dpois(X.val,2)
> EX <- sum(X.val*P.val)
> EX
[1] 2
> sum((X.val-EX)^2*P.val)
[1] 2

In the last expression we have computed the variance of the Poisson distribution
and obtained that it is equal to the expectation. This results can be validated
mathematically. For the Poisson distribution it is always the case that the
variance is equal to the expectation, namely to λ:

E(X) = Var(X) = λ .

In Figure 5.4 you may find the probabilities of the Poisson distribution for
λ = 0.5, λ = 1 and λ = 2. Notice once more that the sample space is the same
for all the Poisson distributions. What varies when we change the value of λ are
the probabilities. Observe that as λ increases then probability of larger values
increases as well.

Example 5.3. A radio active element decays by the release of subatomic parti-
cles and energy. The decay activity is measured in terms of the number of decays
per second in a unit mass. A typical model for the distribution of the number
of decays is the Poisson distribution. Observe that the number of decays in a
second is a integer and, in principle, it may obtain any integer value larger or
equal to zero. The event of a radio active decay of an atom is a relatively rare
event. Therefore, the Poisson model is likely to fit this phenomena3 .

Example 5.4. Consider an overhead power line suspended between two utility
poles. During rain, drops of water may hit the power line. The total number
of drops that hit the line in a one minute period may be modeled by a Poisson
random variable.

5.3 Continuous Random Variable

Many types of measurements, such as height, weight, angle, temperature, etc.,
may in principle have a continuum of possible values. Continuous random vari-
ables are used to model uncertainty regarding future values of such measure-
ments.
The main difference between discrete random variables, which is the type we
examined thus far, and continuous random variable, that are added now to the
list, is in the sample space, i.e., the collection of possible outcomes. The former
3 The number of decays may also be considered in the Binomial(n, p) setting. The number

n is the total number of atoms in the unit mass and p is the probability that an atom decays
within the given second. However, since n is very large and p is very small we get that the
Poisson distribution is an appropriate model for the count.
5.3. CONTINUOUS RANDOM VARIABLE 75

type is used when the possible outcomes are separated from each other as the
integers are. The latter type is used when the possible outcomes are the entire
line of real numbers or when they form an interval (possibly an open ended one)
of real numbers.
The difference between the two types of sample spaces implies differences in
the way the distribution of the random variables is being described. For discrete
random variables one may list the probability associated with each value in the
sample space using a table, a formula, or a bar plot. For continuous random
variables, on the other hand, probabilities are assigned to intervals of values,
and not to specific values. Thence, densities are used in order to display the
distribution.
Densities are similar to histograms, with areas under the plot corresponding
to probabilities. We will provide a more detailed description of densities as we
discuss the different examples of continuous random variables.
In continuous random variables integration replaces summation and the den-
sity replaces the probability in the computation of quantities such as the prob-
ability of an event, the expectation, and the variance.
Hence,Pif the expectation

of a discrete random variable is given in the formula
E(X) = x x × P(x) , which involves the summation over all values of the
product between the value and the probability of the value, then for continuous
random variable the definition becomes:
Z


E(X) = x × f (x) dx ,

where f (x) is the density of X at the value x. Therefore, in the expectation

of a continuous random variable one multiplies the value by the density at the
value. This product is then integrated
Pover the sample space.

Likewise, the formula Var(X) = x (x − E(X))2 × P(x) for the variance
is replaced by: Z
(x − E(X))2 × f (x) dx .


Var(X) =

Nonetheless, the intuitive interpretation of the expectation as the central value

of the distribution that identifies the location and the interpretation of the
standard deviation (the square root of the variance) as the summary of the
total spread of the distribution is still valid.
In this section we will describe two types of continuous random variables:
Uniform and Exponential. In the next chapter another example – the Normal
distribution – will be introduced.

5.3.1 The Uniform Random Variable

The Uniform distribution is used in order to model measurements that may
have values in a given interval, with all values in this interval equally likely to
occur.
For example, consider a random variable X with the Uniform distribution
over the interval [3, 7], denoted by “X ∼ Uniform(3, 7)”. The density function
at given values may be computed with the aid of the function “dunif”. For
instance let us compute the density of the Uniform(3, 7) distribution over the
integers {0, 1, . . . , 10}:
76 CHAPTER 5. RANDOM VARIABLES

0.25
0.20

P(X <= 4.73)

0.15
Density

0.10
0.05

x = 4.73
0.00

0 2 4 6 8 10

Figure 5.5: The Uniform(3,7) Distribution

> dunif(0:10,3,7)
[1] 0.00 0.00 0.00 0.25 0.25 0.25 0.25 0.25 0.00 0.00 0.00

Notice that for the values 0, 1, and 2, and the values 8, 9 and 10 that are outside
of the interval the density is equal to zero, indicating that such values cannot
occur in the given distribution. The values of the density at integers inside the
interval are positive and equal to each other. The density is not restricted to
integer values. For example, at the point 4.73 we get that the density is positive
and of the same height:

> dunif(4.73,3,7)
[1] 0.25

A plot of the Uniform(3, 7) density is given in Figure 5.5 in the form of a

solid line. Observe that the density is positive over the interval [3, 7] where its
height is 1/4. Area under the curve in the density corresponds to probability.
Indeed, the fact that the total probability is one is reflected in the total area
under the curve being equal to 1. Over the interval [3, 7] the density forms a
rectangle. The base of the rectangle is the length of the interval 7 − 3 = 4. The
5.3. CONTINUOUS RANDOM VARIABLE 77

height of the rectangle is thus equal to 1/4 in order to produce a total area of
4 × (1/4) = 1.
The function “punif” computes the cumulative probability of the uniform
distribution. The probability P(X ≤ 4.73), for X ∼ Uniform(3, 7), is given by:
> punif(4.73,3,7)
[1] 0.4325
This probability corresponds to the marked area to the left of the point x = 4.73
in Figure 5.5. This area of the marked rectangle is equal to the length of the
base 4.73 - 3 = 1.73, times the height of the rectangle 1/(7-3) = 1/4. Indeed:
> (4.73-3)/(7-3)
[1] 0.4325
is the area of the marked rectangle and is equal to the probability.
Let us use R in order to plot the density and the cumulative probability
functions of the Uniform distribution. We produce first a large number of points
in the region we want to plot. The points are produced with aid of the function
“seq”. The output of this function is a sequence with equally spaced values.
The starting value of the sequence is the first argument in the input of the
function and the last value is the second argument in the input. The argument
“length=1000” sets the length of the sequence, 1,000 values in this case:
> x <- seq(0,10,length=1000)
> den <- dunif(x,3,7)
> plot(x,den)
The object “den” is a sequence of length 1,000 that contains the density of the
Uniform(3, 7) evaluated over the values of “x”. When we apply the function
“plot” to the two sequences we get a scatter plot of the 1,000 points that is
presented in the upper panel of Figure 5.6.
A scatter plot is a plot of points. Each point in the scatter plot is identify by
its horizontal location on the plot (its “x” value) and by its vertical location on
the plot (its y value). The horizontal value of each point in the plot is determined
by the first argument to the function “plot” and the vertical value is determined
by the second argument. For example, the first value in the sequence “x” is 0.
The value of the Uniform density at this point is 0. Hence, the first value of the
sequence “den” is also 0. A point that corresponds to these values is produced
in the plot. The horizontal value of the point is 0 and the vertical value is 0. In
a similar way the other 999 points are plotted. The last point to be plotted has
a horizontal value of 10 and a vertical value of 0.
The number of points that are plotted is large and they overlap each other
in the graph and thus produce an impression of a continuum. In order to obtain
nicer looking plots we may choose to connect the points to each other with
segments and use smaller points. This may be achieved by the addition of the
argument “type="l"”, with the letter "l" for line, to the plotting function:
> plot(x,den,type="l")
The output of the function is presented in the second panel of Figure 5.6. In
the last panel the cumulative probability of the Uniform(3, 7) is presented. This
function is produced by the code:
78 CHAPTER 5. RANDOM VARIABLES

●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●

0.15
den

0.00

●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
● ●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●●
●
●

0 2 4 6 8 10

x
0.15
den

0.00

0 2 4 6 8 10

x
0.8
cdf

0.4
0.0

0 2 4 6 8 10

Figure 5.6: The Density and Cumulative Probability of Uniform(3,7)

> cdf <- punif(x,3,7)

> plot(x,cdf,type="l")
One can think of the density of the Uniform as an histogram4 . The expecta-
tion of a Uniform random variable is the middle point of it’s histogram. Hence,
if X ∼ Uniform(a, b) then:
a+b
E(X) = .
2
For the X ∼ Uniform(3, 7) distribution the expectation is E(X) = (3+7)/2 = 5.
Observe that 5 is the center of the Uniform density in Plot 5.5.
It can be shown that the variance of the Uniform(a, b) is equal to
(b − a)2
, Var(X) =
12
with the standard deviation being the square root of this value. Specifically, for
X ∼ Uniform(3, 7) we√get that Var(X) = (7−3)2 /12 = 1.333333. The standard
deviation is equal to 1.333333 = 1.154701.
4 If X ∼ Uniform(a, b) then the density is f (x) = 1/(b − a), for a ≤ x ≤ b, and it is equal

to 0 for other values of x.

5.3. CONTINUOUS RANDOM VARIABLE 79

Example 5.5. In Example 5.4 we considered rain drops that hit an overhead
power line suspended between two utility poles. The number of drops that hit
the line can be modeled using the Poisson distribution. The position between
the two poles where a rain drop hits the line can be modeled by the Uniform
distribution. The rain drop can hit any position between the two utility poles.
Hitting one position along the line is as likely as hitting any other position.
Example 5.6. Meiosis is the process in which a diploid cell that contains two
copies of the genetic material produces an haploid cell with only one copy (sperms
or eggs, depending on the sex). The resulting molecule of genetic material is
linear molecule (chromosome) that is composed of consecutive segments: a seg-
ment that originated from one of the two copies followed by a segment from
the other copy and vice versa. The border points between segments are called
points of crossover. The Haldane model for crossovers states that the position
of a crossover between two given loci on the chromosome corresponds to the
Uniform distribution and the total number of crossovers between these two loci
corresponds to the Poisson distribution.

5.3.2 The Exponential Random Variable

The Exponential distribution is frequently used to model times between events.
For example, times between incoming phone calls, the time until a component
becomes malfunction, etc. We denote the Exponential distribution via “X ∼
Exponential(λ)”, where λ is a parameter that characterizes the distribution and
is called the rate of the distribution. The overlap between the parameter used
to characterize the Exponential distribution and the one used for the Poisson
distribution is deliberate. The two distributions are tightly interconnected. As
a matter of fact, it can be shown that if the distribution between occurrences
of a phenomena has the Exponential distribution with rate λ then the total
number of the occurrences of the phenomena within a unit interval of time has
a Poisson(λ) distribution.
The sample space of an Exponential random variable contains all non-negative
numbers. Consider, for example, X ∼ Exponential(0.5). The density of the dis-
tribution in the range between 0 and 10 is presented in Figure 5.7. Observe
that in the Exponential distribution smaller values are more likely to occur in
comparison to larger values. This is indicated by the density being larger at
the vicinity of 0. The density of the exponential distribution given in the plot
is positive, but hardly so, for values larger than 10.
The density of the Exponential distribution can be computed with the aid
of the function “dexp”5 . The cumulative probability can be computed with
the function “pexp”. For illustration, assume X ∼ Exponential(0.5). Say one
is interested in the computation of the probability P(2 < X ≤ 6) that the
random variable obtains a value that belongs to the interval (2, 6]. The required
probability is indicated as the marked area in Figure 5.7. This area can be
computed as the difference between the probability P(X ≤ 6), the area to the
left of 6, and the probability P(X ≤ 2), the area to the left of 2:
> pexp(6,0.5) - pexp(2,0.5)
[1] 0.3180924
5 If X ∼ Exponential(λ) then the density is f (x) = λe−λx , for 0 ≤ x, and it is equal to 0

for x < 0.
80 CHAPTER 5. RANDOM VARIABLES

0.5
0.4
0.3
Density

0.2

P(2 < X < 6)

0.1
0.0

0 2 4 6 8 10

Figure 5.7: The Exponential(0.5) Distribution

The difference is the probability of belonging to the interval, namely the area
marked in the plot.
The expectation of X, when X ∼ Exponential(λ), is given by the equation:

E(X) = 1/λ ,

and the variance is given by:

Var(X) = 1/λ2 .

The standard deviation is the square root of the variance, namely 1/λ. Observe
that the larger is the rate the smaller are the expectation and the standard
deviation.
In Figure 5.8 the densities of the Exponential distribution are plotted for
λ = 0.5, λ = 1, and λ = 2. Notice that with the increase in the value of the
parameter then the values of the random variable tends to become smaller. This
inverse relation makes sense in connection to the Poisson distribution. Recall
that the Poisson distribution corresponds to the total number of occurrences in
a unit interval of time when the time between occurrences has an Exponential
5.3. CONTINUOUS RANDOM VARIABLE 81

2.0

lambda = 0.5
lambda = 1
lambda = 2
1.5
Density

1.0
0.5
0.0

0 2 4 6 8 10

Figure 5.8: The Exponential Distribution for Various Values of λ

distribution. A larger expectation λ of the Poisson corresponds to a larger

number of occurrences that are likely to take place during the unit interval of
time. The larger is the number of occurrences the smaller are the time intervals
between occurrences.

Example 5.7. Consider Examples 5.4 and 5.5 that deal with rain dropping on
a power line. The times between consecutive hits of the line may be modeled by
the Exponential distribution. Hence, the time to the first hit has an Exponential
distribution. The time between the first and the second hit is also Exponentially
distributed, and so on.

Example 5.8. Return to Example 5.3 that deals with the radio activity of some
element. The total count of decays per second is model by the Poisson dis-
tribution. The times between radio active decays is modeled according to the
Exponential distribution. The rate λ of that Exponential distribution is equal to
the expectation of the total count of decays in one second, i.e. the expectation of
the Poisson distribution.
82 CHAPTER 5. RANDOM VARIABLES

5.4 Solved Exercises

Question 5.1. A particular measles vaccine produces a reaction (a fever higher
that 102 Fahrenheit) in each vaccinee with probability of 0.09. A clinic vacci-
nates 500 people each day.
1. What is the expected number of people that will develop a reaction each
day?
2. What is the standard deviation of the number of people that will develop
a reaction each day?
3. In a given day, what is the probability that more than 40 people will
develop a reaction?
4. In a given day, what is the probability that the number of people that will
develop a reaction is between 50 and 45 (inclusive)?
Solution (to Question 5.1.1): The Binomial distribution is a reasonable
model for the number of people that develop high fever as result of the vac-
cination. Let X be the number of people that do so in a give day. Hence,
X ∼ Binomial(500, 0.09). According to the formula for the expectation in the
Binomial distribution, since n = 500 and p = 0.09, we get that:

E(X) = np = 500 × 0.09 = 45 .

Solution (to Question 5.1.2): Let X ∼ Binomial(500, 0.09). Using the

formula for the variance for the Binomial distribution we get that:

Var(X) = np(1 − p) = 500 × 0.09 × 0.91 = 40.95 .

p √
Hence, since V ar(X) = 40.95 = 6.3992, the standard deviation is 6.3992.

Solution (to Question 5.1.3): Let X ∼ Binomial(500, 0.09). The probability

that more than 40 people will develop a reaction may be computed as the
difference between 1 and the probability that 40 people or less will develop a
reaction:
P(X > 40) = 1 − P(X ≤ 40) .
The probability can be computes with the aid of the function “pbinom” that
produces the cumulative probability of the Binomial distribution:
> 1 - pbinom(40,500,0.09)
[1] 0.7556474

Solution (to Question 5.1.4): The probability that the number of people
that will develop a reaction is between 50 and 45 (inclusive) is the difference
between P(X ≤ 50) and P(X < 45) = P(X ≤ 44). Apply the function “pbinom”
to get:
> pbinom(50,500,0.09) - pbinom(44,500,0.09)
[1] 0.3292321
5.4. SOLVED EXERCISES 83

Question 5.2. The Negative-Binomial distribution is yet another example of a

discrete, integer valued, random variable. The sample space of the distribution
are all non-negative integers {0, 1, 2, . . .}. The fact that a random variable X
has this distribution is marked by “X ∼ Negative-Binomial(r, p)”, where r and
p are parameters that specify the distribution.
Consider 3 random variables from the Negative-Binomial distribution:
• X1 ∼ Negative-Binomial(2, 0.5)
• X2 ∼ Negative-Binomial(4, 0.5)
• X3 ∼ Negative-Binomial(8, 0.8)
The bar plots of these random variables are presented in Figure 5.9, re-organizer
in a random order.
1. Produce bar plots of the distributions of the random variables X1 , X2 , X3
in the range of integers between 0 and 15 and thereby identify the pair of
parameters that produced each one of the plots in Figure 5.9. Notice that
the bar plots can be produced with the aid of the function “plot” and the
function “dnbinom(x,r,p)”, where “x” is a sequence of integers and “r”
and “p” are the parameters of the distribution. Pay attention to the fact
that you should use the argument “type = "h"” in the function “plot”
in order to produce the horizontal bars.
2. Below is a list of pairs that includes an expectation and a variance. Each
of the pairs is associated with one of the random variables X1 , X2 , and
X3 :
(a) E(X) = 4, Var(X) = 8.
(b) E(X) = 2, Var(X) = 4.
(c) E(X) = 2, Var(X) = 2.5.
Use Figure 5.9 in order to match the random variable with its associated
pair. Do not use numerical computations or formulae for the expectation
and the variance in the Negative-Binomial distribution in order to carry
out the matching6 . Use, instead, the structure of the bar-plots.
Solution (to Question 5.2.1): The plots can be produced with the following
code, which should be run one line at a time:
> x <- 0:15
> plot(x,dnbinom(x,2,0.5),type="h")
> plot(x,dnbinom(x,4,0.5),type="h")
> plot(x,dnbinom(x,8,0.8),type="h")
The first plot, that corresponds to X1 ∼ Negative-Binomial(2, 0.5), fits
Barplot 3. Notice that the distribution tends to obtain smaller values and that
the probability of the value “0” is equal to the probability of the value “1”.
The second plot, the one that corresponds to X2 ∼ Negative-Binomial(4, 0.5),
is associated with Barplot 1. Notice that the distribution tends to obtain larger
6 It can be shown, or else found on the web, that if X ∼ Negative-Binomial(r, p) then

E(X) = r(1 − p)/p and Var(X) = r(1 − p)/p2 .

84 CHAPTER 5. RANDOM VARIABLES

values. For example, the probability of the value “10” is substantially larger
than zero, where for the other two plots this is not the case.
The third plot, the one that corresponds to X3 ∼ Negative-Binomial(8, 0.8),
matches Barplot 2. Observe that this distribution tends to produce smaller
probabilities for the small values as well as for the larger values. Overall, it is
more concentrated than the other two.

Solution (to Question 5.2.2): Barplot 1 corresponds to a distribution that

tends to obtain larger values than the other two distributions. Consequently,
the expectation of this distribution should be larger. The conclusion is that the
pair E(X) = 4, Var(X) = 8 should be associated with this distribution.
Barplot 2 describes a distribution that produce smaller probabilities for the
small values as well as for the larger values and is more concentrated than the
other two. The expectations of the two remaining distributions are equal to
each other and the variance of the pair E(X) = 2, Var(X) = 2.5 is smaller.
Consequently, this is the pair that should be matched with this box plot.
This leaves only Barplot 3, that should be matched with the pair E(X) = 2,
Var(X) = 4.

5.5 Summary
Glossary
Binomial Random Variable: The number of successes among n repeats of
independent trials with a probability p of success in each trial. The dis-
tribution is marked as Binomial(n, p).

Poisson Random Variable: An approximation to the number of occurrences

of a rare event, when the expected number of events is λ. The distribution
is marked as Poisson(λ).

Density: Histogram that describes the distribution of a continuous random

variable. The area under the curve corresponds to probability.

Uniform Random Variable: A model for a measurement with equally likely

outcomes over an interval [a, b]. The distribution is marked as Uniform(a, b).

Exponential Random Variable: A model for times between events. The

distribution is marked as Exponential(λ).

Discuss in the Forum

This unit deals with two types of discrete random variables, the Binomial and
the Poisson, and two types of continuous random variables, the Uniform and
the Exponential. Depending on the context, these types of random variables
may serve as theoretical models of the uncertainty associated with the outcome
of a measurement.
In your opinion, is it or is it not useful to have a theoretical model for a
situation that occurs in real life?
5.5. SUMMARY 85

When forming your answer to this question you may give an example of a
situation from you own field of interest for which a random variable, possibly
from one of the types that are presented in this unit, can serve as a model.
Discuss the importance (or lack thereof) of having a theoretical model for the
situation.
For example, the Exponential distribution may serve as a model for the
time until an atom of a radio active element decays by the release of subatomic
particles and energy. The decay activity is measured in terms of the number of
decays per second. This number is modeled as having a Poisson distribution.
Its expectation is the rate of the Exponential distribution. For the radioactive
element Carbon-14 (14 C) the decay rate is 3.8394 × 10−12 particles per second.
Computations that are based on the Exponential model may be used in order
to date ancient specimens.

Summary of Formulas
Discrete Random Variable:
X

E(X) = x × P(x)
x
X
(x − E(X))2 × P(x)


Var(X) =
x

Continuous Random Variable:

Z


E(X) = x × f (x) dx
Z
(x − E(X))2 × f (x) dx


Var(X) =

Binomial: E(X) = np , Var(X) = np(1 − p)

Poisson: E(X) = λ , Var(X) = λ
Uniform: E(X) = (a + b)/2 , Var(X) = (b − a)2 /12
Exponential: E(X) = 1/λ , Var(X) = 1/λ2
86 CHAPTER 5. RANDOM VARIABLES

0.10
0.00

0 5 10 15

Barplot 1
0.20
0.10
0.00

0 5 10 15

Barplot 2
0.20
0.10
0.00

0 5 10 15

Barplot 3

Figure 5.9: Bar Plots of the Negative-Binomial Distribution

Chapter 6

The Normal Random

Variable

6.1 Student Learning Objective

This chapter introduces a very important bell-shaped distribution known as
the Normal distribution. Computations associated with this distribution are
discussed, including the percentiles of the distribution and the identification of
intervals of subscribed probability. The Normal distribution may serve as an ap-
proximation to other distributions. We demonstrate this property by showing
that under appropriate conditions the Binomial distribution can be approxi-
mated by the Normal distribution. This property of the Normal distribution
will be picked up in the next chapter where the mathematical theory that estab-
lishes the Normal approximation is demonstrated. By the end of this chapter,
the student should be able to:

• Recognize the Normal density and apply R functions for computing Normal
probabilities and percentiles.

• Associate the distribution of a Normal random variable with that of its

standardized counterpart, which is obtained by centering and re-scaling.

• Use the Normal distribution to approximate the Binomial distribution.

6.2 The Normal Random Variable

The Normal distribution is the most important of all distributions that are used
in statistics. In many cases it serves as a generic model for the distribution of
a measurement. Moreover, even in cases where the measurement is modeled
by other distributions (i.e. Binomial, Poisson, Uniform, Exponential, etc.) the
Normal distribution emerges as an approximation of the distribution of numer-
ical characteristics of the data produced by such measurements.

87
88 CHAPTER 6. THE NORMAL RANDOM VARIABLE

0.12
P(0 < X < 5)
0.10
0.08
Density

0.06
0.04
0.02
0.00

−10 −5 0 5 10

Figure 6.1: The Normal(2,9) Distribution

6.2.1 The Normal Distribution

A Normal random variable has a continuous distribution over the sample space
of all numbers, negative or positive. We denote the Normal distribution via
“X ∼ Normal(µ, σ 2 )”, where µ = E(X) is the expectation of the random vari-
able and σ 2 = Var(X) is it’s variance1 .
Consider, for example, X ∼ Normal(2, 9). The density of the distribution
is presented in Figure 6.1. Observe that the distribution is symmetric about
the expectation 2. The random variable is more likely to obtain its value in
the vicinity of the expectation. Values much larger or much smaller than the
expectation are substantially less likely.
The density of the Normal distribution can be computed with the aid of
the function “dnorm”. The cumulative probability can be computed with the
function “pnorm”. For illustrating the use of the latter function, assume that
X ∼ Normal(2, 9). Say one is interested in the computation of the probability
P(0 < X ≤ 5) that the random variable obtains a value that belongs to the
1 If X ∼ Normal(µ, σ 2 ) then the density of X is given by the formula f (x) =
(x−µ)2
√
exp{− 2σ2 }/ 2πσ 2 , for all x.
6.2. THE NORMAL RANDOM VARIABLE 89

0.8

N(0,1)
N(2,9)
N(−3,0.25)
0.6
Density

0.4
0.2
0.0

−10 −5 0 5 10

Figure 6.2: The Normal Distribution for Various Values of µ and σ 2

interval (0, 5]. The required probability is indicated by the marked area in
Figure 6.1. This area can be computed as the difference between the probability
P(X ≤ 5), the area to the left of 5, and the probability P(X ≤ 0), the area to
the left of 0:
> pnorm(5,2,3) - pnorm(0,2,3)
[1] 0.5888522
The difference is the indicated area that corresponds to the probability of being
inside the interval, which turns out to be approximately equal to 0.589. Notice
that the expectation µ of the Normal distribution is entered as the second
argument to the function. The third argument to the function is the standard
deviation, i.e.
√ the square root of the variance. In this example, the standard
deviation is 9 = 3.
Figure 6.2 displays the densities of the Normal distribution for the combina-
tions µ = 0, σ 2 = 1 (the red line); µ = 2, σ 2 = 9 (the black line); and µ = −3,
σ 2 = 1/4 (the green line). Observe that the smaller the variance the more
concentrated is the distribution of the random variable about the expectation.
90 CHAPTER 6. THE NORMAL RANDOM VARIABLE

Example 6.1. IQ tests are a popular (and controversial) mean for measuring
intelligence. They are produced as (weighted) average of a response to a long
list of questions, designed to test different abilities. The score of the test across
the entire population is set to be equal to 100 and the standard deviation is set
to 15. The distribution of the score is Normal. Hence, if X is the IQ score of a
random subject then X ∼ Normal(100, 152 ).

Example 6.2. Any measurement that is produced as a result of the combination

of many independent influencing factors is likely to poses the Normal distribu-
tion. For example, the hight of a person is influenced both by genetics and by
the environment in which that person grew up. Both the genetic and the envi-
ronmental influences are a combination of many factors. Thereby, it should not
come as a surprise that the heights of people in a population tend to follow the
Normal distribution.

6.2.2 The Standard Normal Distribution

The standard normal distribution is a normal distribution of standardized val-
ues, which are called z-scores. A z-score is the original measurement measured
in units of the standard deviation from the expectation. For example, if √ the
expectation of a Normal distribution is 2 and the standard deviation is 3 = 9,
then the value of 0 is 2/3 standard deviations smaller than (or to the left of)
the expectation. Hence, the z-score of the value 0 is -2/3. The calculation of
the z-score emerges from the equation:

(0 =) x = µ + z · σ (= 2 + z · 3)

The z-score is obtained by solving the equation

0=2+z·3 =⇒ z = (0 − 2)/3 = −2/3 .

In a similar way, the z-score of the value x = 5 is equal to 1, following the

solution of the equation 5 = 2 + z · 3, which leads to z = (5 − 2)/3 = 1.
The standard Normal distribution is the distribution of a standardized Nor-
mal measurement. The expectation for the standard Normal distribution is
0 and the variance is 1. When X ∼ N (µ, σ 2 ) has a Normal distribution
with expectation µ and variance σ 2 then the transformed random variable
Z = (X − µ)/σ produces the standard Normal distribution Z ∼ N (0, 1). The
transformation corresponds to the reexpression of the original measurement in
terms of a new “zero” and a new unit of measurement. The new “zero” is
the expectation of the original measurement and the new unit is the standard
deviation of the original measurement.
Computation of probabilities associated with a Normal random variable X
can be carried out with the aid of the standard Normal distribution. For exam-
ple, consider the computation of the probability P(0 < X ≤ 5) for X ∼ N (2, 9),
that has expectation µ = 2 and standard deviation σ = 3. Consider X’s stan-
dardized values: Z = (X − 2)/3. The boundaries of the interval [0, 5], namely
0 and 5, have standardized z-scores of (0 − 2)/3 = −2/3 and (5 − 2)/3 = 1,
respectively. Clearly, the original measurement X falls between the original
boundaries (0 < X ≤ 5) if, and only if, the standardized measurement Z falls
6.2. THE NORMAL RANDOM VARIABLE 91

0.4

P((0−2)/3 < Z < (5−2)/3)

0.3
Density

0.2
0.1
0.0

−4 −2 0 2 4

Figure 6.3: The Standard Normal Distribution

between the standardized boundaries (−2/3 < Z ≤ 1). Therefore, the proba-
bility that X obtains a value in the range [0, 5] is equal to the probability that
Z obtains a value in the range [−2/3, 1].
The function “pnorm” was used in the previous subsection in order to com-
pute that probability that X obtains values between 0 and 5. The computation
produced the probability 0.5888522. We can repeat the computation by the
application of the same function to the standardized values:
> pnorm((5-2)/3) - pnorm((0-2)/3)
[1] 0.5888522
The value that is being computed, the area under the graph for the standard
Normal distribution, is presented in Figure 6.3. Recall that 3 arguments where
specified in the previous application of the function “pnorm”: the x value, the
expectation, and the standard deviation. In the given application we did not
specify the last two arguments, only the first one. (Notice that the output of
the expression “(5-2)/3” is a single number and, likewise, the output of the
expression “(0-2)/3” is also a single number.)
Most R function have many arguments that enables flexible application in a
92 CHAPTER 6. THE NORMAL RANDOM VARIABLE

wide range of settings. For convenience, however, default values are set to most
of these arguments. These default values are used unless an alternative value for
the argument is set when the function is called. The default value of the second
argument of the function “pnorm” that specifies the expectation is “mean=0”,
and the default value of the third argument that specifies the standard devi-
ation is “sd=1”. Therefore, if no other value is set for these arguments the
function computes the cumulative distribution function of the standard Normal
distribution.

6.2.3 Computing Percentiles

Consider the issue of determining the range that contains 95% of the probability
for a Normal random variable. We start with the standard Normal distribution.
Consult Figure 6.4. The figure displays the standard Normal distribution with
the central region shaded. The area of the shaded region is 0.95.
We may find the z-values of the boundaries of the region, denoted in the
figure as z0 and z1 by the investigation of the cumulative distribution function.
Indeed, in order to have 95% of the distribution in the central region one should
leave out 2.5% of the distribution in each of the two tails. That is, 0.025
should be the area of the unshaded region to the right of z1 and, likewise, 0.025
should be the area of the unshaded region to the left of z0 . In other words, the
cumulative probability up to z0 should be 0.025 and the cumulative distribution
up to z1 should be 0.975.
In general, given a random variable X and given a percent p, the x value with
the property that the cumulative distribution up to x is equal to the probability
p is called the p-percentile of the distribution. Here we seek the 2.5%-percentile
and the 97.5%-percentile of the standard Normal distribution.
The percentiles of the Normal distribution are computed by the function
“qnorm”. The first argument to the function is a probability (or a sequence
of probabilities), the second and third arguments are the expectation and the
standard deviations of the normal distribution. The default values to these argu-
ments are set to 0 and 1, respectively. Hence if these arguments are not provided
the function computes the percentiles of the standard Normal distribution. Let
us apply the function in order to compute z1 and z0 :
> qnorm(0.975)
[1] 1.959964
> qnorm(0.025)
[1] -1.959964
Observe that z1 is practically equal to 1.96 and z0 = −1.96 = −z1 . The fact
that z0 is the negative of z1 results from the symmetry of the standard Normal
distribution about 0. As a conclusion we get that for the standard Normal
distribution 95% of the probability is concentrated in the range [−1.96, 1.96].
The problem of determining the central range that contains 95% of the
distribution can be addresses in the context of the original measurement X (See
Figure 6.5). We seek in this case an interval centered at the expectation 2,
which is the center of the distribution of X, unlike 0 which was the center of the
standardized values Z. One way of solving the problem is via the application of
the function “qnorm” with the appropriate values for the expectation and the
standard deviation:
6.2. THE NORMAL RANDOM VARIABLE 93

0.4

P( z0< Z < z1 ) = 0.95

0.3
Density

0.2
0.1

z0 = −1.96 z1 = 1.96
0.0

−4 −2 0 2 4

Figure 6.4: Central 95% of the Standard Normal Distribution

> qnorm(0.975,2,3)
[1] 7.879892
> qnorm(0.025,2,3)
[1] -3.879892
Hence, we get that x0 = −3.88 has the property that the total probability to
its left is 0.025 and x1 = 7.88 has the property that the total probability to its
right is 0.025. The total probability in the range [−3.88, 7.88] is 0.95.
An alternative approach for obtaining the given interval exploits the inter-
val that was obtained for the standardized values. An interval [−1.96, 1.96] of
standardized z-values corresponds to an interval [2 − 1.96 · 3, 2 + 1.96 · 3] of the
original x-values:
> 2 + qnorm(0.975)*3
[1] 7.879892
> 2 + qnorm(0.025)*3
[1] -3.879892
Hence, we again produce the interval [−3.88, 7.88], the interval that was ob-
tained before as the central interval that contains 95% of the distribution of the
94 CHAPTER 6. THE NORMAL RANDOM VARIABLE

0.12

P( x0< X < x1 ) = 0.95

0.10
0.08
Density

0.06
0.04
0.02

x0 = ? x1 = ?
0.00

−10 −5 0 5 10

Figure 6.5: Central 95% of the Normal(2,9) Distribution

Normal(2, 9) random variable.

In general, if X ∼ N (µ, σ) is a Normal random variable then the interval
[µ − 1.96 · σ, µ + 1.96 · σ] contains 95% of the distribution of the random variable.
Frequently one uses the notation µ ± 1.96 · σ to describe such an interval.

6.2.4 Outliers and the Normal Distribution

Consider, next, the computation of the interquartile range in the Normal distri-
bution. Recall that the interquartile range is the length of the central interval
that contains 50% of the distribution. This interval starts at the first quartile
(Q1), the value that splits the distribution so that 25% of the distribution is
to the left of the value and 75% is to the right of it. The interval ends at the
third quartile (Q3) where 75% of the distribution is to the left and 25% is to
the right.
For the standard Normal the third and first quartiles can be computed with
the aid of the function “qnorm”:

> qnorm(0.75)
6.2. THE NORMAL RANDOM VARIABLE 95

[1] 0.6744898
> qnorm(0.25)
[1] -0.6744898

Observe that for the standard Normal distribution one has that 75% of the
distribution is to the left of the value 0.6744898, which is the third quartile of this
distribution. Likewise, 25% of the standard Normal distribution are to the left
of the value -0.6744898, which is the first quartile. the interquartile range is the
length of the interval between the third and the first quartiles. In the case of the
standard Normal distribution this length is equal to 0.6744898−(−0.6744898) =
1.348980.
In Chapter 3 we considered box plots as a mean for the graphical display
of numerical data. The box plot includes a vertical rectangle that initiates at
the first quartile and ends at the third quartile, with the median marked within
the box. The rectangle contains 50% of the data. Whiskers extends from the
ends of this rectangle to the smallest and to the largest data values that are not
outliers. Outliers are values that lie outside of the normal range of the data.
Outliers are identified as values that are more then 1.5 times the interquartile
range away from the ends of the central rectangle. Hence, a value is an outlier
if it is larger than the third quartile plus 1.5 times the interquartile range or if
it is less than the first quartile minus 1.5 times the interquartile range.
How likely is it to obtain an outlier value when the measurement has the
standard Normal distribution? We obtained that the third quartile of the stan-
dard Normal distribution is equal to 0.6744898 and the first quartile is minus
this value. The interquartile range is the difference between the third and first
quartiles. The upper and lower thresholds for the defining outliers are:

> qnorm(0.75) + 1.5*(qnorm(0.75)-qnorm(0.25))

[1] 2.697959
> qnorm(0.25) - 1.5*(qnorm(0.75)-qnorm(0.25))
[1] -2.697959

Hence, a value larger than 2.697959 or smaller than -2.697959 would be identified
as an outlier.
The probability of being less than the upper threshold 2.697959 in the stan-
dard Normal distribution is computed with the expression “pnorm(2.697959)”.
The probability of being above the threshold is 1 minus that probability, which
is the outcome of the expression “1-pnorm(2.697959)”.
By the symmetry of the standard Normal distribution we get that the prob-
ability of being below the lower threshold -2.697959 is equal to the probability
of being above the upper threshold. Consequently, the probability of obtaining
an outlier is equal to twice the probability of being above the upper threshold:

> 2*(1-pnorm(2.697959))
[1] 0.006976603

We get that for the standard Normal distribution the probability of an outlier
is approximately 0.7%.
96 CHAPTER 6. THE NORMAL RANDOM VARIABLE

6.3 Approximation of the Binomial Distribution

The Normal distribution emerges frequently as an approximation of the dis-
tribution of data characteristics. The probability theory that mathematically
establishes such approximation is called the Central Limit Theorem and is the
subject of the next chapter. In this section we demonstrate the Normal approx-
imation in the context of the Binomial distribution.

6.3.1 Approximate Binomial Probabilities and Percentiles

Consider, for example, the probability of obtaining between 1940 and 2060 heads
when tossing 4,000 fair coins. Let X be the total number of heads. The tossing of
a coin is a trial with two possible outcomes: “Head” and “Tail.” The probability
of a “Head” is 0.5 and there are 4,000 trials. Let us call obtaining a “Head” in a
trial a “Success”. Observe that the random variable X counts the total number
of successes. Hence, X ∼ Binomial(4000, 0.5).
The probability P(1940 ≤ X ≤ 2060) can be computed as the difference
between the probability P(X ≤ 2060) of being less or equal to 2060 and the
probability P(X < 1940) of being strictly less than 1940. However, 1939 is the
largest integer that is still strictly less than the integer 1940. As a result we
get that P(X < 1940) = P(X ≤ 1939). Consequently, P(1940 ≤ X ≤ 2060) =
P(X ≤ 2060) − P(X ≤ 1939).
Applying the function “pbinom” for the computation of the Binomial cumu-
lative probability, namely the probability of being less or equal to a given value,
we get that the probability in the range between 1940 and 2060 is equal to
> pbinom(2060,4000,0.5) - pbinom(1939,4000,0.5)
[1] 0.9442883
This is an exact computation. The Normal approximation produces an ap-
proximate evaluation, not an exact computation. The Normal approximation
replaces Binomial computations by computations carried out for the Normal
distribution. The computation of a probability for a Binomial random variable
is replaced by computation of probability for a Normal random variable that has
the same expectation and standard deviation as the Binomial random variable.
Notice that if X ∼ Binomial(4000, 0.5) then the expectation is E(X) =
4, 000 × 0.5 = 2, 000 and the variance is Var(X) = 4, 000 × 0.5 × 0.5 = 1, 000,
with the standard deviation being the square root of the variance. Repeating
the same computation that we conducted for the Binomial random variable,
but this time with the function “pnorm” that is used for the computation of the
Normal cumulative probability, we get:
> mu <- 4000*0.5
> sig <- sqrt(4000*0.5*0.5)
> pnorm(2060,mu,sig) - pnorm(1939,mu,sig)
[1] 0.9442441
Observe that in this example the Normal approximation of the probability
(0.9442441) agrees with the Binomial computation of the probability (0.9442883)
up to 3 significant digits.
Normal computations may also be applied in order to find approximate per-
centiles of the Binomial distribution. For example, let us identify the central
6.3. APPROXIMATION OF THE BINOMIAL DISTRIBUTION 97

region that contains for a Binomial(4000, 0.5) random variable (approximately)

95% of the distribution. Towards that end we can identify the boundaries of
the region for the Normal distribution with the same expectation and standard
deviation as that of the target Binomial distribution:

> qnorm(0.975,mu,sig)
[1] 2061.980
> qnorm(0.025,mu,sig)
[1] 1938.020

After rounding to the nearest integer we get the interval [1938, 2062] as a pro-
posed central region.
In order to validate the proposed region we may repeat the computation
under the actual Binomial distribution:

> qbinom(0.975,4000,0.5)
[1] 2062
> qbinom(0.025,4000,0.5)
[1] 1938

Again, we get the interval [1938, 2062] as the central region, in agreement with
the one proposed by the Normal approximation. Notice that the function
“qbinom” produces the percentiles of the Binomial distribution. It may not
come as a surprise to learn that “qpois”, “qunif”, “qexp” compute the per-
centiles of the Poisson, Uniform and Exponential distributions, respectively.
The ability to approximate one distribution by the other, when computation
tools for both distributions are handy, seems to be of questionable importance.
Indeed, the significance of the Normal approximation is not so much in its
ability to approximate the Binomial distribution as such. Rather, the important
point is that the Normal distribution may serve as an approximation to a wide
class of distributions, with the Binomial distribution being only one example.
Computations that are based on the Normal approximation will be valid for
all members in the class of distributions, including cases where we don’t have
the computational tools at our disposal or even in cases where we do not know
what the exact distribution of the member is! As promised, a more detailed
discussion of the Normal approximation in a wider context will be presented in
the next chapter.
On the other hand, one need not assume that any distribution is well approx-
imated by the Normal distribution. For example, the distribution of wealth in
the population tends to be skewed, with more than 50% of the people possessing
less than 50% of the wealth and small percentage of the people possessing the
majority of the wealth. The Normal distribution is not a good model for such
distribution. The Exponential distribution, or distributions similar to it, may
be more appropriate.

6.3.2 Continuity Corrections

In order to complete this section let us look more carefully at the Normal ap-
proximations of the Binomial distribution.
In principle, the Normal approximation is valid when n, the number of inde-
pendent trials in the Binomial distribution, is large. When n is relatively small
98 CHAPTER 6. THE NORMAL RANDOM VARIABLE

0.15
0.10
Pr

0.05
0.00

0 5 10 15 20 25 30

Figure 6.6: Normal Approximation of the Binomial Distribution

the approximation may not be so good. Indeed, take X ∼ Binomial(30, 0.3)

and consider the probability P(X ≤ 6). Compare the actual probability to the
Normal approximation:
> pbinom(6,30,0.3)
[1] 0.1595230
> pnorm(6,30*0.3,sqrt(30*0.3*0.7))
[1] 0.1159989
The Normal approximation, which is equal to 0.1159989, is not too close to the
actual probability, which is equal to 0.1595230.
A naı̈ve application of the Normal approximation for the Binomial(n, p) dis-
tribution may not be so good when the number of trials n is small. Yet, a
small modification of the approximation may produce much better results. In
order to explain the modification consult Figure 6.6 where you will find the bar
plot of the Binomial distribution with the density of the approximating Nor-
mal distribution superimposed on top of it. The target probability is the sum
of heights of the bars that are painted in red. In the naı̈ve application of the
Normal approximation we used the area under the normal density which is to
6.3. APPROXIMATION OF THE BINOMIAL DISTRIBUTION 99

the left of the bar associated with the value x = 6.

Alternatively, you may associate with each bar located at x the area under
the normal density over the interval [x − 0.5, x + 0.5]. The resulting correction
to the approximation will use the Normal probability of the event {X ≤ 6.5},
which is the area shaded in red. The application of this approximation, which
is called continuity correction produces:
> pnorm(6.5,30*0.3,sqrt(30*0.3*0.7))
[1] 0.1596193
Observe that the corrected approximation is much closer to the target prob-
ability, which is 0.1595230, and is substantially better that the uncorrected
approximation which was 0.1159989. Generally, it is recommended to apply the
continuity correction to the Normal approximation of a discrete distribution.
Consider the Binomial(n, p) distribution. Another situation where the Nor-
mal approximation may fail is when p, the probability of “Success” in the Bino-
mial distribution, is too close to 0 (or too close to 1). Recall, that for large n the
Poisson distribution emerged as an approximation of the Binomial distribution
in such a setting. One may expect that when n is large and p is small then the
Poisson distribution may produce a better approximation of a Binomial proba-
bility. When the Poisson distribution is used for the approximation we call it a
Poisson Approximation.
Let us consider an example. Let us analyze 3 Binomial distributions. The
expectation in all the distributions is equal to 2 but the number of trials, n,
vary. In the first case n = 20 (and hence p = 0.1), in the second n = 200 (and
p = 0.01), and in the third n = 2, 000 (and p = 0.001. In all three cases we will
be interested in the probability of obtaining a value less or equal to 3.
The Poisson approximation replaces computations conducted under the Bi-
nomial distribution with Poisson computations, with a Poisson distribution that
has the same expectation as the Binomial. Since in all three cases the expec-
tation is equal to 2 we get that the same Poisson approximation is used to the
three probabilities:
> ppois(3,2)
[1] 0.8571235
The actual Binomial probability in the first case (n = 20, p = 0.1) and a
Normal approximation thereof are:
> pbinom(3,20,0.1)
[1] 0.8670467
> pnorm(3.5,2,sqrt(20*0.1*0.9))
[1] 0.8682238
Observe that the Normal approximation (with a continuity correction) is better
than the Poisson approximation in this case.
In the second case (n = 200, p = 0.01) the actual Binomial probability and
the Normal approximation of the probability are:
> pbinom(3,200,0.01)
[1] 0.858034
> pnorm(3.5,2,sqrt(200*0.01*0.99))
[1] 0.856789
100 CHAPTER 6. THE NORMAL RANDOM VARIABLE

Observe that the Poisson approximation that produces 0.8571235 is slightly

closer to the target than the Normal approximation. The greater accuracy of
the Poisson approximation for the case where n is large and p is small is more
pronounced in the final case (n = 2000, p = 0.001) where the target probability
and the Normal approximation are:
> pbinom(3,2000,0.001)
[1] 0.8572138
> pnorm(3.5,2,sqrt(2000*0.001*0.999))
[1] 0.8556984
Compare the actual Binomial probability, which is equal to 0.8572138, to the
Poisson approximation that produced 0.8571235. The Normal approximation,
0.8556984, is slightly off, but is still acceptable.

6.4 Solved Exercises

Question 6.1. Consider the problem of establishing regulations concerning the
maximum number of people who can occupy a lift. In particular, we would like
to assess the probability of exceeding maximal weight when 8 people are allowed
to use the lift simultaneously and compare that to the probability of allowing 9
people into the lift.
Assume that the total weight of 8 people chosen at random follows a normal
distribution with a mean of 560kg and a standard deviation of 57kg. Assume
that the total weight of 9 people chosen at random follows a normal distribution
with a mean of 630kg and a standard deviation of 61kg.
1. What is the probability that the total weight of 8 people exceeds 650kg?
2. What is the probability that the total weight of 9 people exceeds 650kg?
3. What is the central region that contains 80% of distribution of the total
weight of 8 people?
4. What is the central region that contains 80% of distribution of the total
weight of 9 people?
Solution (to Question 6.1.1): Let X be the total weight of 8 people. By
the assumption, X ∼ Normal(560, 572 ). We are interested in the probability
P(X > 650). This probability is equal to the difference between 1 and the
probability P(X ≤ 650). We use the function “pnorm” in order to carry out the
computation:
> 1 - pnorm(650,560,57)
[1] 0.05717406
We get that the probability that the total weight of 8 people exceeds 650kg is
equal to 0.05717406.

Solution (to Question 6.1.2): Let Y be the total weight of 9 people. By

the assumption, Y ∼ Normal(630, 612 ). We are interested in the probability
P(Y > 650). This probability is equal to the difference between 1 and the
probability P(Y ≤ 650). We use again the function “pnorm” in order to carry
out the computation:
6.4. SOLVED EXERCISES 101

> 1 - pnorm(650,630,61)
[1] 0.3715054
We get that the probability that the total weight of 9 people exceeds 650kg is
much higher and is equal to 0.3715054.

Solution (to Question 6.1.3): Again, X ∼ Normal(560, 572 ), where X is the

total weight of 8 people. In order to find the central region that contains 80% of
the distribution we need to identify the 10%-percentile and the 90%-percentile
of X. We use the function “qnorm” in the code:
> qnorm(0.1,560,57)
[1] 486.9516
> qnorm(0.9,560,57)
[1] 633.0484
The requested region is the interval [486.9516, 633.0484].

Solution (to Question 6.1.4): As before, Y ∼ Normal(630, 612 ), where Y is

the total weight of 9 people. In order to find the central region that contains
80% of the distribution we need to identify the 10%-percentile and the 90%-
percentile of Y . The computation this time produces:
> qnorm(0.1,630,61)
[1] 551.8254
> qnorm(0.9,630,61)
[1] 708.1746
and the region is [551.8254, 708.1746].

Question 6.2. Assume X ∼ Binomial(27, 0.32). We are interested in the prob-

ability P(X > 11).

1. Compute the (exact) value of this probability.

2. Compute a Normal approximation to this probability, without a continuity
correction.
3. Compute a Normal approximation to this probability, with a continuity
correction.
4. Compute a Poisson approximation to this probability.

Solution (to Question 6.2.1): The probability P(X > 11) can be computed
as the difference between 1 and the probability P(X ≤ 11). The latter proba-
bility can be computed with the function “pbinom”:
> 1 - pbinom(11,27,0.32)
[1] 0.1203926
Therefore, P(X > 11) = 0.1203926.

Solution (to Question 6.2.2): Refer again to the probability P(X > 11).
A formal application of the Normal approximation replaces in the computation
102 CHAPTER 6. THE NORMAL RANDOM VARIABLE

the Binomial distribution by the Normal distribution with the same mean and
variance. Since E(X) = n · p = 27 · 0.32 = 8.64 and Var(X) = n · p · (1 − p) =
27 · 0.32 · 0.68 = 5.8752. If we take X ∼ Normal(8.64, 5.8752) and use the
function “pnorm” we get:
> 1 - pnorm(11,27*0.32,sqrt(27*0.32*0.68))
[1] 0.1651164
Therefore, the current Normal approximation proposes P(X > 11) ≈ 0.1651164.

Solution (to Question 6.2.3): The continuity correction, that consider inter-
val of radius 0.5 about each value, replace P(X > 11), that involves the values
{12, 13, . . . , 27}, by the event P(X > 11.5). The Normal approximation uses the
Normal distribution with the same mean and variance. Since E(X) = 8.64 and
Var(X) = 5.8752. If we take X ∼ Normal(8.64, 5.8752) and use the function
“pnorm” we get:
> 1 - pnorm(11.5,27*0.32,sqrt(27*0.32*0.68))
[1] 0.1190149
The Normal approximation with continuity correction proposes P(X > 11) ≈
0.1190149.

Solution (to Question 6.2.4): The Poisson approximation replaces the Bi-
nomial distribution by the Poisson distribution with the same expectation. The
expectation is E(X) = n · p = 27 · 0.32 = 8.64. If we take X ∼ Poisson(8.64)
and use the function “ppois” we get:
> 1 - ppois(11,27*0.32)
[1] 0.1635232
Therefore, the Poisson approximation proposes P(X > 11) ≈ 0.1651164.

6.5 Summary
Glossary
Normal Random Variable: A bell-shaped distribution that is frequently used
to model a measurement. The distribution is marked with Normal(µ, σ 2 ).
Standard Normal Distribution: The Normal(0, 1). The distribution of stan-
dardized Normal measurement.
Percentile: Given a percent p · 100% (or a probability p), the value x is the
percentile of a random variable X if it satisfies the equation P(X ≤ x) = p.
Normal Approximation of the Binomial: Approximate computations as-
sociated with the Binomial distribution with parallel computations that
use the Normal distribution with the same expectation and standard de-
viation as the Binomial.
Poisson Approximation of the Binomial: Approximate computations as-
sociated with the Binomial distribution with parallel computations that
use the Poisson distribution with the same expectation as the Binomial.
6.5. SUMMARY 103

Discuss in the Forum

Mathematical models are used as tools to describe reality. These models are
supposed to characterize the important features of the analyzed phenomena and
provide insight. Random variables are mathematical models of measurements.
Some people claim that there should be a perfect match between the mathemat-
ical characteristics of a random variable and the properties of the measurement
it models. Other claim that a partial match is sufficient. What is your opinion?
When forming your answer to this question you may give an example of a
situation from you own field of interest for which a random variable can serve
as a model. Identify discrepancies between the theoretical model and actual
properties of the measurement. Discuss the appropriateness of using the model
in light of these discrepancies.
Consider, for example, testing IQ. The score of many IQ tests are modeled
as having a Normal distribution with an expectation of 100 and a standard
deviation of 15. The sample space of the Normal distribution is the entire line
of real numbers, including the negative numbers. In reality, IQ tests produce
only positive values.
104 CHAPTER 6. THE NORMAL RANDOM VARIABLE
Chapter 7

The Sampling Distribution

7.1 Student Learning Objective

In this section we integrate the concept of data that is extracted from a sample
with the concept of a random variable. The new element that connects between
these two concepts is the notion of sampling distribution. The data we observe
results from the specific sample that was selected. The sampling distribution,
in a similar way to random variables, corresponds to all samples that could have
been selected. (Or, stated in a different tense, to the sample that will be selected
prior to the selection itself.) Summaries of the distribution of the data, such as
the sample mean and the sample standard deviation, become random variables
when considered in the context of the sampling distribution. In this section we
investigate the sampling distribution of such data summaries. In particular, it
is demonstrated that (for large samples) the sampling distribution of the sample
average may be approximated by the Normal distribution. The mathematical
theorem that proves this approximation is called the Central Limit Theory. By
the end of this chapter, the student should be able to:

• Comprehend the notion of sampling distribution and simulate the sam-

pling distribution of the sample average.

• Relate the expectation and standard deviation of a measurement to the

expectation and standard deviation of the sample average.

• Apply the Central Limit Theorem to the sample averages.

7.2 The Sampling Distribution

In Chapter 5 the concept of a random variable was introduced. As part of
the introduction we used an example that involved the selection of a random
person from the population and the measuring of his/her height. Prior to the
action of selection, the height of that person is a random variable. It has the
potential of obtaining any of the heights that are present in the population,
which is the sample space of this example, with a distribution that reflects the
relative frequencies of each of the heights in the population: the probabilities of
the values. After the selection of the person and the measuring of the height

105
106 CHAPTER 7. THE SAMPLING DISTRIBUTION

we get a particular value. This is the observed value and is no longer a random
variable. In this section we extend the concept of a random variable and define
the concept of a random sample.

7.2.1 A Random Sample

The relation between the random sample and the data is similar to the relation
between a random variable and the observed value. The data is the observed
values of a sample taken from a population. The content of the data is known.
The random sample, similarly to a random variable, is the data that will be
selected when taking a sample, prior to the selection itself. The content of the
random sample is unknown, since the sample has not yet been taken. Still, just
like for the case of the random variable, one is able to say what the possible
evaluations of the sample may be and, depending on the mechanism of selecting
the sample, what are the probabilities of the different potential evaluations.
The collection of all possible evaluations of the sample is the sample space of
the random sample and the probabilities of the different evaluations produce
the distribution of the random sample.
(Alternatively, if one prefers to speak in past tense, one can define the sample
space of a random sample to be the evaluations of the sample that could have
taken place, with the distribution of the random sample being the probabilities
of these evaluations.)
A statistic is a function of the data. Example of statistics are the average of
the data, the sample variance and standard deviation, the median of the data,
etc. In each case a given formula is applied to the data. In each type of statistic
a different formula is applied.
The same formula that is applied to the observed data may, in principle, be
applied to random samples. Hence, for example, one may talk of the sample
average, which is the average of the elements in the data. The average, consid-
ered in the context of the observed data, is a number and its value is known.
However, if we think of the average in the context of a random sample then it
becomes a random variable. Prior to the selection of the actual sample we do
not know what values it will include. Hence, we cannot tell what the outcome
of the average of the values will be. However, due to the identification of all
possible evaluations that the sample can possess we may say in advance what is
the collection of values the sample average can have. This is the sample space
of the sample average. Moreover, from the sampling distribution of the random
sample one may identify the probability of each value of the sample average,
thus obtaining the sampling distribution of the sample average.
The same line of argumentation applies to any statistic. Computed in the
context of the observed data, the statistic is a known number that may, for
example, be used to characterize the variation in the data. When thinking of a
statistic in the context of a random sample it becomes a random variable. The
distribution of the statistic is called the sampling distribution of the statistic.
Consequently, we may talk of the sampling distribution of the median, the
sample distribution of the sample variance, etc.
Random variables are also applied as models for uncertainty in future mea-
surements in more abstract settings that need not involve a specific population.
Specifically, we introduced the Binomial and Poisson random variables for set-
tings that involve counting and the Uniform, Exponential, and Normal random
7.2. THE SAMPLING DISTRIBUTION 107

variables for settings where the measurement is continuous.

The notion of a sampling distribution may be extended to a situation where
one is taking several measurements, each measurement taken independently of
the others. As a result one obtains a sequence of measurements. We use the
term “sample” to denote this sequence. The distribution of this sequence is
also called the sampling distribution. If all the measurements in the sequence
are Binomial then we call it a Binomial sample. If all the measurements are
Exponential we call it an Exponential sample and so forth.
Again, one may apply a formula (such as the average) to the content of the
random sequence and produce a random variable. The term sampling distribu-
tion describes again the distribution that the random variable produced by the
formula inherits from the sample.
In the next subsection we examine an example of a sample taken from a
population. Subsequently, we discuss examples that involves a sequence of mea-
surements from a theoretical model.

