Professional Documents
Culture Documents
Quantitative Literacy
Quantitative Literacy
QUANTITATIVE LITERACY
Crauder-4100237 QL_FM November 13, 2014 11:19
Crauder-4100237 QL_FM November 13, 2014 11:19
QUANTITATIVE LITERACY
Thinking Between the Lines Second Edition
ISBN-13: 978-1-4641-2512-6
ISBN-10: 1-4641-2512-0
First printing
www.macmillanhighered.com
Crauder-4100237 QL_FM November 13, 2014 11:19
honors in mathematics, and received his Ph.D. from Columbia in 1981, writing a
dissertation on algebraic geometry. After post-doctoral positions at the Institute for
Advanced Study, the University of Utah, and the University of Pennsylvania, Bruce
went to Oklahoma State University in 1986 where he is now a Professor of Mathe-
matics and also Associate Dean.
Bruce’s research in algebraic geometry has resulted in 10 refereed articles in as
many years in his specialty, three-dimensional birational geometry. He has since
worked on the challenge of the beginning college math curriculum, resulting in the
creation of two new courses with texts to support the courses. He is especially pleased
with these texts, as they combine scholarship with his passion for teaching.
Bruce has two sons, one an electrical engineer and the other a producer in
sports media. He enjoys birdwatching, playing with his dog, and dancing, especially
contradancing.
BENNY EVANS received his Ph.D. in mathematics from the University of Michigan
Benny Evans
in 1971. After a year at the Institute for Advanced Study in Princeton, New Jersey,
he went to Oklahoma State University where he is currently Professor Emeritus of
Mathematics. In his tenure at OSU, he has served as Undergraduate Director, As-
sociate Head, and Department Head. His career has included visiting appointments
at the Institute for Advanced Study, Rice University, Texas A&M, and University of
Nevada at Reno.
Benny’s research interests are in topology and mathematics education. His record
includes 28 papers in refereed journals, numerous books and articles (some in col-
laboration with those three other guys), and 25 grants from the National Science
Foundation, Oklahoma State Board of Regents, and private foundations.
Benny’s hobbies include fishing and spending time with his grandchildren. For-
tunately, it is possible to do both at the same time.
JERRY JOHNSON received his Ph.D. in mathematics in 1969 from the University
Jeanne Johnson
of Illinois, Urbana. From 1969 until 1993 he taught at Oklahoma State University,
with a year off in 1976 for a sabbatical at Pennsylvania State University. In 1993,
he moved to the University of Nevada, Reno, to become the founding director of
their Mathematics Center as well as director of a project called Mathematics across
the Curriculum. From 1995 to 2001, he served as Chairman of the Mathematics
Department. He is now Professor of Mathematics Emeritus.
Over the years, Jerry has published 17 research articles in prominent mathematics
journals and more than 35 articles in various publications related to mathematics ed-
ucation. He wrote and published a commercially successful animated 3-D graphing
program called GyroGraphics and was a founding member of the National Numer-
acy Network, whose goal is to promote quantitative literacy.
Jerry’s wife, Jeanne, is a folklorist with a Ph.D. from Indiana University, Bloom-
Alan Noell
ington. They enjoy traveling (Jerry has been to all 50 states), gardening (especially
tomatoes), cooking (you fry it, I’ll try it), and cats (the kind that meow).
ALAN NOELL received his Ph.D. in mathematics from Princeton University in 1983.
After a postdoctoral position at Caltech, he joined the faculty of Oklahoma State Uni-
versity in 1985. He is currently Professor of Mathematics and Associate Head. His
scholarly activities include research in complex analysis and curriculum development.
Alan and his wife, Liz, have four children, one daughter-in-law, and one son-in-
law. They are very involved with their church, and they also enjoy reading and cats
(the same kind as for Jerry).
Crauder-4100237 QL_FM November 13, 2014 11:19
BRIEF CONTENTS
1 CRITICAL THINKING 3
2 ANALYSIS OF GROWTH 77
6 STATISTICS 393
9 GEOMETRY 603
APPENDIXES A-1
ANSWERS ANS-1
CREDITS C-1
INDEX I-1
vi
Crauder-4100237 QL_FM November 13, 2014 11:19
CONTENTS
always average? 4
Taming those very large and very small
numbers 64
IN THE NEWS 1.2 A Great Example of Simpson’s
Decisions based on estimates 67
Paradox: US Median Wage Decline 4
Exercise Set 1.5 71
Simpson’s paradox and wages 5
The Berkeley gender discrimination case 6 CHAPTER SUMMARY 72
More conditional statements: Converse, IN THE NEWS 2.3 The Hockey Stick Graph Model
inverse, contrapositive 38 of Climate Change 98
Counting using Venn diagrams 52 IN THE NEWS 2.4 TV Drug Ads: The Whole Truth? 113
Venn diagrams with three sets 54 Misleading by choice of axis scale 114
vii
Crauder-4100237 QL_FM November 13, 2014 11:19
viii CONTENTS
Default ranges on graphs generated by 3.3 Logarithmic phenomena: Compressed scales 177
calculators and computers 116 IN THE NEWS 3.5 Record Number of Oklahoma
IN THE NEWS 2.5 State ACT Scores Improve, Tremors Raises Possibility of
Still Low 116 Damaging Earthquakes 177
Misleading by misrepresentation of data: What is a logarithm? 178
Inflation 118 The Richter scale 179
Misleading by using insufficient data 122 SUMMARY 3.7 Meaning of Magnitude Changes 182
Pictorial representations 123 The decibel as a measure of sound 183
SUMMARY 2.5 Cautionary Notes 125 SUMMARY 3.8 Increasing Decibels 184
3.1 Lines and linear growth: What does a constant Exercise Set 3.4 203
rate mean? 138 CHAPTER SUMMARY 208
IN THE NEWS 3.2 Sled Dog Days of Summer 138
KEY TERMS 210
Recognizing linear functions 139
Formulas for linear functions 141
CHAPTER QUIZ 210
Linear data and trend lines 148 IN THE NEWS 4.1 Top Financial Literacy Tips
Exercise Set 3.1 152
for College Freshman 213
3.2 Exponential growth and decay: Constant 4.1 Saving money: The power of compounding 214
percentage rates 158 IN THE NEWS 4.2 Americans Still Aren’t Saving
IN THE NEWS 3.3 How Vulnerable Are the Pools Enough 214
Exercise Set 3.2 172 4.2 Borrowing: How much car can you afford? 234
IN THE NEWS 3.4 Pinterest, Instagram Continue IN THE NEWS 4.3 Paying in Installments 234
Meteoric Growth 175 Installment loans and the monthly payment 234
Crauder-4100237 QL_FM November 13, 2014 11:19
CONTENTS ix
IN THE NEWS 4.5 4 Retirement Tips Young Adults Probability of non-occurrence 313
Retirement income: Annuities 263 IN THE NEWS 5.3 Percentages in Forecasts 323
SUMMARY 4.6 Retiring on an Annuity 266 IN THE NEWS 5.4 ‘FIFTY PERCENT Chance of
Exercise Set 4.3 268 Another 9/11-Style Attack’ in the Next Ten
Years. . . but Likelihood is MUCH Higher
4.4 Credit cards: Paying off consumer debt 271 if the World Gets Less Stable 323
IN THE NEWS 4.6 The Case for College Students
to Have Credit Cards 271 5.2 Medical testing and conditional probability: Ill or not? 327
Credit card basics 272 IN THE NEWS 5.5 Judge Strikes Down Drug Testing
SUMMARY 4.7 Credit Card Basics 274 of Florida Welfare Recipients 327
Making only minimum payments 274 Sensitivity and specificity 328
SUMMARY 4.8 Making Minimum Payments 279 Positive predictive value 330
Further complications of credit card usage 279 SUMMARY 5.3 Prevalence, PPV, and NPV 332
IN THE NEWS 4.7 Consumer Financial Protection A case study: Crohn’s disease 332
Bureau 280 Conditional probability 335
Exercise Set 5.2 336
Exercise Set 4.4 282
x CONTENTS
5.4 More ways of counting: Permuting and combining 356 6.2 The normal distribution: Why the bell curve? 411
IN THE NEWS 5.9 The Human Genome Project IN THE NEWS 6.3 Tumbling Dice & Birthdays 411
Completion: Frequently Asked Questions 356 The bell-shaped curve 413
Permutations 357 Mean and standard deviation for the normal
SUMMARY 5.5 Permutations of n Items 358 distribution 415
More on arrangements 358 z-scores 419
Calculations using factorials 360 Percentile scores 420
The Central Limit Theorem 422
SUMMARY 5.6 Permutations of n Items Taken
Significance of apparently small deviations 424
k at a Time 361
Health issues and the Fallon leukemia cluster 425
Combinations 362
Exercise Set 6.2 427
SUMMARY 5.7 Combinations 363
Hand calculation of combinations 364
6.3 The statistics of polling: Can we believe the polls? 431
Probabilities with permutations
IN THE NEWS 6.4 RC Poll: Virginians’ Views
or combinations 366
on Ethics, Mental Health Reforms, Medicaid
Exercise Set 5.4 369
Expansion Plus Warner vs. Gillespie
IN THE NEWS 5.10 Hands on the iPhone TouchID 370 for US Senate 432
IN THE NEWS 5.12 National Gambling Impact SUMMARY 6.1 Polls and Margin of Error 435
Study Commission Report 374 How big should the sample be? 435
The “statistical dead heat” 440
Lotteries and expected value 375
Expected value and the law of large numbers 376 IN THE NEWS 6.5 Romney and Obama Dead Even
Calculating expected value 377 in Key Swing States, According to New Poll 440
Fair games 379 Methodology of polling 441
SUMMARY 5.8 Expected Value 381 IN THE NEWS 6.6 The American Public Strongly
Risk assessment 381 Supports Social Security Reform 442
Exercise Set 5.5 384
Exercise Set 6.3 444
CHAPTER SUMMARY 388
CONTENTS xi
IN THE NEWS 6.11 Coed Housing Has Higher 8 VOTING AND SOCIAL CHOICE 525
Rates of Sex, Binge Drinking 462
IN THE NEWS 8.1 Most Americans Want Popular
CHAPTER SUMMARY 463 Presidential Vote 525
KEY TERMS 465
8.1 Measuring voting power: Does my vote count? 526
CHAPTER QUIZ 466
IN THE NEWS 8.2 The Disability Voting Coalition
of PA 526
Voting power and coalitions 527
7 GRAPH THEORY 469
Winning coalitions and critical voters 529
IN THE NEWS 7.1 Revealing Social Networks SUMMARY 8.1 Winning Coalitions and Critical
of Spammers 469 Voters 530
Counting coalitions 531
7.1 Modeling with graphs and finding Euler circuits 470
SUMMARY 8.2 Number of Coalitions 531
IN THE NEWS 7.2 NextGen: The Future of Flying 470
The Banzhaf power index 531
Modeling with graphs 472
Euler circuits 474
SUMMARY 8.3 Banzhaf Power Index 534
7.2 Hamilton circuits and traveling salesmen: 8.2 Voting systems: How do we choose a winner? 544
IN THE NEWS 8.3 We Need Instant Runoff Voting:
Efficient routes 490
It’s Time for Election Reform 544
IN THE NEWS 7.3 Traveling Salesman:
Plurality voting and spoilers 545
A Seemingly Unsolvable Problem Offers a
Runoffs, preferential voting, and the
Glimpse of the Limits of Computation 490
Borda count 547
Making Hamilton circuits 492
SUMMARY 8.5 Common Preferential Voting
SUMMARY 7.3 Hamilton Circuits 493
Systems 553
The traveling salesman problem 494
Condorcet’s paradox 553
The complexity of the traveling salesman
SUMMARY 8.6 Condorcet Winners 556
problem 495
Arrow’s impossibility theorem 556
Nearest-neighbor algorithm 496
The cheapest link algorithm 498 SUMMARY 8.7 Arrow’s Impossibility Theorem 557
SUMMARY 7.4 The Traveling Salesman Problem 501 Exercise Set 8.2 558
Exercise Set 7.2 502 8.3 Fair division: What is a fair share? 561
7.3 Trees: Viral e-mails and spell checkers 508 IN THE NEWS 8.4 Tiger Woods Divorce Settlement:
IN THE NEWS 7.4 How Did That Chain Letter Elin Nordegren Received $110M
Get to My Inbox? 508
Payout—Report 561
xii CONTENTS
SUMMARY 9.2 Volumes and Surface Areas 614 Appendix 4 Basic Math A-14
Exercise Set 9.1 615 Appendix 5 Problem Solving A-18
9.2 Proportionality and similarity: Changing the scale 622 ANSWERS ANS-1
IN THE NEWS 9.2 “Lego Proportions are the
CREDITS C-1
Golden Ratio of Architecture,” Says
Bjarke Ingles 622 INDEX I-1
ill Rogers, a popular humorist and actor from the 1920s and 1930s, fa-
W mously quipped, “All I know is just what I read in the papers, and that’s
an alibi for my ignorance.” Today we get our information from a wider
variety of sources, including television broadcasts, magazines, and Internet sites, as
well as newspapers. But the sentiment remains the same—we rely on public media for
the information we require to understand the world around us and to make important
decisions in our daily lives.
In today’s world, the vast amount of information that is available in the form of
print and images is overwhelming. Interpreting and sorting this information, trying
to decide what to believe and what to reject, is ever more daunting in the face of
rapidly expanding sources. In addition, the important news of the day often involves
complicated topics such as economics, finance, political polls, and an array of statis-
tical data—all of which are quantitative in nature.
Billions of dollars
600.0
500.0
400.0
APY. What do these mean? In 650.0
600.0
550.0
300.0
200.0
100.0
these and many other practi- 500.0
450.0
0
cal situations, math matters.
2004 2005 2006 2007 2008 2009 2010 2011 2012
Year
400.0
2004 2005 2006 2007 2008 2009 2010 2011 2012
Year
FIGURE 2.39 Defense spending. A nice example of the role in FIGURE 2.40 Defense spending again.
fense spending.
xiii
Balanced Coverage
Theoretical probability
To discuss theoretical probability, we need to make clear what we mean by an event.
Our audience for this book is students who have not
An event is a collection of specified outcomes of an experiment.2 For example, sup-
pose we roll a standard die (with faces numbered 1 through 6). One event is that
taken math courses past elementary algebra. Unlike
the number showing is even. This event consists of three outcomes: a 2 shows, a 4 some liberal arts math books that may focus too much
shows, or a 6 shows.
To define the probability of an event, it is common to describe the outcomes on either the quantitative or qualitative aspects of key
corresponding to that event as favorable and all other outcomes as unfavorable. If
the event is that an even number shows when we toss a die, there are three favorable mathematical concepts, we offer students and teachers
outcomes (2, 4, or 6) and three unfavorable outcomes (1, 3, or 5).
a book that strikes the right balance between reading
text and doing calculations. For example, in Chapter 5
2 The situations we will consider all have a finite number of outcomes, so we are considering only discrete probabilities.
we discuss the probability of an event not occurring.
The usual formula is presented, but it is also clearly
described in words and in familiar contexts. This bal-
> KEY CONCEPT ance of concepts and calculation engages students in
If each outcome of an experiment is equally likely, the probability of an event
is the fraction of favorable outcomes. We can express this definition using
a formula:
thinking about interesting ideas and also gives them the
Probability of an event =
Number of favorable outcomes opportunity to apply and practice the math underlying
Total number of possible equally likely outcomes
We will often shorten this to
those ideas.
Favorable outcomes
Probability =
Total outcomes
Justin Sullivan/Getty Images
Wherever possible, we use real applications, real sce- These numbers may seem to be conclusive evidence of discrimination. When we
look more closely, however, we find that graduate school admissions are based on
narios, and real data. For example, in Chapter 1, we use departmental selections. There are many departments at Berkeley, but to keep things
simple, we will assume that the university consists of only two departments: Math
the Berkeley gender discrimination case to introduce the and English. Here is how the male and female applicants might have been divided
concept of paradox. among these two departments.
In the News
Throughout the text, explanations, examples, and exercises are often motivated by
everyday events reported in the popular media. Each chapter and each section opens
with real articles gleaned from newspapers, magazines, and Internet sites. They are
presented under the feature titled In the News. These lead naturally to discussions of
mathematics as part of everyday life. For example, In the News 2.1 launches a dis-
cussion of the different ways to present data and the advantages and disadvantages
of each.
F acebook has announced its Q3 earnings, which show not only growth in
mobile ads but also a widening of the social platform’s user base.
Despite its massive size, Facebook is still seeing big gains in user growth. In
its Q3 earnings report, the social media giant said that September 2013 saw an
1200
average of 728 million daily active users, an increase of 25 percent year-over-year.
Number of users (millions)
The company saw huge gains in mobile, too, taking in 874 million monthly active
1000
users as of September 30, 2013---a growth rate of 45 percent from 2012.
The increased user base accounted for an uptick in revenue for the company: 800
Q3 2013 revenue was $2.02 billion, an increase of 60 percent from the Q3 2012
intake of $1.26 billion. Mobile advertising revenue was also a big winner for the 600
company representing approximately 49 percent of its overall advertising revenue
for the third quarter of 2013. The mobile numbers reflect Facebook’s focus on 400
getting those ads right---the company continues to try out new ads, including a
targeted app unit, to find the right value for the space. 200
0
d 04
d 05
ct il 6
Au ber 007
Ja ust 007
br y 08
Ap y 20 9
ril 09
em ly 9
r 9
br d 09
r y 09
y 0
d 10
em ly 0
r 1
M nd 2 11
Se J h 2 11
D tem e 2 12
em r 2
r2 2
2
O Apr 200
r 0
pt Ju 200
be 200
Jul 201
pt Ju 201
be 201
ec be 01
be 201
01
En 20
En 20
Fe nuar 20
ua 20
Fe En 20
ua 20
En 20
E 20
ar 0
p un 0
o 2
g 2
d
c
En
Se
Se
Date
Try It Yourself The APR is 22.8%, and we divide by 12 to get a monthly rate of 1.9%. In decimal
form, this calculation is 0.228/12 = 0.019. Therefore,
Many worked examples are followed by a very
Finance charge = 0.019 × (Previous balance − Payments + Purchases)
similar problem that gives students the oppor-
= 0.019 × ($500 − $200 + $400)
tunity to test their understanding. This feature
= 0.019 × $700 = $13.30
serves to actively engage students in the learn-
ing process. The answers to the Try It Yourself That gives a new balance of
problems are provided at the end of the section New balance = Previous balance − Payments + Purchases + Finance charge
so that students can check their solutions. = $500 − $200 + $400 + $13.30 = $713.30
Your new balance is $713.30. Here is the completed table.
Exercise Sets
The exercise sets are designed to help students appreciate the relevance of section
topics to everyday life, and they often include direct references to popular media.
The early exercises in each set are very similar f g p
to worked examples and are easily accessible 32. Health care and longevity. The graphical presentation in
to all students. But later exercises show a wide Figure 2.66 compares health care spending with life expectancy
variety of applications and challenge students in various countries. Where on this graph is the best place to be
to think more deeply. Exercise sets often end (lower left, lower right, upper left, upper right)? Where on this
with open-ended exercises designed for group graph is the worst place to be (lower left, lower right, upper
left, upper right)? Make a plot with the axes reversed and an-
work and may require outside research.
swer the previous questions. See if you can find another way to
present the data, either of your own devising or from the Web.
8
120
Magnitude
Review, Key Concepts, Chapter Summa- 80 5
CALCULATION TIP 6.1 Calculating Standard Deviation EXAMPLE 4.21 Using the balance formula: Saving money regularly
To find the standard deviation of n data points, we first calculate the mean μ. Suppose we have a savings account earning 7% APR. We deposit $20 to the account
The next step is to complete the following calculation template: at the end of each month for five years. What is the future value for this savings
arrangement? That is, what is the account balance after five years?
Data Deviation Square of deviation SOLUTION
.. .. .. We use the regular deposits balance formula (Formula 4.8). The monthly deposit is
. . .
$20, and the monthly interest rate as a decimal is
xi xi − μ Square of second column
.. .. .. APR 0.07
. . . r= =
12 12
Sum of third column
Divide the above sum by n The number of deposits is t = 5 × 12 = 60. The regular deposits balance formula
and take the square root. gives
Benny Evans
Deposit × ((1 + r )t − 1)
Balance after 60 deposits =
r
$20 × ((1 + 0.07/12)60 − 1)
=
(0.07/12)
= $1431.86
The future value is $1431.86.
Algebraic Spotlights
ALGEBRAIC SPOTLIGHT 4.3 Derivation of the Monthly An effort is made throughout the text to motivate the use of
Payment Formula, Part II
Now we use the account balance formula to derive the monthly payment for-
complicated formulas, but some derivations require algebra
mula. Suppose we borrow B0 dollars at a monthly interest rate of r as a decimal skills that may be beyond the scope of a course. Such deri-
and we want to pay off the loan in t monthly payments. That is, we want the
balance to be 0 after t payments. Using the account balance formula we derived vations are presented in a special discussion called Algebraic
in Algebraic Spotlight 4.2, we want to find the monthly payment M that makes
Bt = 0. That is, we need to solve the equation Spotlight. An example is the derivation of the monthly payment
0 = Bt = B0 (1 + r )t − M
(1 + r )t − 1 formula in Chapter 4. While some instructors like to present
r
for M. Now derivations of formulas to their students, others would prefer
(1 + r )t − 1
0 = B0 (1 + r ) − M
t
r
to avoid them. Therefore, these spotlights are intended to be
(1 + r )t − 1 used at the instructor’s discretion.
M = B0 (1 + r )t
r
B0r (1 + r )t
M=
((1 + r )t − 1)
Because B0 is the amount borrowed and t is the term, this is the monthly pay-
ment formula we stated earlier.
Quick Review Making Graphs
Typically, we locate points on a graph using a table such as the one below,
which shows the number of automobiles sold by a dealership on a given day.
A simple tool for visualizing data sets is the bar graph. See Figure 2.67 for an 4
example of a bar graph. A bar graph is often appropriate for small data sets when Reviews. For example, in 3
Both absolute and percentage changes are easily calculated from data sets, and Chapter 2 there is a Quick 1
such calculations often show behavior not readily apparent from the original data. 0
Change in function
Percentage change = × 100%
Previous function value
One example of the value of calculating such changes is the study of world population
in Example 2.3.
To find the average growth rate of a function over an interval, we calculate the
change in the function divided by the change in the independent variable:
Change in function
Average growth rate =
Change in independent variable
Keeping track of the units helps us to interpret the average growth rate in a given
context.
300
Chapter Summaries
250
Population (millions)
200
150
Chapter summaries are included to wrap up the discussion and give
100
an overview of the major points in the chapter. For example, at the end
50
of Chapter 2, the summary brings together the major points about
0
visualization of data.
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
Year
Key Terms
Prior to each Chapter Quiz is a list of Key
KEY TERMS
Terms from the chapter along with the page logic, p. 16 valid, p. 16 inductive, p. 23 elements, p. 45
number on which each term is introduced. premises, p. 16 informal fallacy, p. 17 or, p. 32 subset, p. 45
hypotheses, p. 16 formal fallacy, p. 17 logically equivalent, p. 39 disjoint, p. 45
This allows students to review vocabulary conclusion, p. 16 deductive, p. 23 set, p. 45 Venn diagram, p. 46
Chapter Quiz
Each chapter ends with a Chapter Quiz, which
CHAPTER QUIZ
consists of a representative sample of exercises. 1. If I draw a card at random from a standard deck of 52 cards, Answer PPV 25.0%; NPV 88.9%
These are intended to give students a review of what is the probability that I draw a red queen? Write your an-
swer as a fraction.
If you had difficulty with this problem, see Example 5.11.
the major points in the chapter and a chance Answer 2/52 = 1/26 6. How many four-letter codes can we make if the first letter
is A, B, or C and the remaining letters are D or E? Letters may
to test themselves on retention of the concepts If you had difficulty with this problem, see Example 5.1. be repeated. One such code is ADDE.
2. Suppose I flip two identical coins. What is the probability Answer 24
they have learned. These exercises are accom- that I get a head and a tail? Write your answer as a fraction.
If you had difficulty with this problem, see Example 5.15.
Answer 1/2
panied by answers and references to worked If you had difficulty with this problem, see Example 5.2.
7. It has been found that only 1 in 100 bags of M&M’s have
five more than the average number of M&Ms. Only 1 in
examples in the text. For instance, the quiz at 3. Suppose a softball team has 25 players. Assume that 13 bat 20 boxes of Cracker Jacks has a really good prize. I buy a box
over .300, 15 are infielders, and 5 are both. What is the prob- of Cracker Jacks and a bag of M&Ms. What is the probability
the end of Chapter 5 contains twelve exercises. ability that a player selected at random is either batting over that I get a really good prize and five more than the average
.300 or an infielder? Write your answer as a fraction. number of M&Ms? Write your answer as a fraction.
Answer 23/25 Answer 1/2000
LaunchPad
Developed with extensive feedback from instructors and students, W. H. Freeman’s
breakthrough online course space offers:
• Pre-built Units for each chapter, curated by experienced educators, with media
for that chapter organized and ready to assign or customize to suit your course.
• All online resources for the text in one location, including an interactive e-Book,
LearningCurve adaptive quizzing (see following), interactive applets, Dynamic Fig-
ures with manipulable variables, MathClips whiteboard videos, and more.
• Powerful online homework options, with algorithmically generated exercises
including LearningCurve, precalculus quizzes, and more.
• Helpful analytics, with a Gradebook that lets you see how your class is doing
individually and as a whole.
• A streamlined and intuitive interface that lets you build an entire course in
minutes.
LearningCurve
In a game-like format, LearningCurve adaptive and formative quizzing pro-
vides an effective way to get students involved in the coursework. It offers:
• A unique learning path for each student, with quizzes shaped by each individual’s
correct and incorrect answers.
• A Personalized Study Plan, to guide students’ preparation for class and for
exams.
• Feedback for each question with live links to relevant e-Book pages, guiding
students to the reading they need to do to improve their areas of weakness.
For Instructors
The Instructor’s Web site (www.whfreeman.com/quantlit) requires user registration
as an instructor and features all of the student Web materials plus:
• PowerPoint Slides containing all textbook figures and tables
• Lecture PowerPoint Slides
• Instructor’s Guide with Solutions
Test Bank
ISBN: 1-4641-2525-2 (digital [CD-ROM])
The Test Bank contains hundreds of multiple-choice, fill-in, and short-answer ques-
tions to generate quizzes and tests for each chapter of the text. The Test Bank is avail-
able on CD-ROM (for Windows and Mac), where questions can be downloaded,
edited, and resequenced to suit each instructor’s needs.
Acknowledgments
We are grateful for the thoughtful comments and insights from the reviewers who
helped us shape the second edition. Their reviews were invaluable, helping us see our
materials through the eyes of others.
We remain grateful to the many instructors who made important contributions
to the creation of the first edition.
Bruce Crauder
Benny Evans
Jerry Johnson
Alan Noell
© Fancy/Alamy
Crauder-4100237 QL November 6, 2014 14:34
1 CRITICAL THINKING
T
here is no need to make a case for the importance of literacy in contemporary
society. The ability to read with comprehension and the ability to communicate
clearly are universally recognized as necessary skills for the average citizen. But
there is another kind of literacy, one that may be just as important in a world filled
with numbers, charts, diagrams, and even formulas: quantitative literacy.
1.1 Public policy and In contemporary society we are confronted with quantitative issues at every turn.
Simpson’s paradox: We are required to make choices and decisions about purchases, credit cards, savings
Is “average” always plans, investments, and borrowing money. In addition, politicians and advertisers
throw figures and graphs at us in an attempt to convince us to vote for them or buy
average? their product. If we want to be informed citizens and savvy consumers, we must be
1.2 Logic and informal quantitatively literate.
At the heart of quantitative literacy is critical thinking, which is necessary to deal
fallacies: Does that intelligently with the information and misinformation that come to us via television,
argument hold water? computers, newspapers, and magazines.
The following excerpt is from an article that appeared in The Tennessean. It shows
1.3 Formal logic and the importance of critical thinking in education.
truth tables:
Do computers think?
1.4 Sets and Venn
diagrams: Pictorial logic
IN THE NEWS 1.1 Search
1.5 Critical thinking and Home TV & Video U. S. World Politics Entertainment Tech Travel
number sense: What do
these figures mean? Reading Academy Teaches Kids to Think Critically
VICKY TRAVIS November 3, 2013
This chapter opens with a phenomenon called Simpson’s paradox, which illus-
trates in a very interesting way the need for critical thinking. We next discuss basic
logic, the use of Venn diagrams, and number sense.
3
Crauder-4100237 QL November 6, 2014 14:34
4 CHAPTER 1
S ince 2000, the median US wage has risen about 1%, adjusted for inflation.
But over the same period, the median wage for:
• high school dropouts,
• high school graduates with no college education,
• people with some college education, and
• people with Bachelor’s or higher degrees
have all decreased. In other words, within every educational subgroup, the me-
dian wage is lower now than it was in 2000.
How can both things be true: overall wages have risen, but wages within
every subgroup have fallen? This is a great example of Simpson’s Paradox. In this
particular case, the explanation lies in the changing educational profile of the
workforce over the past 13 years: there are now many more college graduates
(who get higher-paying jobs) than there were in 2000, but wages for college
graduates collectively have fallen at a much slower rate (down 1.2%) than for
those of lower educational attainment (whose wages have fallen precipitously,
down 7.9% for high school dropouts). The growth in the proportion of college
graduates swamps the wage decline for specific groups.
Simpson’s Paradox isn’t really a paradox at all, it just reflects the fact that
your perception of events can change depending on your viewpoint. From the
points of view of an economist looking at the national economy, the “headline”
rate of overall median wages has slightly increased, largely due to changes in
demographics. But from the point of view of a typical individual, wages have
declined. The New Statesman says that “it’s more accurate to say that wages have
fallen in the last thirty years even though the headline figure shows otherwise,”
but it all depends on your perspective.
According to the article, it is possible to have the overall wages in the United States
increase over a period of time but have the wages within every economic subgroup
decrease over the same period. This phenomenon seems impossible, but the following
example shows that it can happen.
CALVIN AND HOBBES ©1986 Watterson. Dist. By UNIVERSAL UCLICK. Reprinted with permission. All rights reserved.
Year 2000
Number of workers (thousands) Total weekly earnings (thousands)
Less than bachelor’s 61,166 $39,500,093
Bachelor’s or higher 27,564 $33,169,777
Total 88,730 $72,669,870
Year 2013
Number of workers (thousands) Total weekly earnings (thousands)
Less than bachelor’s 57,154 $33,867,652
Bachelor’s or higher 37,013 $44,008,457
Total 94,167 $77,876,109
Show that from 2000 to 2013 the average weekly earnings per worker decreased
in each educational subgroup but increased overall. Round each average to the near-
est dollar.
SOLUTION
First we find the average weekly earnings per worker in 2000 for those with less than
a bachelor’s degree:
Total weekly earnings
Average weekly earnings =
Number of workers
$39,500,093 thousand
=
61,166 thousand
This average is about $646 per worker.
Crauder-4100237 QL November 6, 2014 14:34
6 CHAPTER 1
We calculate the remaining averages in a similar fashion. The results are shown
in the following table.
Thus, for both educational subgroups the average weekly earnings decreased
from 2000 to 2013. But overall the average weekly earnings increased.
Year 2000
Number of workers Total weekly earnings
Less than bachelor’s 400 $260,000
Bachelor’s or higher 700 $532,000
Total 1100 $792,000
Year 2013
Number of workers Total weekly earnings
Less than bachelor’s 200 $128,000
Bachelor’s or higher 1100 $825,000
Total 1300 $953,000
Make a table showing the average weekly earnings per worker for each educa-
tional subgroup and overall in both 2000 and 2013. (Round each average to the
nearest dollar.) Interpret your results.
The answer is provided at the end of the section.
It is easy to see why Simpson’s paradox is called a paradox. Such a result seems
impossible. But the example above shows that combining, or aggregating, data can
mask underlying patterns.
Even though this phenomenon is called a paradox, these results are not so surpris-
ing if we look more carefully. The article notes that the relative number of college
graduates increased significantly from 2000 to 2013. In 2000, the ratio of college
graduates to total workers was 27,564/88,730 or about 31%, whereas in 2013 that
ratio was 37,013/94,167 or about 39%. This shift in the level of educational at-
tainment, plus the fact that better-educated workers are paid higher wages, helps to
explain this phenomenon.
the next are rounded to one decimal place. Consult the end of this section for a quick
review of percentages.)