7.2.2 Sampling From a Population

Consider taking a sample from a population. Let us use again for the illustration
the file “pop1.csv” like we did in Chapter 4. The data frame produced from
the file contains the sex and hight of the 100,000 members of some imaginary
population. Recall that in Chapter 4 we applied the function “sample” to
randomly sample the height of a single subject from the population. Let us
apply the same function again, but this time in order to sample the heights of
100 subjects:
> pop.1 <- read.csv("pop1.csv")
> X.samp <- sample(pop.1$height,100)
> X.samp
[1] 168 177 172 174 154 179 145 160 188 172 175 174 176 144 164
[16] 171 167 158 181 165 166 173 184 174 169 176 168 154 167 175
[31] 178 179 175 187 160 171 175 172 178 167 181 193 163 181 168
[46] 153 200 168 169 194 177 182 167 183 177 155 167 172 176 168
[61] 164 162 188 163 166 156 163 185 149 163 157 155 161 177 176
[76] 153 162 180 177 156 162 197 183 166 185 178 188 198 175 167
[91] 185 160 148 160 174 162 161 178 159 168
In the first line of code we produce a data frame that contains the information
on the entire population. In the second line we select a sample of size 100 from
the population, and in the third line we present the content of the sample.
The first argument to the function “sample” that selects the sample is the
sequence of length 100,000 with the list of heights of all the members of the
population. The second argument indicates the sample size, 100 in this case.
The outcome of the random selection is stored in the object “X.samp”, which is
a sequence that contains 100 heights.
Typically, a researcher does not get to examine the entire population. In-
stead, measurements on a sample from the population are made. In relation to
the imaginary setting we simulate in the example, the typical situation is that
the research does not have the complete list of potential measurement evalu-
ations, i.e. the complete list of 100,000 heights in “pop.1$height”, but only
a sample of measurements, namely the list of 100 numbers that are stored in
108 CHAPTER 7. THE SAMPLING DISTRIBUTION

“X.samp” and are presented above. The role of statistics is to make inference
on the parameters of the unobserved population based on the information that
is obtained from the sample.
For example, we may be interested in estimating the mean value of the
heights in the population. A reasonable proposal is to use the sample average
to serve as an estimate:
> mean(X.samp)
[1] 170.73
In our artificial example we can actually compute the true population mean:
> mean(pop.1$height)
[1] 170.035
Hence, we may see that although the match between the estimated value and
the actual value is not perfect still they are close enough.
The actual estimate that we have obtained resulted from the specific sample
that was collected. Had we collected a different subset of 100 individuals we
would have obtained different numerical value for the estimate. Consequently,
one may wonder: Was it pure luck that we got such good estimates? How likely
is it to get estimates that are close to the target parameter?
Notice that in realistic settings we do not know the actual value of the
target population parameters. Nonetheless, we would still want to have at
least a probabilistic assessment of the distance between our estimates and the
parameters they try to estimate. The sampling distribution is the vehicle that
may enable us to address these questions.
In order to illustrate the concept of the sampling distribution let us select
another sample and compute its average:
> X.samp <- sample(pop.1$height,100)
> X.bar <- mean(X.samp)
> X.bar
[1] 171.87
and do it once more:
> X.samp <- sample(pop.1$height,100)
> X.bar <- mean(X.samp)
> X.bar
[1] 171.02
In each case we got a different value for the sample average. In the first of
the last two iterations the result was more than 1 centimeter away from the
population average, which is equal to 170.035, and in the second it was within
the range of 1 centimeter. Can we say, prior to taking the sample, what is the
probability of falling within 1 centimeter of the population mean?
Chapter 4 discussed the random variable that emerges by randomly sampling
a single number from the population presented by the sequence “pop.1$height”.
The distribution of the random variable resulted from the assignment of the
probability 1/100,000 to each one of the 100,000 possible outcomes. The same
principle applies when we randomly sample 100 individuals. Each possible out-
come is a collection of 100 numbers and each collection is assigned equal prob-
ability. The resulting distribution is called the sampling distribution.
7.2. THE SAMPLING DISTRIBUTION 109

The distribution of the average of the sample emerges from this distribution:
With each sample one may associate the average of that sample. The probability
assigned to that average outcome is the probability of the sample. Hence, one
may assess the probability of falling within 1 centimeter of the population mean
using the sampling distribution. Each sample produces an average that either
falls within the given range or not. The probability of the sample average falling
within the given range is the proportion of samples for which this event happens
among the entire collection of samples.
However, we face a technical difficulty when we attempt to assess the sam-
pling distribution of the average and the probability of falling within 1 cen-
timeter of the population mean. Examination of the distribution of a sample
of a single individual is easy enough. The total number of outcomes, which is
100,000 in the given example, can be handled with no effort by the computer.
However, when we consider samples of size 100 we get that the total number
of ways to select 100 number out of 100,000 numbers is in the order of 10342
(1 followed by 342 zeros) and cannot be handled by any computer. Thus, the
probability cannot be computed.
As a compromise we will approximate the distribution by selecting a large
number of samples, say 100,000, to represent the entire collection, and use
the resulting distribution as an approximation of the sampling distribution.
Indeed, the larger the number of samples that we create the more accurate
the approximation of the distribution is. Still, taking 100,000 repeats should
produce approximations which are good enough for our purposes.
Consider the sampling distribution of the sample average. We simulated
above a few examples of the average. Now we would like to simulate 100,000
such examples. We do this by creating first a sequence of the length of the
number of evaluations we seek (100,000) and then write a small program that
produces each time a new random sample of size 100 and assigns the value of
the average of that sample to the appropriate position in the sequence. Do first
and explain later1 :

> X.bar <- rep(0,10^5)

> for(i in 1:10^5)
+ {
+ X.samp <- sample(pop.1$height,100)
+ X.bar[i] <- mean(X.samp)
+ }
> hist(X.bar)

In the first line we produce a sequence of length 100,000 that contains zeros.
The function “rep” creates a sequence that contains repeats of its first argument
a number of times that is specified by its second argument. In this example,
the numerical value 0 is repeated 100,000 times to produce a sequence of zeros
of the length we seek.
1 Running this simulation, and similar simulations of the same nature that will be consid-

ered in the sequel, demands more of the computer’s resources than the examples that were
considered up until now. Beware that running times may be long and, depending on the
strength of your computer and your patience, too long. You may save time by running less it-
erations, replacing, say, “10^5” by “10^4”. The results of the simulation will be less accurate,
but will still be meaningful.
110 CHAPTER 7. THE SAMPLING DISTRIBUTION

3500
Frequency

1500
0

117 127 137 147 157 167 177 187 197 207 217

height

Histogram of X.bar
Frequency

10000
0

166 168 170 172 174

X.bar

Figure 7.1: Distribution of Height and the Sampling Distribution of Averages

The main part of the program is a “for” loop. The argument of the function
“for” takes the special form: “index.name in index.values”, where index.name
is the name of the running index and index.values is the collection of values over
which the running index is evaluated. In each iteration of the loop the running
index is assigned a value from the collection and the expression that follows the
brackets of the “for” function is evaluated with the given value of the running
index.
In the given example the collection of values is produced by the expression
“1:n”. Recall that the expression “1:n” produces the collection of integers be-
tween 1 and n. Here, n = 100,000. Hence, in the given application the collection
of values is a sequence that contains the integers between 1 and 100,000. The
running index is called “i”. the expression is evaluated 100,000 times, each time
with a different integer value for the running index “i”.
The R system treats a collection of expressions enclosed within curly brackets
as one entity. Therefore, in each iteration of the “for” loop, the lines that are
within the curly brackets are evaluated. In the first line a random sample of size
100 is produced and in the second line the average of the sample is computed and
stored in the i-th position of the sequence “X.bar”. Observe that the specific
7.2. THE SAMPLING DISTRIBUTION 111

position in the sequence is referred to by using square brackets.

The program changes the original components of the sequence, from 0 to
the average of a random sample, one by one. When the loop ends all values are
changed and the sequence “X.bar” contains 100,000 evaluations of the sample
average. The last line, which is outside the curly brackets and is evaluated after
the “for” loop ends, produces an histogram of the averages that were simulated.
The histogram is presented in the lower panel of Figure 7.1.
Compare the distribution of the sample average to the distribution of the
heights in the population that was presented first in Figure 4.1 and is currently
presented in the upper panel of Figure 7.1. Observe that both distributions are
centered at about 170 centimeters. Notice, however, that the range of values
of the sample average lies essentially between 166 and 174 centimeters, whereas
the range of the distribution of heights themselves is between 127 and 217 cen-
timeter. Broadly speaking, the sample average and the original measurement
are centered around the same location but the sample average is less spread.
Specifically, let us compare the expectation and standard deviation of the
sample average to the expectation and standard deviation of the original mea-
surement:

> mean(pop.1$height)
[1] 170.035
> sd(pop.1$height)
[1] 11.23205
> mean(X.bar)
[1] 170.037
> sd(X.bar)
[1] 1.122116

Observe that the expectation of the population and the expectation of the sam-
ple average, are practically the same, the standard deviation of the sample
average is about 10 times smaller than the standard deviation of the popula-
tion. This result is not accidental and actually reflects a general phenomena
that will be seen below in other examples.
We may use the simulated sampling distribution in order to compute an ap-
proximation of the probability of the sample average falling within 1 centimeter
of the population mean. Let us first compute the relevant probability and then
explain the details of the computation:

> mean(abs(X.bar - mean(pop.1$height)) <= 1)

[1] 0.62589

Hence we get that the probability of the given event is about 62.6%.
The object “X.bar” is a sequence of length 100,000 that contains the sim-
ulated sample averages. This sequence represents the distribution of the sam-
ple average. The expression “abs(X.bar - mean(pop.1$height)) <= 1” pro-
duces a sequence of logical “TRUE” or “FALSE” values, depending on the value
of the sample average being less or more than one unit away from the popu-
lation mean. The application of the function “mean” to the output of the last
expression results in the computation of the relative frequency of TRUEs, which
corresponds to the probability of the event of interest.
112 CHAPTER 7. THE SAMPLING DISTRIBUTION

Example 7.1. A poll for the determination of the support in the population
for a candidate was describe in Example 5.1. The proportion in the population
of supporters was denoted by p. A sample of size n = 300 was considered in
order to estimate the size of p. We identified that the distribution of X, the
number of supporters in the sample, is Binomial(300, p). This distribution is
the sampling distribution2 of X. One may use the proportion in the sample
of supporters, the number of supporters in the sample divided by 300, as an
estimate to the parameter p. The sampling distribution of this quantity, X/300,
may be considered in order to assess the discrepancy between the estimate and
the actual value of the parameter.

7.2.3 Theoretical Models

Sampling distribution can also be considered in the context of theoretical distri-
bution models. For example, take a measurement X ∼ Binomial(10, 0.5) from
the Binomial distribution. Assume 64 independent measurements are produced
with this distribution: X1 , X2 , . . . , X64 . The sample average in this case corre-
sponds to the distribution of the random variable produced by averaging these
64 random variables:
64
X1 + X2 + · · · + X64 1 X
X̄ = = Xi .
64 64 i=1

Again, one may wonder what is the distribution of the sample average X̄ in this
case?
We can approximate the distribution of the sample average by simulation.
The function “rbinom” produces a random sample from the Binomial distribu-
tion. The first argument to the function is the sample size, which we take in this
example to be equal to 64. The second and third arguments are the parameters
of the Binomial distribution, 10 and 0.5 in this case. We can use this function
in the simulation:

> X.bar <- rep(0,10^5)

> for(i in 1:10^5)
+ {
+ X.samp <- rbinom(64,10,0.5)
+ X.bar[i] <- mean(X.samp)
+ }

Observe that in this code we created a sequence of length 100,000 with evalua-
tions of the sample average of 64 Binomial random variables. We start with a
sequence of zeros and in each iteration of the “for” loop a zero is replaced by
the average of a random sample of 64 Binomial random variables.
Examine the sampling distribution of the Binomial average:

> hist(X.bar)
2 Mathematically speaking, the Binomial distribution is only an approximation to the sam-
pling distribution of X. Actually, the Binomial is an exact description to the distribution only
in the case where each subject has the chance be represented in the sample more than once.
However, only when the size of the sample is comparable to the size of the population would
the Binomial distribution fail to be an adequate approximation to the sampling distribution.
7.2. THE SAMPLING DISTRIBUTION
Probability 113

0.15
0.00

0 2 4 6 8 10

Histogram of X.bar
Frequency

10000
0

4.5 5.0 5.5

X.bar

Figure 7.2: Distributions of an Average and a Single Binomial(10,0.5)

> mean(X.bar)
[1] 4.999074
> sd(X.bar)
[1] 0.1982219

The histogram of the sample average is presented in the lower panel of Figure 7.2.
Compare it to the distribution of a single Binomial random variable that appears
in the upper panel. Notice, once more, that the center of the two distributions
coincide but the spread of the sample average is smaller. The sample space of
a single Binomial random variable is composed of integers. The sample space
of the average of 64 Binomial random variables, on the other hand, contains
many more values and is closer to the sample space of a random variable with
a continuous distribution.
Recall that the expectation of a Binomial(10, 0.5) random variable is E(X) =
10 · 0.5 = 5 and
√ the variance is Var(X) = 10 · 0.5 · 0.5 = 2.5 (thus, the standard
deviation is 2.5 = 1.581139). Observe that the expectation of the sample
average that we got from the simulation is essentially equal to 5 and the standard
deviation is 0.1982219.
114 CHAPTER 7. THE SAMPLING DISTRIBUTION

One may prove mathematically that the expectation of the sample mean is
equal to the theoretical expectation of its components:

E(X̄) = E(X) .

The results of the simulation for the expectation of the sample average are
consistent with the mathematical statement. The mathematical theory of prob-
ability may also be used in order to prove that the variance of the sample average
is equal to the variance of each of the components, divided by the sample size:

Var(X̄) = Var(X)/n ,

here n is the number of observations in the sample. Specifically, in the Bino-

mial example we get that Var(X̄) = 2.5/64, since the variance of a Binomial
component isp2.5 and there are 64 observations. Consequently, the standard
deviation is 2.5/64 = 0.1976424, in agreement, more or less, with the results
of the simulation (that produced 0.1982219 as the standard deviation).
Consider the problem of identifying the central interval that contains 95%
of the distribution. In the Normal distribution we were able to use the function
“qnorm” in order to compute the percentiles of the theoretical distribution. A
function that can be used for the same purpose for simulated distribution is
the function “quantile”. The first argument to this function is the sequence
of simulated values of the statistic, “X.bar” in the current case. The second
argument is a number between 0 and 1, or a sequence of such numbers:

> quantile(X.bar,c(0.025,0.975))
2.5% 97.5%
4.609375 5.390625

We used the sequence “c(0.025,0.975)” as the input to the second argument.

As a result we obtained the output 4.609375, which is the 2.5%-percentile of the
sampling distribution of the average, and 5.390625, which is the 97.5%-percentile
of the sampling distribution of the average.
Of interest is to compare these percentiles to the parallel percentiles of the
Normal distribution with the same expectation and the same standard deviation
as the average of the Binomials:

> qnorm(c(0.025,0.975),mean(X.bar),sd(X.bar))
[1] 4.611456 5.389266

Observe the similarity between the percentiles of the distribution of the average
and the percentiles of the Normal distribution. This similarity is a reflection of
the Normal approximation of the sampling distribution of the average, which is
formulated in the next section under the title: The Central Limit Theorem.

Example 7.2. The distribution of the number of events of radio active decay
in a second was modeled in Example 5.3 according to the Poisson distribution.
A quantity of interest is λ, the expectation of that Poisson distribution. This
quantity may be estimated by measuring the total number of decays over a period
of time and dividing the outcome by the number of seconds in that period of
time. Let n be this number of second. The procedure just described corresponds
to taking the sample average of Poisson(λ) observations for a sample of size n.
7.3. LAW OF LARGE NUMBERS AND CENTRAL LIMIT THEOREM 115

The expectation of the sample p average is λ and the variance is λ/n, leading to
a standard deviation of size λ/n. The Central Limit Theorem states that the
sampling distribution of this average corresponds, approximately, to the Normal
distribution with this expectation and standard deviation.

7.3 Law of Large Numbers and Central Limit

Theorem
The Law of Large Numbers and the Central Limit Theorem are mathemati-
cal theorems that describe the sampling distribution of the average for large
samples.

7.3.1 The Law of Large Numbers

The Law of Large Numbers states that, as the sample size becomes larger, the
sampling distribution of the sample average becomes more and more concen-
trated about the expectation.
Let us demonstrate the Law of Large Numbers in the context of the Uni-
form distribution. Let the distribution of the measurement X be Uniform(3, 7).
Consider three different sample sizes n: n = 10, n = 100, and n = 1000. Let
us carry out a simulation similar to the simulations of the previous section.
However, this time we run the simulation for the three sample sizes in parallel:

> unif.10 <- rep(0,10^5)

> unif.100 <- rep(0,10^5)
> unif.1000 <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X.samp.10 <- runif(10,3,7)
+ unif.10[i] <- mean(X.samp.10)
+ X.samp.100 <- runif(100,3,7)
+ unif.100[i] <- mean(X.samp.100)
+ X.samp.1000 <- runif(1000,3,7)
+ unif.1000[i] <- mean(X.samp.1000)
+ }

Observe that we have produced 3 sequences of length 100,000 each: “unif.10”,

“unif.100”, and “unif.1000”. The first sequence is an approximation of the
sampling distribution of an average of 10 independent Uniform measurements,
the second approximates the sampling distribution of an average of 100 mea-
surements and the third the distribution of an average of 1000 measurements.
The distribution of single measurement in each of the examples is Uniform(3, 7).
Consider the expectation of sample average for the three sample sizes:

> mean(unif.10)
[1] 4.999512
> mean(unif.100)
[1] 4.999892
> mean(unif.1000)
[1] 4.99996
116 CHAPTER 7. THE SAMPLING DISTRIBUTION

For all sample size the expectation of the sample average is equal to 5, which is
the expectation of the Uniform(3, 7) distribution.
Recall that the variance of the Uniform(a, b) distribution is (b − a)2 /12.
Hence, the variance of the given Uniform distribution is Var(X) = (7−3)2 /12 =
16/12 ≈ 1.3333. The variances of the sample averages are:
> var(unif.10)
[1] 0.1331749
> var(unif.100)
[1] 0.01333089
> var(unif.1000)
[1] 0.001331985
Notice that the variances decrease with the increase of the sample sizes. The
decrease is according to the formula Var(X̄) = Var(X)/n.
The variance is a measure of the spread of the distribution about the expecta-
tion. The smaller the variance the more concentrated is the distribution around
the expectation. Consequently, in agreement with the Law of Large Numbers,
the larger the sample size the more concentrated is the sampling distribution of
the sample average about the expectation.

7.3.2 The Central Limit Theorem (CLT)

The Law of Large Numbers states that the distribution of the sample average
tends to be more concentrated as the sample size increases. The Central Limit
Theorem (CLT in short) provides an approximation of this distribution.
The deviation between the sample average and the expectation of the mea-
surement tend to decreases with the increase in sample size. In order to obtain
a refined assessment of this deviation one needs to magnify it. The appropriate
way to obtain the magnification is to consider the standardized sample average,
in which the deviation of the sample average from its expectation is divided by
the standard deviation of the sample average:
X̄ − E(X̄)
Z= p .
Var(X̄)
Recall that the expectation of the sample average is equal to the expectation
of a single random variable (E(X̄) = E(X)) and that the variance of the sample
average is equal to the variance of a single observation, divided by the sample
size (Var(X̄) = Var(X)/n). Consequently, one may rewrite the standardized
sample average in the form:
√
X̄ − E(X) n(X̄ − E(X))
Z=p = p .
Var(X)/n Var(X)
The second equality follows from placing in the numerator the square root of n
which divides the term in the denominator. Observe that with the increase of the
sample size the decreasing difference between the average and the expectation
is magnified by the square root of n.
The Central Limit Theorem states that, with the increase in sample size,
the sample average converges (after standardization) to the standard Normal
distribution.
7.3. LAW OF LARGE NUMBERS AND CENTRAL LIMIT THEOREM 117

0.4

N(0,1)
n=10
n=100
n=1000
0.3
Density

0.2
0.1
0.0

−3 −2 −1 0 1 2 3

Figure 7.3: The CLT for the Uniform(3,7) Distribution

Let us examine the Central Normal Theorem in the context of the example
of the Uniform measurement. In Figure 7.3 you may find the (approximated)
density of the standardized average for the three sample sizes based on the
simulation that we carried out previously (as red, green, and blue lines). Along
side with these densities you may also find the theoretical density of the standard
Normal distribution (as a black line). Observe that the four curves are almost
one on top of the other, proposing that the approximation of the distribution of
the average by the Normal distribution is good even for a sample size as small
as n = 10.
However, before jumping to the conclusion that the Central Limit Theorem
applies to any sample size, let us consider another example. In this example we
repeat the same simulation that we did with the Uniform distribution, but this
time we take Exponential(0.5) measurements instead:

> exp.10 <- rep(0,10^5)

> exp.100 <- rep(0,10^5)
> exp.1000 <- rep(0,10^5)
> for(i in 1:10^5)
118 CHAPTER 7. THE SAMPLING DISTRIBUTION

0.4
N(0,1)
n=10
n=100
n=1000
0.3
Density

0.2
0.1
0.0

−3 −2 −1 0 1 2 3

Figure 7.4: The CLT for the Exponential(0.5) Distribution

+ {
+ X.samp.10 <- rexp(10,0.5)
+ exp.10[i] <- mean(X.samp.10)
+ X.samp.100 <- rexp(100,0.5)
+ exp.100[i] <- mean(X.samp.100)
+ X.samp.1000 <- rexp(1000,0.5)
+ exp.1000[i] <- mean(X.samp.1000)
+ }
The expectation of an Exponential(0.5) random variable is E(X) = 1/λ =
1/0.5 = 2 and the variance is Var(X) = 1/λ2 = 1/(0.5)2 = 4. Observe below
that the expectations of the sample averages are equal to the expectation of
the measurement and the variances of the sample averages follow the relation
Var(X̄) = Var(X)/n:
> mean(exp.10)
[1] 1.999888
> mean(exp.100)
[1] 2.000195
7.3. LAW OF LARGE NUMBERS AND CENTRAL LIMIT THEOREM 119

> mean(exp.1000)
[1] 1.999968

So the expectations of the sample average are all equal to 2. For the variance
we get:

> var(exp.10)
[1] 0.4034642
> var(exp.100)
[1] 0.03999479
> var(exp.1000)
[1] 0.004002908

Which is in agreement with the decrease proposed by the theory,

However, when one examines the densities of the sample averages in Fig-
ure 7.4 one may see a clear distinction between the sampling distribution of the
average for a sample of size 10 and the normal distribution (compare the red
curve to the black curve. The match between the green curve that corresponds
to a sample of size n = 100 and the black line is better, but not perfect. When
the sample size is as large as n = 1000 (the blue curve) then the agreement with
the normal curve is very good.

7.3.3 Applying the Central Limit Theorem

The conclusion of the Central Limit Theorem is that the sampling distribution of
the sample average can be approximated by the Normal distribution, regardless
what is the distribution of the original measurement, but provided that the
sample size is large enough. This statement is very important, since it allows
us, in the context of the sample average, to carry out probabilistic computations
using the Normal distribution even if we do not know the actual distribution of
the measurement. All we need to know for the computation are the expectation
of the measurement, its variance (or standard deviation) and the sample size.
The theorem can be applied whenever probability computations associated
with the sampling distribution of the average are required. The computation
of the approximation is carried out by using the Normal distribution with the
same expectation and the same standard deviation as the sample average.
An example of such computation was conducted in Subsection 7.2.3 where
the central interval that contains 95% of the sampling distribution of a Binomial
average was required. The 2.5%- and the 97.5%-percentiles of the Normal distri-
bution with the same expectation and variance as the sample average produced
boundaries for the interval. These boundaries were in good agreement with the
boundaries produced by the simulation. More examples will be provided in the
Solved Exercises of this chapter and the next one.
With all its usefulness, one should treat the Central Limit Theorem with a
grain of salt. The approximation may be valid for large samples, but may be
bad for samples that are not large enough. When the sample is small a careless
application of the Central Limit Theorem may produce misleading conclusions.
120 CHAPTER 7. THE SAMPLING DISTRIBUTION

7.4 Solved Exercises

Question 7.1. The file “pop2.csv” contains information associated to the
blood pressure of an imaginary population of size 100,000. The file can be found
on the internet (http://pluto.huji.ac.il/~msby/StatThink/Datasets/pop2.
csv). The variables in this file are:

id: A numerical variable. A 7 digits number that serves as a unique identifier

of the subject.

sex: A factor variable. The sex of each subject. The values are either “MALE”
or “FEMALE”.

age: A numerical variable. The age of each subject.

bmi: A numerical variable. The body mass index of each subject.

systolic: A numerical variable. The systolic blood pressure of each subject.

diastolic: A numerical variable. The diastolic blood pressure of each subject.

group: A factor variable. The blood pressure category of each subject. The
values are “NORMAL” both the systolic blood pressure is within its normal
range (between 90 and 139) and the diastolic blood pressure is within its
normal range (between 60 and 89). The value is “HIGH” if either mea-
surements of blood pressure are above their normal upper limits and it is
“LOW” if either measurements are below their normal lower limits.

Our goal in this question is to investigate the sampling distribution of the sample
average of the variable “bmi”. We assume a sample of size n = 150.

1. Compute the population average of the variable “bmi”.

2. Compute the population standard deviation of the variable “bmi”.

3. Compute the expectation of the sampling distribution for the sample av-
erage of the variable.

4. Compute the standard deviation of the sampling distribution for the sam-
ple average of the variable.

5. Identify, using simulations, the central region that contains 80% of the
sampling distribution of the sample average.

6. Identify, using the Central Limit Theorem, an approximation of the cen-

tral region that contains 80% of the sampling distribution of the sample
average.

Solution (to Question 7.1.1): After placing the file “pop2.csv” in the work-
ing directory one may produce a data frame with the content of the file and
compute the average of the variable “bmi” using the code:

> pop.2 <- read.csv(file="pop2.csv")

> mean(pop.2$bmi)
[1] 24.98446
7.4. SOLVED EXERCISES 121

We obtain that the population average of the variable is equal to 24.98446.

Solution (to Question 7.1.2): Applying the function “sd” to the sequence
of population values produces the population standard deviation:

> sd(pop.2$bmi)
[1] 4.188511

In turns out that the standard deviation of the measurement is 4.188511.

Solution (to Question 7.1.3): In order to compute the expectation under the
sampling distribution of the sample average we conduct a simulation. The sim-
ulation produces (an approximation) of the sampling distribution of the sample
average. The sampling distribution is represented by the content of the sequence
“X.bar”:

> X.bar <- rep(0,10^5)

> for(i in 1:10^5)
+ {
+ X.samp <- sample(pop.2$bmi,150)
+ X.bar[i] <- mean(X.samp)
+ }
> mean(X.bar)
[1] 24.98681

Initially, we produce a vector of zeros of the given lenght (100,000). In each

iteration of the “for” loop a random sample of size 150 is selected from the pop-
ulation. The sample average is computed and stored in the sequence “X.bar”.
At the end of all the iterations all the zeros are replaced by evaluations of the
sample average.
The expectation of the sampling distribution of the sample average is com-
puted by the application of the function “mean” to the sequence that represents
the sampling distribution of the sample average. The result for the current is
24.98681, which is vary similar3 to the population average 24.98446.

Solution (to Question 7.1.4): The standard deviation of the sample average
under the sampling distribution is computed using the function “sd”:

> sd(X.bar)
[1] 0.3422717

The resulting standard deviation is 0.3422717. Recall that the standard devi-
ation of a single measurement is equal to 4.188511 and that the sample size is
n = 150. The ratio between the √standard deviation of the measurement and the
square root of 150 is 4.188511/ 150 = 0.3419905, which is similar in value to
the standard deviation of the sample average4 .
3 Theoretically, the two numbers should coincide. The small discrepancy follows from the

fact that the sequence “X.bar” is only an approximation of the sampling distribution.
4 It can be shown mathematically that the variance of the sample average, in the case of

sampling from a population, is equal to [(N − n)/(N − 1)] · Var(X)/n, where Var(X) is the
population variance of the measurement, n is the sample size, and N is the population size.
The factor [(N − n)/(N − 1)] is called the finite population correction. In the current setting
the finite population correction is equal to 0.99851, which is practically equal to one.
122 CHAPTER 7. THE SAMPLING DISTRIBUTION

Solution (to Question 7.1.5): The central region that contains 80% of the
sampling distribution of the sample average can be identified with the aid of the
function “quantile”:
> quantile(X.bar,c(0.1,0.9))
10% 90%
24.54972 25.42629
The value 24.54972 is the 10%-percentile of the sampling distribution. To the
left of this value are 10% of the distribution. The value 25.42629 is the 90%-
percentile of the sampling distribution. To the right of this value are 10% of the
distribution. Between these two values are 80% of the sampling distribution.

Solution (to Question 7.1.6): The Normal approximation, which is the con-
clusion of the Central Limit Theorem substitutes the sampling distribution of
the sample average by the Normal distribution with the same expectation and
standard deviation. The percentiles are computed with the function “qnorm”:
> qnorm(c(0.1,0.9),mean(X.bar),sd(X.bar))
[1] 24.54817 25.42545
Observe that we used the expectation and the standard deviation of the sample
average in the function. The resulting interval is [24.54817, 25.42545], which is
similar to the interval [24.54972, 25.42629] which was obtained via simulations.

Question 7.2. A subatomic particle hits a linear detector at random locations.

The length of the detector is 10 nm and the hits are uniformly distributed. The
location of 25 random hits, measured from a specified endpoint of the interval,
are marked and the average of the location computed.
1. What is the expectation of the average location?
2. What is the standard deviation of the average location?
3. Use the Central Limit Theorem in order to approximate the probability
the average location is in the left-most third of the linear detector.
4. The central region that contains 99% of the distribution of the average is
of the form 5 ± c. Use the Central Limit Theorem in order to approximate
the value of c.
Solution (to Question 7.2.1): Denote by X the distance from the specified
endpoint of a random hit. Observe that X ∼ Uniform(0, 10). The 25 hits form a
sample X1 , X2 , . . . , X25 from this distribution and the sample average X̄ is the
average of these random locations. The expectation of the average is equal to the
expectation of a single measurement. Since E(X) = (a + b)/2 = (0 + 10)/2 = 5
we get that E(X̄) = 5.

Solution (to Question 7.2.2): The variance of the sample average is equal to
the variance of a single measurement, divided by the sample size. The variance
of the Uniform distribution is Var(X) = (a + b)2 /12 = (10 − 0)2 /12 = 8.333333.
The standard deviation of the sample average is equal to the standard deviation
7.5. SUMMARY 123

of the sample average is equal to the standard deviation of a single measurement,

divided by the square root of the sample size. The p sample size is n = 25. Con-
sequently, the standard deviation of the average is 8.333333/25 = 0.5773503.

Solution (to Question 7.2.3): The left-most third of the detector is the
interval to the left of 10/3. The distribution of the sample average, according
to the Central Limit Theorem, is Normal. The probability of being less than
10/3 for the Normal distribution may be computed with the function “pnorm”:
> mu <- 5
> sig <- sqrt(10^2/(12*25))
> pnorm(10/3,mu,sig)
[1] 0.001946209
The expectation and the standard deviation of the sample average are used in
computation of the probability. The probability is 0.001946209, about 0.2%.

Solution (to Question 7.2.3): The central region in the Normal(µ, σ 2 ) distri-
bution that contains 99% of the distribution is of the form µ±qnorm(0.995)·σ,
where “qnorm(0.995)” is the 99.5%-percentile of the Standard Normal distri-
bution. Therefore, c = qnorm(0.995) · σ:
> qnorm(0.995)*sig
[1] 1.487156
We get that c = 1.487156.

7.5 Summary
Glossary
Random Sample: The probabilistic model for the values of a measurements
in the sample, before the measurement is taken.
Sampling Distribution: The distribution of a random sample.
Sampling Distribution of a Statistic: A statistic is a function of the data;
i.e. a formula applied to the data. The statistic becomes a random variable
when the formula is applied to a random sample. The distribution of this
random variable, which is inherited from the distribution of the sample,
is its sampling distribution.
Sampling Distribution of the Sample Average: The distribution of the sam-
ple average, considered as a random variable.
The Law of Large Numbers: A mathematical result regarding the sampling
distribution of the sample average. States that the distribution of the av-
erage of measurements is highly concentrated in the vicinity of the expec-
tation of a measurement when the sample size is large.
The Central Limit Theorem: A mathematical result regarding the sampling
distribution of the sample average. States that the distribution of the av-
erage is approximately Normal when the sample size is large.
124 CHAPTER 7. THE SAMPLING DISTRIBUTION

Discussion in the Forum

Limit theorems in mathematics deal with the convergence of some property to
a limit as some indexing parameter goes to infinity. The Law of Large Numbers
and the Central Limit Theorem are examples of limit theorems. The property
they consider is the sampling distribution of the sample average. The indexing
parameter that goes to infinity is the sample size n.
Some people say that the Law of Large Numbers and the Central Limit
Theorem are useless for practical purposes. These theorems deal with a sample
size that goes to infinity. However, all sample sizes one finds in reality are
necessarily finite. What is your opinion?
When forming your answer to this question you may give an example of a
situation from your own field of interest in which conclusions of an abstract
mathematical theory are used in order to solve a practical problem. Identify
the merits and weaknesses of the application of the mathematical theory.
For example, in making statistical inference one frequently needs to make
statements regarding the sampling distribution of the sample average. For in-
stant, one may want to identify the central region that contains 95% of the
distribution. The Normal distribution is used in the computation. The justifi-
cation is the Central Limit Theorem.

Summary of Formulas
Expectation of the sample average: E(X̄) = E(X)
Variance of the sample average: Var(X̄) = Var(X)/n
Chapter 8

Overview and Integration

8.1 Student Learning Objective

This section provides an overview of the concepts and methods that where
presented in the first part of the book. We attempt to relate them to each other
and put them in prospective. Some problems are provided. The solutions to
these problems require combinations of many of the tools that were presented
in previous chapters. By the end of this chapter, the student should be able to:

• Have a better understanding of the relation between descriptive statistics,

probability, and inferential statistics.

• Distinguish between the different uses of the concept of variability.

• Integrate the tools that were given in the first part of the book in order
to solve complex problems.

8.2 An Overview
The purpose of the first part of the book was to introduce the fundamentals
of statistics and teach the concepts of probability which are essential for the
understanding of the statistical procedures that are used to analyze data. These
procedures are presented and discussed in the second part of the book.
Data is typically obtained by selecting a sample from a population and taking
measurements on the sample. There are many ways to select a sample, but all
methods for such selection should not violate the most important characteristic
that a sample should posses, namely that it represents the population it came
from. In this book we concentrate on simple random sampling. However, the
reader should be aware of the fact that other sampling designs exist and may
be more appropriate in specific applications. Given the sampled data, the main
concern of the science of statistics is in making inference on the parameter of
the population on the basis of the data collected. Such inferences are carried
out with the aid of statistics, which are functions of the data.
Data is frequently stored in the format of a data frame, in which columns
are the measured variable and the rows are the observations associated with the
selected sample. The main types of variables are numeric, either discrete or not,

125
126 CHAPTER 8. OVERVIEW AND INTEGRATION

and factors. We learned how one can produce data frames and read data into R
for further analysis.
Statistics is geared towards dealing with variability. Variability may emerge
in different forms and for different reasons. It can be summarized, analyzed and
handled with many tools. Frequently, the same tool, or tools that have much
resemblance to each other, may be applied in different settings and for different
forms of variability. In order not to loose track it is important to understand in
each scenario the source and nature of the variability that is being examined.
An important split in term of the source of variability is between descriptive
statistics and probability. Descriptive statistics examines the distribution of
data. The frame of reference is the data itself. Plots, such as the bar plots,
histograms and box plot; tables, such as the frequency and relative frequency
as well as the cumulative relative frequency; and numerical summaries, such as
the mean, median and standard deviation, can all serve in order to understand
the distribution of the given data set.
In probability, on the other hand, the frame of reference is not the data at
hand but, instead, it is all data sets that could have been sampled (the sample
space of the sampling distribution). One may use similar plots, tables, and
numerical summaries in order to analyze the distribution of functions of the
sample (statistics), but the meaning of the analysis is different. As a matter
of fact, the relevance of the probabilistic analysis to the data actually sampled
is indirect. The given sample is only one realization within the sample space
among all possible realizations. In the probabilistic context there is no special
role to the observed realization in comparison to all other potential realizations.
The fact that the relation between probabilistic variability and the observed
data is not direct does not make the relation unimportant. On the contrary,
this indirect relation is the basis for making statistical inference. In statistical
inference the characteristics of the data may be used in order to extrapolate
from the sampled data to the entire population. Probabilistic description of the
distribution of the sample is then used in order to assess the reliability of the
extrapolation. For example, one may try to estimate the value of population
parameters, such as the population average and the population standard devi-
ation, on the basis of the parallel characteristics of the data. The variability
of the sampling distribution is used in order to quantify the accuracy of this
estimation. (See Example 5 below.)
Statistics, like many other empirically driven forms of science, uses theo-
retical modeling for assessing and interpreting observational data. In statistics
this modeling component usually takes the form of a probabilistic model for
the measurements as random variables. In the first part of this book we have
encountered several such models. The model of simple sampling assumed that
each subset of a given size from the population has equal probability to be se-
lected as the sample. Other, more structured models, assumed a specific form
to the distribution of the measurements. The examples we considered were the
Binomial, the Poisson, the Uniform, the Exponential and the Normal distribu-
tions. Many more models may be found in the literature and may be applied
when appropriate. Some of these other models have R functions that can be
used in order to compute the distribution and produce simulations.
A statistic is a function of sampled data that is used for making statistical
inference. When a statistic, such as the average, is computed on a random
sample then the outcome, from a probabilistic point of view, is a random vari-
8.3. INTEGRATED APPLICATIONS 127

able. The distribution of this random variable depends on the distribution of

the measurements that form the sample but is not identical to that distribu-
tion. Hence, for example, the distribution of an average of a sample from the
Uniform distribution does not follow the Uniform distribution. In general, the
relation between the distribution of a measurement and the distribution of a
statistic computed from a sample that is generated from that distribution may
be complex. Luckily, in the case of the sample average the relation is rather
simple, at least for samples that are large enough.
The Central Limit Theorem provides an approximation of the distribution
of the sample average that typically improves with the increase in sample size.
The expectation of the sample average is equal to the expectation of a single
measurement and the variance is equal to the variance of a single measurement,
divided by the sample size. The Central Limit Theorem adds to this observation
the statement that the distribution of the sample average may be approximated
by the Normal distribution (with the same expectation and standard deviation
as those of the sample average). This approximation is valid for practically any
distribution of the measurement. The conclusion is, at least in the case of the
sample average, that the distribution of the statistic depends on the underlying
distribution of the measurements only through their expectation and variance
but not through other characteristics of the distribution.
The conclusion of the theorem extends to quantities proportional to the sam-
ple average. Therefore, since the sum of the sample is obtained by multiplying
the sample average by the sample size n, we get that the theorem can be used in
order to approximate the distribution of sums. As a matter of fact, the theorem
may be generalized much further. For example, it may be shown to hold for a
smooth function of the sample average, thereby increasing the applicability of
the theorem and its importance.
In the next section we will solve some practical problems. In order to solve
these problems you are required to be familiar with the concepts and tools
that were introduced throughout the first part of the book. Hence, we strongly
recommend that you read again and review all the chapters of the book that
preceded this one before moving on to the next section.

8.3 Integrated Applications

The main message of the Central Limit Theorem is that for the sample average
we may compute probabilities based on the Normal distribution and obtain rea-
sonable approximations, provided that the sample size is not too small. All we
need to figure out for the computations are the expectation and variance of the
underlying measurement. Otherwise, the exact distribution of that measure-
ment is irrelevant. Let us demonstrate the applicability of the Central Limit
Theorem in two examples.

8.3.1 Example 1
A study involving stress is done on a college campus among the students. The
stress scores follow a (continuous) Uniform distribution with the lowest stress
score equal to 1 and the highest equal to 5. Using a sample of 75 students, find:
128 CHAPTER 8. OVERVIEW AND INTEGRATION

1. The probability that the average stress score for the 75 students is less
than 2.
2. The 90th percentile for the average stress score for the 75 students.
3. The probability that the total of the 75 stress scores is less than 200.
4. The 90th percentile for the total stress score for the 75 students.

Solution:
Denote by X the stress score of a random student. We are given that X ∼
Uniform(1, 5). We use the formulas E(X) = (a+b)/2 and Var(X) = (b−a)2 /12
in order to obtain the expectation and variance of a single observation and then
we use the relations E(X̄) = E(X) and Var(X̄) = Var(X)/n to translated these
results to the expectation and variance of the sample average:
> a <- 1
> b <- 5
> n <- 75
> mu.bar <- (a+b)/2
> sig.bar <- sqrt((b-a)^2/(12*n))
> mu.bar
[1] 3
> sig.bar
[1] 0.1333333
After obtaining the expectation and the variance of the sample average we
can forget about the Uniform distribution and proceed only with the R functions
that are related to the Normal distribution. By the Central Limit Theorem we
get that the distribution of the sample average is approximately Normal(µ, σ 2 ),
with µ = mu.bar and σ = sig.bar.
In the Question 1.1 we are asked to find the value of the cumulative distri-
bution function of the sample average at x = 2:
> pnorm(2,mu.bar,sig.bar)
[1] 3.190892e-14
The goal of Question 1.2 is to identify the 90%-percentile of the sample
average:
> qnorm(0.9,mu.bar,sig.bar)
[1] 3.170874
The sample average is equal to the total sum divided by the number of
observations, n = 75 in this example. The total sum is less than 200 if, and
only if the average is less than 200/n. Therefore, for Question 1.3:
> pnorm(200/n,mu.bar,sig.bar)
[1] 0.006209665
Finally, if 90% of the distribution of the average is less that 3.170874 then
90% of the distribution of the total sum is less than 3.170874 n. In Question 1.4
we get:
> n*qnorm(0.9,mu.bar,sig.bar)
[1] 237.8155
8.3. INTEGRATED APPLICATIONS 129

8.3.2 Example 2
Consider again the same stress study that was described in Example 1 and
answer the same questions. However, this time assume that the stress score may
obtain only the values 1, 2, 3, 4 or 5, with the same likelihood for obtaining
each of the values.

Solution:
Denote again by X the stress score of a random student. The modified dis-
tribution states that the sample space of X are the integers {1, 2, 3, 4, 5}, with
equal probability for each value. Since the probabilities must sum to 1 we get
that P(X = x) = 1/5, for all x in the sample space. In principle we may repeat
the steps of the solution of previous example, substituting the expectation and
standard deviation of the continuous measurement by the discrete counterpart:

> x <- 1:5

> p <- rep(1/5,5)
> n <- 75
> mu.X <- sum(x*p)
> sig.X <- sum((x-mu.X)^2*p)
> mu.bar <- mu.X
> sig.bar <- sqrt(sig.X/n)
> mu.bar
[1] 3
> sig.bar
[1] 0.1632993

Notice that the expectation of the sample average is the same as before but
the standard deviation is somewhat larger due to the larger variance in the
distribution of a single response.
We may apply the Central Limit Theorem again in order to conclude that
distribution of the average is approximately Normal(µ, σ 2 ), with µ = mu.bar as
before and for the new σ = sig.bar.
For Question 2.1 we compote that the cumulative distribution function of
the sample average at x = 2 is approximately equal:

> pnorm(2,mu.bar,sig.bar)
[1] 4.570649e-10

and the 90%-percentile is:

> qnorm(0.9,mu.bar,sig.bar)
[1] 3.209276

which produces the answer to Question 2.2.

Similarly to the solution of Question 1.3 we may conclude that the total sum
is less than 200 if, and only if the average is less than 200/n. Therefore, for
Question 2.3:

> pnorm(200/n,mu.bar,sig.bar)
[1] 0.02061342
130 CHAPTER 8. OVERVIEW AND INTEGRATION

Observe that in the current version of the question we have the score is
integer-valued. Clearly, the sum of scores is also integer valued. Hence we may
choose to apply the continuity correction for the Normal approximation whereby
we approximate the probability that the sum is less than 200 (i.e. is less than
or equal to 199) by the probability that a Normal random variable is less than
or equal to 199.5. Translating this event back to the scale of the average we get
the approximation1 :

> pnorm(199.5/n,mu.bar,sig.bar)
[1] 0.01866821

Finally, if 90% of the distribution of the average is less that 3.170874 then
90% of the distribution of the total sum is less than 3.170874n. Therefore:

> n*qnorm(0.9,mu.bar,sig.bar)
[1] 240.6957

or, after rounding to the nearest integer we get for Question 2.4 the answer 241.

8.3.3 Example 3
Suppose that a market research analyst for a cellular phone company conducts
a study of their customers who exceed the time allowance included on their
basic cellular phone contract. The analyst finds that for those customers who
exceed the time included in their basic contract, the excess time used follows an
exponential distribution with a mean of 22 minutes. Consider a random sample
of 80 customers and find

1. The probability that the average excess time used by the 80 customers in
the sample is longer than 20 minutes.

2. The 95th percentile for the average excess time for samples of 80 customers
who exceed their basic contract time allowances.

Solution:
Let X be the excess time for customers who exceed the time included in their
basic contract. We are told that X ∼ Exponential(λ). For the Exponential
distribution E(X) = 1/λ. Hence, given that E(X) = 22 we can conclude that
λ = 1/22. For the Exponential we also have that Var(X) = 1/λ2 . Therefore:

> lam <- 1/22

> n <- 80
> mu.bar <- 1/lam
> sig.bar <- sqrt(1/(lam^2*n))
> mu.bar
[1] 22
> sig.bar
[1] 2.459675
1 As a matter of fact, the continuity correction could have been applied in the previous two

sections as well, since the sample average has a discrete distribution.

8.3. INTEGRATED APPLICATIONS 131

Like before, we can forget at this stage about the Exponential distribution
and refer henceforth to the Normal Distribution. In Question 2.1 we are asked
to compute the probability above x = 20. The total probability is 1. Hence,
the required probability is the difference between 1 and the probability of being
less or equal to x = 20:

> 1-pnorm(20,mu.bar,sig.bar)
[1] 0.7919241

The goal in Question 2.2 is to find the 95%-percentile of the sample average:

> qnorm(0.95,mu.bar,sig.bar)
[1] 26.04580

8.3.4 Example 4
A beverage company produces cans that are supposed to contain 16 ounces of
beverage. Under normal production conditions the expected amount of beverage
in each can is 16.0 ounces, with a standard deviation of 0.10 ounces.
As a quality control measure, each hour the QA department samples 50 cans
from the production during the previous hour and measures the content in each
of the cans. If the average content of the 50 cans is below a control threshold
then production is stopped and the can filling machine is re-calibrated.

1. Compute the probability that the amount of beverage in a random can is

below 15.95.

2. Compute the probability that the amount of beverage in a sample average

of 50 cans is below 15.95.

3. Find a threshold with the property that the probability of stopping the
machine in a given hour is 5% when, in fact, the production conditions
are normal.

4. Consider the data in the file “QC.csv”2 . It contains measurement results

of 8 hours. Assume that we apply the threshold that was obtained in
Question 4.3. At the end of which of the hours the filling machine needed
re-calibration?

5. Based on the data in the file “QC.csv”, which of the hours contains mea-
surements which are suspected outliers in comparison to the other mea-
surements conducted during that hour?

Solution
The only information we have on the distribution of each measurement is its
expectation (16.0 ounces under normal conditions) and its standard deviation
(0.10, under the same condition). We do not know, from the information pro-
vided in the question, the actual distribution of a measurement. (The fact
that the production conditions are normal does not imply that the distribution
2 URL for the file: http://pluto.huji.ac.il/~msby/StatThink/Datasets/QC.csv
132 CHAPTER 8. OVERVIEW AND INTEGRATION

16.3
●

●
16.2
16.1
16.0
15.9
15.8

●
● ●
15.7

h1 h2 h3 h4 h5 h6 h7 h8

Figure 8.1: Box Plots

of the measurement in the Normal distribution!) Hence, the correct answer to

Question 4.1 is that there is not enough information to calculate the probability.
When we deal with the sample average, on the other hand, we may apply
the Central Limit Theorem in order to obtain at least an approximation of
the probability. Observe that the expectation
√ of the sample average is 16.0
ounces and the standard deviation is 0.1/ 50. The distribution of the average
is approximately the Normal distribution:

> pnorm(15.95,16,0.1/sqrt(50))
[1] 0.000203476

Hence, we get that the probability of the average being less than 15.95 ounces
is (approximately) 0.0002, which is a solution to Question 4.2.
In order to solve Question 4.3 we may apply the function “qnorm” in order
to compute the 5%-percentile of the distribution of the average:

> qnorm(0.05,16,0.1/sqrt(50))
[1] 15.97674
8.3. INTEGRATED APPLICATIONS 133

Consider the data in the file “QC.csv”. Let us read the data into a data
frame by the by the name “QC” and apply the function “summary” to obtain an
overview of the content of the file:

> QC <- read.csv("QC.csv")

> summary(QC)
h1 h2 h3 h4
Min. :15.86 Min. :15.82 Min. :15.68 Min. :15.75
1st Qu.:15.94 1st Qu.:15.93 1st Qu.:15.84 1st Qu.:15.91
Median :16.02 Median :16.01 Median :15.91 Median :15.99
Mean :16.02 Mean :16.01 Mean :15.91 Mean :15.99
3rd Qu.:16.09 3rd Qu.:16.08 3rd Qu.:15.98 3rd Qu.:16.06
Max. :16.22 Max. :16.22 Max. :16.11 Max. :16.31
h5 h6 h7 h8
Min. :15.75 Min. :15.73 Min. :15.73 Min. :15.78
1st Qu.:15.93 1st Qu.:15.93 1st Qu.:15.94 1st Qu.:15.92
Median :15.99 Median :15.98 Median :15.99 Median :15.97
Mean :15.99 Mean :15.98 Mean :15.99 Mean :15.97
3rd Qu.:16.05 3rd Qu.:16.06 3rd Qu.:16.05 3rd Qu.:16.04
Max. :16.17 Max. :16.23 Max. :16.18 Max. :16.25

Observe that the file contains 8 quantitative variables that are given the names
h1, . . . , h8. Each of these variables contains the 50 measurements conducted in
the given hour.
Observe that the mean is computed as part of the summary. The threshold
that we apply to monitor the filling machine is 15.97674. Clearly, the average
of the measurements at the third hour “h3” is below the threshold. Not enough
significance digits of the average of the 8th hour are presented to be able to
say whether the average is below or above the threshold. A more accurate
presentation of the computed mean is obtained by the application of the function
“mean” directly to the data:

> mean(QC$h8)
[1] 15.9736

Now we can see that the average is below the threshold. Hence, the machine
required re-calibration after the 3rd and the 8th hours, which is the answer to
Question 4.4.
In Chapter 3 it was proposed to use box plots in order to identify points that
are suspected to be outliers. We can use the expression “boxplot(QC$h1)” in
order to obtain the box plot of the data of the first hour and go through the
names of the variable one by one in order to screen all variable. Alternatively,
we may apply the function “boxplot” directly to the data frame “QC” and get
a plot with box plots of all the variables in the data frame plotted side by side
(see Figure 8.1):

> boxplot(QC)

Examining the plots we may see that evidence for the existence of outliers can be
spotted on the 4th, 6th, 7th, and 8th hours, providing an answer to Question 4.5
134 CHAPTER 8. OVERVIEW AND INTEGRATION

8.3.5 Example 5
A measurement follows the Uniform(0, b), for an unknown value of b. Two
statisticians propose two distinct ways to estimate the unknown quantity b with
the aid of a sample of size n = 100. Statistician A proposes to use twice the
sample average (2X̄) as an estimate. Statistician B proposes to use the largest
observation instead.
The motivation for the proposal made by Statistician A is that the expecta-
tion of the measurement is equal to E(X) = b/2. A reasonable way to estimate
the expectation is to use the sample average X̄. Thereby, a reasonable way to
estimate b, twice the expectation, is to use 2X̄. A motivation for the proposal
made by Statistician B is that although the largest observation is indeed smaller
that b, still it may not be much smaller than that value.
In order to choose between the two options they agreed to prefer the statistic
that tends to have values that are closer to b. (with respect to the sampling
distribution). They also agreed to compute the expectation and variance of each
statistic. The performance of a statistic is evaluated using the mean square error
(MSE), which is defined as the sum of the variance and the squared difference
between the expectation and b. Namely, if T is the statistic (either the one
proposed by Statistician A or Statistician B) then

M SE = Var(T ) + (E(T ) − b)2 .

A smaller mean square error corresponds to a better, more accurate, statistic.

1. Assume that the actual value of b is 10 (b = 10). Use simulations to com-

pute the expectation, the variance and the MSE of the statistic proposed
by Statistician A.

2. Assume that the actual value of b is 10 (b = 10). Use simulations to

compute the expectation, the variance and the MSE of the statistic pro-
posed by Statistician B. (Hint: the maximal value of a sequence can be
computed with the function “max”.)

3. Assume that the actual value of b is 13.7 (b = 13.7). Use simulations

to compute the expectation, the variance and the MSE of the statistic
proposed by Statistician A.

4. Assume that the actual value of b is 13.7 (b = 13.7). Use simulations

to compute the expectation, the variance and the MSE of the statistic
proposed by Statistician B. (Hint: the maximal value of a sequence can
be computed with the function “max”.)

5. Based on the results in Questions 5.1–4, which of the two statistics seems
to be preferable?

Solution
In Questions 5.1 and 5.2 we take the value of b to be equal to 10. Consequently,
the distribution of a measurement is Uniform(0, 10). In order to generate the
sampling distributions we produce two sequences, “A” and “B”, both of length
100,000, with the evaluations of the statistics:
8.3. INTEGRATED APPLICATIONS 135

> A <- rep(0,10^5)

> B <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X.samp <- runif(100,0,10)
+ A[i] <- 2*mean(X.samp)
+ B[i] <- max(X.samp)
+ }
Observe that in each iteration of the “for” loop a sample of size n = 100
from the Uniform(0, 10) distribution is generated. The statistic proposed by
Statistician A (“2*mean(X.samp)”) is computed and stored in sequence “A”
and the statistic proposed by Statistician B (“max(X.samp)”) is computed and
stored in sequence “B”.
Consider the statistic proposed by Statistician A:
> mean(A)
[1] 9.99772
> var(A)
[1] 0.3341673
> var(A) + (mean(A)-10)^2
[1] 0.3341725
The expectation of the statistic is 9.99772 and the variance is 0.3341673. Con-
sequently, we get that the mean square error is equal to
0.3341673 + (9.99772 − 10)2 = 0.3341725 .
Next, deal with the statistic proposed by Statistician B:
> mean(B)
[1] 9.901259
> var(B)
[1] 0.00950006
> var(B) + (mean(B)-10)^2
[1] 0.01924989
The expectation of the statistic is 9.901259 and the variance is 0.00950006.
Consequently, we get that the mean square error is equal to
0.00950006 + (9.901259 − 10)2 = 0.01924989 .
Observe that the mean square error of the statistic proposed by Statistician B
is smaller.
For Questions 5.3 and 5.4 we run the same type of simulations. All we
change is the value of b (from 10 to 13.7):
> A <- rep(0,10^5)
> B <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X.samp <- runif(100,0,13.7)
+ A[i] <- 2*mean(X.samp)
+ B[i] <- max(X.samp)
+ }
136 CHAPTER 8. OVERVIEW AND INTEGRATION

Again, considering the statistic proposed by Statistician A we get:

> mean(A)
[1] 13.70009
> var(A)
[1] 0.6264204
> var(A) + (mean(A)-13.7)^2
[1] 0.6264204

The expectation of the statistic in this setting is 13.70009 and the variance is
0.6264204. Consequently, we get that the mean square error is equal to

0.6264204 + (13.70009 − 13.7)2 = 0.6264204 .

For the statistic proposed by Statistician B we obtain:

> mean(B)
[1] 13.56467
> var(B)
[1] 0.01787562
> var(B) + (mean(B)-13.7)^2
[1] 0.03618937

The expectation of the statistic is 13.56467 and the variance is 0.01787562.

Consequently, we get that the mean square error is equal to

0.01787562 + (13.56467 − 13.7)2 = 0.03618937 .

Once more, the mean square error of the statistic proposed by Statistician B is
smaller.
Considering the fact that the mean square error of the statistic proposed
by Statistician B is smaller in both cases we may conclude that this statistic
seems to be better for estimation of b in this setting of Uniformly distributed
measurements3 .

Discussion in the Forum

In this course we have learned many subjects. Most of these subjects, espe-
cially for those that had no previous exposure to statistics, were unfamiliar. In
this forum we would like to ask you to share with us the difficulties that you
encountered.
What was the topic that was most difficult for you to grasp? In your opinion,
what was the source of the difficulty?
When forming your answer to this question we will appreciate if you could
elaborate and give details of what the problem was. Pointing to deficiencies
in the learning material and confusing explanations will help us improve the
presentation for the future application of this course.

3 As a matter of fact, it can be proved that the statistic proposed by Statistician B has a

smaller mean square error than the statistic proposed by Statistician A, for any value of b
 Part II

Statistical Inference

137
Chapter 9

Introduction to Statistical
Inference

9.1 Student Learning Objectives

The next section of this chapter introduces the basic issues and tools of statistical
inference. These tools are the subject matter of the second part of this book. In
Chapters 9–15 we use data on the specifications of cars in order to demonstrate
the application of the tools for making statistical inference. In the third section
of this chapter we present the data frame that contains this data. The fourth
section reviews probability topics that were discussed in the first part of the
book and are relevant for the second part. By the end of this chapter, the
student should be able to:
• Define key terms that are associated with inferential statistics.
• Recognize the variables of the “cars.csv” data frame.
• Revise concepts related to random variables, the sampling distribution
and the Central Limit Theorem.