For example, here is the way to calculate the percentage of male applicants who
were accepted:
Males accepted
Percentage of male applicants accepted = × 100%
Male applicants
3714
= × 100%
8442
This is about 44.0%.
Justin Sullivan/Getty Images
Males Females
Applicants Accepted % accepted Applicants Accepted % accepted
Math 2000 500 25.0% 3000 780 26.0%
English 6442 3214 49.9% 1321 732 55.4%
Total 8442 3714 44.0% 4321 1512 35.0%
Note that each of the two departments actually accepted a larger percentage of
female applicants than male applicants: The Math Department accepted 25.0% of
males and 26.0% of females, and the English Department accepted 49.9% of males
Crauder-4100237 QL November 6, 2014 14:34
8 CHAPTER 1
and 55.4% of females. The explanation of why a larger percentage of men overall
were accepted lies in the fact that the Math Department accepts significantly fewer
applicants than does English, both for men (25.0% compared to 49.9%) and for
women (26.0% compared to 55.4%). Most women applied to the Math Department,
where it is more difficult to be accepted, but most men applied to English, where it
is less difficult.
A similar type of mind-bending phenomenon occurs in sports, as the following
example shows. These instances of Simpson’s paradox are taken from a post on the
blog Natural Blogarithms.
Indoors Outdoors
At-bats Hits At-bats Hits
Chicago Cubs 462 134 5050 1280
Oakland Athletics 113 33 5335 1363
To calculate the batting average we divide the number of hits by the number of at-
bats. Which team had the higher average playing indoors? Which had the higher
average playing outdoors? Which had the higher average overall? Round all figures
to three decimal places.
SOLUTION
We calculate the average for the Chicago Cubs playing indoors as follows:
Hits
Average indoors =
At-bats
134
=
462
This is about 0.290. We calculate averages for the Chicago Cubs and the Oakland
Athletics both indoors and outdoors in a similar fashion. The results are shown in
the following table.
Indoors Outdoors
Average for Chicago Cubs 134/462 or 0.290 1280/5050 or 0.253
Average for Oakland Athletics 33/113 or 0.292 1363/5335 or 0.255
Now we find the overall averages. Altogether the Chicago Cubs got a total of
134 + 1280 = 1414 hits out of 462 + 5050 = 5512 trips to the plate. Therefore,
Hits
Overall average for Chicago Cubs =
At-bats
1414
=
5512
A similar calculation for the Oakland Athletics gives the average overall of
1396/5448 or about 0.256. Hence, the Oakland Athletics had the higher batting
average both indoors and outdoors, but the Chicago Cubs had the higher batting
average overall.
Indoors Outdoors
At-bats Hits At-bats Hits
Colorado Rockies 272 66 5258 1386
Milwaukee Brewers 1075 271 4531 1200
Which team had the higher average playing indoors? Which had the higher average
playing outdoors? Which had the higher average overall? Round all figures to three
decimal places.
The answer is provided at the end of the section.
As we have seen in the other examples in this section, the explanation for this
seemingly paradoxical situation with batting averages lies in the absolute numbers
involved in calculating the averages. Each team had a higher batting average indoors
than outdoors. But the Chicago Cubs had a higher percentage of their at-bats indoors,
462/(462 + 5050) or about 8%, than did the Oakland Athletics, 113/(113 + 5335)
or about 2%. Simpson’s paradox makes the use of overall batting averages alone
somewhat suspect in the debate about which team is better.
The examples of Simpson’s paradox presented here show the importance of criti-
cal thinking. The use of a single number such as an average to describe complex data
can be perilous. Further, combining data can mask underlying patterns when some
factor distorts the overall picture but not the underlying patterns. To understand in-
formation presented to us, it is important for us to know at least the basics of how
the information is produced and what it means.
Crauder-4100237 QL November 14, 2014 12:3
10 CHAPTER 1
result for batting averages is entirely possible. Aggregate averages can be influenced
by unusual subgroups of data, and five-year batting averages can be biased by an
odd year’s average. As an example, suppose player 1 has 1 at-bat and 1 hit per year
for the first 4 years. In the fifth year, she has 100 at-bats and only 1 hit. Her overall
average is very low because of the poor performance in the fifth year. Compare this
with player 2, who has 100 at-bats and 99 hits for each of the first four years and
then has no hit in her 1 at-bat in the fifth year. Her overall average is very good, but
the average each year is not as good as player 1. The apparent contradiction comes
from the fact that almost all of player 1’s at-bats came in the fifth year but most of
player 2’s at-bats were in the first four years.
Further questions of this type may be found in the What Do You Think section of
exercises.
For both educational subgroups, the average weekly earnings decreased from 2000
to 2013. But overall the average weekly earnings increased.
Try It Yourself 1.2: Calculating averages: Batting averages The Milwaukee Brew-
ers had a higher average than the Colorado Rockies indoors, 0.252 to 0.243, and also
outdoors, 0.265 to 0.264. But the Colorado Rockies had the higher average overall,
0.263 to 0.262.
12 CHAPTER 1
20. Complete the following table for all students: teristics similar to those who were not traced from the initial
group.” Are the percentages of those with insurance similar
between the two groups in the table? If not, what conclusion
All students in both groups would you draw on the basis of this quotation?
Passed Failed Total % passing 25. Here is the table showing the results for whites. Complete
School A the table. Round all percentages to whole numbers.
School B
White participants
21. The principal of School B argued that, based on the over- Had No % with
all figures in the table from Exercise 20, her school came out insurance insurance Total insurance
ahead and therefore is doing a better job than School A. How
Children not 104 22
would you respond? traced
22. You are a parent trying to decide to which school to send Five-year group 10 2
your child. On the basis of the tables in Exercises 14, 15, and
20, which school would you choose? Explain.
26. Here is the table showing the results for blacks. Complete
23–28. Children in South Africa. Complete Exercises 23 the table. Round all percentages to whole numbers.
through 28, in order, using the following information. In 1990,
scientists initiated a study of the development of children in
the greater Johannesburg/Soweto metropolitan area of South Black participants
Africa.2 They collected information on certain children at Had No % with
birth, and five years later they invited these children and their insurance insurance Total insurance
caregivers to be interviewed about health-related issues. Rel-
atively few participated in the follow-up interviews, and the Children not 91 957
researchers wanted to know whether those who participated traced
had similar characteristics to those who did not participate. In Five-year group 36 368
the following tables, these are referred to as the five-year group
and the children not traced, respectively. One factor used to 27. For whites, compare the percentages having insurance be-
compare these groups was whether the mother had health in- tween the two groups: children not traced and the five-year
surance at the time the child was born. Here are the results for group. For blacks, compare the percentages having insurance
this factor, including only participants who were either white between the two groups: children not traced and the five-year
or black. group.
28. On the basis of your answer to Exercise 27, revisit the ques-
All participants tion at the end of Exercise 24. Do the tables for whites and
blacks provide any evidence that the five-year sample cannot
Had No % with be used to draw conclusions about the entire group? Note: The
insurance insurance Total insurance author of the summary observed that in South Africa whites
Children not 195 979 generally were much more likely to have health insurance than
traced blacks. Further, relatively few whites were included in the orig-
Five-year group 46 370 inal study. As a result, there were relatively few in the first of
the preceding tables who had health insurance. Another factor
in the paradoxical outcome is that, in contrast to blacks, rela-
23. Fill in the blanks in the preceding table. Round all percent- tively few of the whites who were initially involved participated
ages to whole numbers. in the interviews and were included in the five-year group.
24. In the summary of the cited study, the author noted, “If 29. History of Simpson’s paradox. Research the history of
the five-year sample is to be used to draw conclusions about Simpson’s paradox and describe some of the early examples
the entire birth cohort, the five-year group should have charac- given.
2 Christopher H. Morrell, “Simpson’s Paradox: An Example from a Longitudinal Study in South Africa,” Journal of Statistics Education 7 (1999).
Crauder-4100237 QL November 6, 2014 14:34
14 CHAPTER 1
1.2 Logic and informal fallacies: Does that argument hold water? 15
Political debates and commercials are often delivered in sound bites to evoke an
emotional reaction. Critical examination of what these “pitches” actually say can
reveal exaggerations, half-truths, or outright fallacies.
In the face of such heated rhetoric and tenuous logic, we appeal to an older source,
one that remains relevant today.
In this section, we begin a study of logic and examine some informal logical
fallacies.
Crauder-4100237 QL November 6, 2014 14:34
16 CHAPTER 1
New York City Democratic mayoral candidate Bill de Blasio, right, and Republican
mayoral candidate Joe Lhota participate in their first televised debate on October 15,
2013 in New York City.
1.2 Logic and informal fallacies: Does that argument hold water? 17
The term fallacy, as used by logicians, refers to an argument that may on the
surface seem to be correct but is in fact incorrect. There are many types of falla-
cies. Some are easy to spot, and others are more subtle. Although there are many
ways in which incorrect inferences can be constructed, fallacies fall into two general
categories: informal and formal.
Fallacies of relevance
In fallacies of relevance, the premises are logically irrelevant to, and hence incapable
of establishing the truth of their conclusions. We offer several examples, but this
collection is by no means exhaustive.
Appeal to ignorance This type of fallacy occurs whenever it is argued that a statement
is true or not true simply on the basis of a lack of proof. This is a fallacy of relevance
because the lack of proof of a statement is not by itself relevant to the truth or falsity
of that statement. The structure of this fallacy is as follows:
• A certain statement is unproven.
• Therefore, the statement must be false.
3 Irving M. Copi and Carl Cohen, Introduction to Logic, 13th ed. (Upper Saddle River, NJ: Prentice-Hall, 2008).
Crauder-4100237 QL November 6, 2014 14:34
18 CHAPTER 1
Example: For over 75 years people have tried and failed to show that aliens have
not visited Earth. So we must finally accept the fact that at least some of the UFO
reports are based on actual alien visits.
In this example, the inability to disprove alien visits is not relevant to the existence
of such visits. Perhaps it would help to consider the opposite argument: For over 75
years people have not been able to show that aliens have visited Earth. So we must
finally accept the fact that no such visits have ever occurred. The same set of facts
is used to establish both that aliens have visited Earth and that they have not. The
point is that an absence of proof cannot establish either that aliens have visited or
that they have not.
Dismissal based on personal attack This type of fallacy occurs when we simply
attack the person making an assertion instead of trying to prove or disprove the
assertion itself. This is a fallacy of relevance because the character of the person
making a claim is not necessarily relevant to the claim being made. This fallacy has
the following structure:
• A person presents an argument or point of view.
• The character of that person is brought into question.
• Based on the character attack, it is concluded that the argument or point of view
is incorrect.
Example: My political opponent is against government-funded health care, and she
also has a reputation for being heartless. She refused to seek medical treatment for
her own father when he was ill. So government-funded health care is a good idea.
In this argument, the conclusion is that a certain health-care proposal is good,
but no information is provided regarding the merits of the proposal itself. Rather,
the argument is that the proposal must be good because of the claim that one of its
opponents is suspect. Even if that accusation is true, it does not necessarily follow that
the proposal is good. We cannot evaluate the proposal without information about
what it is.
False authority This fallacy arises whenever the validity of an argument is accepted
based solely on the fact that it is supported by someone, but that person’s credentials
or expertise are not relevant to the argument. This is clearly a fallacy of relevance.
This fallacy has the following structure:
• A person makes a claim based on his or her authority.
• The claim is outside the scope of that person’s authority.
• The truth of the argument is concluded based on the authority of the claimant.
Example: Over the past few years, I have starred in a number of the most popular
movies in America, so I can assure you that Johnson and Johnson’s new anti-nausea
drug is medically safe and effective.
The fallacy here is obvious—starring in movies in no way qualifies anyone to
judge the efficacy of any medical product. The prevalence of commercials where
sports figures or movie stars tell us which clothes to buy, which foods to eat, and
which candidates to vote for speaks volumes about the effectiveness of such falla-
cious arguments.
Of course, appropriate authority does indeed carry logical weight. For exam-
ple, the authority of Stephen Hawking lends credence to a claim regarding physics
because he is a world-renowned physicist. On the other hand, his authority would
probably add nothing to a toothpaste commercial.
Straw man This fallacy occurs whenever an argument or position is dismissed based
on the refutation of a distorted or completely different argument or position (the
straw man)—presumably one that is easier to refute. This is a fallacy of relevance
Crauder-4100237 QL November 6, 2014 14:34
1.2 Logic and informal fallacies: Does that argument hold water? 19
because the refutation does not even address the original position. The structure of
this fallacy is as follows:
• A position or point of view is presented.
• The case for dismissing a distorted or different position or point of view
(the straw man) is offered.
• The original position is dismissed on the basis of the refutation of the straw-man
position.
Example: A group of my fellow senators is proposing a cut in military expenditures.
I cannot support such a cut because leaving our country defenseless in these troubled
times is just not acceptable to me.
A U.S. Air Force MQ-1 Predator unmanned aerial vehicle assigned to the California
Air National Guard’s 163rd Reconnaissance Wing flies near the Southern California
Logistics Airport in Victorville, California.
The “straw man” here is the proposition that our military defense program should
be dismantled. It is easy to get agreement that this is a bad idea. The fallacy is that
no such proposal was made. The proposal was to cut military expenditures, not to
dismantle the military.
Political speech is often full of this type of blatant fallacy. Rather than engaging
in honest, reasoned debate of the issues, we often see distortions of an opponent’s
position (the straw man) refuted. The politicians who do this are trying to appeal
to emotion rather than reason. Critical thinking is our best defense against such
tactics.
Appeal to common practice This fallacy occurs when we justify a position based on
its popularity. It is a fallacy of relevance because common practice does not imply
correctness. This fallacy has the following structure:
• The claim is offered that a position is popular.
• The validity of the claim is based on its popularity.
Example: It is OK to cheat on your income tax because everybody does it.
Of course, it isn’t true that everybody cheats on their income taxes, but even if
it were, that is not justification for cheating. (That argument certainly will not carry
much weight with the IRS if you’re audited.) This type of argument is often used in
advertising.
Crauder-4100237 QL November 6, 2014 14:34
20 CHAPTER 1
Fallacies of presumption
In fallacies of presumption, false or misleading assumptions are either tacitly or ex-
plicitly assumed, and these assumptions are the basis of the conclusion.
False dilemma This fallacy occurs when a conclusion is based on an inaccurate or
incomplete list of choices. This is a fallacy of presumption because the (incorrect) as-
sumption is being made that the list offered is complete. This fallacy has the following
structure:
• An incomplete or inaccurate list of consequences of not accepting an argument is
presented.
• A conclusion is drawn based on the best (or least bad) of these consequences.
Example: You’d better buy this car or your wife will have to walk to work and your
kids will have to walk to school. I know you don’t want to inconvenience your family,
so let’s start the paperwork on the automobile purchase.
The salesperson is telling us we have two choices, “Buy this car or inconvenience
your family.” In fact, we probably have more than these two choices. There may be a
less expensive car that will serve my family’s needs, or perhaps public transportation
is an option in my neighborhood. The dilemma presumed by the salesperson is not
necessarily true. The conclusion that we should purchase the car is based on choosing
the best of the options offered by the salesperson, but in fact there are additional
choices available.
An argument of this sort is valid only if the list of choices is indeed complete. This
type of argument is commonly presented by public officials: We must either increase
your taxes or cut spending on education. So you must support an increase in taxes.
False cause The fallacy of false cause occurs whenever we conclude a causal rela-
tionship based solely on common occurrence. It is an error of presumption because
we are tacitly assuming that a causal relationship must exist. The structure of this
fallacy is as follows:
Crauder-4100237 QL November 6, 2014 14:34
1.2 Logic and informal fallacies: Does that argument hold water? 21
• The conclusion that it is generally true or always true is drawn based on the few
examples.
Example: I know the quarterback, the tight end, and the center on our football team,
and all three are excellent students. The athletes at our university do not shirk their
scholarly duties.
Crauder-4100237 QL November 6, 2014 14:34
22 CHAPTER 1
Larry Downing/Reuters/Landov
U.S. President Barack Obama speaks in support of the U.S. Senate’s bipartisan immigration
reform bill while in the East Room of the White House in Washington, June 11, 2013.
Here we are making a judgment about athletes at the university based on the good
qualities of three football players. The exemplary behavior of these three athletes tells
us nothing about how the remaining athletes are behaving.
Finding three athletes who are scholars does not mean that all athletes are, but
these examples might prompt us to explore further. Examples cannot prove the gen-
eral case, but they can help us formulate ideas that may merit further investigation.
For example, I might be very pleased to learn that three of my daughter’s friends
are honor students. Perhaps she is wise in her choice of friends. It is certainly worth
further attention.
1.2 Logic and informal fallacies: Does that argument hold water? 23
There are other types of fallacies. Some of these will be examined in the exercises.
24 CHAPTER 1
In some cases inductive arguments are extremely convincing. Take this one for
example:
Premise: Every human being who has ever lived has eventually died.
Inductive conclusion: All human beings are mortal.
Because the evidence is so overwhelming, some people might say that the con-
clusion is deduced from the premise. Technically speaking, however, this is still an
inductive argument. The conclusion is based on evidence, not on deductive logic.
Here’s another inductive argument with lots of evidence:
Premise: Every college football referee I have ever seen is male.
Inductive conclusion: All college football referees are male.
Rick Osentoski/Cal Sport Media/Newscom
This argument may sound convincing, but in 2009 Sarah Thomas became the first
woman to officiate a bowl game.
The weight of an inductive argument depends to a great extent on just how much
evidence is in the premises. Such argumentation also has value in that it may suggest
further investigation. In mathematics, this is often how we discover things—we ob-
serve patterns displayed by specific examples. This may lead us to propose a more
general result, but in mathematics no such result is accepted as true until a proper
deductive argument is found.
To show what we mean, let’s look at the following table, which shows the squares
of the first few odd integers:
Odd integer 1 3 5 7 9
Odd integer squared 1 9 25 49 81
Note that each of the squares turns out to be odd. Reasoning inductively, we
Sarah Thomas officiating a game. propose that the square of every odd integer is odd. Note that the pattern in the
table supports this claim, but the table does not prove the claim. The table does not,
for example, guarantee that the square of 163 is an odd integer. Mathematicians are
indeed able to give a (relatively simple) deductive proof of this proposal. Thus, the
pattern that we saw in the table is, in fact, true for all odd integers.
Inductive reasoning of this type can occasionally lead us astray. Recall that a
prime number is a number (other than 1) that has no divisors other than itself and
1. So 7 is prime because its only divisors are 1 and 7. On the other hand, 15 is not
prime because both 3 and 5 are divisors of 15. It has divisors other than 1 and 15.
Let’s consider the expression n2 + n + 41. The table below shows some values for
this expression.
n 1 2 3 4 5 6 7
n2 + n + 41 43 47 53 61 71 83 97
After we observe that 41, 43, 47, 53, 61, 71, 83, and 97 are all prime, we might
be led to propose that the expression n2 + n + 41 always gives a prime number. If
you explore further, you will find that the result is prime for n = 1 up through
n = 39. But if we put in n = 40, the result is not prime:
402 + 40 + 41 = 1681 = 41 × 41
In this case, the pattern we observed in the table is not true for all numbers.
Inductive reasoning can lead us to important discoveries. But as we have seen, it can
sometimes lead us to wrong conclusions.
The idea of an inductive argument is closely tied to pattern recognition. We ex-
plore this connection further in the following example.
Crauder-4100237 QL November 6, 2014 14:34
1.2 Logic and informal fallacies: Does that argument hold water? 25
Hours 0 1 2 3 4 5 6
Number of cells 1 2 4 8 16 32 64
Describe the pattern shown by the table and suggest a general rule for finding the
number of cells in terms of the number of hours elapsed.
SOLUTION
Looking closely at the table, we see a pattern. The number of cells present is always
a power of 2.
Hours 0 1 2 3 4 5 6
Number of cells 1 = 20 2 = 21 4 = 22 8 = 23 16 = 24 32 = 25 64 = 26
The pattern suggests that, in general, we find the number of cells present by raising
2 to the power of the number of hours elapsed.
Use this table to determine a pattern that leads to a general relationship between
yard size and fertilizer needed.
The answer is provided at the end of the section.
26 CHAPTER 1
11. The senate has a number of dishonest members. We just 25. No one in my family has tested positive for HIV. So my
elected a new senator. Therefore, that senator is dishonest. family is free of AIDS.
12. If I have diabetes, my blood-glucose level will be abnormal. 26. My history teacher claims that accepted statistics for the
My blood-glucose level is normal, so I do not have diabetes. population of South America before Columbus are, in fact, far
Crauder-4100237 QL November 6, 2014 14:34
1.2 Logic and informal fallacies: Does that argument hold water? 27
too small. I have heard that he is something of a racist with 37. Gambler’s fallacy. I have lost on 20 straight lottery tickets.
regard to Native Americans. His conclusions about the Indian I’m due for a win, so I’m buying extra tickets today. Note: This
population cannot be valid. fallacy is explored further in Section 5 of Chapter 5.
27. My math teacher says studying can improve my grade. I 38. Currency conversion. The euro is the currency used by
don’t believe I will make an “A” no matter how hard I study, most member states of the European Union. Here is a table of
so I’m skipping the review session tonight. values from the Web for converting euros to American dollars.
28. Everybody wants to enroll in Professor Smith’s section. He
Euro 1 2 3 4 5 6 7 8 9
must be the best teacher on campus. Dollar 1.43 2.86 4.29 5.72 7.15 8.58 10.01 11.44 12.87
29. If you don’t go with me, you are going to miss the biggest
party of the year. So hop in, and let’s go. Based on this table, what would be the cost in American dollars
of a bottle of wine marked 16 euros?
30. I bought that product, and I lost 25 pounds. It is the best
weight-loss product on the market. 39. Children’s blocks. The following chart shows the number
of children’s blocks that will fit in cubes of various sizes. Size
31. I know that my income tax is too high because I have to measurements are in inches.
pay too much.
32. The three smartest students in my math class are of Asian Size of 1 by 1 2 by 2 3 by 3 4 by 4 5 by 5
descent. Asians are better at mathematics than Americans. cube by 1 by 2 by 3 by 4 by 5
33. The president says we need to help people in Third World Number of 1 8 27 64 125
countries who don’t have sufficient food or medical care. Be- blocks
cause all of our money is going overseas, I don’t support foreign
aid. What pattern is shown in this chart? In general, how can you
34–37. Additional informal fallacies. Exercises 34 through find the number of children’s blocks that will fit in a cube of a
37 introduce additional informal fallacies. In each case, explain given size? How many blocks will fit in a cube that is 50 inches
why the example given is a fallacy. on a side?
34. Appeal to fear. You should support my point of view. I am 40–42. Finding patterns. In Exercises 40 through 42 find a
having dinner with your boss tonight, and I might have to tell pattern suggested by the given sequence.
her what you said last week. 40. 2, 4, 6, 8, 10, . . .
35. Fallacy of accident. All men are created equal. So you can 41. 1, 3, 5, 7, 9, . . .
throw the football just as well as the quarterback for the Dallas 1 1 1 1 1
Cowboys. 42. 1, , , , , , ...
2 4 8 16 32
36. Slippery slope. They keep cutting state taxes. Eventually, 43. Research. Find and report on an argument presented in the
the state won’t have enough revenue to repair the roads. popular media that is an informal fallacy.
28 CHAPTER 1
W-6. Astrophysics. Astrophysical theory has been successful at explaining most ob-
served events in the universe, but there are things it has yet to explain (like “dark
matter”). Should the theory be rejected because of this? Explain.
W-7. Warming. I read about a witch doctor in South America who believes that
global warming is real and is caused by evil spirits. That shows me that global warm-
ing is obviously false. Do you agree? Explain.
W-8. Terrorists. Every terrorist attempt I’ve read about since 9/11 involved a Muslim.
That proves to me that all Muslims are terrorists. Do you agree? Explain.
W-9. Hasty induction? Explain the difference between the fallacy of hasty general-
ization and the sort of reasoning used in pattern recognition. Give an example (not
necessarily from mathematics) of pattern recognition that is useful.
S cientists in the US have devised an enzymatic logic system that could be used
for releasing drugs. This is the first report of a man-made biomolecular system
that can process a series of physiological signals, without the use of electronics.
Biocomputer-based logic systems that process biomolecular signals could rev-
olutionise drug administration. By harnessing signal-responsive electrode surfaces
This new chip developed by that respond to biochemical signals, widespread personalised medicine takes a
researchers at Oregon State step closer to reality.
University may be a relatively ...
easy and cost-effective way to Expanding on their recent work with glucose sensing electrodes, Evgeny Katz
monitor vital signs. and Shay Mailloux at Clarkson University, in collaboration with Jan Halámek from
the State University of New York at Albany, have developed a logical biomolecule
release system.
...
Katz says the novelty of their approach is the release of the biomolecule
on receiving particular signals, which could eventually be those from a patient’s
body. He emphasises that this is the first system integrating biocomputing with
biomolecule release: `The study demonstrates the use of biocomputing—in other
words, information processing by biochemical means—for triggering another
process. This might be compared to connecting a computer with a printer. In
this combination the computer is processing the information and the printer is
printing the results. Our biochemical system conceptually works in this way.’
Crauder-4100237 QL November 6, 2014 14:34
Symbols and rules can be used to analyze logical relationships in much the same
way they are used to analyze relationships in algebra. This formalism has the ad-
vantage of allowing us to focus on the structure of logical arguments without being
distracted by the actual statements themselves.
Formal logic deals only with statements that can be classified as either true or
false (but not both). To clarify what we mean, consider the following sentences:
This sentence is false. Is this sentence true or is it false? If the sentence is true,
then it’s telling us the truth when it says that it’s false—so the sentence must be
false. However, if the sentence is false, then it’s not telling us the truth when it
says that it’s false—so the sentence must be true. This is rather paradoxical: It
appears that there is no reasonable way to classify this sentence as either true
or false. Therefore, the sentence does not qualify as a statement to which the
rules of logic can be applied.
Please turn down the music. This sentence is a request for action. It cannot be
classified as either true or false.
I am 6 feet tall. Although you may not know my height, it is the case that either
I am 6 feet tall or I am not. The statement is either true or false (and not both),
so it is a statement in the formal logical sense. We emphasize that it is necessary
only that we know that a statement can be classified as either true or false—it
is not necessary that we know which of the two applies.
By its definition, the negation of a true statement is false, and the negation of a false
statement is true. For example, if the statement p that He likes dogs is true, then the
statement NOT p that He does not like dogs is false. We show this below using a
truth table. This table lists the possibilities for the truth of p and the corresponding
value for the truth of NOT p.
p NOT p
T F This means that the negation of p is false when p is true.
F T This means that the negation of p is true when p is false.
30 CHAPTER 1
In this statement, the speaker is telling us that, in fact, he does like other animals.
If q is the statement that I like animals other than dogs, then the preceding quote is
represented by NOT NOT q, which has the same meaning as q.
The Internet search engine Google will perform the NOT operation, but it uses
the minus sign (preceded by a space) to indicate negation. For example, if you search
on Syria, you will get a number of hits, including at least one from the Central Intel-
ligence Agency. If you want to refine your search to exclude references to the CIA,
you use Syria -CIA. (The space before the minus sign is necessary.) The result is a list
of pages about Syria that don’t mention the CIA.
© 2014 Google
Conjunction The conjunction of two statements is the assertion that both are true.
In common English usage, and signifies conjunction. We will use the symbol AND to
denote conjunction. The following example shows how:
Assume that
p = Baseball is 90% mental
q = The other half of baseball is physical
The conjunction of these two statements is one of the many humorous sayings of
Yogi Berra. In symbols, p AND q represents Baseball is 90% mental and the other
half is physical.
Because the conjunction of two statements is the assertion that both are true, the
conjunction is true when, and only when, both statements are true. For example,
Yogi Berra’s preceding statement cannot be true because the truth of either of the
two statements implies that the other is false. Hence, one or the other is false—so the
conjunction is false.
The truth table for the conjunction shows the truth value of p AND q for each
possible combination of the truth values of p and q.
p q p AND q
T T T This means that p AND q is true when p is true and q is true.
T F F This means that p AND q is false when p is true and q is false.
F T F This means that p AND q is false when p is false and q is true.
F F F This means that p AND q is false when p is false and q is false.
The following four statements illustrate the four different cases in the conjunction
truth table:
True conjunction: A fever may accompany a cold and a headache may accompany a
cold.
False conjunction: You are required to pay income tax and pigs often fly.
False conjunction: Aspirin cures cancer and water is wet.
False conjunction: The president’s policies are always best for America and Congress
never passes an unwise bill.
Internet search engines understand the use of the AND operator. If you search for Syria
and Turkey, you will see sites that mention both Syria and Turkey—for example,
refugees from Syria who have fled to Turkey.
In the past, many health insurance policies did not cover preexisting conditions.
That is, they did not cover illness that existed prior to the purchase of the policy.
A salesman for such a policy stated, If you buy this policy, it will cover cases of flu in
your family next winter, and it will cover treatment for your wife’s chronic arthritis.
Was the salesman telling the truth?
SOLUTION
The policy did not cover the preexisting arthritic condition, so that part of the con-
junction was not true. Conjunctions are true only when both parts are true. The
Health insurance card. salesman did not speak truthfully.
Crauder-4100237 QL November 6, 2014 14:34
32 CHAPTER 1
Disjunction The disjunction of two statements is the assertion that either one or
the other is true (or possibly both). Compare this with the conjunction, where both
statements must be true. Disjunction in common English usage is indicated with or.
For symbolic representations, we use OR. Here is an example.
Assume that
p = This medication may cause dizziness
q = This medication may cause fatigue
We use the disjunction to state the warning on my medication: p OR q stands for the
statement This medication may cause dizziness or fatigue.
In common English usage, there is no doubt as to what is meant when we combine
statements using and. There can be a question, however, when we consider the usage
of the word or. In the sample warning, it is clearly intended that the medicine may
cause either or both conditions. This is the inclusive use of or, which means either or
both of the combined statements may be true. By way of contrast, in June of 2012,
Republicans in New Jersey had the opportunity to vote for Ron Paul or Mitt Romney
(the other contenders had withdrawn). That definitely did not mean they could vote
for both. This is an illustration of what is called the exclusive or, meaning that one
choice or the other is allowed, but not both.
In logic, a disjunction is true when either one part is true or both parts are true.
It is false only when both parts are false. Here is the truth table for disjunction.
p q p OR q
T T T This means that p OR q is true when p is true and q is true.
T F T This means that p OR q is true when p is true and q is false.
F T T This means that p OR q is true when p is false and q is true.
F F F This means that p OR q is false when p is false and q is false.
The following four statements illustrate the four different cases in the disjunction
truth table:
True disjunction: A fever may accompany a cold or a headache may accompany a
cold.
True disjunction: You are required to pay income tax or pigs often fly.
True disjunction: Aspirin cures cancer or water is wet.
False disjunction: The president’s policies are always best for America or Congress
never passes an unwise bill.
Internet searches use OR in the inclusive sense. For example, red or blue Toyota will
bring up information about Toyotas that have something to do with red, or blue, or
both.
Crauder-4100237 QL November 6, 2014 14:34
34 CHAPTER 1
Here’s an old joke that uses the mathematical interpretation of the conditional state-
ment:
Sue: “If you had two billion dollars, would you give me one billion?”
Bill: “Sure.”
Sue: “If you had two million dollars, would you give me one million?”
Bill: “Of course.”
Sue: “If you had twenty dollars, would you give me ten?”
Bill: “No!”
Sue: “Why not?”
Bill: “Because I have twenty dollars!”
The point of the joke is that one can promise anything based on an unfulfilled con-
dition and not be accused of lying.
The truth table for the conditional statement is as follows:
p q p→q
T T T This means that p → q is true when p is true and q is true.
T F F This means that p → q is false when p is true and q is false.
F T T This means that p → q is true when p is false and q is true.
F F T This means that p → q is true when p is false and q is false.
President Barack Obama (2ndL), U.S. Vice President Joe Biden (L) and Director
of the White House Office of Management and Budget Sylvia Mathews
Burwell (L) applaud HHS Secretary Kathleen Sebelius for her service during
an event in the Rose Garden at the White House, on April 11, 2014 in
Washington, DC. President Obama announced that Burwell will replace
Secretary Sebelius as secretary of Health and Human Services.