9.2 Key Terms

The first part of the book deals with descriptive statistics and with probability.
In descriptive statistics one investigates the characteristics of the data by using
graphical tools and numerical summaries. The frame of reference is the observed
data. In probability, on the other hand, one extends the frame of reference to
include all data sets that could have potentially emerged, with the observed
data as one among many.
The second part of the book deals with inferential statistics. The aim of
statistical inference is to gain insight regarding the population parameters from
the observed data. The method for obtaining such insight involves the applica-
tion of formal computations to the data. The interpretation of the outcome of
these formal computations is carried out in the probabilistic context, in which
one considers the application of these formal computations to all potential data
sets. The justification for using the specific form of computation on the observed

139
140 CHAPTER 9. INTRODUCTION TO STATISTICAL INFERENCE

data stems from the examination of the probabilistic properties of the formal
computations.
Typically, the formal computations will involve statistics, which are functions
of the data. The assessment of the probabilistic properties of the computations
will result from the sampling distribution of these statistics.
An example of a problem that requires statistical inference is the estimation
of a parameter of the population using the observed data. Point estimation
attempts to obtain the best guess to the value of that parameter. An estimator
is a statistic that produces such a guess. One may prefer an estimator whose
sampling distribution is more concentrated about the population parameter
value over another estimator whose sampling distribution is less so. Hence, the
justification for selecting a specific statistic as an estimator is a consequence of
the probabilistic characteristics of this statistic in the context of the sampling
distribution.
For example, a car manufacture may be interested in the fuel consumption
of a new type of car. In order to do so the manufacturer may apply a standard
test cycle to a sample of 10 new cars of the given type and measure their fuel
consumptions. The parameter of interest is the average fuel consumption among
all cars of the given type. The average consumption of the 10 cars is a point
estimate of the parameter of interest.
An alternative approach for the estimation of a parameter is to construct
an interval that is most likely to contain the population parameter. Such an
interval, which is computed on the basis of the data, is called the a confidence
interval. The sampling probability that the confidence interval will indeed con-
tain the parameter value is called the confidence level. Confidence intervals are
constructed so as to have a prescribed confidence level.
A different problem in statistical inference is hypothesis testing. The scientific
paradigm involves the proposal of new theories and hypothesis that presumably
provide a better description for the laws of Nature. On the basis of these hy-
pothesis one may propose predictions that can be examined empirically. If the
empirical evidence is consistent with the predictions of the new hypothesis but
not with those of the old theory then the old theory is rejected in favor of the
new one. Otherwise, the established theory maintains its status. Statistical hy-
pothesis testing is a formal method for determining which of the two hypothesis
should prevail that uses this paradigm.
Each of the two hypothesis, the old and the new, predicts a different dis-
tribution for the empirical measurements. In order to decide which of the dis-
tributions is more in tune with the data a statistic is computed. This statistic
is called the test statistic. A threshold is set and, depending on where the test
statistic falls with respect to this threshold, the decision is made whether or not
to reject the old theory in favor of the new one.
This decision rule is not error proof, since the test statistic may fall by
chance on the wrong side of the threshold. Nonetheless, by the examination of
the sampling distribution of the test statistic one is able to assess the probability
of making an error. In particular, the probability of erroneously rejecting the
currently accepted theory (the old one) is called the significance level of the test.
Indeed, the threshold is selected in order to assure a small enough significance
level.
Returning to the car manufacturer. Assume that the car in question is man-
ufactured in two different factories. One may want to examine the hypothesis
9.3. THE CARS DATA SET 141

that the car’s fuel consumption is the same for both factories. If 5 of the tested
cars were manufactured in one factory and the other 5 in the other factory
then the test may be based on the absolute value of the difference between the
average consumption of the first 5 and the average consumption of the other 5.
The method of testing hypothesis is also applied in other practical settings
where it is required to make decisions. For example, before a new treatment
to a medical condition is approved for marketing by the appropriate authorities
it must undergo a process of objective testing through clinical trials. In these
trials the new treatment is administered to some patients while other obtain the
(currently) standard treatment. Statistical tests are applied in order to compare
the two groups of patient. The new treatment is released to the market only if
it is shown to be beneficial with statistical significance and it is shown to have
no unacceptable side effects.
In subsequent chapters we will discuss in more details the computation of
point estimation, the construction of confidence intervals, and the application
of hypothesis testing. The discussion will be initiated in the context of a single
measurement but will later be extended to settings that involve comparison of
measurements.
An example of such analysis is the analysis of clinical trials where the re-
sponse of the patients treated with the new procedure is compared to the re-
sponse of patients that were treated with the conventual treatment. This com-
parison involves the same measurement taken for two sub-samples. The tools of
statistical inference – hypothesis testing, point estimation and the construction
of confidence intervals – may be used in order to carry out this comparison.
Other comparisons may involve two measurements taken for the entire sam-
ple. An important tool for the investigation of the relations between two mea-
surements, or variables, is regression. Models of regression describe the change
in the distribution in one variable as a function of the other variable. Again,
point estimation, confidence intervals, and hypothesis testing can be carried out
in order to examine regression models. The variable who’s distribution is the
target of investigation is called the response. The other variable that may affect
that distribution is called the explanatory variable.

9.3 The Cars Data Set

Statistical inference is applied to data in order to address specific research ques-
tion. We will demonstrate different inferential procedures using a specific data
set with the aim of making the discussion of the different procedures more con-
crete. The same data set will be used for all procedures that are presented in
Chapters 10–151 . This data set contains information on various models of cars
and is stored in the CVS file “cars.csv”2 . The file can be found on the in-
ternet at http://pluto.huji.ac.il/~msby/StatThink/Datasets/cars.csv.
You are advised to download this file to your computer and store it in the
working directory of R.
1 Other data sets will be used in Chapter 16 and in the quizzes and assignments.
2 The original “Automobiles” data set is accessible at the UCI Machine Learning Repository
(http://archive.ics.uci.edu/ml). This data was assembled by Jeffrey C. Schlimmer, using
as source the 1985 Model Import Car and Truck Specifications, 1985 Ward’s Automotive
Yearbook. The current file “cars.csv” is based on all 205 observations of the original data
set. We selected 17 of the 26 variables available in the original source.
142 CHAPTER 9. INTRODUCTION TO STATISTICAL INFERENCE

Let us read the content of the CSV file into an R data frame and produce a
brief summary:
> cars <- read.csv("cars.csv")
> summary(cars)
make fuel.type num.of.doors body.style
toyota : 32 diesel: 20 four:114 convertible: 6
nissan : 18 gas :185 two : 89 hardtop : 8
mazda : 17 NA’s: 2 hatchback :70
honda : 13 sedan :96
mitsubishi: 13 wagon :25
subaru : 12
(Other) :100
drive.wheels engine.location wheel.base length
4wd: 9 front:202 Min. : 86.60 Min. :141.1
fwd:120 rear : 3 1st Qu.: 94.50 1st Qu.:166.3
rwd: 76 Median : 97.00 Median :173.2
Mean : 98.76 Mean :174.0
3rd Qu.:102.40 3rd Qu.:183.1
Max. :120.90 Max. :208.1

width height curb.weight engine.size

Min. :60.30 Min. :47.80 Min. :1488 Min. : 61.0
1st Qu.:64.10 1st Qu.:52.00 1st Qu.:2145 1st Qu.: 97.0
Median :65.50 Median :54.10 Median :2414 Median :120.0
Mean :65.91 Mean :53.72 Mean :2556 Mean :126.9
3rd Qu.:66.90 3rd Qu.:55.50 3rd Qu.:2935 3rd Qu.:141.0
Max. :72.30 Max. :59.80 Max. :4066 Max. :326.0

horsepower peak.rpm city.mpg highway.mpg

Min. : 48.0 Min. :4150 Min. :13.00 Min. :16.00
1st Qu.: 70.0 1st Qu.:4800 1st Qu.:19.00 1st Qu.:25.00
Median : 95.0 Median :5200 Median :24.00 Median :30.00
Mean :104.3 Mean :5125 Mean :25.22 Mean :30.75
3rd Qu.:116.0 3rd Qu.:5500 3rd Qu.:30.00 3rd Qu.:34.00
Max. :288.0 Max. :6600 Max. :49.00 Max. :54.00
NA’s : 2.0 NA’s : 2
price
Min. : 5118
1st Qu.: 7775
Median :10295
Mean :13207
3rd Qu.:16500
Max. :45400
NA’s : 4
Observe that the first 6 variables are factors, i.e. they contain qualitative
data that is associated with categorization or the description of an attribute.
The last 11 variable are numeric and contain quantitative data.
Factors are summarized in R by listing the attributes and the frequency of
each attribute value. If the number of attributes is large then only the most
9.3. THE CARS DATA SET 143

frequent attributes are listed. Numerical variables are summarized in R with

the aid of the smallest and largest values, the three quartiles (Q1, the median,
and Q3) and the average (mean).
The third factor variable, “num.of.doors”, as well as several of the numeri-
cal variables have a special category titled “NA’s”. This category describes the
number of missing values among the observations. For a given variable, the
observations for which a value for the variable is not recorded, are marked as
missing. R uses the symbol “NA” to identify a missing value3 .
Missing observations are a concern in the analysis of statistical data. If
the relative frequency of missing values is substantial and the reason for not
obtaining the data for specific observations is related to the phenomena under
investigation than naı̈ve statistical inference may produce biased conclusions.
In the “cars” data frame missing values are less of a concern since their relative
frequency is low.
One should be on the lookout for missing values when applying R to data
since the different functions may have different ways for dealing with missing
values. One should make sure that the appropriate way is applied for the specific
analysis.
Consider the variables of the data frame “cars”:
make: The name of the produces of the car (a factor).
fuel.type: The type of fuel used by the car, either diesel or gas (a factor).
num.of.doors: The number of passenger doors, either two or four (a factor).
body.style: The type of the car (a factor).
drive.wheels: The wheels powered by the engine (a factor).
engine.location: The location in the car of the engine (a factor).
wheel.base: The distance between the centers of the front and rear wheels in
inches (numeric).
length: The length of the body of the car in inches (numeric).
width: The width of the body of the car in inches (numeric).
height: The height of the car in inches (numeric).
curb.weight: The total weight in pounds of a vehicle with standard equipment
and a full tank of fuel, but with no passengers or cargo (numeric).
engine.size: The volume swept by all the pistons inside the cylinders in cubic
inches (numeric).
horsepower: The power of the engine in horsepowers (numeric).
peak.rpm: The top speed of the engine in rounds-per-minute (numeric).
city.mpg: The fuel consumption of the car in city driving conditions, measured
as miles per gallon of fuel (numeric).
3 Indeed, if you scan the CSV file directly by opening it with a spreadsheet then every here

and there you will encounter this symbol.

144 CHAPTER 9. INTRODUCTION TO STATISTICAL INFERENCE

highway.mpg: The fuel consumption of the car in highway driving conditions,

measured as miles per gallon of fuel (numeric).
price: The retail price of the car in US Dollars (numeric).

9.4 The Sampling Distribution

9.4.1 Statistics
Statistical inferences, be it point estimation, confidence intervals, or testing hy-
pothesis, are based on statistics computed from the data. Examples of statistics
are the sample average and the sample standard deviation. These are impor-
tant examples, but clearly not the only ones. Given numerical data, one may
compute the smallest value, the largest value, the quartiles, and the median.
All are examples of statistics. Statistics may also be associated with factors.
The frequency of a given attribute among the observations is a statistic. (An
example of such statistic is the frequency of diesel cars in the data frame.) As
part of the discussion in the subsequent chapters we will consider these and
other types of statistics.
Any statistic, when computed in the context of the data frame being ana-
lyzed, obtains a single numerical value. However, once a sampling distribution
is being considered then one may view the same statistic as a random variable.
A statistic is a function or a formula which is applied to the data frame. Conse-
quently, when a random collection of data frames is the frame of reference then
the application of the formula to each of the data frames produces a random
collection of values, which is the sampling distribution of the statistic.
We distinguish in the text between the case where the statistic is computed
in the context of the given data frame and the case where the computation is
conducted in the context of the random sample. This distinguishing is empha-
sized by the use of small letters for the former and capital letters for the later.
Consider, for example, the sample average. In the context of the observed data
we denote the data values for a specific variable by x1 , x2 , . . . , xn . The sample
average computed for these values is denoted by
x1 + x2 + · · · + xn
x̄ = .
n
On the other hand, if the discussion of the sample average is conducted in the
context of a random sample then the sample is a sequence X1 , X2 , . . . , Xn of
random variables. The sample average is denoted in this context as
X1 + X2 + · · · + Xn
X̄ = .
n
The same formula that was applied to the data values is applied now to the
random components of the random sample. In the first context x̄ is an observed
non-random quantity. In the second context X̄ is a random variable, an abstract
mathematical concept.
A second example is the sample variance. When we compute the sample
variance for the observed data we use the formula:
Pn
2 Sum of the squares of the deviations (xi − x̄)2
s = = i=1 .
Number of values in the sample − 1 n−1
9.4. THE SAMPLING DISTRIBUTION 145

However, when we discuss the sampling distribution of the sample variance we

apply the same formula to the random sample:
Pn
Sum of the squares of the deviations (Xi − X̄)2
S2 = = i=1 .
Number of values in the sample − 1 n−1
Again, S 2 is a random variable whereas s2 is a non-random quantity: The
evaluation of the random variable at the specific sample that is being observed.

9.4.2 The Sampling Distribution

The sampling distribution may emerge as random selection of samples from a
particular population. In such a case, the sampling distribution of the sample,
and hence of the statistic, is linked to the distribution of values of the variable
in the population.
Alternatively, one may assign theoretical distribution to the measurement
associated with the variable. In this other case the sampling distribution of the
statistic is linked to the theoretical model.
Consider, for example, the variable “price” that describes the prices of
the 205 car types (with 4 prices missing) in the data frame “cars”. In order
to define a sampling distribution one may imagine a larger population of car
types, perhaps all the car types that were sold during the 80’s in the United
States, or some other frame of reference, with the car types that are included
in the data frame considered as a random sample from that larger population.
The observed sample corresponds to car types that where sold in 1985. Had
one chosen to consider car types from a different year then one may expect to
obtain other evaluations of the price variable. The reference population, in this
case, is the distribution of the prices of the car types that were sold during the
80’s and the sampling distribution is associated with a random selection of a
particular year within this period and the consideration of prices of car types
sold in that year. The data for 1985 is what we have at hand. But in the
sampling distribution we take into account the possibility that we could have
obtained data for 1987, for example, rather than the data we did get.
An alternative approach for addressing sampling distribution is to consider
a theoretical model. Referring again to the variable “price” one may propose
an Exponential model for the distribution of the prices of cars. This model
implies that car types in the lower spectrum of the price range are more frequent
than cars with a higher price tag. With this model in mind, one may propose
the sampling distribution to be composed of 205 unrelated copies from the
Exponential distribution (or 201 if we do not want to include the missing values).
The rate λ of the associated Exponential distribution is treated as an unknown
parameter. One of the roles of statistical inference is to estimate the value of
this parameter with the aid of the data at hand.
Sampling distribution is relevant also for factor variables. Consider the vari-
able “fuel.type” as an example. In the given data frame the frequency of
diesel cars is 20. However, had one considered another year during the 80’s one
may have obtained a different frequency, resulting in a sampling distribution.
This type of sampling distribution refers to all cars types that were sold in the
United States during the 80’s as the frame of reference.
Alternatively, one may propose a theoretical model for the sampling distribu-
tion. Imagine there is a probability p that a car runs on diesel (and probability
146 CHAPTER 9. INTRODUCTION TO STATISTICAL INFERENCE

1 − p that it runs on gas). Hence, when one selects 205 car types at random
then one obtains that the distribution of the frequency of car types that run on
diesel has the Binomial(205, p) distribution. This is the sampling distribution
of the frequency statistic. Again, the value of p is unknown and one of our tasks
is to estimate it from the data we observe.
In the context of statistical inference the use of theoretical models for the
sampling distribution is the standard approach. There are situation, such as
the application surveys to a specific target population, where the consideration
of the entire population as the frame of reference is more natural. But, in
most other applications the consideration of theoretical models is the method
of choice. In this part of the book, where we consider statistical inference, we
will always use the theoretical approach for modeling the sampling distribution.

9.4.3 Theoretical Distributions of Observations

In the first part of the book we introduced several theoretical models that may
describe the distribution of an observation. Let us take the opportunity and
review the list of models:

Binomial: The Binomial distribution is used in settings that involve counting

the number of occurrences of a particular outcome. The parameters that
determine the distribution are n, the number of observations, and p, the
probability of obtaining the particular outcome in each observation. The
expression “Binomial(n, p)” is used to mark the Binomial distribution.
The sample space for this distribution is formed by the integer values
{0, 1, 2, . . . , n}. The expectation of the distribution is np and the variance
is np(1 − p). The functions “dbinom”, “pbinom”, and “qbinom” may be
used in order to compute the probability, the cumulative probability, and
the percentiles, respectively, for the Binomial distribution. The function
“rbinom” can be used in order to simulate a random sample from this
distribution.
Poisson: The Poisson distribution is also used in settings that involve count-
ing. This distribution approximates the Binomial distribution when the
number of examinations n is large but the probability p of the particular
outcome is small. The parameter that determines the distribution is the
expectation λ. The expression “Poisson(λ)” is used to mark the Poisson
distribution. The sample space for this distribution is the entire collection
of natural numbers {0, 1, 2, . . .}. The expectation of the distribution is λ
and the variance is also λ. The functions “dpois”, “ppois”, and “qpois”
may be used in order to compute the probability, the cumulative proba-
bility, and the percentiles, respectively, for the Poisson distribution. The
function “rpois” can be used in order to simulate a random sample from
this distribution.
Uniform: The Uniform distribution is used in order to model measurements
that may have values in a given interval, with all values in this interval
equally likely to occur. The parameters that determine the distribution are
a and b, the two end points of the interval. The expression “Uniform(a, b)”
is used to identify the Uniform distribution. The sample space for this
distribution is the interval [a, b]. The expectation of the distribution is
9.4. THE SAMPLING DISTRIBUTION 147

(a+b)/2 and the variance is (b−a)2 /12. The functions “dunif”, “punif”,
and “qunif” may be used in order to compute the density, the cumulative
probability, and the percentiles for the Uniform distribution. The function
“runif” can be used in order to simulate a random sample from this
distribution.

Exponential: The Exponential distribution is frequently used to model times

between events. It can also be used in other cases where the outcome of
the measurement is a positive number and where a smaller value is more
likely than a larger value. The parameter that determines the distribution
is the rate λ. The expression “Exponential(λ)” is used to identify the
Exponential distribution. The sample space for this distribution is the
collection of positive numbers. The expectation of the distribution is 1/λ
and the variance is 1/λ2 . The functions “dexp”, “pexp”, and “qexp”
may be used in order to compute the density, the cumulative probability,
and the percentiles, respectively, for the Exponential distribution. The
function “rexp” can be used in order to simulate a random sample from
this distribution.

Normal: The Normal distribution frequently serves as a generic model for the
distribution of a measurement. Typically, it also emerges as an approx-
imation of the sampling distribution of statistics. The parameters that
determine the distribution are the expectation µ and the variance σ 2 . The
expression “Normal(µ, σ 2 )” is used to mark the Normal distribution. The
sample space for this distribution is the collection of all numbers, negative
or positive. The expectation of the distribution is µ and the variance is
σ 2 . The functions “dnorm”, “pnorm”, and “qnorm” may be used in order
to compute the density, the cumulative probability, and the percentiles for
the Normal distribution. The function “rnorm” can be used in order to
simulate a random sample from this distribution.

9.4.4 Sampling Distribution of Statistics

Theoretical models describe the distribution of a measurement as a function of a
parameter, or a small number of parameters. For example, in the Binomial case
the distribution is determined by the number of trials n and by the probability
of success in each trial p. In the Poisson case the distribution is a function of
the expectation λ. For the Uniform distribution we may use the end-points of
the interval, a and b, as the parameters. In the Exponential case, the rate λ
is a natural parameter for specifying the distribution and in Normal case the
expectation µ and the variance σ 2 my be used for that role.
The general formulation of statistical inference problems involves the iden-
tification of a theoretical model for the distribution of the measurements. This
theoretical model is a function of a parameter who’s value is unknown. The
goal is to produce statements that refer to this unknown parameter. These
statements are based on a sample of observations from the given distribution.
For example, one may try to guess the value of the parameter (point estima-
tion), one may propose an interval which contains the value of the parameter
with some subscribed probability (confidence interval) or one may test the hy-
pothesis that the parameter obtains a specific value (hypothesis testing).
148 CHAPTER 9. INTRODUCTION TO STATISTICAL INFERENCE

The vehicles for conducting the statistical inferences are statistics that are
computed as a function of the measurements. In the case of point estimation
these statistics are called estimators. In the case where the construction of an
interval that contains the value of the parameter is the goal then the statistics
are called confidence interval. In the case of testing hypothesis these statistics
are called test statistics.
In all cases of inference, The relevant statistic possesses a distribution that it
inherits from the sampling distribution of the observations. This distribution is
the sampling distribution of the statistic. The properties of the statistic as a tool
for inference are assessed in terms of its sampling distribution. The sampling
distribution of a statistic is a function of the sample size and of the parameters
that determine the distribution of the measurements, but otherwise may be of
complex structure.
In order to assess the performance of the statistics as agents of inference
one should be able to determine their sampling distribution. We will apply two
approaches for this determination. One approach is to use a Normal approxima-
tion. This approach relies on the Central Limit Theorem. The other approach
is to simulate the distribution. This other approach relies on the functions
available in R for the simulation of a random sample from a given distribution.

9.4.5 The Normal Approximation

In general, the sampling distribution of a statistic is not the same as the sam-
pling distribution of the measurements from which it is computed. For example,
if the measurements are from the Uniform distributed then the distribution of
a function of the measurements will, in most cases, not possess the Uniform
distribution. Nonetheless, in many cases one may still identify, at least approx-
imately, what the sampling distribution of the statistic is.
The most important scenario where the limit distribution of the statistic has
a known shape is when the statistic is the sample average or a function of the
sample average. In such a case the Central Limit Theorem may be applied in
order to show that, at least for a sample size not too small, the distribution of
the statistic is approximately Normal.
In the case where the Normal approximation may be applied then a proba-
bilistic statement associated with the sampling distribution of the statistic can
be substituted by the same statement formulated for the Normal distribution.
For example, the probability that the statistic falls inside a given interval may
be approximated by the probability that a Normal random variable with the
same expectation and the same variance (or standard deviation) as the statistic
falls inside the given interval.
For the special case of the sample average one may use the fact that the
expectation of the average of a sample of measurements is equal to the expecta-
tion of a single measurement and the fact that the variance of the average is the
variance of a single measurement, divided by the sample size. Consequently,
the probability that the sample average falls within a given interval may be
approximate by the probability of the same interval according to the Normal
distribution. The expectation that is used for the Normal distribution is the ex-
pectation of the measurement. The standard deviation is the standard deviation
of the measurement, divided by the square root of the number of observations.
9.4. THE SAMPLING DISTRIBUTION 149

The Normal approximation of the distribution of a statistic is valid for cases

other than the sample average or functions thereof. For example, it can be
shown (under some conditions) that the Normal approximation applies to the
sample median, even though the sample median is not a function of the sample
average.
On the other hand, one need not always assume that the distribution of a
statistic is necessarily Normal. In many cases it is not, even for a large sample
size. For example, the minimal value of a sample that is generated from the
Exponential distribution can be shown to follow the Exponential distribution
with an appropriate rate4 , regardless of the sample size.

9.4.6 Simulations
In most problems of statistical inference that will be discussed in this book we
will be using the Normal approximation for the sampling distribution of the
statistic. However, every now and then we may want to check the validity of
this approximation in order to reassure ourselves of its appropriateness. Com-
puterized simulations can be carried out for that checking. The simulations are
equivalent to those used in the first part of the book.
A model for the distribution of the observations is assumed each time a
simulation is carried out. The simulation itself involves the generation of random
samples from that model for the given sample size and for a given value of the
parameter. The statistic is evaluated and stored for each generated sample.
Thereby, via the generation of many samples, an approximation of the sampling
distribution of the statistic is produced. A probabilistic statement inferred from
the Normal approximation can be compared to the results of the simulation.
Substantial disagreement between the Normal approximation and the outcome
of the simulations is an evidence that the Normal approximation may not be
valid in the specific setting.
As an illustration, assume the statistic is the average price of a car. It is
assumed that the price of a car follows an Exponential distribution with some
unknown rate parameter λ. We consider the sampling distribution of the average
of 201 Exponential random variables. (Recall that in our sample there are 4
missing values among the 205 observations.) The expectation of the average
is 1/λ, which is the expectation of a single Exponential random variable. The
2
variance of a single
p observation is 1/λ .√Consequently, the standard deviation
2
of the average is (1/λ )/201 = (1/λ)/ 201 = (1/λ)/14.17745 = 0.0705/λ.
In the first part of the book we found out that for Normal(µ, σ 2 ), the Normal
distribution with expectation µ and variance σ 2 , the central region that contains
95% of the distribution takes the form µ ± 1.96 σ (namely, the interval [µ −
1.96 σ, µ + 1.96 σ]). Thereby, according to the Normal approximation for the
sampling distribution of the average price we state that the region 1/λ ± 1.96 ·
0.0705/λ should contain 95% of the distribution.
We may use simulations in order to validate this approximation for selected
values of the rate parameter λ. Hence, for example, we may choose λ = 1/12, 000
(which corresponds to an expected price of $12,000 for a car) and validate the
approximation for that parameter value.
4 If the rate of an Exponential measurement is λ then the rate of the minimum of n such

measurements is nλ.
150 CHAPTER 9. INTRODUCTION TO STATISTICAL INFERENCE

The simulation itself is carried out by the generation of a sample of size

n = 201 from the Exponential(1/1200) distribution using the function “rexp”
for generating Exponential samples5 . The function for computing the average
(mean) is applied to each sample and the result stored. We repeat this process a
large number of times (100,000 is the typical number we use) in order to produce
an approximation of the sampling distribution of the sample average. Finally,
we check the relative frequency of cases where the simulated average is within
the given range6 . This relative frequency is an approximation of the required
probability and may be compared to the target value of 0.95.
Let us run the proposed simulation for the sample size of n = 201 and for
a rate parameter equal to λ = 1/12000. Observe that the expectation of the
sample average is equal to 12, 000 and the standard deviation is 0.0705 × 12000.
Hence:

> X.bar <- rep(0,10^5)

> for(i in 1:10^5)
+ {
+ X <- rexp(201,1/12000)
+ X.bar[i] <- mean(X)
+ }
> mean(abs(X.bar-12000) <= 1.96*0.0705*12000)
[1] 0.9496

Observe that the simulation produces 0.9496 as the probability of the interval.
This result is close enough to the target probability of 0.95, proposing that the
Normal approximation is adequate in this example.
The simulation demonstrates the appropriateness of the Normal approxima-
tion for the specific value of the parameter that was used. In order to gain more
confidence in the approximation we may want to consider other values as well.
However, simulations in this book are used only for demonstration. Hence, in
most cases where we conduct a simulation experiment, we conduct it only for
a single evaluation of the parameters. We leave it to the curiosity of the reader
to expand the simulations and try other evaluations of the parameters.
Simulations may also be used in order to compute probabilities in cases
where the Normal approximation does not hold. As an illustration, consider
the mid-range statistic. This statistic is computed as the average between the
largest and the smallest values in the sample. This statistic is discussed in the
next chapter.
Consider the case where we obtain 100 observations. Let the distribution
of each observation be Uniform. Suppose we are interested as before in the
central range that contains 95% of the distribution of the mid-range statistic.
The Normal approximation does not apply in this case. Yet, if we specify the
parameters of the Uniform distribution then we may use simulations in order to
compute the range.
As a specific example let the distribution of an observation be Uniform(3, 7).
In the simulation we generate a sample of size n = 100 from this distribution7
5 The expression for generating a sample is “rexp(201,1/12000)”
6 In the case where the simulated averages are stored in the sequence “X.bar” then we may
use the expression “mean(abs(X.bar - 12000) <= 1.96*0.0705*12000)” in order to compute
the relative frequency.
7 With the expression “runif(100,3,7)”.
9.4. THE SAMPLING DISTRIBUTION 151

and compute the mid-range for the sample.

For the computation of the statistic we need to obtain the minimal and the
maximal values of the sample. The minimal value of a sequence is compute with
the function “min”. The input to this function is a sequence and the output is
the minimal value of the sequence. Similarly, the maximal value is computed
with the function “max”. Again, the input to the function is a sequence and the
output is the maximal value in the sequence. The statistic itself is obtained by
adding the two extreme values to each other and dividing the sum by two8 .
We produce, just as before, a large number of samples and compute the
value of the statistic to each sample. The distribution of the simulated values
of the statistic serves as an approximation of the sampling distribution of the
statistic. The central range that contains 95% of the sampling distribution may
be approximated with the aid of this simulated distribution.
Specifically, we approximate the central range by the identification of the
0.025-percentile and the 0.975-percentile of the simulated distribution. Between
these two values are 95% of the simulated values of the statistic. The percentiles
of a sequence of simulated values of the statistic can be identified with the aid of
the function “quantile” that was presented in the first part of the book. The
first argument to the function is a sequence of values and the second argument
is a number p between 0 and 1. The output of the function is the p-percentile
of the sequence9 . The p-percentile of the simulated sequence serves as an ap-
proximation of the p-percentile of the sampling distribution of the statistic.
The second argument to the function “quantile” may be a sequence of val-
ues between 0 and 1. If so, the percentile for each value in the second argument
is computed10 .
Let us carry out the simulation that produces an approximation of the central
region that contains 95% of the sampling distribution of the mid-range statistic
for the Uniform distribution:

> mid.range <- rep(0,10^5)

> for(i in 1:10^5)
+ {
+ X <- runif(100,3,7)
+ mid.range[i] <- (max(X)+min(X))/2
+ }
> quantile(mid.range,c(0.025,0.975))
2.5% 97.5%
4.941680 5.059004

Observe that (approximately) 95% of the sampling distribution of the statistic

are in the range [4.941680, 5.059004].
Simulations can be used in order to compute the expectation, the standard
deviation or any other numerical summary of the sampling distribution of a
8 If the sample is stored in an object by the name “X” then one may compute the mid-range

statistic with the expression “(max(X)+min(X))/2”.

9 The p-percentile of a sequence is a number with the property that the proportion of entries

with values smaller than that number is p and the proportion of entries with values larger
than the number is 1 − p.
10 If the simulated values of the statistic are stored in a sequence by the name “mid.range”

then the 0.025-percentile and the 0.975-percentile of the sequence can be computed with the
expression “quantile(mid.range,c(0.025,0.975))”.
152 CHAPTER 9. INTRODUCTION TO STATISTICAL INFERENCE

statistic. All one needs to do is compute the required summary for the simulated
sequence of statistic values and hence obtain an approximation of the required
summary. For example, we my use the sequence “mid.range” in order to obtain
the expectation and the standard deviation of the mid-range statistic of a sample
of 100 observations from the Uniform(3, 7) distribution:

> mean(mid.range)
[1] 5.000168
> sd(mid.range)
[1] 0.02767719

The expectation of the statistic is obtained by the application of the function

“mean” to the sequence. Observe that it is practically equal to 5. The stan-
dard deviation is obtained by the application of the function “sd”. Its value is
approximately equal to 0.028.

9.5 Solved Exercises

Magnetic fields have been shown to have an effect on living tissue and were
proposed as a method for treating pain. In the case study presented here,
Carlos Vallbona and his colleagues11 sought to answer the question “Can the
chronic pain experienced by postpolio patients be relieved by magnetic fields
applied directly over an identified pain trigger point?”
A total of 50 patients experiencing post-polio pain syndrome were recruited.
Some of the patients were treated with an active magnetic device and the others
were treated with an inactive placebo device. All patients rated their pain before
(score1) and after application of the device (score2). The variable “change”
is the difference between “score1” and “score2. The treatment condition is
indicated by the variable “active.” The value “1” indicates subjects receiving
treatment with the active magnet and the value “2” indicates subjects treated
with the inactive placebo.
This case study is taken from the Rice Virtual Lab in Statistics. More details
on this case study can be found in the case study Magnets and Pain Relief that
is presented in that site.

Question 9.1. The data for the 50 patients is stored in file “magnets.csv”.
The file can be found on the internet at http://pluto.huji.ac.il/~msby/
StatThink/Datasets/magnets.csv. Download this file to your computer and
store it in the working directory of R. Read the content of the file into an R data
frame. Produce a summary of the content of the data frame and answer the
following questions:

1. What is the sample average of the change in score between the patient’s
rating before the application of the device and the rating after the appli-
cation?

2. Is the variable “active” a factor or a numeric variable?

11 Vallbona,Carlos, Carlton F. Hazlewood, and Gabor Jurida. (1997). Response of pain to
static magnetic fields in postpolio patients: A double-blind pilot study. Archives of Physical
and Rehabilitation Medicine 78(11): 1200-1203.
9.5. SOLVED EXERCISES 153

3. Compute the average value of the variable “change” for the patients that
received and active magnet and average value for those that received an
inactive placebo. (Hint: Notice that the first 29 patients received an active
magnet and the last 21 patients received an inactive placebo. The sub-
sequence of the first 29 values of the given variables can be obtained via
the expression “change[1:29]” and the last 21 vales are obtained via the
expression “change[30:50]”.)

4. Compute the sample standard deviation of the variable “change” for the
patients that received and active magnet and the sample standard devia-
tion for those that received an inactive placebo.

5. Produce a boxplot of the variable “change” for the patients that received
and active magnet and for patients that received an inactive placebo.
What is the number of outliers in each subsequence?

Solution (to Question 9.1.1): Let us read the data into a data frame by the
name “magnets” and apply the function “summary” to the data frame:

> magnets <- read.csv("magnets.csv")

> summary(magnets)
score1 score2 change active
Min. : 7.00 Min. : 0.00 Min. : 0.0 "1":29
1st Qu.: 9.25 1st Qu.: 4.00 1st Qu.: 0.0 "2":21
Median :10.00 Median : 6.00 Median : 3.5
Mean : 9.58 Mean : 6.08 Mean : 3.5
3rd Qu.:10.00 3rd Qu.: 9.75 3rd Qu.: 6.0
Max. :10.00 Max. :10.00 Max. :10.0

The variable “change” contains the difference between the patient’s rating be-
fore the application of the device and the rating after the application. The
sample average of this variable is reported as the “Mean” for this variable and
is equal to 3.5.

Solution (to Question 9.1.2): The variable “active” is a factor. Observe

that the summary of this variable lists the two levels of the variable and the fre-
quency of each level. Indeed, the levels are coded with numbers but, nonetheless,
the variable is a factor12 .

Solution (to Question 9.1.3): Based on the hint we know that the expres-
sions “change[1:29]” and “change[30:50]” produce the values of the variable
“change” for the patients that were treated with active magnets and by inactive
placebo, respectively. We apply the function “mean” to these sub-sequences:

> mean(magnets$change[1:29])
[1] 5.241379
> mean(magnets$change[30:50])
[1] 1.095238
12 The number codes are read as character strings into R. Notice that the codes are given in

the data file “magnets.csv” between double quotes.

154 CHAPTER 9. INTRODUCTION TO STATISTICAL INFERENCE

10
●

5
●
8

4
●
6

3
4

2
2

1
0

Figure 9.1: Two Box-plots

The sample average for the patients that were treated with active magnets is
5.241379 and sample average for the patients that were treated with inactive
placebo is 1.095238.

Solution (to Question 9.1.4): We apply the function “sd” to these sub-
sequences:
> sd(magnets$change[1:29])
[1] 3.236568
> sd(magnets$change[30:50])
[1] 1.578124
The sample standard deviation for the patients that were treated with active
magnets is 3.236568 and sample standard deviation for the patients that were
treated with inactive placebo is 1.578124.

Solution (to Question 9.1.5): We apply the function “boxplot” to each

sub-sequences:
> boxplot(magnets$change[1:29])
9.5. SOLVED EXERCISES 155

> boxplot(magnets$change[30:50])

The box-plots are presented in Figure 9.1. The box-plot on the left correspond
to the sub-sequence of the patients that received an active magnet. There are no
outliers in this plot. The box-plot on the right correspond to the sub-sequence
of the patients that received aa inactive placebo. Three values, the values “3”,
“4”, and “5” are associated with outliers. Let us see what is the total number
of observations that receive these values:

> table(magnets$change[30:50])

0 1 2 3 4 5
11 5 1 1 2 1

One may see that a single observation obtained the value “3”, another one
obtained the value “5” and 2 observations obtained the value “4”, a total of
4 outliers13 . Notice that the single point that is associated with the value “4”
actually represents 2 observations and not one.

Question 9.2. In Chapter 13 we will present a statistical test for testing if

there is a difference between the patients that received the active magnets and
the patients that received the inactive placebo in terms of the expected value
of the variable that measures the change. The test statist for this problem is
taken to be
X̄1 − X̄2
p ,
2
S1 /29 + S22 /21
where X̄1 and X̄2 are the sample averages for the 29 patients that receive active
magnets and for the 21 patients that receive inactive placebo, respectively. The
quantities S12 and S22 are the sample variances for each of the two samples. Our
goal is to investigate the sampling distribution of this statistic in a case where
both expectations are equal to each other and to compare this distribution to
the observed value of the statistic.

1. Assume that the expectation of the measurement is equal to 3.5, regard-

less of what the type of treatment that the patient received. We take the
standard deviation of the measurement for patients the receives an active
magnet to be equal to 3 and for those that received the inactive placebo
we take it to be equal to 1.5. Assume that the distribution of the measure-
ments is Normal and there are 29 patients in the first group and 21 in the
second. Find the interval that contains 95% of the sampling distribution
of the statistic.

2. Does the observed value of the statistic, computed for the data frame
“magnets”, falls inside or outside of the interval that is computed in 1?

Solution (to Question 9.2.1): Let us run the following simulation:

> mu1 <- 3.5

> sig1 <- 3
13 An alternative method for obtaining the total count of the observations with values larger

or equal to “3” is to run the expression “sum(magnets$change[30:50] >= 3)”.

156 CHAPTER 9. INTRODUCTION TO STATISTICAL INFERENCE

> mu2 <- 3.5

> sig2 <- 1.5
> test.stat <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X1 <- rnorm(29,mu1,sig1)
+ X2 <- rnorm(21,mu2,sig2)
+ X1.bar <- mean(X1)
+ X2.bar <- mean(X2)
+ X1.var <- var(X1)
+ X2.var <- var(X2)
+ test.stat[i] <- (X1.bar-X2.bar)/sqrt(X1.var/29 + X2.var/21)
+ }
> quantile(test.stat,c(0.025,0.975))
2.5% 97.5%
-2.014838 2.018435

Observe that each iteration of the simulation involves the generation of two
samples. One sample is of size 29 and it is generated from the Normal(3.5, 32 )
distribution and the other sample is of size 21 and it is generated from the
Normal(3.5, 1.52 ) distribution. The sample average and the sample variance are
computed for each sample. The test statistic is computed based on these aver-
ages and variances and it is stored in the appropriate position of the sequence
“test.stat”.
The values of the sequence “test.stat” at the end of all the iterations rep-
resent the sampling distribution of the static. The application of the function
“quantile” to the sequence gives the 0.025-percentiles and the 0.975-percentiles
of the sampling distribution, which are -2.014838 and 2.018435. It follows that
the interval [−2.014838, 2.018435] contains about 95% of the sampling distribu-
tion of the statistic.

Solution (to Question 9.2.2): In order to evaluate the statistic for the given
data set we apply the same steps that were used in the simulation for the
computation of the statistic:

> x1.bar <- mean(magnets$change[1:29])

> x2.bar <- mean(magnets$change[30:50])
> x1.var <- var(magnets$change[1:29])
> x2.var <- var(magnets$change[30:50])
> (x1.bar-x2.bar)/sqrt(x1.var/29 + x2.var/21)
[1] 5.985601

In the first line we compute the sample average for the first 29 patients and in
the second line we compute it for the last 21 patients. In the third and fourth
lines we do the same for the sample variances of the two types of patients.
Finally, in the fifth line we evaluate the statistic. The computed value of the
statistic turns out to be 5.985601, a value that does not belong to the interval
[−2.014838, 2.018435].
9.6. SUMMARY 157

9.6 Summary
Glossary
Statistical Inferential: Methods for gaining insight regarding the population
parameters from the observed data.
Point Estimation: An attempt to obtain the best guess of the value of a
population parameter. An estimator is a statistic that produces such a
guess. The estimate is the observed value of the estimator.
Confidence Interval: An interval that is most likely to contain the population
parameter. The confidence level of the interval is the sampling probability
that the confidence interval contains the parameter value.
Hypothesis Testing: A method for determining between two hypothesis, with
one of the two being the currently accepted hypothesis. A determination is
based on the value of the test statistic. The probability of falsely rejecting
the currently accepted hypothesis is the significance level of the test.
Comparing Samples: Samples emerge from different populations or under
different experimental conditions. Statistical inference may be used to
compare the distributions of the samples to each other.
Regression: Relates different variables that are measured on the same sample.
Regression models are used to describe the effect of one of the variables
on the distribution of the other one. The former is called the explanatory
variable and the later is called the response.
Missing Value: An observation for which the value of the measurement is not
recorded. R uses the symbol “NA” to identify a missing value.

Discuss in the forum

A data set may contain missing values. Missing value is an observation of a
variable for which the value is not recorded. Most statistical procedures delete
observations with missing values and conduct the inference on the remaining
observations.
Some people say that the method of deleting observations with missing values
is dangerous and may lead to biased analysis. The reason is that missing values
may contain information. What is your opinion?
When you formulate your answer to this question it may be useful to come
up with an example from you own field of interest. Think of an example in
which a missing value contains information relevant for inference or an example
in which it does not. In the former case try to assess the possible effects on
the analysis that may emerge due to the deletion of observations with missing
values.
For example, the goal in some clinical trials is to assess the effect of a new
treatment on the survival of patients with a life-threatening illness. The trial
is conducted for a given duration, say two years, and the time of death of the
patients is recorded. The time of death is missing for patients that survived the
entire duration of the trial. Yet, one is advised not to ignore these patients in
the analysis of the outcome of the trial.
158 CHAPTER 9. INTRODUCTION TO STATISTICAL INFERENCE
Chapter 10

Point Estimation

10.1 Student Learning Objectives

The subject of this chapter is the estimation of the value of a parameter on the
basis of data. An estimator is a statistic that is used for estimation. Criteria
for selecting among estimators are discussed, with the goal of seeking an esti-
mator that tends to obtain values that are as close as possible to the value of
the parameter. Different examples of estimation problems are presented and
analyzed. By the end of this chapter, the student should be able to:

• Recognize issues associated with the estimation of parameters.

• Define the notions of bias, variance and mean squared error (MSE) of an
estimator.

• Estimate parameters from data and assess the performance of the estima-
tion procedure.

10.2 Estimating Parameters

Statistic is the science of data analysis. The primary goal in statistic is to draw
meaningful and solid conclusions on a given phenomena on the basis of observed
data. Typically, the data emerges as a sample of observations. An observation
is the outcome of a measurement (or several measurements) that is taken for a
subject that belongs to the sample. These observations may be used in order
to investigate the phenomena of interest. The conclusions are drawn from the
analysis of the observations.
A key aspect in statistical inference is the association of a probabilistic model
to the observations. The basic assumption is that the observed data emerges
from some distribution. Usually, the assumption is that the distribution is linked
to a theoretical model, such as the Normal, Exponential, Poisson, or any other
model that fits the specifications of the measurement taken.
A standard setting in statistical inference is the presence of a sequence of
observations. It is presumed that all the observations emerged from a common
distribution. The parameters one seeks to estimate are summaries or character-
istics of that distribution.

159
160 CHAPTER 10. POINT ESTIMATION

For example, one may be interested in the distribution of price of cars. A

reasonable assumption is that the distribution of the prices is the Exponential(λ)
distribution. Given an observed sample of prices one may be able to estimate
the rate λ that specifies the distribution.
The target in statistical point estimation of a parameter is to produce the
best possible guess of the value of a parameter on the basis of the available data.
The statistic that tries to guess the value of the parameter is called an estimator.
The estimator is a formula applied to the data that produces a number. This
number is the estimate of the value of the parameter.
An important characteristic of a distribution, which is always of interest, is
the expectation of the measurement, namely the central location of the distri-
bution. A natural estimator of the expectation is the sample average. However,
one may propose other estimators that make sense, such as the sample mid-range
that was presented in the previous chapter. The main topic of this chapter is
the identification of criteria that may help us choose which estimator to use for
the estimation of which parameter.
In the next section we discuss issues associated with the estimation of the
expectation of a measurement. The following section deals with the estimation
of the variance and standard deviation – summaries that characterize the spread
of the distribution. The last section deals with the theoretical models of distri-
bution that were introduced in the first part of the book. It discusses ways by
which one may estimate the parameters that characterize these distributions.

10.3 Estimation of the Expectation

A natural candidate for the estimation of the expectation of a random variable
on the basis of a sample of observations is the sample average. Consider, as an
example, the estimation of the expected price of a car using the information in
the data file “cars.csv”. Let us read the data into a data frame named “cars”
and compute the average of the variable “price”:

> cars <- read.csv("cars.csv")

> mean(cars$price)
[1] NA

The application of the function “mean” for the computation of the sample
average produced a missing value. The reason is that the variable “price”
contains 4 missing values. As default, when applied to a sequence that contains
missing values, the function “mean” produce as output a missing value.
The behavior of the function “mean” at the presence of missing values is
determined by the argument “na.rm”1 . If we want to compute the average of the
non-missing values in the sequence we should specify the argument “na.rm” as
“TRUE”. This can be achieved by the inclusion of the expression “na.rm=TRUE”
in the arguments of the function:
1 The name of the argument stands for “NA remove”. If the value of the argument is set to
“TRUE” then the missing values are removed in the computation of the average. Consequently,
the average is computed for the sub-sequence of non-missing values. The default specification
of the argument in the definition of the function is “na.rm=FALSE”, which implies a missing
value for the mean when computed on a sequence that contains missing values.
10.3. ESTIMATION OF THE EXPECTATION 161

> mean(cars$price,na.rm=TRUE)
[1] 13207.13

The resulting average price is, approximately, $13,000.

10.3.1 The Accuracy of the Sample Average

How close is the estimated value of the expectation – the average price – to the
expected price?
There is no way of answering this question on the basis of the data we ob-
served. Indeed, we think of the price of a random car as a random variable.
The expectation we seek to estimate is the expectation of that random variable.
However, the actual value of that expectation is unknown. Hence, not know-
ing what is the target value, how can we determine the distance between the
computed average 13207.13 and that unknown value?
As a remedy for not being able to answer the question we would like to
address we, instead, change the question. In the new formulation of the question
we ignore the data at hand altogether. The new formulation considers the
sample average as a statistic and the question is formulated in terms of the
sampling distribution of that statistic. The question, in its new formulation is:
How close is the sample average of the price, taken as a random variable, to the
expected price?
Notice that in the new formulation of the question the observed average price
x̄ = 13207.13 has no special role. The question is formulated in terms of the
sampling distribution of the sample average (X̄). The observed average value
is only one among many in the sampling distribution of the average.
The advantage of the new formulation of the question is that it can be ad-
dressed. We do have means for investigating the closeness of the estimator to
the parameter and thereby producing meaningful answers. Specifically, consider
the current case where the estimator is the sample average X̄. This estimator
attempts to estimate the expectation E(X) of the measurement, which is the
parameter. Assessing the closeness of the estimator to the parameter corre-
sponds to the comparison between the distribution of the random variable, i.e.
the estimator, and the value of the parameter.
For this comparison we may note that the expectation E(X) is also the
expectation of the sample average X̄. Consequently, in this problem the as-
sessment of the closeness of the estimator to the parameter is equivalent to the
investigation of the spread of the distribution of the sample average about its
expectation.
Consider an example of such investigation. Imagine that the expected price
of a car is equal to $13,000. A question one may ask is how likely it is that
the estimator’s guess at the value is within $1,000 of the actual value? In
other words, what is the probability that sample average falls in the range
[12, 000, 14, 000]?
Let us investigate this question using simulations. Recall our assumption
that the distribution of the price is Exponential. An expectation of 13,000
corresponds to a rate parameter of λ = 1/13, 000. We simulate the sampling
distribution of the estimator by the generation of a sample of 201 Exponential
random variables with this rate. The sample average is computed for each sam-
ple and stored. The sampling distribution of the sample average is approximated
162 CHAPTER 10. POINT ESTIMATION

via the production of a large number of sample averages:

> lam <- 1/13000
> X.bar <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X <- rexp(201,lam)
+ X.bar[i] <- mean(X)
+ }
> mean(abs(X.bar - 1/lam) <= 1000)
[1] 0.7247
In the last line of the code we compute the probability of being within $1,000
of the expected price. Recall that the expected price in the Exponential case
is the reciprocal of the rate λ. In this simulation we obtained 0.7247 as an
approximation of the probability.
In the case of the sample average we may also apply the Normal approx-
imation in order to assess the probability under consideration. In particu-
lar, if λ = 1/13, 000 then the expectation of an Exponential observation is
E(X) = 1/λ = 13, 000 and the variance is Var(X) = 1/λ2 = (13, 000)2 . The
expectation of the sample average is equal to the expectation of the measure-
ment, 13,000 in this example. The variance of the sample average is equal to the
variance of the observation, divided by the sample size. In the current setting
it is equal to (13, 000)2 /201. The standard deviation is equal to the square root
of the variance.
The Normal approximation uses the Normal distribution in order to compute
probabilities associated with the sample average. The Normal distribution that
is used has the same expectation and standard deviation as the sample average:
> mu <- 13000
> sig <- 13000/sqrt(201)
> pnorm(14000,mu,sig) - pnorm(12000,mu,sig)
[1] 0.7245391
The probability of falling within the interval [12000, 14000] is computed as
the difference between the cumulative Normal probability at 14,000 and the
cumulative Normal probability at 12,000.
These cumulative probabilities are computed with the function “pnorm”.
Recall that this function computes the cumulative probability for the Normal
distribution. If the first argument is 14,000 then the function produces the prob-
ability that a Normal random variable is less than or equal to 14,000. Likewise,
if the first argument is 12,000 then the computed probability is the probability
of being less than or equal to 12,000. The expectation of the Normal distribu-
tion enters in the second argument of the function and the standard deviation
enters in the third argument.
The Normal approximation of falling in the interval [12000, 14000], computed
as the difference between the two cumulative probabilities, produces 0.7245391
as the probability2 . Notice that the probability 0.7247 computed in the simu-
lations is in agreement with the Normal approximation.
2 As a matter of fact, the difference is the probability of falling in the half-open interval

(12000, 14000]. However, for continuous distributions the probability of the end-points is zero
and they do not contribute to the probability of the interval.
10.3. ESTIMATION OF THE EXPECTATION 163

If we wish to assess the accuracy of the estimator at other values of the

parameter, say E(X) = 12, 000 (which corresponds to λ = 1/12, 000) or E(X) =
14, 033, (which corresponds to λ = 1/14, 033) all we need to do is change the
expression “lam <- 1/13000” to the new value and rerun the simulation.
Alternatively, we may use a Normal approximation with modified interval,
expectation, and standard deviation. For example, consider the case where the
expected price is equal to $12,000. In order to asses the probability that the
sample average falls within $1,000 of the parameter we should compute the
probability of the interval [11, 000, 13, 000] and change the entries to the first
argument of the function ”pnorm” accordingly. √ The new expectation is 12,000
and the new standard deviation is 12, 000/ 201:

> mu <- 12000

> sig <- 12000/sqrt(201)
> pnorm(13000,mu,sig) - pnorm(11000,mu,sig)
[1] 0.7625775

This time we get that the probability is, approximately, 0.763.

The fact that the computed value of the average 13,207.13 belongs to the
interval [12, 000, 14, 000] that was considered in the first analysis but does not
belong to the interval [11, 000, 13, 000] that was considered in the second analysis
is irrelevant to the conclusions drawn from the analysis. In both cases the
theoretical properties of the sample average as an estimator were considered
and not its value at specific data.
In the simulation and in the Normal approximation we applied one method
for assessing the closeness of the sample average to the expectation it estimates.
This method involved the computation of the probability of being within $1,000
of the expected price. The higher this probability, the more accurate is the
estimator.
An alternative method for assessing the accuracy of an estimator of the
expectation may involve the use of an overall summary of the spread of the
distribution. A standard mean for quantifying the spread of a distribution about
the expectation is the variance (or its square root, the standard deviation).
Given an estimator of the expectation of a measurement, the sample average
for example, we may evaluate the accuracy of the estimator by considering its
variance. The smaller the variance the more accurate is the estimator.
Consider again the case where the sample average is used in order to estimate
the expectation of a measurement. In such a situation the variance of the
estimator, i.e. the variance of the sample average, is obtained as the ratio
between the variance of the measurement Var(X), divided by the sample size
n:
Var(X̄) = Var(X)/n .

Notice that for larger sample sizes the estimator is more accurate. The lager
the sample size n the smaller is the variance of the estimator, in which case the
values of the estimator tend to be more concentrated about the expectation.
Hence, one may make the estimator more accurate by increasing the sample
size.
Another method for improving the accuracy of the average of measurements
in estimating the expectation is the application of a more accurate measurement
164 CHAPTER 10. POINT ESTIMATION

device. If the variance Var(X) of the measurement device decreases so does the
variance of the sample average of such measurements.
In the sequel, when we investigate the accuracy of estimators, we will gener-
ally use overall summaries of the spread of their distribution around the target
value of the parameter.

10.3.2 Comparing Estimators

Notice that the formulation of the accuracy of estimation that we use replaces
the question: “How close is the given value of the estimator to the unknown
value of the parameter?” by the question: “How close are the unknown (and
random) values of the estimator to a given value of the parameter?” In the
second formulation the question is completely academic and unrelated to ac-
tual measurement values. In this academic context we can consider different
potential values of the parameter. Once the value of the parameter has been
selected it can be treated as known in the context of the academic discussion.
Clearly, this does not imply that we actually know what is the true value of the
parameter.
The sample average is a natural estimator of the expectation of the mea-
surement. However, one may propose other estimators. For example, when the
distribution of the measurement is symmetric about the expectation then the
median of the distribution is equal to the expectation. The sample median,
which is a natural estimator of the measurement median, is an alternative es-
timator of the expectation in such case. Which of the two alternatives, the
sample average or the sample median, should we prefer as an estimator of the
expectation in the case of a symmetric distribution?
The straightforward answer to this question is to prefer the better one, the
one which is more accurate. As part of the solved exercises you are asked to
compare the sample average to the sample median as estimators of the expecta-
tion. Here we compare the sample average to yet another alternative estimator
– the mid-range estimator – which is the average between the smallest and the
largest observations.
In the comparison between estimators we do not evaluate them in the context
of the observed data. Rather, we compare them as random variables. The
comparison deals with the properties of the estimators in a given theoretical
context. This theoretical context is motivated by the realities of the situation
as we know them. But, still, the frame of reference is the theoretical model and
not the collected data.
Hence, depending on the context, we may assume in the comparison that the
observations emerge from some distribution. We may specify parameter values
for this distribution and select the appropriate sample size. After setting the
stage we can compare the accuracy of one estimator against that of the other.
Assessment at other parameter values in the context of the given theoretical
model, or of other theoretical models, may provide insight and enhance our
understanding regarding the relative merits and weaknesses of each estimator.
Let us compare the sample average to the sample mid-range as estimators of
the expectation in a situation that we design. Consider a Normal measurement
X with expectation E(X) = 3 and variance that is equal to 2. Assume that
the sample size is n = 100. Both estimators, due to the symmetry of the
Normal distribution, are centered at the expectation. Hence, we may evaluate
10.3. ESTIMATION OF THE EXPECTATION 165

the accuracy of the two estimators using their variances. These variances are
the measure of the spread of the distributions of each estimator about the target
parameter value.
We produce the sampling distribution and compute the variances using a
simulation. Recall that the distribution of the mid-range statistic was simulated
in the previous chapter. In the computation of the mid-range statistic we used
the function “max” that computes the maximum value of its input and the
function “min” that computes the minimum value:

> mu <- 3
> sig <- sqrt(2)
> X.bar <- rep(0,10^5)
> mid.range <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X <- rnorm(100,mu,sig)
+ X.bar[i] <- mean(X)
+ mid.range[i] <- (max(X)+min(X))/2
+ }
> var(X.bar)
[1] 0.02020161
> var(mid.range)
[1] 0.1850595

We get that the variance of the sample average3 is approximately equal to 0.02.
The variance of the mid-range statistic is approximately equal to 0.185, more
than 9 times as large. We see that the accuracy of the sample average is better
in this case than the accuracy of the mid-range estimator. Evaluating the two
estimators at other values of the parameter will produce the same relation.
Hence, in the current example it seems as if the sample average is the better of
the two.
Is the sample average necessarily the best estimator for the expectation?
The next example will demonstrate that this need not always be the case.
Consider again a situation of observing a sample of size n = 100. How-
ever, this time the measurement X is Uniform and not Normal. Say X ∼
Uniform(0.5, 5.5) has the Uniform distribution over the interval [0.5, 5.5]. The
expectation of the measurement is equal to 3 like before, since E(X) = (0.5 +
5.5)/2 = 3. The variance on an observation is Var(X) = (5.5 − 0.5)2 /12 =
2.083333, not much different from the variance that was used in the Normal
case. The Uniform distribution, like the Normal distribution, is a symmetric
distribution about the center of the distribution. Hence, using the mid-range
statistic as an estimator of the expectation makes sense4 .
We re-run the simulations, using the function “runif” for the simulation
of a sample from the Uniform distribution and the parameters of the Uniform
distribution instead of the function “rnorm” that was used before:
3 As a matter of fact, the variance of the sample average is exactly Var(X)/100 = 0.02.
Due to the inaccuracy of the simulation we got a slightly different variance.
4 Observe that the middle range of the Uniform(a, b) distribution, the middle point between

the maximum value of the distribution b and the minimal value a, is (a + b)/2, which is equal
to the expectation of the distribution
166 CHAPTER 10. POINT ESTIMATION

> a <- 0.5

> b <- 5.5
> X.bar <- rep(0,10^5)
> mid.range <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X <- runif(100,a,b)
+ X.bar[i] <- mean(X)
+ mid.range[i] <- (max(X)+min(X))/2
+ }
> var(X.bar)
[1] 0.02074304
> var(mid.range)
[1] 0.001209732

Again, we get that the variance of the sample average is approximately equal to
0.02, which is close to the theoretical value5 . The variance of mid-range statistic
is approximately equal to 0.0012.
Observe that in the current comparison between the sample average and
the mid-range estimator we get that the latter is a clear winner. Examination
of other values of a and b for the Uniform distribution will produce the same
relation between the two competitors. Hence, we may conclude that for the case
of the Uniform distribution the sample average is an inferior estimator.
The last example may serve as yet another reminder that life is never simple.
A method that is good in one situation may not be as good in a different
situation.
Still, the estimator of choice of the expectation is the sample average. In-
deed, in some cases we may find that other methods may produce more accurate
estimates. However, in most settings the sample average beats its competitors.
The sample average also possesses other useful benefits. Its sampling distribu-
tion is always centered at the expectation it is trying to estimate. Its variance
has a simple form, i.e. it is equal to the variance of the measurement divided by
the sample size. Moreover, its sampling distribution can be approximated by
the Normal distribution. Henceforth, due to these properties, we will use the
sample average whenever estimation of the expectation is required.