36 CHAPTER 1
p q NOT p (NOT p) → q
T T p is true, and q is true.
T F p is true, and q is false.
F T p is false, and q is true.
F F p is false, and q is false.
When parentheses are present, we always work from “the inside out,” so the next
step is to complete the column corresponding to the “inside,” that is, NOT p. The
truth values of this statement are the opposite of those for p.
p q NOT p (NOT p) → q
T T F NOT p is false.
T F F NOT p is false.
F T T NOT p is true.
F F T NOT p is true.
We complete the table by applying the truth table for → to the last column. Re-
member that the conditional is true except when the premise is true but the conclusion
is false.
Crauder-4100237 QL November 6, 2014 14:34
p q NOT p (NOT p) → q
T T F T NOT p is false, and q is true, so the conditional is true.
T F F T NOT p is false, and q is false, so the conditional is true.
F T T T NOT p is true, and q is true, so the conditional is true.
F F T F NOT p is true, and q is false, so the conditional is false.
The truth value of the compound statement (NOT p) → q is shown in the last
column. Note, for example, that the table tells us that (NOT p) → q is true when p
is true and q is true. This fact is relevant to those people who remember the past and
are condemned to repeat it. In these cases, Santayana’s statement is true.
Then under Proposition 187, the condition that would deny services is
NOT (C OR A). Use a truth table to analyze this statement.
SOLUTION
To begin making the truth table, we list the four possibilities for C and A.
C A C OR A NOT (C OR A)
T T C is true, and A is true.
T F C is true, and A is false.
F T C is false, and A is true.
F F C is false, and A is false.
Crauder-4100237 QL November 6, 2014 14:34
38 CHAPTER 1
Working from the inside out, we next fill in the column for C OR A.
C A C OR A NOT (C OR A)
T T T C is true, and A is true, so C OR A is true.
T F T C is true, and A is false, so C OR A is true.
F T T C is false, and A is true, so C OR A is true.
F F F C is false, and A is false, so C OR A is false.
C A C OR A NOT (C OR A)
T T T F C OR A is true, so NOT (C OR A) is false.
T F T F C OR A is true, so NOT (C OR A) is false.
F T T F C OR A is true, so NOT (C OR A) is false.
F F F T C OR A is false, so NOT (C OR A) is true.
The final truth value of the statement NOT (C OR A) is the last column in the truth
table.
There are three common variations of the conditional that may lead to confusion.
The converse: q → p. If you are a patriotic American, then you support my bill. The
© Hipix/Alamy
This table shows that the conditional and its contrapositive have the same truth
value. Thus, they are termed logically equivalent. Similarly, the converse and inverse
are logically equivalent.
40 CHAPTER 1
EXAMPLE 1.13 Using logic gates and logical operations: Input and output
The logic gate AND takes two inputs, each of them 0 or 1. It corresponds to the
logical operation of conjunction. If the inputs to the AND gate are 1 and 1, what is
the output?
SOLUTION
This situation corresponds to the conjunction p AND q when both p and q are true.
In that case, p AND q is true, so the output of the logic gate is 1.
T T T F
T F F T
F T F T
F F F T
Try It Yourself 1.12: Formulating variants of the conditional: Politics The con-
verse is If you are a conservative, then you are a Republican. The inverse is If you
are not a Republican, then you are not a conservative. The contrapositive is If you
are not a conservative, then you are not a Republican.
Try It Yourself 1.13: Using logic gates and logical operations: Input and output
The output is 0.
42 CHAPTER 1
says to Congresswoman Jones, “Well, you were wrong, we 30. If pigs can fly, then oceans are water.
didn’t pass your bill and the economy has recovered just fine.”
31. If oceans are water, then pigs can fly.
Is there anything wrong here? Explain.
32. Mickey Mouse is president of the United States and
7. Taxing implications. Consider the statement If you don’t
5 + 5 = 10.
pay your taxes by April 15, then you owe a penalty. A tax-
payer is confused because she paid her taxes by April 15 but 33. Mickey Mouse is president of the United States or
she still owes a penalty. Explain her confusion. 5 + 5 = 10.
8–12. More on representing statements. Let p represent the 34. If Mickey Mouse is president of the United States, then
statement You live in Utah, and let q represent the statement 5 + 5 = 10.
You work in Nevada. Use this information in Exercises 8 35. Mickey Mouse is a cartoon character and 5 + 5 = 10.
through 12 to express the given statement in symbolic form.
36. Mickey Mouse is president of the United States or
8. You do not live in Utah and you work in Nevada. 5 + 5 = 11.
9. If you work in Nevada, then you cannot live in Utah.
37. If 5+5 = 12, then Mickey Mouse is president of the United
10. You do not live in Utah or you do not work in Nevada. States.
11. Nobody who works in Nevada can live in Utah. 38–45. The inverse. In Exercises 38 through 45, formulate
12. You do not live in Utah only if you work in Nevada. the inverse of the given statement.
13–19. Translating into symbolic form. In Exercises 13 38. All dogs chase cats.
through 19, translate the given sentence into symbolic form by 39. If you don’t clean your room, I’ll tell your father.
assigning letters to statements that do not contain a negative.
40. I’ll give you a dollar if you clean your room and take out
State clearly what your letters represent.
the trash.
13. If we pass my bill, the economy will recover.
41. I’ll go downtown with you if you agree to go to a movie
14. We didn’t pass your bill, and the economy recovered. or a restaurant.
15. For breakfast I want cereal and eggs. 42. If I am elected, then taxes will be cut.
16. I want cereal or eggs for breakfast. 43. All math courses are important.
17. If you don’t clean your room, I’ll tell your father. 44. If you don’t exercise regularly, your health will decline.
18. I’ll give you a dollar if you clean your room and take out
45. If you take your medicine, you will get well.
the trash.
46–53. The converse. In Exercises 46 through 53, formulate
19. I’ll go downtown with you if we can go to a movie or a
the converse of the given statement.
restaurant.
46. All dogs chase cats.
20–22. Translating into English. Let p represent the state-
ment He’s American, and let q represent the statement He’s 47. If you don’t clean your room, I’ll tell your father.
Canadian. Each of Exercises 20 through 22 has two logically 48. I’ll give you a dollar if you clean your room and take out
equivalent statements. That is, they are just different ways of the trash.
saying the same thing. Write each of these statements in plain
English. 49. I’ll go downtown with you if you agree to go to a movie
or a restaurant.
20. (NOT p) → q and p OR q
50. If I am elected, then taxes will be cut.
21. NOT ( p OR q) and (NOT p) AND (NOT q)
51. All math courses are important.
22. NOT ( p AND q) and (NOT p) OR (NOT q)
52. If you don’t exercise regularly, your health will decline.
23–37. True or false. In Exercises 23 through 37, determine
whether the given statement is true or false. Be sure to explain 53. If you take your medicine, you will get well.
your answer. 54–65. Contrapositives. In Exercises 54 through 65, formu-
23. 3 + 3 = 7 or 2 + 2 = 5. late the contrapositive of the following statements:
24. 3 + 3 = 6 or 2 + 2 = 4. 54. If there is life on Mars, then there must be traces of water
there.
25. If pigs fly, then Mickey Mouse is president.
55. If you are neutral in situations of injustice, you have chosen
26. If George Washington was the first U.S. president, then
the side of the oppressor. (Desmond Tutu)
Mickey Mouse is the current U.S. president.
56. If you have ten thousand regulations, you destroy all re-
27. I will be president if pigs fly.
spect for the law. (Winston Churchill)
28. If pigs can fly, then Earth is flat.
57. If you drink and drive, you will get arrested.
29. If Earth is flat, then pigs can fly.
Crauder-4100237 QL November 6, 2014 14:34
58. If aliens had visited Earth, we would have found concrete 71. NOT ( p AND q) and (NOT p) OR (NOT q)
evidence by now. 72. (NOT p) AND q and NOT ( p OR (NOT q))
59. If you don’t clean your room, I’ll tell your father. 73. Logic gate AND. If the inputs to the AND gate are 0 and
60. I’ll give you a dollar if you clean your room and take out 1, what is the output?
the trash. 74. Logic gate NAND. The logical operation NAND cor-
61. I’ll go downtown with you if you agree to go to a movie responds to negating the conjunction of two statements:
or a restaurant. p NAND q means NOT ( p AND q). The corresponding logic
gate NAND is one of two basic gates, in the sense that every
62. If we continue to produce greenhouse gases, then global
other logic gate can be constructed using it. (The other basic
warming will occur.
gate NOR is defined in Exercise 75.) The gate NAND takes
63. My friends all drive Porsches. two inputs, each of them 0 or 1. If the inputs to the NAND
64. Pneumonia leads to severe respiratory distress. gate are 0 and 1, what is the output?
65. If thunderstorms are approaching, they will be seen on 75. Logic gate NOR. The logical operation NOR corresponds
radar. to negating the disjunction of two statements: p NOR q means
NOT ( p OR q). The corresponding logic gate NOR is one of
66–69. Truth tables. In Exercises 66 through 69, make the two basic gates, in the sense that every other logic gate can be
truth table for the given compound statement. constructed using it. (The other basic gate NAND is defined in
66. p AND (NOT q) Exercise 74.) The gate NOR takes two inputs, each of them 0
or 1. If the inputs to the NOR gate are 0 and 0, what is the
67. p → ( p OR q)
output?
68. p → ( p AND q)
76. History of truth tables. Research the development and use
69. p AND ( p OR q) of truth tables in the late nineteenth and early twentieth cen-
70–72. Equivalence. In each of Exercises 70 through 72, use turies. Be sure to include the contributions of Charles Lutwidge
truth tables to show that the given two statements are logically Dodgson (better known as Lewis Carroll).
equivalent. That is, show that the two statements have the same 77. More research. Find and report on an argument presented
truth values. in the popular media that makes a fallacious argument using
70. (NOT p) → q and p OR q the converse.
44 CHAPTER 1
you answer your friend? Exercise 28 provides additional information relevant to this
topic.
W-8. Contrapositive. The contrapositive of a conditional statement is logically equiv-
alent to the original statement. Give an example in which the contrapositive is more
useful than the original statement for a specific purpose. Exercise 62 provides addi-
tional information relevant to this topic.
W-9. Converse confusion. Why are fallacious arguments involving the converse made
so often?
A nyone who has ever spent time in a conference room equipped with an over-
head projector is familiar with the basic Venn diagram—three overlapping
circles whose eight regions represent every possible intersection of three given
sets, the eighth region being the space around the diagram. Although it resem-
bled the intertwined rings familiar in Christian (and more recently, Led Zeppelin)
iconography, when Cambridge University logician John Venn devised the diagram
in 1880 it was hailed as an innovative way to represent complex logical problems
in two dimensions.
There was just one dilemma, according to British statistician and geneticist
A.W.F. Edwards, author of the entertaining new book Cogwheels of the Mind
(Johns Hopkins): Venn’s diagram didn’t scale up. When a problem involves four
sets, circles don’t have enough possible combinations of overlaps. Ovals work for
four sets, Venn found, but after that one winds up drawing unreadable messes.
What to do?
In Venn’s day, a rival lecturer in mathematics at Oxford by the name of Charles
Dodgson—better known to us as Lewis Carroll—tried to come up with a superior
logical diagram using rectangles but failed (though not before producing an 1887
board game based on his design).
As the article indicates, Venn diagrams are used for representing and solving log-
ical problems. Thus, they are an important tool for critical thinking.
Sets
Underlying the Venn diagram discussed in the preceding article is the idea of a set,
which is nothing more than a collection of objects. This is where we begin.
Crauder-4100237 QL November 6, 2014 14:34
A set is specified by describing its elements. This is sometimes done in words and
sometimes simply by listing the elements and enclosing them in braces. For example,
we can specify the set of all U.S. states that touch the Pacific Ocean by listing the
elements of this set as
{Alaska, California, Hawaii, Oregon, Washington}
California is an element of this set, but Nebraska is not. If we let P stand for the set
of Pacific Ocean states, we would write
P = {Alaska, California, Hawaii, Oregon, Washington}
Washington
Alaska
Oregon
California
Hawaii
The states in the set P of Pacific Ocean states are also elements of the larger set
U of all 50 U.S. states. This is an example of the idea of a subset: P is a subset of U.
At the other extreme, some sets have nothing in common. For example, the set
of U.S. states that touch the Atlantic Ocean shares no members with the set of Pa-
cific states. Whenever two sets have no elements in common, they are said to be
disjoint.
46 CHAPTER 1
d. If V is the collection of all vowels in the alphabet, specify the set as a list of
elements.
SOLUTION
a. There are 2 senators from each of the 50 states, making a total of 100 U.S.
senators. Therefore, there are 100 elements in S.
b. Both senators and representatives are elected officials. So S is a subset of G
and R is a subset of G.
c. The Senate and House have no members in common, so S and R are disjoint.
d. V = {a, e, i, o, u}. (Sometimes y is a vowel, depending on its usage. See
Exercise 45 on page 59.)
A Venn diagram typically consists of either two or three circles inside a rectangle.
The area inside the rectangle represents all the objects that are under consideration
at the time, and the areas inside the circles represent certain sets of these objects.
These diagrams can be a useful tool for answering questions about the number
of elements in a set and for understanding relationships between the elements. For
example, Figure 1.1 shows a Venn diagram4 in which the rectangle represents all
elected government officials. One circle represents senators, and the other circle rep-
resents members of the House of Representatives. As we noted in the solution to
part c of Example 1.14, these two sets are disjoint, and that relationship is indicated
by the fact that the circles do not overlap. The region outside the two circles rep-
resents elected officials who are neither senators nor representatives. The president
and vice president would go in this region.
In Figure 1.2, we have added a circle that represents Republican senators. This
new circle is inside the “Senators” circle, indicating that the set of Republican sena-
tors is a subset of the set of all senators.
4 Technically, Figure 1.1 and Figure 1.2 are known as Euler diagrams, but we will not bother with that distinction here.
Crauder-4100237 QL November 6, 2014 14:34
Senators
FIGURE 1.1 Senators and FIGURE 1.2 Republican senators are a subset
representatives are disjoint. of senators.
Democrats Women
Male
Democrats
Female
Male Republicans
Republicans
Female Democrats
A more common example of a Venn diagram is shown in Figure 1.3, where the
two circles overlap. In this diagram, the category under discussion is all U.S. rep-
resentatives. One circle represents the set of Democratic members, and the other
circle represents female members. There are four distinct regions in the diagram. (At
the time of this writing, all members of the House of Representatives were either
Democrats or Republicans.)
• The region common to both circles represents female Democrats.
• The region inside the “Democrat” circle but outside the “Women” circle
represents male Democrats.
• The region inside the “Women” circle but outside the “Democrat” circle
represents female Republicans.
• The region outside both circles represents members who are neither Democrats
nor female. These are male Republicans.
48 CHAPTER 1
4,
2, 3 5, 7, 8
6
B
1, 9, 10
To further clarify what the various regions in a Venn diagram represent, let the
category be the whole numbers from 1 to 10. Let A = {2, 3, 4, 5, 6} and B =
{4, 5, 6, 7, 8}. The corresponding Venn diagram is shown in Figure 1.4. It shows
which elements go in each part of the diagram.
positive
True
False False
negative positive
True
negative
FIGURE 1.5 A Venn diagram of medical tests FIGURE 1.6 The completed diagram.
for hepatitis C.
Venn diagrams can be useful in helping to count the number of items in a given
category.
5 A placebo is a harmless substance that contains no medication and that is outwardly indistinguishable from the real
drug. We discuss clinical trials in Section 4 of Chapter 6.
Crauder-4100237 QL November 6, 2014 14:34
50 CHAPTER 1
30 20 10
40
FIGURE 1.7 The test group and FIGURE 1.8 The completed Venn diagram.
improvement group.
Our conditional can be recast as everyone who is stopped for speeding will get
a traffic ticket. This means that S is a subset of T. Thus, to represent this statement
using a Venn diagram, we put the circle for S inside the circle for T. The result is
shown in Figure 1.9. This figure makes it clear why we cannot necessarily conclude
the converse: If you get a traffic ticket, then you were stopped for speeding. Any
driver in between the two circles is a driver who got a ticket but was not stopped for
speeding (Figure 1.10). The Venn diagram for the converse is shown in Figure 1.11.
FIGURE 1.9 Venn diagram for a FIGURE 1.10 Why the converse may not FIGURE 1.11 Diagram for the converse:
conditional: If you are stopped be true. If you get a traffic ticket, then you were
for speeding, you get a ticket. stopped for speeding.
Crauder-4100237 QL November 6, 2014 14:34
Note that the roles of the circles are reversed. The only time a statement and its
converse are both true is when the two circles in the Venn diagram are the same.
We can also represent the conjunction and disjunction using Venn diagrams. For
the conjunction He is a math major and a music major, let A denote the set of all
math majors and B the set of all music majors:
A = {Math majors}
B = {Music majors}
To say he is a math and a music major means that he belongs to both sets. These
double majors are shown in Figure 1.12. The shaded region in that figure is the
region of the Venn diagram where the conjunction is true.
The disjunction: Joe is a math major or a music major means that Joe belongs
to one set or the other. Figure 1.13 illustrates the corresponding region of the Venn
diagram.
FIGURE 1.12 A conjunction is true in the FIGURE 1.13 A disjunction is true in either
overlap of the sets. one or both sets.
a. Make a Venn diagram for F and T, and shade in the region for which the
statement He voted for the tax bill but not the farm bill is true.
b. Let S denote the set of all Southern senators. Suppose it is true that all
Southern senators voted for both the tax and farm bills. Add the circle
representing Southern senators to the Venn diagram you made in part a.
SOLUTION
a. The statement is true for all senators who are in the set T but not in the set
F . That group of senators is inside the “Tax” circle but outside the “Farm”
circle. We have shaded this region in the Venn diagram shown in Figure 1.14.
b. Because all Southern senators voted for both bills, they all belong to both F
and T. That makes S a subset of the overlap of the two circles. Thus, the circle
representing S is inside the region common to both F and T, as shown in
Figure 1.15.
Crauder-4100237 QL November 6, 2014 14:34
52 CHAPTER 1
F = Voted for farm bill T= Voted for tax bill F = Voted for farm bill T= Voted for tax bill
FIGURE 1.14 Senators voting for the tax bill FIGURE 1.15 All Southern senators (S) voted
but not the farm bill. for both bills.
Using these sets, we make the Venn diagram shown in Figure 1.16. The rectangle
represents all the tourists.
The next step is to fill in the numbers we know. Now 10 tourists visit Mexico and
not Costa Rica, so this number goes in the “Mexico” circle but outside the “Costa
Rica” circle. Similarly, 5 people visit both countries, and this is the number that goes
in the overlap of the two circles. In Figure 1.17, we have entered these numbers
and used n to denote the number who visited Costa Rica but not Mexico. Because
the sum of all the numbers in the two circles gives the total number of tourists,
we find
10 + 5 + n = 17
10 5 n
FIGURE 1.16 Venn diagram for tourists. FIGURE 1.17 Entering known values.
Crauder-4100237 QL November 6, 2014 14:34
which gives n = 2. So 2 tourists visited Costa Rica but not Mexico. Adding this
number to the 5 tourists who visited both countries gives 7 visitors to Costa Rica.
8 + 19 + 41 + n = 100
This gives n = 32. We know that 41 passed the math exam but not the science exam
and that 19 passed both. Hence, 41 + 19 = 60 passed the math exam. Also, 32
students passed the science exam but not the math exam, and 19 passed both. That
gives 32 + 19 = 51 who passed the science exam.
41 n Passed
19 neither
Passed
Passed math, science,
not science Passed both not math
FIGURE 1.18 Venn diagram for science FIGURE 1.19 Entering given data.
and math exams.
54 CHAPTER 1
Road (34%)
Other (1%) Road only
Football only 24
Basketball only 13
Baseball only 12
Football and baseball, but not basketball 9
Football and basketball, but not baseball 5
Basketball and baseball, but not football 10
All three sports 6
6 This information is taken from the California 2006 Five Percent Report. It is available at the Federal Highway
Administration Web site.
Crauder-4100237 QL November 6, 2014 14:34
Now we can answer the question in part a. There is one region that has no
number in it. That is the one that is outside all three circles—the number who
56 CHAPTER 1
F = Football B = Basketball
24 5 13
6
9 10
Played
no sports
12
31
110 surveyed S = Baseball
FIGURE 1.22 Adding the number who play none of these sports.
play none of these three sports. To find this number, we note that if we add up
all the numbers, including those who play none of these sports, we get the total
number surveyed: 110. So the number of students who play none of these
sports is
110 − 24 − 13 − 12 − 9 − 5 − 10 − 6 = 31
for healthy people. A large percentage in this category is cause for concern. Positive
results for healthy people are called false positives. Tests with a large percentage of
false positives are to some degree unreliable.
Further questions of this type may be found in the What Do You Think section of
exercises.
Texas
Oregon Florida
Gulf States
Try It Yourself 1.16: Making Venn diagrams with numbers: Reading proficiency
The rectangle represents all who were tested.
I = Inner city
20 30 40
10
P = Passed
Try It Yourself 1.18: Counting using Venn diagrams: Exam results 65.
58 CHAPTER 1
7. A is the set of Republican governors, B is the set of gover- H of history majors and the set P of political science majors at
nors, and C is the set of elected state officials. a certain college.
8. Miscounting. A store sells only bottled water and sodas
(some of which come in bottles). A clerk wants to know how
many items a customer wants to buy, so he asks how many H = History
sodas and how many bottles there are. The customer answers
truthfully that there are 10 sodas and 10 bottles. Should the
clerk conclude that there are 20 items altogether? Can the clerk
determine the number of items using the given information?
Use a Venn diagram to explain your answers.
9. Countries. Make a Venn diagram where the category is
countries of the world. Use A to denote countries that touch
the Atlantic Ocean and N to denote countries of the Northern
Hemisphere. On your diagram, locate these countries: United P = Political science
States, Brazil, Japan, and Australia.
10. Plants. Make a Venn diagram where the category is an- FIGURE 1.24 History and political
imals. Use M for mammals and P for predators. Locate the science majors.
following on your diagram: tiger, elephant, turtle, and shark.
20. Shade in the region of Figure 1.24 where the statement She
11. Fruit. For this exercise, the category is fruit. Make a Venn is a political science major or a history major is true.
diagram using the set Y of yellow fruit and the set C of cit-
rus fruit. Locate the following fruits in your diagram: banana, 21. Shade in the region of Figure 1.24 where the statement She
cherry, lemon, and orange. is a political science major but not a history major is true.
12–19. Using a Venn diagram. Figure 1.23 shows a Venn di- 22. Shade in the region of Figure 1.24 where the statement She
agram for students at a certain college. Let H denote the set is a political science major and a history major is false.
of history majors, M the set of male students, and A the set of 23. Let G be the set of all students from Greece. Add a circle
students receiving financial aid. Exercises 12 through 19 refer for G to the Venn diagram in Figure 1.24 so that the statement
to this diagram. All Greek students are history majors but none are political
science majors is true.
H = History major A = Financial aid 24. Let G be the set of all students from Greece. Add a “circle”
for G to the Venn diagram in Figure 1.24 so that the statement
Some Greek students are political science majors and some
23 30 821 are history majors but none are double majors is true. (This
10 “circle” will not be round!)
5 405
25–28. PSA test. The following table presents partial data
525 concerning the accuracy of the PSA (Prostate Specific Antigen)
test for prostate cancer. Complete Exercises 25 through 28, in
170
order, using the table.7
M = Males
• False negatives: These are people who test negative but do Sodas only 28
have prostate cancer. Coffee only 17
26. Use the table to fill in the number of people in as many of Water 51
Sodas and water 26
the regions of your Venn diagram as you can. For which region
Coffee and water 27
can you not determine a number? Sodas and coffee but not water 9
27. Assume that the total number of people tested was 10,000. Coffee and water, not sodas 6
Fill in the number you were missing in Exercise 26. All three 21
28. The sensitivity of a medical test is defined to be the number 33. Use S for soda, W for water, and C for coffee, and make a
of true positives divided by the number who have the disease. Venn diagram for drinks. The completed diagram should show
Find the sensitivity of the PSA test. Express your answer as a the number in each region.
percentage. Note: The sensitivity is one measure of the accu-
34. How many liked coffee only?
racy of a medical test—accurate tests have a high sensitivity.
These topics are discussed in Section 2 of Chapter 5. 35. How many liked water but not coffee?
29. A drug trial. In a drug trial, a test group of 50 patients is 36. How many liked none of the three?
given an experimental drug, and a control group of 40 patients 37. How many liked coffee and water?
is given a placebo. Assume that 15 patients from the control
38. How many liked exactly two of the three?
group improved and 30 from the test group improved. Let C
denote the control group and I the group of patients who im- 39. How many liked exactly one of the three?
proved. Make a Venn diagram using C and I. Show the number 40–44. News. Complete Exercises 40 through 44, in order,
of patients in each region of the diagram. using the following information. A survey asked a group of
30. Weather. Over a period of 100 days, conditions were people where they got their news: a newspaper, TV, or the In-
recorded as sunny or cloudy and warm or cold. Let S denote ternet. The responses are recorded in the following table.
sunny days and W warm days. Of 75 sunny days, 55 were
warm. There were only 10 cold and cloudy days. Make a Venn Newspaper and TV 46
diagram showing the numbers in each region of the diagram. TV only 20
Internet only 12
How many days were warm?
Newspaper and Internet 57
31. Cars. Suppose that 178 cars were sorted by model and TV and Internet 50
color. Assume that 10 were Fords but not blue, 72 were blue Newspapers 81
but not Fords, and 84 were neither blue nor Fords. Make a TV and Internet, but not newspapers 19
All three 31
Venn diagram using Fords and blue cars, and include the num-
None of the three 8
bers in each region of the diagram. How many blue cars were
observed?
40. Using N for a newspaper, T for TV, and I for the Inter-
32. Employed freshmen. A certain college has 11,989 students. net, make a Venn diagram for news sources. The completed
Assume that 6120 are unemployed sophomores, juniors, or se- diagram should give the number in each region of the diagram.
niors and that 4104 are employed sophomores, juniors, or se- 41. How many were surveyed?
niors. There are 745 employed freshmen. Use F for freshmen
and E for employed students, and make a Venn diagram mark- 42. How many got their news from only one source?
ing the number of students in each region of the diagram. How 43. How many got some of their news from the Internet?
many freshmen are there?
44. How many got none of their news from TV?
33–39. Beverage survey. Complete Exercises 33 through 39, 45. Research on vowels. In the solution to Example 1.14, it
in order, using the following information. A survey of 150 stu- was stated that the set of vowels is V = {a, e, i, o, u} and that
dents asked which of the following three beverages they liked: sometimes y is a vowel, depending on its usage. Look up the
sodas, coffee, or bottled water. The results are given in the fol- meaning of the word “vowel” and write a report on what you
lowing table. discover.
Crauder-4100237 QL November 6, 2014 14:34
60 CHAPTER 1
1.5 Critical thinking and number sense: What do these figures mean? 61
A t $680 billion, the fiscal 2013 deficit is 51% less than it was in 2009, when
it hit a record high nominally of $1.4 trillion.
As a percent of the economy, it’s also considerably smaller than it’s been in
the past five years, coming in at 4.1% of gross domestic product. By contrast,
the annual deficit in 2009 topped 10% of GDP. And last year it was 6.8%.
Overall, Treasury said higher receipts accounted for 79% of the decline in the
deficit from last year.
$680B
$455B
Amounts of money like the $680 billion deficit are so large that their meaning
may be lost.8 Perhaps it helps to note that if every man, woman, and child in the state
of Oklahoma bought a $25,000 automobile, they wouldn’t come close to spending
$680 billion. In fact, there would be enough money left over to buy $25,000 au-
tomobiles for each living soul in the states of Iowa, Kansas, Louisiana, Minnesota,
Nebraska, New Mexico, North Dakota, and South Dakota.
Making informed decisions often depends on understanding what very large or
very small numbers really mean. Putting such numbers into a personal context helps
us get a sense of their true scale. We can also use estimation to sort our way through
otherwise intimidating calculations. For a discussion about problem solving, see
Appendix 5.
8 Everett Dirksen, a senator from Illinois several decades ago, is often quoted as saying, “A billion here, a billion there,
and pretty soon you’re talking real money.”
Crauder-4100237 QL November 6, 2014 14:34
62 CHAPTER 1
Examples of powers of 10
Positive powers of 10
• 103 = 1000 (three zeros) is a thousand.
• 106 = 1,000,000 (six zeros) is a million, which is a thousand thousands.
• 109 = 1,000,000,000 (nine zeros) is a billion, which is a thousand millions.
• 1012 = 1,000,000,000,000 (twelve zeros) is a trillion, which is a thousand
billions.
Negative powers of 10
• 10−2 = 0.01 (two digits to the right of the decimal) is a hundredth.
• 10−3 = 0.001 (three digits to the right of the decimal) is a thousandth.
• 10−6 = 0.000 001 (six digits to the right of the decimal) is a millionth.
• 10−9 = 0.000 000 001 (nine digits to the right of the decimal) is a billionth.
The gigantic number 10100 is a “one” followed by 100 zeros and actually has the
funny name googol. (It’s an interesting fact that the name of the company Google
is a misspelling of googol.) Physicists believe there are only about 1080 atoms in the
universe—far fewer than a googol.
To get a better grasp of these magnitudes, we refer to the figures on the facing
page. The first figure is a view of space through a frame that is 1021 meters on each
side.9 That frame is large enough to show the entire Milky Way galaxy, which has
a diameter of about 100,000 lightyears. (A lightyear is the distance light travels in
a year, about 1016 meters.) The pictures decrease in scale by powers of 10 until we
finally view the structure of a DNA molecule through a frame that is 10−9 meter on
a side.
Dealing with powers of 10 sometimes requires a bit of algebra with exponents.
The following quick review is included for those who need a refresher. Additional in-
formation about exponents, including scientific notation, is provided in Appendix 2.
ap 106
= a p−q = 106−4 = 102 = 100
aq 104
2
(a p )q = a pq 103 = 103×2 = 106 = 1,000,000
1.5 Critical thinking and number sense: What do these figures mean? 63
Photodisc
NASA
San Francisco Bay is about 105 meters, or Some lily flowers are about 0.1, or 10−1 ,
100 kilometers, long. meter across.
Dalyne A. Murawski/National Geographic Creative
Scott Camazine/Alamy
The eye of a bee is about 10−3 meter across. A DNA molecule is about 10−9 meter across.
64 CHAPTER 1
Thus, there are 1000 megabytes in a gigabyte. Similarly, we calculate that there are
a million megabytes in a terabyte.
1.5 Critical thinking and number sense: What do these figures mean? 65
Harley Schwadron/cartoonstock.com
At the other end of the scale from numbers describing the national debt or com-
puter memory are extremely small numbers. Microchips that can be implanted in
pets are about 0.0015 meter in diameter. That is 1.5 millimeters—about the size of
Crauder-4100237 QL November 6, 2014 14:34
66 CHAPTER 1
an uncooked grain of rice. The eye of a bee, which is about 1 millimeter across (see
the figure on p. 63), may provide additional context for this magnitude.
A smaller magnitude is a micron, which is one-millionth of a meter or 10−6 meter.
In this scale, you could get a close-up view of a bacterium.
Even smaller are the particles used in the developing field of nanotechnology,
which is the control of matter on an atomic or molecular scale. It may be possible
to develop nanoparticles in the 10- to 100-nanometer range that can seek out and
selectively destroy cancer cells. One nanometer is 10−9 meter, a billionth of a meter.
On this scale, we can view the structure of DNA, as we saw in the illustration on
p. 63. We further develop our sense of these very small sizes in the next example.
Thus, 5000 of the nanoparticles would be needed to span the diameter of a human
hair.
1.5 Critical thinking and number sense: What do these figures mean? 67
68 CHAPTER 1
In Canada (and most of the rest of the world), gasoline is measured in liters
rather than gallons. Also, the dollar sign refers not to U.S. dollars but to Canadian
dollars. So the sign on the Canadian side of the border means that gasoline costs 1.10
Canadian dollars per liter.
In parts a and b, use an exchange rate of 0.91 U.S. dollar per Canadian dollar.
a. Use the fact that a quart and a liter are nearly the same and that the
Canadian dollar is worth a little less than the U.S. dollar to get a quick estimate
of the cost of gasoline in Canada measured in U.S. dollars per gallon.
b. There are 3.79 liters in a gallon. Use this fact to find the actual cost in U.S.
dollars per gallon of gasoline at the Canadian station. Was your estimate in part
a good enough to tell you where you should buy your gas?
SOLUTION
a. Because 1 Canadian dollar is worth about 1 U.S. dollar and 1 liter costs
about 1 Canadian dollar, gas costs about 1 U.S. dollar per liter. Because a quart
is about the same as a liter, there are about 4 liters in a gallon. This means that
gas in Canada costs approximately 4 U.S. dollars per gallon. This estimate
suggests that gas is cheaper at the station in the United States.
b. Because there are 3.79 liters in a gallon, gasoline in Canada costs
——
liters Canadian dollars Canadian dollars
3.79 × 1.10 = 4.169
gallon liter
—– gallon
One way to check that the units are converted correctly is to include them in the
computation, as we did here, and combine or cancel them using the usual rules
of algebra. Note that the units involving liters cancel. Further information on
unit conversion is in Appendix 1.
Crauder-4100237 QL November 6, 2014 14:34
1.5 Critical thinking and number sense: What do these figures mean? 69
Areas and volumes are important measurements for consumers. For example,
carpet is priced in both square feet and square yards. There are 3 feet in a yard. How
many square feet are in a square yard? In Figure 1.25, we have made a picture of a
square yard. It has a length and width of 3 feet. Recall that the area of a rectangle is
the length times the width. So the area in square feet of a square yard is
Area of 1 square yard = Length × Width
= 3 feet × 3 feet
= 9 square feet
Concrete is usually priced by the cubic yard.11 Figure 1.26 shows that 1 cubic
yard is a cube that is 3 feet on each side. Recall that the volume of a box is the length
times the width times the height. So the volume in cubic feet of a cubic yard is
Volume of 1 cubic yard = Length × Width × Height
= 3 feet × 3 feet × 3 feet
= 27 cubic feet
So 1 square yard is 9 square feet, and 1 cubic yard is 27 cubic feet.
The next example illustrates that savvy consumers need both the skill of estima-
tion and the ability to measure areas and volumes.
t
1 foot fee
3
1 foot
3 feet
3 feet
11 Often sellers of concrete quote the price “per yard.” What they really mean is the price per cubic yard.
12 Often sellers of carpet quote the price “per yard.” What they really mean is the price per square yard.
Crauder-4100237 QL November 6, 2014 14:34
70 CHAPTER 1
a. You need to decide right away whether you should be looking at hardwood
or carpet for your floor. Use the fact that the cost of hardwood is about $10 per
square foot to estimate the cost of a square yard of hardwood. Use your
estimate to decide how the cost of hardwood compares with the cost of carpet.
b. Find the actual cost of a square yard of hardwood.
SOLUTION
a. A common (and ultimately expensive) consumer error is to think that,
because 1 yard is 3 feet, 1 square yard is 3 square feet. But as we noted above,
1 square yard is, in fact, 9 square feet. Because hardwood costs about $10 per
square foot, 9 square feet costs about $90. Because the carpet is $28 per yard,
we estimate that hardwood is more than three times as expensive as carpet.
b. To find the actual cost of a square yard of hardwood, we multiply the exact
cost of $10.40 per square foot by 9 square feet:
$10.40 × 9 = $93.60
The result is a bit higher than the estimate of $90 we found in part a, and it
confirms our conclusion that hardwood is much more expensive than carpet.
1.5 Critical thinking and number sense: What do these figures mean? 71
13 See www.businessinsider.com.
Crauder-4100237 QL November 6, 2014 14:34
72 CHAPTER 1
24–26. World population. As of early 2014, the world’s pop- Texas, how many people would there be in each square mile?
ulation was about 7.2 billion people. The population of the Round your answer to the nearest whole number.
United States was 317 million. This information is used in Ex- 26. The area of New York City is 305 square miles. If in early
ercises 24 through 26. 2014 every person in the world moved into New York City,
24. In early 2014, what percent of the world’s population was how many square feet of space would each person have? Round
the U.S. population? Round your answer as a percentage to one your answer to one decimal place. Note: There are 5280 feet
decimal place. in a mile. First determine how many square feet are in a square
25. The land area of the state of Texas is about 261,797 square mile.
miles. If in early 2014 all the world’s people were put into
CHAPTER SUMMARY
Today’s media sources bombard us with information and misinformation, often with
the aim of selling us products, ideas, or political positions. Sorting the good from the
bad is a constant challenge that requires a critical eye. Solid reasoning is a crucial
tool for success.
Chapter Summary 73
Males Females
Applicants Accepted % accepted Applicants Accepted % accepted
Medical school 10 2 20.0% 20 4 20.0%
Law school 20 5 25.0% 12 3 25.0%
Total 30 7 23.3% 32 7 21.9%
Note that the medical school accepts 20% of male applicants and 20% of female
applicants. The law school accepts 25% of male applicants and 25% of female appli-
cants. But overall 23.3% of male applicants are accepted and only 21.9% of female
applicants are accepted. The discrepancy is explained by the fact that females tended
to apply to the medical school and males tended to apply to the law school—and the
medical school accepts fewer applicants (on a percentage basis) than does the law
school. The overall average might lead one to suspect gender bias where none exists.
74 CHAPTER 1
Get tickets
Females Democrats
Speeders
KEY TERMS
logic, p. 16 valid, p. 16 inductive, p. 23 elements, p. 45
premises, p. 16 informal fallacy, p. 17 or, p. 32 subset, p. 45
hypotheses, p. 16 formal fallacy, p. 17 logically equivalent, p. 39 disjoint, p. 45
conclusion, p. 16 deductive, p. 23 set, p. 45 Venn diagram, p. 46
CHAPTER QUIZ
1. The following tables show, for two declining towns, the Year 2005
number of workers and the total weekly earnings at the be-
ginning of the given year. Number of Total weekly Average weekly
workers earnings earnings
Town A 500 $320,000
Town B 100 $70,000
Total
Crauder-4100237 QL November 6, 2014 14:34
Chapter Quiz 75
Year 2015 5. Let p represent the statement You study for your math exam
and q the statement You pass your math exam. Use these let-
Number of Total weekly Average weekly ters to express symbolically the statement If you don’t study
workers earnings earnings for your math exam, you fail the math exam.
Town A 50 $31,000 Answer (NOT p) → (NOT q).
Town B 60 $42,000 If you had difficulty with this problem, see Example 1.10.
Total 6. Make the truth table for p → (NOT q).
Answer
Complete the table by filling in the blank spaces. Round each
average to the nearest dollar. p q NOT q p → (NOT q)
T T F F
Answer T F T T
F T F T
Year 2005 F F T T
Number of Total weekly Average weekly
workers earnings earnings If you had difficulty with this problem, see Example 1.11.
7. Tests for drug use are not always accurate. Make a Venn
Town A 500 $320,000 $640
Town B 100 $70,000 $700 diagram using the categories of drug users and those who test
positive. Shade the area of the diagram that represents false
Total 600 $390,000 $650 negatives, that is, drug users who test negative.
Answer The region for false negatives is shaded.
Year 2015
Number of Total weekly Average weekly Test positive
workers earnings earnings
Town A 50 $31,000 $620
Town B 60 $42,000 $700
Total 110 $73,000 $664
2 ANALYSIS OF GROWTH
T
he following article appears at the Gigaom Web site.
2.1 Measurements of
growth: How fast is it IN THE NEWS 2.1 Search
This article, along with Figure 2.1, makes clear that the growth rate of Facebook
is dramatic and has resulted in a tremendous increase in its value---but without the
figure, the growth might not be as vivid.
In this technological age, it is easy to be overwhelmed with information, which
frequently comes in the form of numerical data: the number of cases of flu each year,
the variation in the inflation rate over the past decade, the growth rate of Facebook,
and so forth. Presenting data graphically, as in Figure 2.1, makes them easier to
digest.
In this chapter, we will explore further how information can be presented using
different kinds of graphs and tables, and we will examine their advantages and dis-
advantages. We will learn to analyze data, especially growth rates, and to make pre-
dictions. We will also see that one should learn to be wary of graphical presentations
and to recognize when data are misrepresented and when inaccurate conclusions are
drawn. Empowering you to analyze tables and graphs critically is the ultimate goal
of this chapter.
77
Crauder-4100237 QL November 6, 2014 15:7
78 CHAPTER 2
1200
800
600
400
200
d 04
d 05
ct il 6
Au ber 007
Ja ust 007
br y 08
Ap y 20 9
ril 09
em ly 9
r 9
br d 09
r y 09
y 0
d 10
em ly 0
r 1
M d 2 11
Se J ch 2 11
D tem e 2 12
em r 2
r2 2
2
O pr 200
r 0
pt Ju 200
be 200
Jul 201
pt Ju 201
be 201
ec be 01
be 201
01
En 20
En 20
Fe nuar 20
ua 20
Fe En 20
ua 20
En 20
En 20
ar 0
p un 0
o 2
g 2
d
En
Se
Se
Date
I NDIANAPOLIS---The NCAA says football and men’s basketball players are be-
coming more productive in the classroom.
A one-year measurement, released Thursday, showed that 70 percent or more
of Division I athletes who were freshmen in 2005--06 in those sports earned their
diplomas---the first time that has happened since the governing body started col-
lecting data 11 years ago for the annual Graduation Success Rate.
Athletes in men’s basketball graduated at a rate of 74 percent, a 6 percentage-
point jump over the 2004–05 freshman class. Football Bowl Subdivision ath-
letes improved their scores by 1 percentage point over the previous year, hitting
70 percent.
The focus of this article is the trend in graduation rates—how they change over
time, which is crucial for understanding the current situation.
Although raw information can seem overwhelming, certain basic principles are
helpful for mining data. In this section, we examine and analyze quantitative informa-
tion presented in tables and bar graphs. The basic measures of growth are important
keys for analysis.
Crauder-4100237 QL November 6, 2014 15:7
1 The terminology of independent and dependent variables should not be interpreted to imply that one variable
somehow causes the other.
Crauder-4100237 QL November 6, 2014 15:7
80 CHAPTER 2
When we look at the census table, it is natural to pick the date and then look
up the population for that date. This makes sense because the population depends
on the date. In the case of the census, the date is the independent variable and the
population is a function of this variable because it depends on the date. In this
table, the first column lists the values of the independent variable and the second
column lists the values of the dependent variable. When data tables list the variables
vertically in columns, it is customary for the first column to represent the indepen-
dent variable and the second column to represent the dependent variable. When a
data table lists the values of the variables in rows, it is typical for the first row to
represent the independent variable and the second row to represent the dependent
variable.
What would you label as the independent and dependent variables for this table?
Explain what the corresponding function means.
SOLUTION
We would choose the year to be the independent variable and the number of medals
won to be the dependent variable. Then the number of medals won is a function of
the year. The function gives the most medals won by any country in a given year.
Crauder-4100237 QL November 6, 2014 15:7
Silver medalists Elana Meyers and Lauryn Williams of the United States
team 1, gold medalists Kaillie Humphries and Heather Moyse of Canada
team 1, and bronze medalists Jamie Greubel and Aja Evans of the United
States team 2 celebrate on the podium during the medal ceremony for the
Women’s Bobsleigh on day thirteen of the Sochi 2014 Winter Olympics at
Medals Plaza on February 20, 2014 in Sochi, Russia.
Note that in this example it would not be appropriate to label the date as the
dependent variable because there were three different years in which 29 medals were
won. Therefore, the year doesn’t depend only on the number of medals won. Another
way to say this is that the year is not a function of the number of medals won.
Choose independent and dependent variables for this table and explain what the
corresponding function means.
The answer is provided at the end of this section.
Data tables are meant to convey information. But the tables themselves are some-
times dry and difficult to decipher. There are important tools that help us understand
Crauder-4100237 QL November 6, 2014 15:7
82 CHAPTER 2
300
250
Population (millions)
200
150
100
50
0
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010
Year
information presented through a visual display. One of these is the bar graph. Bar
graphs are useful in displaying data from tables of relatively modest size. For large
data tables, other types of visual representation may be more appropriate. To make
a bar graph of the census data in Table 2.1, we let the horizontal axis represent the
independent variable (the date in this case), and we let the heights of the vertical bars
show the value of the function (the U.S. population, in millions) for the correspond-
ing date. The completed bar graph is in Figure 2.3. It shows how the population
increased with time.
Percentage change
The apparent regularity of the graph in Figure 2.3 hides some important information.
We can find out more about U.S. population growth by calculating the percentage
change, that is, the percentage increase or percentage decrease. Population growth is
often measured in this way, and in the article opening this section, the growth rate
of Facebook was measured in terms of percentages.
The formula for percentage change from one function value to another is
Change in function
Percentage change = × 100%
Previous function value
In the case of the census data, this change is the percentage increase in population
from decade to decade. The following Quick Review is provided as a reminder of how
to make percentage calculations.
Crauder-4100237 QL November 6, 2014 15:7
84 CHAPTER 2
This is about 0.35 × 100% or 35%. Between 1790 and 1800, the population in-
creased by 35%.
The percentage change during each decade is shown in the accompanying table
and as a bar graph in Figure 2.4.
As we look at each percentage change, some things stand out. Note that between
1790 and 1860, the percentage changes in each decade are approximately the same—
between 33% and 36%. That is, the percentage increase in population was nearly
constant over this period.
40
35
30
Percent growth
25
15
10
0
1790–1800
1800–1810
1810–1820
1820–1830
1830–1840
1840–1850
1850–1860
1860–1870
1870–1880
1880–1890
1890–1900
1900–1910
1910–1920
1920–1930
1930–1940
1940–1950
1950–1960
1960–1970
1970–1980
1980–1990
1990–2000
2000–2010
Decade
FIGURE 2.4 Left: Percentage change in U.S. population. Right: Engineers of the 8th N.Y. State Militia, 1861.
Crauder-4100237 QL November 6, 2014 15:7
The decade of the Civil War (1860–1870) jumps out. Even if we didn’t know the
historical significance of this decade, the bar graph of percentage change would alert
us that something important may have occurred in that decade. Observe that the
significance of this decade is not apparent from the graph of population shown in
Figure 2.3. It is interesting to note that percentage population growth never returned
to its pre-Civil War levels. The percentage change generally declined to its current
level of 10% to 13% per decade.
a. Identify the independent variable and the function, and make a bar graph
that displays the data.
b. Make a table and bar graph showing percentage changes between decades.
c. Comment on what your graphs tell you. Compare the second graph with the
corresponding graph for U.S. population growth.
SOLUTION
a. The independent variable is the date, and the function gives the world
population (in billions) on that date. To create a bar graph, we list dates across
the horizontal axis and make bars representing population for each date. The
finished product is in Figure 2.5.
b. We calculate the percentage change from one date to the next using
Change in population
Percentage change = × 100%
Previous population
So the percentage increase from 1950 to 1960 is
3.04 − 2.56
Percentage change = × 100%
2.56
or about 19%.
The percentage changes for other decades are calculated in a similar fashion.
Crauder-4100237 QL November 6, 2014 15:7
86 CHAPTER 2
The table is given below, and the bar graph is in Figure 2.6.
7 25
Population (billions)
Percentage increase
6 20
5
4 15
3 10
2
5
1
0 0
1950 1960 1970 1980 1990 2000 2010 1950– 1960– 1970– 1980– 1990– 2000–
1960 1970 1980 1990 2000 2010
Year Year
c. The world population is steadily increasing, but from 1960 on, the
percentage change is declining. Starting from 1960, the percentage change drops
from 22% to 12% per decade. The percentage increase for the world
population is still larger than that for the United States.
Change in function
Average growth rate =
Change in independent variable
This result tells us that from 1940 to 1950 the U.S. population grew, on average, by
about 1.92 million people each year.
Note that the average growth rate from 1940 to 1950 is not the same as that
from 1950 to 1960. In general, average growth rates vary from decade to decade. In
some cases, the average growth rate may be negative. This indicates a decline rather
than an increase. The following example illustrates this possibility.
88 CHAPTER 2
In practical settings, the average growth rate almost always has a familiar mean-
ing, and proper use of units can help determine that meaning. For example, if the
function gives the distance driven in an automobile and the independent variable is
time, the average growth rate is the average velocity, which typically is measured in
miles per hour. Thus, if you make a 100-mile trip and the trip takes 2 hours, the
average growth rate is 50 miles per hour. In other settings, the average growth rate
may represent the average number of traffic accidents per week, the average number
of births per thousand women, or the average number of miles per gallon.
Interpolation
Once we know the average rate of change, we can obtain more information from the
data. Suppose we want to estimate the U.S. population in the year 1945. According to
Example 2.4, between 1940 and 1950, the U.S. population grew at an average rate of
1.917 million people per year. So it is reasonable to assume that over the 5-year period
from 1940 to 1945, the population increased by approximately 5 × 1.917 = 9.585
million. We know that the population in 1940 was 132.16 million. We can then
estimate that the population in 1945 was
Population in 1940 + 5 years of increase = 132.16 + 9.585 = 141.745 million
or about 141.75 million.
Using average growth rates to estimate values between data points is known as
interpolation.
3 Data from the annual CIRP Freshman Surveys by the Higher Education Research Institute at UCLA.
Crauder-4100237 QL November 6, 2014 15:7
SOLUTION
a. We want to interpolate, so first we find the average growth rate. The change in
the independent variable from 2007 to 2011 is 4 years. The change in the
dependent variable over that period is 49.1 − 38.2 = 10.9 percentage points.
Hence, the average growth rate from 2007 to 2011 was:
This means that, on average, the increase was 2.725 percentage points each
year over this period. The year 2008 represents 1 year of increase from 2007, so the
estimated percentage in 2008 is about
b. The year 2009 represents 2 years of increase from 2007, so the estimated
percentage in 2009 is about
Our estimate of 43.7% for 2009 is less than the actual figure of 45.6%. This means
that the growth rate of the percentage was lower from 2007 to 2009 than it was
from 2009 to 2011.
90 CHAPTER 2
states that the national debt has continued to increase an average of $2690 million
per day since September 30, 2012. Use this information to estimate the national debt
as of November 30, 2013, 18 days after November 12.
The answer is provided at the end of this section.
a. Calculate the average growth rate of cases from March 26 to March 31. Be
sure to express your answer using proper units.
b. Use your answer from part a to estimate the cumulative number of SARS
cases by March 28. The actual cumulative number on March 28 was 1485.
What does this say about how the growth rate varied over time?
Benny Evans
SOLUTION
a. The independent variable is the date, and the function is the cumulative
number of SARS cases reported. The average growth rate from March 26 to
March 31 is
b. There were 1323 cases on March 26, and we expect to see about 59.8
new cases on March 27 and again on March 28. Thus, we make our estimate
Crauder-4100237 QL November 6, 2014 15:7
as follows:
Estimated cases on March 28 = Cases on March 26 + 2 × Average new cases
per day = 1323 + 2 × 59.8 = 1442.6
or about 1443 cases. Our estimate using interpolation is relatively good but
somewhat lower than the actual value of 1485. The fact that interpolation
gives an underestimate indicates that the growth rate between March 26 and
March 28 was higher than the growth rate between March 26 and March
31—the number of new cases per day was higher over the first part of the
five-day period.
Extrapolation
It is also possible to use average growth rates to make estimates beyond the limits
of available data and thus to forecast trends. This process is called extrapolation. As
with interpolation, extrapolation requires a little judgment and common sense.
Now let’s see what happens if we try to extrapolate farther beyond the limits of
the SARS data by predicting the number of cases on April 30. We start with 3235
Crauder-4100237 QL November 6, 2014 15:7
92 CHAPTER 2
cases on April 15 and use the rate of change of 90.8 new cases per day for 15 days
to obtain
The actual number of SARS cases on April 30, as reported by the World Health
Organization, was 5663. That’s 1066 (or about 23%) more cases than our estimate.
This example illustrates why one should be cautious in using extrapolation to
make estimates well beyond the limits of the data.
94 CHAPTER 2
Date 2010 2011 2012 2013 2014 7--8. Children in kindergarten. The following table shows
Value $100.00 $110.00 $121.00 $145.00 $159.50 the percentage of children in the United States between the ages
of 4 and 6 who are enrolled in full-day kindergarten programs.
a. Make a table that shows percentage increase from each Use this table to answer Exercises 7 and 8.
year to the next.
b. Make a bar graph that shows percentage increase from
Date 2006 2008 2010 2012
each year to the next.
Percentage 71.9 71.9 72.8 76.1
c. Over what period was percentage growth the greatest?
2--4. Units for average growth rate. In Exercises 2 through 4,
state the units of the average growth rate and explain in prac- 7. Use interpolation to estimate the percentage of children en-
tical terms the meaning of the average growth rate. rolled in 2011.
2. The independent variable is the day of the year. The func- 8. Use extrapolation to estimate the percentage of children en-
tion is the cumulative number of accidents since the first of the rolled in 2014.
year at a certain dangerous intersection up to that day. State the 9--11. Customer complaints. The following table shows the
units of the average growth rate and give its practical meaning. number of customer complaints against airlines operating in
3. The independent variable is the day of the year. The function the United States during the given year. Use this table to an-
is the cumulative number of cases of a disease since the first of swer Exercises 9 through 11.
the year. State the units of the average growth rate and give its
practical meaning.
Year 2002 2004 2006 2008 2010
4. The independent variable is the hour of the day. The func- Number of complaints 9466 7452 8325 10,648 16,508
tion is the distance west of New York City of an automobile
since it left the city on the freeway. State the units of the aver-
age growth rate and give its practical meaning. 9. Use interpolation to estimate the number of complaints in
5. Home runs. The following table shows the total number of 2007.
home runs scored by major league baseball players in the given 10. Complete the table below with the absolute numbers and
year: the percentages by which the complaints changed over each
two-year period. Round your final answers to the nearest whole
Year Home runs
number.
2004 5451
2005 5017
2006 5386 Range 2002--2004 2004--2006 2006--2008 2008--2010
2007 4957
2008 4878 Change in
2009 5042 complaints
2010 4613 Percentage
2011 4552 change
2012 4934
2013 4661
11. Use the second row of the table you completed in Exer-
Choose independent and dependent variables for this table and cise 10 to estimate the number of complaints in 2001. The ac-
explain what the corresponding function means. tual number was 16,508. What events in 2001 might cause the
6. More basketball. The following table shows the number of actual number to be higher than your estimate? Explain.
bids to the NCAA Men’s Basketball Tournament for the Big 12 12--15. Grades. The bar graph in Figure 2.7 shows the per-
Conference in the given year: centage of college freshmen in the United States, as of the fall
of the given year, whose average grade in high school was an
Year Number of bids by Big 12 A-, an A, or an A+. Exercises 12 through 15 refer to this bar
2008 6 graph.
2009 6 12. Describe in general how the percentage changed over time.
2010 7
2011 5 13. Find the average yearly growth rate in the percentage from
2012 6 2005 to 2010.
2013 5 14. Use your answer to Exercise 13 to estimate the percentage
2014 7 in the year 2012.
Crauder-4100237 QL November 6, 2014 15:7
Percentage
30 of your dollar. For example, if inflation this year is 5%, then
25 a given item would be expected to cost 5% more at the end of
20 the year than at the beginning of the year. Figure 2.8 shows the
15
10 rate of inflation in the United States for the given years. This
5 information is needed for Exercises 23 through 26.
0
1980 1985 1990 1995 2000 2005 2010
Year
14
13
FIGURE 2.7 Freshmen with high school 12
11
grades of “A”.
Rate of inflation
10
9
8
15. The actual percentage in 2012 was 49.5%. Compare this 7
with your answer to Exercise 14. What does this say about the 6
5
rate of growth in the percentage after 2010, as compared with 4
the rate of growth in the percentage during the last part of the 3
2
preceding decade? 1
0
16--19. College expenditures. The table below shows total 1975 1980 1985 1990 1995 2000 2005 2010
expenditures for colleges and universities in millions of dol- Year
lars. All data are in current dollars. Use this table to answer
Exercises 16 through 19. FIGURE 2.8 Rate of inflation.
Year 1970 1980 1990 2000 2010 23. Use interpolation to estimate the rate of inflation in the
Expenditures 21,043 56,914 134,656 236,784 461,000 United States for each of the years 1996 through 1999. Round
your answer to one decimal point as a percentage.
16. What, in general, does the table tell you about spending on 24. Assume that men’s clothing followed the rate of inflation.
colleges and universities from 1970 through 2010? If a man’s suit cost $300 at the beginning of 1995, what did
it cost at the end of 1995? Round your answer to the nearest
17. Use interpolation to complete the following table:
dollar.
25. It is a fact that in 1949 the rate of inflation in the United
Year 1970 1975 1980 1990 1991 2000 2010 States was negative, and the value was −2%. If a man’s suit
Expenditures 21,043 56,914 134,656 236,784 461,000 cost $100 at the beginning of 1949, what did it cost at the end
of 1949?
18. Make a table that shows the absolute change in expendi- 26. Extreme inflation can have devastating effects. In 1992 in-
tures over each 10-year period. flation was 100% in Lebanon and Bulgaria.4 If a man’s suit
19. Estimate expenditures in 2015. Explain your reasoning. cost $100 at the beginning of a year with that rate of inflation,
what would it cost at the end of the year?
20--22. Attitudes on abortion. The following table shows the
percentage of U.S. college freshmen in the given year who be- Exercises 27 through 31 are suitable for group work.
lieved abortion should be legal. Use this table to complete Ex- 27--31. School costs. Exercises 27 through 31 are based on
ercises 20 through 22. the following table, which shows the average total cost, mea-
sured as dollars expended per pupil, of operating a high school
Year 1996 2000 2004 2008 2012 with the given enrollment. This table is based on a study of
Percentage 56.3 53.9 53.9 58.2 61.1 economies of scale in high school operations from the 1960s
20. Calculate the average growth rate per year from 1996 to Enrollment 170 250 450 650
2000 in the percentage of college freshmen who believed that Average total cost 532 481 427 413
abortion should be legal.
21. Use your calculation from Exercise 20 to estimate the per-
27. Estimate the average cost per student for a high school with
centage of college freshmen in 1997 who believed that abortion
an enrollment of 350 students.
should be legal.
96 CHAPTER 2
28. The actual cost per student with an enrollment of 350 35. Explain in general terms what the data on enrollments in
is $446 per pupil. Use this information to compare how the French show.
average cost changes from an enrollment of 250 to 350 with 36. For each period shown in the table (1995--1998, 1998--
how the average cost changes from an enrollment of 350 2002, etc.), calculate the average growth rate per year in en-
to 450. rollments in French. Note that the time periods are not of the
29. What is the average growth rate of the average cost per ad- same length.
ditional student enrolled, when moving from an enrollment of 37. Make a bar graph displaying the average growth rate per
450 students to an enrollment of 650 students? year in enrollments in French.
30. What would you estimate for the average cost per student 38--39. AP examinations. The following table shows the per-
at an enrollment of 750 students? centage of Advanced Placement examinations in which stu-
31. Some policy analysts use tables such as this to argue that dents achieved the scores of 1 through 5 for the years 2010
small schools should be consolidated to make the operations and 2013. Exercises 38 and 39 refer to this table.
more cost-effective. On the basis of your answers to previous
parts, determine which is likely to save more money (in terms of
average total cost): consolidation that results in increasing the Score 5 4 3 2 1
enrollment by 100 starting at the low end of the scale (say, from 2010 15.1 19.5 23.4 20.5 21.6
250 to 350) or such consolidation starting at a higher level (say, 2013 14.3 19.9 24.7 22.2 18.8
from 350 to 450). Note: These computations, of course, repre-
sent only one side of a complex problem. School consolidation
can be devastating to small rural communities. 38. Make a bar graph showing the data. Suggestion: Use the
different scores (1 through 5) for categories and put a bar for
32--34. Historical events. The table for percentage change in each year.
U.S. population on page 84 contains some anomalies that we
did not discuss in the text. Exercises 32 through 34 refer to this 39. For which year were the scores above 3 higher?
table. 40--41. Buying power. The following table shows how the
32. The percentage increase from 1930 to 1940 is markedly buying power of a dollar is reduced with inflation. The sec-
reduced. What historical event may account for the fact that ond line in the table gives the percent reduction of a dollar’s
this percentage is much smaller than expected? Suggestion: buying power in one year due to the corresponding inflation
Use the Internet or the library to investigate what happened on rate. Exercises 40 and 41 refer to this table.
“Black Tuesday.”
33. The population increased by a smaller percentage from Annual inflation rate 10% 20% 30% 40% 50%
1910 to 1920 than the preceding or following decades. There Buying power reduction 9.1% 16.7% 23.1% 28.6% 33.3%
were two historical events that may have accounted for this.
Annual inflation rate 60% 70% 80% 90% 100%
One was a great “pandemic” of influenza. Look up the his- Buying power reduction 37.5% 41.2% 44.4% 47.4% 50.0%
tory of the great influenza pandemic and write a brief ac-
count of it. What do you think was the other great historical
event that may have influenced the population growth in this 40. Make a bar graph for this table with the inflation rate on
decade? the horizontal axis.
34. Look at the table on page 84. What might account for the 41. If the buying power of a dollar declined by 25% during
large percentage increase from 1950 to 1960? a year, what was the inflation rate during that year? Sugges-
The following exercises are designed to be solved using tech- tion: First note that 25% is between 23.1% and 28.6%, so the
nology such as calculators or computer spreadsheets. answer must be an inflation rate between 30% and 40%. In-
terpolate to find the answer more accurately.
35--37. Enrollment in French. The table below shows course
enrollments (in thousands) in French in U.S. institutions of
higher education for the given year.
W-7. Percentage change. When we studied population growth in the United States
and the world in this section, we calculated the percentage change. If you were given
a table showing how food prices in the United States changed with time, would it
be helpful to calculate the percentage change? Can you think of other situations in
which the percentage change gives important information?
Exercise 20 provides additional information relevant to this topic.
W-8. No change? The population of a town is 500. Then the population decreases
by 100 over the year, and the population grows by 100 over the next year. What is
the average yearly growth rate in the population over the two-year period? Does this
growth rate fully represent what happened over this period? If not, what calculations
would you make to give a better representation?
W-9. Which would you choose? One pair of jeans is advertised at 10% off the regular
price. At another store, the same pair of jeans is advertised at $5.00 off the regular
price. Which is the better buy? Do you need additional information to make your
decision? If so, what additional information?
Crauder-4100237 QL November 6, 2014 15:7
98 CHAPTER 2
0.5
anomaly (°C)
Temperature
–0.5
–1
–1.5
10,000 8000 6000 4000 2000 0
Age (yr BP)
At the center of the public controversy over global warming is the graph shown in
Figure 2.9. In this graph the black line represents the temperature relative to a long-
term average. This black line is produced by a complicated method that involves
averaging certain measurements, which are shown by colored lines.
There is controversy regarding both the accuracy of the graph and its interpre-
tation. But at the heart of the controversy is how the graph changes and what that
change means. From 1900 on (the “blade” portion of the “hockey stick”), the char-
acter of the graph is different from the earlier part—the “hockey stick handle.” It is
this change in character that concerns scientists, politicians, and the general public.
Crauder-4100237 QL November 6, 2014 15:7
Why are graphs so often used by both the scientific community and the popular
media? A graph is a good example of the old saying that a picture is worth a thousand
words. Graphs allow us to visualize information, which helps us see patterns that may
not be readily apparent from a table. For example, a graph reveals visually where the
function is increasing or decreasing and how rapid the change is. In this section, we
study various types of graphs and learn how to interpret these patterns.
In the preceding section, we used bar graphs to present data from tables. In this
section, we consider two other ways to present data: scatterplots and line graphs.
These ways of presenting data are often seen in the media, and it is important to be
familiar with them, to interpret them, to analyze them critically, and to be aware of
their advantages and disadvantages.
7
6
5
Cars sold
4
3
2
1
0
1 2 3 4
Day
Scatterplots
Often we have a collection of isolated data points and want to make a visual display
in order to search for patterns. When the values of the independent variable are evenly
spaced, a bar graph may be used. When the values of the independent variable are not
evenly spaced, a scatterplot is often used to represent the data. Scatterplots are also
Crauder-4100237 QL November 6, 2014 15:7
100 CHAPTER 2
used to represent large numbers of data points that cannot be reasonably displayed
using a bar graph.