10.4 Estimation of the Variance and Standard

Deviation
The spread of the measurement about its expected value may be measured by
the variance or by the standard deviation, which is the square root of the vari-
ance. The standard estimator for the variance of the measurement is the sample
variance and the square root of the sample variance is the default estimator of
the standard deviation.
The computation of the sample variance from the data is discussed in Chap-

5 Actually, the exact value of the variance of the sample average is Var(X)/100 =

0.02083333. The results of the simulation are consistent with this theoretical computation.
10.4. VARIANCE AND STANDARD DEVIATION 167

ter 3. Recall that the sample variance is computed via the formula:
Pn
2 Sum of the squares of the deviations (xi − x̄)2
s = = i=1 ,
Number of values in the sample − 1 n−1

Pn xi − x̄ is the
where x̄ is the sample average and n is the sample size. The term
deviation from the sample average of the ith observation and i=1 (xi − x̄)2 is
the sum of the squares of deviations. It is pointed out in Chapter 3 that the
reason for dividing the sum of squares by (n − 1), rather than n, stems from
considerations of statistical inference. A promise was made that these reasonings
will be discussed in due course. Now we want to deliver on this promise.
Let us compare between two competing estimators for the variance, both
considered as random variables. One is the estimator S 2 , which is equal to the
formula for the sample variance applied to a random sample:
Pn
Sum of the squares of the deviations (Xi − X̄)2
S2 = = i=1 ,
Number of values in the sample − 1 n−1
The computation of this statistic can be carried out with the function “var”.
The second estimator is the one obtained when the sum of squares is divided
by the sample size (instead of the sample size minus 1):
Pn
Sum of the squares of the deviations (Xi − X̄)2
= i=1 .
Number of values in the sample n
Observe that the second estimator can be represented in the form:
Pn 2
Pn
i=1 (Xi − X̄) n−1 (Xi − X̄)2
= · i=1 = [(n − 1)/n]S 2 .
n n n−1
Hence, the second estimator may be obtained by the multiplication of the first
estimator S 2 by the ratio (n − 1)/n. We seek to compare between S 2 and
[(n − 1)/n]S 2 as estimators of the variance.
In order to make the comparison concrete, let us consider it in the context
of a Normal measurement with expectation µ = 5 and variance σ 2 = 3. Let us
assume that the sample is of size 20 (n = 20).
Under these conditions we carry out a simulation. Each iteration of the
simulation involves the generation of a sample of size n = 20 from the given
Normal distribution. The sample variance S 2 is computed from the sample with
the application of the function “var”. The resulting estimate of the variance is
stored in an object that is called “X.var”:
> mu <- 5
> std <- sqrt(3)
> X.var <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X <- rnorm(20,mu,std)
+ X.var[i] <- var(X)
+ }
The content of the object “X.var”, at the end of the simulation, approximates
the sampling distribution of the estimator S 2 .
168 CHAPTER 10. POINT ESTIMATION

Our goal is to compare between the performance of the estimator of the

variance S 2 and that of the alternative estimator. In this alternative estimator
the sum of squared deviations is divided by the sample size (n = 20) and not by
the sample size minus 1 (n − 1 = 19). Consequently, the alternative estimator is
obtained by multiplying S 2 by the ratio 19/20. The sampling distribution of the
values of S 2 is approximated by the content of the object “X.var”. It follows
that the sampling distribution of the alternative estimator is approximated by
the object “(19/20)*X.var”, in which each value of S 2 is multiplied by the
appropriate ratio. The comparison between the sampling distribution of S 2 and
the sampling distribution of the alternative estimator is obtained by comparing
between “X.var” and “(19/20)*X.var”.
Let us start by the investigation of the expectation of the estimators. Recall
that when we analyzed the sample average as an estimator of the expectation of
a measurement we obtained that the expectation of the sampling distribution of
the estimator is equal to the value of the parameter it is trying to estimate. One
may wonder: What is the situation for the estimators of the variance? Is it or
is it not the case that the expectation of their sampling distribution equals the
value of the variance? In other words, is the distribution of either estimators of
the variance centered at the value of the parameter they are trying to estimate?
Compute the expectations of the two estimators:

> mean(X.var)
[1] 2.995400
> mean((19/20)*X.var)
[1] 2.845630

Note that 3 is the value of the variance of the measurement that was used in the
simulation. Observe that the expectation of S 2 is essentially equal to 3, whereas
the expectation of the alternative estimator is less than 3. Hence, at least in
the example that we consider, the center of the distribution of S 2 is located on
the target value. On the other hand, the center of the sampling distribution of
the alternative estimator is located off that target value.
As a matter of fact it can be shown mathematically that the expectation of
the estimator S 2 is always equal to the variance of the measurement. This holds
true regardless of what is the actual value of the variance. On the other hand
the expectation of the alternative estimator is always off the target value6 .
An estimator is called unbiased if its expectation is equal to the value of
the parameter that it tries to estimate. We get that S 2 is an unbiased estima-
tor of the variance. Similarly, the sample average is an unbiased estimator of
the expectation. Unlike these two estimators, the alternative estimator of the
variance is a biased estimator.
The default is to use S 2 as the estimator of the variance of the measurement
and to use its square root as the estimator of the standard deviation of the
measurement. A justification, which is frequently quoted to justify this selection,
is the fact that S 2 is an unbiased estimator of the variance7 .
6 For the estimator S 2 we get that E(S 2 ) = Var(X). On the other hand, for the alternative

estimator we get that E([(n − 1)/n] · S 2 ) = [(n − 1)/n]Var(X) 6= Var(X). This statement
holds true also in the cases where the distribution of the measurement is not Normal.
7 As part of your homework assignment you are required to investigate the properties of

S, the square root of S 2 , as an estimator of the standard deviation of the measurement. A

conclusion of this investigation is that S is a biased estimator of the standard deviation.
10.4. VARIANCE AND STANDARD DEVIATION 169

In the previous section, when comparing two competing estimators of the

expectation, or main concern was the quantification of the spread of the sam-
pling distribution of either estimator about the target value of the parameter.
We used that spread as a measure of the distance between the estimator and the
value it tries to estimate. In the setting of the previous section both estimators
were unbiased. Consequently, the variance of the estimators, which measures
the spread of the distribution about its expectation, could be used in order to
quantify the distance between the estimator and the parameter. (Since, for
unbiased estimators, the parameter is equal to the expectation of the sampling
distribution.)
In the current section one of the estimators (S 2 ) is unbiased, but the other
(the alternative estimator) is not. In order to compare their accuracy in esti-
mation we need to figure out a way to quantify the distance between a biased
estimator and the value it tries to estimate.
Towards that end let us recall the definition of the variance. Given a random
variable X with an expectation E(X), we consider the square of the deviations
(X − E(X))2 , which measure the (squared) distance between each value of the
random variable and the expectation. The variance is defined as the expectation
of the squared distance: Var(X) = E[(X − E(X))2 ]. One may think of the
variance as an overall measure of the distance between the random variable and
the expectation.
Assume now that the goal is to assess the distance between an estimator and
the parameter it tries to estimate. In order to keep the discussion on an abstract
level let us use the Greek letter θ (read: theta) to denote this parameter8 . The
estimator is denoted by θ̂ (read: theta hat). It is a statistic, a formula applied
to the data. Hence, with respect to the sampling distribution, θ̂ is a random
variable9 . The issue is to measure the distance between the random variable θ̂
and the parameter θ.
Motivated by the method that led to the definition of the variance we con-
sider the deviations between the estimator and the parameter. The square
deviations (θ̂ − θ)2 may be considered in the current context as a measure of the
(squared) distance between the estimator and the parameter. When we take the
expectation of these square deviations we get an overall measure of the distance
between the estimator and the parameter. This overall distance is called the
mean square error of the estimator and is denoted by MSE:

MSE = E (θ̂ − θ)2 .

 

The mean square error of an estimator is tightly linked to the bias and the
variance of the estimator. The bias of an estimator θ̂ is the difference between
8 The letter θ is frequently used in the statistical literature to denote a parameter of the

distribution. In the previous section we considered θ = E(X) and in this section we consider
θ = Var(X).
9 Observe that we diverge here slightly form our promise to use capital letters to denote

random variables. However, denoting the parameter by θ and denoting the estimator of the
parameter by θ̂ is standard in the statistical literature. As a matter of fact, we will use the
“hat” notation, where a hat is placed over a Greek letter that represents the parameter, in
other places in this book. The letter with the hat on top will represent the estimator and will
always be considered as a random variable. For Latin letters we will still use capital letters,
with or without a hat, to represent a random variable and small letter to represent evaluation
of the random variable for given data.
170 CHAPTER 10. POINT ESTIMATION

the expectation of the estimator and the parameter it seeks to estimate:

Bias = E(θ̂) − θ .
In an unbiased estimator the expectation of the estimator and the estimated
parameter coincide, i.e. the bias is equal to zero. For a biased estimator the
bias is either negative, as is the case for the alternative estimator of the variance,
or else it is positive.
The variance of the estimator, Variance = Var(θ̂), is a measure of the spread
of the sampling distribution of the estimator about its expectation.
The link between the mean square error, the bias, and the variance is de-
scribed by the formula:
MSE = Variance + (Bias)2 .
Hence, the mean square error of an estimator is the sum of its variance, the
(squared) distance between the estimator and its expectation, and the square of
the bias, the square of the distance between the expectation and the parameter.
The mean square error is influenced both by the spread of the distribution about
the expected value (the variance) and by the distance between the expected
value and the parameter (the bias). The larger either of them become the
larger is the mean square error, namely the distance between the estimator and
the parameter.
Let us compare between the mean square error of the estimator S 2 and
the mean square error of the alternative estimator [19/20]S 2 . Recall that we
have computed their expectations and found out that the expectation of S 2 is
essentially equal to 3, the target value of the variance. The expectation of the
alternative estimator turned out to be equal to 2.845630, which is less than the
target value10 . It turns out that the bias of S 2 is zero (or essentially zero in
the simulations) and the bias of the alternative estimator is 2.845630 − 3 =
−0.15437 ≈ −0.15.
In order to compute the mean square errors of both estimators, let us com-
pute their variances:
> var(X.var)
[1] 0.9361832
> var((19/20)*X.var)
[1] 0.8449054
Observe that the variance of S 2 is essentially equal to 0.936 and the variance of
the alternative estimator is essentially equal to 0.845.
The estimator S 2 is unbiased. Consequently, the mean square error of S 2 is
equal to its variance. The bias of the alternative is -0.15. As a result we get
that the mean square error of this estimator, which is the sum of the variance
and the square of the bias, is essentially equal to
0.845 + (−0.15)2 = 0.845 + 0.0225 = 0.8675 .
Observe that the mean square error of the estimator S 2 , which is equal to 0.936,
is larger than the mean square error of the alternative estimator.
10 It can be shown mathematically that E([(n − 1)/n]S 2 ) = [(n − 1)/n]E(S 2 ). Consequently,

the actual value of the expectation of the alternative estimator in the current setting is [19/20]·
3 = 2.85 and the bias is −0.15. The results of the simulation are consistent with this fact.
10.5. ESTIMATION OF OTHER PARAMETERS 171

Notice that even though the alternative estimator is biased it still has a
smaller mean square error than the default estimator S 2 . Indeed, it can be
prove mathematically that when the measurement has a Normal distribution
then the mean square error of the alternative estimator is always smaller than
the mean square error of the sample variance S 2 .
Still, although the alternative estimator is slightly more accurate than S 2
in the estimation of the variance, the tradition is to use the latter. Obeying
this tradition we will henceforth use S 2 whenever estimation of the variance is
required. Likewise, we will use S, the square root of the sample variance, to
estimate the standard deviation.
In order to understand how is it that the biased estimator produced a smaller
mean square error than the unbiased estimator let us consider the two compo-
nents of the mean square error. The alternative estimator is biased but, on the
other hand, it has a smaller variance. Both the bias and the variance contribute
to the mean square error of an estimator. The price for reducing the bias in es-
timation is usually an increase in the variance and vice versa. The consequence
of producing an unbiased estimator such as S 2 is an inflated variance. A bet-
ter estimator is an estimator that balances between the error that results form
the bias and the error that results from the variance. Such is the alternative
estimator.
We will use S 2 in order to estimate the variance of a measurement. A
context in which an estimate of the variance of a measurement is relevant is in
the assessment of the variance of the sample mean. Recall that the variance of
the sample mean is equal to Var(X)/n, where Var(X) is the variance of the
measurement and n is the size of the sample. In the case where the variance
of the measurement is not known one may estimate it from the sample using
S 2 . It follows√that the estimator of the variance of the sample average is S 2 /n.
Similarly, S/ n can be used as an estimator of the standard deviation of the
sample average.

10.5 Estimation of Other Parameters

In the previous two section we considered the estimation of the expectation and
the variance of a measurement. The proposed estimators, the sample average
for the expectation and the sample variance for the variance, are not tied to any
specific model for the distribution of the measurement. They may be applied to
data whether or not a theoretical model for the distribution of the measurement
is assumed.
In the cases where a theoretical model for the measurement is assumed one
may be interested in the estimation of the specific parameters associated with
this model. In the first part of the book we introduced the Binomial, the Poisson,
the Uniform, the Exponential, and the Normal models for the distribution of
measurements. In this section we consider the estimation of the parameters that
determine each of these theoretical distributions based on a sample generated
from the same distribution. In some cases the estimators coincide with the
estimators considered in the previous sections. In other cases the estimators are
different.
Start with the Binomial distribution. We will be interested in the special
case X ∼ Binomial(1, p). This case involves the outcome of a single trial. The
172 CHAPTER 10. POINT ESTIMATION

trial has two possible outcomes, one of them is designated as “success” and
the other as “failure”. The parameter p is the probability of the success. The
Binomial(1, p) distribution is also called the Bernoulli distribution. Our concern
is the estimation of the parameter p based on a sample of observations from this
Bernoulli distribution.
This estimation problem emerges in many settings that involve the assess-
ment of the probability of an event based on a sample of n observations. In
each observation the event either occurs or not. A natural estimator of the
probability of the event is its relative frequency in the sample. Let us show that
this estimator can be represented as an average of a Bernoulli sample and the
sample average is used for the estimation of a Bernoulli expectation.
Consider an event, one may code a measurement X, associated with an
observation, by 1 if the event occurs and by 0 if it does not. Given a sample of
size n, one thereby produces n observations with values 0 or 1. An observation
has the value 1 if the event occurs for that observation or, else, the value is 0.
Notice that E(X) = 1 · p = p. Consequently, the probability of the event is
equal to the expectation of the Bernoulli measurement11 . It turns out that the
parameter one seeks to estimate is the expectation of a Bernoulli measurement.
The estimation is based on a sample of size n of Bernoulli observations.
In Section 10.3 it was proposed to use the sample average as an estimate
of the expectation. The sample average is the sum of the observations, divided
by the number of observation. In the specific case of a sample of Bernoulli
observations, the sum of observation is the sum of zeros and one. The zeros
do not contribute to the sum. Hence, the sum is equal to the number of times
that 1 occurs, namely the frequency of the occurrences of the event. When
we divide by the sample size we get the relative frequency of the occurrences.
The conclusion is that the sample average of the Bernoulli observations and
the relative frequency of occurrences of the event in the sample are the same.
Consequently, the sample relative frequency of the event is also a sample average
that estimates the expectation of the Bernoulli measurement.
We seek to estimate p, the probability of the event. The estimator is the
relative frequency of the event in the sample. We denote this estimator by P̂ .
This estimator is a sample average of Bernoulli observations that is used in order
to estimate the expectation of the Bernoulli distribution. From the discussion
in Section 10.3 one may conclude that this estimator is an unbiased estimator
of p (namely, E(P̂ ) = p) and that its variance is equal to:
Var(P̂ ) = Var(X)/n = p(1 − p)/n ,
where the variance of the measurement is obtained form the formula for the
variance of a Binomial(1, p) distribution12 .
The second example of an integer valued random variable that was consid-
ered in the first part of the book is the Poisson(λ) distribution. Recall that λ
is the expectation of a Poisson measurement. Hence, one may use the sample
average of Poisson observations in order to estimate this parameter.
The first example of a continuous distribution that was discussed in the first
part of the book is the Uniform(a, b) distribution. This distribution is param-
11 The expectation of X ∼ Binomial(n, p) is E(X) = np. In the Bernoulli case n = 1.

Therefore, E(X) = 1 · p = p.
12 The variance of X ∼ Binomial(n, p) is Var(X) = np(1 − p). In the Bernoulli case n = 1.

Therefore, Var(X) = 1 · p(1 − p) = p(1 − p).

10.6. SOLVED EXERCISES 173

eterized by a and b, the end-points of the interval over which the distribution
is defined. A natural estimator of a is the smallest value observed and a natu-
ral estimator of b is the largest value. One may use the function “min” for the
computation of the former estimate from the sample and use the function “max”
for the computation of the later. Both estimators are slightly biased but have
a relatively small mean square error.
Next considered the X ∼ Exponential(λ) random variable. This distribution
was applied in this chapter to model the distribution of the prices of cars. The
distribution is characterized by the rate parameter λ. In order to estimate
the rate one may notice the relation between it and the expectation of the
measurement:
E(X) = 1/λ =⇒ λ = 1/E(X) .
The rate is equal to the reciprocal of the expectation. The expectation can
be estimated by the sample average. Hence a natural proposal is to use the
reciprocal of the sample average as an estimator of the rate:

λ̂ = 1/X̄ .

The final example that we mention is the Normal(µ, σ 2 ) case. The parameter
µ is the expectation of the measurement and may be estimated by the sample
average X̄. The parameter σ 2 is the variance of a measurement, and can be
estimated using the sample variance S 2 .

10.6 Solved Exercises

Question 10.1. In Subsection 10.3.2 we compare the average against the mid-
range as estimators of the expectation of the measurement. The goal of this
exercise is to repeat the analysis, but this time compare the average to the
median as estimators of the expectation in symmetric distributions.
1. Simulate the sampling distribution of average and the median of a sample
of size n = 100 from the Normal(3, 2) distribution. Compute the expec-
tation and the variance of the sample average and of the sample median.
Which of the two estimators has a smaller mean square error?
2. Simulate the sampling distribution of average and the median of a sample
of size n = 100 from the Uniform(0.5, 5.5) distribution. Compute the
expectation and the variance of the sample average and of the sample
median. Which of the two estimators has a smaller mean square error?
Solution (to Question 10.1.1): We simulate the sampling distribution of the
average and the median in a sample generated from the Normal distribution. In
order to do so we copy the code that was used in Subsection 10.3.2, replacing
the object “mid.range” by the object “X.med” and using the function “median”
in order to compute the sample median instead of the computation of the mid-
range statistic:
> mu <- 3
> sig <- sqrt(2)
> X.bar <- rep(0,10^5)
> X.med <- rep(0,10^5)
174 CHAPTER 10. POINT ESTIMATION

> for(i in 1:10^5)

+ {
+ X <- rnorm(100,mu,sig)
+ X.bar[i] <- mean(X)
+ X.med[i] <- median(X)
+ }

The sequence “X.bar” represents the sampling distribution of the sample

average and the sequence “X.med” represents the sampling distribution of the
sample median. We apply the function “mean” to these sequences in order to
obtain the expectations of the estimators:

> mean(X.bar)
[1] 3.000010
> mean(X.med)
[1] 3.000086

The expectation of the measurement, the parameter of interest is equal to 3.

Observe that expectations of the estimators are essentially equal to the expec-
tation13 . Consequently, both estimators are unbiased estimators of the expec-
tation of the measurement.
In order to obtain the variances of the estimators we apply the function
“var” to the sequences that represent their sampling distributions:

> var(X.bar)
[1] 0.02013529
> var(X.med)
[1] 0.03120206

Observe that the variance of the sample average is essentially equal to 0.020
and the variance of the sample median is essentially equal to 0.0312. The mean
square error of an unbiased estimator is equal to its variance. Hence, these
numbers represent the mean square errors of the estimators. It follows that the
mean square error of the sample average is less than the mean square error of the
sample median in the estimation of the expectation of a Normal measurement.

Solution (to Question 10.1.2): We repeat the same steps as before for the
Uniform distribution. Notice that we use the parameters a = 0.5 and b = 5.5 the
same way we did in Subsection 10.3.2. These parameters produce an expectation
E(X) = 3 and a variance Var(X) = 2.083333:

> a <- 0.5

> b <- 5.5
> X.bar <- rep(0,10^5)
> X.med <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X <- runif(100,a,b)
+ X.bar[i] <- mean(X)
13 It can be proved mathematically that for a symmetric distribution the expectation of the

sample average and the expectation of the sample median are both equal to the expectation
of the measurement. The Normal distribution is a symmetric distribution.
10.6. SOLVED EXERCISES 175

+ X.med[i] <- median(X)

+ }

Applying the function “mean” to the sequences that represent the sampling
distribution of the estimators we obtain that both estimators are essentially
unbiased14 :

> mean(X.bar)
[1] 3.000941
> mean(X.med)
[1] 3.001162

Compute the variances:

> var(X.bar)
[1] 0.02088268
> var(X.med)
[1] 0.06069215

Observe 0.021 is, essentially, the value of the variance of the sample average15 .
The variance of the sample median is essentially equal to 0.061. The variance
of each of the estimators is equal to it’s mean square error. This is the case
since the estimators are unbiased. Consequently, we again obtain that the mean
square error of the sample average is less than that of the sample median.

Question 10.2. The goal in this exercise is to assess estimation of a proportion

in a population on the basis of the proportion in the sample.
The file “pop2.csv” was introduced in Exercise 7.1 of Chapter 7. This file
contains information associated to the blood pressure of an imaginary popu-
lation of size 100,000. The file can be found on the internet (http://pluto.
huji.ac.il/~msby/StatThink/Datasets/pop2.csv). One of the variables in
the file is a factor by the name “group” that identifies levels of blood pressure.
The levels of this variable are “HIGH”, “LOW”, and “NORMAL”.
The file “ex2.csv” contains a sample of size n = 150 taken from the given
population. This file can also be found on the internet (http://pluto.huji.
ac.il/~msby/StatThink/Datasets/ex2.csv). It contains the same variables
as in the file “pop2.csv”. The file “ex2.csv” corresponds in this exercise to
the observed sample and the file “pop2.csv” corresponds to the unobserved
population.
Download both files to your computer and answer the following questions:

1. Compute the proportion in the sample of those with a high level of blood
pressure16 .

2. Compute the proportion in the population of those with a high level of

blood pressure.
14 The Uniform distribution is symmetric. Consequently, both estimators are unbiased.
15 As a matter of fact, the variance is equal to 0.02. The discrepancy results from the fact
that simulations serves only as an approximation to the sampling distribution.
16 Hint: You may use the function summary or you may note that the expression “vari-

able==level” produces a sequence with logical “TRUE” or “FALSE” entries that identify entries
in the sequence “variable” that obtain the value “level”.
176 CHAPTER 10. POINT ESTIMATION

3. Simulate the sampling distribution of the sample proportion and compute

its expectation.
4. Compute the variance of the sample proportion.
5. It is proposed in Section 10.5 that the variance of the sample proportion
is Var(P̂ ) = p(1 − p)/n, where p is the probability of the event (having a
high blood pressure in our case) and n is the sample size (n = 150 in our
case). Examine this proposal in the current setting.
Solution (to Question 10.2.1): Assuming that the file “ex2.csv” is saved
in the working directory, one may read the content of the file into a data frame
and produce a summary of the content of the data frame using the code:
> ex2 <- read.csv("ex2.csv")
> summary(ex2)
id sex age bmi
Min. :1024982 FEMALE:74 Min. :26.00 Min. :15.12
1st Qu.:3172783 MALE :76 1st Qu.:32.00 1st Qu.:22.02
Median :5200484 Median :35.00 Median :25.16
Mean :5463304 Mean :35.09 Mean :24.76
3rd Qu.:7982902 3rd Qu.:37.00 3rd Qu.:27.49
Max. :9934175 Max. :45.00 Max. :35.24
systolic diastolic group
Min. :100.8 Min. : 51.98 HIGH : 37
1st Qu.:118.1 1st Qu.: 75.02 LOW : 3
Median :124.3 Median : 83.19 NORMAL:110
Mean :125.3 Mean : 82.44
3rd Qu.:132.6 3rd Qu.: 88.83
Max. :154.5 Max. :112.71
Examine the variable “group”. Observe that the sample contains 37 subjects
with high levels of blood pressure. Dividing 37 by the sample size we get:
> 37/150
[1] 0.2466667
Consequently, the sample proportion is 0.2466667.
Alternatively, we compute the sample proportion using the code:
> mean(ex2$group == "HIGH")
[1] 0.2466667
Notice that the expression “ex2$group == "HIGH"” produces a sequence of
length 150 with logical entries. The entry is equal to “TRUE” if the equality
holds and it is equal to “FALSE” if it dos not17 . When the function “mean” is
applied to a sequence with logical entries it produces the relative frequeny of the
TRUEs in the sequence. This corresponds, in the corrent context, to the sample
proportion of the level “HIGH” in the variable “ex2$group”.

Solution (to Question 10.2.2): Make sure that the file “pop2.csv” is saved
in the working directory. In order to compute the proportion in the population
17 Pay attention to the fact that we use “==” in order to express equation and not “=”.
10.6. SOLVED EXERCISES 177

we read the content of the file into a data frame and compute the relative
frequency of the level “HIGH” in the variable “group”:

> pop2 <- read.csv("pop2.csv")

> mean(pop2$group == "HIGH")
[1] 0.28126

We get that the proporion in the population is p = 0.28126.

Solution (to Question 10.2.3): The simulation of the sampling distribution

involves a selection of a random sample of size 150 from the population and
the computation of the proportion of the level “HIGH” in that sample. This
procedure is iterated 100,000 times in order to produce an approximation of the
distribution:

> P.hat <- rep(0,10^5)

> for(i in 1:10^5)
+ {
+ X <- sample(pop2$group,150)
+ P.hat[i] <- mean(X == "HIGH")
+ }
> mean(P.hat)
[1] 0.2812307

Observe that the sampling distribution is stored in the object “P.hat”. The
function “sample” is used in order to sample 150 observation from the sequence
“pop2$group”. The sample is stored in the object “X”. The expression “mean(X
== "HIGH")” computes the relative frequency of the level “HIGH” in the sequence
“X”.
At the last line, after the production of the sequence “P.hat” is completed,
the function “mean” is applied to the sequence. The result is the expected value
of estimator P̂ , which is equal to 0.2812307. This expectation is essentially
equal to the probability of the event p = 0.28126.18

Solution (to Question 10.2.4): The application of the function “var” to the
sequence “P.hat” produces:

> var(P.hat)
[1] 0.001350041

Hence, the variance of the estimator is (approximately) equal to 0.00135.

Solution (to Question 10.2.5): Compute the variance according to the for-
mula that is proposed in Section:

> p <- mean(pop2$group == "HIGH")

> p*(1-p)/150
[1] 0.001347685
18 It can be shown mathematically that for random sampling from a population we have

E(P̂ ) = p. The discrepancy from the mathematical theory results from the fact that simula-
tions serves only as an approximation to the sampling distribution.
178 CHAPTER 10. POINT ESTIMATION

We get that the proposed variance in Section 10.5 is 0.0013476850, which is in

good agreement with the value 0.00135 that was obtained in the simulation19 .

10.7 Summary
Glossary
Point Estimation: An attempt to obtain the best guess of the value of a
population parameter. An estimator is a statistic that produces such a
guess. The estimate is the observed value of the estimator.

Bias: The difference between the expectation of the estimator and the value
of the parameter. An estimator is unbiased if the bias is equal to zero.
Otherwise, it is biased.

Mean Square Error (MSE): A measure of the concentration of the distribu-

tion of the estimator about the value of the parameter. The mean square
error of an estimator is equal to the sum of the variance and the square of
the bias. If the estimator is unbiased then the mean square error is equal
to the variance.

Bernoulli Random Variable: A random variable that obtains the value “1”
with probability p and the value “0” with probability 1 − p. It coincides
with the Binomial(1, p) distribution. Frequently, the Bernoulli random
variable emerges as the indicator of the occurrence of an event.

Discuss in the forum

Performance of estimators is assessed in the context of a theoretical model for
the sampling distribution of the observations. Given a criteria for optimality,
an optimal estimator is an estimator that performs better than any other esti-
mator with respect to that criteria. A robust estimator, on the other hand, is
an estimator that is not sensitive to misspecification of the theoretical model.
Hence, a robust estimator may be somewhat inferior to an optimal estimator in
the context of an assumed model. However, if in actuality the assumed model
is not a good description of reality then robust estimator will tend to perform
better than the estimator denoted optimal.
Some say that optimal estimators should be preferred while other advocate
the use of more robust estimators. What is your opinion?
When you formulate your answer to this question it may be useful to come up
with an example from you own field of interest. Think of an estimation problem
and possible estimators that can be used in the context of this problem. Try to
identify a model that is natural to this problem an ask yourself in what ways
may this model err in its attempt to describe the real situation in the estimation
problem.
19 It can be shown theoretically that the variance of the sample proportion, in the case of
sampling from a population, is equal to [(N − n)/(N − 1)] · p(1 − p)/n, where n is the sample
size, and N is the population size. The factor [(N − n)/(N − 1)] is called the finite population
correction. In the current setting the finite population correction is equal to 0.99851, which
is practically equal to one.
10.7. SUMMARY 179

As an example consider estimation of the expectation of a Uniform mea-

surement. We demonstrated that the mid-range estimator is better than the
sample average if indeed the measurements emerge from the Uniform distribu-
tion. However, if the modeling assumption is wrong then this may no longer be
the case. If the distribution of the measurement in actuality is not symmetric or
if the distribution is more concentrated in the center than in the tails then the
performance of the mid-range estimator may deteriorate. The sample average,
on the other hand is not sensitive to the distribution not being symmetric.

Formulas:
• Bias: Bias = E(θ̂) − θ.
 
• Variance: Var(θ̂) = E (θ̂ − E(θ̂))2 .
 
• Mean Square Error: MSE = E (θ̂ − θ)2 .
180 CHAPTER 10. POINT ESTIMATION
Chapter 11

Confidence Intervals

11.1 Student Learning Objectives

A confidence interval is an estimate of an unknown parameter by a range of
values. This range contains the value of the parameter with a prescribed proba-
bility, called the confidence level. In this chapter we discuss the construction of
confidence intervals for the expectation and for the variance of a measurement
as well as for the probability of an event. In some cases the construction will
apply the Normal approximation suggested by the Central Limit Theorem. This
approximation is valid when the sample size is large enough. The construction
of confidence intervals for a small sample is considered in the context of Normal
measurements. By the end of this chapter, the student should be able to:
• Define confidence intervals and confidence levels.
• Construct a confidence interval for the expectation of a measurement and
for the probability of an event.
• Construct a confidence interval for expectation and for the variance of a
Normal measurement.
• Compute the sample size that will produce a confidence interval of a given
width.

11.2 Intervals for Mean and Proportion

A confidence interval, like a point estimator, is a method for estimating the
unknown value of a parameter. However, instead of producing a single number,
the confidence interval is an interval of numbers. The interval of values is cal-
culated from the data. The confidence interval is likely to include the unknown
population parameter. The probability of the event of inclusion is denoted as
the confidence level of the confidence intervals.
This section presents a method for the computation of confidence intervals
for the expectation of a measurement and a similar method for the computation
of a confidence interval for the probability of an event. These methods rely on
the application of the Central Limit Theorem to the sample average in the one
case, and to the sample proportion in the other case.

181
182 CHAPTER 11. CONFIDENCE INTERVALS

In the first subsection we compute a confidence interval for the expectation of

the variable “price” and a confidence interval for the proportion of diesel cars.
The confidence intervals are computed based on the data in the file “cars.csv”.
In the subsequent subsections we discuss the theory behind the computation of
the confidence intervals and explain the meaning of the confidence level. Sub-
section 11.2.2 does so with respect to the confidence interval for the expectation
and Subsection 11.2.3 with respect to the confidence interval for the proportion.

11.2.1 Examples of Confidence Intervals

A point estimator of the expectation of a measurement is the sample average of
the variable that is associated with the measurement. A confidence interval is
an interval of numbers that is likely to contain the parameter value. A natural
interval to consider is an interval centered at the sample average x̄. The interval
is set to have a width that assures the inclusion of the parameter value in the
prescribed probability, namely the confidence level.
Consider the confidence interval for the expectation. √ The structure of√the
confidence interval of confidence level 95% is [x̄ − 1.96 · s/ n, x̄ + 1.96 · s/ n],
where s is the estimated standard deviation of the measurement (namely, the
sample standard deviation) and n is the sample size. This interval may be
expressed in the form:
√
x̄ ± 1.96 · s/ n .
As an illustration, let us construct a 0.95-confidence interval for the expected
price of a car. :

> cars <- read.csv("cars.csv")

> x.bar <- mean(cars$price,na.rm=TRUE)
> s <- sd(cars$price,na.rm=TRUE)
> n <- 201

In the first line of code the data in the file “cars.csv” is stored in a data frame
called “cars”. In the second line the average x̄ is computed for the variable
“price” in the data frame “cars”. This average is stored under the name
“x.bar”. Recall that the variable “price” contains 4 missing values. Hence, in
order to compute the average of the non-missing values we should set a “TRUE”
value to the argument “na.rm”. The sample standard deviation “s” is computed
in the third line by the application of the function “sd”. We set once more the
argument “na.rm=TRUE” in order to deal with the missing values. Finally, in
the last line we store the sample size “n”, the number of non-missing values.
Let us compute the lower and the upper limits of the confidence interval for
the expectation of the price:

> x.bar - 1.96*s/sqrt(n)

[1] 12108.47
> x.bar + 1.96*s/sqrt(n)
[1] 14305.79

The lower limit of the confidence interval turns out to be $12,108.47 and the
upper limit is $14,305.79. The confidence interval is the range of values between
these two numbers.
11.2. INTERVALS FOR MEAN AND PROPORTION 183

Consider, next, a confidence interval for the probability of an event. The

estimate of the probability p is p̂, the relative proportion of occurrences of the
event in the sample. Again, we construct an interval about this estimate.  In
this case, a confidence interval of confidence level 95% is of the form p̂ − 1.96 ·
p p 
p̂(1 − p̂)/n, p̂ + 1.96 · p̂(1 − p̂)/n , where n is the sample size. Observe that
p̂ replaces x̄ as the estimate of the parameter and that p̂(1 − p̂)/n replace s2 /n
as the estimate of the variance of the estimator. The confidence interval for the
probability may be expressed in the form:
p
p̂ ± 1.96 · p̂(1 − p̂)/n .

As an example, let us construct a confidence interval for the proportion of

car types that use diesel fuel. The variable “fuel.type” is a factor that records
the type of fuel the car uses, either diesel or gas:

> table(cars$fuel.type)

diesel gas
20 185

Only 20 of the 205 types of cars are run on diesel in this data set. The point
estimation of the probability of such car types and the confidence interval for
this probability are:

> n <- 205

> p.hat <- 20/n
> p.hat
[1] 0.09756098
> p.hat - 1.96*sqrt(p.hat*(1-p.hat)/n)
[1] 0.05694226
> p.hat + 1.96*sqrt(p.hat*(1-p.hat)/n)
[1] 0.1381797

The point estimation of the probability is p̂ = 20/205 ≈ 0.098 and the confidence
interval, after rounding up, is [0.057, 0.138].

11.2.2 Confidence Intervals for the Mean

In the previous subsection we computed a confidence interval for the expected
price of a car and a confidence interval for the probability that a car runs on
diesel. In this subsection we explain the theory behind the construction of
confidence intervals for the expectation. The theory provides insight to the way
confidence intervals should be interpreted. In the next subsection we will discuss
the theory behind the construction of confidence intervals for the probability of
an event.
Assume one is interested in a confidence interval for the expectation of a
measurement X. For a sample of size n, one may compute the sample average
X̄, which is the point estimator for the expectation. The expected value of the
sample average is the expectation E(X), for which we are trying to produce the
confidence interval. Moreover, the variance of the sample average is Var(X)/n,
where Var(X) is the variance of a single measurement and n is the sample size.
184 CHAPTER 11. CONFIDENCE INTERVALS

The construction of a confidence interval for the expectation relies on the

Central Limit Theorem and on estimation of the variance of the measurement.
The Central Limit Theorem states pthat the distribution of the (standardized)
sample average Z = (X̄ − E(X)/ Var(X)/n is approximately standard Nor-
mal for a large enough sample size. The variance of the measurement can be
estimated using the sample variance S 2 .
Supposed that we are interested in a confidence interval with a confidence
level of 95%. The value 1.96 is the 0.975-percentile of the standard Normal.
Therefore, about 95% of the distribution of the standardized sample average is
concentrated in the range [−1.96, 1, 96]:
  
 X̄ − E(X) 
P p
  ≤ 1.96 ≈ 0.95
Var(X)/n 
The event, the probability of which is being described in the last display,
states that the absolute value of deviation of the sample average from the ex-
pectation, divided by the standard deviation of the sample average, is no more
than 1.96. In other words, the distance between the sample average and the
expectation is at most 1.96 units of standard deviation. One may rewrite this
event in a form that puts the expectation within an interval that is centered at
the sample average1 :
n p o
|X̄ − E(X)| ≤ 1.96 · Var(X)/n ⇐⇒
n p p o
X̄ − 1.96 · Var(X)/n ≤ E(X) ≤ X̄ + 1.96 · Var(X)/n .

Clearly, the probability of the later event is (approximately) 0.95 since we

are considering the same event, each time represented in a different form. The
second representation states that thepexpectation E(X) belongs to an interval
about the sample average: X̄ ± 1.96 Var(X)/n. This interval is, almost, the
confidence interval we seek.
The difficulty is that we do not know the value of the variance Var(X), hence
we cannot compute the interval in the proposed form from the data. In order
to overcome this difficulty we recall that the unknown variance may nonetheless
be estimated from the data:
p √
S 2 ≈ Var(X) =⇒ Var(X)/n ≈ S/ n ,

where S is the sample standard deviation2 .

When the sample size is sufficiently large, so that S is very close to the
value of p
the standard deviation of an observation,
√ we obtain that the interval
X̄±1.96 Var(X)/n and the interval X̄±1.96·S/ n almost coincide. Therefore:
 
S S
P X̄ − 1.96 · √ ≤ E(X) ≤ X̄ + 1.96 · √ ≈ 0.95 .
n n
p
1 Observe

that |X̄ −E(X)| = |E(X)− X̄| and therefore |X̄ −E(X)| ≤ 1.96· Var(X)/n =
 p
|E(X) − X̄| ≤ 1.96 · Var(X)/n . From the definition of the absolute value we obtain that
 p p
the last expression is equal to − 1.96 · Var(X)/n ≤ E(X) − X̄ ≤ 1.96 · Var(X)/n .
Moving the average to the otherpside of the inequality (for both inequalities
p involved) produces

the representation X̄ − 1.96 · Var(X)/n ≤ E(X) ≤ X̄ + 1.96 · Var(X)/n .
2 The sample variance, that serves as the estimator of the variance, is computed from the

sample using the formula: S 2 = n 2

P
i=1 (Xi − X̄) /(n − 1).
11.2. INTERVALS FOR MEAN AND PROPORTION 185

√
Hence, X̄ ± 1.96 · S/ n is an (approximate) confidence interval of the (approx-
imate) confidence level 0.95.
Let us demonstrate the issue of confidence level by running a simulation.
We are interested in a confidence interval for the expected price of a car. In the
simulation we assume that the distribution of the price is Exponential(1/13000).
(Consequently, E(X) = 13, 000). We take the sample size to be equal to n = 201
and compute the actual probability of the confidence interval containing the
value of the expectation:
> lam <- 1/13000
> n <- 201
> X.bar <- rep(0,10^5)
> S <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X <- rexp(n,lam)
+ X.bar[i] <- mean(X)
+ S[i] <- sd(X)
+ }
> LCL <- X.bar - 1.96*S/sqrt(n)
> UCL <- X.bar + 1.96*S/sqrt(n)
> mean((13000 >= LCL) & (13000 <= UCL))
[1] 0.94518
Below we will go over the code and explain the simulation. But, before doing
so, notice that the actual probability that the confidence interval contains the
expectation is about 0.945, which is slightly below the nominal confidence level
of 0.95. Still quoting the nominal value as the confidence level of the confidence
interval is not too far from reality.
Let us look now at the code that produced the simulation. In each iteration
of the simulation a sample is generated. The sample average and standard
deviations are computed and stored in the appropriate locations of the sequences
“X.bar” and “S”. At the end of all the iterations the content of these two
sequences represents the sampling distribution of the sample average X̄ and the
sample standard deviation S, respectively.
The lower and the upper end-points of the confidence interval are computed
in the next two lines of code. The lower level of the confidence interval is stored
in the object “LCL” and the upper level is stored in “UCL”. Consequently, we
obtain the sampling distribution of the confidence interval. This distribution is
approximated by 100,000 random confidence intervals that are generated by the
sampling distribution. Some of these random intervals contain the value of the
expectation, namely 13,000, and some do not. The proportion of intervals that
contain the expectation is the (simulated) confidence level. The last expression
produces this confidence level, which turns out to be equal to about 0.945.
The last expression involves a new element, the term “&”, which calls for
more explanations. Indeed, let us refer to the last expression in the code. This
expression involves the application of the function “mean”. The input to this
function contains two sequences with logical values (“TRUE” or “FALSE”), sepa-
rated by the character “&”. The character “&” corresponds to the logical “AND”
operator. This operator produces a “TRUE” if a “TRUE” appears at both sides.
Otherwise, it produces a “FALSE”. (Compare this operator to the operator
186 CHAPTER 11. CONFIDENCE INTERVALS

“OR”, that is expressed in R with the character “|”, that produces a “TRUE” if
at least one “TRUE” appears at either sides.)
In order to clarify the behavior of the terms “&” and “|” consider the fol-
lowing example:

> a <- c(TRUE, TRUE, FALSE, FALSE)

> b <- c(FALSE, TRUE, TRUE, FALSE)
> a & b
[1] FALSE TRUE FALSE FALSE
> a | b
[1] TRUE TRUE TRUE FALSE

The term “&” produces a “TRUE” only if parallel components in the sequences
“a” and “b” both obtain the value “TRUE”. On the other hand, the term “|”
produces a “TRUE” if at least one of the parallel components are “TRUE”. Ob-
serve, also, that the output of the expression that puts either of the two terms
between two sequences with logical values is a sequence of the same length (with
logical components as well).
The expression “(13000 >= LCL)” produces a logical sequence of length
100,000 with “TRUE” appearing whenever the expectation is larger than the lower
level of the confidence interval. Similarly, the expression “(13000 <= UCL)”
produces “TRUE” values whenever the expectation is less than the upper level
of the confidence interval. The expectation belongs to the confidence interval
if the value in both expressions is “TRUE”. Thus, the application of the term
“&” to these two sequences identifies the confidence intervals that contain the
expectation. The application of the function “mean” to a logical vector produces
the relative frequency of TRUE’s in the vector. In our case this corresponds to the
relative frequency of confidence intervals that contain the expectation, namely
the confidence level.
We calculated before the confidence interval [12108.47, 14305.79] for the ex-
pected price of a car. This confidence interval was obtained via the application
of the formula for the construction of confidence intervals with a 95% confi-
dence level to the variable “price” in the data frame “cars”. Casually speak-
ing, people frequently refer to such an interval as an interval that contains the
expectation with probability of 95%.
However, one should be careful when interpreting the confidence level as a
probabilistic statement. The probability computations that led to the method
for constructing confidence intervals were carried out in the context of the sam-
pling distribution. Therefore, probability should be interpreted in the context
of all data sets that could have emerged and not in the context of the given
data set. No probability is assigned to the statement “The expectation belongs
to the interval [12108.47, 14305.79]”. The probability is assigned
√ to the state-
ment “The expectation belongs to the interval X̄ ± 1.96 · S/ n”, where X̄ and
S are interpreted as random variables. Therefore the statement that the in-
terval [12108.47, 14305.79] contains the expectation with probability of 95% is
meaningless. What is meaningful is the statement that the given interval was
constructed using a procedure that produces, when applied to random samples,
intervals that contain the expectation with the assigned probability.
11.2. INTERVALS FOR MEAN AND PROPORTION 187

11.2.3 Confidence Intervals for a Proportion

The next issue is the construction of a confidence interval for the probability
of an event. Recall that a probability p of some event can be estimated by
the observed relative frequency of the event in the sample, denoted P̂ . The
estimation is associated with the Bernoulli random variable X, that obtains
the value 1 when the event occurs and the value 0 when it does not. In the
estimation problem p is the expectation of X and P̂ is the sample average of
this measurement. With this formulation we may relate the problem of the
construction of a confidence interval for p to the problem of constructing a
confidence interval for the expectation of a measurement. The latter problem
was dealt with in the previous subsection.
Specifically, the discussion regarding the steps in the construction – staring
with an application of the Central Limit Theorem in order to produce an interval
that depends on the sample average and its variance and proceeding by the
replacement of the unknown variance by its estimate – still apply and may be
taken as is. However, in the specific case we have a particular expression for the
variance of the estimate P̂ :

Var(P̂ ) = p(1 − p)/n ≈ P̂ (1 − P̂ )/n .

The tradition is to estimate this variance by using the estimator P̂ for the un-
known p instead of using the sample variance. The resulting confidence interval
of significance level 0.95 takes the form:
q
P̄ ± 1.96 · P̂ (1 − P̂ )/n .

Let us run a simulation in order to assess the confidence level of the con-
fidence interval for the probability. Assume that n = 205 and p = 0.12. The
simulation we run is very similar to the simulation of Subsection 11.2.2. In the
first stage we produce the sampling distribution of P̂ (stored in the sequence
“P.hat”) and in the second stage we compute the relative frequency in the sim-
ulation of the intervals that contain the actual value of p that was used in the
simulation:
> p <- 0.12
> n <- 205
> P.hat <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X <- rbinom(n,1,p)
+ P.hat[i] <- mean(X)
+ }
> LCL <- P.hat - 1.96*sqrt(P.hat*(1-P.hat)/n)
> UCL <- P.hat + 1.96*sqrt(P.hat*(1-P.hat)/n)
> mean((p >= LCL) & (p <= UCL))
[1] 0.95131
In this simulation we obtained that the actual confidence level is approximately
0.951, which is slightly above the
√ nominal confidence level of 0.95.
The formula X̄ ± 1.96 · S/ n that is used for a confidence interval for the
expectation and the formula P̂ ± 1.96 · {P̂ (1 − P̂ )/n}1/2 for the probability both
188 CHAPTER 11. CONFIDENCE INTERVALS

refer to a confidence intervals with confidence level of 95%. If one is interested

in a different confidence level then the width of the confidence interval should
be adjusted: a wider interval for higher confidence and a narrower interval for
smaller confidence level.
Specifically, if we examine the derivation of the formulae for confidence in-
tervals we may notice that the confidence level is used to select the number
1.96, which is the 0.975-percentile of the standard Normal distribution (1.96
=qnorm(0.975)). The selected number satisfies that the interval [−1.96, 1.96]
contains 95% of the standard Normal distribution by leaving out 2.5% on both
tails. For a different confidence level the number 1.96 should be replace by a
different number.
For example, if one is interested in a 90% confidence level then one should
use 1.645, which is the 0.95-percentile of the standard Normal distribution
(qnorm(0.95)), leaving out 5% in both √ tails. The resulting confidence inter-
val for an expectation is X̄ ± 1.645 · S/ n and the confidence interval for a
probability is P̂ ± 1.645 · {P̂ (1 − P̂ )/n}1/2 .

11.3 Intervals for Normal Measurements

In the construction of the confidence intervals in the previous section it was
assumed that the sample size is large enough. This assumption was used both
in the application of the Central Limit Theorem and in the substitution of the
unknown variance by its estimated value. For a small sample size the reasoning
that was applied before may no longer be valid. The Normal distribution may
not be a good enough approximation of the sampling distribution of the sample
average and the sample variance may differ substantially from the actual value
of the measurement variance.
In general, making inference based on small samples requires more detailed
modeling of the distribution of the measurements. In this section we will make
the assumption that the distribution of the measurements is Normal. This
assumption may not fit all scenarios. For example, the Normal distribution is a
poor model for the price of a car, which is better modeled by the Exponential
distribution. Hence, a blind application of the methods developed in this section
to variables such as the price when the sample size is small may produce dubious
outcomes and is not recommended.
When the distribution of the measurements is Normal then the method dis-
cussed in this section will produce valid confidence intervals for the expectation
of the measurement even for a small sample size. Furthermore, we will ex-
tend the methodology to enable the construction of confidence intervals for the
variance of the measurement.
Before going into the details of the methods let us present an example of
inference that involves a small sample. Consider the issue of fuel consumption.
Two variables in the “cars” data frame describe the fuel consumption. The first,
“city.mpg”, reports the number of miles per gallon when the car is driven in
urban conditions and the second, “highway.mpg”, reports the miles per gallon in
highway conditions. Typically, driving in city conditions requires more stopping
and change of speed and is less efficient in terms of fuel consumption. Hence,
one expects to obtained a reduced number of miles per gallon when driving in
urban conditions compared to the number when driving in highway conditions.
11.3. INTERVALS FOR NORMAL MEASUREMENTS 189

●
10
8
6
4
2

●
0

diesel gas

Figure 11.1: Box Plots of Differences in MPG

For each car type we calculate the difference variable that measures the
difference between the number of miles per gallon in highway conditions and
the number in urban conditions. The cars are sub-divided between cars that
run on diesel and cars that run on gas. Our concern is to estimate, for each fuel
type, the expectation of difference variable and to estimate the variance of that
variable. In particular, we are interested in the construction of a confidence
intervals for the expectation and a confidence interval for the variance.
Box plots of the difference in fuel consumption between highway and urban
conditions are presented in Figure 11.1. The box plot on the left hand side
corresponds to cars that run on diesel and the box plot on the right hand side
corresponds to cars that run on gas. Recall that 20 of the 205 car types use
diesel and the other 185 car types use gas. One may suspect that the fuel
consumption characteristics vary between the two types of fuel. Indeed, the
measurement tends to have sightly higher values for vehicles that use gas.
We conduct inference for each fuel type separately. However, since the sam-
ple size for cars that run on diesel is only 20, one may have concerns regarding
the application of methods that assume a large sample size to a sample size this
small.
190 CHAPTER 11. CONFIDENCE INTERVALS

11.3.1 Confidence Intervals for a Normal Mean

Consider the construction of a confidence interval for the expectation of a Nor-
mal measurement. In the previous section, when dealing with the construction
of a confidence interval for the expectation, we exploited the Central Limit
Theorem in order to identify
p that the distribution of the standardized sample
average (X̄ − E(X))/ Var(X)/n is, approximately, standard Normal. After-
wards, we substituted the standard deviation of the measurement by the sample
standard deviation S, which was an accurate estimator of the former due to the
magnitude sample size.
In the case where the measurements themselves are Normally distributed
one can identify the exact distribution of the standardized sample average,
with the sample
√ variance substituting the variance of the measurement: (X̄ −
E(X))/(S/ n). This specific distribution is called the Student’s t-distribution,
or simply the t-distribution.
The t-distribution is bell shaped and symmetric. Overall, it looks like the
standard Normal distribution but it has wider tails. The t-distribution is char-
acterized by a parameter called the number of degrees of freedom. In the current
setting, where we deal with the standardized sample average (with the sample
variance substituting the variance of the measurement) the number of degrees
of freedom equals the number of observations associated with the estimation of
the variance, minus 1. Hence, if the sample size is n and if the measurement is
Normally distributed then the standardized sample average (with S substitut-
ing the standard deviation of the measurement) has a t-distribution on (n − 1)
degrees of freedom. We use t(n−1) to denote this t-distribution.
The R system contains functions for the computation of the density, the
cumulative probability function and the percentiles of the t-distribution, as well
as for the simulation of a random sample from this distribution. Specifically, the
function “qt” computes the percentiles of the t-distribution. The first argument
to the function is a probability and the second argument is the number of degrees
of freedom. The output of the function is the percentile associated with the
probability of the first argument. Namely, it is a value such that the probability
that the t-distribution is below the value is equal to the probability in the first
argument.
For example, let “n” be the sample size. The output of the expression
“qt(0.975,n-1)” is the 0.975-percentile of the t-distribution on (n − 1) degrees
of freedom. By definition, 97.5% of the t-distribution are below this value and
2.5% are above it. The symmetry of the t distribution implies that 2.5% of the
distribution is below the negative of this value. The middle part of the distribu-
tion is bracketed by these two values: [−qt(0.975,n-1), qt(0.975,n-1)], and
it contains 95% of the distribution.
Summarizing the above claims in a single formula produces the statement:
  
X̄ − E(X)  X̄ − E(X) 
√ ∼ t(n−1) =⇒ P 
 √  ≤ qt(0.975,n-1) = 0.95 .
S/ n S/ n 

Notice that the equation associated with the probability is not an approximation
but an exact relation3 . Rewriting the event that is described in the probability
3 When the measurement is Normally distributed.
11.3. INTERVALS FOR NORMAL MEASUREMENTS 191

in the form of a confidence interval, produces

√
X̄ ± qt(0.975,n-1) · S/ n

as a confidence interval for the expectation of the Normal measurement with a

confidence level of 95%.
The structure of the confidence interval for the expectation of a Normal
measurement is essentially identical to the structure proposed in the previous
section. The only difference is that the number 1.96, the percentile of the stan-
dard Normal distribution, is substituted by the percentile of the t-distribution.
Consider the construction of a confidence interval for the expected difference
in fuel consumption between highway and urban driving conditions. In order to
save writing we created two new variables; a factor called “fuel” that contains
the data on the fuel type of each car, and a numerical vector called “dif.mpg”
that contains the difference between highway and city fuel consumption for each
car type:
> fuel <- cars$fuel.type
> dif.mpg <- cars$highway.mpg - cars$city.mpg
We are interested in confidence intervals based on the data stored in the
variable “dif.mpg”. One confidence interval will be associated with the level
“diesel” of the factor “fuel” and the other will be associated with the level
“gas” of the same factor.
In order to compute these confidence intervals we need to compute, for each
level of the factor “fuel”, the sample average and the sample standard deviation
of the data points of the variable “dif.mpg” that are associated with that level.
It is convenient to use the function “tapply” for this task. This function
uses three arguments. The first argument is the sequence of values over which
we want to carry out some computation. The second argument is a factor.
The third argument is a name of a function that is used for the computation.
The function “tapply” applies the function in the third argument to each sub-
collection of values of the first argument. The sub-collections are determined
by the levels of the second argument.
Sounds complex but it is straightforward enough to apply:
> tapply(dif.mpg,fuel,mean)
diesel gas
4.450000 5.648649
> tapply(dif.mpg,fuel,sd)
diesel gas
2.781045 1.433607
Sample averages are computed in the first application of the function “tapply”.
Observe that an average was computed for cars that run on diesel and an average
was computed for cars that run on gas. In both cases the average corresponds
to the difference in fuel consumption. Similarly, the standard deviations were
computed in the second application of the function. We obtain that the point
estimates of the expectation for diesel and gas cars are 4.45 and 5.648649, re-
spectively and the point estimates for the standard deviation of the variable are
2.781045 and 1.433607.
Let us compute the confidence interval for each type of fuel:
192 CHAPTER 11. CONFIDENCE INTERVALS

> x.bar <- tapply(dif.mpg,fuel,mean)

> s <- tapply(dif.mpg,fuel,sd)
> n <- c(20,185)
> x.bar - qt(0.975,n-1)*s/sqrt(n)
diesel gas
3.148431 5.440699
> x.bar + qt(0.975,n-1)*s/sqrt(n)
diesel gas
5.751569 5.856598

The objects “x.bar” and “s” contain the sample averages and sample stan-
dard deviations, respectively. Both are sequences of length two, with the first
component referring to “diesel” and the second component referring to “gas”.
The object “n” contains the two sample sizes, 20 for “diesel” and 185 for
“gas”. In the expression next to last the lower boundary for each of the
confidence intervals is computed and in the last expression the upper bound-
ary is computed. The confidence interval for the expected difference in diesel
cars is [3.148431, 5.751569]. and the confidence interval for cars using gas is
[5.440699, 5.856598].
The 0.975-percentiles of the t-distributions are computed with the expres-
sions “qt(0.025,n-1)”:

> qt(0.975,n-1)
[1] 2.093024 1.972941

The second argument of the function “qt” is a sequence with two components,
the number 19 and the number 184. Accordingly, The first position in the
output of the function is the percentile associated with 19 degrees of freedom
and the second position is the percentile associated to 184 degrees of freedom.
Compare the resulting percentiles to the 0.975-percentile of the standard
Normal distribution, which is essentially equal to 1.96. When the sample size
is small, 20 for example, the percentile of the t-distribution is noticeably larger
than the percentile of the standard Normal. However, for a larger sample size
the percentiles, more or less, coincide. It follows that for a large sample the
method proposed in Subsection 11.2.2 and the method discussed in this subsec-
tion produce essentially the same confidence intervals.

11.3.2 Confidence Intervals for a Normal Variance

The next task is to compute confidence intervals for the variance of a Normal
measurement. The main idea in the construction of a confidence interval is to
identify the distribution of a random variable associated with the parameter of
interest. A region that contains 95% of the distribution of the random variable
(or, more generally, the central part of the distribution of probability equal
to the confidence level) is identified. The confidence interval results from the
reformulation of the event associated with that region. The new formulation
puts the parameter between a lower limit and an upper limit. These lower and
the upper limits are computed from the data and they form the boundaries of
the confidence interval. Pn
We start with the sample variance, S 2 = i=1 (Xi −X̄)2 /(n−1), which serves
as a point estimator of the parameter of interest. When the measurements are
11.3. INTERVALS FOR NORMAL MEASUREMENTS 193

Normally distributed then the random variable (n − 1)S 2 /σ 2 possesses a special

distribution called the chi-square distribution. (Chi is the Greek letter χ, which
is read “Kai”.) This distribution is associated with the sum of squares of Normal
variables. It is parameterized, just like the t-distribution, by a parameter called
the number of degrees of freedom. This number is equal to (n − 1) in the
situation we discuss. The chi-square distribution on (n − 1) degrees of freedom
is denoted with the symbol χ2(n−1) .
The R system contains functions for the computation of the density, the cu-
mulative probability function and the percentiles of the chi-square distribution,
as well as for the simulation of a random sample from this distribution. Specifi-
cally, the percentiles of the chi-square distribution are computed with the aid of
the function “qchisq”. The first argument to the function is a probability and
the second argument is the number of degrees of freedom. The output of the
function is the percentile associated with the probability of the first argument.
Namely, it is a value such that the probability that the chi-square distribution
is below the value is equal to the probability in the first argument.
For example, let “n” be the sample size. The output of the expression
“qt(0.975,n-1)” is the 0.975-percentile of the chi-square distribution. By def-
inition, 97.5% of the chi-square distribution are below this value and 2.5%
are above it. Similarly, the expression “qchisq(0.025,n-1)” is the 0.025-
percentile of the chi-square distribution, with 2.5% of the distribution below
this value. Notice that between these two percentiles, namely within the inter-
val [qchisq(0.025,n-1), qchisq(0.975,n-1)], are 95% of the chi-square dis-
tribution.
We may summarize that for Normal measurements:

(n − 1)S 2 /σ 2 ∼ χ2(n−1) =⇒


P qchisq(0.025,n-1) ≤ (n − 1)S 2 /σ 2 ≤ qchisq(0.975,n-1) = 0.95 .