90
80
70
60
50
0
1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Eruption duration (minutes)
100 100
Running speed
Running speed
60 60
40 40
20 20
0 0
10 20 30 40 50 10 20 30 40 50
Length (inches) Length (inches)
FIGURE 2.11 Scatterplot of running speed FIGURE 2.12 Line graph produced by
versus length. adding segments.
An interesting example comes from the following table, which shows running
speed of various animals as a function of length:5
Length Speed
Animal (inches) (feet per second)
Deermouse 3.5 8.2
Chipmunk 6.3 15.7
Desert crested lizard 9.4 24.0
Grey squirrel 9.8 24.9
Red fox 24.0 65.6
Cheetah 47.0 95.1
We make the line graph in two steps. First we make the scatterplot shown in
Figure 2.11. Then we join adjacent points with line segments to get the line graph
shown in Figure 2.12.
One advantage of a line graph over a scatterplot is that it allows us to estimate
the values between the points in the scatterplot. Suppose, for example, we want to
know how fast a 15-inch-long animal can run. There is no data point in the table
corresponding to 15 inches, but we can use the graph to estimate it. In Figure 2.13,
100
(feet per second)
80
Running speed
60
40
20
0
10 15 20 30 40 50
Length (inches)
5 This table is adapted from J. T. Bonner, Size and Cycle (Princeton, NJ: Princeton University Press, 1965). We will
encounter it again in Chapter 3 when we study trend lines.
Crauder-4100237 QL November 6, 2014 15:7
102 CHAPTER 2
we find the 15-inch point on the horizontal axis (halfway between 10 and 20) and
then go up to the line graph. We cross the graph at a level corresponding to about
40 feet per second on the vertical axis. So, these segments allow us to fill gaps in data
by estimation.
An interesting fact about filling in gaps using line segments is that the value of
the function (running speed in this case) determined in this way is exactly the same
as that obtained by interpolation, which we discussed in Section 2.1. Line graphs
provide a graphical representation of the average growth rate from one point to the
next.
One must take care not to use a line graph when there is nothing between cat-
egories. For example, Figure 2.14 shows two line graphs. These graphs show the
number of people playing various card games at the Yahoo! Web site on a certain
Wednesday and a certain Sunday.6 The line segment joining blackjack to bridge
would indicate that there are card games in between them. A bar graph would be
appropriate here.
David Lane is licensed under a Creative Commons Attribution License 1.0 license.
da6b98f93045@13. © Jun 27, 2003 David Lane. Textbook content produced by
Wednesday
Download for free at http://cnx.org/contents/9531d109-432b-4922-9787-
Sunday
8000
7000
6000
5000
4000
3000
2000
1000
0
Poker
Blackjack
Bridge
Gin
Cribbage
Hearts
Canasta
Pinochle
Euchre
Spades
FIGURE 2.14 When a line graph is not appropriate.
and temperature
The running speed of ants varies with the ambient temperature. Here are data col-
lected at various temperatures:
Temperature Speed
(degrees Celsius) (centimeters per second)
25.6 2.62
27.5 3.03
30.4 3.56
33.0 4.17
First make a scatterplot of the data showing the speed as the function and the
temperature as the independent variable, then make a line graph using these data.
6 This figure is taken from http://cnx.org/content/m10927/latest/, where it is used to illustrate the misuse of line graphs.
Crauder-4100237 QL November 6, 2014 15:7
4.5 4.5
4.0 4.0
Running speed
3.5 3.5
3.0 3.0
2.5 2.5
0 0
20 25 30 35 20 25 30 35
Temperature (degrees Celsius) Temperature (degrees Celsius)
FIGURE 2.15 Running speed of ants FIGURE 2.16 Running speed of ants
versus temperature: scatterplot. versus temperature: line graph.
SOLUTION
To make a scatterplot, we plot the data points, each of which corresponds to a point
on the graph: (25.6, 2.62), . . . , (33.0, 4.17). The scatterplot is the graph of these
points, as shown in Figure 2.15. We join the points with line segments to get the
line graph in Figure 2.16.
104 CHAPTER 2
400
350
200
Black females
150
White females
100
50
0
1950–54
1955–59
1960–64
1965–69
1970–74
1975–79
1980–84
1985–89
1990–94
1995–99
2000–04
2005–09
Years
Figure 2.17 shows line graphs of these data. This figure reveals several trends
that are not readily apparent from the table. For one, it is immediately clear that
cancer mortality rates among black males are much higher than for any of
the other categories presented. These rates were on the rise from 1970–74 to
1990–94 and then fell sharply through 2005–09, ending at a level below that of
1970–74.
The graph shows that cancer mortality rates among white women reached a min-
imum of about 160 per 100,000 at the point representing the period 1970–74. Those
rates began increasing slightly after that time and then decreased to the lowest level
in 2005–09. Also, we see that cancer mortality rates for white males reached their
peak at the point representing the period 1985–89. It would take considerable effort
to see these trends using only the table.
It is important to note not only how high a graph is but also how steep it is:
A steeper graph indicates a growth rate of greater magnitude. For example, we
observe that the graph of mortality rates for black males begins to level off from
1980–84 to 1990–94. That is, the rate of increase appears to be slowing. Rates for
black women appear to have increased at a fairly steady rate from 1975–79 to 1990–
94, and then decreased at a steady rate from 1990–94 to 2005–09. The line graph
makes apparent these changes in growth rates, far more so than the numerical data
themselves do.
The sign of the growth rate for a graph can be seen by examining where the graph
increases or decreases. A positive growth rate is indicated when the graph increases
as we move from left to right along the horizontal axis, and a negative growth rate is
indicated by a decreasing graph. For example, the mortality rate for white women is
decreasing from 1950–54 to 1970–74 and also from 1990–94 to 2005–09, indicating
Crauder-4100237 QL November 6, 2014 15:7
negative growth rates during those periods. The increasing graph from 1970–74 to
1990–94 indicates a positive growth rate.
$90
Income (thousands)
85
80
75
70
65
60
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Year
SOLUTION
Income increased between the years 2002 and 2004. After that it remained constant
until 2006, when it began to decline. Income reached a minimum in 2009 and in-
creased after that.
The growth rate of income was positive from 2002 to 2004 and from 2009 to
2011. The fact that the graph is steeper from 2009 to 2011 than from 2002 to 2004
means that income was growing at a faster rate from 2009 to 2011.
Because income declined from 2006 to 2009, the growth rate was negative over
that period. From 2004 to 2006, income did not change, meaning there was no
growth. So the growth rate over that period was zero.
106 CHAPTER 2
$120
100
Sales (thousands)
80
60
40
20
0
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Year
A smoothed line graph may be appropriate when the growth rate is continuous.
For example, Figure 2.20 shows the weight of a baby over the first 12 months of life.
Weight changes continuously, not in jerks. Hence, a smooth graph is appropriate.
Smoothed line graphs that are accurate contain more information about growth
rates than do straight line graphs. For example, Figure 2.21 shows the growth of
money in a savings account, and Figure 2.22 shows a man’s height as a function of
his age. Both graphs are increasing and so display a positive growth rate. But the
shapes of the two curves are different. The graph in Figure 2.21 gets steeper as we
move to the right, indicating an increasing growth rate. That is, the value of the
account is increasing at an increasing rate. This is typical of how savings accounts
grow, as we shall see in Chapter 4.
In contrast, the graph in Figure 2.22 gets less steep as we move to the right,
indicating a decreasing growth rate. That is, the graph is increasing at a decreasing
rate. Practically speaking, this means that our growth in height slows as we age.
$500 7
25
6
400
Weight (pounds)
20
5
Height (feet)
Dollars
15 300 4
200 3
10
2
5 100
1
0 0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Months Years Years
FIGURE 2.20 Smoothed graph FIGURE 2.21 Money in a savings FIGURE 2.22 Height versus age.
showing a child’s weight. account.
Crauder-4100237 QL November 6, 2014 15:7
SOLUTION
The graph should be decreasing, but it should become less steep as we move to the
right. One possibility is shown in Figure 2.23.
Time
108 CHAPTER 2
Positive Negative
growth rate growth rate
300
250
Population
200
150
100
50
0
0 1 2 3 4 5
Years
200
150
Population
100
50
0
0 1 2 3 4 5 6 7 8 9 10
Year
a. Give a brief general description of how the deer population changes over the
10-year period shown. Your description should include the growth rate of the
population and how that rate changes. Explain what this means for the
population in the long run.
b. Estimate the time at which the population is growing at the fastest rate.
SOLUTION
a. The population grew throughout the 10-year observation. Over about the
first 4 years of the observation, the graph is increasing and getting steeper, so the
population shows increasingly rapid growth over this period. This means that
not only is the population increasing over this period, but also the rate of
population growth is increasing. From about 4 years on, we see that the graph
continues to increase but becomes less steep. Thus, the population continues to
grow, but at a decreasing rate. The graph levels off as we approach Year 10.
7 One interesting source is Dale R. McCullough, The George Reserve Deer Herd (Ann Arbor: University of Michigan
Press, 1979).
Crauder-4100237 QL November 6, 2014 15:7
This flattening of the graph indicates that the growth rate is declining to near 0,
and the population is stabilizing at about 175 deer.
b. The population is growing at the fastest rate where the graph is the steepest.
We estimate that this occurs at about Year 4.
250
250
200 200
Height (cm)
Height (cm)
150 150
100 100
50 50
0 0
0 20 40 60 80 0 20 40 60 80
Age (days) Age (days)
Try It Yourself 2.12: Interpreting line graphs: Sales growth The rate of growth
in sales is negative from 2002 to 2006 and from 2008 to 2011. The growth rate is
more negative over the second period. From 2006 to 2008, the growth rate is 0.
Try It Yourself 2.13: Making graphs with varying growth rates: Gas mileage The
graph should be decreasing and become steeper as we move to the right. The accom-
panying figure gives one example.
Mileage
Time
Crauder-4100237 QL November 6, 2014 15:7
110 CHAPTER 2
18
Represent these data with a bar graph, a scatterplot, and a line
Expense (thousands)
16
14
graph. Which type of graph do you think best represents the
12
data?
10 6. Graduation rates. According to the NCAA,8 86.1% of fe-
8 male athletes at Division I schools who entered college in 1998
6 graduated within 6 years. This increased to 87.9% in 2002
4 and by 2006 it had reached 88.4%. Of their male counter-
2
parts, only 69.8% of students who entered in 1998 graduated
0
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 within 6 years. This number increased to 72.1% in 2002 and
Year by 2006 it had reached 75.7%. On one set of axes, sketch two
line graphs displaying this information, one for females and
another for males.
FIGURE 2.26 Figure for Exercise 1.
7. An investment. You invest $1000 in a risky fund. Over the
first six months, your investment loses half its value. Over the
2. Varying growth rates. A savings account increases in value
next six months, the value increases until it reaches $1000.
at an ever increasing rate over the period from 2005 through
Make a smoothed line graph that shows your investment over
2015. Sketch a graph of the value of the account versus time
the 12-month period.
that shows growth of this type.
8. Paper airplanes. The following table shows the gliding dis-
3. An endangered population. The graph in Figure 2.27 shows
tance in meters of a certain type of paper airplane and its
the population of a certain endangered species. Give a general
wingspan in centimeters:
description of how the population changes over the 10-year pe-
riod shown. Your description should include the growth rate of
the population and how the rate changes. Wingspan (centimeters) 5 10 15 20
Glide distance (meters) 5 8 9 13
16,000
14,000
Make a scatterplot of these data with wingspan on the hori-
zontal axis.
12,000
Population
8 www.ncaa.org
Crauder-4100237 QL November 6, 2014 15:7
is initially decreasing at an increasing rate, but after a bit the 20. Population growth. The graphs in Figures 2.29 and 2.30
graph decreases at a decreasing rate. show the population (in thousands) of a certain animal in two
13. Making a graph. Sketch a graph that is increasing at a de- different reserves, reserve A and reserve B. In each case, the
creasing rate from 0 to 3, decreasing at an increasing rate from horizontal axis represents the number of years since observa-
3 to 6, and decreasing at a decreasing rate from 6 on. tion began.
Population
Population
200
increased each day until they hit a high of 800 copies on June 1. 150
150
From there sales decreased each day to 400 copies on December 100
100
1. Sales declined most rapidly at some point in August. Sketch a 50 50
graph reflecting this information. Suggestion: Locate the points
0 0
you know (January 1, June 1, and December 1) first. 0 2 4 6 8 10 0 2 4 6 8 10
Year Year
15. Bears. A population of bears is introduced into a game pre-
serve. Over the first five years, the bear population shows a
FIGURE 2.29 Population in FIGURE 2.30 Population in
negative growth rate. The growth rate is positive over the next reserve A. reserve B.
five years. Sketch a possible graph of the bear population as a
function of time. In Figures 2.31 and 2.32 are two graphs of population
16. Wolves. A population of wolves is introduced into a game growth rates. Let’s call them growth rate I and growth rate
preserve. Over the first five years, the wolf population shows a II. One of the graphs of growth rate describes the population
positive growth rate. The growth rate is negative over the next in reserve A, and the other describes the population in reserve
five years. Sketch a possible graph of the wolf population as a B. Which is which? For each reserve, give a careful explanation
function of time. of how the graph of population is related to the appropriate
17–19. Language enrollments. The line graph in Figure 2.28 graph of growth rate.
shows total course enrollments in languages other than English
in U.S. institutions of higher education from 1960 to 2009. (En-
Population growth rate
1,800,000
1,629,326
1,600,000 1,522,770
0 2 4 6 8 10 0 2 4 6 8 10
1,400,000 1,347,036 Year Year
1,138,880 1,151,283
1,200,000
1,073,097 1,067,217 FIGURE 2.31 Population FIGURE 2.32 Population
1,000,000 975.777 963,930
883,222
922,439 1,096,603 growth rate I. growth rate II.
960,588
800,000 897,077 877,691
21. Magazine sales. An executive for a company that publishes
600,000
608,749 a magazine is presented with two graphs, one showing sales (in
400,000 thousands of dollars), and the other showing the rate of growth
200,000
in sales (in thousands of dollars per month). Someone forgot to
label the vertical axis for each graph, and the executive doesn’t
0
know which is which. Let’s call the graphs mystery curve 1 and
1960 1965 1968 1972 1977 1980 1983 1986 1990 1995 1998 2002 2006 2009
1970 1974 mystery curve 2. They are shown in Figures 2.33 and 2.34. (In
FIGURE 2.28 Enrollments in languages other than English in each case, the horizontal axis represents the number of months
U.S. institutions of higher education (2009). since the start of 2010.)
112 CHAPTER 2
Help the executive out by identifying which curve shows vehicle, as measured by speedometer readings. In each case, the
sales and which shows the rate of growth in sales. Give a care- horizontal axis represents the time in hours since measurements
ful explanation of how the graph of sales is related to the graph began.
of growth in sales. In this situation, the speedometer reading measures the
22. Location and speed. The graph in Figure 2.35 shows the growth rate of distance from your starting point. In Exam-
location of a vehicle moving east on a straight road. The lo- ple 2.14, we saw how the graph of population growth rate is
cation is measured as the distance east of a fixed observation related to the graph of population. Explain how the graph of
point. The graph in Figure 2.36 shows the speed of the same speed in Figure 2.36 is related to the graph of location in Fig-
ure 2.35, and write a travel story that matches these graphs.
23. Historical figures: Descartes and Fermat. Graphs and
500 70
charts have been used throughout recorded history. Their in-
Distance (miles)
60
400
corporation into mathematics should be credited to any num-
Speed (mph)
50
300 40 ber of early mathematicians. Certainly, two on any such list are
200 30 René Descartes and Pierre de Fermat. Write a brief report on
20
100 their development of graphs in mathematics.
10
0 0 24. Historical figures: Oresme. Descartes and Fermat are
0 4 8 12 16 20 0 4 8 12 16 20
Time (hours) Time (hours)
famous early mathematicians, but lesser known is Nicole
Oresme, who preceded both Fermat and Descartes by 300
FIGURE 2.35 Location of FIGURE 2.36 Speed of years. He employed graphs in mathematics long before either.
vehicle. vehicle. Write a brief report on his use of graphs in mathematics.
W-4. A trip. Refer to Figure 2.38. Explain what happened at the five-hour mark.
Where did the trip end?
W-5. Scatterplots and line graphs. A fellow student has found a scatterplot of data
in a newspaper. He wants to create a line graph by connecting the dots. Describe
ways in which the resulting line graph could be more useful than the original
scatterplot.
W-6. Scatterplots and line graphs again. A fellow student has found a scatterplot of
data in a newspaper. He wants to create a line graph by connecting the dots. What
cautions would you give about doing that?
Population growth rate
10 25
Miles from home
8 20
6 15
4 10
2 5
0 0
0 1 2 3 4 0 2 4 6 8 10
Months Hours
W-7. Extrapolation. In this section we noted that filling in gaps in a scatterplot us-
ing line segments gives estimates for function values that agree with those obtained
by interpolation. What graphical procedure corresponds to extrapolation? Draw a
graph to illustrate your answer.
W-8. Death rate. You and a friend are discussing the good news that the annual
number of deaths due to a certain disease has been decreasing over the last decade.
Your friend says that it is even more encouraging that the rate of decrease of this
annual number is decreasing. How would you reply?
W-9. Endangered species. A line graph shows the population of an endangered
species over the past few years. Explain what features you would look for in the
graph that might offer encouraging news about the species.
114 CHAPTER 2
This article refers to a study of advertisements for drugs on the news and warns
consumers to be wary when watching such advertisements. In an earlier study, a
significant percentage of graphs associated with drug advertisements in medical jour-
nals were found to distort facts. Graphs can be helpful visual tools, but sometimes
they can be misleading (intentionally or otherwise).
There are many ways that graphs can mislead us into drawing incorrect, inac-
curate, or inappropriate conclusions about data.9 We illustrate some of the most
common types of misleading graphs in this section.
800.0 750.0
700.0 700.0
Billions of dollars
Billions of dollars
600.0 650.0
500.0
600.0
400.0
550.0
300.0
200.0 500.0
100.0 450.0
0 400.0
2004 2005 2006 2007 2008 2009 2010 2011 2012 2004 2005 2006 2007 2008 2009 2010 2011 2012
Year Year
The key to understanding the dramatic difference between these two graphs is
the range on the vertical axis. In Figure 2.39 the vertical scale goes from 0 to 800
billion dollars, and in Figure 2.40 that scale goes from 400 to 750 billion dollars.
The shorter vertical range exaggerates the changes from year to year.
One way to assess the actual change in defense spending is to find the percentage
change. Let’s consider the change from 2004 to 2006. In both graphs, we see that
cartoonstock.com
9 A classic reference here is Darrell Huff’s How to Lie with Statistics (New York: W.W. Norton, 1954).
Crauder-4100237 QL November 6, 2014 15:7
the amount spent in 2004 was about 456 billion dollars, and the amount spent in
2006 was about 522 billion dollars. First we find the absolute change:
Change = 522 − 456 = 66 billion dollars
Now, to find the percentage change, we divide by the spending in 2004:
66
Percentage change = × 100%
456
This is about 14.5%. The percentage increase is fairly significant, but it is not nearly
as large as that suggested by Figure 2.40, which gives the impression that spending
has almost doubled—that is, increased by 100%.
It is striking to see how easily someone with a particular point of view can make
visual displays to suit his or her own position. If we wanted others to believe that
defense spending remained nearly stable over this time period, which graph would
we use? On the other hand, which graph would we use if we wanted others to believe
that defense spending showed a dramatic increase over this time period?
39.6% 35
40
Top tax rate (%)
30
25
38
20
15
36 10
35%
5
34 0
Now January 1, 2013 Now January 1, 2013
10 This article and the graphics shown here appear at the Web site
http://mediamatters.org/research/2012/10/01/a-history-of-dishonest-fox-charts/190225
11 We will discuss marginal tax rates and tax brackets in Section 4.5.
Crauder-4100237 QL November 6, 2014 15:7
116 CHAPTER 2
SOLUTION
a. The graph shows that the top tax bracket at the time labeled “Now” in the
graphs is 35% and the top tax bracket if the Bush tax cuts had been allowed to
expire is 39.6%.
b. To find the answer, we divide the difference of 39.6 − 35 = 4.6 percentage
points by the bracket “Now:”
Difference in percentages
Percentage change = × 100%
Percentage “Now”
4.6
= × 100%
35
or about 13%. We conclude that the bracket if the rates expired is about 13%
larger than the bracket “Now.”
c. In the original graph, the bar representing expired tax cuts is about 6 units
high, and the bar representing “Now” is 1 unit high. This gives the impression
that the rate for the higher tax bracket is 6 times as large as that “Now,” which
is 500% more instead of 13% more.
O klahoma students scored higher on the ACT in the spring than students
who took it last year, according to a report released today, but Oklahoma
students scored lower than the national average in all areas.
Though an estimated 72 percent of 2006 Oklahoma graduates took the test,
the average student couldn’t be accepted to either of the state’s two research
universities based on ACT scores alone.
According to data released today by ACT Inc., the national average score was
21.1 this year, compared with Oklahoma’s average score of 20.5.
Last year’s national average score was 20.9; the average Oklahoma score was
20.4 last year.
Crauder-4100237 QL November 6, 2014 15:7
22.00
21.75 State
National
21.50 21.4
21.25
21.1 21.1
21.00 20.9
20.8
20.75
20.6
20.5
20.50 20.4
20.3
20.25
20.00
19.75 19.7
19.50
Composite English Math Reading Science
The article is accurate. It states, among other things, that the national average
was 21.1 and the Oklahoma average was 20.5. Now consider Figure 2.43, which
shows a graph of the ACT scores for these Oklahoma students. This bar graph is
based on data that compare the subject area test scores in Oklahoma to those in the
nation as a whole. It certainly gives the impression that Oklahoma scores are far
below national scores—especially in math.
Is this impression accurate? Are the Oklahoma scores really as far below the na-
tional scores as the bar graph makes it appear? We’ll explore that question further in
Exercises 10 through 12, where the scale on the vertical axis is examined more closely.
Even though the graph may be a misleading representation of the data, the re-
porter might be trying to give a fair and accurate presentation. It is worth noting
that the graph shown in Figure 2.43 is exactly the graph obtained by accepting the
default ranges offered by one of the most popular graphing packages.
We emphasize that, although a graph may or may not be intended to deceive, the
effect is the same. It is up to us to think critically so as not to allow ourselves to be
misled.
Your debate team must be prepared to present a case for the following proposition
and against it.
Crauder-4100237 QL November 6, 2014 15:7
118 CHAPTER 2
SOLUTION
To support the proposition, you want a graph that emphasizes the increase in gas
prices. A narrow range on the vertical scale emphasizes the differences in data. In
Figure 2.44, we have made a graph to support the proposition using a vertical range
from $1.75 to $3.50. To argue against the proposition, we want a graph that deem-
phasizes the differences in data. A wider vertical range will accomplish this. In Fig-
ure 2.45, we have used a vertical range from $0 to $10.00. This graph would be used
in an argument against the proposition.
$3.50 $10.00
Average price per gallon
2.75 6.00
2.50 4.00
2.25
2.00
2.00
1.75 0
2007 2008 2009 2010 2011 2012 2013 2007 2008 2009 2010 2011 2012 2013
Year Year
Year 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
ACT scores 20.6 20.6 20.6 20.7 20.8 20.8 20.9 21.0 21.0 21.0 21.0
Make a bar graph that you would use to argue that average ACT scores increased
significantly during this period. Make a second bar graph that you would use to argue
that average ACT scores did not increase significantly over this period.
The answer is provided at the end of this section.
$120 $120
100 100
Currency (billions,
80 80
60 60
40 40
20 20
0 0
1950 1955 1960 1965 1970 1975 1980 1950 1955 1960 1965 1970 1975 1980
Year Year
currency (in billions of dollars) in circulation. The plot appeared with the caption,
“Over a quarter-century the currency pile-up steepens.” The article uses this graph
to bolster the case that “the amount of currency in individual hands is soaring.”
On the surface, the graph makes a strong case for the author, and his conclusion is
technically accurate—that is, the amount of currency in circulation was increasing.
But this conclusion ignores the value of the currency.
The price of many commodities increases over time. For instance, the median
price of an existing one-family house in 1968 was $20,100, the median price of a
house in 1982 was $67,800, and the median price of a house in 2014 was $188,900.12
This increase in home values is for the most part a reflection of increases in prices
overall. The overall increase in prices is measured by the inflation rate.13
Another way of looking at this question is in terms of the value of a dollar. For
example, from 1950 to 2000 prices, on average, increased by 596%. This means
that a dollar would buy the same amount of goods in 1950 as the dollar plus $5.96,
or $6.96, would buy in 2000. So a “1950 dollar” has the same buying power as
6.96 “2000 dollars.” When we adjust a graph for inflation, we report all currency
amounts in constant dollars—for example, we could state the buying power in terms
of the year 2000.
In Figure 2.47, we have reproduced the scatterplot after adjusting for inflation.
Using constant dollars means we are using the same measure of value in 1950 and
1980. Before adjusting for inflation, the vertical axis represents the amount of cur-
rency. After the adjustment, the vertical axis represents the buying power of the cur-
rency, which is a better measure of wealth. We see that, in terms of buying power,
the amount of currency increased much less dramatically than the original graph
suggested. It could be argued that the graph in Figure 2.46 presents a somewhat dis-
torted view of the amount of currency because of the changing value of the dollar
over time.
Another example is shown in Figure 2.48. This figure shows the federal minimum
wage by year. The blue (lower) graph shows the minimum wage (referred to as nom-
inal), which increased from 1938 through 2012. The red graph shows the minimum
wage in constant 2012 dollars. It is the blue graph adjusted to 2012 dollars. Because
the red graph shows the minimum wage in terms of a fixed value of the dollar, it
provides a better comparison of minimum wages earned from year to year than does
the blue graph. It is interesting that the value of the minimum wage did not always
increase (as one might be led to believe by the blue graph) but actually peaked in
1968.
12 Prices don’t always go up. Housing prices fell quite sharply from 2007 to 2009.
13 Inflation is calculated using the consumer price index. We will look more closely at inflation in Chapter 4.
Crauder-4100237 QL November 6, 2014 15:7
120 CHAPTER 2
$11
Real
Nominal
10
0
1938
1943
1948
1953
1958
1963
1968
1973
1978
1983
1988
1993
1998
2003
2008
2013
Year
Let’s see how we find the graph showing constant 2012 dollars. For example, the
nominal minimum wage in 1968 was $1.60. According to the U.S. Bureau of Labor,
the inflation rate from 1968 to 2012 was 546%, meaning that prices increased by
546%. Hence, one “1968 dollar” has the same buying power as the dollar plus 5.46,
or 6.46 “2012 dollars.” Therefore, we convert the value of $1.60 for the minimum
wage in 1968 to 2012 dollars as follows:
1.60 “1968 dollars” = 1.60 × 6.46 = 10.34 “2012 dollars.”
We note that this result agrees with the location of the peak of the red graph in
Figure 2.48.
$2.80
a. Use the graph in Figure 2.49 to determine, without adjusting for inflation, in
what year gasoline was most expensive.
b. Complete the following table showing gasoline prices in constant 2010
dollars:
So the price of gas in 1970 was $1.98 in “2010 dollars.” This gives the first
entry in the table. We find the remaining entries in a similar fashion. The table
below shows the result.
122 CHAPTER 2
$3.00
2.00
Price
1.50
1.00
0.50
0
1970 1980 1990 2000 2010
Year
d. The price of gasoline, adjusted for inflation, was highest in 1980. Comparing
this answer with the answer to part a shows the value of adjusting for inflation.
8000 10,000
7000 9000
Passenger cars
8000
Passenger cars
6000
(thousands)
(thousands)
7000
5000 6000
4000 5000
3000 4000
3000
2000
2000
1000 1000
0 0
1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010
Year Year
8000 10,000
7000 9000
8000
Passenger cars
Passenger cars
6000
(thousands)
(thousands)
7000
5000 6000
4000 5000
3000 4000
3000
2000
2000
1000 1000
0 0
1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010
Year Year
FIGURE 2.52 Smoothed line graph over 10-year periods. FIGURE 2.53 Smoothed line graph over 5-year periods.
A smoothed line graph using data points once per decade is shown in Figure 2.52,
and a graph using data from five-year intervals is shown in Figure 2.53. Unlike the
preceding bar graphs, there is no way to tell how many data points were used to
make the new graphs, and hence there is no way to determine which gives a more
accurate picture of automobile production.
Jens Meyer/AP Photo
If we see a smoothed line graph in the media, we may not know how many data
points were used in producing it. If there’s a reason to question the accuracy, the
wisest course of action is to look for the original data source.
Pictorial representations
Often data are presented using images designed to grab our attention. These images
can be helpful if they are properly constructed. But they can be misleading if they are
not properly presented. One common graphical device is the pie chart. Pie charts are
used to display how constituent parts make up a whole. They are circular regions
divided into slices like pieces of a pie. The pie represents the whole, and the slices
represent individual parts of the whole. For example, if one item is 30% of the whole,
then its corresponding slice should make up about 30% of the area of the pie.
124 CHAPTER 2
SOLUTION
First we calculate the percentage of the whole represented by each category of stu-
dent. For example, the 6274 students in Agricultural and Environmental Sciences
represent
6274
× 100%
26,533
or about 23.6% of the whole undergraduate population of 26,533. The resulting
percentages are shown in the table below.
Our pie must be divided in these proportions. The result is shown in Figure 2.54.
Engineering
Pie charts such as the one in Figure 2.54 can be helpful in understanding data. As
a contrast, consider the pie chart on the next page. It presumably tells us something
about repeat offenders. Recall that the sections of a pie chart represent percentages
of a whole. It is unclear what the sections in this chart are intended to represent.
Other types of graphical descriptions can also be misleading. For example, the
currency representation on the left is supposed to show the value of the dollar un-
der various presidents. But notice that the “Carter dollar” is labeled as being worth
44 cents, compared to $1.00 for the “Eisenhower dollar.” Hence, the “Carter dol-
lar” should be almost half the size of the “Eisenhower dollar.” In fact, it is less than
an eighth of the size in area of the “Eisenhower dollar.” The picture itself gives
a false impression of the relative values of currency during these two presidential
A visual currency administrations.
representation claiming to There is a similar difficulty in a fuel economy chart. Note that the length of the
show the value of a dollar. line segment corresponding to 18 miles per gallon is less than a quarter of the length
Crauder-4100237 QL November 6, 2014 15:7
Public-order Violent
Offenses Offenses
59% 62%
Property
Offenses
54%
of the line segment for 27.5 miles per gallon. The figure gives an inflated view of the
progress in automobile efficiency.
126 CHAPTER 2
There are many other ways of presenting graphs that distort our perception. You
will be asked to comment on some of these in the exercises.
20.9 18
16
20.8 14
12
10
20.7 8
6
20.6 4
2
20.5 0
90
91
92
93
94
95
96
97
98
99
00
90
91
92
93
94
95
96
97
98
99
00
19
19
19
19
19
19
19
19
19
19
20
19
19
19
19
19
19
19
19
19
19
20
Year Year
14 www.ncaa.org
Crauder-4100237 QL November 6, 2014 15:7
Number of people
Carpenter and R. V. Acosta. It shows the percentage of female 32 20
28
head coaches for women’s teams (all divisions, all sports) in
How many
24 15
the given year. Make a line graph of the data that deempha- 20
16 10
sizes the decline in the percentage of female head coaches from 12
1980 through 2010. 8 5
4
0 0
Year 1986 1992 1998 2004 2010
So ll
Ba er
sk ll
Te ll
is
7
am 35
,5
ba
Ba eba
ba
nn
,2
cc
s,
s,
rs
Percent 50.6 48.3 47.4 44.1 42.6
ot
et
ts,
h,
rd
or
s
ste
s
og
Fis
Fo
Ca
Bi
at
D
lig
Al
H
5. Informative graph? Figure 2.55 shows the distribution of ed- Pet types
ucation level among the staff positions at a certain university.
Discuss the usefulness of this graph. Suggestion: Look at the FIGURE 2.57 Pets. FIGURE 2.58 Sports.
first bar.
ol
Co e
so e
Ba tes
M s
oc rs
te
g
ow
elo
e
ho
ra
lle
cia
t
ific
as
to
kn
Sc
ch
rt
igh
Ce
As
e
D
m
H
So
20
6. Donations. An article in the Oklahoma State University
newspaper The Daily O’Collegian looked at the seven largest 15
ov
d
to
to
to
to
to
65 0 to
d
d
Un
60,000,000
50,000,000 10--12. Oklahoma ACT scores. Exercises 10 through 12 re-
40,000,000 fer to the In the News feature 2.5 on page 116 and the graph
30,000,000
in Figure 2.43, which accompanied the original article.
20,000,000
10,000,000 10. Calculate, for the composite and each of the subject areas,
0 how many percentage points lower than the national average
0 1 2 3 4 5 6 7 8 the Oklahoma scores are.
11. Sketch a bar graph of the data in Figure 2.43 using a scale
FIGURE 2.56 Donations to OSU.
of 0 to 22 on the vertical axis.
7. Pets. The graph in Figure 2.57 shows the number of pets 12. Explain how someone might prefer the original bar graph
of various types on my street. What information do the lines or the bar graph you made in Exercise 11, depending on the
connecting the dots convey? Can you suggest a better way to impression to be conveyed. Which graph do you think more
present these data? accurately reflects the numbers you found in Exercise 10?
Crauder-4100237 QL November 6, 2014 15:7
128 CHAPTER 2
trademarks of Yahoo.
YAHOO! and the YAHOO! logo are registered
Nov 15, 2013 2:54 PM - 2:59 PM EST: ^DJI 15938.00
$5.00
$2.24
$.00 16,000
1945 1990 1999 2002 2005 2006 2007 2008 2009
15,950
15,850
13. Wedding costs. The graph in Figure 2.60 was taken from 15,800
the Web site of The Wedding Report. It shows average wedding
costs in thousands of dollars for the given year. Thu Nov 14, 2013 Fri Nov 15 Mon Nov 18 Tue Nov 19 Wed Nov 20
a. Over which of the time periods indicated by the hori-
zontal axis do wedding costs appear to be increasing most FIGURE 2.62 Dow Jones Industrial Average.
rapidly?
b. What is the average growth rate per year in wedding
costs from 1945 to 1990? (First express your answer in 17--19. Tulsa State Fair. The bar graph in Figure 2.63
thousands of dollars per year rounded to two decimal appeared in a report of the Tulsa County Public Facilities
places, and then express it in dollars per year.) Authority. The graph shows Tulsa State Fair attendance in mil-
lions from 1998 through 2002. Exercises 17 through 19 refer
c. What is the average growth rate per year in wedding to this graph.
costs from 2005 to 2006?
d. In view of your answers to parts b and c, do you think
1.12
the graph in Figure 2.60 is misleading? 1.104
Attendance (millions)
1.10 1.091
14--15. Web conferencing. The graph in Figure 2.61 shows
the worldwide revenue in millions of dollars for the Web con- 1.08
ferencing market from 2001 through 2007. It appeared in the 1.06 1.048 1.043
December 2003/January 2004 edition of Technology Review. 1.04
Exercises 14 and 15 refer to this graph.
1.02 11.009
009
1
1998 1999 2000 2001 2002
1400 Year
Revenue (millions)
1200
1000 FIGURE 2.63 Tulsa State Fair attendance.
800
600 17. In a few sentences, explain what you think the graph in
400 Figure 2.63 appears to show regarding State Fair attendance.
200
18. Make a table of percentage change in attendance for each
0
2001 2002 2003 2004 2005 2006 2007
year.
19. Sketch a bar graph using 0 to 1.2 million on the vertical
FIGURE 2.61 Revenue for Web conferencing. axis. Does this give a more accurate display of the data? Why
or why not?
20. Minorities employed. The bar graph in Figure 2.64 shows
14. In light of the date of publication of the article, comment how the percent of federal employees who are minorities has
on your estimation of the reliability of the graph in Figure 2.61. changed with time.15 By adjusting the vertical range, make a
15. Can you think of a reason why the person who produced bar graph that might accompany an article with the title “Fed-
the graph in Figure 2.61 made it a straight line? eral employment of minorities shows little change.”
15 Data, but not graph, from Statistical Abstract of the United States. Only federal civilian nonpostal employees included.
Crauder-4100237 QL November 6, 2014 15:7
Percent minority
32.0 in constant 2011 dollars:
31.5
31.0 Year 2005 2007 2009 2011
30.5
30.0
Salary $67,800 $82,200 $83,600 $93,000
29.5
2011 dollars
29.0
2000 2001 2002 2003 2004 2005 2006 2007 c. Make a bar graph of salaries in constant 2011 dollars.
Year
d. Compare the graphs from parts a and c.
FIGURE 2.64 Minorities employed at 27. The following table shows the average price per ounce of
federal level. gold in the given year:
21--24. Aging employees. The following table shows the av- Year 1970 1980 1990 2000 2010
erage age in years of federal employees at various times. It is Price per ounce $35.94 $612.56 $383.51 $279.11 $1224.53
used in Exercises 21 through 24.
a. Make a bar graph of gold prices.
Year 2000 2001 2002 2003 2004 2005 2006 2007 b. Complete the following table showing gold prices in
Average age 46.3 46.5 46.5 46.7 46.8 46.9 46.9 47.0 constant 2010 dollars:
21. Make a bar graph that emphasizes the increase in average Year 1970 1980 1990 2000 2010
age. Price per ounce $35.94 $612.56 $383.51 $279.11 $1224.53
22. Make a bar graph that gives an impression of relative sta- 2010 dollars
bility in average age.
23. By what percentage did the average age increase from 2000 c. Make a bar graph of gold prices in constant 2010
to 2007? dollars.
24. In light of the result of Exercise 23, which of the graphs 28. The following table shows the average price per barrel of
you made in Exercises 21 and 22 is a better representation of crude oil in the given year:
the change in average age?
Year 1970 1980 1990 2000 2010
25. Adjusting defense spending for inflation. The amount in Price per barrel $3.39 $37.42 $23.19 $27.39 $71.21
billions of constant 2004 dollars spent on defense in the given
year is shown in the table below. Make a bar graph of the data a. Complete the following table showing oil prices in
and compare the information provided by this graph with the constant 2010 dollars:
graphs in Figures 2.39 and 2.40.
Year 1970 1980 1990 2000 2010
Year 2004 2005 2006 2007 2008 2009 2010 2011 2012
2004 455.8 480.4 490.5 501.7 542.2 581.7 603.3 592.7 555.9 Price per barrel $3.39 $37.42 $23.19 $27.39 $71.21
dollars 2010 dollars
130 CHAPTER 2
b. Make a bar graph of ticket prices in constant 2010 Represent the data in this table in two ways: by sketching
dollars. a bar graph and by sketching a pie chart.
c. Compare ticket prices in constant dollars for 1970 and
Exercise 32 is suitable for group work.
2010.
32. Health care and longevity. The graphical presentation in
30. Federal receipts. The pie chart in Figure 2.65 shows gov-
Figure 2.66 compares health care spending with life expectancy
ernment revenue in the United States in 2013. The total amount
in various countries. Where on this graph is the best place to be
of federal revenue was about 5.36 trillion dollars. Complete the
(lower left, lower right, upper left, upper right)? Where on this
table below.
graph is the worst place to be (lower left, lower right, upper
left, upper right)? Make a plot with the axes reversed and an-
Social Ad Fees Business swer the previous questions. See if you can find another way to
Income insurance valorem and and present the data, either of your own devising or from the Web.
Source taxes taxes taxes charges other revenue
Percent
Amount 83
ITA JPN
AUS
82 ESP
SUE
GBR NLD
74
CHN
FIGURE 2.65 Government revenue in 2013. BRA
73
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 $10,000
31. A cake. The following table shows ingredients in a hypo-
thetical recipe for a cake as a percent of the total by weight. Health care spending per person (U.S. dollars)
(The authors cannot recommend actual use of this recipe.)
FIGURE 2.66 Health-care spending. Source: Organisation for
Ingredients Flour Milk Sugar Eggs Economic Cooperation and Development (OECD) Health Data
Percent 50 25 15 10 2013.
W-4. Danger signs. You see a graph where the vertical scale begins with a large
number rather than zero. Explain why this should always give you reason to be
suspicious.
W-5. Pie charts. When you see a pie chart that you might be suspicious about, you
could try adding up the numerical values assigned to the pieces of the pie. Explain
why this might help detect an error or something misleading.
W-6. Better graph? Elvis says he’s going to use a pie chart to help illustrate the
growth of social media over the last four years. How would you explain to Elvis
that this is probably not the right kind of chart to use? What would you recommend
as a better choice?
W-7. Poverty. A county reports that the number of people in the county living be-
low the poverty threshold has increased from 100 to 200 over a decade. This looks
like bad news. The report also notes, however, that the population of the county
has increased from 1000 to 2000. Use percentages to describe in another way how
the poverty rate has changed. How is this procedure similar to adjusting monetary
quantities for inflation?
W-8. Pictorial representations. In this section we have noted several ways in which
pictorial representations can be misleading. What good reasons are there for using
pictorial representations?
W-9. How to do it. If you were going to make a graph that emphasizes a certain
point of view, perhaps by distorting real data, what techniques might you employ?
CHAPTER SUMMARY
In our age, we are confronted with an overwhelming amount of information. Analyz-
ing quantitative data is an important aspect of our lives. Often this analysis requires
the ability to understand visual displays such as graphs, and to make sense of tables
of data. A key point to keep in mind is that growth rates are reflected in graphs and
tables.
132 CHAPTER 2
300
250
Population (millions)
200
150
100
50
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
Year
Data sets often have gaps. With no further information, there is no way to accu-
rately fill in the gaps. But interpolation using the average growth rate is available to
provide estimates. Such estimates should always be viewed critically.
Extending data sets by extrapolation can be helpful in predicting function values,
but doing so is a risky business, especially if we go far beyond the limits of the data
set. This is illustrated in Example 2.10, a study of the age of first-time mothers.
100 100
(feet per second)
80 80
Running speed
Running speed
60 60
40 40
20 20
0 0
10 20 30 40 50 10 20 30 40 50
Length (inches) Length (inches)
800.0 750.0
700.0 700.0
Billions of dollars
Billions of dollars
600.0 650.0
500.0
600.0
400.0
550.0
300.0
200.0 500.0
100.0 450.0
0 400.0
2004 2005 2006 2007 2008 2009 2010 2011 2012 2004 2005 2006 2007 2008 2009 2010 2011 2012
Year Year
FIGURE 2.70 Defense spending. FIGURE 2.71 The same defense spending data.
$80 $100
(in year 2010 dollars)
70 90
Dollars per barrel
80
60
70
50 60
40 50
30 40
30
20
20
10 10
0 0
1970 1980 1990 2000 2010 1970 1980 1990 2000 2010
Year Year
FIGURE 2.72 Price per barrel of oil. FIGURE 2.73 Crude oil price adjusted for
inflation.
Any graph can be misleading if insufficient data are plotted. It may be particularly
difficult to determine how many data points were used to produce a smoothed line
graph.
Often data are presented using images designed to grab our attention. One com-
mon graphical representation is the pie chart, which is used to display how con-
stituent parts make up a whole. In general, visual representations can be misleading
if the relative sizes of the items pictured do not correspond to the relative sizes of the
data.
Crauder-4100237 QL November 6, 2014 15:7
134 CHAPTER 2
KEY TERMS
independent variable, p. 79 relative change, p. 82 scatterplot, p. 100
dependent variable, p. 79 average growth rate, p. 86 line graph, p. 100
function, p. 79 interpolation, p. 88 smoothed line graph, p. 106
percentage change, p. 82 extrapolation, p. 91
CHAPTER QUIZ
1. The following table shows the population (in millions) of b. Interpolation gives 88.5, so the cumulative number of cases
greater New York City in the given year: is about 89.
If you had difficulty with this problem, see Example 2.8.
Date 1900 1920 1940 1960 1980 2000 3. The following table shows the barometric pressure in (mil-
Population 3.44 5.62 7.46 7.78 7.07 8.01
limeters of mercury) at various altitudes above sea level:
(millions)
Altitude (thousand feet) 0 20 40 60 80 100
Make a table and a bar graph showing the percentage change
Pressure (mm mercury) 760 350 141 54 21 8
over each 20-year period. Round your answers in percentage
form to the nearest whole number. Make a scatterplot and a line graph of the data. Use the altitude
for the independent variable and the pressure for the dependent
Answer
variable.
Date range 1900–20 1920–40 1940–60 1960–80 1980–2000 Answer See Figures 2.74 and 2.75.
Percentage 63% 33% 4% −9% 13%
change
Millimeters of mercury
Millimeters of mercury
800 800
700 700
If you had difficulty with this problem, see Example 2.3. 600 600
500 500
400 400
70 300 300
60 200 200
Percentage change
100 100
50
0 0
40 0 20 40 60 80 100 120 0 20 40 60 80 100
30 Thousands of feet Thousands of feet
20
10
FIGURE 2.74 Pressure versus FIGURE 2.75 Pressure versus
0 altitude: scatterplot. altitude: line graph.
–10
–20
If you had difficulty with this problem, see Example 2.11.
20
40
60
80
00
19
19
19
19
20
–
00
20
40
60
80
19
19
19
19
19
5. The following table shows total sales in the given month: a. Complete the following table showing the cost in constant
2010 dollars:
Month Jan. Feb. March April May
Sales $5322 $5483 $6198 $6263 $6516
Year 1970 1980 1990 2000 2010
Make a bar graph that emphasizes the increase in sales over Cost $23.00 $35.00 $65.00 $80.00 $116.00
the five-month period, and make a second bar graph that de- 2010 dollars
emphasizes the increase.
Answer To emphasize the increase, we use a narrow range on b. Make a bar graph showing the cost in constant 2010 dollars.
the vertical axis. One possibility is shown in Figure 2.77. To c. When was the item most expensive, taking inflation into
deemphasize the increase, we use a wide range on the vertical account?
axis. One possibility is shown in Figure 2.78.
Answer
$7000 $7000 a.
Sales (dollars)
Sales (dollars)
6000
6500 5000
4000 Year 1970 1980 1990 2000 2010
6000
3000
5500 2000 Cost $23.00 $35.00 $65.00 $80.00 $116.00
1000 2010 dollars $126.73 $88.90 $106.60 $100.80 $116.00
5000 0
.
.
b.
b.
Jan
Jan
ch
ch
ay
ay
r il
r il
Fe
Fe
ar
ar
M
M
Ap
Ap
b.
M
2010 dollars
100
growth in sales. growth in sales.
80
Time span 1970–2010 1980–2010 1990–2010 2000–2010 c. The maximum cost, adjusted for inflation, occurred in 1970.
Inflation 451% 154% 64% 26%
If you had difficulty with this problem, see Example 2.17.
Crauder-4100237 QL November 6, 2014 15:45
T
he following article is an official statement from the United Nations.
Concern over world hunger is nothing new. The English demographer and politi-
© Interphoto/Alamy
cal economist Thomas Robert Malthus (1766–1834) is best known for his influential
views on population growth. In 1798 he wrote the following in An Essay on the Prin-
ciple of Population.
I say that the power of population is indefinitely greater than the power in the
earth to produce subsistence for man. Population, when unchecked, increases in
a geometrical ratio whereas the food-supply grows at an arithmetic rate.
The power of population is so superior to the power of the earth to produce
subsistence for man, that premature death must in some shape or other visit the
human race.
Note the distinction Malthus makes between a “geometrical ratio” and an “arith-
metic rate.” In contemporary terms, we would say that population growth is
exponential and food production growth is linear. Inherent properties of these two
types of growth led Malthus to his dire predictions. His analysis is regarded today
Thomas Robert Malthus, as an oversimplification of complex issues, but these specific types of growth remain
1766--1834. crucial tools in analyzing demand versus availability of Earth’s resources.
Change can follow various patterns, but linear and exponential patterns are of
particular importance because they occur so often and in such significant applications.
137
Crauder-4100237 QL November 6, 2014 15:45
138 CHAPTER 3
We will examine what they mean in this chapter. We will also consider the logarithm
and its applications.
constant growth Home TV & Video U. S. World Politics Entertainment Tech Travel
rate.
Sled Dog Days of Summer
National Snow and Ice Data Center, Boulder, CO
August 6, 2014
A rctic sea ice extent declined at a fairly rapid rate through the first three weeks
of July, but the loss rate then slowed due to a shift in weather patterns. In
Antarctica, the advance of sea ice nearly halted for about a week in early July,
and then resumed. At the end of the month, Antarctic extent was at or near a
record high for this time of year.
July 2014 average ice extent was 8.25 million square kilometers (3.19 million
square miles). This is 1.85 million square kilometers (714,000 square miles) below
the 1981 to 2010 average for the month.
Ice extent is below average in nearly all sectors of the Arctic. Open water
continued to grow in the Laptev and Beaufort Seas, reaching well north of 80◦ N
in the Laptev Sea. By the end of the month, the Alaskan Coast was essentially
free of ice except for small patches of very diffuse ice off Barrow. The Barents
Sea, Hudson Bay, and Baffin Bay/Davis Strait are now essentially ice free. Large
areas of low concentration ice in the central Beaufort Sea are likely to melt out
in coming weeks. The Northwest Passage through the channels of the Canadian
Arctic Archipelago remains choked with ice. Parts of the Northern Sea Route are
still difficult to traverse because of high-concentration, near-shore ice between
the Laptev and East Siberian seas and also north of the Taymyr Peninsula.
For July 2014 as a whole, ice extent declined at an average rate of 86,900
square kilometers (33,600 square miles) per day, close to the 1981 to 2010 aver-
age July rate of 86,500 square kilometers (33,400 square miles) per day. However,
this averages together a fairly fast rate of decline over the first three weeks of the
month with a slower rate of decline over the remainder of the month.
The slower ice loss later in the month reflects a shift in weather patterns.
For much of the month, high pressure at sea level dominated the central Arctic
Ocean and the Barents Sea. However, this pattern broke down and was replaced
by lower-than-average pressure over the central Arctic Ocean. A low pressure
pattern tends to bring cool conditions and the counterclockwise winds associated
with this pattern also tend to spread the ice out.
July 2014 is the 4th lowest Arctic sea ice extent in the satellite record, 340,000
square kilometers (131,000 square miles) above the previous record lows in July
2011, 2012, and 2007.
The analysis in this article is accompanied by Figure 3.1, which shows how the
monthly July ice extent declined from 1979 to 2014. The figure uses a straight line
to help us understand the decline. Straight lines and the relationships they represent
frequently occur in the media, and every critical reader needs a basic understanding
of them.
Crauder-4100237 QL November 6, 2014 15:45
3.1 Lines and linear growth: What does a constant rate mean? 139
12.0
Year
FIGURE 3.1 Arctic sea ice extent in July, 1979 to 2014.
$400 $400
350 350
300 300
Balance (dollars)
Balance (dollars)
250 250
200 200
150 150
100 100
50 50
0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Number of months Number of months
FIGURE 3.2 Adding money to a cookie FIGURE 3.3 Taking money from a cookie
jar: Positive growth rate corresponds to jar: Negative growth rate corresponds to a
an increasing linear function. decreasing linear function.
Crauder-4100237 QL November 6, 2014 15:45
140 CHAPTER 3
75
Height (inches)
50
25
0
0 5 10 15 20 25 30
Age (years)
preceding examples. For comparison, a graph of typical height versus age for females
is shown in Figure 3.4. It is not a straight line, and this is further evidence that height
as a function of age is not a linear function.
cartoonstock.com
3.1 Lines and linear growth: What does a constant rate mean? 141
$9 $90
FIGURE 3.5 Cost is a linear function of FIGURE 3.6 Salary is not a linear
number of wedding guests. function of time.
SOLUTION
a. The growth rate in this case is the extra cost incurred for each additional
guest, which is $15. The additional cost is the same for each additional guest,
so the growth rate is constant. Because the growth rate is constant, the total
cost of the reception is a linear function of the number of guests. The graph of
cost versus number of guests is shown in Figure 3.5. It is a straight line, as
expected.
b. The growth rate in this case is the increase per year in my salary. The first
year my salary increased by 10% of $30,000, or $3000, so my new annual
salary is $33,000. The second year my salary increased by 10% of $33,000, or
$3300. The growth rate is not the same each year, so my salary is not a linear
function of time in years. A graph showing salary over several years with a
10% increase each year is illustrated in Figure 3.6. The graph is not a straight
line, which is further verification that salary is not a linear function of time in
years.
1 1 SQUARE
1 UNIT
TRY IT YOURSELF 3.1
2
4 SQUARE One of the following is a linear function, and one is not. In each case, determine
UNITS
the practical meaning of the growth rate, then determine whether or not the given
2
function is linear.
3
9 SQUARE a. One inch is the same as 2.54 centimeters. Consider distance in centimeters to
UNITS
be a function of distance in inches.
3
b. Consider the area of a square to be a function of the length of a side.
FIGURE 3.7 Areas Figure 3.7 may be helpful here.
of squares of
different sizes. The answer is provided at the end of this section.
142 CHAPTER 3
x +b
y =m m units
1 unit
3.1 Lines and linear growth: What does a constant rate mean? 143
SOLUTION
The initial value of the linear function y = mx + b is b, and the growth rate or slope
is m. So, for L = 0.14n + 7.20, the initial value is 7.20 meters. It is the (approximate)
length of the winning long jump in the 1900 Olympic Games. The growth rate is
0.14 meter per Olympic Game. This means that the length of the winning long jump
The Olympic symbol. increased by approximately 0.14 meter from one game to the next.
Note that in Example 3.2, we used the variables n and L rather than x and y.
There is nothing special about the names x and y for the variables, and it is common
to use variable names that help us remember their practical meanings.
If we know the initial value and growth rate for a linear function, we can imme-
diately write its formula. The method is shown in the following example.
A rocket starting from an orbit 30,000 kilometers above the surface of Earth blasts
off and flies at a constant speed of 1000 kilometers per hour away from Earth.
Explain why the function giving the rocket’s distance from Earth in terms of time
is linear. Identify the initial value and growth rate, and find a linear formula for the
distance.
SOLUTION
First we choose letters to represent the function and variable. Let d denote the
distance in kilometers from Earth after t hours. The growth rate is the velocity,
1000 kilometers per hour. (The velocity is positive because the distance is increas-
ing.) Because the growth rate is constant, d is a linear function of t. The initial value
is 30,000 kilometers, which is the height above Earth at blastoff (that is, at time
t = 0). We find the formula for d using
The Saturn V carried the first
men to the moon in 1969. d = Growth rate × t + Initial value
d = 1000t + 30,000
144 CHAPTER 3
the function giving the object’s distance from Earth in terms of time is linear. Identify
the initial value and growth rate, and find a linear formula for the distance.
The answer is provided at the end of this section.
Change in function
Slope = Growth rate =
Change in independent variable
Change in y
m = Slope = Growth rate =
Change in x
The slope is measured in inches per hour, which highlights the fact that the slope
represents the rate of snowfall.
pth
de
w
Sno Change of
12 inches
+b
mx
y= Change in y
4 hours 20 inches
8 inches
Change in x
FIGURE 3.9 How slope is calculated. FIGURE 3.10 Growth rate of snow accumulation.
Crauder-4100237 QL November 6, 2014 15:45
3.1 Lines and linear growth: What does a constant rate mean? 145
SOLUTION
We need to know the change in the amount of gas in the tank. When the tank is
three-quarters full, there are 15 gallons of gas left in the tank. The amount of gas in
the tank has decreased by 5 gallons. This means that the change in gas is −5 gallons.
This occurred over two hours of driving, so the slope is
Change in gas
m=
Change in time
−5 gallons
= = −2.5 gallons per hour
2 hours
In practical terms, this means we are using 2.5 gallons of gas each hour.
VIN MORGAN/AFP/Getty Images
An ice core from Antarctica. The answer is provided at the end of this section.
T = 0.25n + 39
Note that the slope of this linear function is 0.25. The slope is measured in degrees
Fahrenheit per chirp per minute.
Suppose initially the temperature is 50 degrees Fahrenheit. An hour later we find
that the number of chirps per minute has increased by 20. What is the new tempera-
ture? An increase of one chirp per minute corresponds to an increase of 0.25 degree
Fahrenheit because 0.25 is the slope. Therefore, an increase by 20 chirps per minute
corresponds to an increase in temperature of 20 × 0.25 = 5 degrees. The new tem-
perature is 50 + 5 = 55 degrees Fahrenheit.
Crauder-4100237 QL November 6, 2014 15:45
146 CHAPTER 3
The next example illustrates the importance of correctly interpreting the slope of
a linear function.
3.1 Lines and linear growth: What does a constant rate mean? 147
a. What is the slope of the linear function giving the temperature in degrees
Fahrenheit in terms of the temperature in degrees Celsius? Use your answer to
determine what increase in degrees Fahrenheit corresponds to a 1-degree
increase on the Celsius scale.
b. Choose variable and function names, and find a linear formula that converts
degrees Celsius to degrees Fahrenheit. Make a graph of the linear function.
c. A news story released by Reuters on March 19, 2002, said that the Antarctic
peninsula had warmed by 36 degrees Fahrenheit over the past half-century.
Such temperature increases would result in catastrophic climate changes
worldwide, and it is surprising that an error of this magnitude could have
slipped by the editorial staff of Reuters. We can’t say for certain how the error
occurred, but it is likely that the British writer saw a report that the temperature
had increased by 2.2 degrees Celsius. Verify that a temperature of 2.2 degrees
Celsius is about 36 degrees Fahrenheit.
d. The report described in part c did not say that the temperature was 2.2
degrees Celsius but that the temperature increased by 2.2 degrees Celsius. What
increase in Fahrenheit temperature should the writer have reported?
SOLUTION
a. To find the slope, we need the change in degrees Fahrenheit for a given
change in degrees Celsius. An increase on the Celsius scale from 0 to 100
degrees corresponds to an increase on the Fahrenheit scale from 32 to 212, a
180-degree increase. Therefore,
Change in degrees Fahrenheit
Slope =
Change in degrees Celsius
180
= = 1.8 degrees Fahrenheit per degree Celsius
100
Therefore, a 1-degree increase on the Celsius scale corresponds to a 1.8-degree
increase on the Fahrenheit scale.
b. We use F for the temperature in degrees Fahrenheit and C for the
temperature in degrees Celsius. The linear relationship we seek is expressed by
the formula F = Slope × C + Initial value. We found in part a that the slope is
1.8. Now we need the initial value, which is the Fahrenheit temperature when
the Celsius temperature is 0 degrees. That is 32 degrees Fahrenheit. Therefore,
the formula is F = 1.8C + 32. We have graphed this function in Figure 3.11.
225
200
Degrees Fahrenheit
175
150
125
100
75
50
25
0
-20 -10 0 10 20 30 40 50 60 70 80 90 100
Degrees Celsius
148 CHAPTER 3
c. We put 2.2 degrees Celsius into the formula for converting Celsius to
Fahrenheit:
F = 1.8C + 32
= 1.8 × 2.2 + 32 = 35.96 degrees Fahrenheit
Rounding gives 36 degrees Fahrenheit.
d. The slope of the linear function that converts from Celsius to Fahrenheit is
1.8 degrees Fahrenheit per degree Celsius. As we noted in the solution to part a,
this tells us that each 1-degree increase on the Celsius scale corresponds to a
1.8-degree increase on the Fahrenheit scale. So an increase of 2.2 degrees
Celsius corresponds to an increase of 2.2 × 1.8 = 3.96 degrees Fahrenheit. The
writer should have reported a warming of about 4 degrees Fahrenheit.
In Figure 3.12, we have plotted these data points. The data points appear to
fall on a straight line, which indicates that we can use a linear function to describe
taxable income, at least over the range given in the table. To verify this, we calculate
the average growth rate over each of the income ranges. Recall from Chapter 2 that
we find the average growth rate using
Change in function Change in tax owed
Average growth rate = =
Change in independent variable Change in taxable income
These calculations are shown in the following table:
The fact that the average growth rate over each interval is the constant 0.15 tells
us that the tax owed is a linear function of income with slope $0.15 per dollar. (If
we had obtained different values for the average growth rate over different intervals,
we would have concluded that the data are not related by a linear formula.)
This slope is known as the marginal tax rate. It tells us the additional tax owed
on each additional dollar of income. For example, if you have $100 more in taxable
income than someone else, your tax burden will be greater by 0.15 × 100 = 15
dollars. Let T be the tax owed in dollars, and let I be the taxable income (also in
dollars) in excess of $41,000. The tax owed when I = 0 is $5261, and this is the
1 The actual table shows taxable income in increments of $50 rather than $100.
Crauder-4100237 QL November 6, 2014 15:45
3.1 Lines and linear growth: What does a constant rate mean? 149
$5330 $5330
5320 5320
5310 5310
FIGURE 3.12 Tax data fall on a straight line. FIGURE 3.13 Line added to plot of tax data.
In Figure 3.13, we have added the graph of the line to the plot of the tax data.
The graph passes through each of the data points, which verifies the validity of our
formula.
cartoonstock.com
The linear function fits the tax data exactly. Many data sets are not perfectly linear
like this, but they may be fairly well approximated by a linear function. And we may
be able to get useful information from such an approximation. This is especially true
when we have good reason to suspect that the relationship is approximately linear.
As a case in point, let’s return to the example of running speed versus length
discussed in Section 2.2. We recall the following table, which shows the running
speed of various animals versus their length:2
2 This table is adapted from J. T. Bonner, Size and Cycle (Princeton, NJ: Princeton University Press, 1965).
Crauder-4100237 QL November 6, 2014 15:45
150 CHAPTER 3
Figure 3.14 shows the scatterplot. We note that the points do not fall on a straight
line, so the data in the table are not exactly linear. Nevertheless, the points seem to
fall nearly on a straight line, so it may make sense to approximate the data using a
linear function. Statisticians, scientists, and engineers frequently use a line called a
regression line or trend line to make such approximations.
The procedure for calculating a trend line for given data is called linear regres-
sion. An explanation of this procedure is outside the scope of this book, but linear
regression is programmed into most calculators and commercial spreadsheets. All
one has to do is enter the data.
In Figure 3.15, we have added the trend line produced by the spreadsheet program
Excel© . We see that the points lie close to the line, so it appears to be a reasonable
approximation to the data. Note that the spreadsheet gives both the graph and the
linear formula.
The equation of the trend line is y = 2.03x + 5.09. (We have rounded the initial
value and the slope to two decimal places.) This means that running speed S in feet
per second can be closely estimated by
S = 2.03L + 5.09
where L is the length measured in inches. We can use this formula to make estimates
about animals not shown in the table. For example, we would expect a 20-inch-long
120 120
100 100
y = 2.0298x + 5.0873
(feet per second)
80 80
Running speed
60 60
40 40
20 20
0 0
10 20 30 40 50 10 20 30 40 50
Length (inches) Length (inches)
FIGURE 3.14 Scatterplot of running speed FIGURE 3.15 Trend line added.
versus length.
Crauder-4100237 QL November 6, 2014 15:45
3.1 Lines and linear growth: What does a constant rate mean? 151
animal to run
S = 2.03 × 20 + 5.09 = 45.69 feet per second
or about 45.7 feet per second. The most important bit of information we obtain from
the trend line is its slope. In this case, that is 2.03 feet per second per inch. This value
for the slope means that, in general, an animal that is 1 inch longer than another
would be expected to run about 2.03 feet per second faster.
We can also gain information by studying the position of the data points relative
to the trend line. For example, in Figure 3.15, the point for the red fox is above the
trend line, so the red fox is faster than one might expect for an animal of its size.
EXAMPLE 3.7 Meaning of the equation for the trend line: Home runs
A graph of the data giving home run production over time shows a steady increase.
The formula for the trend line is H = 0.1t +5.1, where H is the number of home runs
per 500 plate appearances and t is the time in years since 1919. Explain in practical
terms the meaning of the slope.
SOLUTION
The slope is 0.1, so each year the number of home runs per 500 plate appearances
increases, on average, by 0.1.
152 CHAPTER 3
b. The growth rate for the area is the increase in area for each unit increase in
the length of a side. The area of a square is not a linear function of the length of
a side.