The chi-square distribution is not symmetric. Therefore, in order to identify the

region that contains 95% of the distribution region we have to compute both
the 0.025- and the 0.975-percentiles of the distribution.
The event associated with the 95% region is rewritten in a form that puts
the parameter σ 2 in the center:

(n − 1)S 2 /qchisq(0.975,n-1) ≤ σ 2 ≤ (n − 1)S 2 /qchisq(0.025,n-1) .



The left most and the right most expressions in this event mark the end points
of the confidence interval. The structure of the confidence interval is:

{(n − 1)/qchisq(0.975,n-1)} × S 2 , {(n − 1)/qchisq(0.025,n-1)} × S 2 .

 

Consequently, the confidence interval is obtained by the multiplication of the

estimator of the variance by a ratio between the number of degrees of freedom
(n − 1) and an appropriate percentile of the chi-square distribution. The per-
centile on the left hand side is associated with the larger probability (making
the ratio smaller) and the percentile on the right hand side is associated with
the smaller probability (making the ratio larger).
Consider, specifically, the confidence intervals for the variance of the mea-
surement “diff.mpg” for cars that run on diesel and for cars that run on gas.
Here, the size of the samples is 20 and 185, respectively:
194 CHAPTER 11. CONFIDENCE INTERVALS

> (n-1)/qchisq(0.975,n-1)
[1] 0.5783456 0.8234295
> (n-1)/qchisq(0.025,n-1)
[1] 2.133270 1.240478
The ratios that are used in the left hand side of the intervals are 0.5783456 and
0.8234295, respectively. Both ratios are less than one. On the other hand, the
ratios associated with the other end of the intervals, 2.133270 and 1.240478, are
both larger than one.
Let us compute the point estimates of the variance and the associated con-
fidence intervals. Recall that the object “s” contains the sample standard de-
viations of the difference in fuel consumption for diesel and for gas cars. The
object “n” contains the two sample sizes:
> s^2
diesel gas
7.734211 2.055229
> (n-1)*s^2/qchisq(0.975,n-1)
diesel gas
4.473047 1.692336
> (n-1)*s^2/qchisq(0.025,n-1)
diesel gas
16.499155 2.549466
The variance of the difference in fuel consumption for diesel cars is estimated
to be 7.734211 with a 95%-confidence interval of [4.473047, 16.499155] and for
cars that use gas the estimated variance is 2.055229, with a confidence interval
of [1.692336, 2.549466].
As a final example in this section let us simulate the confidence level for
a confidence interval for the expectation and for a confidence interval for the
variance of a Normal measurement. In this simulation we assume that the ex-
pectation is equal to µ = 3 and the variance is equal to σ 2 = 32 = 9. The sample
size is taken to be n = 20. We start by producing the sampling distribution of
the sample average X̄ and of the sample standard deviation S:
> mu <- 4
> sig <- 3
> n <- 20
> X.bar <- rep(0,10^5)
> S <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X <- rnorm(n,mu,sig)
+ X.bar[i] <- mean(X)
+ S[i] <- sd(X)
+ }
Consider first the confidence interval for the expectation:
> mu.LCL <- X.bar - qt(0.975,n-1)*S/sqrt(n)
> mu.UCL <- X.bar + qt(0.975,n-1)*S/sqrt(n)
> mean((mu >= mu.LCL) & (mu <= mu.UCL))
[1] 0.95033
11.4. CHOOSING THE SAMPLE SIZE 195

The nominal significance level of the confidence interval is 95%, which is prac-
tically identical to the confidence level that was computed in the simulation.
The confidence interval for the variance is obtained in a similar way. The
only difference is that we apply now different formulae for the computation of
the upper and lower confidence limits:
> var.LCL <- (n-1)*S^2/qchisq(0.975,n-1)
> var.UCL <- (n-1)*S^2/qchisq(0.025,n-1)
> mean((sig^2 >= var.LCL) & (sig^2 <= var.UCL))
[1] 0.94958
Again, we obtain that the nominal confidence level of 95% coincides with the
confidence level computed in the simulation.

11.4 Choosing the Sample Size

One of the more important contributions of Statistics to research is providing
guidelines for the design of experiments and surveys. A well planed experiment
may produce accurate enough answers to the research questions while optimizing
the use of resources. On the other hand, poorly planed trials may fail to produce
such answers or may waste valuable resources.
Unfortunately, in this book we do not cover the subject of experiment design.
Still, we would like to give a brief discussion of a narrow aspect in design: The
selection of the sample size.
An important consideration at the stage of the planning of an experiment or
a survey is the number of observations that should be collected. Indeed, having
a larger sample size is usually preferable form the statistical point of view. How-
ever, an increase in the sample size typically involves an increase in expenses.
Thereby, one would prefer to collect the minimal number of observations that
is still sufficient in order to reach a valid conclusion.
As an example, consider an opinion poll aimed at the estimation of the
proportion in the population of those that support a specific candidate that
considers running for an office. How large the sample must be in order to
assure, with high probability, that the percentage in the sample of supporters
is within 0.5% of the percentage in the population? Within 0.25%?
A natural way to address this problem is via a confidence interval for the
proportion. If the radius of the confidence interval is no more than 0.05 (or
0.025 in the other case) then with a probability equal to the confidence level
it is assured that the population relative frequency is within the given distance
from the sample proportion.
Consider a confidence level of 95%. Recall that the structure of the confi-
dence interval for the proportion is P̂ ± 1.96 · {P̂ (1 − P̂ )/n}1/2 . The radius of
the confidence interval is 1.96 · {P̂ (1 − P̂ )/n}1/2 . How large should n be in order
to guarantee that the radius is no more than 0.05?
The answer to this question depends on the magnitude of P̂ (1 − P̂ ), which
is a random quantity. Luckily, one may observe that the maximal value4 of the
quadratic function f (p) = p(1 − p) is 1/4. It follows that
√
1.96 · {P̂ (1 − P̂ )/n}1/2 ≤ 1.96 · {0.25/n}1/2 = 0.98/ n .
4 The derivative is f 0 (p) = 1 − 2p. Solving f 0 (p) = 0 produces p = 1/2 as the maximizer.

Plugging this value in the function gives 1/4 as the maximal value of the function.
196 CHAPTER 11. CONFIDENCE INTERVALS

Finally,
√ √
0.98/ n ≤ 0.05 =⇒ n ≥ 0.98/0.05 = 19.6 =⇒ n ≥ (19.6)2 = 384.16 .

The conclusion is that n should be larger than 384 in order to assure the given
radius. For example, n = 385 should be sufficient.
If the request is for an interval of radius 0.025 then the last line of reasoning
should be modified accordingly:
√ √ 0.98
0.98/ n ≤ 0.025 =⇒ n≥ = 39.2 =⇒ n ≥ (39.2)2 = 1536.64 .
0.025
Consequently, n = 1537 will do. Increasing the accuracy by 50% requires a
sample size that is 4 times larger.
More examples that involve selection of the sample size will be considered
as part of the homework.

11.5 Solved Exercises

Question 11.1. This exercise deals with an experiment that was conducted
among students. The aim of the experiment was to assess the effect of rumors
and prior reputation of the instructor on the evaluation of the instructor by her
students. The experiment was conducted by Towler and Dipboye5 . This case
study is taken from the Rice Virtual Lab in Statistics. More details on this
case study can be found in the case study “Instructor Reputation and Teacher
Ratings” that is presented in that site.
The experiment involved 49 students that were randomly assigned to one of
two conditions. Before viewing the lecture, students were give one of two “sum-
maries” of the instructor’s prior teaching evaluations. The first type of summary,
i.e. the first condition, described the lecturer as a charismatic instructor. The
second type of summary (second condition) described the lecturer as a punitive
instructor. We code the first condition as “C” and the second condition as “P”.
All subjects watched the same twenty-minute lecture given by the exact same
lecturer. Following the lecture, subjects rated the lecturer.
The outcomes are stored in the file “teacher.csv”. The file can be found
on the internet at http://pluto.huji.ac.il/~msby/StatThink/Datasets/
teacher.csv. Download this file to your computer and store it in the working
directory of R. Read the content of the file into an R data frame. Produce a
summary of the content of the data frame and answer the following questions:
1. Identify, for each variable in the file “teacher.csv”, the name and the
type of the variable (factor or numeric).
2. Estimate the expectation and the standard deviation among all students
of the rating of the teacher.
3. Estimate the expectation and the standard deviation of the rating only
for students who were given a summary that describes the teacher as
charismatic.
5 Towler, A. and Dipboye, R. L. (1998). The effect of instructor reputation and need for

cognition on student behavior (poster presented at American Psychological Society conference,

May 1998).
11.5. SOLVED EXERCISES 197

4. Construct a confidence interval of 99% confidence level for the expectation

of the rating among students who were given a summary that describes the
teacher as charismatic. (Assume the ratings have a Normal distribution.)

5. Construct a confidence interval of 90% confidence level for the variance of

the rating among students who were given a summary that describes the
teacher as charismatic. (Assume the ratings have a Normal distribution.)

Solution (to Question 11.1.1): We read the content of the file “teacher.csv”
into a data frame by the name “teacher” and produce a summary of the content
of the data frame:

> teacher <- read.csv("teacher.csv")

> summary(teacher)
condition rating
C:25 Min. :1.333
P:24 1st Qu.:2.000
Median :2.333
Mean :2.429
3rd Qu.:2.667
Max. :4.000

There are two variables: The variable “condition” is a factor with two lev-
els, “C” that codes the Charismatic condition and “P” that codes the Punitive
condition. The second variable is “rating”, which is a numeric variable.

Solution (to Question 11.1.2): The sample average for the variable “rating”
can be obtained from the summary or from the application of the function
“mean” to the variable. The standard deviation is obtained from the application
of the function “sd” to the variable:

> mean(teacher$rating)
[1] 2.428567
> sd(teacher$rating)
[1] 0.5651949

Observe that the sample average is equal to 2.428567 and the sample standard
deviation is equal to 0.5651949.

Solution (to Question 11.1.3): The sample average and standard deviation
for each sub-sample may be produced with the aid of the function “tapply”.
We apply the function in the third argument, first “mean” and then “sd” to the
variable rating, in the first argument, over each level of the factor “condition”
in the second argument:

> tapply(teacher$rating,teacher$condition,mean)
C P
2.613332 2.236104
> tapply(teacher$rating,teacher$condition,sd)
C P
0.5329833 0.5426667
198 CHAPTER 11. CONFIDENCE INTERVALS

4.0

●
3.5

●
3.0
2.5
2.0
1.5

C P

Figure 11.2: Box Plots of Ratings

Obtain that average for the condition “C” is 2.613332 and the standard deviation
is 0.5329833.
You may note that the rating given by students that were exposed to the
description of the lecturer as charismatic is higher on the average than the rating
given by students that were exposed to a less favorable description. The box
plots of the ratings for the two conditions are presented in Figure 11.2.

Solution (to Question 11.1.4): The 99% confidence interval √ for the expec-
tation is computed by the formula x̄ ± qt(0.995,n-1) · s/ n. Only 0.5% of
the t-distribution on (n − 1) degrees of freedom resides above the percentile
“qt(0.995,n-1)”. Using this percentile leaves out a total of 1% in both tails
and keeps 99% of the distribution inside the central interval.
For the students that were exposed to Condition “C”, x̄ = 2.613332, s =
0.5329833, and n = 25:

> 2.613332 - qt(0.995,24)*0.5329833/25

[1] 2.553703
> 2.613332 + qt(0.995,24)*0.5329833/25
11.5. SOLVED EXERCISES 199

[1] 2.672961
The confidence interval for the expectation is [2.553703, 2.672961].

Solution (to Question 11.1.5): The 90% confidence interval for the variance
n−1 n−1
 
is computed by the formula s2 , s2 .
qchisq(0.95,n-1) qchisq(0.05,n-1)
Observe that 5% of the chi-square distribution on (n − 1) degrees of freedom
is above the percentile “qchisq(0.95,n-1)” and 5% are below the percentile
“qchisq(0.05,n-1)”.
For the students that were exposed to Condition “C”, s = 0.5329833, and
n = 25:
> (24/qchisq(0.95,24))*0.5329833^2
[1] 0.1872224
> (24/qchisq(0.05,24))*0.5329833^2
[1] 0.4923093
The point estimate of the variance is s2 = 0.53298332 = 0.2840712. The confi-
dence interval for the variance is [0.18722243, 0.4923093].

Question 11.2. Twenty observations are used in order to construct a confidence

interval for the expectation. In one case, the construction is based on the Normal
approximation of the sample average and in the other case it is constructed under
the assumption that the observations are Normally distributed. Assume that in
reality the measurement is distributed Exponential(1/4).
1. Compute, via simulation, the actual confidence level for the first case of a
confidence interval with a nominal confidence level of 95%.
2. Compute, via simulation, the actual confidence level for the second case
of a confidence interval with a nominal confidence level of 95%.
3. Which of the two approaches would you prefer?
Solution (to Question 11.2.1): Let us produce the sampling distribution of
the sample average and the sample standard deviation for 20 observations from
the Exponential(1/4) distribution:
> lam <- 1/4
> n <- 20
> X.bar <- rep(0,10^5)
> S <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X <- rexp(n,lam)
+ X.bar[i] <- mean(X)
+ S[i] <- sd(X)
+ }
We compute the confidence level for a confidence interval with a nominal
confidence level of 95%. Observe that using the Normal approximation of the
sample average corresponds to the application of the Normal percentile in the
construction of the confidence interval.
200 CHAPTER 11. CONFIDENCE INTERVALS

> norm.LCL <- X.bar - qnorm(0.975)*S/sqrt(n)

> norm.UCL <- X.bar + qnorm(0.975)*S/sqrt(n)
> mean((4 >= norm.LCL) & (4 <= norm.UCL))
[1] 0.9047
The expectation of the measurement is equal to 4. This number belongs to the
confidence interval 90.47% of the times. Consequently, the actual confidence
level is 90.47%.

Solution (to Question 11.2.2): Using the same sampling distribution that
was produced in the solution to Question 1 we now compute the actual con-
fidence level of a confidence interval that is constructed under the assumption
that the measurement has a Normal distribution:
> t.LCL <- X.bar - qt(0.975,n-1)*S/sqrt(n)
> t.UCL <- X.bar + qt(0.975,n-1)*S/sqrt(n)
> mean((4 >= t.LCL) & (4 <= t.UCL))
[1] 0.91953
Based on the assumption we used the percentiles of the t-distribution. The
actual significance level is 91.953% ≈ 92%, still short of the nominal 95% con-
fidence level.

Solution (to Question 11.2.3): It would be preferable to use the (incorrect)

assumption that the observations have a Normal distribution than to apply the
Normal approximation to such a small sample. In the current setting the former
produced a confidence level that is closer to the nominal one. In general, using
the percentiles of the t-distribution will produce wider and more conservative
confidence intervals than those produces under the Normal approximation of
the average. To be on the safer size, one typically prefers the more conservative
confidence intervals.

Question 11.3. Insurance companies are interested in knowing the population

percent of drivers who always buckle up before riding in a car.
1. When designing a study to determine this proportion, what is the mini-
mal sample size that is required for a 99% confident that the population
proportion is accurately estimated, up to an error of 0.03?
2. Suppose that the insurance companies did conduct the study by surveying
400 drivers. They found that 320 of the drives claim to always buckle
up. Produce an 80% confidence interval for the population proportion of
drivers who claim to always buckle up.
Solution (to Question 11.2.1): The radius of the confidence interval with
99% confidence interval is bounded by
√
qnorm(0.995) · {P̂ (1 − P̂ )/n}1/2 ≤ 2.575829 · {0.25/n}1/2 = 1.287915/ n .

Consequently, the sample size n should satisfy the inequality:

√ √
1.287915/ n ≤ 0.03 =⇒ n ≥ 1.287915/0.03 = 42.9305
=⇒ n ≥ (42.9305)2 = 1843.028 .
11.6. SUMMARY 201

The smallest integer larger than the lower bound is n = 1844.

Solution (to Question 11.2.1): The 80% confidence p interval for the proba-
bility is computed by the formula p̂ ± qnorm(0.90) · p̂(1 − p̂)/n:
> n <- 400
> p.hat <- 320/400
> p.hat - qnorm(0.90)*sqrt(p.hat*(1-p.hat)/n)
[1] 0.774369
> p.hat + qnorm(0.90)*sqrt(p.hat*(1-p.hat)/n)
[1] 0.825631
We obtain a confidence interval of the form [0.774369, 0.825631].

11.6 Summary
Glossary
Confidence Interval: An interval that is most likely to contain the population
parameter.
Confidence Level: The sampling probability that random confidence intervals
contain the parameter value. The confidence level of an observed interval
indicates that it was constructed using a formula that produces, when
applied to random samples, such random intervals.
t-Distribution: A bell-shaped distribution that resembles the standard Nor-
mal distribution but has wider tails. The distribution is characterized by
a positive parameter called degrees of freedom.
Chi-Square Distribution: A distribution associated with the sum of squares
of Normal random variable. The distribution obtains only positive values
and it is not symmetric. The distribution is characterized by a positive
parameter called degrees of freedom.

Discuss in the forum

When large samples are at hand one may make fewer a-priori assumptions re-
garding the exact form of the distribution of the measurement. General limit
theorems, such as the Central Limit Theorem, may be used in order to estab-
lish the validity of the inference under general conditions. On the other hand,
for small sample sizes one must make strong assumptions with respect to the
distribution of the observations in order to justify the validity of the procedure.
It may be claimed that making statistical inferences when the sample size
is small is worthless. How can one trust conclusions that depend on assump-
tions regarding the distribution of the observations, assumptions that cannot
be verified? What is your opinion?
For illustration consider the construction of a confidence interval. Confi-
dence interval for the expectation is implemented with a specific formula. The
significance level of the interval is provable when the sample size is large or when
the sample size is small but the observations have a Normal distribution. If the
202 CHAPTER 11. CONFIDENCE INTERVALS

sample size is small and the observations have a distribution different from the
Normal then the nominal significance level may not coincide with the actual
significance level.

Formulas for Confidence Intervals, 95% Confidence Level:

√
• Expectation: x̄ ± qnorm(0.975) · s/ n.
√
• Probability: p̄ ± qnorm(0.975) · p̂(1 − p̂)/ n.
√
• Normal Expectation: x̄ ± qt(0.975,n-1) · s/ n.
n−1 n−1
 
• Normal Expectation: s2 , s2 .
qchisq(0.975,n-1) qchisq(0.025,n-1)
Chapter 12

Testing Hypothesis

12.1 Student Learning Objectives

Hypothesis testing emerges as a crucial component in decision making where
one of two competing options needs to be selected. Statistical hypothesis testing
provides formal guidelines for making such a selection. This chapter deals with
the formulation of statistical hypothesis testing and describes the associated
decision rules. Specifically, we consider hypothesis testing in the context of the
expectation of a measurement and in the context of the probability of an event.
In subsequent chapters we deal with hypothesis testing in the context of other
parameters as well. By the end of this chapter, the student should be able to:

• Formulate statistical hypothesis for testing.

• Test, based on a sample, hypotheses regarding the expectation of the

measurement and the probability of an event.

• Identify the limitations of statistical hypothesis testing and the danger of

misinterpretation of the test’s conclusions.

12.2 The Theory of Hypothesis Testing

Statistical inference is used in order to detect and characterize meaningful phe-
nomena that may be hidden in an environment contaminated by random noise.
Hypothesis testing is an important step, typically the first, in the process of
making inferences. In this step one tries to answer the question: “Is there a
phenomena at all?”. The basic approach is to determine whether the observed
data can or cannot be reasonably explained by a model of randomness that does
not involve the phenomena.
In this section we introduce the structure and characteristics of statistical
hypothesis testing. We start with an informal application of a statistical test and
proceed with formal definitions. In the next section we discuss in more detail
the testing of hypotheses on the expectation of a measurement and the testing
of hypotheses on the probability of an event. More examples are considered in
subsequent chapters.

203
204 CHAPTER 12. TESTING HYPOTHESIS

12.2.1 An Example of Hypothesis Testing

The variable “price” in the file “cars.csv” contains data on the prices of
different types of cars that were sold in the United States during 1985. The
average price of a car back then — the average of the variable “price” — was
$13,207. One may be interested in the question: Do Americans pay today for
cars a different price than what they used to pay in the 80’s? Has the price of
cars changed significantly since 1985?
The average price of a car in the United States in 2009 was $27,9581 . Clearly,
this figure is higher than $13,207. However, in order to produce a fair answer
to the question we have to take into account that, due to inflation, the prices
of all products went up during these years. A more meaningful comparison will
involve the current prices of cars in terms of 1985 Dollars. Indeed, if we take
into account inflation then we get that, on the average, the cost of today’s cars
corresponds to an average price of $13,662 in 1985 values2 . This price is still
higher than the prices in the 1985 but not as much. The question we are asking
is: “Is the difference between $13,207 and $13,662 significant or is it not so?”.
In order to give a statistical answer to this question we carry out a statistical
test. The specific test is conducted with the aid of the function “t.test”. Later
we will discuss in more details some of the arguments that may be used in
this function. Currently, we simply apply it to the data stored in the variable
“price” to test that the expected price is different than the $13,662, the average
price of a car in 2009, adjusted for inflation:

> cars <- read.csv("cars.csv")

> t.test(cars$price,mu=13662)

One Sample t-test

data: cars$price
t = -0.8115, df = 200, p-value = 0.4181
alternative hypothesis: true mean is not equal to 13662
95 percent confidence interval:
12101.80 14312.46
sample estimates:
mean of x
13207.13

The data in the file “cars.csv” is read into a data frame that is given the
name “cars”. Afterwards, the data on prices of car types in 1985 is entered
as the first argument to the function “t.test”. The other argument is the
expected value that we want to test, the current average price of cars, given in
terms of 1985 Dollar value. The output of the function is reported under the
title: “One Sample t-test”.
Let us read the report from the bottom up. The bottom part of the report
describes the confidence interval and the point estimate of the expected price
of a car in 1985, based on the given data. Indeed, the last line reports the
1 Source: “http://wiki.answers.com/Q/Average_price_of_a_car_in_2009”.
2 Source: “http://www.westegg.com/inflation/”. The interpretation of adjusting prices
to inflation is that our comparison will correspond to changes in the price of cars, relative to
other items that enter into the computation of the Consumer Price Index.
12.2. THE THEORY OF HYPOTHESIS TESTING 205

sample average of the price, which is equal to 13,207.13. This number, the
average of the 201 non-missing values of the variable “price”, serves as the
estimate of the expected price of a car in 1985. The 95% confidence interval of
the expectation, the interval [12101.80, 14312.46], is presented on the 4th line
from the bottom. This is the confidence interval for the expectation that was
computed in Subsection 11.2.13 .
The information relevant to conducting the statistical test itself is given
in the upper part of the report. Specifically, it is reported that the data in
“cars$price” is used in order to carry out the test. Based on this data a test
statistic is computed and obtains the value of “t = -0.8115”. This statistic
is associated with the t-distribution with “df = 200” degrees of freedom. The
last quantity that is being reported is denoted the p-value and it obtains the
value “p-value = 0.4181”. The test may be carried out with the aid of the
value of the t statistic or, more directly, using the p-value. Currently we will
use the p-value.
The test itself examines the hypothesis that the expected price of a car
in 1985 was equal to $13,662, the average price of a car in 2009, given in 1985
values. This hypothesis is called the null hypothesis. The alternative hypothesis
is that the expected price of a car in 1985 was not equal to that figure. The
specification of the alternative hypothesis is reported on the third line of the
output of the function “t.test”.
One may decide between the two hypothesis on the basis of the size of the
p-value. The rule of thumb is to reject the null hypothesis, and thus accept the
alternative hypothesis, if the p-value is less than 0.05. In the current example the
p-value is equal 0.4181 and is larger than 0.05. Consequently, we may conclude
that the expected price of a car in 1985 was not significantly different than the
current price of a car.
In the rest of this section we give a more rigorous explanation of the theory
and practice of statistical hypothesis testing.

12.2.2 The Structure of a Statistical Test of Hypotheses

The initial step in statistical inference in general, and in statistical hypothesis
testing in particular, is the formulation of the statistical model and the identifi-
cation of the parameter/s that should be investigated. In the current situation
the statistical model may correspond to the assumption that the data in the
variable “price” are an instant of a random sample (of size n = 201). The
parameter that we want to investigate is the expectation of the measurement
that produced the sample. The variance of the measurement is also relevant for
the investigation.
After the statistical model has been set, one may split the process of testing
a statistical hypothesis into three steps: (i) formulation of the hypotheses, (ii)
specification of the test, and (iii) reaching the final conclusion. The first two
steps are carried out on the basis of the probabilistic characteristics of the
3 As a matter of fact, the confidence interval computed in Subsection 11.2.1 is [12108.47,

14305.79], which is not identical to the confidence that appears in the report. The reason for
the discrepancy is that we used the 0.975-percentile of the Normal distribution, 1.96, whereas
the confidence interval computed here uses the 0.975-percentile of the t-distribution on 201-
1=200 degrees of freedom. The latter is equal to 1.971896. Nonetheless, for all practical
purposes, the two confidence intervals are the same.
206 CHAPTER 12. TESTING HYPOTHESIS

statistical model and in the context of the sampling distribution. In principal,

the first two steps may be conducted in the planning stage prior to the collection
of the observations. Only the third step involves the actual data. In the example
that was considered in the previous subsection the third step was applied to the
data in the variable “price” using the function “t.test”.

(i) Formulating the hypotheses: A statistical model involves a parameter

that is the target of the investigation. In principle, this parameter may obtain
any value within a range of possible values. The formulation of the hypothesis
corresponds to splitting the range of values into two sub-collections: a sub-
collection that emerges in response to the presence of the phenomena and a
sub-collection that emerges in response to the situation when the phenomena is
absent. The sub-collection of parameter values where the phenomena is absent
is called the null hypothesis and is marked as “H0 ”. The other sub-collection,
the one reflecting the presence of the phenomena, is denoted the alternative
hypothesis and is marked “H1 ”.
For example, consider the price of cars. Assume that the phenomena one
wishes to investigate is the change in the relative price of a car in the 80’s as
compared to prices today. The parameter of interest is the expected price of
cars back then, which we denote by E(X). The formulation of the statement
that the expected price of cars has changed is “E(X) 6= 13, 662”. This statement
corresponds to the presence of a phenomena, to a change, and is customarily
defined as the alternative hypothesis. On the other hand, the situation “E(X) =
13, 662” corresponds to not having any change in the price of cars. Hence, this
situation corresponds to the absence of the phenomena and is denoted the null
hypothesis. In summary, in order to investigate the change in the relative price
of cars we my consider the null hypothesis “H0 : E(X) = 13, 662” and test it
against the alternative hypothesis “H1 : E(X) 6= 13, 662”.
A variation in the formulation of the phenomena can change the definition
of the null and alternative hypotheses. For example, if the intention is to in-
vestigate the rise in the price of cars then the phenomena will correspond to
the expected price in 1985 being less than $13,662. Accordingly, the alternative
hypothesis should be defined as H1 : E(X) < 13, 662, with the null hypoth-
esis defined as H0 : E(X) ≥ 13, 662. Observe that in this case an expected
price larger than $13,662 relates to the phenomena of rising (relative) prices not
taking place.
On the other hand, if one would wants to investigate a decrease in the price
then one should define the alternative hypothesis to be H1 : E(X) > 13, 662,
with the null hypothesis being H0 : E(X) ≤ 13, 662.
The type of alternative that was considered in the example, H1 : E(X) 6=
13, 622 is called a two-sided alternative. The other two types of alternative
hypotheses that were considered thereafter, H1 : E(X) < 13, 662 and H1 :
E(X) > 13, 662, are both called one-sided alternatives.
In summary, the formulation of the hypothesis is a reflection of the phe-
nomena one wishes to examine. The setting associated with the presence of the
phenomena is denoted the alternative hypothesis and the complimentary set-
ting, the setting where the phenomena is absent, is denoted the null hypothesis.

(ii) Specifying the test: The second step in hypothesis testing involves the
selection of the decision rule, i.e. the statistical test, to be used in order to decide
12.2. THE THEORY OF HYPOTHESIS TESTING 207

between the two hypotheses. The decision rule is composed of a statistic and
a subset of values of the statistic that correspond to the rejection of the null
hypothesis. The statistic is called the test statistic and the subset of values is
called the rejection region. The decision is to reject the null hypothesis (and
consequently choose the alternative hypothesis) if the test statistic falls in the
rejection region. Otherwise, if the test statistic does not fall in the rejection
region then the null hypothesis is selected.
Return to the example in which we test between H0 : E(X) = 13, 662 and
H1 : E(X) 6= 13, 662. One may compute the statistic:
X̄ − 13, 662
T = √ ,
S/ n
where X̄ is the sample average (of the variable “price”), S is the sample stan-
dard deviation, and n is the sample size (n = 201 in the current example).
The sample average X̄ is an estimator of a expected price of the car. In prin-
ciple, the statistic T measures the discrepancy between the estimated value of
the expectation (X̄) and the expected value under the null hypothesis (E(X) =
13, 662). This discrepancy is measured in units of the (estimated) standard
deviation of the sample average4 .
If the null hypothesis H0 : E(X) = 13, 662 is true then the sampling distri-
bution of the sample average X̄ should be concentrated about the value 13,662.
Values of the sample average much larger or much smaller than this value may
serve as evidence against the null hypothesis.
In reflection, if the null hypothesis holds true then the values of the sampling
distribution of the statistic T should tend to be in the vicinity of 0. Values with
a relative small absolute value are consistent with the null hypothesis. On
the other hand, extremely positive or extremely negative values of the statistic
indicate that the null hypothesis is probably false.
It is natural to set a value c and to reject the null hypothesis whenever the
absolute value of the statistic T is larger than c. The resulting rejection region
is of the form {|T | > c}. The rule of thumb, again, is to take threshold c to be
equal the 0.975-percentile of the t-distribution on n−1 degrees of freedom, where
n is the sample size. In the current example, the sample size is n = 201 and the
percentile of the t-distribution is qt(0.975,200) = 1.971896. Consequently,
the subset {|T | > 1.971896} is the rejection region of the test.
A change in the hypotheses that are being test may lead to a change in
the test statistic and/or the rejection region. For example, for testing H0 :
E(X) ≥ 13, 662 versus H1 : E(X) < 13, 662 one may still use the same test
statistic T as before. However, only very negative values of the statistic are
inconsistent with the null hypothesis. It turns out that the rejection region in
this case is of the form {T < −1.652508}, where qt(0.05,200) = -1.652508
is the 0.05-percentile of the t-distribution on 200 degrees of freedom. On the
other hand, the rejection region for testing between H0 : E(X) ≤ 13, 662 and
H1 : E(X) > 13, 662 is {T > 1.652508}. In this case, qt(0.95,200) = 1.652508
is the 0.95-percentile of the t-distribution on 200 degrees of freedom.
4 If
the pvariance of the measurement Var(X) was known one could have use Z = (X̄ −
−13, 662)/ VarX/n as a test statistic. This statistic corresponds to the discrepancy of the
sample average from the null expectation in units of its standard deviation, i.e. the z-value of
the sample average. Since the variance of the observation is unknown, we use an estimator of
the variance (S 2 ) instead.
208 CHAPTER 12. TESTING HYPOTHESIS

Selecting the test statistic and deciding what rejection region to use specifies
the statistical test and completes the second step.

(iii) Reaching a conclusion: After the stage is set, all that is left is to apply
the test to the observed data. This is done by computing the observed value of
the test statistic and checking whether or not the observed value belongs to the
rejection region. If it does belong to the rejection region then the decision is
to reject the null hypothesis. Otherwise, if the statistic does not belong to the
rejection region, then the decision is to accept the null hypothesis.
Return to the example of testing the price of car types. The observed value of
the T statistic is part of the output of the application of the function “t.test”
to the data. The value is “t = -0.8115”. As an exercise, let us recompute
directly from the data the value of the T statistic:

> x.bar <- mean(cars$price,na.rm=TRUE)

> x.bar
[1] 13207.13
> s <- sd(cars$price,na.rm=TRUE)
> s
[1] 7947.066

The observed value of the sample average is x̄ = 13207.13 and the observed
value of the sample standard deviation is s = 7947.066. The sample size (due
to having 4 missing values) is n = 201. The formula for √ the computation of
the test statistic in this example is t = [x̄ − 13, 662]/[s/ n]. Plugging in this
formula the sample size and the computed values of the sample average and
standard deviation produces:

> (x.bar - 13662)/(s/sqrt(201))

[1] -0.8114824

This value, after rounding up, is equal to the value “t = -0.8115” that is
reported in the output of the function “t.test”.
The critical threshold for the absolute value of the T statistic on 201−1 = 200
degrees of freedom is qt(0.975,200) = 1.971896. Since the absolute observed
value (|t| = 0.8114824) is less then the threshold we get that the value of the
statistic does not belong to the rejection region (which is composed of absolute
values larger than the threshold). Consequently, we accept the null hypothesis.
This null hypothesis declares that the expected price of a car was equal to
the current expected price (after adjusting for the change in Consumer Price
Index)5 .

12.2.3 Error Types and Error Probabilities

The T statistic was proposed for testing a change in the price of a car. This
statistic measures the discrepancy between the sample average price of a car and
5 Previously, we carried out the same test using the p-value. The computed p-value in this

example is 0.4181. The null hypothesis was accepted since this value is larger than 0.05. As a
matter of fact, the test that uses the T statistic as a test statistic and reject the null hypothesis
for absolute values larger than qt(0.975,n-1) is equivalent to the test that uses the p-value
and rejects the null hypothesis for p-values less than 0.05. Below we discuss the computation
of the p-value.
12.2. THE THEORY OF HYPOTHESIS TESTING 209

the expected value of the sample average, where the expectation is computed
under the null hypothesis. The structure of the rejection region of the test is
{|T | > c}, where c is an appropriate threshold. In the current example the value
of the threshold c was set to be equal to qt(0.975,200) = 1.971896. In general,
the specification of the threshold c depends on the error probabilities that are
associated with the test. In this section we describe these error probabilities.
The process of making decisions may involve errors. In the case of hypothesis
testing one may specify two types of error. On the one hand, the case may be
that the null hypothesis is correct (in the example, E(X) = 13, 662). However,
the data is such that the null hypothesis is rejected (here, |T | > 1.971896). This
error is called a Type I error.
A different type of error occurs when the alternative hypothesis holds (E(X) 6=
13, 662) but the null hypothesis is not rejected (|T | ≤ 1.971896). This other type
of error is called Type II error. A summary of the types of errors can be found
in Table 12.1:
H0 : E(X) = 13, 662 H1 : E(X) 6= 13, 662
Accept H0 : |T | ≤ 1.971896 X Type II Error
Reject H0 : |T | > 1.971896 Type I Error X

Table 12.1: Error Types

In statistical testing of hypothesis the two types of error are not treated
symmetrically. Rather, making a Type I error is considered more sever than
making a Type II error. Consequently, the test’s decision rule is designed so
as to assure an acceptable probability of making a Type I error. Reducing the
probability of a Type II error is desirable, but is of secondary importance.
Indeed, in the example that deals with the price of car types the thresh-
old was set as high as qt(0.975,200) = 1.971896 in order to reject the null
hypothesis. It is not sufficient that the sample average is not equal to 13,662
(corresponding to a threshold of 0), but it has to be significantly different from
the expectation under the null hypothesis, the distance between the sample av-
erage and the null expectation should be relatively large, in order to exclude H0
as an option.
The significance level of the evidence for rejecting the null hypothesis is
based on the probability of the Type I error. The probabilities associated with
the different types of error are presented in Table 12.2:

H0 : E(X) = 13, 662 H1 : E(X) 6= 13, 662

P(|T | ≤ c) Prob. of Type II Error
P(|T | > c) Significance Level Statistical Power

Table 12.2: Error Probabilities

Observe that the probability of a Type I error is called the significance level.
The significance level is set at some pre-specified level such as 5% or 1%, with
5% being the most widely used level. In particular, setting the threshold in the
example to be equal to qt(0.975,200) = 1.971896 produces a test with a 5%
significance level.
This lack of symmetry between the two hypothesis proposes another interpre-
tation of the difference between the hypothesis. According to this interpretation
210 CHAPTER 12. TESTING HYPOTHESIS

the null hypothesis is the one in which the cost of making an error is greater.
Thus, when one separates the collection of parameter values into two subsets
then the subset that is associated with a more sever error is designated as the
null hypothesis and the other subset becomes the alternative.
For example, a new drug must pass a sequence of clinical trials before it
is approved for distribution. In these trials one may want to test whether the
new drug produces beneficial effect in comparison to the current treatment.
Naturally, the null hypothesis in this case would be that the new drug is no
better than the current treatment and the alternative hypothesis would be that
it is better. Only if the clinical trials demonstrates a significant beneficiary
effect of the new drug would it be released for marketing.
In scientific research, in general, the currently accepted theory, the conser-
vative explanation, is designated as the null hypothesis. A claim for novelly in
the form of an alternative explanation requires strong evidence in order for it to
be accepted and be favored over the traditional explanation. Hence, the novel
explanation is designated as the alternative hypothesis. It replaces the current
theory only if the empirical data clearly supports its. The test statistic is a
summary of the empirical data. The rejection region corresponds to values that
are unlikely to be observed according to the current theory. Obtaining a value
in the rejection region is an indication that the current theory is probably not
adequate and should be replaced by an explanation that is more consistent with
the empirical evidence.
The second type of error probability in Table 12.2 is the probability of a
Type II error. Instead of dealing directly with this probability the tradition is
to consider the complementary probability that corresponds to the probability
of not making a Type II error. This complementary probability is called the
statistical power :
Statistical Power = 1 − Probability of Type II Error
The statistical power is the probability of rejecting the null hypothesis when the
state of nature is the alternative hypothesis. (In comparison, the significance
level is the probability of rejecting the null hypothesis when the state of nature is
the null hypothesis.) When comparing two decision rules for testing hypothesis,
both having the same significance level, the one that possesses a higher statistical
power should be favored.

12.2.4 p-Values
The p-value is another test statistic. It is associated with a specific test statistic
and a structure of the rejection region. The p-value is equal to the significance
level of the test in which the observed value of the statistic serves as the thresh-
old. In the current example, where the T is the underlying test statistic and
the structure of the rejection region is of the form {|T | > c} then the p-value is
equal to the probability of rejecting the null hypothesis in the case where the
threshold c is equal to the observed absolute value of the T statistic. In other
words:
p-value = P(|T | > |t|) = P(|T | > | − 0.8114824|) = P(|T | > 0.8114824) ,
where t = −0.8114824 is the observed value of the T statistic and the compu-
tation of the probability is conducted under the null hypothesis.
12.3. TESTING HYPOTHESIS ON EXPECTATION 211

Specifically, under the null hypothesis H0 : E(X)

√ = 13, 662 we get that the
distribution of the statistic T = [X̄ − 13, 662]/[S/ n] is the t-distribution on
n − 1 = 200 degrees of freedom. The probability of the event {|T | > 0.8114824}
corresponds to the sum of the probabilities of both tails of the distribution. By
the symmetry of the t-distribution this equals twice the probability of the upper
tail:

P(|T | > 0.8114824) = 2 · P(T > 0.8114824) = 2 · [1 − P(|T | ≤ 0.8114824)] .

When we compute this probability in R we get:

> 2*(1-pt(0.8114824,200))
[1] 0.4180534

This probability is equal, after rounding up, to the probability “p-value =

0.4181” that is reported in the output of the function “t.test”.
The p-value is a function of the data. In the particular data set the computed
value of the T statistic was -0.8114824. For a different data set the evaluation
of the statistic would have produced a different value. As a result, the thresh-
old that would have been used in the computation would have been different,
thereby changing the numerical value of the p-value. Being a function of the
data, we conclude that the p-value is a statistic.
The p-value is used as a test statistic by comparing its value to the pre-
defined significance level. If the significance level is 1% then the null hypothesis
is rejected for p-values less that 0.01. Likewise, if the significance level is set at
the 5% level then the null hypothesis is rejected for p-values less than 0.05.
The statistical test that is based directly on the T statistic and the statistical
test that is based on the p-value are equivalent to each other. The one rejects
the null hypothesis if, and only if, the other does so. The advantage of using
the p-value as the test statistic is that no further probabilistic computations
are required. The p-value is compared directly to the significance level we seek.
For the test that examines the T statistic we still need to identify the threshold
associated with the given significance level.
In the next 2 sections we extend the discussion of the t-test and give further
examples to the use of the function “t.test”. We also deal with tests on
probabilities of events and introduce the function “prop.test” for conducting
such tests.

12.3 Testing Hypothesis on Expectation

Let us consider the variable “dif.mpg” that contains the difference in fuel con-
sumption between highway and city conditions. This variable was considered in
Chapter 11. Examine the distribution of this variable:

> dif.mpg <- cars$highway.mpg - cars$city.mpg

> summary(dif.mpg)
Min. 1st Qu. Median Mean 3rd Qu. Max.
0.000 5.000 6.000 5.532 7.000 11.000
> plot(table(dif.mpg))
212 CHAPTER 12. TESTING HYPOTHESIS

60
50
40
table(dif.mpg)

30
20
10
0

0 1 2 3 4 5 6 7 8 9 11

dif.mpg

Figure 12.1: The Difference Between Highway and City MPG

In the first expression we created the variable “dif.mpg” that contains the
difference in miles-per-gallon. The difference is computed for each car type be-
tween highway driving conditions and urban driving condition. The summary
of this variable is produced in the second expression. Observe that the values of
the variable range between 0 and 11, with 50% of the distribution concentrated
between 5 and 7. The median is 6 and the mean is 5.532. The last expres-
sion produces the bar plot of the distribution. This bar plot is presented in
Figure 12.1. It turns out that the variable “dif.mpg” obtains integer values.
In this section we test hypotheses regarding the expected difference in fuel
consumption between highway and city conditions.
Energy is required in order to move cars. For heavier cars more energy is
required. Consequently, one may conjecture that milage per gallon for heavier
cars is less than the milage per gallon for lighter cars.
The relation between the weight of the car and the difference between the
milage-per-gallon in highway and city driving conditions is less clear. On the one
hand, urban traffic involves frequent changes in speed in comparison to highway
conditions. One may presume that this change in speed is a cause for reduced
efficiency in fuel consumption. If this is the case then one may predict that
12.3. TESTING HYPOTHESIS ON EXPECTATION 213

heavier cars, which require more energy for acceleration, will be associated with
a bigger difference between highway and city driving conditions in comparison
to lighter cars.
One the other hand, heavier cars do less miles per gallon overall. The differ-
ence between two smaller numbers (the milage per gallon in highway and in city
conditions for heavier cars) may tend to be smaller than the difference between
two larger numbers (the milage per gallon in highway and in city conditions for
lighter cars). If this is the case then one may predict that heavier cars will be
associated with a smaller difference between highway and city driving conditions
in comparison to lighter cars.
The average difference between highway and city conditions is approximately
5.53 for all cars. Divide the cars into to two groups of equal size: One group
is composed of the heavier cars and the other group is composed of the lighter
cars. We will examine the relation between the weight of the car and difference
in miles per gallon between the two driving conditions by testing hypotheses
separately for each weight group6 . For each such group we start by testing
the two-sided hypothesis H1 : E(X) 6= 5.53, where X is the difference between
highway and city miles-per-gallon in cars that belong to the given weight group.
After carrying the test for the two-sided alternative we will discuss results of
the application of tests for one-sided alternatives.
We start by the definition of the weight groups. The variable “curb.weight”
measures the weight of the cars in the data frame “cars”. Let us examine the
summary of the content of this variable:

> summary(cars$curb.weight)
Min. 1st Qu. Median Mean 3rd Qu. Max.
1488 2145 2414 2556 2935 4066

Half of the cars in the data frame weigh less than 2,414 lb and half the cars weigh
more. The average weight of a car is 2,556 lb. Let us take 2,414 as a threshold
and denote cars below this weight as “light” and cars above this threshold as
“heavy”:

> heavy <- cars$curb.weight > 2414

> table(heavy)
heavy
FALSE TRUE
103 102

The variable “heavy” indicates for each car type whether its weight is above or
below the threshold weight of 2,414 lb. The variable is composed of a sequence
with as many components as the number of observations in the data frame
“cars” (n = 205). Each component is a logical value: “TRUE” if the car is heavier
than the threshold and “FALSE” if it is not. When we apply the function “table”
to this sequence we get that 102 of the cars are heavier than the threshold and
103 are not so.
6 In the next chapters we will consider a more direct ways for comparing the effect of one

variable (curb.weight in this example) on the distribution of another variable (dif.mpg in this
example). Here, instead, we investigate the effect indirectly by the investigation of hypotheses
on the expectation of the variable dif.mpg separately for heavier cars and for lighter cars.
214 CHAPTER 12. TESTING HYPOTHESIS

We would like to apply the t-test first to the subset of all cars with weight
above 2,414 lb (cars that are associated with the value “TRUE” in the vari-
able “heavy”), and then to all cars with weights not exceeding the threshold
(cars associated with value “FALSE”). In the past we showed that one may ad-
dress components of a sequence using its position in the sequence7 . Here we
demonstrate an alternative approach for addressing specific locations by using
a sequence with logical components.
In order to illustrate this second approach consider the two sequences:
> w <- c(5,3,4,6,2,9)
> d <- c(13,22,0,12,6,20)
Say we want to select the components of the sequence “d” in all the locations
where the components of the sequence “w” obtain values larger than 5. Consider
the code:
> w > 5
[1] FALSE FALSE FALSE TRUE FALSE TRUE
> d[w > 5]
[1] 12 20
The expression “w > 5” is a sequence of logical components, with the value
“TRUE” at the positions where “w” is above the threshold and the value “FALSE”
at the positions where “w” is below the threshold. We may use the sequence
with logical components as an index to the sequence of the same length “d”.
The relevant expression is “d[w > 5]”. The output of this expression is the
sub-sequence of elements from “d” that are associated with the “TRUE” values
of the logical sequence. Indeed, “TRUE” values are present at the 4th and the
6th positions of the logical sequence. Consequently, the output of the expression
“d[w > 5]” contains the 4th and the 6th components of the sequence “d”.
The operator “!”, when applied to a logical value, reverses the value. A
“TRUE” becomes “FALSE” and a “FALSE” becomes “TRUE”. Consider the code:
> !(w > 5)
[1] TRUE TRUE TRUE FALSE TRUE FALSE
> d[!(w > 5)]
[1] 13 22 0 6
Observe that the sequence “!(w > 5)” obtains a value of “TRUE” at the positions
where “w” is less or equal to 5. Consequently, the output of the expression
“d[!(w > 5)]” are all the values of “d” that are associated with components
of “w” that are less or equal to 5.
The variable “dif.mpg” contains data on the difference in miles-per-gallon
between highway and city driving conditions for all the car types. The se-
quence “heavy” identifies the car types with curb weight above the threshold
of 2,414 lb. The components of this sequence are logical with the value “TRUE”
at positions associated with the heavier car types and the “FALSE” at positions
associated with the lighter car types. Observe that the output of the expres-
sion “dif.mpg[heavy]” is the subsequence of differences in miles-per-gallon for
7 For example, in Question 9.1 we referred to the first 29 observations of the sequence

“change” using the expression “change[1:29]” and to the last 21 observations using the
expression “change[30:50]”.
12.3. TESTING HYPOTHESIS ON EXPECTATION 215

the cars with curb weight above the given threshold. We apply the function
“t.test” to this expression in order to conduct the t-test on the expectation of
the variable “dif.mpg” for the heavier cars:

> t.test(dif.mpg[heavy],mu=5.53)

One Sample t-test

data: dif.mpg[heavy]
t = -1.5385, df = 101, p-value = 0.1270
alternative hypothesis: true mean is not equal to 5.53
95 percent confidence interval:
4.900198 5.609606
sample estimates:
mean of x
5.254902

The target population are the heavier car types. Notice that we test the null hy-
pothesis that expected difference among he heavier cars is equal to 5.53 against
the alternative hypothesis that the expected difference among heavier cars is not
equal to 5.53. The null hypothesis is not rejected at the 5% significance level
since the p-value, which is equal to 0.1735, is larger than 0.05. Consequently,
based on the data at hand, we cannot conclude that the expected difference
in miles-per-gallon for heavier cars is significantly different than the average
difference for all cars.
Observe also that the estimate of the expectation, the sample mean, is equal
to 5.254902, with a confidence interval of the form [4.900198, 5.609606].
Next, let us apply the same test to the lighter cars. The expression “dif.pmg[!heavy]”
produces the subsequence of differences in miles-per-gallon for the cars with curb
weight below the given threshold. The application of the function “t.test” to
this subsequence gives:

> t.test(dif.mpg[!heavy],mu=5.53)

One Sample t-test

data: dif.mpg[!heavy]
t = 1.9692, df = 102, p-value = 0.05164
alternative hypothesis: true mean is not equal to 5.53
95 percent confidence interval:
5.528002 6.083649
sample estimates:
mean of x
5.805825

Again, the null hypothesis is not rejected at the 5% significance level since a p-
value of 0.05164 is still larger than 0.05. However, unlike the case for heavier cars
where the p-value was undeniably larger than the threshold. In this example it
is much closer to the threshold of 0.05. Consequently, we may almost conclude
that the expected difference in miles-per-gallon for lighter cars is significantly
different than the average difference for all car.
216 CHAPTER 12. TESTING HYPOTHESIS

Why did we not reject the null hypothesis for the heavier cars but almost did
so for the lighter cars? Both tests are based on the T statistic, which measures
the ratio between the deviation of the sample average from its expectation under
the null, divided by the estimate of the standard deviation of the sample average.
The value of this statistic is “t = -1.5385” for heavier cars and it is “t =
1.9692” for lighter cars, an absolute value of about 25% higher.
The deviation of the sample average for the heavier cars and the expectation
under the null is 5.254902−5.53 = −0.275098. On the other hand, the deviation
of the sample average for the lighter cars and the expectation under the null is
5.805825 − 5.53 = 0.275825. The two deviations are practically equal to each
other in the absolute value.
√
The estimator of the standard deviation of the sample average is S/ n,
where S is the sample standard deviation and n is the sample size. The sample
sizes, 103 for lighter cars and 102 for heavier cars, are almost equal. Therefore,
the reason for the difference in the values of the T statistics for both weight
groups must be differences in the sample standard deviations. Indeed, when
we compute the sample standard deviation for lighter and heavier cars8 we get
that the standard deviation for lighter cars (1.421531) is much smaller than the
standard deviation for heavier cars (1.805856):

> tapply(dif.mpg,heavy,sd)
FALSE TRUE
1.421531 1.805856

The important lesson to learn from this exercise is that simple minded notion
of significance and statistical significance are not the same. A simple minded
assessment of the discrepancy from the null hypothesis will put the evidence
from the data on lighter cars and the evidence from the data on heavier cars on
the same level. In both cases the estimated value of the expectation is the same
distance away from the null value.
However, statistical assessment conducts the analysis in the context of the
sampling distribution. The deviation of the sample average from the expectation
is compared to the standard deviation of the sample average. Consequently, in
statistical testing of hypothesis a smaller deviation of the sample average from
the expectation under the null may be more significant than a larger one if the
sampling variability of the former is much smaller than the sampling variability
of the later.
Let us proceed with the demonstration of the application of the t-test by
the testing of one-sided alternatives in the context of the lighter cars. One may
test the one-sided alternative H1 : E(X) > 5.53 that the expected value of the
difference in miles-per-gallon among cars with curb weight no more than 2,414
lb is greater than 5.53 by the application of the function “t.test” to the data
on lighter cars. This data is the output of the expression “dif.mpg[!heavy]”.
As before, we specify the null value of the expectation by the introduction of the
8 The function “tapply” applies the function that is given as its third argument (the func-

tion “sd” in this case) to each subset of values of the sequence that is given as its first argument
(the sequence “dif.mpg” in the current application). The subsets are determined by the levels
of the second arguments (the sequence “heavy” in this case). The output is the sample stan-
dard deviation of the variable “dif.mpg” for lighter cars (the level “FALSE”) and for heavier
cars (the level “TRUE”).
12.3. TESTING HYPOTHESIS ON EXPECTATION 217

expression “mu=5.53”. The fact that we are interested in the testing of the spe-
cific alternative is specified by the introduction of a new argument of the form:
“alternative="greater"”. The default value of the argument “alternative”
is “"two.sided"”, which produces a test of a two-sided alternative. By changing
the value of the argument to “"greater"” we produce a test for the appropriate
one-sided alternative:

> t.test(dif.mpg[!heavy],mu=5.53,alternative="greater")

One Sample t-test

data: dif.mpg[!heavy]
t = 1.9692, df = 102, p-value = 0.02582
alternative hypothesis: true mean is greater than 5.53
95 percent confidence interval:
5.573323 Inf
sample estimates:
mean of x
5.805825

The value of the test statistic (t = 1.9692) is the same as for the test of
the two-sided alternative and so is the number of degrees of freedom associ-
ated with the statistic (df = 102). However, the p-value is smaller (p-value
= 0.02582), compared to the p-value in the test for the two-sided alternative
(p-value = 0.05164). The p-value for the one-sided test is the probability un-
der the sampling distribution that the test statistic obtains vales larger than
the observed value of 1.9692. The p-value for the two-sided test is twice that
figure since it involves also the probability of being less than the negative of the
observes value.
The estimated value of the expectation, the sample average, is unchanged.
However, instead of producing a confidence interval for the expectation the re-
port produces a one-sided confidence interval of the form [5.573323, ∞). Such an
interval corresponds to the smallest value that the expectation may reasonably
obtain on the basis of the observed data.
Finally, consider the test of the other one-sided alternative H1 : E(X) < 5.53:

> t.test(dif.mpg[!heavy],mu=5.53,alternative="less")

One Sample t-test

data: dif.mpg[!heavy]
t = 1.9692, df = 102, p-value = 0.9742
alternative hypothesis: true mean is less than 5.53
95 percent confidence interval:
-Inf 6.038328
sample estimates:
mean of x
5.805825

The alternative here is determined by the expression “alternative="less"”.

The p-value is equal to 0.9742, which is the probability that the test statistic
218 CHAPTER 12. TESTING HYPOTHESIS

obtains values less than the observed value of 1.9692. Clearly, the null hypothesis
is not rejected in this test.

12.4 Testing Hypothesis on Proportion

Consider the problem of testing hypothesis on the probability of an event. Recall
that a probability p of some event can be estimated by the observed relative
frequency of the event in the sample, denoted P̂ . The estimation is associated
with the Bernoulli random variable X, that obtains the value 1 when the event
occurs and the value 0 when it does not. The statistical model states that p is the
expectation of X. The estimator P̂ is the sample average of this measurement.
With this formulation we may relate the problem of testing hypotheses for-
mulated in terms of p to the problem of tests associated to the expectation of a
measurement. For the latter problem we applied the t-test. A similar, though
not identical, test is used for the problem of testing hypothesis on proportions.
Assume that one in interested in testing the null hypothesis that the prob-
ability of the event is equal to some specific value, say one half, versus the
alternative hypothesis that the probability is not equal to this value. These
hypotheses are formulated as H0 : p = 0.5 and H1 : p 6= 0.5.
The sample proportion of the event P̂ is the basis for the construction of the
test statistic. Recall that the variance of the estimator P̂ is given by Var(P̂ ) =
p(1 − p)/n. Under the null hypothesis we get that the variance is equal to
Var(P̂ ) = 0.5(1 − 0.5)/n. A natural test statistic is the standardized sample
proportion:
P̂ − 0.5
Z=p ,
0.5(1 − 0.5)/n
that measures the ratio between the deviation of the estimator from its null
expected value and the standard deviation of the estimator. The standard
deviation of the sample proportion is used in the ratio.
If the null hypothesis that p = 0.5 holds true then one gets that the value
0 is the center of the sampling distribution of the test statistic Z. Values of
the statistic that are much larger or much smaller than 0 indicate that the
null hypothesis is unlikely. Consequently, one may consider a rejection region
of the form {|Z| > c}, for some threshold value c. The threshold c is set
at a high enough level to assure the required significance level, namely the
probability under the null hypothesis of obtaining a value in the rejection region.
Equivalently, the rejection region can be written in the form {Z 2 > c2 }.
As a result of the Central Limit Theorem one may conclude that the dis-
tribution of the test statistic is approximately Normal. Hence, Normal compu-
tations may be used in order to produce an approximate threshold or in order
to compute an approximation for the p-value. Specifically, if Z has the stan-
dard Normal distribution then Z 2 has a chi-square distribution on one degree
of freedom.
In order to illustrate the application of hypothesis testing for proportion
consider the following problem: In the previous section we obtained the curb
weight of 2,414 lb as the sample median. The weights of half the cars in the
sample were above that level and the weights of half the cars were below this
level. If this level was actually the population median then the probability that
the weight of a random car is not exceeding this level would be equal to 0.5.
12.4. TESTING HYPOTHESIS ON PROPORTION 219

Let us test the hypothesis that the median weight of cars that run on diesel
is also 2,414 lb. Recall that 20 out of the 205 car types in the sample have diesel
engines. Let us use the weights of these cars in order to test the hypothesis.
The variable “fuel.type” is a factor with two levels “diesel” and “gas”
that identify the fuel type of each car. The variable “heavy” identifies for each
car whether its weight is above the level of 2414 or not. Let us produce a 2 × 2
table that summarizes the frequency of each combination of weight group and
the fuel type:

> fuel <- cars$fuel.type

> table(fuel,heavy)
heavy
fuel FALSE TRUE
diesel 6 14
gas 97 88

Originally the function “table” was applied to a single factor and produced
a sequence with the frequencies of each level of the factor. In the current ap-
plication the input to the function are two factors9 . The output is a table of
frequencies. Each entry to the table corresponds to the frequency of a combi-
nation of levels, one from the first input factor and the other from the second
input factor. In this example we obtain that 6 cars use diesel and their curb
weight was below the threshold. There are 14 cars that use diesel and their curb
weight is above the threshold. Likewise, there are 97 light cars that use gas and
88 heavy cars with gas engines.
The function “prop.test” produces statistical tests for proportions. The
relevant information for the current application of the function is the fact that
frequency of light diesel cars is 6 among a total number of 20 diesel cars. The
first entry to the function is the frequency of the occurrence of the event, 6 in
this case, and the second entry is the relevant sample size, the total number of
diesel cars which is 20 in the current example:

> prop.test(6,20)

1-sample proportions test with continuity correction

data: 6 out of 20, null probability 0.5

X-squared = 2.45, df = 1, p-value = 0.1175
alternative hypothesis: true p is not equal to 0.5
95 percent confidence interval:
0.1283909 0.5433071
sample estimates:
p
0.3

The function produces a report that is printed on the screen. The title
identifies the test as a one-sample test of proportions. In later chapters we
will apply the same function to more complex data structures and the title will
9 To be more accurate, the variable “heavy” is not a factor but a sequence with logical

components. Nonetheless, when the function “table” is applied to such a sequence it treats
it as a factor with two levels, “TRUE” and “FALSE”.
220 CHAPTER 12. TESTING HYPOTHESIS

change accordingly. The title also identifies the fact that a continuity correction
is used in the computation of the test statistic.
The line under the title indicates the frequency of the event in the sample
and the sample size. (In the current example, 6 diesel cars with weights below
the threshold among a total of 20 diesel cars.) The probability of the event,
under the null hypothesis, is described. The default value of this probability
is “p = 0.5”, which is the proper value in the current example. This default
value can be modified by replacing the value 0.5 by the appropriate probability.
The next line presents the information relevant for the test itself. The test
statistic, which is essentially the square of the Z statistic described above10 ,
obtains the value 2.45. The sampling distribution of this statistic under the null
hypothesis is, approximately, the chi-square distribution on 1 degree of freedom.
The p-value, which is the probability that chi-square distribution on 1 degree of
freedom obtains a value above 2.45, is equal to 0.1175. Consequently, the null
hypothesis is not rejected at the 5% significance level.
The bottom part of the report provides the confidence interval and the point
estimate for the probability of the event. The confidence interval for the given
data is [0.1283909, 0.5433071] and the point estimate is p̂ = 6/20 = 0.3.
It is interesting to note that although the deviation between the estimated
proportion p̂ = 0.3 and the null value of the probability p = 0.5 is relatively large
still the null hypothesis was not rejected. The reason for that is the smallness
of the sample, n = 20, that was used in order to test the hypothesis. Indeed, as
an exercise let us examine the application of the same test to a setting where
n = 200 and the number of occurrences of the event is 60:

> prop.test(60,200)

1-sample proportions test with continuity correction

data: 60 out of 200, null probability 0.5

X-squared = 31.205, df = 1, p-value = 2.322e-08
alternative hypothesis: true p is not equal to 0.5
95 percent confidence interval:
0.2384423 0.3693892
sample estimates:
p
0.3

The estimated value of the probability is the same as before since p̂ = 60/200 =
0.3. However, the p-value is 2.322 × 10−8 , which is way below the significance
threshold of 0.05. In this scenario the null hypothesis is rejected with flying
colors.
This last example is yet another demonstration of the basic characteristic
of statistical hypothesis testing. The consideration is based not on the discrep-
ancy of the estimator of the parameter from the value of the parameter under
10 The test statistic that is computed by default is based on Yates’ correction for continuity,

which is very similar to the continuity correction that was used in Chapter 6 for the Normal
approximation of the Binomial distribution. Specifically, the test statistic to the continuity
correction for testing H0 : p = p0 takes the form [|p̂ − p0 | − 0.5/n]2 /[p0 (1 − p0 )/n]. Compare
this statistic with the statistic proposed in the text that takes the form [p̂−p0 ]2 /[p0 (1−p0 )/n].
The latter statistic is used if the argument “correct = FALSE” is added to the function.
12.5. SOLVED EXERCISES 221

the null. Instead, it is based on the relative discrepancy in comparison to the

sampling variability of the estimator. When the sample size is larger the vari-
ability is smaller. Hence, the chances of rejecting the null hypothesis for the
same discrepancy increases.

12.5 Solved Exercises

Question 12.1. Consider a medical condition that does not have a standard
treatment. The recommended design of a clinical trial for a new treatment
to such condition involves using a placebo treatment as a control. A placebo
treatment is a treatment that externally looks identical to the actual treatment
but, in reality, it does not have the active ingredients. The reason for using
placebo for control is the “placebo effect”. Patients tent to react to the fact that
they are being treated regardless of the actual beneficial effect of the treatment.
As an example, consider the trial for testing magnets as a treatment for pain
that was described in Question 9.1. The patients that where randomly assigned
to the control (the last 21 observations in the file “magnets.csv”) were treated
with devises that looked like magnets but actually were not. The goal in this
exercise is to test for the presence of a placebo effect in the case study “Magnets
and Pain Relief” of Question 9.1 using the data in the file “magnets.csv”.

1. Let X be the measurement of change, the difference between the score of

pain before the treatment and the score after the treatment, for patients
that were treated with the inactive placebo. Express, in terms of the
expected value of X, the null hypothesis and the alternative hypothesis
for a statistical test to determine the presence of a placebo effect. The null
hypothesis should reflect the situation that the placebo effect is absent.

2. Identify the observations that can be used in order to test the hypotheses.

3. Carry out the test and report your conclusion. (Use a significance level of
5%.)

Solution (to Question 12.2.1): The null hypothesis of no placebo effect

corresponds to the expectation of the change equal to 0 (H0 : E(X) = 0). The
alternative hypothesis may be formulated as the expectation not being equal to
0 (H1 : E(X) 6= 0). This corresponds to a two sided alternative. Observe that
a negative expectation of the change still corresponds to the placebo having an
effect.

Solution (to Question 12.2.2): The observations that can be used in order
to test the hypothesis are those associated with patients that were treated with
the inactive placebo, i.e. the last 21 observations. We extracts these values form
the data frame using the expression “magnets$change[30:50]”.

Solution (to Question 12.2.3): In order to carry out the test we read the
data from the file “magnets.csv” into the data frame “magnets”. The function
“t.test” is applied to the observations extracted from the data frame. Note
that the default expectation value of tested by the function is “mu = 0”:

> magnets <- read.csv("magnets.csv")

222 CHAPTER 12. TESTING HYPOTHESIS

> t.test(magnets$change[30:50])

One Sample t-test

data: magnets$change[30:50]
t = 3.1804, df = 20, p-value = 0.004702
alternative hypothesis: true mean is not equal to 0
95 percent confidence interval:
0.3768845 1.8135916
sample estimates:
mean of x
1.095238
The computed p-value is 0.004702, which is below 0.05. Consequently, we reject
the null hypothesis and conclude that a placebo effect seems to be present.

Question 12.2. It is assumed, when constructing the t-test, that the measure-
ments are Normally distributed. In this exercise we examine the robustness of
the test to divergence from the assumption. You are required to compute the
significance level of a two-sided t-test of H0 : E(X) = 4 versus H1 : E(X) 6= 4.
Assume there are n = 20 observations and use a t-test with a nominal 5%
significance level.
1. Consider the case where X ∼ Exponential(1/4).
2. Consider the case where X ∼ Uniform(0, 8).
Solution (to Question 12.2.1): We simulate the sampling distribution of the
sample average and standard deviation. The sample is composed of n = 20
observations from the given Exponential distribution:
> lam <- 1/4
> n <- 20
> X.bar <- rep(0,10^5)
> S <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X <- rexp(n,lam)
+ X.bar[i] <- mean(X)
+ S[i] <- sd(X)
+ }
> T <- (X.bar - 4)/(S/sqrt(n))
> mean(abs(T) > qt(0.975,n-1))
[1] 0.08047
We compute the test statistic “T” from the sample average “X.bar” and the
sample standard deviation “S”. In the last expression we compute the probabil-
ity that the absolute value of the test statistic is larger than “qt(0.975,19)”,
which is the threshold that should be used in order to obtain a significance level
of 5% for Normal measurements.
We obtain that the actual significance level of the test is 0.08047, which is
substantially larger than the nominal significance level.
12.5. SOLVED EXERCISES 223

Solution (to Question 12.2.2): We repeat essentially the same simulations

as before. We only change the distribution of the sample from the Exponential
to the Uniform distribution:
> a <- 0
> b <- 8
> n <- 20
> X.bar <- rep(0,10^5)
> S <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X <- runif(n,a,b)
+ X.bar[i] <- mean(X)
+ S[i] <- sd(X)
+ }
> T <- (X.bar - 4)/(S/sqrt(n))
> mean(abs(T) > qt(0.975,n-1))
[1] 0.05163
The actual significance level of the test is 0.05163, much closer to the nominal
significance level of 5%.
A possible explanation for the difference between the two cases is that the
Uniform distribution is symmetric like the Normal distribution, whereas the
Exponential is skewed. In any case, for larger sample sizes one may expect the
Central Limit Theorem to kick in and produce more satisfactory results, even
for the Exponential case.

Question 12.3. Assume that you are interested in testing H0 : E(X) = 20

versus H1 : E(X) 6= 20 with a significance level of 5% using the t-test. Let the
sample average, of a sample of size n = 55, be equal to x̄ = 22.7 and the sample
standard deviation be equal to s = 5.4.
1. Do you reject the null hypothesis?
2. Use the same information. Only now you are interested in a significance
level of 1%. Do you reject the null hypothesis?
3. Use the information the presentation of the exercise. But now you are
interested in testing H0 : E(X) = 24 versus H1 : E(X) 6= 24 (with a
significance level of 5%). Do you reject the null hypothesis?
Solution (to Question 12.3.1): We input the data to R and then compute
the test statistic and the appropriate percentile of the t-distribution:
> n <- 55
> x.bar <- 22.7
> s <- 5.4
> t <- (x.bar - 20)/(s/sqrt(n))
> abs(t)
[1] 3.708099
> qt(0.975,n-1)
[1] 2.004879
224 CHAPTER 12. TESTING HYPOTHESIS

Observe that the absolute value of the statistic (3.708099) is larger than the
threshold for rejection (2.004879). Consequently, we reject the null hypothesis.

Solution (to Question 12.3.2): We recompute the percentile of the t-distribution:

> qt(0.995,n-1)
[1] 2.669985

Again, the absolute value of the statistic (3.708099) is larger than the threshold
for rejection (2.669985). Consequently, we reject the null hypothesis.

Solution (to Question 12.3.3): In this question we should recompute the

test statistic:

> t <- (x.bar - 24)/(s/sqrt(n))

> abs(t)
[1] 1.785381

The absolute value of the new statistic (1.785381) is smaller than the threshold
for rejection (2.004879). Consequently, we do not reject the null hypothesis.

12.6 Summary
Glossary
Hypothesis Testing: A method for determining between two hypothesis, with
one of the two being the currently accepted hypothesis. A determination is
based on the value of the test statistic. The probability of falsely rejecting
the currently accepted hypothesis is the significance level of the test.

Null Hypothesis (H0 ): A sub-collection that emerges in response to the sit-

uation when the phenomena is absent. The established scientific theory
that is being challenged. The hypothesis which is worse to erroneously
reject.

Alternative Hypothesis (H1 ): A sub-collection that emerges in response to

the presence of the investigated phenomena. The new scientific theory
that challenges the currently established theory.

Test Statistic: A statistic that summarizes the data in the sample in order to
decide between the two alternative.

Rejection Region: A set of values that the test statistic may obtain. If the
observed value of the test statistic belongs to the rejection region then the
null hypothesis is rejected. Otherwise, the null hypothesis is not rejected.

Type I Error The null hypothesis is correct but it is rejected by the test.

Type II Error The alternative hypothesis holds but the null hypothesis is not
rejected by the test.
12.6. SUMMARY 225

Significance Level: The probability of a Type I error. The probability, com-

puted under the null hypothesis, of rejecting the null hypothesis. The
test is constructed to have a given significance level. A commonly used
significance level is 5%.
Statistical Power: The probability, computed under the alternative hypoth-
esis, of rejecting the null hypothesis. The statistical power is equal to 1
minus the probability of a Type II error.

p-value: A form of a test statistic. It is associated with a specific test statistic

and a structure of the rejection region. The p-value is equal to the signif-
icance level of the test in which the observed value of the statistic serves
as the threshold.

Discuss in the forum

In statistical thinking there is a tenancy towards conservatism. The investi-
gators, enthusiastic to obtain positive results, may prefer favorable conclusions
and may tend over-interpret the data. It is the statistician’s role to add to the
objectivity in the interpretation of the data and to advocate caution.
On the other hand, the investigators may say that conservatism and science
are incompatible. If one is too cautious, if one is always protecting oneself
against the worst-case scenario, then one will not be able to make bold new
discoveries.
Which of the two approach do you prefer?
When you formulate your answer to this question it may be useful to recall
cases in your past in which you where required to analyze data or you were
exposed to other people’s analysis. Could the analysis benefit or be harmed by
either of the approaches?
For example, many scientific journal will tend to reject a research paper
unless the main discoveries are statistically significant (p-value < 5%). Should
one not publish also results that show a significance level of 10%?

Formulas:
√
• Test Statistic for Expectation: t = (x̄ − µ0 )/(s/ n).
• Two-Sided Test: Reject H0 if {|t| > qt(0.975,n-1)}.

• Greater Than: Reject H0 if {t > qt(0.95,n-1)}.

• Less Than: Reject H0 if {t < qt(0.05,n-1)}.
226 CHAPTER 12. TESTING HYPOTHESIS
Chapter 13

Comparing Two Samples

13.1 Student Learning Objectives

The next 3 chapters deal with the statistical inference associated with the rela-
tion between two variables. The relation corresponds to the effect of one variable
on the distribution of the other. The variable whose distribution is being in-
vestigated is called the response. The variable which may have an effect on the
distribution of the response is called the explanatory variable.
In this section we consider the case where the explanatory variable is a
factor with two levels. This factor splits the sample into two sub-samples. The
statistical inference compares between the distributions of the response variable
in the two sub-samples. The statistical inference involves point estimation,
confidence intervals, and hypothesis testing. R functions may be used in order
to carry out the statistical inference. By the end of this chapter, the student
should be able to:

• Define estimators, confidence intervals, and tests for comparing the distri-
bution of a numerical response between two sub-populations.

• Apply the function “t.test” in order to investigate the difference between

the expectations of the response variable in the two sub-samples.

• Apply the function “var.test” in order to investigate the ratio between

the variances of the response variable in the two sub-samples.

13.2 Comparing Two Distributions

Up until this point in the book we have been considering tools for the inves-
tigation of the characteristics of the distribution of a single measurement. In
most applications, however, one is more interested in inference regarding the
relationships between several measurements. In particular, one may want to
understand how the outcome of one measurement effects the outcome of an-
other measurement.
A common form of a mathematical relation between two variables is when
one of the variables is a function of the other. When such a relation holds then

227
228 CHAPTER 13. COMPARING TWO SAMPLES

the value of the first variably is determined by the value of the second. How-
ever, in the statistical context relations between variables are more complex.
Typically, a statistical relation between variables does not make one a direct
function of the other. Instead, the distribution of values of one of the variables
is affected by the value of the other variable. For a given value of the second
variable the first variable may have one distribution, but for a different value
of the second variable the distribution of the first variable may be different. In
statistical terminology the second variable in this setting is called an explana-
tory variable and the first variable, with a distribution affected by the second
variable, is called the response.
As an illustration of the relation between the response and the explana-
tory variable consider the following example. In a clinical trial, which is a
precondition for the marketing of a new medical treatment, a group of pa-
tients is randomly divided into a treatment and a control sub-groups. The new
treatment is anonymously administered to the treatment sub-group. At the
same time, the patients in the control sub-group obtain the currently standard
treatment. The new treatment passes the trial and is approved for marketing
by the Health Authorities only if the response to the medical intervention is
better for the treatment sub-group than it is for the control sub-group. This
treatment-control experimental design, in which a response is measured under
two experimental conditions, is used in many scientific and industrial settings.
In the example of a clinical trial one may identify two variables. One vari-
able measures the response to the medical intervention for each patient that
participated in the trial. This variable is the response variable, the distribu-
tion of which one seeks to investigate. The other variable indicates to which
sub-group, treatment or control, each patient belongs. This variable is the ex-
planatory variable. In the setting of a clinical trial the explanatory variable
is a factor with two levels, “treatment” and “control”, that splits the sample
into two sub-samples. The statistical inference compares the distribution of
the response variable among the patients in the treatment sub-sample to the
distribution of the response among the patients in the control sub-group.
The analysis of experimental settings such as the treatment-control trial ia a
special case that involves the investigation of the effect an explanatory variable
may have on the response variable. In this special case the explanatory variable
is a factor with two distinct levels. Each level of the factor is associate with
a sub-sample, either treatment or control. The analysis seeks to compare the
distribution of the response in one sub-sample with the distribution in the other
sub-sample. If the response is a numeric measurement then the analysis may
take the form of comparing the response’s expectation in one sub-group to the
expectation in the other. Alternatively, the analysis may involve comparing the
variance. In a different case, if the response is the indicator of the occurrence
of an event then the analysis may compare two probabilities, the probability of
the event in the treatment group to the probability of the same event in the
control group.
In this chapter we deal with statistical inference that corresponds to the
comparison of the distribution of a numerical response variable between two
sub-groups that are determined by a factor. The inference includes testing hy-
potheses, mainly the null hypothesis that the distribution of the response is
the same in both subgroups versus the alternative hypothesis that the distribu-
tion is not the same. Another element in the inference is point estimation and
13.3. COMPARING THE SAMPLE MEANS 229

confidence intervals of appropriate parameters.

In the next chapter we will consider the case where the explanatory variable
is numeric and in the subsequent chapter we describe the inference that is used
in the case that the response is the indicator of the occurrence of an event.

13.3 Comparing the Sample Means

In this section we deal with the issue of statistical inference when comparing
the expectation of the response variable in two sub-samples. The inference is
used in order to address questions such as the equality of the two expectations
to each other and, in the case they are not equal, the assessment of the differ-
ence between the expectations. For the first question one may use statistical
hypothesis testing and for the assessment one may use point estimates and/or
confidence intervals.
In the first subsection we provide an example of a test of the hypothesis that
the expectations are equal. A confidence interval for the difference between
expectations is given in the output of the report of the R function that applies
the test. The second subsection considers the construction of the confidence
interval and the third subsection deals with the theory behind the statistical
test.

13.3.1 An Example of a Comparison of Means

In order to illustrate the statistical inference that compars two expectations let
us return to an example that was considered in Chapter 12. The response of
interest is the difference in miles-per-gallon between driving in highway condi-
tions and driving in city conditions. This response is produced as the difference
between the variable “cars$highway.mpg” and the variable “cars$city.mpg”.
It is stored in the object “dif.mpg”, which is a numerical sequence:
> cars <- read.csv("cars.csv")
> dif.mpg <- cars$highway.mpg - cars$city.mpg
The object “heavy” was defined in the previous chapter as a sequence with
logical components. A component had the value “TRUE” if the curb weight of
the car type associated with this component was above the median level of
2,414 lb. The component obtained the value “FALSE” if the curb weight did not
exceed that level. The logical sequence “heavy” was used in order to select the
subsequences associated with each weight sub-group. Statistical inference was
applied separately to each subsequence.
In the current analysis we want to examine directly the relation between the
response variable “dif.mpg” and an explanatory factor variable “heavy”. In
order to do so we redefine the variable “heavy” to be a factor:
> heavy <- factor(cars$curb.weight > 2414)
The variable “curb.weight” is numeric and the expression “cars$curb.weight
> 2414” produces a sequence with logical “TRUE” or “FALSE” components. This
sequence is not a factor. In order to produce a factor we apply the function
“factor” to the sequence. The function “factor” transforms its input into a
factor. Specifically, the application of this function to a sequence with logical
230 CHAPTER 13. COMPARING TWO SAMPLES

●
10

●
8
dif.mpg

6
4

●
2

● ●

●
0

FALSE TRUE

heavy

Figure 13.1: Distributions of Responses for the Weight Groups

components produces a factor with two levels that are given the names “TRUE”
and “FALSE”1 .
We want to examine the relation between the response variable “dif.mpg”
and the explanatory factor “heavy”. Towards that end we produce a plot of the
relation with the function “plot” and test for the equality of the expectations
of the response with the function “t.test”. First the plot:
> plot(dif.mpg~heavy)
The application of the function “plot” to the expression “dif.mpg ˜ heavy”
produces the plot that is given in Figure 13.1.
Observe that the figure contains two box plots, one associated with the level
“FALSE” of the explanatory factor and the other with the level “TRUE” of that
factor. The box plots describe the distribution of the response variable for
each level of the explanatory factor. Overall, the distribution of the response
for heavier cars (cars associated with the level “TRUE”) tends to obtain smaller
1 It should be noted that the redefined sequence “heavy” is no longer a sequence with

logical components. It cannot be used, for example, as an index to another sequence in order
to sellect the components that are associated with the “TRUE” logical value.
13.3. COMPARING THE SAMPLE MEANS 231

values than the distribution of the response for lighter cars (cars associated with
the level “FALSE”).
The input to the function “plot” is a formula expression of the form:
“response˜explanatory.variable”. A formula identifies the role of variables. The
variable to the left of the tilde character (˜) in a formula is the response and the
variable to the right is the explanatory variable. In the current case the vari-
able “dif.mpg” is the response and the variable “dif.mpg” is the explanatory
variable.
Let us use a formal test in order to negate the hypothesis that the expectation
of the response for the two weight groups is the same. The test is provided by
the application of the function “t.test” to the formula “dif.mpg˜heavy”:

> t.test(dif.mpg~heavy)

Welch Two Sample t-test

data: dif.mpg by heavy

t = 2.4255, df = 191.561, p-value = 0.01621
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
0.1029150 0.9989315
sample estimates:
mean in group FALSE mean in group TRUE
5.805825 5.254902

The function “t.test”, when applied to a formula that describes the relation
between a numeric response and a explanatory factor with two level, produces
a special form of a t-test that is called the Welch Two Sample t-test. The sta-
tistical model associated with this test assumes the present of two independent
sub-samples, each associated with a level of the explanatory variable. The rele-
vant parameters for this model are the two expectations and the two variances
associated with the sub-samples.
The hypotheses tested in the context of the Welch test are formulated in
terms of the difference between the expectation of the first sub-sample and the
expectation of the second sub-sample. In the default application of the test the
null hypothesis is that the difference is equal to 0 (or, equivalently, that the
expectations are equal to each other). The alternative is that the difference is
not equal to 0 (hence, the expectations differ).
The test is conducted with the aid of a test statistic. The computed value of
the test statistic in this example is “t = 2.4255”. Under the null hypothesis the
distribution of the test statistic is (approximately) equal to the t-distribution
on “df = 191.561” degrees of freedom. The resulting p-value is “p-value =
0.01621”. Since the computed p-value is less than 0.05 we reject the null hy-
pothesis with a significance level of 5% and declare that the expectations are
not equal to each other.
The bottom part of the report presents points estimates and a confidence
interval. The point estimates of the two expectations are the sub-samples av-
erages. The estimated value of the expected difference in miles-per-gallon for
lighter cars is 5.805825, which is the average of the measurements associated
with the level “FALSE”. The estimated value of the expected difference for
232 CHAPTER 13. COMPARING TWO SAMPLES

heavier cars is 5.254902, the average of measurements associated with the level
“TRUE”.
The point estimate for the difference between the two expectations is the
difference between the two sample averages: 5.805825 − 5.254902 = 0.550923.
A confidence interval for the difference between the expectations is reported
under the title “95 percent confidence interval:”. The computed value of
the confidence interval is [0.1029150, 0.9989315].
In the rest of this section we describe the theory behind the construction of
the confidence interval and the statistical test.

13.3.2 Confidence Interval for the Difference

Consider the statistical model that is used for the construction of the confidence
interval. The main issue is that the model actually deals with two populations
rather than one population. In previous theoretical discussions we assumed the
presence of a single population and a measurement taken for the members of
this population. When the measurement was considered as a random variable it
was denoted by a capital Latin letter such as X. Of concern were characteristics
of the distribution of X such as E(X), the expectation of X, and Var(X), the
variance.
In the current investigation two populations are considered. One popu-
lation is the sub-population associated with the first level of the factor and
the other population is associated with the second level. The measurement is
taken for the members of both sub-populations. However, the measurement
involves two random variables, one associated with the first sub-population and
the other associated with the second sub-population. Moreover, the distribu-
tion of the measurement for one population may differ from the distribution for
the other population. We denote the random variable associated with the first
sub-population by Xa and the one associated with the other sub-population by
Xb .
Consider the example in which the measurement is the difference in miles-
per-gallon between highway and city driving conditions. In this example Xa is
the measurement for cars with curb weight up to 2,414 lb and Xb is the same
measurement for cars with curb weight above that threshold.
The random variables Xa and Xb may have different distributions. Con-
sequently, the characteristics of their distributions may also vary. Denote by
E(Xa ) and E(Xb ) the expectations of the first and second random variable,
respectively. Likewise, Var(Xa ) and Var(Xb ) are the variances of the two ran-
dom variables. These expectations and variances are subjects of the statistical
inference.
The sample itself may also be divided into two sub-samples according to
the sub-population each observation originated from. In the example, one sub-
sample is associated with the lighter car types and the other sub-sample with
the heavier ones. These sub-samples can be used in order to make inference with
respect to the parameters of Xa and Xb , respectively. For example, the average
of the observations from first sub-sample, X̄a , can serve as the estimator of the
of the expectation E(Xa ) and the second sub-sample‘s average X̄b may be used
in order to estimate E(Xb ).
Our goal in this section is to construct a confidence interval for the differ-
ence in expectations E(Xa ) − E(Xb ). A natural estimator for this difference in
13.3. COMPARING THE SAMPLE MEANS 233

expectations is the difference in averages X̄a − X̄b . The average difference will
also serve as the basis for the construction of a confidence interval.
Recall that the construction of the confidence interval for a signal expec-
tation was based on the sample p average X̄. We exploited the fact that the
distribution of Z = (X̄ − E(X)/ Var(X)/n, the standardized sample aver-
age, is approximately standard Normal. From this Normal approximation we
obtained an approximate 0.95 probability for the event
 p p
− 1.96 · Var(X)/n ≤ X̄ − E(X) ≤ 1.96 · Var(X)/n ,

where 1.96 = qnorm(0.975) is the 0.975-percentile of the standard Normal

distribution2 . Substituting the estimator S for the unknown variance of the
measurement and rewriting the event in a format that puts the expectation
E(X) in the center, between two boundaries, produced the confidence interval:
√
X̄ ± 1.96 · S/ n .

Similar considerations can be used in the construction of a confidence interval

for the difference between expectations on the basis of the difference between
sub-sample averages. The deviation {X̄a − X̄b } − {E(Xa ) − E(Xb )} between
the difference of the averages and the difference of the expectations that they
estimate can be standardized. By the Central Limit Theorem one may obtain
that the distribution of the standardized deviation is approximately standard
Normal.
Standardization is obtained by dividing by the standard deviation of the
estimator. In the current setting the estimator is the difference between the
averages. The variance of the difference is given by
Var(Xa ) Var(Xb )
Var(X̄a − X̄b ) = Var(X̄a ) + Var(X̄b ) = + ,
na nb
where na is the size of the sub-sample that produces the sample average X̄a and
nb is the size of the sub-sample that produces the sample average X̄b . Observe
that both X̄a and X̄b contribute to the variability of the difference. The total
variability is the sum of the two contributions3 . Finally, we use the fact that the
variance of the sample average is equal to he variance of a single measurement
divided by the sample size. This fact is used for both averages in order to obtain
a representation of the variance of the estimator in terms of the variances of the
measurement in the two sub-population and the sizes of the two sub-samples.
The standardized deviation takes the form:
X̄a − X̄b − {E(Xa ) − E(Xb )}
Z=p .
Var(Xa )/na + Var(Xb )/nb
When both sample sizes na and nb are large then the distribution of Z is ap-
proximately standard Normal. As a corollary from the Normal approximation
one gets that P(−1.96 ≤ Z ≤ 1.96) ≈ 0.95.
2 In the case where the sample size is small and the observations are Normally distributed we

used the t-distribution instead. The percentile that was used in that case was qt(0.975,n-1),
the 0.975 percentile of the t-distribution on n − 1 degrees of freedom.
3 It can be proved mathematically that the variance of a difference (or a sum) of two

independent random variables is the sum of the variances. The situation is different when the
two random variables are correlated.
234 CHAPTER 13. COMPARING TWO SAMPLES

The values of variances Var(Xa ) and Var(Xb ) that appear in the definition
of Z are unknown. However, these values can be estimated using the sub-
samples variances Sa2 and Sb2 . When the size of both sub-samples is large then
these estimators will produce good approximations of the unknown variances:

Var(Xa ) Var(Xb ) S2 S2
Var(Xa ) ≈ Sa2 , Var(Xb ) ≈ Sb2 =⇒ + ≈ a + b .
na nb na nb

The event {−1.96 ≤ Z ≤ 1.96} may be approximated by the event:

( s s )
Sa2 Sb2 Sa2 Sb2
− 1.96 · + ≤ X̄a − X̄b − {E(Xa ) − E(Xb )} ≤ 1.96 · + ,
na nb na nb

The approximation results from the use of the sub-sample variances as a substi-
tute for the unknown variances of the measurement in the two sub-populations.
When the two sample sizes na and nb are large then the probability of the given
event will also be approximately equal to 0.95.
Finally, reexpressing the least event in a format that puts the parameter
E(Xa )−E(Xb ) in the center will produce the confidence interval with boundaries
of the form: q
X̄a − X̄b ± 1.96 · Sa2 /na + Sb2 /nb
In order to illustrate the computations that are involved in the construction
of a confidence interval for the difference between two expectations let us return
to the example of difference in miles-per-gallon for lighter and for heavier cars.
Compute the two sample sizes, sample averages, and sample variances:

> table(heavy)
heavy
FALSE TRUE
103 102
> tapply(dif.mpg,heavy,mean)
FALSE TRUE
5.805825 5.254902
> tapply(dif.mpg,heavy,var)
FALSE TRUE
2.020750 3.261114

Observe that there 103 lighter cars and 102 heavier ones. These counts were
obtained by the application of the function “table” to the factor “heavy”. The
lighter cars are associated with the level “FALSE” and heavier cars are associated
with the level “TRUE”.
The average difference in miles-per-gallon for lighter cars is 5.805825 and
the variance is 2.020750. The average difference in miles-per-gallon for heavier
cars is 5.254902 and the variance is 3.261114. These quantities were obtained
by the application of the functions “mean” or “var” to the values of the vari-
able “dif.mpg” that are associated with each level of the factor “heavy”. The
application was carried out using the function “tapply”.
The computed values of the means are equal to the vales reported in the out-
put of the application of the function “t.test” to the formula “dif.mpg˜heavy”.
The difference between the averages is x̄a −x̄b = 5.805825−5.254902 = 0.550923.
13.3. COMPARING THE SAMPLE MEANS 235

This value is the center of the confidence interval. The estimate of the standard
deviation of the difference in averages is:
q p
s2a /na + s2b /nb = 2.020750/103 + 3.261114/102 = 0.227135 .

Therefore, the confidence interval for the difference in expectations is

s
s2a s2
x̄a − x̄b ±1.96· + b = 0.550923±1.96·0.227135 = [0.1057384, 0.9961076] ,
na nb

which is (essentially) the confidence interval that is presented in the report4 .

13.3.3 The t-Test for Two Means

The statistical model that involves two sub-populations may be considered also
in the context of hypothesis testing. Hypotheses can be formulated regarding
the relations between the parameters of the model. These hypotheses can be
tested using the data. For example, in the current application of the t-test,
the null hypothesis is H0 : E(Xa ) = E(Xb ) and the alternative hypothesis is
H1 : E(Xa ) 6= E(Xb ). In this subsection we explain the theory behind this test.
Recall that the construction of a statistical test included the definition of a
test statistic and the determination of a rejection region. The null hypothesis is
rejected if, and only if, the test statistic obtains a value in the rejection region.
The determination of the rejection region is based on the sampling distribution
of the test statistic under the null hypothesis. The significance level of the test
is the probability of rejecting the null hypothesis (i.e., the probability that the
test statistic obtains a value in the rejection region) when the null hypothesis
is correct (the distribution of the test statistic is the distribution under the null
hypothesis). The significance level of the test is set at a given value, say 5%,
thereby restricting the size of the rejection region.
In the previous chapter we consider the case where there is one population.
For review, consider testing the hypothesis that the expectation of the measure-
ment is equal to zero (H0 : E(X) = 0) against the alternative hypothesis that
it is not (H1 : E(X) 6= 0). A sample of size n is obtained from this population.
Based on the sample one may compute a test statistic:
X̄ − 0 X̄
T = √ = √ ,
S/ n S/ n
where X̄ is the sample average and S is the sample standard deviation. The
rejection region of this test is {|T | > qt(0.975,n-1)}, for “qt(0.975,n-1)”
the 0.975-percentile of the t-distribution on n − 1 degrees of freedom.
Alternatively, one may compute the p-value and reject the null hypothesis
if the p-value is less than 0.05. The p-value in this case is equal to P(|T | > |t|),
4 The confidence interval given in the output of the function “t.test” is

[0.1029150, 0.9989315], which is very similar, but not identical, to the confidence interval
that we computed. The discrepancy stems from the selection of the percentile. We used the
percentile of the normal distribution 1.96 = qnorm(0.975). The function “t.test”, on the
other hand, uses the percentile of the t-distribution 1.972425 = qt(0.975,191.561). Using
this value instead would give 0.550923 ± 1.972425 · 0.227135, which coincides with the interval
reported by “t.test”. For practical applications the difference between the two confidence
intervals are not negligible.
236 CHAPTER 13. COMPARING TWO SAMPLES

where t is the computed value of the test statistic. The distribution of T is the
t-distribution of n − 1 degrees of freedom.
A similar approach can be used in the situation where two sub-population
are involved and one wants to test the null hypothesis that the expectations
are equal versus the alternative hypothesis that they are not. The null hypoth-
esis can be written in the form H0 : E(Xa ) − E(Xb ) = 0 with the alternative
hypothesis given as H1 : E(Xa ) − E(Xb ) 6= 0.
It is natural to base the test static on the difference between sub-samples
averages X̄a − X̄b . The T statistic is the ratio between the deviation of the
estimator from the null value of the parameter, divided by the (estimated)
standard deviation of the estimator. In the current setting the estimator is
difference in sub-samples averages X̄a − X̄b , the null value of the parameter, the
difference between the
p expectations, is 0, and the (estimated) standard deviation
of the estimator is Sa2 /na + Sb2 /nb . It turns out that the test statistic in the
current setting is:

X̄a − X̄b − 0 X̄a − X̄b

T =p 2
=p .
2
Sa /na + Sb /nb Sa /na + Sb2 /nb
2

Consider as a measurement the difference in miles-per-gallon. Define the

sub-population a to the lighter cars and the sub-population b to be the heavier
cars. Recall that the sub-sample sizes are na = 103 and nb = 102. Also, the
sub-sample averages are x̄a = 5.805825 and x̄b = 5.254902, and the sub-sample
variances are s2a = 2.020750 and s2b = 5.254902.
In order to calculate the observed value of the test statistic we use once
more the fact that the difference between the averages is x̄a − x̄b = 5.805825 −
5.254902 = 0.550923 and the estimated value of the standard deviation of the
sub-samples average difference is:
q p
s2a /na + s2b /nb = 2.020750/103 + 3.261114/102 = 0.227135 .

It follows that the observed value of the T statistic is

x̄a − x̄b 0.550923
t= p = = 2.425531 ,
s2a /na + s2b /nb 0.227135

which, after rounding up, is equal to the value presented in the report that was
produced by the function “t.test”.
The p-value is computed as the probability of obtaining values of the test
statistic more extreme than the value that was obtained in our data. The
computation is carried out under the assumptions of the null hypothesis. The
limit distribution of the T statistic, when both sub-sample sizes na and nb are
large, is standard Normal. In the case when the measurements are Normally
distributed then a refined approximation of the distribution of the statistic is the
t-distribution. Both the standard Normal and the t-distribution are symmetric
about the origin.
The probability of obtaining a value in either tails for a symmetric distribu-
tion is equal to twice the probability of obtaining a value in the upper tail:
 
P(|T | > 2.4255) = 2 × P(T > 2.4255) = 2 × 1 − P(T ≤ 2.4255) .
13.4. COMPARING SAMPLE VARIANCES 237

The function “t.test” computes the p-value using the t-distribution. For
the current data, the number of degrees of freedom that are used in this approxi-
mation5 is df = 191.561. When we apply the function “pt” for the computation
of the cumulative probability of the t-distribution we get:

> 2*(1-pt(2.4255,191.561))
[1] 0.01621458

which (after rounding) is equal to the reported p-value of 0.01621. This p-value
is less than 0.05, hence the null hypothesis is rejected in favor of the alternative
hypothesis that assumes an effect of the weight on the expectation.

13.4 Comparing Sample Variances

In the previous section we discussed inference associated with the comparison
of the expectations of a numerical measurement between two sub-population.
Inference included the construction of a confidence interval for the difference
between expectations and the testing of the hypothesis that the expectations
are equal to each other.
In this section we consider a comparisons between variances of the mea-
surement in the two sub-populations. For this inference we consider the ratio
between estimators of the variances and introduce a new distribution, the F -
distribution, that is associated with this ratio.
Assume, again, the presence of two sub-populations, denoted a and b. A
numerical measurement is taken over a sample. The sample can be divided
into two sub-samples according to the sub-population of origin. In the previous
section we were interested in inference regarding the relation between the expec-
tations of the measurement in the two sub-populations. Here we are concerned
with the comparison of the variances.
Specifically, let Xa be the measurement at the first sub-population and let
Xb be the measurement at the second sub-population. We want to compare
Var(Xa ), the variance in the first sub-population, to Var(Xb ), the variance in
the second sub-population. As the basis for the comparison we may use Sa2 and
Sb2 , the sub-samples variances, which are computed from the observations in the
first and the second sub-sample, respectively.
Consider the confidence interval for the ratio of the variances. In Chapter 11
we discussed the construction of the confidence interval for the variance in a sin-
gle sample. The derivation was based on the sample variance S 2 that serves as
an estimator of the population variance Var(X). In particular, the distribution
of the random variable (n − 1)S 2 /Var(X) was identified as the chi-square dis-
tribution on n − 1 degrees of freedom6 . A confidence interval for the variance
was obtained as a result of the identification of a central region in the chi-square
distribution that contains a pre-subscribed probability7 .
5 The Weltch t-test for the comparison of two means uses the t-distribution as an approx-

imation of the null distribution of the T test statistic. The number of degrees of freedom is
computed by the formula: df = (va + vb )2 /{va2 /(na − 1) + vb2 /(nb − 1)}, where va = s2a /na
and vb = s2b /nb .
6 This statement holds when the distribution of the measurement is Normal.
7 Use P(qchisq(0.025,n-1) ≤ (n − 1)S 2 /Var(X) ≤ qchisq(0.975,n-1)) = 0.95 and

rewrite the event in a format that puts the parameter in the center. The resulting 95%
confidence interval is [(n − 1)S 2 /qchisq(0.975,n-1), (n − 1)S 2 /qchisq(0.025,n-1)].
238 CHAPTER 13. COMPARING TWO SAMPLES

In order to construct a confidence interval for the ratio of the variances we

consider the random variable that is obtained as a ratio of the estimators of the
variances:
Sa2 /Var(Xa )
∼ F(na −1,nb −1) .
Sb2 /Var(Xb )
The distribution of this random variable is denoted the F -distribution8 . This
distribution is characterized by the number of degrees of freedom associated with
the estimator of the variance at the numerator and by the number of degrees of
freedom associated with the estimator of the variance at the denominator. The
number of degrees of freedom associated with the estimation of each variance is
the number of observation used for the computation of the estimator, minus 1.
In the current setting the numbers of degrees of freedom are na − 1 and nb − 1,
respectively.
The percentiles of the F -distribution can be computed in R using the function
“qf”. For example, the 0.025-percentile of the distribution for the ratio between
sample variances of the response for two sub-samples is computed by the expres-
sion “qf(0.025,dfa,dfb)”, where dfa = na − 1 and dfb = nb − 1. Likewise,
the 0.975-percentile is computed by the expression “qf(0.975,dfa,dfb)”. Be-
tween these two numbers lie 95% of the given F -distribution. Consequently, the
probability that the random variable {Sa2 /Var(Xa )}/{Sb2 /Var(Xb )} obtains its
values between these two percentiles is equal to 0.95:
Sa2 /Var(Xa )
∼ F(na −1,nb −1) =⇒
Sb2 /Var(Xb )
Sa2 /Var(Xa )

P qf(0.025,dfa,dfb) ≤ Sb2 /Var(Xb )
≤ qf(0.975,dfa,dfb) = 0.95 .

A confidence interval for the ratio between Var(Xa ) and Var(Xb ) is obtained
by reformulation of the last event. In the reformulation, the ratio of the variances
is placed in the center:
n Sa2 /Sb2 Var(Xa ) Sa2 /Sb2 o
≤ ≤ .
qf(0.975,dfa,fdb) Var(Xb ) qf(0.025,dfa,dfb)
This confidence interval has a significance level of 95%.
Next, consider testing hypotheses regarding the relation between the vari-
ances. Of particular interest is testing the equality of the variances. One may
formulate the null hypothesis as H0 : Var(Xa )/Var(Xb ) = 1 and test it against
the alternative hypothesis H1 : Var(Xa )/Var(Xb ) 6= 1.
The statistic F = Sa2 /Sb2 can used in order to test the given null hypothesis.
Values of this statistic that are either much larger or much smaller than 1 are
evidence against the null hypothesis and in favor of the alternative hypothesis.
The sampling distribution, under that null hypothesis, of this statistic is the
F(na −1,nb −1) distribution. Consequently, the null hypothesis is rejected either if
F < qf(0.025,dfa,dfb) or if F > qf(0.975,dfa,dfb), where dfa = na − 1
and dfb = nb − 1. The significance level of this test is 5%.
Given an observed value of the statistic, the p-value is computed as the
significance level of the test which uses the observed value as the threshold. If
8 The F distribution is obtained when the measurement has a Normal distribution. When

the distribution of the measurement is not Normal then the distribution of the given random
variable will not be the F -distribution.
13.4. COMPARING SAMPLE VARIANCES 239

the observed value f is less than 1 then the p-value is twice the probability of
the lower tail: 2 · P(F < f ). On the other hand, if f is larger than 1 one takes
twice the upper tail as the p-value: 2 · P(F > f ) = 2 · [1 − P(F ≤ f )]. The null
hypothesis is rejected with a significance level of 5% if the p-value is less than
0.05.
In order to illustrate the inference that compares variances let us return
to the variable “dif.mpg” and compare the variances associated with the two
levels of the factor “heavy”. The analysis will include testing the hypothesis
that the two variances are equal and an estimate and a confidence interval for
their ratio.
The function “var.test” may be used in order to carry out the required
tasks. The input to the function is a formula such “dif.mpg ˜ heavy”, with
a numeric variable on the left and a factor with two levels on the right. The
default application of the function to the formula produces the desired test and
confidence interval:
> var.test(dif.mpg~heavy)

F test to compare two variances

data: dif.mpg by heavy

F = 0.6197, num df = 102, denom df = 101, p-value = 0.01663
alternative hypothesis: true ratio of variances is not equal to 1
95 percent confidence interval:
0.4189200 0.9162126
sample estimates:
ratio of variances
0.6196502
Consider the report produced by the function. The observed value of the test
statistic is “F = 0.6197”, and it is associated with the F -distribution on “num
df = 102” and “denom df = 101” degrees of freedom. The test statistic can be
used in order to test the null hypothesis H0 : Var(Xa )/Var(Xb ) = 1, that states
that the two variance are equal, against the alternative hypothesis that they are
not. The p-value for this test is “p-value = 0.01663”, which is less than 0.05.
Consequently, the null hypothesis is rejected and the conclusion is that the
two variances are significantly different from each other. The estimated ratio
of variances, given at the bottom of the report, is 0.6196502. The confidence
interval for the ratio is reported also and is equal to [0.4189200, 0.9162126].
In order to relate the report to the theoretical discussion above let us recall
that the sub-samples variances are s2a = 2.020750 and s2b = 3.261114. The sub-
samples sizes are na = 103 and nb = 102, respectively. The observed value of
the statistic is the ratio s2a /s2b = 2.020750/3.261114 = 0.6196502, which is the
value that appears in the report. Notice that this is the estimate of the ration
between the variances that is given at the bottom of the report.
The p-value of the two-sided test is equal to twice the probability of the tail
that is associated with the observed value of the test statistic as a threshold.
The number of degrees of freedom is dfa = na −1 = 102 and dfb = nb −1 = 101.
The observed value of the ratio test statistic is f = 0.6196502. This value is
less than one. Consequently, the probability P(F < 0.6196502) enters into the
computation of the p-value, which equals twice this probability:
240 CHAPTER 13. COMPARING TWO SAMPLES

> 2*pf(0.6196502,102,101)
[1] 0.01662612

Compare this value to the p-value that appears in the report and see that, after
rounding up, the two are the same.
For the confidence interval of the ratio compute the percentiles of the F
distribution:

> qf(0.025,102,101)
[1] 0.676317
> qf(0.975,102,101)
[1] 1.479161

The confidence interval is equal to:

h s2a /s2b s2a /s2b i h 0.6196502 0.6196502 i
, = ,
qf(0.975,102,101) qf(0.025,102,101) 1.479161 0.676317
= [0.4189200, 0.9162127] ,

which coincides with the reported interval.

13.5 Solved Exercises

Question 13.1. In this exercise we would like to analyze the results of the
trial that involves magnets as a treatment for pain. The trial is described in
Question 9.1. The results of the trial are provided in the file “magnets.csv”.
Patients in this trail where randomly assigned to a treatment or to a control.
The responses relevant for this analysis are either the variable “change”, which
measures the difference in the score of pain reported by the patients before and
after the treatment, or the variable “score1”, which measures the score of pain
before a device is applied. The explanatory variable is the factor “active”.
This factor has two levels, level “1” to indicate the application of an active
magnet and level “2” to indicate the application of an inactive placebo.
In the following questions you are required to carry out tests of hypotheses.
All tests should conducted at the 5% significance level:

1. Is there a significance difference between the treatment and the control

groups in the expectation of the reported score of pain before the appli-
cation of the device?

2. Is there a significance difference between the treatment and the control

groups in the variance of the reported score of pain before the application
of the device?

3. Is there a significance difference between the treatment and the control

groups in the expectation of the change in score that resulted from the
application of the device?

4. Is there a significance difference between the treatment and the control

groups in the variance of the change in score that resulted from the appli-
cation of the device?
13.5. SOLVED EXERCISES 241

Solution (to Question 13.1.1): The score of pain before the application of
the device is measured in the variable “score1”. This variable is used as the
response. We apply the function “t.test” in order to test the equality of the
expectation of the response in the two groups. First we read in the data from
the file into a data frame and then we apply the test:

> magnets <- read.csv("magnets.csv")

> t.test(magnets$score1 ~ magnets$active)

Welch Two Sample t-test

data: magnets$score1 by magnets$active

t = 0.4148, df = 38.273, p-value = 0.6806
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
-0.3757896 0.5695498
sample estimates:
mean in group "1" mean in group "2"
9.62069 9.52381

The computed p-value is 0.6806, which is above 0.05. Consequently, we do not

reject the null hypothesis that the expectations in the two groups are equal.
This should not come as a surprise, since patients were assigned to the groups
randomly and without knowledge to which group they belong. Prior to the
application of the device, the two groups should look alike.

Solution (to Question 13.1.2): Again, we use the variable “score1” as the
response. Now apply the function “var.test” in order to test the equality of
the variances of the response in the two groups:

> var.test(magnets$score1 ~ magnets$active)

F test to compare two variances

data: magnets$score1 by magnets$active

F = 0.695, num df = 28, denom df = 20, p-value = 0.3687
alternative hypothesis: true ratio of variances is not equal to 1
95 percent confidence interval:
0.2938038 1.5516218
sample estimates:
ratio of variances
0.6950431

The computed p-value is 0.3687, which is once more above 0.05. Consequently,
we do not reject the null hypothesis that the variances in the two groups are
equal. This fact is reassuring. Indeed, prior to the application of the device, the
two groups have the same characteristics. Therefore, any subsequent difference
between the two groups can be attributed to the difference in the treatment.

Solution (to Question 13.1.3): The difference in score between the treatment
and the control groups is measured in the variable “change”. This variable is
242 CHAPTER 13. COMPARING TWO SAMPLES

used as the response for the current analysis. We apply the function “t.test”
in order to test the equality of the expectation of the response in the two groups:
> t.test(magnets$change ~ magnets$active)

Welch Two Sample t-test

data: magnets$change by magnets$active

t = 5.9856, df = 42.926, p-value = 3.86e-07
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
2.749137 5.543145
sample estimates:
mean in group "1" mean in group "2"
5.241379 1.095238
The computed p-value is 3.86 × 10−7 , which is much below 0.05. Consequently,
we reject the null hypothesis that the expectations in the two groups are equal.
The conclusion is that, according to this trial, magnets do have an effect on the
expectation of the response9 .

Solution (to Question 13.1.4): Once more we consider the variable “change”
as the response. We apply the function “var.test” in order to test the equality
of the variances of the response in the two groups:
> var.test(magnets$change ~ magnets$active)

F test to compare two variances

data: magnets$change by magnets$active

F = 4.2062, num df = 28, denom df = 20, p-value = 0.001535
alternative hypothesis: true ratio of variances is not equal to 1
95 percent confidence interval:
1.778003 9.389902
sample estimates:
ratio of variances
4.206171
The computed p-value is 0.001535, which is much below 0.05. Consequently, we
reject the null hypothesis that the variances in the two groups are equal. Hence,
magnets also affect the variance of the response.

Question 13.2. It is assumed, when constructing the F -test for equality of

variances, that the measurements are Normally distributed. In this exercise we
what to examine the robustness of the test to divergence from the assumption.
You are required to compute the significance level of a two-sided F -test of
H0 : Var(Xa ) = Var(Xb ) versus H1 : Var(Xa ) 6= Var(Xb ). Assume there are
na = 29 observations in one group and nb = 21 observations in the other group.
Use an F -test with a nominal 5% significance level.
9 The evaluation of magnets as a treatment for pain produced mixed results and there is a

debate regarding their effectiveness. More information can be found in the NIH NCCAM site.
13.5. SOLVED EXERCISES 243

1. Consider the case where X ∼ Normal(4, 42 ).

2. Consider the case where X ∼ Exponential(1/4).
Solution (to Question 13.2.1): We simulate the sampling distribution of
the sample standard deviation for two samples, one sample of size na = 29 and
the other of size nb = 21. Both samples are simulated from the given Normal
distribution:
> mu <- 4
> sig <- 4
> n.a <- 29
> n.b <- 21
> S.a <- rep(0,10^5)
> S.b <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X.a <- rnorm(n.a,mu,sig)
+ X.b <- rnorm(n.b,mu,sig)
+ S.a[i] <- sd(X.a)
+ S.b[i] <- sd(X.b)
+ }
> F <- S.a^2/S.b^2
> mean((F < qf(0.025,n.a-1,n.b-1))|(F > qf(0.975,n.a-1,n.b-1)))
[1] 0.05074
We compute the test statistic “F” as the ratio of the two sample standard devi-
ations “S.a” and “S.b”. The last expression computes the probability that the
test statistic is either less than “qt(0.025,n.a-1,n.b-1)”, or it is larger than
“qt(0.975,n.a-1,n.b-1)”. The term “qt(0.025,n.a-1,n.b-1)” is the 0.025-
percentile of the F -distribution on 28 and 20 degrees of freedom and the term
“qt(0.975,n.a-1,n.b-1)” is the 0.975-percentile of the same F -distribution.
The result of the last expression is the actual significance level of the test.
We obtain that the actual significance level of the test when the measure-
ments are Normally distributed is 0.05074, which in agreement with the nominal
significance level of 5%. Indeed, the nominal significance level is computed under
the assumption that the distribution of the measurement is Normal.

Solution (to Question 13.2.2): We repeat essentially the same simulations

as before. We only change the distribution of the samples from the Normal to
the Exponential distribution:
> lam <- 1/4
> n.a <- 29
> n.b <- 21
> S.a <- rep(0,10^5)
> S.b <- rep(0,10^5)
> for(i in 1:10^5)
+ {
+ X.a <- rexp(n.a,lam)
+ X.b <- rexp(n.b,lam)
+ S.a[i] <- sd(X.a)
244 CHAPTER 13. COMPARING TWO SAMPLES

+ S.b[i] <- sd(X.b)

+ }
> F <- S.a^2/S.b^2
> mean((F < qf(0.025,n.a-1,n.b-1))|(F > qf(0.975,n.a-1,n.b-1)))
[1] 0.27596
The actual significance level of the test is 0.27596, which is much higher than
the nominal significance level of 5%.
Through this experiment we may see that the F -test is not robust to the
divergence from the assumed Normal distribution of the measurement. If the
distribution of the measurement is skewed (the Exponential distribution is an
example of such skewed distribution) then the application of the test to the data
may produce unreliable conclusions.

Question 13.3. The sample average in one sub-sample is x̄a = 124.3 and the
sample standard deviation is sa = 13.4. The sample average in the second sub-
sample is x̄b = 80.5 and the sample standard deviation is sb = 16.7. The size of
the first sub-sample is na = 15 and this is also the size of the second sub-sample.
We are interested in the estimation of the ratio of variances Var(Xa )/Var(Xb ).
1. Compute the estimate of parameter of interest.
2. Construct a confidence interval, with a confidence level of 95%, to the
value of the parameter of interest.
3. It is discovered that the size of each of the sub-samples is actually equal to
150, and no to 15 (but the values of the other quantities are unchanged).
What is the corrected estimate? What is the corrected confidence interval?
Solution (to Question 13.3.1): We input the data to R and then compute
the estimate:
> s.a <- 13.4
> s.b <- 16.7
> s.a^2/s.b^2
[1] 0.6438381
The estimate is equal to the ratio of the sample variances s2a /s2b . It obtains
the value 0.6438381. Notice that the information regarding the sample averages
and the sizes of the sub-samples is not relevant for the point estimation of the
parameter.

Solution (to Question 13.3.2): We use the formula:

 2 2
(sa /sb )/qf(0.975,14,14), (s2a /s2b )/qf(0.025,14,14)


in order to compute the confidence interval:

> n.a <- 15
> n.b <- 15
> (s.a^2/s.b^2)/qf(0.975,n.a-1,n.b-1)
[1] 0.2161555
> (s.a^2/s.b^2)/qf(0.025,n.a-1,n.b-1)
[1] 1.917728
13.6. SUMMARY 245

The confidence interval we obtain is [0.2161555, 1.917728].

Solution (to Question 13.3.3): The estimate of the parameter is not affected
by the change in the sample sizes and it is still equal to 0.6438381. For the
confidence interval we use now the formula:
 2 2
(sa /sb )/qf(0.975,149,149), (s2a /s2b )/qf(0.025,149,149) :


> n.a <- 150

> n.b <- 150
> (s.a^2/s.b^2)/qf(0.975,n.a-1,n.b-1)
[1] 0.466418
> (s.a^2/s.b^2)/qf(0.025,n.a-1,n.b-1)
[1] 0.8887467
The corrected confidence interval is [0.466418, 0.8887467].

13.6 Summary
Glossary
Response: The variable who‘s distribution one seeks to investigate.
Explanatory Variable: A variable that may affect the distribution of the re-
sponse.

Discuss in the forum

Statistics has an important role in the analysis of data. However, some claim
that the more important role of statistics is in the design stage when one decides
how to collect the data. Good design may improve the chances that the eventual
inference of the data will lead to a meaningful and trustworthy conclusion.
Some say that the quantity of data that is collected is most important. Other
say that the quality of the data is more important than the quantity. What is
your opinion?
When formulating your answer it may be useful to come up with an example
from your past experience where the quantity of data was not sufficient. Else,
you can describe a case where the quality of the data was less than satisfactory.
How did these deficiencies affected the validity of the conclusions of the analysis
of the data?
For illustration consider the surveys. Conducting the survey by the telephone
may be a fast way to reach a large number of responses. However, the quality of
the response may be less that the response obtained by face-to-face interviews.

Formulas:
p
• Test statistic for equality of expectations: t = (x̄a − x̄b )/ s2a /na + s2b /nb .
p
• Confidence interval: (x̄a − x̄b ) ± qnorm(0.975) s2a /na + s2b /nb .
• Test statistic for equality of variances: f = s2a /s2b .
246 CHAPTER 13. COMPARING TWO SAMPLES

• Confidence interval:
 2 2
(sa /sb )/qf(0.975,dfa,dfb), (s2a /s2b )/qf(0.025,dfa,dfb) .

Chapter 14

Linear Regression

14.1 Student Learning Objectives

In the previous chapter we examined the situation whe the response is numeric
and the explanatory variable is a factor with two levels. This chapter deals with
the case where both the response and the explanatory variables are numeric.
The method that is used in order to describe the relations between the two
variables is regression. Here we apply linear regression to deal with a linear
relation between two numeric variables. This type of regression fits a line to
the data. The line summarizes the effect of the explanatory variable on the
distribution of the response.
Statistical inference can be conducted in the context of regression. Specif-
ically, one may fit the regression model to the data. This corresponds to the
point estimation of the parameters of the model. Also, one may produce con-
fidence intervals for the parameters and carry out hypotheses testing. Another
issue that is considered is the assessment of the percentage of variability of the
response that is explained by the regression model.
By the end of this chapter, the student should be able to:

• Produce scatter plots of the response and the explanatory variable.

• Explain the relation between a line and the parameters of a linear equation.
Add lines to a scatter plot.

• Fit the linear regression to data using the function “lm” and conduct
statistical inference on the fitted model.

• Explain the relations among R2 , the percentage of response variability ex-

plained by the regression model, the variability of the regression residuals,
and the variance of the response.

14.2 Points and Lines

In this section we consider the graphical representation of the response and
the explanatory variables on the same plot. The data associated with both
variables is plotted as points in a two-dimensional plane. Linear equations

247
248 CHAPTER 14. LINEAR REGRESSION

can be represented as lines on the same two-dimensional plane. This section

prepares the background for the discussion of the linear regression model. The
actual model of linear regression is introduced in the next section.

14.2.1 The Scatter Plot

Consider two numeric variables. A scatter plot can be used in order to display
the data in these two variables. The scatter plot is a graph in which each
observation is represented as a point. Examination of the scatter plot may
revile relations between the two variables.
Consider an example. A marine biologist measured the length (in millime-
ters) and the weight (in grams) of 10 fish that where collected in one of her
expeditions. The results are summarized in a data frame that is presented in
Table 14.2.1. Notice that the data frame contains 10 observations. The variable
x corresponds to the length of the fish and the variable y corresponds to the
weight.