Try It Yourself 3.2: Interpreting formulas: Athletic records The initial value of
3.3 meters is the (approximate) height of the winning pole vault in the 1900 Olympic
Games. The growth rate of 0.20 meter per Olympic game means that the height of
the winning pole vault increased by approximately 0.20 meter from one game to the
next.
Try It Yourself 3.3: Finding linear formulas: Distance from Earth Let d denote the
distance from Earth in kilometers after t hours. Then d is a linear function of t because
the growth rate (the velocity) is constant. The initial value is 1,000,000 kilometers,
and the growth rate is −2000 kilometers per hour (negative because the object is
traveling toward Earth, so the distance is decreasing). The formula is d = −2000t +
1,000,000.
Try It Yourself 3.4: Calculating the slope: Ice depth 0.6 foot per year. The depth
increased by 0.6 foot each year.
Try It Yourself 3.5: Using the slope: Temperature and speed of sound 3.3 feet
per second.
Try It Yourself 3.7: Meaning of the equation for the trend line: Walks Each year
the number of walks per 500 plate appearances increases, on average, by 0.2.
3.1 Lines and linear growth: What does a constant rate mean? 153
11. My savings account pays 2% per year, and the interest is izontal miles from the base. At the top, you see a sign that
compounded yearly. That is, each year accrued interest is added reads “Elevation 4100 Feet.” What is the growth rate in your
to my account balance. Is the amount of money in my savings elevation with respect to horizontal distance as you drive up
account a linear function of the number of years since it was the road? Use E for elevation in feet and h for horizontal dis-
opened? Suggestion: Compare the interest earned the first year tance in miles, and find a formula that gives your elevation as
with the interest earned the second year. a linear function of your horizontal distance from the base of
12. If we drop a rock near the surface of Earth, its velocity in- the hill.
creases by 32 feet per second for each second the rock falls. Is 22--25. Price of Amazon’s Kindle. Figure 3.16 shows how
the velocity of the rock a linear function of the time since it was the price of Amazon’s Kindle has decreased over time. The
dropped? green line is a trend line. (You can check that the hash marks
13. If we drop a rock near the surface of Earth, physics tells on the horizontal axis are one-month intervals.) Complete Ex-
us that the distance in feet that it falls is given by 16t 2 , where ercises 22 through 25 using this information.
t is the number of seconds since the rock was dropped. Is the
distance a rock falls a linear function of the time since it was
$450
dropped? Suggestion: Compare the distance fallen over the first
$400
second to that fallen over the next second.
$350
14. In order to ship items I bought on eBay, I pay a delivery
$300
company a flat fee of $10, plus $1.50 per pound. Is the ship-
ping cost a linear function of the number of pounds shipped? $250
$200
15. My car gets 25 miles per gallon at 60 miles per hour. Is
$150
the number of miles I can drive at 60 miles per hour a linear
function of the number of gallons of gas in the tank? $100
$50
16. A certain type of bacterium reproduces by dividing each
hour. If we start with just one bacterium, is the number of bac- $0
Nov Mar Jul Nov Mar Jul Nov Mar Jul Nov Mar Jul Nov Mar Jul
teria present a linear function of time? ’07 ’08 ’08 ’08 ’09 ’09 ’09 ’10 ’10 ’10 ’11 ’11 ’11 ’12 ’12
17. On a typical day, the temperature increases as the sun rises.
It is hottest around midday, and then things cool down toward FIGURE 3.16 Price of the Kindle. Source: Company Filings
evening. Is the relationship between temperature and time of (Sept 2011). Reprinted by permission of Business Insiders, Inc.
day linear?
18. The method of calculating the compensation paid to re- 22. The trend line estimates that the price was $432 in Novem-
tired teachers in the state of Oklahoma is as follows.3 Yearly ber 2007 and $80 in September 2011. (That is an interval of
retirement income is 0.02 times the number of years of service 46 months.) Use this information to determine the slope of the
times the average of the last three years of salary. Is retirement trend line. Round your answer in dollars per month to two
income for Oklahoma teachers a linear function of the number decimal places. Be careful about the sign.
of years of service? 23. Use your answer to Exercise 22 to determine by how much
19. Income tax. One federal tax schedule states that if you the price of a Kindle should decrease every six months.
are single and your taxable income is between $35,350 and 24. Use the information in Exercise 22 to find a formula for
$85,650, then you owe $4867.50 plus 25% of your taxable the price, in dollars, as a linear function of the time in months
income over $35,350. Is the tax you owe a linear function of since November 2007.
your taxable income if the taxable income is between $35,350 25. Use your answer to Exercise 24 to determine the predic-
and $85,650? If so, find a formula giving tax owed T as a lin- tion of the trend line for the price in July 2012 (56 months
ear function of taxable income I over $35,350. Both T and I after November 2007). What do you conclude from this un-
are measured in dollars. reasonable result?
20. Working on commission. A certain man works in sales and 26--31. Speed of sound in the ocean. Measuring the speed
earns a base salary of $1000 per month, plus 5% of his total of sound in the ocean is an important part of marine research.4
sales for the month. Is his monthly income a linear function of One application is the study of climate change. The speed
his monthly sales? If so, find a linear formula that gives monthly of sound depends on the temperature, salinity, and depth be-
income in terms of sales. low the surface. For a fixed temperature of 25 degrees Celsius
21. Driving up a hill. Suppose that you are at the base of a and salinity of 35 parts per thousand, the speed of sound is a
hill and see a sign that reads “Elevation 3500 Feet.” The road function of the depth. At the surface, the speed of sound is 1534
you are on goes straight up the hill to the top, which is 3 hor- meters per second. For each increase in depth by 1 kilometer,
154 CHAPTER 3
the speed of sound increases by 17 meters per second. This in- with rent, staff salaries, utilities, and supplies. In the month of
formation is used in Exercises 26 through 31. Complete these August, total sales were $832,000 and net income (after paying
exercises in order. fixed costs) was $9703. Because each dollar increase in sales re-
26. Explain why the function expressing the speed of sound in sults in the same increase (a certain percentage of that dollar)
terms of depth is linear. in net income, net income is a linear function of total sales.
Find a linear formula that gives net income in terms of total
27. Identify the slope and initial value of the linear function sales. What percentage commission on sales is this company
that gives the speed of sound in terms of depth. Explain in prac- charging?
tical terms what each means.
40. Meaning of marginal tax rate. Recall from the discussion
28. What increase in the speed of sound is caused by a 2- on page 148 that the marginal tax rate is the additional tax
kilometer increase in depth? you expect to pay for each additional dollar of taxable income.
29. Use your answer to Exercise 28 to determine the speed of The marginal tax rate is usually expressed as a percentage. In
sound when the depth is 2 kilometers. the example, the marginal tax rate was 15%. Does that mean
that your total tax is 15% of your taxable income? Specifi-
30. Use D for depth (in kilometers) and S for the speed of
cally, if your taxable income is $41,300, is your total tax 15%
sound (in meters per second), and find a linear formula for S
of $41,300? Explain why or why not.
as a function of D.
41--44. More on tax tables. The following table shows the
31. Use your formula from Exercise 30 to calculate the speed
federal income tax owed by a single taxpayer for the given level
of sound when the depth is 2 kilometers.
of taxable income for 2013. Both are measured in dollars.
32--35. Speed on a curve. On rural highways, the average
speed S (in miles per hour) is related to the amount of curva- Taxable 97,000 97,050 97,100 97,150 97,200 97,250 97,300
ture C (in degrees) of the road.5 On a straight road (C = 0), the income
average speed is 46.26 miles per hour. This decreases by 0.746 Tax owed 20,460 20,474 20,488 20,502 20,516 20,530 20,544
mile per hour for each additional degree of curvature. This
information is used in Exercises 32 through 35. Complete these This information is used in Exercises 41 through 44. Complete
exercises in order. these exercises in order.
32. Explain why the function expressing S in terms of C is 41. Show that, over the range of taxable income shown in the
linear. table, the tax owed is a linear function of taxable income.
33. Find the slope of the linear function expressing S in terms 42. What tax do you owe if you have a taxable income of
of C. Explain in practical terms its meaning. Suggestion: Be $97,000?
careful about the sign of the slope. 43. How much additional tax is due on each dollar of taxable
34. Find a formula expressing S as a linear function of C. income over $97,000?
35. What is the average speed if the degree of curvature is 44. Let A denote the amount (in dollars) of your taxable in-
15 degrees? come in excess of $97,000 and T the tax (in dollars) you owe.
Find a linear formula6 that gives T in terms of A.
36--38. The IRS. In 2003, the Internal Revenue Service col-
lected about 1.952 trillion dollars. Based on the trend line for 45--48. The flu. The following table shows the total number
the data from 2003 through 2008, collections increased by (ap- of patients diagnosed with the flu in terms of the time in days
proximately) 178 billion (or 0.178 trillion) dollars per year. In since the outbreak started:
Exercises 36 through 38 we assume that collections increased
by exactly 178 billion dollars per year from 2003 through Time in days 0 5 10 15 20 25
2008. Complete these exercises in order. Number of flu patients 35 41 47 53 59 65
36. Explain how the preceding assumption would show that
This information is used in Exercises 45 through 48. Complete
the function expressing tax collections in terms of time from
these exercises in order.
2003 through 2008 is linear. Why do we have to limit the time
interval to 2003 through 2008? 45. Show that the function giving the number of diagnosed flu
cases in terms of time is linear.
37. Find a formula for tax collections as a linear function of
time from 2003 through 2008. 46. Find the slope of the linear function from Exercise 45. Ex-
plain in practical terms the meaning of the slope.
38. What would you estimate for the collections in 2006?
47. Find a formula for the linear function in Exercise 45.
39. A real estate agency. Suppose that a real estate agency
makes its money by taking a percentage commission of total 48. What would you expect to be the total number of diag-
sales. It also has fixed costs of $15,257 per month associated nosed flu cases after 17 days?
5 A. Taragin, “Driver Performance on Horizontal Curves,” Proceedings of the Highway Research Board 33
(Washington, DC: Highway Research Board, 1954), 446–466.
6 The formula does not apply for taxable incomes above $100,000.
Crauder-4100237 QL November 6, 2014 15:45
3.1 Lines and linear growth: What does a constant rate mean? 155
CPI
CPI
210 210
205 205
1980 1981 1982 1983 1984 1985 200 200
2747.7 2725.3 2704.8 2597.6 2494.8 2414.0 195 195
190 190
1986 1987 1988 1989 1990 1991 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
2382.6 2428.8 2500.2 2458.8 2320.3 2234.9 Time in years since 2005 Time in years since 2005
1992 1993 1994 1995 1996 1997
2226.0 2233.2 2220.8 2273.5 2273.1 2358.4 FIGURE 3.19 Plot of FIGURE 3.20 Plot of
consumer price index. consumer price index with
1998 1999 2000 2001 2002 2003 trend line added.
2439.1 2485.6 2553.8 2569.0 2621.5 2719.9
2004 2005 2006 2007 2008
2753.4 2799.3 2815.5 2892.4 2999.5 52. What CPI does the trend line predict for December 2013?
Round your answer to one decimal place.
Exercises 53 through 57 are suitable for group work.
The scatterplot of the data is shown in Figure 3.17. In Fig-
ure 3.18, the trend line has been added. Does the trend line ap- 53--57. A closer look at tax tables. The table below gives se-
pear to offer an appropriate way to analyze the data? Explain lected entries from the 2009 federal income tax table. It applies
your reasoning. to a married couple filing jointly. Taxable income and tax owed
are both in dollars.
156 CHAPTER 3
W-2. Is it linear? You see an article about an interesting biological phenomenon. The
article is accompanied by a graph. What properties of the graph might lead you to
believe that the phenomenon is expressed using a linear relationship? What properties
of the graph might lead you to believe that the phenomenon is not expressed using a
linear relationship?
W-3. When does it apply? There are many settings in which linear functions model
relationships over limited ranges only. For example, a factory owner says he has
determined that the number of units U produced is related to the number W of
workers by the linear function U = 3W + 2. Is this relationship reasonable if there
are no workers? Use the Internet to investigate other settings where linear func-
tions apply over limited ranges only. Suggestion: There are examples of this sort in
this section.
Exercise 25 provides additional information relevant to this topic.
W-4. Marginal tax rate. Recall that the marginal tax rate is the slope of the linear
function giving your tax as a function of your taxable income. Often the marginal
Crauder-4100237 QL November 6, 2014 15:45
3.1 Lines and linear growth: What does a constant rate mean? 157
tax rate is expressed as a percentage. For example, a slope of $0.15 per dollar cor-
responds to a marginal tax rate of 15%. Does this fact mean that if your taxable
income is $100, then your tax will be 15% of $100, or $15? If not, explain carefully
how this percentage can be used to describe how your tax depends on your taxable
income.
Exercise 43 provides additional information relevant to this topic.
W-5. Appropriate use of trend lines. A trend line can be found for any set of nu-
merical data points. What are some ways to determine whether the trend line offers
an appropriate way to analyze the data?
Exercise 49 provides additional information relevant to this topic.
W-6. Trend lines. Explain how an appropriately used trend line can help us under-
stand data.
Exercise 51 provides additional information relevant to this topic.
W-7. Piecewise linear functions. Sometimes a linear growth rate changes abruptly to
a larger or smaller linear growth rate. When this happens, the graph takes the form
of a series of straight lines pasted together, as shown in Figure 3.22. Find examples
of common relationships that are piecewise-linear. Suggestion: You might start by
looking at federal income tax rates or postal rates.
W-8. Is it linear? Look at historical records of tuition rates at your university. Deter-
mine whether it is reasonable to model these rates by a linear function. If so, propose
one.
W-9. Positive, negative, or zero. Suppose you are looking at three separate relation-
ships that are modeled by linear functions. One linear model has a positive slope,
one has a negative slope, and one has a slope of zero. What do these slopes tell you
about the relationships they model? Find practical examples of each of the three.
Crauder-4100237 QL November 6, 2014 15:45
158 CHAPTER 3
M. Daley from Point Richmond, CA, asks “The nuclear accident in Fukushima
raised concerns about the safety of spent fuel storage pools in nuclear
reactors. How vulnerable are the pools at U.S. reactors, especially in light of the
Kyodo/AP Photo
threat of a possible terrorist attack, and what, if anything, can be done to make
spent fuel storage safer?” and is answered by UCS Global Security Program Co-
Director David Wright, Ph.D.
The highly radioactive fuel rods in nuclear power plants are moved from the
reactor core to cooling pools when they are “spent” and are replaced by fresh
fuel to drive the nuclear chain reaction that generates electricity. You are right
that spent fuel pools have raised safety concerns because an event like a severe
accident or terrorist attack that caused a loss of water from the pool could result
in a self-sustaining fire that would damage the fuel and could cause a massive
radiological release.
In the United States, because there is no permanent storage site for the spent
fuel from the nation’s nuclear power plants, spent fuel has been kept at the
reactor sites, with the vast majority of it stored in spent fuel pools. In the past,
Storage for spent nuclear most analysis focused on the safety of spent fuel pools rather than their security
fuel—the spent fuel pool at the against terrorist attack. That changed after September 11, 2001, when studies
No. 3 reactor of the Tomari began to look at the possibility of terrorism. A 2006 National Academy of Sciences
nuclear power plant in the study, for example, concluded that “under some conditions, a terrorist attack
village of Tomari, Hokkaido, that partially or completely drained a spent fuel pool could lead to a propagating
Japan’s northernmost prefecture, zirconium cladding fire and the release of large quantities of radioactive materials
was shown to the press on to the environment.”
March 12, 2014.
Dangerous radioactive materials, such as those referred to in the article, will even-
tually decay into relatively harmless matter. The problem is that this process takes an
extremely long time because the way in which these materials decay is exponential,
not linear. We will return to this issue later in this section.
Recall that linear change occurs at a constant rate. Exponential change, in con-
trast, is characterized by a constant percentage rate. The rates may be positive or
negative. In certain idealized situations, a population of living organisms, such as
bacteria, grows at a constant percentage rate. Radioactive substances provide an ex-
ample of a negative rate because the amount decays rather than grows.
petri dish and the population doubles every hour. We refer to 2000 as the initial
value because it is the amount with which we start. After one hour the population
doubles, so
Population after 1 hour = 2 × Initial population
= 2 × 2000 = 4000 bacteria
After one more hour the population doubles again, so
Population after 2 hours = 2 × Population after 1 hour
= 2 × 4000 = 8000 bacteria
In general, we find the population for the next hour by multiplying the current pop-
ulation by 2:
Population next hour = 2 × Current population
Let’s find the percentage change in the population. Over the first hour the pop-
ulation grows from 2000 to 4000, an increase of 2000. The percentage change is
then
Change in population 2000
Percentage change = × 100% = × 100% = 100%
Previous population 2000
Over the second hour the population grows from 4000 to 8000, another increase of
100%. In fact, doubling always corresponds to a 100% increase. Thus, the popula-
tion shows a 100% increase each hour. This means that the population grows not by
a constant number but by a constant percentage, namely 100%. Growth by a con-
stant percentage characterizes exponential growth. In this example, the population
size is an exponential function of time.
160 CHAPTER 3
The pattern is evident. To calculate the population after t hours, we multiply the
initial value 2000 by 2 raised to the power t:
N = 2000 × 2t
In Figure 3.23, we have plotted the data points we calculated for the population.
In Figure 3.24, we have added the graph of N = 2000×2t . This curve passes through
each of the data points. This gives further evidence that the formula we proposed is
correct.
70,000 70,000
60,000 60,000
50,000 50,000
Population
Population
40,000 40,000
30,000 30,000
20,000 20,000
10,000 10,000
0 0
0 1 2 3 4 5 0 1 2 3 4 5
Hours Hours
FIGURE 3.23 A population doubles each FIGURE 3.24 Adding the graph of the
hour. exponential function.
Crauder-4100237 QL November 6, 2014 15:45
The initial value is $800 because that is the original investment. The base is 1.07
because that is the multiplier we use to find next year’s balance. This gives the formula
B = 800 × 1.07t
162 CHAPTER 3
$2500
2000
Balance (dollars)
1500
1000
500
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Year
Find the percentage increase each month. If the initial investment is $450, find an
exponential formula for the balance as a function of time. What is the balance after
two years?
The answer is provided at the end of this section.
150,000 150,000
125,000 125,000
Growth
100,000 100,000 from 5th hour
Population
Population
to 6th hour
75,000 75,000
0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Hours Hours
FIGURE 3.26 Graph of population growth. FIGURE 3.27 Comparing growth rates.
Crauder-4100237 QL November 6, 2014 15:45
This number was calculated in the eleventh century by the Persian philosopher and
mathematician al-Biruni.
There are about a million grains of wheat in a bushel, so the king’s promise forces
him to put more than 18 trillion bushels of wheat on the 64th square. At $3.50 per
bushel, that amount of wheat would be worth about 65 trillion dollars—about five
times the national debt of the United States on January 1, 2010. This illustrates how
all exponential growth eventually reaches gargantuan proportions.
To calculate the growth from year 40 to year 50, we put t = 40 and t = 50 into
the formula B = 800 × 1.07t :
164 CHAPTER 3
EXAMPLE 3.11 Finding a formula from the growth rate: Health care
U.S. health-care expenditures in 2010 reached 2.47 trillion dollars. It is difficult to
predict future health-care costs, especially with the advent of the Affordable Care
Act (also known as Obamacare). One prediction is that, in the near term, they will
grow by 6.5% each year. Assuming that this growth rate continues, find a formula
that gives health-care expenditures as a function of time. If this trend continues, what
will health-care expenditures be in 2030?
SOLUTION
Because health-care expenditures are increasing at a constant percentage rate, they
represent an exponential function of time. Let H denote the expenditures, in trillions
of dollars, t years after 2010. We find the formula using
H = Initial value × (1 + r )t
The initial value is the expenditures in 2010, namely 2.47 trillion dollars. Also, 6.5%
as a decimal is 0.065, and this is the value we use for r :
1 + r = 1 + 0.065 = 1.065
So the formula is
H = 2.47 × 1.065t trillion dollars
Now, 2030 is 20 years after 2010. Thus, to predict health-care expenditures in 2030,
we use t = 20 in the formula for H:
Expenditures in 2030 = 2.47 × 1.06520 trillion dollars
The result is about 8.7 trillion dollars.
In Figure 3.28, we have plotted the graph of health-care expenditures from Ex-
ample 3.11.
$9
Health care spending (trillions)
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Years since 2010
166 CHAPTER 3
continues, and find a formula that gives China’s defense spending (in billion yuan)
as a function of time from 2013 onward. What does this formula predict for China’s
defense spending in 2016?
The answer is provided at the end of this section.
Exponential decay
At the beginning of this section, we saw a news article about the storage of nuclear
waste. The decay of nuclear material is an example of exponential change that is
decreasing instead of increasing.
Let’s examine how this works out in a simple example. Suppose we have a bacte-
ria population that is dying off due to contaminants. Assume that initially the popu-
lation size is 2000 and that the population is declining by 25% each hour. This means
that 75% of the present population will still be there one hour from now. Because
the change is by a constant percentage, this is an example of an exponential function.
But because the population is declining, we call this process exponential decay.
Because the population after one hour is 75% of the current population, we find
the population for the next hour by multiplying the current population by 0.75:
Population next hour = 0.75 × Current population
This means that the population as an exponential function of time in hours has a base
of 0.75. We obtain the formula for exponential decay just as we did for exponential
growth. If N denotes the population size after t hours,
N = Initial value × Baset
N = 2000 × 0.75t
We used the formula for this population to obtain the graph in Figure 3.29. Its
shape is characteristic of graphs of exponential decay. Note that the graph is decreas-
ing. It declines rapidly at first but then the graph levels off, showing a much smaller
rate of decrease.
A key observation from this example of population decline is that if the rate of
decline is 25%, we find the base of 0.75 for the resulting exponential function using
1 − 0.25 = 0.75. This procedure is typical of exponential decay.
2200
2000
1800
1600
1400
Population
1200
1000
800
600
400
200
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time
Recall Summary 3.4, where the percentage increase per period was r as a decimal.
In that case, the base was 1 + r . If a quantity is declining at a constant percentage
rate of r as a decimal, we use the same idea. But now the rate r is replaced by −r ,
which means that the base is 1 − r . For exponential decay, the base is always less
than 1.
In summary, we see that the formula for the amount left after t periods is
A = Initial value × (1 − r )t
A = 70 × 0.51t
To find the amount of amoxicillin in the blood after five hours, we put t = 5 into
this formula:
The result is about 2.4 milligrams, which is less than the minimum of 10 milligrams.
Therefore, another injection will be needed before five hours.
Crauder-4100237 QL November 6, 2014 15:45
168 CHAPTER 3
80
70
60
Amount (milligrams)
50
40
30
20
10
0
0 0.5 1.0 1.5 2.0 2.5 3.0 3.,5 4.0 4.5 5.0
Figure 3.30 shows the amount of amoxicillin in the blood as a function of time
for Example 3.12. The graph shows that the level declines to 10 milligrams in about
3 hours.
170 CHAPTER 3
120
80
60
40
20
0
0 1 2 3 4 5
Time (half-lives)
amount remaining when time is measured in years, not in half-lives. For example,
© Ashley Cooper/Alamy
suppose we ask how much Pu–239 will remain after 36,000 years. The key here is to
determine how many half-lives 36,000 years represents. We know that one half-life
is 24,000 years and 36,000/24,000 = 1.5, so 36,000 years is 1.5 half-lives. Thus,
to find the amount remaining after 36,000 years, we use h = 1.5 in the preceding
formula:
1.5
1
Amount remaining after 36,000 years = 100 × grams
2
This is about 35.4 grams.
Plutonium decays in a fashion typical of all radioactive substances. The half-life
is different for each radioactive substance.
years has been about constant. As long as an organism is alive, it ingests air, and
the level of carbon-14 in the organism remains the same. When it dies, it no longer
absorbs carbon-14 from the air, and the carbon-14 in the organism decays, with a
half-life of 5770 years.
Now, when a bit of charcoal or bone from a prehistoric site is found, the percent of
carbon-14 remaining is measured. By comparing this with the percentage of carbon-
14 in a living tree today, scientists can estimate the time of death of the tree from
which the charcoal came.
Suppose a tree contained C0 grams of carbon-14 when it was cut down. What
percentage of the original amount of carbon-14 would we find if it was cut down
30,000 years ago?
SOLUTION
The amount C (in grams) remaining after h half-lives is given by
h
1
C = C0 ×
2
Now 5770 years is one half-life, so 30,000 years is 30,000/5770 or about 5.20 half-
lives. We use this value for h to calculate the amount remaining after 30,000 years:
5.2
1
Amount after 30,000 years = C0 × grams
2
This is about 0.027C0 grams. Thus, about 2.7% of the original amount of carbon-14
remains after 30,000 years.
In Example 3.14, we determined the percentage of carbon-14 from the age of the
sample. In the next section, we return to the topic of radiocarbon dating and see how
to estimate the age of the sample based on the percentage of carbon-14.
cartoonstock.com
172 CHAPTER 3
Questions of this sort may invite many correct responses. We present one possibility:
Constant percentage change may result in significant absolute decrease initially, but in
the long term the absolute decrease is negligible. If there are 100 items present subject
to constant 50% decrease, then over the first term the number is decreased by 50.
But over the second term the number is decreased by only 25. Because the decrease
is exponential, low levels of nuclear waste persist almost indefinitely. Depending on
the waste product in question, they may remain dangerous for hundreds or even
thousands of years. The graph referred to here is decreasing at a decreasing rate, and
it gives a visual display of this type of decay. In the case of linear decay, we can point
to a time when all of the substance will be gone. That never occurs with exponential
decay.
Further questions of this type may be found in the What Do You Think section of
exercises.
8. Linear or exponential. Water is pumped into a tank at a rate unredeemed in 2024, what will be their total value? Give your
of 10 gallons per minute. Determine which type of function de- answers in billions of dollars rounded to two decimal places.
scribes the volume of water in the tank: linear or exponential. 16. Headway. For traffic that flows on a highway, the headway
9. Heart health. One study measured the benefits of more sleep is the average time between vehicles. On four-lane highways,
for healthier hearts.7 Heart health was measured in terms of the probability P (as a decimal) that the headway is at least
calcification of the coronary arteries. The main conclusion of t seconds when there are 500 vehicles per hour traveling one
the study was, “One hour more of sleep decreased the estimated way is given by8
odds of calcification by 33%.” Rephrase this conclusion using P = 0.87t
the terminology of this section.
Under these circumstances, what is the probability that the
10. Large percent change. Chinese textile imports went up by
headway is at least 20 seconds? Round your answer to two
1000%. Were the imports
decimal places.
a. 10 times the old figure?
b. 11 times? c. 100 times? 17--18. Newton’s law of heating. Exercises 17 and 18 refer
to the following situation. If we put a cool object into a pre-
d. 110 times? e. none of these? heated oven, Newton’s law of heating tells us that the differ-
11. Balance. An investment grows according to the rule ence between the temperature of the oven and the temperature
of the object decreases exponentially with time. The percentage
Next month’s balance = 1.002 × Current balance rate of decrease depends on the material that is being heated.
Find the percentage increase each month, and explain why the Suppose a potato initially has a temperature of 75 degrees and
balance is an exponential function of time. the oven is preheated to 375 degrees. Use the formula
12--14. Internet domain hosts. With new technologies and D = 300 × 0.98t
new types of successful business ventures, it is common for
growth to be approximately exponential. An example is the where D is the temperature difference between the oven and
growth in the number of Internet domain hosts. One model is the potato, t is the time in minutes the potato has been in the
that the number of domain hosts grew according to the rule oven, and all temperatures are measured in degrees Fahrenheit.
Round your answers to the nearest degree.
Next year’s number = 1.43 × Current number
17. What is the temperature difference after 30 minutes?
This information is used in Exercises 12 through 14.
18. What is the temperature of the potato after 30 minutes?
12. Find the percentage increase each year, and explain why
the number of hosts is an exponential function of time. 19--20. Tsunami waves. Exercises 19 and 20 refer to the fol-
lowing information. Crescent City, California, is in an area
13. The number of domain hosts in 1995 was 8.2 million. Find that is subject to tsunamis. Its harbor was severely damaged
an exponential formula that gives the number of hosts in terms by the tsunami caused by the earthquake off the coast of Japan
of the time, in years since 1995. on March 11, 2011. The probability P (as a decimal) that no
14. What number of Internet domains does your formula in tsunami with waves of 15 feet or higher will strike Crescent
Exercise 13 give for 2005? Round your answer to the nearest City over a period of t years is given by the formula9
million.
P = 0.98t
15. German gold bonds. A certain gentleman acquired 54 gold
bonds issued by the Weimar Republic in 1934. The value of 19. If you move to Crescent City and stay there for 20 years,
each was 2500 troy ounces of gold. The bonds matured in what is the probability that you will witness no tsunami waves
1954, at which time the value of each was 2500 troy ounces of of 15 feet or higher? Round your answer to two decimal places.
gold at 1954 prices. The price of gold at that time was around 20. What is the percentage decrease of the probability for each
$350 per troy ounce. The bonds could not be redeemed because one-year increase in the time interval?
in 1954 there was no single German country—it had been di-
21. Population growth. Initially, a population is 500, and it
vided into two nations. But provisions of the bond stated that
grows by 2% each year. Explain why the population is an ex-
from 1954 onward the unredeemed bonds grew in value ac-
ponential function of time. Find a formula for the population
cording to the exponential formula
at any time, and determine the population size after 5 years.
Value = Initial value × 1.00019365t 22. World population. The world population in 2013 was
where t is the number of years since 1954. What was the to- estimated to be 7.1 billion people, and it is increasing by
tal value of this man’s 54 bonds in 2012? If the bonds remain about 1.1% per year. Assume that this percentage growth rate
7 C. R. King et al., “Short Sleep Duration and Incident Coronary Artery Calcification,” Journal of the
American Medical Association 300 (2008), 2859–2866.
8 See Institute of Traffic Engineers, Transportation and Traffic Engineering Handbook, ed. John E.
Baerwald (Englewood Cliffs, NJ: Prentice-Hall 1976), p. 102.
9 Taken from data provided on p. 294 of Robert L. Wiegel, ed., Earthquake Engineering
(Englewood Cliffs, NJ: Prentice-Hall, 1970).
Crauder-4100237 QL November 6, 2014 15:45
174 CHAPTER 3
remains constant through 2025.10 Explain why the population 29. What would be the thickness in kilometers if you were
is an exponential function of time. What would you expect the physically able to fold it 50 times? Note: For comparison,
world population to be in 2025? the distance from Earth to the Sun is about 149,600,000 kilo-
23. United States population. According to the U.S. Census Bu- meters.
reau, the population of the United States in 2010 was 308.75 30--31. Inflation. The rate of inflation measures the percent-
million people. The rate of growth in population was 0.57% age increase in the price of consumer goods. The rate of infla-
per year. Assume that this rate of growth remains the same tion in the year 2012 was 3% per year. To get a sense of what
through 2025. Explain why the population is an exponential this rate would mean in the long run, let’s suppose that it per-
function of time. What would you predict the U.S. population sists through 2032. Exercises 30 and 31 refer to this inflation.
to be in 2025? 30. What would be the cost in 2013 of an item that costs $100
24. Epidemics. In many epidemics, the cumulative number of in 2012?
cases grows exponentially, at least over a limited time. A re- 31. What would be the cost in 2032 of an item that costs $100
cent example is the spread of the influenza A (H1N1) virus, in 2012?
commonly called the swine flu, which affected Mexico and the
United States especially. On April 24, 2009, there were seven 32--33. Deflation. From 1929 through the early 1930s, the
cumulative cases of swine flu in the United States reported to prices of consumer goods actually decreased. Economists call
the World Health Organization. Over the next two weeks, the this phenomenon deflation. The rate of deflation during this
cumulative number of cases in the United States increased by period was about 7% per year, meaning that prices decreased
about 40% each day. If this growth rate had continued, what by 7% per year. To get a sense of what this rate would mean in
would have been the cumulative number of cases in the United the long run, let’s suppose that this rate of deflation persisted
States reported 30 days after April 24? over a period of 20 years. This is the situation for Exercises 32
and 33.