Observation x y
1 4.5 9.5
2 3.7 8.2
3 1.8 4.9
4 1.3 6.7
5 3.2 12.9
6 3.8 14.1
7 2.5 5.6
8 4.5 8.0
9 4.1 12.6
10 1.1 7.2

Table 14.1: Data

Let us display this data in a scatter plot. Towards that end, let us read the
length data into an object by the name “x” and the weight data into an object
by the name “y”. Finally, let us apply the function “plot” to the formula that
relates the response “y” to the explanatory variable “x”:
> x <- c(4.5,3.7,1.8,1.3,3.2,3.8,2.5,4.5,4.1,1.1)
> y <- c(9.5,8.2,4.9,6.7,12.9,14.1,5.6,8.0,12.6,7.2)
> plot(y~x)
The scatter plot that is produced by the last expression is presented in Fig-
ure 14.1.
A scatter plot is a graph that displays jointly the data of two numerical
variables. The variables (“x” and “y” in this case) are represented by the x-
axis and the y-axis, respectively. The x-axis is associated with the explanatory
variable and the y-axis is associated with the response.
Each observation is represented by a point. The x-value of the point cor-
responds to the value of the explanatory variable for the observation and the
y-value corresponds to the value of the response. For example, the first obser-
vation is represented by the point (x = 4.5, y = 9.5). The two rightmost points
have an x value of 4.5. The higher of the two has a y value of 9.5 and is therefore
14.2. POINTS AND LINES 249

14

●
●
12
10

●
y

●
●
8

●
6

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Figure 14.1: A Scatter Plot

point associated with the first observation. The lower of the two has a y value
of 8.0, and is thus associated with the 8th observation. Altogether there are 10
points in the plot, corresponding to the 10 observations in the data frame.
Let us consider another example of a scatter plot. The file “cars.csv”
contains data regarding characteristics of cars. Among the variables in this
data frame are the variables “horsepower” and the variable “engine.size”.
Both variables are numeric.
The variable “engine.size” describes the volume, in cubic inches, that
is swept by all the pistons inside the cylinders. The variable “horsepower”
measures the power of the engine in units of horsepower. Let us examine the
relation between these two variables with a scatter plot:

> cars <- read.csv("cars.csv")

> plot(horsepower ~ engine.size, data=cars)

In the first line of code we read the data from the file into an R data frame that
is given the name “cars”. In the second line we produce the scatter plot with
“horsepower” as the response and “engine.size” as the explanatory variable.
250 CHAPTER 14. LINEAR REGRESSION

●
250

●
200

●
horsepower

● ●●
● ●

● ●
● ● ● ●
● ● ● ●
150

●
● ●
●●
● ●

● ● ●
● ●●●
● ● ● ●●
●
100

● ● ●
●
●
●
● ●
● ●
●
● ●●
●
●
● ●
●
● ●
●
●
●
●●● ●
●
● ●
●
● ●
● ●●
50

●
●

100 150 200 250 300

engine.size

Figure 14.2: The Scatter Plot of Power versus Engine Size

Both variables are taken from the data frame “cars”. The plot that is produced
by the last expression is presented in Figure 14.2.
Consider the expression “plot(horsepower~engine.size, data=cars)”.
Both the response variable and the explanatory variables that are given in this
expression do not exist in the computer’s memory as independent objects, but
only as variables within the object “cars”. In some cases, however, one may
refer to these variables directly within the function, provided that the argument
“data=data.frame.name” is added to the function. This argument informs the
function in which data frame the variables can be found, where data.frame.name
is the name of the data frame. In the current example, the variables are located
in the data frame “cars”.
Examine the scatter plot in Figure 14.2. One may see that the values of
the response (horsepower) tend to increase with the increase in the values of
the explanatory variable (engine.size). Overall, the increase tends to follow
a linear trend, a straight line, although the data points are not located exactly
on a single line. The role of linear regression, which will be discussed in the
subsequent sections, is to describe and assess this linear trend.
14.2. POINTS AND LINES 251

14

●
a=7 , b=1
a=14 , b=−2
a=8.97 , b=0 ●
●
12
10
y

●
●
8

●
●
6

0 1 2 3 4 5

Figure 14.3: Lines

14.2.2 Linear Equation

Linear regression describes linear trends in the relation between a response and
an explanatory variable. Linear trends may be specified with the aid of linear
equations. In this subsection we discuss the relation between a linear equation
and a linear trend (a straight line).
A linear equation is an equation of the form:

y =a+b·x,

where y and x are variables and a and b are the coefficients of the equation.
The coefficient a is called the intercept and the coefficient b is called the slope.
A linear equation can be used in order to plot a line on a graph. With
each value on the x-axis one may associate a value on the y-axis: the value
that satisfies the linear equation. The collection of all such pairs of points, all
possible x values and their associated y values, produces a straight line in the
two-dimensional plane.
As an illustration consider the three lines in Figure 14.3. The green line is
produced via the equation y = 7+x, the intercept of the line is 7 and the slope is
252 CHAPTER 14. LINEAR REGRESSION

1. The blue is a result of the equation y = 14 − 2x. For this line the intercept is
14 and the slope is -2. Finally, the red line is produced by the equation y = 8.97.
The intercept of the line is 8.97 and the slope is equal to 0.
The intercept describes the value of y when the line crosses the y-axis. Equiv-
alently, it is the result of the application of the linear equation for the value
x = 0. Observe in Figure 14.3 that the green line crosses the y-axis at the level
y = 7. Likewise, the blue line crosses the y-axis at the level y = 14. The red
line stays constantly at the level y = 8.97, and this is also the level at which it
crosses the y-axis.
The slope is the change in the value of y for each unit change in the value
of x. Consider the green line. When x = 0 the value of y is y = 7. When x
changes to x = 1 then the value of y changes to y = 8. A change of one unit in
x corresponds to an increase in one unit in y. Indeed, the slope for this line is
b = 1. As for the blue line, when x changes from 0 to 1 the value of y changes
from y = 14 to y = 12; a decrease of two units. This decrease is associated with
the slope b = −2. Lastly, for the constant red line there is no change in the
value of y when x changes its value from x = 0 to x = 1. Therefore, the slope is
b = 0. A positive slope is associated with an increasing line, a negative slope is
associated with a decreasing line and a zero slope is associated with a constant
line.
Lines can be considered in the context of scatter plots. Figure 14.3 contains
the scatter plot of the data on the relation between the length of fish and their
weight. A regression line is the line that best describes the linear trend of the
relation between the explanatory variable and the response. Neither of the lines
in the figure is the regression line, although the green line is a better description
of the trend than the blue line. The regression line is the best description of the
linear trend.
The red line is a fixed line that is constructed at a level equal to the average
value1 of the variable y. This line partly reflects the information in the data. The
regression line, which we fit in the next section, reflects more of the information
by including a description of the trend in the data.
Lastly, let us see how one can add lines to a plot in R. Functions to produce
plots in R can be divided into two categories: high level and low level plotting
functions. High level functions produce an entire plot, including the axes and the
labels of the plot. The plotting functions that we encountered in the past such
as “plot”, “hist”, “boxplot” and the like are all high level plotting functions.
Low level functions, on the other hand, add features to an existing plot.
An example of a low level plotting function is the function “abline”. This
function adds a straight line to an existing plot. The first argument to the
function is the intercept of the line and the second argument is the slope of the
line. Other arguments may be used in order to specify the characteristics of
the line. For example, the argument “col=color.name” may be used in order
to change the color of the line from its default black color. A plot that is very
similar to plot in Figure 14.3 may be produced with the following code2 :
> plot(y~x)
1 Run the expression “mean(y)” to obtain ȳ = 8.97 as the value of the sample average.
2 The actual plot in Figure 14.3 is produced by a slightly modified code. First an empty
plot is produced with the expression “plot(c(0,5),c(5,15),type="n",xlab="x",ylab="y")”
and then the points are added with the expression “points(y~x)”. The lines are added as in
the text. Finally, a legend is added with the function “legend”.
14.3. LINEAR REGRESSION 253

> abline(7,1,col="green")
> abline(14,-2,col="blue")
> abline(mean(y),0,col="red")

Initially, the scatter plot is created and the lines are added to the plot one after
the other. Observe that color of the first line that is added is green, it has an
intercept of 7 and a slope of 1. The second line is blue, with a intercept of 14
and a negative slope of -2. The last line is red, and its constant value is the
average of the variable y.
In the next section we discuss the computation of the regression line, the line
that describes the linear trend in the data. This line will be added to scatter
plots with the aid of the function “abline”.

14.3 Linear Regression

Data that describes the joint distribution of two numeric variables can be rep-
resented with a scatter plot. The y-axis in this plot corresponds to the response
and the x-axis corresponds to the explanatory variable. The regression line de-
scribes the linear trend of the response as a function of the explanatory variable.
This line is characterized by a linear equation with an intercept and a slope that
are computed from the data.
In the first subsection we present the computation of the regression linear
equation from the data. The second subsection discusses regression as a statis-
tical model. Statistical inference can be carried out on the basis of this model.
In the context of the statistical model, one may consider the intercept and the
slope of the regression model that is fitted to the data as point estimates of the
model’s parameter. Based on these estimates, one may test hypotheses regard-
ing the regression model and construct confidence intervals for parameters.

14.3.1 Fitting the Regression Line

The R function that fits the regression line to data is called “lm”, an acronym
for Linear Model. The input to the function is a formula, with the response
variable to the left of the tilde character and the explanatory variable to the
right of it. The output of the function is the fitted linear regression model.
Let us apply the linear regression function to the data on the weight and
the length of fish. The output of the function is saved by us in a object called
“fit”. Subsequently, the content of the object “fit” is displayed:

> fit <- lm(y~x)

> fit

Call:
lm(formula = y ~ x)

Coefficients:
(Intercept) x
4.616 1.427
254 CHAPTER 14. LINEAR REGRESSION

14
●

●
●
12
10

●
y

●
●
8

●
6

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Figure 14.4: A Fitted Regression Line

When displayed, the output of the function “lm” shows the formula that was
used by the function and provides the coefficients of the regression linear equa-
tion. Observe that the intercept of the line is equal to 4.616. The slope of the
line, the coefficient that multiplies “x” in linear equation, is equal to 1.427.
One may add the regression line to the scatter plot with the aid of the
function “abline”:
> plot(y~x)
> abline(fit)
The first expression produces the scatter plot of the data on fish. The second
expression adds the regression line to the scatter plot. When the input to
the graphical function “abline” is the output of the function “lm” that fits
the regression line, then the result is the addition of the regression line to the
existing plot. The line that is added is the line characterized by the coefficients
that are computed by the function “lm”. The coefficients in the current setting
are 4.616 for the intercept and 1.427 for the slope.
The scatter plot and the added regression line are displayed in Figure 14.4.
Observe that line passes through the points, balancing between the points that
14.3. LINEAR REGRESSION 255

are above the line and the points that are below. The line captures the linear
trend in the data.
Examine the line in Figure 14.4. When x = 1 then the y value of the line
is slightly above 6. When the value of x is equal to 2, a change of one unit,
then value of y is below 8, and is approximately equal to 7.5. This observation
is consistent with the fact that the slop of the line is 1.427. The value of x is
decreased by 1 when changing from x = 1 to x = 0. Consequently, the value of
y when x = 0 should decrease by 1.427 in comparison to its value when x = 1.
The value at x = 1 is approximately 6. Therefore, the value at x = 0 should be
approximately 4.6. Indeed, we do get that the intercept is equal to 4.616.
The coefficients of the regression line are computed from the data and are
hence statistics. Specifically, the slope of the regression line is computed as
the ratio between the covariance of the response and the explanatory variable,
divided by the variance of the explanatory variable. The intercept of the re-
gression line is computed using the sample averages of both variables and the
computed slope.
Start with the slope. The main ingredient in the formula for the slope, the
numerator in the ratio, is the covariance between the two variables. The covari-
ance measures the joint variability of two variables. Recall that the formula for
the sample variance of the variable x is equal to::
Pn
2 Sum of the squares of the deviations (xi − x̄)2
s = = i=1 .
Number of values in the sample − 1 n−1
The formula of the sample covariance between x and y replaces the square of the
deviations by the product of deviations. The product is between an y deviation
and the parallel x deviation:
Pn
Sum of products of the deviations (yi − ȳ)(xi − x̄)
covariance = = i=1 .
Number of values in the sample − 1 n−1
The function “cov” computes the sample covariance between two numeric
variables. The two variables enter as arguments to the function and the sample
covariance is the output. Let us demonstrate the computation by first applying
the given function to the data on fish and then repeating the computations
without the aid of the function:
> cov(y,x)
[1] 2.386111
> sum((y-mean(y))*(x-mean(x)))/9
[1] 2.386111
In both cases we obtained the same result. Notice that the sum of products of
deviations in the second expression was divided by 9, which is the number of
observations, minus 1.
The slope of the regression line is the ratio between the covariance and the
variance of the explanatory variable.
The regression line passes through the point (x̄, ȳ), a point that is determined
by the means of the both the explanatory variable and the response. It follows
that the intercept should obey the equation:

ȳ = a + b · x̄ =⇒ a = ȳ − b · x̄ ,
256 CHAPTER 14. LINEAR REGRESSION

The left-hand-side equation corresponds to the statement that the value of the
regression line at the average x̄ is equal to the average of the response ȳ. The
right-hand-side equation is the solution to the left-hand-side equation.
One may compute the coefficients of the regression model manually by com-
puting first the slope as a ratio between the covariance and the variance of
explanatory variable. The intercept can then be obtained by the equation that
uses the computed slope and the averages of both variables:

> b <- cov(x,y)/var(x)

> a <- mean(y) - b*mean(x)
> a
[1] 4.616477
> b
[1] 1.427385

Applying the manual method we obtain, after rounding up, the same coefficients
that were produced by the application of the function “lm” to the data.
As an exercise, let us fit the regression model to the data on the relation be-
tween the response “horsepower” and the explanatory variable “engine.size”.
Apply the function “lm” to the data and present the results:

> fit.power <- lm(horsepower ~ engine.size, data=cars)

> fit.power

Call:
lm(formula = horsepower ~ engine.size, data = cars)

Coefficients:
(Intercept) engine.size
6.6414 0.7695

The fitted regression model is stored in an object called “fit.power”. The

intercept in the current setting is equal to 6.6414 and the slope is equal to
0.7695.
Observe that one may refer to variables that belong to a data frame, provided
that the name of the data frame is entered as the value of the argument “data”
in the function “lm”. Here we refer to variables that belong to the data frame
“cars”.
Next we plot the scatter plot of the data and add the regression line:

> plot(horsepower ~ engine.size, data=cars)

> abline(fit.power)

The output of the plotting functions is presented in Figure 14.5. Again, the
regression line describes the general linear trend in the data. Overall, with the
increase in engine size one observes increase in the power of the engine.

14.3.2 Inference
Up to this point we have been considering the regression model in the context
of descriptive statistics. The aim in fitting the regression line to the data was to
characterize the linear trend observed in the data. Our next goal is to deal with
14.3. LINEAR REGRESSION 257

●
250

●
200

●
horsepower

● ●●
● ●

● ●
● ● ● ●
● ● ● ●
150

●
● ●
●●
● ●

● ● ●
● ●●●
● ● ● ●●
●
100

● ● ●
●
●
●
● ●
● ●
●
● ●●
●
●
● ●
●
● ●
●
●
●
●●● ●
●
● ●
●
● ●
● ●●
50

●
●

100 150 200 250 300

engine.size

Figure 14.5: A Regression Model of Power versus Engine Size

regression in the context of inferential statistics. The goal here is to produce

statements on characteristics of an entire population on the basis of the data
contained in the sample.
The foundation for statistical inference in a given setting is a statistical
model that produces the sampling distribution in that setting. The sampling
distribution is the frame of reference for the analysis. In this context, the ob-
served sample is a single realization of the sampling distribution, one realization
among infinitely many potential realizations that never take place. The setting
of regression involves a response and an explanatory variable. We provide a
description of the statistical model for this setting.
The relation between the response and the explanatory variable is such that
the value of the later affects the distribution of the former. Still, the value of the
response is not uniquely defined by the value of the explanatory variable. This
principle also hold for the regression model of the relation between the response
Y and the explanatory variable X. According to the model of linear regression
the value of the expectation of the response for observation i, E(Yi ), is a lin-
ear function of the value of the explanatory variable for the same observation.
Hence, there exist and intercept a and a slope b, common for all observations,
258 CHAPTER 14. LINEAR REGRESSION

such that if Xi = xi then

E(Yi ) = a + b · xi .
The regression line can thus be interpreted as the average trend of the response
in the population. This average trend is a linear function of the explanatory
variable.
The intercept a and the slope b of the statistical model are parameters of
the sampling distribution. One may test hypotheses and construct confidence
intervals for these parameters based on the observed data and in relation to the
sampling distribution.
Consider testing hypothesis. A natural null hypothesis to consider is the
hypothesis that the slope is equal to zero. This hypothesis corresponds to state-
ment that the expected value of the response is constant for all values of the
explanatory variable. In other words, the hypothesis is that the explanatory
variable does not affect the distribution of the response3 . One may formulate
this null hypothesis as H0 : b = 0 and test it against the alternative H1 : b 6= 0
that states that the explanatory variable does affect the distribution of the re-
sponse.
A test of the given hypotheses can be carried out by the application of the
function “summary” to the output of the function “lm”. Recall that the function
“lm” was used in order to fit the linear regression to the data. In particular,
this function was applied to the data on the relation between the size of the
engine and the power that the engine produces. The function fitted a regression
line that describes the linear trend of the data. The output of the function was
saved in an object by the name “fit.power”. We apply the function “summary”
to this object:
> summary(fit.power)

Call:
lm(formula = horsepower ~ engine.size, data = cars)

Residuals:
Min 1Q Median 3Q Max
-59.643 -12.282 -5.515 10.251 125.153

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 6.64138 5.23318 1.269 0.206
engine.size 0.76949 0.03919 19.637 <2e-16 ***
---
Signif. codes: 0 "***" 0.001 "**" 0.01 "*" 0.05 "." 0.1 " " 1

Residual standard error: 23.31 on 201 degrees of freedom

(2 observations deleted due to missingness)
Multiple R-squared: 0.6574, Adjusted R-squared: 0.6556
F-statistic: 385.6 on 1 and 201 DF, p-value: < 2.2e-16
3 According to the model of linear regression, the only effect of the explanatory variable on

the distribution of the response is via the expectation. If such an effect, according to the null
hypothesis, is also excluded then the so called explanatory variable is not effecting at all the
distribution of the response.
14.3. LINEAR REGRESSION 259

The output produced by the application of the function “summary” is long and
detailed. We will discuss this output in the next section. Here we concentrate
on the table that goes under the title “Coefficients:”. The said table is made
of 2 rows and 4 columns. It contains information for testing, for each of the
coefficients, the null hypothesis that the value of the given coefficient is equal to
zero. In particular, the second row may be used is order to test this hypothesis
for the slope of the regression line, the coefficient that multiplies the explanatory
variable.
Consider the second row. The first value on this row is 0.76949, which is
equal (after rounding up) to the slope of the line that was fitted to the data
in the previous subsection. However, in the context of statistical inference this
value is the estimate of the slope of the population regression coefficient, the
realization of the estimator of the slope4 .
The second value is 0.03919. This is an estimate of the standard deviation
of the estimator of the slope. The third value is the test statistic. This statis-
tic is the ratio between the deviation of the sample estimate of the parameter
(0.76949) from the value of the parameter under the null hypothesis (0), di-
vided by the estimated standard deviation (0.03919): (0.76949 − 0)/0.03919 =
0.76949/0.03919 = 19.63486, which is essentially the value given in the report5 .
The last value is the computed p-value for the test. It can be shown that
the sampling distribution of the given test statistic, under the null distribution
which assumes no slope, is asymptotically the standard Normal distribution.
If the distribution of the response itself is Normal then the distribution of the
statistic is the t-distribution on n−2 degrees of freedom. In the current situation
this corresponds to 201 degrees of freedom6 . The computed p-value is extremely
small, practically eliminating the possibility that the slope is equal to zero.
The first row presents information regarding the intercept. The estimated
intercept is 6.64138 with an estimated standard deviation of 5.23318. The value
of the test statistic is 1.269 and the p-value for testing the null hypothesis that
the intercept is equal to zero against the two sided alternative is 0.206. In this
case the null hypothesis is not rejected since the p-value is larger than 0.05.
The report contains an inference for the intercept. However, one is advised
to take this inference in the current case with a grain of salt. Indeed, the
intercept is the expected value of the response when the explanatory variable
is equal to zero. Here the explanatory variable is the size of the engine and
the response is the power of that engine. The power of an engine of size zero
is a quantity that has no physical meaning! In general, unless the intercept is
in the range of observations (i.e. the value 0 is in the range of the observed
explanatory variable) one should treat the inference on the intercept cautiously.
Such inference requires extrapolation and is sensitive to the miss-specification
of the regression model.
Apart from testing hypotheses one may also construct confidence intervals
for the parameters. A crude confidence interval may be obtained by taking
4 The estimator of the slope is obtained via the application of the P
formula for the compu-
1 Pn 1 n 2
tation of the slope to the sample: n−1 i=1 (Yi − Ȳ )(Xi − X̄)/ n−1 i=1 (Xi − X̄) .
5 Our computation involves rounding up errors, hence the small discrepancy between the

value we computed and the value in the report.

6 Notice that the “horsepower” measurement is missing for two observation. These obser-

vations are deleted for the analysis, leaving a total of n = 203 observations. The number of
degrees of freedom is n − 2 = 203 − 2 = 201.
260 CHAPTER 14. LINEAR REGRESSION

1.96 standard deviations on each side of the estimate of the parameter. Hence,
a confidence interval for the slope is approximately equal to 0.76949 ± 1.96 ×
0.03919 = [0.6926776, 0.8463024]. In a similar way one may obtain a confidence
interval for the slope7 : 6.64138 ± 1.96 × 5.23318 = [−3.615653, 16.89841].
Alternatively, one may compute confidence intervals for the parameters of
the linear regression model using the function “confint”. The input to this
function is the fitted model and the output is a confidence interval for each of
the parameters:

> confint(fit.power)
2.5 % 97.5 %
(Intercept) -3.6775989 16.9603564
engine.size 0.6922181 0.8467537

Observe the similarity between the confidence intervals that are computed by
the function and the crude confidence intervals that were produced by us. The
small discrepancies that do exist between the intervals result from the fact that
the function “confint” uses the t-distribution whereas we used the Normal
approximation.

14.4 R-squared and the Variance of Residuals

In this section we discuss the residuals between the values of the response and
their estimated expected value according to the regression model. These resid-
uals are the regression model equivalence of the deviations between the obser-
vations and the sample average. We use these residuals in order compute the
variability that is not accounted for by the regression model. Indeed, the ratio
between the total variability of the residuals and the total variability of the
deviations from the average serves as a measure of the variability that is not
explained by the explanatory variable. R-squared, which is equal to 1 minus
this ratio, is interpreted as the fraction of the variability of the response that is
explained by the regression model.
We start with the definition of residuals. Let us return to the artificial
example that compared length of fish to their weight. The data for this example
was given in Table 14.2.1 and was saved in the objects “x” and “y”. The
regression model was fitted to this data by the application of the function “lm”
to the formula “y~x” and the fitted model was saved in an object called “fit”.
Let us apply the function “summary” to the fitted model:

> summary(fit)

Call:
lm(formula = y ~ x)

Residuals:
Min 1Q Median 3Q Max
7 The warning message that was made in the context of testing hypotheses on the intercept

should be applied also to the construction of confidence intervals. If the value 0 is not in the
range of the explanatory variable then one should be careful when interpreting a confidence
interval for the intercept.
14.4. R-SQUARED AND THE VARIANCE OF RESIDUALS 261

-3.0397 -2.1388 -0.6559 1.8518 4.0595

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 4.6165 2.3653 1.952 0.0868 .
x 1.4274 0.7195 1.984 0.0826 .
---
Signif. codes: 0 "***" 0.001 "**" 0.01 "*" 0.05 "." 0.1 " " 1

Residual standard error: 2.791 on 8 degrees of freedom

Multiple R-squared: 0.3297, Adjusted R-squared: 0.246
F-statistic: 3.936 on 1 and 8 DF, p-value: 0.08255
The given report contains a table with estimates of the regression coefficients
and information for conducting hypothesis testing. The report contains other
information that is associated mainly with the notion of the residuals from
regression line. Our current goal is to understand what is that other information.
The residual from regression for each observation is the difference between
the value of the response for the observation and the estimated expectation of
the response under the regression model8 . An observation is a pair (xi , yi ), with
yi being the value of the response. The expectation of the response according
to the regression model is a + b · xi , where a and b are the coefficients of the
model. The estimated expectation is obtained by using, in the formula for the
expectation, the coefficients that are estimated from the data. The residual is
the difference between yi and a + b · xi .
Consider an example. The first observation on the fish is (4.5, 9.5), where
x1 = 4.5 and y1 = 9.5. The estimated intercept is 4.6165 and the estimated
slope is 1.4274. The estimated expectation of the response for the first variable
is equal to
4.6165 + 1.4274 · x1 = 4.6165 + 1.4274 · 4.5 = 11.0398 .
The residual is the difference between the observes response and this value:
y1 − (4.6165 + 1.4274 · x1 ) = 9.5 − 11.0398 = −1.5398 .
The residuals for the other observations are computed in the same manner.
The values of the intercept and the slope are kept the same but the values of
the explanatory variable and the response are changed.
Consult the upper plot in Figure 14.6. This is a scatter plot of the data,
together with the regression line in black and the line of the average in red. A
vertical arrow extends from each data point to the regression line. The point
where each arrow hits the regression line is associated with the estimated value
of the expectation for that point. The residual is the difference between the
value of the response at the origin of the arrow and the value of the response
at the tip of its head. Notice that there are as many residuals as there are
observations.
The function “residuals” computes the residuals. The input to the function
is the fitted regression model and the output is the sequence of residuals. When
we apply the function to the object “fit”, which contains the fitted regression
model for the fish data, we get the residuals:
8 The estimated expectation of the response is also called the predicted response.
262 CHAPTER 14. LINEAR REGRESSION

Residuals

●
● ●
12

●
y

● ●
6 8

●
●
●
●

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Deviations from Mean

●
● ●
12

●
y

● ●
6 8

●
●
●
●

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Figure 14.6: Residuals and Deviations from the Mean

> residuals(fit)
1 2 3 4 5
-1.5397075 -1.6977999 -2.2857694 0.2279229 3.7158923
6 7 8 9 10
4.0594616 -2.5849385-3.0397075 2.1312463 1.0133998
Indeed, 10 residuals are produced, one for each observation. In particular, the
residual for the first observation is -1.5397075, which is essentially the value that
we obtained9 .
Return to the report produced by the application of the function “summary”
to the fitted regression model. The first component in the report is the formula
that identifies the response and the explanatory variable. The second compo-
nent, the component that comes under the title “Residuals:”, gives a summary
of the distribution of the residuals. This summary includes the smallest and the
largest values in the sequence of residuals, as well as the first and third quartiles
9 The discrepancy between the value that we computed and the value computed by the

function results from rounding up errors. We used the vales of the coefficients that appear
in the report. These values are rounded up. The function “residuals” uses the coefficients
without rounding.
14.4. R-SQUARED AND THE VARIANCE OF RESIDUALS 263

and the median. The average is not reported since the average of the residuals
from the regression line is always equal to 0.
The table that contains information on the coefficients was discussed in the
previous section. Let us consider the last 3 lines of the report.
The first of the three lines contains the estimated value of the standard
deviation of the response from the regression model. If the expectations of
the measurements of the response are located on the regression line then the
variability of the response corresponds to the variability about this line. The
resulting variance is estimated by the sum of squares of the residuals from the
regression line, divided by the number of observations minus 2. A division
by the number of observation minus 2 produces an unbiased estimator of the
variance of the response about the regression model. Taking the square root of
the estimated variance produces an estimate of the standard deviation:

> sqrt(sum(residuals(fit)^2)/8)
[1] 2.790787

The last computation is a manual computation of the estimated standard devi-

ation. It involves squaring the residuals and summing the squares. This sum is
divided by the number of observations minus 2 (10 − 2 = 8). Taking the square
root produces estimate. The value that we get for the estimated standard de-
viation is 2.790787, which coincides with the value that appears in the first of
the last 3 lines of the report.
The second of these lines reports the R-squared of the linear fit. In order to
explain the meaning of R-squared let us consider Figure 14.6 once again. The
two plots in the figure present the scatter plot of the data together with the
regression line and the line of the average. Vertical black arrows that represent
the residuals from the regression are added to the upper plot. The lower plot
contains vertical red arrows that extend from the data points to the line of the
average. These arrows represent the deviations of the response from the average.
Consider two forms of variation. One form is the variation of the response
from its average value. This variation is summarized by the sample variance, the
sum of the squared lengths of the red arrows divided by the number of observa-
tions minus 1. The other form of variation is the variation of the response from
the fitted regression line. This variation is summarized by the sample variation
of the residuals, the sum of squared lengths of the black arrows divided by the
number of observations minus 1. The ratio between these two quantities gives
the relative variability of the response that remains after fitting the regression
line to the data.
The line of the average is a straight line. The deviations of the observations
from this straight line can be thought of as residuals from that line. The vari-
ability of these residuals, the sum of squares of the deviations from the average
divided by the number of observations minus 1, is equal to the sample variance.
The regression line is the unique straight line that minimizes the variability
of its residuals. Consequently, the variability of the residuals from the regression,
the sum of squares of the residuals from the regression divided by the number
of observations minus 1, is the smallest residual variability produced by any
straight line. It follows that the sample variance of the regression residuals is
less than the sample variance of the response. Therefore, the ratio between the
variance of the residuals and the variance of the response is less than 1.
264 CHAPTER 14. LINEAR REGRESSION

R-squared is the difference between 1 and the ratio of the variances. Its
value is between 0 and 1 and it represents the fraction of the variability of the
response that is explained by the regression line. The closer the points are to
the regression line the larger the value of R-squared becomes. On the other
hand, the less there is a linear trend in the data the closer to 0 is the value of
R-squared. In the extreme case of R-squared equal to 1 all the data point are
positioned exactly on a single straight line. In the other extreme, a value of 0
for R-squared implies no linear trend in the data.
Let us compte manually the difference between 1 and the ratio between the
variance of the residuals and the variance of the response:

> 1-var(residuals(fit))/var(y)
[1] 0.3297413

Observe that the computed value of R-squared is the same as the value “Multiple
R-squared: 0.3297” that is given in the report.
The report provides another value of R-squared, titled Adjusted R-squared.
The difference between the adjusted and unadjusted quantities is that in the
former the sample variance of the residuals from the regression is replaced by
an unbiased estimate of the variability of the response about the regression
line. The sum of squares in the unbiased estimator is divided by the number of
observations minus 2. Indeed, when we re-compute the ratio using the unbiased
estimate, the sum of squared residuals divided by 10 − 2 = 8, we get:

> 1-(sum(residuals(fit)^2)/8)/var(y)
[1] 0.245959

The value of this adjusted quantity is equal to the value “Adjusted R-squared:
0.246” in the report.
Which value of R-squared to use is a matter of personal taste. In any case, for
a larger number of observations the difference between the two values becomes
negligible.
The last line in the report produces an overall goodness of fit test for the
regression model. In the current application of linear regression this test reduces
to a test of the slope being equal to zero, the same test that is reported in the
second row of the table of coefficients10 . The F statistic is simply the square of
the t value that is given in the second row of the table. The sampling distribution
of this statistic under the null hypothesis is the F -distribution on 1 and n − 2
degrees of freedom, which is the sampling distribution of the square of the test
statistic for the slope. The computed p-value, “p-value: 0.08255” is the
identical (after rounding up) to the p-value given in the second line of the table.
Return to the R-squared coefficient. This coefficient is a convenient measure
of the goodness of fit of the regression model to the data. Let us demon-
strate this point with the aid of the “cars” data. In Subsection 14.3.2 we
fitted a regression model to the power of the engine as a response and the size
of the engine as an explanatory variable. The fitted model was saved in the
object called “fit.power”. A report of this fit, the output of the expression
“summary(fit.power)” was also presented. The null hypothesis of zero slope
10 In more complex applications of linear regression, applications that are not considered in

this book, the test in the last line of the report and the tests of coefficients do not coincide.
14.4. R-SQUARED AND THE VARIANCE OF RESIDUALS 265

was clearly rejected. The value of R-squared for this fit was 0.6574. Conse-
quently, about 2/3 of the variability in the power of the engine is explained by
the size of the engine.
Consider trying to fit a different regression model for the power of the engine
as a response. The variable “length” describes the length of the car (in inches).
How well would the length explain the power of the car? We may examine this
question using linear regression:
> summary(lm(horsepower ~ length, data=cars))

Call:
lm(formula = horsepower ~ length, data = cars)

Residuals:
Min 1Q Median 3Q Max
-53.57 -20.35 -6.69 14.45 180.72

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -205.3971 32.8185 -6.259 2.30e-09 ***
length 1.7796 0.1881 9.459 < 2e-16 ***
---
Signif. codes: 0 "***" 0.001 "**" 0.01 "*" 0.05 "." 0.1 " " 1

Residual standard error: 33.12 on 201 degrees of freedom

(2 observations deleted due to missingness)
Multiple R-squared: 0.308, Adjusted R-squared: 0.3046
F-statistic: 89.47 on 1 and 201 DF, p-value: < 2.2e-16
We used one expression to fit the regression model to the data and to summarize
the outcome of the fit.
A scatter plot of the two variables together with the regression line is pre-
sented in Figure 14.7. This plot may be produced using the code:
> plot(horsepower ~ length, data=cars)
> abline(lm(horsepower ~ length, data=cars))
From the examination of the figure we may see that indeed there is a linear
trend in the relation between the length and the power of the car. Longer cars
tend to have more power. Testing the null hypothesis that the slope is equal
to zero produces a very small p-value and leads to the rejection of the null
hypothesis.
The length of the car and the size of the engine are both statistically signif-
icant in their relation to the response. However, which of the two explanatory
variables produces a better fit?
An answer to this question may be provided by the examination of values
of R-squared, the ratio of the variance of the response explained by each of the
explanatory variable. The R-squared for the size of the engine as an explanatory
variable is 0.6574, which is approximately equal to 2/3. The value of R-squared
for the length of the car as an explanatory variable is 0.308, less than 1/3. It
follows that the size of the engine explains twice as much of the variability of
the power of the engine than the size of car and is a better explanatory variable.
266 CHAPTER 14. LINEAR REGRESSION

●
250

●
200

●
horsepower

● ● ● ● ●
● ●

● ●
● ● ● ●
●
●
150

● ●● ● ●
● ● ● ●
● ●

●● ●● ● ●
●● ●
● ● ● ●● ● ●
●
●
●
100

● ● ● ●●●
●●● ● ●
● ●
●
● ● ●
●● ● ●
● ● ● ●
●●
● ● ● ● ●
●●● ●
●●
●● ● ●●●●● ● ● ● ● ●
● ● ●
● ●
●●
50

●
●

140 150 160 170 180 190 200 210

length

Figure 14.7: A Regression Model of Power versus Length

14.5 Solved Exercises

Question 14.1. Figure 14.8 presents 10 points and three lines. One of the lines
is colored red and one of the points is marked as a red triangle. The points in
the plot refer to the data frame in Table 14.1 and the three lines refer to the
linear equations:

1. y = 4

2. y = 5 − 2x

3. y = x

You are asked to match the marked line to the appropriate linear equation and
match the marked point to the appropriate observation:

1. Which of the three equations, 1, 2 or 3, describes the line marked in red ?

2. The poind marked with a red triangle represents which of the observations.
(Identify the observation number.)
14.5. SOLVED EXERCISES 267

Observation x y
1 2.3 -3.0
2 -1.9 9.8
3 1.6 4.3
4 -1.6 8.2
5 0.8 5.9
6 -1.0 4.3
7 -0.2 2.0
8 2.4 -4.7
9 1.8 1.8
10 1.4 -1.1

Table 14.2: Points

Solution (to Question 14.1.1): The line marked in red is increasing and at
x = 0 it seems to obtain the value y = 0. An increasing line is associated with
a linear equation with a positive slope coefficient (b > 0). The only equation
with that property is Equation 3, for which b = 1. Notice that the intercept of
this equation is a = 0, which agrees with the fact that the line passes through
the origin (x, y) = (0, 0). If the x-axis and the y-axis were on the same scale
then one would expect the line to be tilted in the 45 degrees angle. However,
here the axes are not on the same scale, so the tilting is different.

Solution (to Question 14.1.2): The x-value of the line marked with a red
triangle is x = −1. The y-value is below 5. The observation that has an x-value
of -1 is Observation 6. The y-value of this observation is y = 4.3. Notice that
there is another observation with the same y-value, Observation 3. However,
the x-value of that observation is x = 1.6. Hence it is the point that is on the
same level as the marked point, but it is placed to the right of it.

Question 14.2. Assume a regression model that describes the relation between
the expectation of the response and the value of the explanatory variable in the
form:
E(Yi ) = 2.13 · xi − 3.60 .
1. What is the value of the intercept and what is the value of the slope in
the linear equation that describes the model?
2. Assume the x1 = 5.5, x2 = 12.13, x3 = 4.2, and x4 = 6.7. What is the
expected value of the response of the 3rd observation?
Solution (to Question 14.2.1): The intercept is equal to a = −3.60 and
the slope is equal to b = 2.13. Notice that the lope is the coefficient that
multiplies the explanatory variable and the intercept is the coefficient that does
not multiply the explanatory variable.

Solution (to Question 14.2.2): The value of the explanatory variable for the
3rd observation is x3 = 4.2. When we use this value in the regression formula
we obtain that:

E(Y3 ) = 2.13 · x3 − 3.60 = 2.13 · 4.2 − 3.60 = 5.346 .

268 CHAPTER 14. LINEAR REGRESSION

10
●

●
5

●
y

●
●
0

●
−5

−2 −1 0 1 2

Figure 14.8: Lines and Points

In words, the expectation of the response of the 3rd observation is equal to 5.346

Question 14.3. The file “aids.csv” contains data on the number of diag-
nosed cases of Aids and the number of deaths associated with Aids among
adults and adolescents in the United States between 1981 and 200211 . The file
can be found on the internet at http://pluto.huji.ac.il/~msby/StatThink/
Datasets/aids.csv.
The file contains 3 variables: The variable “year” that tells the relevant
year, the variable “diagnosed” that reports the number of Aids cases that were
diagnosed in each year, and the variable “deaths” that reports the number of
Aids related deaths in each year. The following questions refer to the data in
the file:
1. Consider the variable “deaths” as response and the variable “diagnosed”
11 The data is taken from Table 1 in section “Practice in Linear Regression”
of the online Textbook “Collaborative Statistics” (Connexions. March 22, 2010.
http://cnx.org/content/col10522/1.38/) by Barbara Illowsky and Susan Dean.
14.5. SOLVED EXERCISES 269

as an explanatory variable. What is the slope of the regression line?

Produce a point estimate and a confidence interval. Is it statistically
significant (namely, significantly different than 0)?
2. Plot the scatter plot that is produced by these two variables and add
the regression line to the plot. Does the regression line provided a good
description of the trend in the data?
3. Consider the variable “diagnosed” as the response and the variable “year”
as the explanatory variable. What is the slope of the regression line? Pro-
duce a point estimate and a confidence interval. Is the slope in this case
statistically significant?
4. Plot the scatter plot that is produced by the later pair of variables and
add the regression line to the plot. Does the regression line provided a
good description of the trend in the data?
Solution (to Question 14.3.1): After saving the file “aids.csv” in the work-
ing directory of R we read it’s content into a data frame by the name “aids”.
We then produce a summary of the fit of the linear regression model of “deaths”
as a response and “diagnosed” as the explanatory variable:
> aids <- read.csv("aids.csv")
> fit.deaths <- lm(deaths~diagnosed,data=aids)
> summary(fit.deaths)

Call:
lm(formula = deaths ~ diagnosed, data = aids)

Residuals:
Min 1Q Median 3Q Max
-7988.73 -680.86 23.94 1731.32 7760.67

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 88.7161 1370.7191 0.065 0.949
diagnosed 0.6073 0.0312 19.468 1.81e-14 ***
---
Signif. codes: 0 "***" 0.001 "**" 0.01 "*" 0.05 "." 0.1 " " 1

Residual standard error: 3589 on 20 degrees of freedom

Multiple R-squared: 0.9499, Adjusted R-squared: 0.9474
F-statistic: 379 on 1 and 20 DF, p-value: 1.805e-14
The estimated value of the slope 0.6073. The computed p-value associated
with this slope is 1.81 × 10−14 , which is much smaller than the 5% threshold.
Consequently, the slope is statistically significant.
For confidence intervals apply the function “confint” to the fitted model:
> confint(fit.deaths)
2.5 % 97.5 %
(Intercept) -2770.5538947 2947.986092
diagnosed 0.5422759 0.672427
270 CHAPTER 14. LINEAR REGRESSION

50000
● ●

●
40000

●
●
●
30000

●
deaths

20000

● ●
● ●
●
●●
10000

●
●
●●
0

0 20000 40000 60000 80000

diagnosed

Figure 14.9: Aids Related Deaths versus Diagnosed Cases of Aids

We get that the confidence interval for the slope is [0.5422759, 0.672427].

Solution (to Question 14.3.2): A scatter plot of the two variables is produced
by the application of the function “plot” to the formula that involves these two
variables. The regression line is added to the plot using the function “abline”
with the fitted model as an input:
> plot(deaths~diagnosed,data=aids)
> abline(fit.deaths)
The plot that is produced is given in Figure 14.9. Observe that the points are
nicely placed very to a line that characterizes the linear trend of the regression.

Solution (to Question 14.3.3): We fit a linear regression model of “diagnosed”

as a response and “year” as the explanatory variable and save the fit in the ob-
ject “fit.diagnosed”. A summary of the model is produced by the application
of the function “summary” to the fit:
> fit.diagnosed <- lm(diagnosed~year,data=aids)
14.5. SOLVED EXERCISES 271

> summary(fit.diagnosed)

Call:
lm(formula = diagnosed ~ year, data = aids)

Residuals:
Min 1Q Median 3Q Max
-28364 -18321 -3909 14964 41199

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -3448225.0 1535037.3 -2.246 0.0361 *
year 1749.8 770.8 2.270 0.0344 *
---
Signif. codes: 0 "***" 0.001 "**" 0.01 "*" 0.05 "." 0.1 " " 1

Residual standard error: 22940 on 20 degrees of freedom

Multiple R-squared: 0.2049, Adjusted R-squared: 0.1651
F-statistic: 5.153 on 1 and 20 DF, p-value: 0.03441
The estimated value of the slope 1749.8. The computed p-value associated with
this slope is 0.0344, which is less than the 0.05. Consequently, one may declare
the slope to be statistically significant. Confidence intervals are produced using
the function “confint”:
> confint(fit.diagnosed)
2.5 % 97.5 %
(Intercept) -6650256.6654 -246193.429
year 141.9360 3357.618
We get that the 95% confidence interval for the slope is [141.9360, 3357.618].

Solution (to Question 14.3.4): A scatter plot of the two variables is produced
by the application of the function “plot” and the regression line is added with
function “abline”:
> plot(diagnosed~year,data=aids)
> abline(fit.diagnosed)
The plot is given in Figure 14.10. Observe that the points do not follow a linear
trend. It seems that the number of diagnosed cases increased in an exponential
rate during the first years after Aids was discovered. The trend changed in
the mid 90’s with a big drop in the number of diagnosed Aids patients. This
drop may be associated with the administration of therapies such as AZT to
HIV infected subjects that reduced the number of such subjects that developed
Aids. In the late 90’s there seams to be yet again a change in the trend and
the possible increase in numbers. The line of linear regression misses all these
changes and is a poor representation of the historical development.
The take home message from this exercise is to not use models blindly. A
good advise is to plot the data. An examination of the plot provides a warning
that the linear model is probably not a good model for the given problem.
272 CHAPTER 14. LINEAR REGRESSION

80000 ● ●

●
●
60000

● ●

●
●
diagnosed

40000

●
●

●
● ● ●
●
20000

●
●
● ●
0

1985 1990 1995 2000

year

Figure 14.10: Diagnosed Cases of Aids versus Year of Report

Question 14.4. Below are the percents of the U.S. labor force (excluding self-
employed and unemployed) that are members of a labor union12 . We use this
data in order to practice the computation of the regression coefficients.

1. Produce the scatter plot of the data and add the regression line. Is the
regression model reasonable for this data?

2. Compute the sample averages and the sample standard deviations of both
variables. Compute the covariance between the two variables.

3. Using the summaries you have just computed, recompute the coefficients
of the regression model.

Solution (to Question 14.4.1): We read the data in the table into R. The
variable “year” is the explanatory variable and the variable “percent” is the
12 Taken from Homework section in the chapter on linear regression of the

online Textbook “Collaborative Statistics” (Connexions. March 22, 2010.

http://cnx.org/content/col10522/1.38/) by Barbara Illowsky and Susan Dean.
14.5. SOLVED EXERCISES 273

year percent
1945 35.5
1950 31.5
1960 31.4
1970 27.3
1980 21.9
1986 17.5
1993 15.8

Table 14.3: Percent of Union Members

response. The scatter plot is produced using the function “plot” and the re-
gression line, fitted to the data with the function “lm”, is added to the plot
using the function “abline”:

> year <- c(1945,1950,1960,1970,1980,1986,1993)

> percent <- c(35.5,31.5,31.4,27.3,21.9,17.5,15.8)
> plot(percent~year)
> abline(lm(percent~year))

The scatter plot and regression line are presented in Figure 14.11. Observe that
a linear trend is a reasonable description of the reduction in the percentage of
workers that belong to labor unions in the post World War II period.

Solution (to Question 14.4.2): We compute the averages, standard devia-

tions and the covariance:

> mean.x <- mean(year)

> mean.y <- mean(percent)
> sd.x <- sd(year)
> sd.y <- sd(percent)
> cov.x.y <- cov(year,percent)
> mean.x
[1] 1969.143
> mean.y
[1] 25.84286
> sd.x
[1] 18.27957
> sd.y
[1] 7.574927
> cov.x.y
[1] -135.6738

The average of the variable “year” is 1969.143 and the standard deviation is
18.27957. The average of the variable “percent” is 25.84286 and the standard
deviation is 7.574927. The covariance between the two variables is −135.6738.

Solution (to Question 14.4.3): The slope of the regression line is the ratio
between the covariance and the variance of the explanatory variable. The inter-
cept is the solution of the equation that states that the value of regression line
at the average of the explanatory variable is the average of the response:
274 CHAPTER 14. LINEAR REGRESSION

●
35

● ●
30

●
percent

25

●
20

1950 1960 1970 1980 1990

year

Figure 14.11: Percent of Union Workers

> b <- cov.x.y/sd.x^2

> a <- mean.y - b*mean.x
> a
[1] 825.3845
> b
[1] -0.4060353

We get that the intercept is equal to 825.3845 and the slope is equal to −0.4060353.
In order to validate these figures, let us apply the function “lm” to the data:

> lm(percent~year)

Call:
lm(formula = percent ~ year)

Coefficients:
(Intercept) year
825.384 -0.406
14.5. SOLVED EXERCISES 275

Indeed, we get the same numbers that we got from the manual computation.
Question 14.5. Assume a regression model was fit to some data that de-
scribes the relation between the explanatory variable x and the response y.
Assume that the coefficients of the fitted model are a = 2.5 and b = −1.13, for
the intercept and the slope, respectively. The first 4 observations in the data
are (x1 , y1 ) = (5.5, 3.22), (x2 , y2 ) = (12.13, −6.02), (x3 , y3 ) = (4.2, −8.3), and
(x4 , y4 ) = (6.7, 0.17).
1. What is the estimated expectation of the response for the 4th observation?
2. What is the residual from the regression line for the 4th observation?
Solution (to Question 14.5.1): The estimate for the expected value for the
ith observation is obtained by the evaluation of the expression a + b · xi , where
a and b are the coefficients of the fitted regression model and xi is the value of
the explanatory variable for the ith observation. In our case i = 4 and x4 = 6.7:

a + b · x4 = 2.5 − 1.13 · x4 = 2.5 − 1.13 · 6.7 = −5.071 .

Therefore, the estimate expectation of the response is −5.071.

Solution (to Question 14.5.2): The residual from the regression line is the
difference between the observed value of the response and the estimated expec-
tation of the response. For the 4th observation we have that the observed value
of the response is y4 = 0.17. The estimated expectation was computed in the
previous question. Therefore, the residual from the regression line for the 4th
observation is:

y4 − (a + b · x4 ) = 0.17 − (−5.071) = 5.241 .

Question 14.6. In Chapter 13 we analyzed an example that involved the dif-

ference between fuel consumption in highway and city driving conditions as the
response13 . The explanatory variable was a factor that was produced by split-
ting the cars into two weight groups. In this exercise we would like to revisit this
example. Here we use the weight of the car directly as an explanatory variable.
We also consider the size of the engine as an alternative explanatory variable
and compare between the two regression models.
1. Fit the regression model that uses the variable “curb.weight” as an ex-
planatory variable. Is the slope significantly different than 0? What frac-
tion of the standard deviation of the response is explained by a regression
model involving this variable?
2. Fit the regression model that uses the variable “engine.size” as an ex-
planatory variable. Is the slope significantly different than 0? What frac-
tion of the standard deviation of the response is explained by a regression
model involving this variable?
3. Which of the two models fits the data better?
13 The response was computed using the expression “cars$highway.mpg - cars$city.mpg”
276 CHAPTER 14. LINEAR REGRESSION

● ●
10

10
●● ●●

●●●
●●
●● ●
●●● ●
8

8
●●
●●
●
●●
●●
●●
●●
●●
●●● ●
●●●
● ●●
●●●
●●
●
● ●●
dif.mpg

dif.mpg
● ●
●
●●
●●●
●●
●●
●●
●
●●
●●
●●●
●●●
●●
●●● ● ●● ●●●
●●●
●
●●● ● ●
6

6
●●
●●●●●
●
●●
●
●●
●●●
●●●●●●
●●
●●
●●
●●
●
●●●
●●
●● ● ●
●●
●●
●●
●●●●●
●●●●
●●
●● ●

●●
●●
●●
● ●● ●●●
●● ●● ●●
●
●●● ●● ● ● ●
4

4
●
●● ● ●
● ●
● ●● ● ●

● ●
●
●● ● ● ●
●
2

● ● ● ●● ●

● ●● ● ●
0

1500 2500 3500 100 200 300

cars$curb.weight cars$engine.size

Figure 14.12: Percent of Union Workers

Solution (to Question 14.6.1): We create the response and then fit a model
and apply the summarizing function to the model:

> dif.mpg <- cars$highway.mpg - cars$city.mpg

> summary(lm(dif.mpg ~ cars$curb.weight))

Call:
lm(formula = dif.mpg ~ cars$curb.weight)

Residuals:
Min 1Q Median 3Q Max
-5.4344 -0.7755 0.1633 0.8844 6.3035

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 8.1653491 0.5460856 14.953 < 2e-16 ***
cars$curb.weight -0.0010306 0.0002094 -4.921 1.77e-06 ***
---
14.5. SOLVED EXERCISES 277

Signif. codes: 0 "***" 0.001 "**" 0.01 "*" 0.05 "." 0.1 " " 1

Residual standard error: 1.557 on 203 degrees of freedom

Multiple R-squared: 0.1066, Adjusted R-squared: 0.1022
F-statistic: 24.22 on 1 and 203 DF, p-value: 1.775e-06
The p-value associated with the slope, 1.77 × 10−6 , is much smaller than the
5% threshold proposing a significant (negative) trend. The value of R-squared,
the fraction of the variability of the response that is explained by a regression
model, is 0.1066.
The standard deviation is the square root of the variance. It follows that
the fraction of
√ the standard deviation of the response that is explained by the
regression is 0.1066 = 0.3265.
Following our own advice, we plot the data and the regression model:
> plot(dif.mpg ~ cars$curb.weight)
> abline(lm(dif.mpg ~ cars$curb.weight))
The resulting plot is presented on the left-hand side of Figure 14.12. One may
observe that although there seems to be an overall downward trend, there is
still allot of variability about the line of regression.

Solution (to Question 14.6.2): We now fit and summarize the regression
model with the size of engine as the explanatory variable:
> summary(lm(dif.mpg ~ cars$engine.size))

Call:
lm(formula = dif.mpg ~ cars$engine.size)

Residuals:
Min 1Q Median 3Q Max
-5.7083 -0.7083 0.1889 1.1235 6.1792

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 6.717304 0.359385 18.691 < 2e-16 ***
cars$engine.size -0.009342 0.002691 -3.471 0.000633 ***
---
Signif. codes: 0 "***" 0.001 "**" 0.01 "*" 0.05 "." 0.1 " " 1

Residual standard error: 1.601 on 203 degrees of freedom

Multiple R-squared: 0.05603, Adjusted R-squared: 0.05138
F-statistic: 12.05 on 1 and 203 DF, p-value: 0.0006329
The regression slope is negative. The p-value is 0.000633, which is statistically
significant. The value of R-squared is 0.05603. Consequently, the fraction of the
standard √deviation of the response that is explained by the current regression
model is 0.05603 = 0.2367.
Produce the scatter plot with the line of regression:
> plot(dif.mpg ~ cars$engine.size)
> abline(lm(dif.mpg ~ cars$engine.size))
278 CHAPTER 14. LINEAR REGRESSION

The plot is given on the right-hand side of Figure 14.12. Again, there is vari-
ability about the line of regression.

Solution (to Question 14.6.3): Of the two models, the model that uses the
curb weigh as the explanatory variable explains a larger portion of the variability
in the response. Hence, unless other criteria tells us otherwise, we will prefer
this model over the model that uses the size of engine as an explanatory variable.

14.6 Summary
Glossary
Regression: Relates different variables that are measured on the same sample.
Regression models are used to describe the effect of one of the variables
on the distribution of the other one. The former is called the explanatory
variable and the later is called the response.
Linear Regression: The effect of a numeric explanatory variable on the dis-
tribution of a numeric response is described in terms of a linear trend.
Scatter Plot: A plot that presents the data in a pair of numeric variables. The
axes represents the variables and each point represents an observation.
Intercept: A coefficient of a linear equation. Equals the value of y when the
line crosses the y-axis.
Slope: A coefficient of a linear equation. The change in the value of y for each
unit change in the value of x. A positive slope corresponds to an increasing
line and a negative slope corresponds to a decreasing line.
Covariance: A measures the joint variability of two numeric variables. It is
equal to the sum of the product of the deviations from the mean, divided
by the number of observations minus 1.
Residuals from Regression: The residual differences between the values of
the response for the observation and the estimated expectations of the
response under the regression model (the predicted response).
R-Square: is the difference between 1 and the ratio between the variance of
the residuals from the regression and the variance of the response. Its
value is between 0 and 1 and it represents the fraction of the variability
of the response that is explained by the regression line.

Discuss in the Forum

The topic for discussion in the Forum of Chapter 6 was mathematical models
and how good they should fit reality. In this Forum we would like to return to
the same topic subject, but consider it specifically in the context of statistical
models.
Some statisticians prefer complex models, models that try to fit the data as
closely as one can. Others prefer a simple model. They claim that although
14.6. SUMMARY 279

simpler models are more remote from the data yet they are easier to interpret
and thus provide more insight. What do you think? Which type of model is
better to use?
When formulating your answer to this question you may thing of a situation
that involves inference based on data conducted by yourself for the sack of
others. What would be the best way to report your findings and explain them
to the others?

Formulas:
• A Linear Equation: y = a + b · x.
Sum of products of the deviations
Pn
i=1 (yi −ȳ)(xi −x̄)
• Covariance: = .
Number of values in the sample−1 n−1

• Regression Slope: b = Covariance(x, y)/Var(x).

• Regression Intercept: a = ȳ − bx̄.
• The Regression Model: E(Yi ) = a + b · xi , a and b population parameters.
• Residuals: yi − (a + bxi ), a and b estimated from the data.
Pn 2
• Estimate of Residual Variance: i=1 (yi − (a + bxi )) /(n − 2), a and b
estimated from the data.
Pn Pn
• R-Squared: 1 − i=1 (yi − (a + bxi ))2 / i=1 (yi − ȳ)2 , a and b estimated
from the data.
280 CHAPTER 14. LINEAR REGRESSION
Chapter 15

A Bernoulli Response

15.1 Student Learning Objectives

Chapters 13 and 14 introduced statistical inference that involves a response
and an explanatory variable that may affect the distribution of the response.
In both chapters the response was numeric. The two chapters differed in the
data type of the explanatory variable. In Chapter 13 the explanatory variable
was a factor with two levels that splits the sample into two sub-samples. In
Chapter 14 the explanatory variable was numeric and produced, together with
the response, a linear trend. The aim in this chapter is to consider the case
where the response is a Bernoulli variable. Such a variable may emerge as the
indicator of the occurrence of an event associated with the response or as a
factor with two levels. The explanatory variable is a factor with two levels in
one case or a numerical variable in the other case.
Specifically, when the explanatory variable is a factor with two levels then
we may use the function “prop.test”. This function was used in Chapter 12
for the analysis of the probability of an event in a single sample. Here we use
it in order to compare between two sub-samples. This is similar to the way the
function “t.test” was used for a numeric response for both a single sample and
for the comparison between sub-samples. For the case where the explanatory
variable is numeric we may use the function “glm”, acronym for Generalized
Linear Model, in order to fit an appropriate regression model to the data.
By the end of this chapter, the student should be able to:

• Produce mosaic plots of the response and the explanatory variable.

• Apply the function “prop.test” in order to compare the probability of

an event between two sub-populations

• Define the logistic regression model that relates the probability of an event
in the response to a numeric explanatory variable.

• Fit the logistic regression model to data using the function “glm” and
produce statistical inference on the fitted model.

281
282 CHAPTER 15. A BERNOULLI RESPONSE

15.2 Comparing Sample Proportions

In this chapter we deal with a Bernoulli response. Such a response has two levels,
“TRUE” or “FALSE”1 , and may emerges as the indicator of an event. Else, it may
be associated with a factor with two levels and correspond to the indication of
one of the two levels. Such response was considered in Chapters 11 and 12 where
confidence intervals and tests for the probability of an event where discussed in
the context of a single sample. In this chapter we discuss the investigation of
relations between a response of this form and an explanatory variable.
We start with the case where the explanatory variable is a factor that has
two levels. These levels correspond to two sub-populations (or two settings).
The aim of the analysis is to compare between the two sub-populations (or
between the two settings) the probability of the even.
The discussion in this section is parallel to the discussion in Section 13.3.
That section considered the comparison of the expectation of a numerical re-
sponse between two sub-populations. We denoted these sub-populations a and
b with expectations E(Xa ) and E(Xb ), respectively. The inference used the av-
erage X̄a , which was based on a sub-sample of size na , and the average X̄b ,
which was based on the other sub-sample of size nb . The sub-samples variances
Sa2 and Sb2 participated in the inference as well. The application of a test for
the equality of the expectations and a confidence interval where produced by
the application of the function “t.test” to the data.
The inference problem, which is considered in this chapter, involves an event.
This event is being examined in two different settings that correspond to two
different sub-population a and b. Denote the probabilities of the event in each
of the sub-populations by pa and pb . Our concern is the statistical inference
associated with the comparison of these two probabilities to each other.
Natural estimators of the probabilities are P̂a and P̂b , the sub-samples pro-
portions of occurrence of the event. These estimators are used in order to carry
out the inference. Specifically, we consider here the construction of a confi-
dence interval for the difference pa − pb and a test of the hypothesis that the
probabilities are equal.
The methods for producing the confidence intervals for the difference and
for testing the null hypothesis that the difference is equal to zero are similar
is principle to the methods that were described in Section 13.3 for making
parallel inferences regarding the relations between expectations. However, the
derivations of the tools that are used in the current situation are not identical to
the derivations of the tools that were used there. The main differences between
the two cases is the replacement of the sub-sample averages by the sub-sample
proportions, a difference in the way the standard deviation of the statistics are
estimated, and the application of a continuity correction. We do not discuss in
this chapter the theoretical details associated with the derivations. Instead, we
demonstrate the application of the inference in an example.
The variable “num.of.doors” in the data frame “cars” describes the num-
ber of doors a car has. This variable is a factor with two levels, “two” and
“four”. We treat this variable as a response and investigate its relation to
explanatory variables. In this section the explanatory variable is a factor with
two levels and in the next section it is a numeric variable. Specifically, in this
1 The levels are frequently coded as 1 or 0, “success” or “failure”, or any other pair of levels.
15.2. COMPARING SAMPLE PROPORTIONS 283

section we use the factor “fuel.type” as the explanatory variable. Recall that
this variable identified the type of fuel, diesel or gas, that the car uses. The aim
of the analysis is to compare the proportion of cars with four doors between
cars that run on diesel and cars that run on gas.
Let us first summarize the data in a 2 × 2 frequency table. The function
“table” may be used in order to produce such a table:
> cars <- read.csv("cars.csv")
> table(cars$fuel.type,cars$num.of.doors)

four two
diesel 16 3
gas 98 86
When the function “table” is applied to a combination of two factors then
the output is a table of joint frequencies. Each entry in the table contains the
frequency in the sample of the combination of levels, one from each variable,
that is associated with the entry. For example, there are 16 cars in the data
set that have the level “four” for the variable “num.of.doors” and the level
“diesel” for the variable “fuel.type”. Likewise, there are 3 cars that are
associated with the combination “two” and “diesel”. The total number of
entries to the table is 16 + 3 + 98 + 86 = 203, which is the number of cars in the
data set, minus the two missing values in the variable “num.of.doors”.
A graphical representation of the relation between the two factors can be
obtained using a mosaic plot. This plot is produced when the input to the
function “plot” is a formula where both the response and the explanatory
variables are factors:
> plot(num.of.doors ~ fuel.type,data=cars)
The resulting mosaic plot is presented in Figure 15.1.
The box plot describes the distribution of the explanatory variable and the
distribution of the response for each level of the explanatory variable. In the
current example the explanatory variable is the factor “fuel” that has 2 levels.
The two levels of this variable, “diesel” and “gas”, are given at the x-axis.
A vertical rectangle is associated with each level. These 2 rectangles split the
total area of the square. The total area of the square represents the total
relative frequency (which is equal to 1). Consequently, the area of each rectangle
represents the relative frequency of the associated level of the explanatory factor.
A rectangle associated with a given level of the explanatory variable is further
divided into horizontal sub-rectangles that are associated with the response
factor. In the current example each darker rectangle is associated with the level
“four” of the response “num.of.door” and each brighter rectangle is associated
with the level “two”. The relative area of the horizontal rectangles within each
vertical rectangle represent the relative frequency of the levels of the response
within each subset associated with the level of the explanatory variable.
Looking at the plot one may appreciate the fact that diesel cars are less
frequent than cars that run on gas. The graph also displays the fact that the
relative frequency of cars with four doors among diesel cars is larger than the
relative frequency of four doors cars among cars that run on gas.
The function “prop.test” may be used in order test the hypothesis that, at
the population level, the probability of the level “four” of the response within the
284 CHAPTER 15. A BERNOULLI RESPONSE

1.0
two

0.8
0.6
num.of.doors

four

0.4
0.2
0.0
diesel gas

fuel.type

Figure 15.1: Number of Doors versus Fuel Type

sub-population of diesel cars (the height of the leftmost darker rectangle in the
theoretic mosaic plot that is produced for the entire population) is equal to the
probability of the same level of the response with in the sub-population of cars
that run on gas (the height of the rightmost darker rectangle in that theoretic
mosaic plot). Specifically, let us test the hypothesis that the two probabilities
of the level “four”, one for diesel cars and one for cars that run on gas, are equal
to each other.
The output of the function “table” may serve as the input to the function
“prop.test”2 . The Bernoulli response variable should be the second variable
in the input to the table whereas the explanatory factor is the first variable in
the table. When we apply the test to the data we get the report:

> prop.test(table(cars$fuel.type,cars$num.of.doors))

2 The
function “prop.test” was applied in Section 12.4 in order to test that the probability
of an event is equal to a given value (“p = 0.5” by default). The input to the function was a
pair of numbers: the total number of successes and the sample size. In the current application
the input is a 2 × 2 table. When applied to such input the function carries out a test of the
equality of the probability of the first column between the rows of the table.
15.3. LOGISTIC REGRESSION 285

2-sample test for equality of proportions with continuity correction

data: table(cars$fuel.type, cars$num.of.doors)

X-squared = 5.5021, df = 1, p-value = 0.01899
alternative hypothesis: two.sided
95 percent confidence interval:
0.1013542 0.5176389
sample estimates:
prop 1 prop 2
0.8421053 0.5326087
The two sample proportions of cars with four doors among diesel and gas cars
are presented at the bottom of the report and serve as estimates of the sub-
populations probabilities. Indeed, the relative frequency of cars with four doors
among diesel cars is equal to p̂a = 16/(16 + 3) = 16/19 = 0.8421053. Likewise,
the relative frequency of cars with four doors among cars that ran on gas is
equal to p̂b = 98/(98 + 86) = 98/184 = 0.5326087. The confidence interval
for the difference in the probability of a car with four doors between the two
sub-populations, pa − pb , is reported under the title “95 percent confidence
interval” and is given as [0.1013542, 0.5176389].
The null hypothesis, which is the subject of this test, is H0 : pa = pb . This
hypothesis is tested against the two-sided alternative hypothesis H1 : pa 6= pb .
The test itself is based on a test statistic that obtains the value X-squared =
5.5021. This test statistic corresponds essentially to the deviation between the
estimated value of the parameter (the difference in sub-sample proportions of
the event) and the theoretical value of the parameter (pa − pb = 0). This devi-
ation is divided by the estimated standard deviation and the ratio is squared.
The statistic itself is produced via a continuity correction that makes its null dis-
tribution closer to the limiting chi-square distribution on one degree of freedom.
The p-value is computed based on this limiting chi-square distribution.
Notice that the computed p-value is equal to p-value = 0.01899. This
value is smaller than 0.05. Consequently, the null hypothesis is rejected at the
5% significance level in favor of the alternative hypothesis. This alternative
hypothesis states that the sub-populations probabilities are different from each
other.

15.3 Logistic Regression

In the previous section we considered a Bernoulli response and a factor with
two levels as an explanatory variable. In this section we use a numeric variable
as the explanatory variable. The discussion in this section is parallel to the
discussion in Chapter 14 that presented the topic of linear regression. However,
since the response is not of the same form, it is the indicator of a level of a factor
and not a regular numeric response, then the tools the are used are different.
Instead of using linear regression we use another type of regression that is called
Logistic Regression.
Recall that linear regression involved fitting a straight line to the scatter
plot of data points. This line corresponds to the expectation of the response as
a function of the explanatory variable. The estimated coefficients of this line
are computed from the data and used for inference.
286 CHAPTER 15. A BERNOULLI RESPONSE

1.0
0.8
two

0.6
num.of.doors

0.4
four

0.2
0.0
140 160 170 180 190 210

length

Figure 15.2: Number of Doors Versus Fuel Type

In logistic regression, instead of the consideration of the expectation of a

numerical response, one considers the probability of an event associated with the
response. This probability is treated as a function of the explanatory variable.
Parameters that determine this function are estimated from the data and are
used for inference regarding the relation between the explanatory variable and
the response. Again, we do not discuss the theoretical details involved in the
derivation of logistic regression. Instead, we apply the method to an example.
We consider the factor “num.of.doors” as the response and the probability
of a car with four doors as the probability of the response. The length of the
car will serve as the explanatory variable. Measurements of lengths of the cars
are stored in the variable “length” in the data frame “cars”.
First, let us plot the relation between the response and the explanatory
variable:
> plot(num.of.doors ~ length,data=cars)
The plot that is produced by the given expression is displayed in Figure 15.2.
It is a type of a mosaic plot and it is produced when the input to the function
“plot” is a formula with a factor as a response and a numeric variable as the
15.3. LOGISTIC REGRESSION 287

70
60
50
Frequency

40
30
20
10
0

140 150 160 170 180 190 200 210

length

Figure 15.3: Histogram of the Length of Cars

explanatory variable. The plot presents, for interval levels of the explanatory
variable, the relative frequencies of each interval. It also presents the relative fre-
quency of the levels of the response within each interval level of the explanatory
variable.
In order to get a better understanding of the meaning of the given mosaic
plot one may consider the histogram of the explanatory variable. This histogram
is presented in Figure 15.3. The histogram involves the partition of the range of
variable length into intervals. These interval are the basis for rectangles. The
height of the rectangles represent the frequency of cars with lengths that fall in
the given interval.
The mosaic plot in Figure 15.2 is constructed on the basis of this histogram.
The x-axis in this plot corresponds to the explanatory variable “length”. The
total area of the square in the plot is divided between 7 vertical rectangles. These
vertical rectangles correspond to the 7 rectangles in the histogram of Figure 15.3,
turn on their sides. Hence, the width of each rectangle in Figure 15.2 correspond
to the hight of the parallel rectangle in the histogram. Consequently, the area
of the vertical rectangles in the mosaic plot represents the relative frequency of
the associated interval of values of the explanatory variable.
288 CHAPTER 15. A BERNOULLI RESPONSE

The rectangle that is associated with each interval of values of the explana-
tory variable is further divided into horizontal sub-rectangles that are associated
with the response factor. In the current example each darker rectangle is asso-
ciated with the level “four” of the response “num.of.door” and each brighter
rectangle is associated with the level “two”. The relative area of the horizontal
rectangles within each vertical rectangle represent the relative frequency of the
levels of the response within each interval of values of the explanatory variable.
From the examination of the mosaic plot one may identify relations between
the explanatory variable and the relative frequency of an identified level of the
response. In the current example one may observe that the relative frequency
of the cars with four doors is, overall, increasing with the increase in the length
of cars.
Logistic regression is a method for the investigation of relations between the
probability of an event and explanatory variables. Specifically, we use it here
for making inference on the number of doors as a response and the length of the
car as the explanatory variable.
Statistical inference requires a statistical model. The statistical model in
logistic regression relates the probability pi , the probability of the event for
observation i, to xi , the value of the response for that observation. The relation
between the two in given by the formula:

ea+b·xi
pi = ,
1 + ea+b·xi
where a and b are coefficients common to all observations. Equivalently, one
may write the same relation in the form:

log(pi /[1 − pi ]) = a + b · xi ,

that states that the relation between a (function of) the probability of the event
and the explanatory variable is a linear trend.
One may fit the logistic regression to the data and test the null hypothesis
by the use of the function “glm”:

> fit.doors <- glm(num.of.doors=="four"~length,

+ family=binomial,data=cars)
> summary(fit.doors)

Call:
glm(formula = num.of.doors == "four" ~ length, family = binomial,
data = cars)

Deviance Residuals:
Min 1Q Median 3Q Max
-2.1646 -1.1292 0.5688 1.0240 1.6673

Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -13.14767 2.58693 -5.082 3.73e-07 ***
length 0.07726 0.01495 5.168 2.37e-07 ***
---
15.4. SOLVED EXERCISES 289

Signif. codes: 0 "***" 0.001 "**" 0.01 "*" 0.05 "." 0.1 " " 1

(Dispersion parameter for binomial family taken to be 1)

Null deviance: 278.33 on 202 degrees of freedom

Residual deviance: 243.96 on 201 degrees of freedom
(2 observations deleted due to missingness)
AIC: 247.96

Number of Fisher Scoring iterations: 3

Generally, the function “glm” can be used in order to fit regression models in
cases where the distribution of the response has special forms. Specifically, when
the argument “family=binomial” is used then the model that is being used in
the model of logistic regression. The formula that is used in the function involves
a response and an explanatory variable. The response may be a sequence with
logical “TRUE” or “FALSE” values as in the example3 . Alternatively, it may be
a sequence with “1” or “0” values, “1” corresponding to the event occurring to
the subject and “0” corresponding to the event not occurring. The argument
“data=cars” is used in order to inform the function that the variables are
located in the given data frame.
The “glm” function is applied to the data and the fitted model is stored in
the object “fit.doors”.
A report is produced when the function “summary” is applied to the fitted
model. Notice the similarities and the differences between the report presented
here and the reports for linear regression that are presented in Chapter 14. Both
reports contain estimates of the coefficients a and b and tests for the equality of
these coefficients to zero. When the coefficient b, the coefficient that represents
the slope, is equal to 0 then the probability of the event and the explanatory
variable are unrelated. In the current case we may note that the null hypothesis
H0 : b = 0, the hypothesis that claims that there is no relation between the
explanatory variable and the response, is clearly rejected (p-value 2.37 × 10−7 ).
The estimated values of the coefficients are −13.14767 for the intercept a
and 0.07726 for the slope b. One may produce confidence intervals for these
coefficients by the application of the function “confint” to the fitted model:

> confint(fit.doors)
Waiting for profiling to be done...
2.5 % 97.5 %
(Intercept) -18.50384373 -8.3180877
length 0.04938358 0.1082429

15.4 Solved Exercises

Question 15.1. This exercise deals with a comparison between Mediterranean
diet and low-fat diet recommended by the American Heart Association in the
3 The response is the output of the expression “num.of.doors=="four"”. This expression

produces logical values. “TRUE” when the car has 4 doors and “FALSE” when it has 2 doors.
290 CHAPTER 15. A BERNOULLI RESPONSE

context of risks for illness or death among patients that survived a heart attack4 .
This case study is taken from the Rice Virtual Lab in Statistics. More details
on this case study can be found in the case study “Mediterranean Diet and
Health” that is presented in that site.
The subjects, 605 survivors of a heart attack, were randomly assigned fol-
low either (1) a diet close to the “prudent diet step 1” of the American Heart
Association (AHA) or (2) a Mediterranean-type diet consisting of more bread
and cereals, more fresh fruit and vegetables, more grains, more fish, fewer deli-
catessen food, less meat.
The subjects‘ diet and health condition were monitored over a period of four-
year. Information regarding deaths, development of cancer or the development
of non-fatal illnesses was collected. The information from this study is stored in
the file “diet.csv”. The file “diet.csv” contains two factors: “health” that
describes the condition of the subject, either healthy, suffering from a non-fatal
illness, suffering from cancer, or dead; and the “type” that describes the type
of diet, either Mediterranean or the diet recommended by the AHA. The file
can be found on the internet at http://pluto.huji.ac.il/~msby/StatThink/
Datasets/diet.csv. Answer the following questions based on the data in the
file:
1. Produce a frequency table of the two variable. Read off from the table the
number of healthy subjects that are using the Mediterranean diet and the
number of healthy subjects that are using the diet recommended by the
AHA.
2. Test the null hypothesis that the probability of keeping healthy following
an heart attack is the same for those that use the Mediterranean diet and
for those that use the diet recommended by the AHA. Use a two-sided
alternative and a 5% significance level.
3. Compute a 95% confidence interval for the difference between the two
probabilities of keeping healthy.
Solution (to Question 15.1.1): First we save the file “diet.csv” in the work-
ing directory of R and read it’s content. Then we apply the function “table”
to the two variables in the file in order to obtain the requested frequency table:
> diet <- read.csv("diet.csv")
> table(diet$health,diet$type)

aha med
cancer 15 7
death 24 14
healthy 239 273
illness 25 8
The resulting table has two columns and 4 rows. The third row corresponds to
healthy subjects. Of these, 239 subjects used the AHA recommended diet and
273 used the Mediterranean diet. We may also plot this data using a mosaic
plot:
4 De Lorgeril, M., Salen, P., Martin, J., Monjaud, I., Boucher, P., Mamelle, N. (1998).

Mediterranean Dietary pattern in a Randomized Trial. Archives of Internal Medicine, 158,

1181-1187.
15.4. SOLVED EXERCISES 291

1.0
illness

0.8
healthy

0.6
health

0.4
death

0.2
cancer

0.0
aha med

type

Figure 15.4: Health Condition Versus Type of Diet

> plot(health~type,data=diet)
The mosaic plot produced by the function “plot” is presented in Figure 15.4.
Examining this plot one may appreciate the fact that the vast majority of the
subjects were healthy and the relative proportion of healthy subjects among
users of the Mediterranean diet is higher than the relative proportion among
users of the AHA recommended diet.

Solution (to Question 15.1.2): In order to test the hypothesis that the
probability of keeping healthy following an heart attack is the same for those
that use the Mediterranean diet and for those that use the diet recommended by
the AHA we create a 2 × 2. This table compares the response of being healthy
or not to the type of diet as an explanatory variable. A sequence with logical
components, “TRUE” for healthy and “FALSE” for not, is used as the response.
Such a sequence is produced via the expression “diet$health=="healthy"”.
The table may serve as input to the function “prop.test”:
> table(diet$health=="healthy",diet$type)
292 CHAPTER 15. A BERNOULLI RESPONSE

aha med
FALSE 64 29
TRUE 239 273

> prop.test(table(diet$health=="healthy",diet$type))

2-sample test for equality of proportions with continuity correction

data: table(diet$health == "healthy", diet$type)

X-squared = 14.5554, df = 1, p-value = 0.0001361
alternative hypothesis: two.sided
95 percent confidence interval:
0.1114300 0.3313203
sample estimates:
prop 1 prop 2
0.6881720 0.4667969
The function “prop.test” conducts the test that compares between the prob-
abilities of keeping healthy. In particular, the computed p-value for the test is
0.0001361, which is less than 0.05. Therefore, we reject the null hypothesis that
both diets have the same effect on the chances of remaining healthy following
an heart attack.

Solution (to Question 15.1.3): The confidence interval for the difference
in probabilities is equal to [0.1114300, 0.3313203]. The point estimation of the
difference between the probabilities is p̂a − p̂b = 0.6881720 − 0.4667969 ≈ 0.22 in
favor of a Mediterranean diet. The confidence interval proposes that a difference
as low as 0.11 or as high as 0.33 are not excluded by the data.

Question 15.2. Cushing’s syndrome disorder results from a tumor (adenoma)

in the pituitary gland that causes the production of high levels of cortisol. The
symptoms of the syndrome are the consequence of the elevated levels of this
steroid hormone in the blood. The syndrome was first described by Harvey
Cushing in 1932.
The file “coshings.csv" contains information on 27 patients that suffer from
Cushing’s syndrome5 . The three variables in the file are “tetra”, “pregn”,
and “type”. The factor “type” describes the underlying type of syndrome,
coded as “a”, (adenoma), ”b” (bilateral hyperplasia), “c” (carcinoma) or “u”
for unknown. The variable “tetra” describe the level of urinary excretion rate
(mg/24hr) of Tetrahydrocortisone, a type of steroid, and the variable “pregn”
describes urinary excretion rate (mg/24hr) of Pregnanetriol, another type of
steroid. The file can be found on the internet at http://pluto.huji.ac.il/
~msby/StatThink/Datasets/coshings.csv. Answer the following questions
based on the information in this file:
1. Plot the histogram of the variable “tetra” and the mosaic plot that de-
scribes the relation between the variable “type” as a response and the
variable “tetra”. What is the information that is conveyed by the second
vertical triangle from the right (the third from the left) in the mosaic plot.
5 The source of the data is the data file “Cushings” from the package “MASS” in R.
15.4. SOLVED EXERCISES 293

Histogram of cushings$tetra

1.0
u
c

15
0.8 0.6
Frequency

10
b
type

0.4

5
0.2
a

0
0.0

0 10 20 0 20 40 60

tetra cushings$tetra

Figure 15.5: Health Condition Versus Type of Diet

2. Test the null hypothesis that there is no relation between the variable
“tetra” as an explanatory variable and the indicator of the type being
equal to “b” as a response. Compute a confidence interval for the param-
eter that describes the relation.
3. Repeat the analysis from 2 using only the observations for which the type
is known. (Hint: you may fit the model to the required subset by the
inclusion of the argument “subset=(type!="u")” in the function that
fits the model.) Which of the analysis do you think is more appropriate?
Solution (to Question 15.2.1): We save the data of the file in a data frame
by the name “cushings”, produce a mosaic plot with the function “plot” and
an histogram with the function “hist”:
> cushings <- read.csv("cushings.csv")
> plot(type~tetra,data=cushings)
> hist(cushings$tetra)
The mosaic plot describes the distribution of the 4 levels of the response within
the different intervals of values of the explanatory variable. The intervals coin-
294 CHAPTER 15. A BERNOULLI RESPONSE

cide with the intervals that are used in the construction of the histogram. In
particular, the third vertical rectangle from the left in the mosaic is associated
with the third interval from the left in the histogram6 . This interval is asso-
ciated with the range of values between 20 and 30. The height of the given
interval in the histogram is 2, which is the number of patients with “terta”
levels that belong to the interval.
There are 4 shades of grey in the first vertical rectangle from the left. Each
shade is associated with a different level of the response. The lightest shade
of grey, the upmost one, is associated with the level “u”. Notice that this is
also the shade of grey of the entire third vertical rectangle from the left. The
conclusion is that the 2 patients that are associated with this rectangle have
Tetrahydrocortisone levels between 2 and 30 and have an unknown type of
syndrome.

Solution (to Question 15.2.2): We fit the logistic regression to the en-
tire data in the data frame “cushings” using the function “glm”, with the
“family=binomial” argument. The response is the indicator that the type is
equal to “b”. The fitted model is saved in an object called “cushings.fit.all”.
The application of the function “summary” to the fitted model produces a report
that includes the test of the hypothesis of interest:
> cushings.fit.all <- glm((type=="b")~tetra,family=binomial,
+ data=cushings)
> summary(cushings.fit.all)

Call:
glm(formula = (type == "b") ~ tetra, family = binomial,
data = cushings)

Deviance Residuals:
Min 1Q Median 3Q Max
-1.0924 -1.0461 -0.8652 1.3427 1.5182

Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.12304 0.61330 -0.201 0.841
tetra -0.04220 0.05213 -0.809 0.418

(Dispersion parameter for binomial family taken to be 1)

Null deviance: 35.594 on 26 degrees of freedom

Residual deviance: 34.739 on 25 degrees of freedom
AIC: 38.739

Number of Fisher Scoring iterations: 4

The test of interest examines the coefficient that is associated wit the explana-
tory variable “tetra”. The estimated value of this parameter is −0.04220. The
6 This is also the third interval from the left in the histogram. However, since the second

and third intervals, counting from the right, in the histogram are empty, it turns out that the
given interval is the second rectangle from the right in the mosaic plot.
15.4. SOLVED EXERCISES 295

p-value for testing that the coefficient is 0 is equal to 0.418. Consequently, since
the p-value is larger than 0.05, we do not reject the null hypothesis that states
that the response and the explanatory variable are statistically unrelated.
Confidence intervals may be computed by applying the function “confint”
to the fitted model:
> confint(cushings.fit.all)
Waiting for profiling to be done...
2.5 % 97.5 %
(Intercept) -1.2955624 1.18118256
tetra -0.1776113 0.04016772
Specifically, the confidence interval for the coefficient that is associated with the
explanatory variable is equal to [−0.1776113, 0.04016772]

Solution (to Question 15.2.3): If we want to fit the logistic model to a

partial subset of the data, say all the observations with values of the response
other that “u”, we may apply the argument “subset”7 . Specifically, adding
the expression “subset=(type!="u")” would do the job8 . We repeat the same
analysis as before. The only difference is the addition of the given expression
to the function that fits the model to the data. The fitted model is saved in an
object we call “cushings.fit.known”:
> cushings.fit.known <- glm((type=="b")~tetra,family=binomial,
+ data=cushings,subset=(type!="u"))
> summary(cushings.fit.known)

Call:
glm(formula = (type == "b") ~ tetra, family = binomial,
data = cushings, subset = (type != "u"))

Deviance Residuals:
Min 1Q Median 3Q Max
-1.2078 -1.1865 -0.7548 1.2033 1.2791

Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) 0.11457 0.59947 0.191 0.848
tetra -0.02276 0.04586 -0.496 0.620

(Dispersion parameter for binomial family taken to be 1)

Null deviance: 29.065 on 20 degrees of freedom

Residual deviance: 28.789 on 19 degrees of freedom
AIC: 32.789

7 This argument may be used in other functions. For example, it may be used in the

function “lm” that fits the linear regression.

8 The value of the argument “subset” is a sequence with logical components that indicate

which of the observations to include in the analysis. This sequence is formed with the aid of
“!=”, which corresponds to the relation “not equal to”. The expression “type!="u"” indicates
all observations with a “type” value not equal to “u”.
296 CHAPTER 15. A BERNOULLI RESPONSE

Number of Fisher Scoring iterations: 4

The estimated value of the coefficient when considering only subject with a
known type of the syndrome is slightly changed to −0.02276. The new p-value,
which is equal to 0.620, is larger than 0.05. Hence, yet again, we do not reject
the null hypothesis.

> confint(cushings.fit.known)
Waiting for profiling to be done...
2.5 % 97.5 %
(Intercept) -1.0519135 1.40515473
tetra -0.1537617 0.06279923

For the modified confidence interval we apply the function “confint”. We get
now [−0.1537617, 0.06279923] as a confidence interval for the coefficient of the
explanatory variable.
We started with the fitting the model to all the observations. Here we use
only the observations for which the type of the syndrome is known. The practical
implication of using all observations in the fit is equivalent to announcing that
the type of the syndrome for observations of an unknown type is not type “b”.
This is not appropriate and may introduce bias, since the type may well be “b”.
It is more appropriate to treat the observations associated with the level “u”
as missing observations and to delete them from the analysis. This approach is
the approach that was used in the second analysis.

Glossary
Mosaic Plot: A plot that describes the relation between a response factor and
an explanatory variable. Vertical rectangles represent the distribution of
the explanatory variable. Horizontal rectangles within the vertical ones
represent the distribution of the response.

Logistic Regression: A type of regression that relates between an explanatory

variable and a response of the form of an indicator of an event.

Discuss in the forum

In the description of the statistical models that relate one variable to the other
we used terms that suggest a causality relation. One variable was called the
“explanatory variable” and the other was called the “response”. One may get
the impression that the explanatory variable is the cause for the statistical be-
havior of the response. In negation to this interpretation, some say that all that
statistics does is to examine the joint distribution of the variables, but casual-
ity cannot be inferred from the fact that two variables are statistically related.
What do you think? Can statistical reasoning be used in the determination of
casuality?
As part of your answer in may be useful to consider a specific situation
where the determination of casuality is required. Can any of the tools that were
discussed in the book be used in a meaningful way to aid in the process of such
determination?
15.4. SOLVED EXERCISES 297

Notice that the last 3 chapters dealt with statistical models that related
an explanatory variable to a response. We considered tools that can be used
when both variable are factors and when both are numeric. Other tools may
be used when one of the variables is a factor and the other is numeric. An
analysis that involves one variable as the response and the other as explanatory
variable can be reversed, possibly using a different statistical tool, with the roles
of the variables exchanged. Usually, a significant statistical finding will be still
significant when the roles of a response and an explanatory variable are reversed.

Formulas:
ea+b·xi
• Logistic Regression, (Probability): pi = 1+ea+b·xi
.

• Logistic Regression, (Predictor): log(pi /[1 − pi ]) = a + b · xi .

298 CHAPTER 15. A BERNOULLI RESPONSE
Chapter 16

Case Studies

16.1 Student Learning Objective

This chapter concludes this book. We start with a short review of the topics that
were discussed in the second part of the book, the part that dealt with statistical
inference. The main part of the chapter involves the statistical analysis of 2 case
studies. The tools that will be used for the analysis are those that were discussed
in the book. We close this chapter and this book with some concluding remarks.
By the end of this chapter, the student should be able to:

• Review the concepts and methods for statistical inference that were pre-
sented in the second part of the book.

• Apply these methods to requirements of the analysis of real data.

• Develop a resolve to learn more statistics.

16.2 A Review
The second part of the book dealt with statistical inference; the science of
making general statement on an entire population on the basis of data from a
sample. The basis for the statements are theoretical models that produce the
sampling distribution. Procedures for making the inference are evaluated based
on their properties in the context of this sampling distribution. Procedures with
desirable properties are applied to the data. One may attach to the output of
this application summaries that describe these theoretical properties.
In particular, we dealt with two forms of making inference. One form was
estimation and the other was hypothesis testing. The goal in estimation is
to determine the value of a parameter in the population. Point estimates or
confidence intervals may be used in order to fulfill this goal. The properties of
point estimators may be assessed using the mean square error (MSE) and the
properties of the confidence interval may be assessed using the confidence level.
The target in hypotheses testing is to decide between two competing hy-
pothesis. These hypotheses are formulated in terms of population parameters.
The decision rule is called a statistical test and is constructed with the aid of a
test statistic and a rejection region. The default hypothesis among the two, is

299
300 CHAPTER 16. CASE STUDIES

rejected if the test statistic falls in the rejection region. The major property a
test must possess is a bound on the probability of a Type I error, the probability
of erroneously rejecting the null hypothesis. This restriction is called the signif-
icance level of the test. A test may also be assessed in terms of it’s statistical
power, the probability of rightfully rejecting the null hypothesis.
Estimation and testing were applied in the context of single measurements
and for the investigation of the relations between a pair of measurements. For
single measurements we considered both numeric variables and factors. For
numeric variables one may attempt to conduct inference on the expectation
and/or the variance. For factors we considered the estimation of the probability
of obtaining a level, or, more generally, the probability of the occurrence of an
event.
We introduced statistical models that may be used to describe the relations
between variables. One of the variables was designated as the response. The
other variable, the explanatory variable, is identified as a variable which may
affect the distribution of the response. Specifically, we considered numeric vari-
ables and factors that have two levels. If the explanatory variable is a factor with
two levels then the analysis reduces to the comparison of two sub-populations,
each one associated with a level. If the explanatory variable is numeric then a
regression model may be applied, either linear or logistic regression, depending
on the type of the response.
The foundations of statistical inference are the assumption that we make in
the form of statistical models. These models attempt to reflect reality. However,
one is advised to apply healthy skepticism when using the models. First, one
should be aware what the assumptions are. Then one should ask oneself how
reasonable are these assumption in the context of the specific analysis. Finally,
one should check as much as one can the validity of the assumptions in light of
the information at hand. It is useful to plot the data and compare the plot to
the assumptions of the model.

16.3 Case Studies

Let us apply the methods that were introduced throughout the book to two
examples of data analysis. Both examples are taken from the case studies of the
Rice Virtual Lab in Statistics can be found in their Case Studies section. The
analysis of these case studies may involve any of the tools that were described
in the second part of the book (and some from the first part). It may be useful
to read again Chapters 9–15 before reading the case studies.

16.3.1 Physicians’ Reactions to the Size of a Patient

Overweight and obesity is common in many of the developed contrives. In some
cultures, obese individuals face discrimination in employment, education, and
relationship contexts. The current research, conducted by Mikki Hebl and Jing-
ping Xu1 , examines physicians’ attitude toward overweight and obese patients
in comparison to their attitude toward patients who are not overweight.
1 Hebl, M. and Xu, J. (2001). Weighing the care: Physicians’ reactions to the size of a

patient. International Journal of Obesity, 25, 1246-1252.

16.3. CASE STUDIES 301

The experiment included a total of 122 primary care physicians affiliated

with one of three major hospitals in the Texas Medical Center of Houston.
These physicians were sent a packet containing a medical chart similar to the
one they view upon seeing a patient. This chart portrayed a patient who was
displaying symptoms of a migraine headache but was otherwise healthy. Two
variables (the gender and the weight of the patient) were manipulated across six
different versions of the medical charts. The weight of the patient, described in
terms of Body Mass Index (BMI), was average (BMI = 23), overweight (BMI
= 30), or obese (BMI = 36). Physicians were randomly assigned to receive one
of the six charts, and were asked to look over the chart carefully and complete
two medical forms. The first form asked physicians which of 42 tests they would
recommend giving to the patient. The second form asked physicians to indicate
how much time they believed they would spend with the patient, and to describe
the reactions that they would have toward this patient.
In this presentation, only the question on how much time the physicians
believed they would spend with the patient is analyzed. Although three patient
weight conditions were used in the study (average, overweight, and obese) only
the average and overweight conditions will be analyzed. Therefore, there are two
levels of patient weight (average and overweight) and one dependent variable
(time spent).
The data for the given collection of responses from 72 primary care physicians
is stored in the file “discriminate.csv”2 . We start by reading the content of
the file into a data frame by the name “patient” and presenting the summary
of the variables:

> patient <- read.csv("discriminate.csv")

> summary(patient)
weight time
BMI=23:33 Min. : 5.00
BMI=30:38 1st Qu.:20.00
Median :30.00
Mean :27.82
3rd Qu.:30.00
Max. :60.00

Observe that of the 72 “patients”, 38 are overweight and 33 have an average

weight. The time spend with the patient, as predicted by physicians, is dis-
tributed between 5 minutes and 1 hour, with a average of 27.82 minutes and a
median of 30 minutes.
It is a good practice to have a look at the data before doing the analysis.
In this examination on should see that the numbers make sense and one should
identify special features of the data. Even in this very simple example we may
want to have a look at the histogram of the variable “time”:

> hist(patient$time)

The histogram produced by the given expression is presented in Figure 16.1.

A feature in this plot that catches attention is the fact that there is a high
concventration of values in the interval between 25 and 30. Together with the
2 The file can be found on the internet at http://pluto.huji.ac.il/ msby/StatThink/
~
Datasets/discriminate.csv.
302 CHAPTER 16. CASE STUDIES

Histogram of patient$time

30
25
20
Frequency

15
10
5
0

10 20 30 40 50 60

patient$time

Figure 16.1: Histogram of “time”

fact that the median is equal to 30, one may suspect that, as a matter of fact,
a large numeber of the values are actually equal to 30. Indeed, let us produce a
table of the response:
> table(patient$time)

5 15 20 25 30 40 45 50 60
1 10 15 3 30 4 5 2 1
Notice that 30 of the 72 physicians marked “30” as the time they expect to
spend with the patient. This is the middle value in the range, and may just be
the default value one marks if one just needs to complete a form and do not
really place much importance to the question that was asked.
The goal of the analysis is to examine the relation between overweigh and
the Doctor’s response. The explanatory variable is a factor with two levels. The
response is numeric. A natural tool to use in order to test this hypothesis is the
t-test, which is implemented with the function “t.test”.
First we plot the relation between the response and the explanatory variable
and then we apply the test:
16.3. CASE STUDIES 303

60

●
50
40
30
20
10

BMI=23 BMI=30

Figure 16.2: Time Versus Weight Group

> boxplot(time~weight,data=patient)
> t.test(time~weight,data=patient)

Welch Two Sample t-test

data: time by weight

t = 2.8516, df = 67.174, p-value = 0.005774
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
1.988532 11.265056
sample estimates:
mean in group BMI=23 mean in group BMI=30
31.36364 24.73684

The box plots that describe the distribution of the response for each level of
the explanatory variable are presented in Figure 16.2. Nothing seems problem-
atic in this plot. The two distributions, as they are reflected in the box plots,
look fairly symmetric.
304 CHAPTER 16. CASE STUDIES

1.0
0.8
factor(patient$time >= 30)

TRUE

0.6
0.4
0.2
FALSE

0.0
BMI=23 BMI=30

weight

Figure 16.3: At Least 30 Minutes Versus Weight Group

When we consider the report that produced by the function “t.test” we

may observe that the p-value is equal to 0.005774. This p-value is computed in
testing the null hypothesis that the expectation of the response for both types
of patients are equal against the two sided alternative. Since the p-value is less
than 0.05 we do reject the null hypothesis.
The estimated value of the difference between the expectation of the response
for a patient with BMI=23 and a patient with BMI=30 is 31.36364−24.73684 ≈
6.63 minutes. The confidence interval is (approximately) equal to [1.99, 11.27].
Hence, it looks as if the physicians expect to spend more time with the average
weight patients.
After analyzing the effect of the explanatory variable on the expectation of
the response one may want to examine the presence, or lack thereof, of such
effect on the variance of the response. Towards that end, one may use the
function ”var.test”:

> var.test(time~weight,data=patient)

F test to compare two variances

16.3. CASE STUDIES 305

data: time by weight

F = 1.0443, num df = 32, denom df = 37, p-value = 0.893
alternative hypothesis: true ratio of variances is not equal to 1
95 percent confidence interval:
0.5333405 2.0797269
sample estimates:
ratio of variances
1.044316

In this test we do not reject the null hypothesis that the two variances of the
response are equal since the p-value is larger than 0.05. The sample variances are
almost equal to each other (their ratio is 1.044316), with a confidence interval
for the ration that essentially ranges between 1/2 and 2.
The production of p-values and confidence intervals is just one aspect in the
analysis of data. Another aspect, which typically is much more time consum-
ing and requires experience and healthy skepticism is the examination of the
assumptions that are used in order to produce the p-values and the confidence
intervals. A clear violation of the assumptions may warn the statistician that
perhaps the computed nominal quantities do not represent the actual statistical
properties of the tools that were applied.
In this case, we have noticed the high concentration of the response at the
value “30”. What is the situation when we split the sample between the two
levels of the explanatory variable? Let us apply the function “table” once more,
this time with the explanatory variable included:
> table(patient$time,patient$weight)

BMI=23 BMI=30
5 0 1
15 2 8
20 6 9
25 1 2
30 14 16
40 4 0
45 4 1
50 2 0
60 0 1

Not surprisingly, there is still high concentration at that level “30”. But one
can see that only 2 of the responses of the “BMI=30” group are above that value
in comparison to a much more symmetric distribution of responses for the other
group.
The simulations of the significance level of the one-sample t-test for an Ex-
ponential response that were conducted in Question 12.2 may cast some doubt
on how trustworthy are nominal p-values of the t-test when the measurements
are skewed. The skewness of the response for the group “BMI=30” is a reason to
be worry.
We may consider a different test, which is more robust, in order to validate
the significance of our findings. For example, we may turn the response into
a factor by setting a level for values larger or equal to “30” and a different
306 CHAPTER 16. CASE STUDIES

level for values less than “30”. The relation between the new response and the
explanatory variable can be examined with the function “prop.test”. We first
plot and then test:
> plot(factor(patient$time>=30)~weight,data=patient)
> prop.test(table(patient$time>=30,patient$weight))

2-sample test for equality of proportions with continuity correction

data: table(patient$time >= 30, patient$weight)

X-squared = 3.7098, df = 1, p-value = 0.05409
alternative hypothesis: two.sided
95 percent confidence interval:
-0.515508798 -0.006658689
sample estimates:
prop 1 prop 2
0.3103448 0.5714286
The mosaic plot that presents the relation between the explanatory variable and
the new factor is given in Figure 16.3. The level “TRUE” is associated with a
value of the predicted time spent with the patient being 30 minutes or more.
The level “FALSE” is associated with a prediction of less than 30 minutes.
The computed p-value is equal to 0.05409, that almost reaches the signifi-
cance level of 5%3 . Notice that the probabilities that are being estimated by
the function are the probabilities of the level “FALSE”. Overall, one may see
the outcome of this test as supporting evidence for the conclusion of the t-test.
However, the p-value provided by the t-test may over emphasize the evidence in
the data for a significant difference in the physician attitude towards overweight
patients.

16.3.2 Physical Strength and Job Performance

The next case study involves an attempt to develop a measure of physical ability
that is easy and quick to administer, does not risk injury, and is related to how
well a person performs the actual job. The current example is based on study
by Blakely et al. 4 , published in the journal Personnel Psychology.
There are a number of very important jobs that require, in addition to
cognitive skills, a significant amount of strength to be able to perform at a high
level. Construction worker, electrician and auto mechanic, all require strength
in order to carry out critical components of their job. An interesting applied
problem is how to select the best candidates from amongst a group of applicants
for physically demanding jobs in a safe and a cost effective way.
The data presented in this case study, and may be used for the development
of a method for selection among candidates, were collected from 147 individuals
3 One may propose splinting the response into two groups, with one group be-

ing associated with values of “time” strictly larger than 30 minutes and the other
with values less or equal to 30. The resulting p-value from the expression
“prop.test(table(patient$time>30,patient$weight))” is 0.01276. However, the number
of subjects in one of the cells of the table is equal only to 2, which is problematic in the
context of the Normal approximation that is used by this test.
4 Blakley, B.A., Quiñones, M.A., Crawford, M.S., and Jago, I.A. (1994). The validity of

isometric strength tests. Personnel Psychology, 47, 247-274.

16.3. CASE STUDIES 307

Histogram of job$ratings Histogram of job$sims

30
25
Frequency

Frequency

20
15

10
5
0

0
20 30 40 50 60 −4 −2 0 2 4 6

job$ratings job$sims

Histogram of job$grip Histogram of job$arm

40

25
Frequency

Frequency
30

15
20
10

5
0

50 100 150 200 20 40 60 80 120

job$grip job$arm

Figure 16.4: Histograms of Variables

working in physically demanding jobs. Two measures of strength were gathered

from each participant. These included grip and arm strength. A piece of equip-
ment known as the Jackson Evaluation System (JES) was used to collect the
strength data. The JES can be configured to measure the strength of a number
of muscle groups. In this study, grip strength and arm strength were measured.
The outcomes of these measurements were summarized in two scores of physical
strength called “grip” and “arm”.
Two separate measures of job performance are presented in this case study.
First, the supervisors for each of the participants were asked to rate how well
their employee(s) perform on the physical aspects of their jobs. This measure
is summarizes in the variable “ratings”. Second, simulations of physically de-
manding work tasks were developed. The summary score of these simulations
are given in the variable “sims”. Higher values of either measures of perfor-
mance indicates better performance.
The data for the 4 variables and 147 observations is stored in “job.csv”5 .

5 The file can be found on the internet at http://pluto.huji.ac.il/ msby/StatThink/

~
Datasets/job.csv.
308 CHAPTER 16. CASE STUDIES

We start by reading the content of the file into a data frame by the name “job”,
presenting a summary of the variables, and their histograms:

> job <- read.csv("job.csv")

> summary(job)
grip arm ratings sims
Min. : 29.0 Min. : 19.00 Min. :21.60 Min. :-4.1700
1st Qu.: 94.0 1st Qu.: 64.50 1st Qu.:34.80 1st Qu.:-0.9650
Median :111.0 Median : 81.50 Median :41.30 Median : 0.1600
Mean :110.2 Mean : 78.75 Mean :41.01 Mean : 0.2018
3rd Qu.:124.5 3rd Qu.: 94.00 3rd Qu.:47.70 3rd Qu.: 1.0700
Max. :189.0 Max. :132.00 Max. :57.20 Max. : 5.1700
> hist(job$grip)
> hist(job$arm)
> hist(job$ratings)
> hist(job$sims)

All variables are numeric. Their histograms are presented in Figure 16.5. Exam-
ination of the 4 summaries and histograms does not produce interest findings.
All variables are, more or less, symmetric with the distribution of the variable
“ratings” tending perhaps to be more uniform then the other three.
The main analyses of interest are attempts to relate the two measures of
physical strength “grip” and “arm” with the two measures of job performance,
“ratings” and “sims”. A natural tool to consider in this context is a linear
regression analysis that relates a measure of physical strength as an explanatory
variable to a measure of job performance as a response.
Let us consider the variable “sims” as a response. The first step is to plot
a scatter plot of the response and explanatory variable, for both explanatory
variables. To the scatter plot we add the line of regression. In order to add the
regression line we fit the regression model with the function “lm” and then apply
the function “abline” to the fitted model. The plot for the relation between
the response and the variable “grip” is produced by the code:

> plot(sims~grip,data=job)
> sims.grip <- lm(sims~grip,data=job)
> abline(sims.grip)

The plot that is produced by this code is presented on the upper-left panel of
Figure 16.5.
The plot for the relation between the response and the variable “arm” is
produced by this code:

> plot(sims~arm,data=job)
> sims.arm <- lm(sims~arm,data=job)
> abline(sims.arm)

The plot that is produced by the last code is presented on the upper-right panel
of Figure 16.5.
Both plots show similar characteristics. There is an overall linear trend in
the relation between the explanatory variable and the response. The value of
the response increases with the increase in the value of the explanatory variable
16.3. CASE STUDIES 309

● ● ● ● ● ●
● ● ● ●
4

4
● ●
● ●●●● ● ●●●● ●
●
●●● ● ● ● ● ●●
●
●●●
2

2
●
●
●● ● ●● ●
●● ●●●● ● ●● ●●● ● ● ●●●● ● ●●●
●● ●●
sims

sims
● ●●●
●● ●●● ●●
● ●●● ●● ●● ● ● ●●●●●●
●● ●
● ●
● ●●●●
● ●●●● ● ● ●
●● ●●● ●●● ●●
● ●●
● ● ● ●
● ●● ●●●●● ●●●● ● ● ● ●● ●● ●●●●● ●●
0

0
● ●●
●●● ●● ●●
●●●● ●
●
●
●● ●●
●●● ● ●● ● ●
● ● ●
● ●● ● ●●● ●●● ●
●●●
●● ● ● ●●● ● ● ● ● ●●● ●
● ● ●
● ●
● ●●● ●●●●● ●● ●●●●
●● ●
● ●●● ●● ● ●●●●● ●
−4 −2

−4 −2
● ● ●
● ●● ●
● ● ● ●
● ●
● ● ● ●

50 100 150 20 40 60 80 100

grip arm

● ● ●
●
● ● ● ●
4

●
150

●●●●● ● ● ● ● ● ●
● ● ● ● ●
● ●● ● ●●● ●●● ●
2

●●● ● ●● ● ●
● ●●● ● ●
●● ● ●
● ●●● ●● ● ●
●●●●●● ●
●● ● ●● ● ●●●●
●●●● ●
●
sims

● ●● ● ● ● ● ●
grip

● ● ● ●●●●●
● ● ● ● ● ●
●● ●●●● ●●● ●● ● ●● ●● ● ●
●●● ● ● ● ●● ● ●●
100

●● ●● ●● ● ● ●
● ●
●●● ●● ●●
●
● ● ●●●
● ● ● ● ●●●
0

●● ●
●● ● ●● ● ● ● ●
● ● ●● ● ●
● ●● ● ● ●● ●● ●●● ● ●
●
●●● ●● ●
● ● ● ● ● ●● ● ●
● ●● ●●
● ● ●● ●
●● ●●
●
●● ●
● ●●
●
● ● ● ●● ● ●● ● ●
● ●●●● ●
−4 −2

● ● ● ● ● ●
●
●● ●
50

● ●●
●
● ● ●

−4 −2 0 2 20 40 60 80 100

score arm

Figure 16.5: Scatter Plots and Regression Lines

(a positive slope). The regression line seems to follow, more or less, the trend
that is demonstrated by the scatter plot.
A more detailed analysis of the regression model is possible by the applica-
tion of the function “summary” to the fitted model. First the case where the
explanatory variable is “grip”:

> summary(sims.grip)

Call:
lm(formula = sims ~ grip, data = job)

Residuals:
Min 1Q Median 3Q Max
-2.9295 -0.8708 -0.1219 0.8039 3.3494

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -4.809675 0.511141 -9.41 <2e-16 ***
310 CHAPTER 16. CASE STUDIES

grip 0.045463 0.004535 10.03 <2e-16 ***

---
Signif. codes: 0 "***" 0.001 "**" 0.01 "*" 0.05 "." 0.1 " " 1

Residual standard error: 1.295 on 145 degrees of freedom

Multiple R-squared: 0.4094, Adjusted R-squared: 0.4053
F-statistic: 100.5 on 1 and 145 DF, p-value: < 2.2e-16
Examination of the report reviles a clear statistical significance for the effect
of the explanatory variable on the distribution of response. The value of R-
squared, the ration of the variance of the response
√ explained by the regression
is 0.4094. The square root of this quantity, 0.4094 ≈ 0.64, is the proportion
of the standard deviation of the response that is explained by the explanatory
variable. Hence, about 64% of the variability in the response can be attributed
to the measure of the strength of the grip.
For the variable “arm” we get:
> summary(sims.arm)

Call:
lm(formula = sims ~ arm, data = job)

Residuals:
Min 1Q Median 3Q Max
-3.64667 -0.75022 -0.02852 0.68754 3.07702

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -4.095160 0.391745 -10.45 <2e-16 ***
arm 0.054563 0.004806 11.35 <2e-16 ***
---
Signif. codes: 0 "***" 0.001 "**" 0.01 "*" 0.05 "." 0.1 " " 1

Residual standard error: 1.226 on 145 degrees of freedom

Multiple R-squared: 0.4706, Adjusted R-squared: 0.467
F-statistic: 128.9 on 1 and 145 DF, p-value: < 2.2e-16
This variable is also statistically significant. The value of R-squared is 0.4706.
The proportion
√ of the standard deviation that is explained by the strength of
the are is 0.4706 ≈ 0.69, which is slightly higher than the proportion explained
by the grip.
Overall, the explanatory variables do a fine job in the reduction of the vari-
ability of the response “sims” and may be used as substitutes of the response in
order to select among candidates. A better prediction of the response based on
the values of the explanatory variables can be obtained by combining the infor-
mation in both variables. The production of such combination is not discussed
in this book, though it is similar in principle to the methods of linear regression
that are presented in Chapter 14. The produced score6 takes the form:
score = −5.434 + 0.024 · grip + 0.037 · arm .
6 The score is produced by the application of the function “lm” to both variables as explana-

tory variables. The code expression that can be used is “lm(sims ~ grip + arm, data=job)”.
16.3. CASE STUDIES 311

We use this combined score as an explanatory variable. First we form the score
and plot the relation between it and the response:

> score <- -5.434 + 0.024*job$grip+ 0.037*job$arm

> plot(sims~score,data=job)
> sims.score <- lm(sims~score,data=job)
> abline(sims.score)
The scatter plot that includes the regression line can be found at the lower-
left panel of Figure 16.5. Indeed, the linear trend is more pronounced for this
scatter plot and the regression line a better description of the relation between
the response and the explanatory variable. A summary of the regression model
produces the report:
> summary(sims.score)

Call:
lm(formula = sims ~ score, data = job)

Residuals:
Min 1Q Median 3Q Max
-3.18897 -0.73905 -0.06983 0.74114 2.86356

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.07479 0.09452 0.791 0.43
score 1.01291 0.07730 13.104 <2e-16 ***
---
Signif. codes: 0 "***" 0.001 "**" 0.01 "*" 0.05 "." 0.1 " " 1

Residual standard error: 1.14 on 145 degrees of freedom

Multiple R-squared: 0.5422, Adjusted R-squared: 0.539
F-statistic: 171.7 on 1 and 145 DF, p-value: < 2.2e-16

Indeed, the score is highly significant. More important, the R-squared coefficient
that is associated with the score is 0.5422, which corresponds
√ to a ratio of the
standard deviation that is explained by the model of 0.5422 ≈ 0.74. Thus,
almost 3/4 of the variability is accounted for by the score, so the score is a
reasonable mean of guessing what the results of the simulations will be. This
guess is based only on the results of the simple tests of strength that is conducted
with the JES device.
Before putting the final seal on the results let us examine the assumptions of
the statistical model. First, with respect to the two explanatory variables. Does
each of them really measure a different property or do they actually measure the
same phenomena? In order to examine this question let us look at the scatter
plot that describes the relation between the two explanatory variables. This
plot is produced using the code:
> plot(grip~arm,data=job)
It is presented in the lower-right panel of Figure 16.5. Indeed, one may see that
the two measurements of strength are not independent of each other but tend
312 CHAPTER 16. CASE STUDIES

Histogram of residuals(sims.score)
Frequency

20
10
0

−3 −2 −1 0 1 2 3

residuals(sims.score)

Normal Q−Q Plot

Theoretical Quantiles

●
●
●●●
2

●●
● ●●●●
●
●●
● ●
● ●●●● ●●●
●
●●●●
●●
●●
● ●●
●
●●
●● ●
●●
●●●
●
●●●
●
0

●
●●
●●
●●
● ●
●●
●●
●●
●●
●
●
●
●●●
●●
●●
● ●
●
●
●● ●● ●●●
●●
●● ●
−2

● ●●
●
●

−3 −2 −1 0 1 2 3

Sample Quantiles

Figure 16.6: An Histogram and a QQ-Plot of Residuals

to produce an increasing linear trend. Hence, it should not be surprising that

the relation of each of them with the response produces essentially the same
goodness of fit. The computed score gives a slightly improved fit, but still, it
basically reflects either of the original explanatory variables.
In light of this observation, one may want to consider other measures of
strength that represents features of the strength not captures by these two
variable. Namely, measures that show less joint trend than the two considered.
Another element that should be examined are the probabilistic assumptions
that underly the regression model. We described the regression model only
in terms of the functional relation between the explanatory variable and the
expectation of the response. In the case of linear regression, for example, this
relation was given in terms of a linear equation. However, another part of the
model corresponds to the distribution of the measurements about the line of
regression. The assumption that led to the computation of the reported p-values
is that this distribution is Normal.
A method that can be used in order to investigate the validity of the Normal
assumption is to analyze the residuals from the regression line. Recall that
these residuals are computed as the difference between the observed value of
16.4. SUMMARY 313

the response and its estimated expectation, namely the fitted regression line.
The residuals can be computed via the application of the function “residuals”
to the fitted regression model.
Specifically, let us look at the residuals from the regression line that uses the
score that is combined from the grip and arm measurements of strength. One
may plot a histogram of the residuals:

> hist(residuals(sims.score))

The produced histogram is represented on the upper panel of Figure 16.6. The
histogram portrays a symmetric distribution that my result from Normally dis-
tributed observations. A better method to compare the distribution of the
residuals to the Normal distribution is to use the Quantile-Quantile plot. This
plot can be found on the lower panel of Figure 16.6. We do not discuss here the
method by which this plot is produced7 . However, we do say that any deviation
of the points from a straight line is indication of violation of the assumption of
Normality. In the current case, the points seem to be on a single line, which is
consistent with the assumptions of the regression model.
The next task should be an analysis of the relations between the explanatory
variables and the other response “ratings”. In principle one may use the same
steps that were presented for the investigation of the relations between the
explanatory variables and the response “sims”. But of course, the conclusion
may differ. We leave this part of the investigation as an exercise to the students.

16.4 Summary
16.4.1 Concluding Remarks
The book included a description of some elements of statistics, element that we
thought are simple enough to be explained as part of an introductory course to
statistics and are the minimum that is required for any person that is involved
in academic activities of any field in which the analysis of data is required. Now,
as you finish the book, it is as good time as any to say some words regarding
the elements of statistics that are missing from this book.
One element is more of the same. The statistical models that were presented
are as simple as a model can get. A typical application will required more com-
plex models. Each of these models may require specific methods for estimation
and testing. The characteristics of inference, e.g. significance or confidence
levels, rely on assumptions that the models are assumed to possess. The user
should be familiar with computational tools that can be used for the analysis of
these more complex models. Familiarity with the probabilistic assumptions is
required in order to be able to interpret the computer output, to diagnose pos-
sible divergence from the assumptions and to assess the severity of the possible
effect of such divergence on the validity of the findings.
Statistical tools can be used for tasks other than estimation and hypothesis
testing. For example, one may use statistics for prediction. In many applica-
tions it is important to assess what the values of future observations may be
7 Generally speaking, the plot is composed of the empirical percentiles of the residuals,

plotted against the theoretical percentiles of the standard Normal distribution. The current
plot is produced by the expression “qqnorm(residuals(sims.score))”.
314 CHAPTER 16. CASE STUDIES

and in what range of values are they likely to occur. Statistical tools such as
regression are natural in this context. However, the required task is not testing
or estimation the values of parameters, but the prediction of future values of
the response.
A different role of statistics in the design stage. We hinted in that direction
when we talked about in Chapter 11 about the selection of a sample size in order
to assure a confidence interval with a given accuracy. In most applications, the
selection of the sample size emerges in the context of hypothesis testing and the
criteria for selection is the minimal power of the test, a minimal probability to
detect a true finding. Yet, statistical design is much more than the determina-
tion of the sample size. Statistics may have a crucial input in the decision of
how to collect the data. With an eye on the requirements for the final analysis,
an experienced statistician can make sure that data that is collected is indeed
appropriate for that final analysis. Too often is the case where researcher steps
into the statistician’s office with data that he or she collected and asks, when
it is already too late, for help in the analysis of data that cannot provide a
satisfactory answer to the research question the researcher tried to address. It
may be said, with some exaggeration, that good statisticians are required for
the final analysis only in the case where the initial planning was poor.
Last, but not least, is the theoretical mathematical theory of statistics. We
tried to introduce as little as possible of the relevant mathematics in this course.
However, if one seriously intends to learn and understand statistics then one
must become familiar with the relevant mathematical theory. Clearly, deep
knowledge in the mathematical theory of probability is required. But apart
from that, there is a rich and rapidly growing body of research that deals with
the mathematical aspects of data analysis. One cannot be a good statistician
unless one becomes familiar with the important aspects of this theory.
I should have started the book with the famous quotation: “Lies, damned
lies, and statistics”. Instead, I am using it to end the book. Statistics can be
used and can be misused. Learning statistics can give you the tools to tell the
difference between the two. My goal in writing the book is achieved if reading
it will mark for you the beginning of the process of learning statistics and not
the end of the process.

16.4.2 Discussion in the Forum

In the second part of the book we have learned many subjects. Most of these
subjects, especially for those that had no previous exposure to statistics, were
unfamiliar. In this forum we would like to ask you to share with us the difficulties
that you encountered.
What was the topic that was most difficult for you to grasp? In your opinion,
what was the source of the difficulty?
When forming your answer to this question we will appreciate if you could
elaborate and give details of what the problem was. Pointing to deficiencies
in the learning material and confusing explanations will help us improve the
presentation for the future editions of this book.