25. Library costs. The library at a certain university reported
that journal prices had increased by 150% over a period of 10 32. What would be the cost after one year of an item that costs
years. The report concluded that this represented a price in- $100 initially?
crease of 15% each year. If journal prices had indeed increased 33. What would be the cost after 20 years of an item that costs
by 15% each year, what percentage increase would that give $100 initially?
over 10 years? Round your answer as a percentage to the near- 34--36. Blogosphere growth. The number of blogs (or web-
est whole number. logs) grew rapidly for several years. According to one report,
26. A nuclear waste site. Cesium-137 is a particularly danger- there were about 2 million blogs in March 2004, after which
ous by-product of nuclear reactors. It has a half-life of 30 years. the number of blogs doubled approximately every six months.
It can be readily absorbed into the food chain and is one of the Exercises 34 through 36 refer to these facts. Note: Growth ta-
materials that would be stored in the proposed waste site at pered off after 2007, and at the end of 2011 there were about
Yucca Mountain, Nevada. Suppose we place 3000 grams of 175 million blogs. Complete Exercises 34 to 36 in order using
cesium-137 in a nuclear waste site. this information.
a. How much cesium-137 will be present after 30 years, or
34. Let N denote the number (in millions) of blogs and d the
one half-life? After 60 years, or two half-lives?
number of doubling periods after March 2004. Use a formula
b. Find an exponential formula that gives the amount of to express N as an exponential function of d.
cesium-137 remaining in the site after h half-lives.
35. How many blogs does the formula derived in Exercise 34
c. How many half-lives of cesium-137 is 100 years? Round project for September 2006?
your answer to two decimal places.
36. The Nielsen Company11 says that the actual number of
d. Use your answer to part c to determine how much blogs in October 2006 was about 35,800,000. What conclu-
cesium-137 will be present after 100 years. Round your an- sions do you draw from this about the reliability of projections
swer to the nearest whole number. over time?
27--29. Folding paper. Each time you fold an ordinary sheet 37. Retirement options. This exercise illustrates just how fast
of paper in half, you double the thickness. Exercises 27 through exponential functions grow in the long term. Suppose you start
29 refer to folding this paper. work for a company at age 25. You are offered two rather un-
27. A single sheet of 20-pound bond paper is about a tenth of a likely retirement options:
millimeter thick. How would you come up with this estimate? Retirement option 1: When you retire, you will receive
(Hint: Paper often come in reams of 500 pages.) $20,000 for each year of service.
28. Find an exponential formula that gives the thickness of the Retirement option 2: When you start work, the company
sheet of paper in Exercise 27 after f folds in: deposits $2500 into a savings account that pays a monthly
a. millimeters b. kilometers rate of 1.5%. When you retire, the account will be closed
and the balance given to you.
10 In fact, for almost all of human history, the human population growth rate has been slowly increasing. However, in 1962
the rate reached a peak of 2.19% and declined steadily to its 2010 level of 1.11%.
11 http://www.nielsen.com/us/en/newswire/2012/buzz-in-the-blogosphere-millions-more-bloggers-and-blog-readers.html
Crauder-4100237 QL November 6, 2014 15:45
How much will you have under the second plan at age 55? At 45. Use your answer to Exercise 44 to determine what percent-
age 65? Which retirement option is better if you plan to retire age of the original amount of carbon-14 remains after 12,000
at age 55? Which if you plan to retire at age 65? years. Round your answer as a percentage to the nearest whole
38--40. Cleaning dirty water. Exercises 38 through 40 refer number.
to the following information. Many physical processes are ex- 46. Research project. The method of radiocarbon dating de-
ponential in nature. A typical way of cleaning a tank of con- scribed in Example 3.14 is an example of a general technique
taminated water is to run in clean water at a constant rate, stir, known as radiometric dating. Investigate this general tech-
and let the mixture run out at the same rate. Suppose there are nique. What radioactive substances are used, and what are their
initially 100 pounds of a contaminant in a large tank of water. half-lives? What are the limitations of this technique?
Assume that the cleaning method described here removes 10% 47. Growth of social networks. Consider the following article
of the remaining contaminant each hour. Round your answers from the Los Angeles Times.
to one decimal place. Complete Exercises 38 to 40 in order us-
ing this information.
38. Find an exponential formula that gives the number of
pounds of contaminant left in the tank after t hours. IN THE NEWS 3.4 Search
12 https://www.cia.gov/library/publications/the-world-factbook/
Crauder-4100237 QL November 6, 2014 15:45
176 CHAPTER 3
W-2. Paypal. The mathematical notion of exponential growth refers to a precise type
of growth that happens to be (eventually) very rapid. But the phrase is often applied
to any fast-growing phenomenon. The headline at venturebeat.com/2012/01/10/
paypals-mobile-payments-4b-2011/ says this: “Paypal’s mobile payments growing
exponentially, reached $4B in 2011.” Do you think this headline is really describing
exponential growth, or is it just using a figure of speech? Explain your thinking.
W-3. Is it exponential? You see a news report about taxes, inflation, or other eco-
nomic issues and wonder whether you are seeing exponential growth. How would
you make a judgment about this question? What factors might lead you to believe
that you are seeing exponential growth, and what might lead you to believe that you
are not seeing exponential growth?
W-5. Shape of exponential growth. The graph of exponential growth gets steeper
as we move to the right. What is it about the definition of exponential growth that
guarantees this shape?
W-6. Base and percentage growth. If a population grows by 5% each year, then
1.05 is the base for the exponential formula expressing the population as a function
of time in years. Often beginning students ask, “Where does the ‘1’ come from in the
expression 1.05?” How would you answer that question in terms understandable to
beginning students?
Exercise 21 provides additional information relevant to this topic.
W-7. Exponential decay. Many phenomena, radioactivity for example, display ex-
ponential decay with time. For such a phenomenon, would you expect the change to
be greater from the fifth to the sixth year or from the fifty-ninth to the sixtieth year?
Explain your answer.
W-8. Examples of exponential decay. The decay of nuclear material is a prime ex-
ample of exponential decay. Use the Internet to find other examples and report on
them. Suggestion: You might look at cleanup of sites contaminated by dangerous
chemicals.
Exercise 26 provides additional information relevant to this topic.
W-9. A long-term investment. You have the option of making a long-term invest-
ment in either of two companies. Company A shows rapid linear growth. Company B
shows moderate exponential growth. Which do you think would be the better invest-
ment? What additional information might you need in order to make your decision?
Exercise 37 provides additional information relevant to this topic.
Crauder-4100237 QL November 6, 2014 15:45
and in solving Home TV & Video U. S. World Politics Entertainment Tech Travel
exponential
equations. Record Number of Oklahoma Tremors Raises Possibil-
ity of Damaging Earthquakes
Updated USGS-Oklahoma Geological Survey Joint Statement on
Oklahoma Earthquakes
May 2, 2014
178 CHAPTER 3
magnitude of earthquakes referred to in the preceding article. Logarithms are also the
basis for measuring the loudness of sound, acidity of substances, and brightness of
stars. Logarithms are closely related to exponential functions, as we will see shortly.
In addition, the logarithm is the mathematical tool needed to solve certain expo-
nential equations. Solving such equations is important for the applications that were
discussed in the preceding section, such as the growth of investments and populations
and finding the age of archaeological artifacts.
What is a logarithm?
We have noted that population growth is often modeled using an exponential func-
tion of time. But we could turn the model around and think of the time as a function
of the population. This function would tell us, for example, when the population
reaches 1000 and when it reaches 1,000,000. This kind of function, which reverses
the effect of an exponential function, is called a logarithmic function. The corre-
sponding graphs for a population of bacteria that doubles each hour are shown in
Figures 3.32 and 3.33. As expected, the exponential graph shown in Figure 3.32
gets steeper as we move to the right. For example, over the first hour the population
grows by 2 (from 2 to 4), but over the last hour shown the population grows by 8
(from 8 to 16). If we think of time as a function of population, this says that growth
in population from 2 to 4 takes one hour, and growth in population from 8 to 16
also takes one hour. This observation explains the growth rate of the logarithmic
function shown in Figure 3.33: Growth is rapid at first but slows as we move to the
right.
16 3
14
12
2
Time (hours)
10
Population
8
6
1
4
2
0 0
0 1 2 3 0 2 4 6 8 10 12 14 16
Time (hours) Population
We have said that the logarithm reverses the action of an exponential function.
Here is a more precise definition: The common logarithm13 of a number, often short-
ened to just “the log,” is the exponent of 10 that gives that number. For example,
the log of 100 is 2 because 100 = 102 . There are other kinds of logarithms besides
the common logarithm, but in this section, we will use the term log or logarithm to
mean the common logarithm.
Because the logarithmic equation log x = t means exactly the same thing as the
exponential equation x = 10t , we will often write both forms side by side to remind
us of this fact. Thus, for example,
log 10 = 1 because 101 = 10; 1 is the exponent of 10 that gives 10
log 100 = 2 because 102 = 100; 2 is the exponent of 10 that gives 100
log 1000 = 3 because 103 = 1000; 3 is the exponent of 10 that gives 1000
1 1 1
log = −1 because 10−1 = ; −1 is the exponent of 10 that gives
10 10 10
For quantities that are not whole-number powers of 10, we use a scientific calcu-
lator or computer to find the logarithm.
14 There are other methods for measuring the magnitude of earthquakes. See Exercise 40 for one alternative method.
Crauder-4100237 QL November 6, 2014 15:45
180 CHAPTER 3
140 9
Magnitude
80 5
60 4
3
40
2
20
1
0 0
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
the resulting tsunami killed about 230,000 people. The earthquake that struck Haiti
on January 12, 2010, was as devastating in terms of loss of life but much smaller in
size (see Exercise 39).
The following table provides more information on interpreting the Richter scale
in practical terms. Note that a small increase in the magnitude yields dramatic effects.
That is one result of using a compressed scale.
Estimated number of
Richter magnitude Effects earthquakes per year
2.5 or less Generally not felt 900,000
2.6 to 5.4 Often felt, only minor damage 30,000
5.5 to 6.0 Slight damage to buildings 500
6.1 to 6.9 May cause lots of damage in 100
populated areas
7.0 to 7.9 Major quake, serious damage 20
8.0 to 8.9 Can totally destroy communities 1
9.0 or higher Rare, great quake. May cause One every 20 years
major damage to areas as much
as 1000 miles away
182 CHAPTER 3
which has a base of 10, we see that a 1-point increase in magnitude corresponds to
multiplying the relative intensity by 10. More generally, increasing the magnitude by
t points multiplies relative intensity by 10t . For example, a magnitude 6.3 quake is
10 times as intense as a magnitude 5.3 quake, and a magnitude 7.3 quake is 102 =
100 times as intense as a magnitude 5.3 quake. This shows why, as noted in the
preceding table, a relatively small increase on the Richter scale corresponds to a large
increase in the relative intensity of a quake.
The graph in Figure 3.36 shows the decibel reading in terms of the relative inten-
sity. This graph shows how our brains actually perceive loudness. This means that a
unit increase in the relative intensity has more effect on our perception of soft sounds
than it does on our perception of louder sounds. Our brains can distinguish small
differences in soft sounds but find the same difference almost indistinguishable for
16 The relative intensity of a sound is a measure of the sound power per unit area as compared to the power per unit
area of a barely audible sound. Physicists use 10−12 watts per square meter as this base intensity.
17 The loudness of a sound decreases as we move away from the source. See Exercise 22 for a discussion.
Crauder-4100237 QL November 6, 2014 15:45
184 CHAPTER 3
© Elvinstar | Dreamstime.com
25
20
Decibel reading
15
10
0
0 10 20 30 40 50 60 70 80 90 100
Relative intensity
FIGURE 3.36 Left: Decibel reading in terms of relative intensity. Right: A whisper is about 20 decibels.
louder sounds. That explains why doubling the relative intensity does not double our
perception of a sound.
Because relative intensity is an exponential function of decibels, we can determine
what happens to relative intensity when decibels increase. The base of the exponential
equation
Relative intensity = 1.26Decibel
is 1.26. Hence, an increase of one decibel multiplies relative intensity by 1.26, and
an increase of t decibels multiplies relative intensity by 1.26t .
Now we see in a practical example how doubling the intensity affects the loudness.
The conclusion of this example may seem counter intuitive: Adding a second
60-decibel speaker does not double the decibel level, which is the perceived loud-
ness. Far from doubling loudness, adding another speaker only increases the decibel
reading from 60 to 63. This interesting phenomenon is roughly true of all five hu-
man senses, including sight, and is known as Fechner’s law.18 This law states that
psychological sensation is a logarithmic function of physical stimulus.
Other common logarithmic phenomena, such as apparent brightness of stars and
acidity of solutions, will be introduced in the exercises.
186 CHAPTER 3
Properties of Logarithms
Now let’s return to the problem of population growth to illustrate how logarithms
are used to solve an equation.
Suppose we have a population that is initially 500 and grows at a rate of 0.5%
per month. How long will it take for the population to reach 800?
Because the population is growing at a constant percentage rate, it is an exponen-
tial function. The monthly percentage growth rate as a decimal is r = 0.005. Hence,
the base of the exponential function is 1 + r = 1.005. Therefore, the population size
N after t months is given by
N = Initial value × (1 + r )t
= 500 × 1.005t
To find out when the population N reaches 800, we need to solve the equation
To complete the solution, we need to find the unknown exponent t. This is where
logarithms play a crucial role. We apply the logarithm function to both sides:
Therefore,
log 1.6
t=
log 1.005
B = Initial value × (1 + r )t
A stock certificate.
B = 5000 × 1.1t
20,000 5000
= × 1.1t
5000 5000
4 = 1.1t
log A log 4
t= =
log B log 1.1
188 CHAPTER 3
and always over the same time span. Suppose we want to find how long it takes the
exponential function
Initial value × Baset
to double. For this, we need to solve the equation
2 × Initial value = Initial value × Baset
Dividing by the initial value gives
2 = Baset
We solve this equation using Summary 3.9 with A = 2 and B = Base:
log 2
Doubling time = t =
log (Base)
A similar formula applies for any multiple of the initial value.
Suppose the charcoal from an ancient campfire is found to contain only one-third
of the carbon-14 of a living tree. How long ago did the tree that was the source of
the charcoal die? Give the answer first in half-lives and then in years rounded to the
nearest hundred.
SOLUTION
The amount of carbon-14 present is an exponential function of time in half-lives,
and the base is 1/2. Because we want to know when only one-third of the original
amount is left, we use K = 1/3 in the formula from Summary 3.10:
log K
Time to multiply by 1/3 is t =
log(Base)
log(1/3)
=
log(1/2)
This is about 1.58, so the time required is 1.58 half-lives. Because each half-life is
5770 years, 1.58 half-lives is
1.58 × 5770 = 9116.6 years
Thus, the tree died about 9100 years ago.
190 CHAPTER 3
sound. It seems to be the case that the human brain performs a logarithmic adjust-
ment for light and sound. Thus, doubled light intensity does not appear to us to be
doubly bright. Scales such as the logarithmic decibel scale accurately reflect how we
perceive loudness.
Further questions of this type may be found in the What Do You Think section of
exercises.
California. How do the relative intensities of the two quakes from a sound source such as a vacuum cleaner multiplies the
compare? relative intensity of the sound by 1/4. If the sound from a vac-
14. New Madrid quake. On December 16, 1811, an earth- uum cleaner has a reading of 80 decibels at a distance of 3 feet,
quake occurred near New Madrid, Missouri, that temporar- what is the decibel reading at a distance of 6 feet?
ily reversed the course of the Mississippi River. The quake 23. Apparent brightness. The apparent brightness on Earth of
had a magnitude of 8.8. On October 17, 1989, a calamitous a star is measured on the magnitude scale.19 The apparent mag-
quake measuring 7.1 on the Richter scale occurred in San Fran- nitude m of a star is defined by
cisco during a World Series baseball game on live TV. How
many times as intense as the San Francisco quake was the New m = 2.5 log I
Madrid quake?
where I is the relative intensity. The accompanying exponential
15--17. Energy of earthquakes. The energy released by an
formula is
earthquake is related to the magnitude by an exponential func-
tion. The formula is I = 2.51m
Energy = 25,000 × 31.6Magnitude To find the relative intensity, we divide the intensity of light
from the star Vega by the intensity of light from the star we
The unit of energy in this equation is a joule. One joule is are studying, with both intensities measured on Earth. Thus,
approximately the energy expended in lifting 3/4 of a pound apparent magnitude is a logarithmic function of relative in-
1 foot. Exercises 15 through 17 refer to this formula. tensity. The star Fomalhaut has a relative intensity of 2.9, in
15. On April 19, 1906, San Francisco was devastated by a 7.9- the sense that on Earth light from Vega is 2.9 times as intense
magnitude earthquake. How much energy was released by the as light from Fomalhaut. What is the apparent magnitude of
San Francisco earthquake? Fomalhaut? Round your answer to two decimal places. Note:
The magnitude scale is perhaps the reverse of what you might
16. Use the fact that energy is an exponential function of mag- think. The higher the magnitude, the dimmer the star. Some
nitude to determine how a 1-unit increase in magnitude affects very bright stars have a negative apparent magnitude.
the energy released by an earthquake.
24. Finding relative brightness. This is a continuation of Ex-
17. We noted earlier that on December 26, 2004, an earth- ercise 23. A star has an apparent magnitude of 4. Use the ex-
quake of magnitude 9.1 struck the Indian Ocean near Suma- ponential formula in Exercise 23 to find its relative intensity.
tra, Indonesia. How did the energy released by the Indonesian Round your answer to one decimal place.
quake compare with that of the San Francisco quake of Exer-
cise 15? 25. Acid rain. Acid rain can do serious damage to the environ-
ment as well as human health. Acidity is measured according to
18. Jet engines. A jet engine close up produces sound at the pH scale. (For more information on the pH scale, see Exer-
155 decibels. What is the decibel reading of a pair of nearby cise 26.) Pure water has a pH of 7, which is considered neutral.
jet engines? In light of the table on page 183, what does your A pH less than 7 indicates an acidic solution, and a pH higher
answer suggest about those who work on the tarmac to support than 7 indicates a basic solution. (Lye and baking soda solu-
jet aircraft? tions are very basic; vinegar and lemon juice are acidic.) An
19. Stereo speakers. A speaker is playing music at 80 decibels. increase of 1 on the pH scale causes acidity to be multiplied by
A second speaker playing the same music at the same decibel a factor of 1/10. Equivalently, a decrease of 1 on the pH scale
reading is placed beside the first. What is the decibel reading of causes acidity to be multiplied by a factor of 10. If rainfall has
the pair of speakers? a pH of 4, how many times as acidic as pure water is it? (It is
20. More on stereo speakers. Consider the lone stereo speaker worth noting that most acid rain in the United States has a pH
in Exercise 19. Someone has the none-too-clever idea of dou- of about 4.3.)
bling the decibel level. Would 100 of these speakers do the job? 26. More on pH. Acidity is measured on the pH scale, as
How about 1000? You should see a pattern developing. If you mentioned in Exercise 25. Here, pH stands for “potential of
don’t see it yet, try 10,000 speakers. How many speakers are Hydrogen.” Here is how the scale works.
required to double the decibel level? The acidity of a solution is determined by the concentration
21. Adding two speakers. Suppose a speaker is playing music H of hydrogen ions.20 The formula is
at 70 decibels. If two more identical speakers (playing the same
pH = − log H
music) are placed beside the first, what is the resulting decibel
level of the sound?
The accompanying exponential formula is
22. Decibels and distance. The loudness of a sound decreases
as the distance from the source increases. Doubling the distance H = 0.1pH
19 The notion of magnitude for stars goes back to the ancient Greeks, who grouped the stars into six magnitude classes, with the brightest
stars being of the first magnitude. The use of the term “magnitude” for stars motivated Richter to use the same term in his scale for earthquakes.
20 The concentration is measured in moles per liter of the solution.
Crauder-4100237 QL November 6, 2014 15:45
192 CHAPTER 3
Lower pH values indicate a more acidic solution. Normal rain- 36. How long will it be before your investment decreases to
fall has a pH of 5.6. Rain in the eastern United States often has half its original value? Round your answer to one decimal
a pH level of 3.8. How many times as acidic as normal rain is place.
this?
37--38. Economic growth. The CIA World Factbook21 esti-
27. Saving for a computer. You have $200 and wish to buy a mates the economic performance of countries. One such figure
computer. You find an investment that increases by 0.6% each is the gross domestic product (GDP), which is the annual mar-
month, and you put your $200 into the account. When will the ket value of the goods and services produced in that country.
account enable you to purchase a computer costing $500? Exercises 37 and 38 refer to these figures.
28--29. Spent fuel rods. The half-life of cesium-137 is 30 37. In 2012, the GDP of the United States was about 15.68 tril-
years. Suppose that we start with 50 grams of cesium-137 in lion dollars. It is predicted to grow by about 2.2% each year.
a storage pool. Exercises 28 and 29 refer to this cesium. What is the projected value of the GDP in 2030 (to the nearest
28. Find a formula that gives the amount C of cesium-137 re- trillion dollars)?
maining after h half-lives. 38. In 2012, the GDP of China was about 8.23 trillion dol-
29. How many half-lives will it take for there to be 10 grams lars. It is predicted to grow by about 7.8% each year. What is
of cesium-137 in the storage pool? (Round your answer to the projected value of the GDP in 2030 (to the nearest trillion
two decimal places.) How many years is that? dollars)? Use this prediction and your answer to Exercise 37 to
determine whether China will have a larger economy than the
30. Cobalt-60. Cobalt-60 is subject to radioactive decay, and
United States by the year 2030. If so, determine in what year the
each year the amount present is reduced by 12.3%.
economy of China will reach the level projected in Exercise 37.
a. The amount of cobalt-60 present is an exponential func- (Round up to the next year.)
tion of time in years. What is the base of this exponential 39. Earthquake in Haiti. The earthquake that devastated Haiti
function? on January 12, 2010, had a magnitude of 7.0 and killed tens of
b. What is the half-life of cobalt-60? Round your answer thousands of people. Six weeks later a quake of magnitude 8.8
to one decimal place. struck southern Chile and killed about 500 people. Investigate
why the much smaller earthquake in Haiti was so much more
31--33. Inflation. Suppose that inflation is 2.5% per year. This
devastating.
means that the cost of an item increases by 2.5% each year.
Suppose that a jacket cost $100 in the year 2011. Exercises 31 40. Beyond Richter. There are other methods than the Richter
through 33 refer to this situation. scale for measuring the magnitude of earthquakes. One alterna-
tive is the moment magnitude scale. Investigate how this scale
31. Find a formula that gives the cost C in dollars of the jacket is defined. Does the definition involve a logarithm?
after t years.
Exercises 41 through 46 are suitable for group work.
32. How long will it take for the jacket to cost $200? Round
your answer to the nearest year. 41. Yahoo stock graphs. If you go to Yahoo’s finance site at
http://finance.yahoo.com and look up a stock, you will see a
33. How long will it take for the jacket to cost $400? Sugges- chart, or graph, showing the price plotted over time. Among
tion: Once you have completed Exercise 32, you do not need the chart settings is the option to set the chart scale to either
a calculator for this exercise, because you have calculated the logarithmic or linear. (Google has the same feature.) Selecting
doubling time for this exponential function. the logarithmic option yields a plot of the logarithm of the
34. World population. If the per capita growth rate of the stock price (note the scale on the vertical axis). Try this with
world population continues to be what it was in the year 2010, Starbucks’ stock (symbol SBUX). Print a copy of the graphs
the world population t years after July 1, 2010, will be of SBUX over a five-year period using the Linear option and
using the Log option. Compare the graphs and discuss the dif-
6.852 × 1.0111t billion ferences. Pick out a couple of other stocks and do the same
thing with them.
According to this formula, when will the world population
42--44. The logarithmic Dow. For graphs that have sharp
reach 8 billion?
jumps, applying the logarithm can dampen the jumps and give
35--36. A poor investment. Suppose you make an investment a clearer visual presentation of the data. As mentioned in Ex-
of $1000 that you are not allowed to cash in for 10 years. Un- ercise 41, Yahoo and Google finance sites both allow this.
fortunately, the value of the investment decreases by 10% per Figure 3.37 shows the logarithmic Dow from 1897 through
year. Exercises 35 and 36 refer to this investment. 2013.22 Exercises 42 through 44 refer to the logarithmic
Dow.
35. How much money will be left after the end of the 10-year
term?
21 https://www.cia.gov/library/publications/the-world-factbook/
22 © G. William Schwert, 2000–2013. http//schwert.ssb.rochester.edu/dj.pdf
Crauder-4100237 QL November 6, 2014 15:45
W-3. Fechner’s law. In this section we stated Fechner’s law, which says that psycho-
logical sensation is a logarithmic function of physical stimulus. For example, the eye
perceives brightness as a logarithmic function of the intensity of light. Draw a graph
of perceived brightness as a function of the intensity of light. Explain how the shape
of the graph reflects the underlying logarithm.
Exercise 23 provides additional information relevant to this topic.
W-4. Logarithmic and exponential functions. Explain the relationship between log-
arithmic and exponential functions.
Exercise 23 provides additional information relevant to this topic.
W-5. Spent fuel rods. There is currently a great deal of controversy regarding the
disposal of spent nuclear fuel rods. Why does this issue cause such controversy?
Would you be willing to have a nuclear disposal site in your state?
Exercise 29 provides additional information relevant to this topic.
194 CHAPTER 3
island, taking a huge toll in terms of both damage to physical structures and loss of
life. Do some research to determine the factors, in addition to the magnitude, that
contribute to damage from earthquakes. Report on some moderate quakes, like the
Haiti quake, that resulted in catastrophic damage.
Exercise 39 provides additional information relevant to this topic.
W-7. Measuring earthquakes. Over the years scientists have used a number of scales,
not just the Richter scale, to measure earthquakes. Investigate some of the other scales
that have been used. Which are in current use?
Exercise 40 provides additional information relevant to this topic.
W-8. The logarithmic Dow. The Dow Jones Industrial Average from 1897 through
2013 is graphed on a logarithmic scale in Figure 3.37. Use the Internet to find a graph
of the Dow over the same period on an uncompressed (regular) scale. Compare the
two graphs. For example, which graph shows the trends more clearly, and which
shows significant economic events more clearly?
Exercise 42 provides additional information relevant to this topic.
W-9. A mathematical question. In this section we have restricted our attention to
the common logarithm, but there are in fact many other logarithms (just as there are
many exponential functions). Mathematicians prefer the natural logarithm. On your
calculator it is likely to be displayed as ln. Investigate the natural logarithm and why
mathematicians prefer it.
M ore solar energy reaches the earth in one hour than the combined world-
wide consumption of energy by human activities in one year. Even though
the sun supplies far more energy than we can possibly use, the challenge is mak-
ing solar energy cost-competitive with fossil fuels and other alternative energy
sources.
Today’s concentrating solar power (CSP) systems use mirrors (sometimes
100,000 or more) and sophisticated tracking systems to reflect and concentrate
sunlight, which is then converted to heat to generate electricity.
[The parabolic trough] type of linear concentrator is the most mature CSP
technology, with over 500 megawatts operating worldwide. Rows of reflectors
concentrate sunlight onto tubes that contain synthetic oil.
Because solar energy is an inconsistent source of power, the best way to make
it more competitive is to increase its storage capacity—this would increase a CSP
plant’s flexibility in meeting utility power demands. It would also make CSP more
competitive with photovoltaic technologies, which have become less expensive
in recent years.
Crauder-4100237 QL November 6, 2014 15:45
This article refers to parabolic trough collectors, which use curved mirrors to
collect the energy of the sun in order to produce electricity (see the photo of the line
focus water heater). These troughs have a parabolic shape because mirrors with that
shape focus incoming sunlight at a single point. Parabolas are graphs of quadratic
functions, and we start by looking at those functions.
196 CHAPTER 3
Some quadratics are difficult to factor. For these, we need the quadratic formula.
A derivation of this formula is given at the end of this section in Algebraic Spot-
light 3.1.
Suppose, for example, that we wish to solve the quadratic equation x2 − 3x− 2 =
0. The factors are not apparent, so we use the quadratic formula with a, b, and c as
indicated:
a b c
1 x2 −3 x −2 = 0
b b2 4×a×c
− (−3) ± (−3)2 − 4 × 1 × (−2)
x=
2×a
2×1
3± 9+8
=
2
3 ± 17
=
2
It is worth emphasizing that the ± symbol means that there are two solutions:
3 + 17 3 − 17
x= or x=
2 2
These are the exact solutions. Using a calculator, we can see that they are about
x = 3.56 or x = −0.56.
a. Solve x2 + 4x − 5 = 0 by factoring.
b. Use the quadratic formula to solve x2 + 2x − 2 = 0.
SOLUTION
a. The first step is to factor:
x2 + 4x − 5 = (x + 5)(x − 1)
Rewriting the equation gives
(x + 5)(x − 1) = 0
Now we set each factor equal to 0:
x+5 = 0 gives the solution x = −5
x−1 = 0 gives the solution x=1
Hence, the solution is x = 1 or x = −5.
Crauder-4100237 QL November 6, 2014 15:45
x2 + 2x − 2 = 0
−2 ± 22 − 4 × 1 × (−2)
x=
2×1
−2 ± 12
=
2
−2 ± 4 × 3
=
2
−2 ± 2 3
=
2
= −1 ± 3
Hence, the solution is x = −1 + 3 or x = −1 − 3.
There are many practical phenomena that are appropriately modeled by quadratic
functions, and solving quadratic equations is an important skill for analysis of such
models. This fact is illustrated in the following two examples.
D = 16t 2 + 32t
If the rock is thrown from the top of a tower that is 128 feet tall, how long does
it take for the rock to strike the ground?
SOLUTION
We need to know how long it takes for the rock to travel 128 feet. That is, we need
to solve the equation 16t 2 + 32t = 128 or 16t 2 + 32t − 128 = 0. This equation is
easier to solve if we first divide both sides by 16 to get t 2 + 2t − 8 = 0. Now we
factor the left-hand side to obtain
t 2 + 2t − 8 = (t + 4)(t − 2)
(t + 4)(t − 2) = 0
We discard the negative solution because it makes no sense in this context. We con-
clude that the rock hits the ground after 2 seconds.
Crauder-4100237 QL November 6, 2014 15:45
198 CHAPTER 3
G = 1 + N − 0.2N2
The population level is stable when it is neither increasing nor decreasing—that is,
when the growth rate is 0. At what level is the population stable?
SOLUTION
We need to find when G = 0, which says 1 + N − 0.2N2 = 0. This is a quadratic
equation, but it is not written in the standard form. We rearrange it to get −0.2N2 +
N + 1 = 0. Now we use the quadratic formula to solve the equation:
−0.2N2 + N + 1 = 0
−1 ± 12 − 4 × (−0.2) × 1
N=
2 × (−0.2)
−1 ± 1.8
N=
−0.4
N = −0.85 or 5.85
Here, we have rounded to two decimal places. The negative solution makes no sense
for a population size, so we discard it. We conclude that the population level is stable
when there are 5.85 thousand individuals present.
G = 2 + 2N − 0.3N2
Once again, assume that G is measured in thousands of individuals per year and N
is measured in thousands.
The answer is provided at the end of this section.
Parabolas
The graph of a quadratic function has a special shape known as a parabola. Typical
parabolas are shown in Figures 3.38 and 3.39. Note that for y = ax2 + bx + c, the
parabola opens upward when a is positive and downward when a is negative.
A key bit of information about a parabola is the location of its tip, called the
vertex.
Crauder-4100237 QL November 6, 2014 15:45
We can use the symmetry of a parabola to locate the x-value of the vertex. We
start with a typical upward-opening parabola, the top graph in Figure 3.40. It is the
graph of ax2 + bx + c with a positive. If we remove the constant term c, the graph is
shifted up or down (depending on the sign of c), but the x-value of the vertex does
not change. This fact is illustrated in Figure 3.40.
Crauder-4100237 QL November 6, 2014 15:45
200 CHAPTER 3
ax2 + bx + c
Halfway ax2 + bx
between
x = –b/2a
x=0
x = –b/a
Therefore, we may as well consider the graph of ax2 + bx. Next we find where
this graph crosses the horizontal axis. The solutions of ax2 + bx = 0 are easy to find
by factoring:
ax2 + bx = 0
x(ax + b) = 0
x = 0 or ax + b = 0
b
x = 0 or x = −
a
Finally, the symmetry of the parabola means that the x-value of the vertex is
halfway between the two
solutions,
x = 0 and x = −b/a. Therefore, the x-value of
1 b b
the vertex is x = × − = − . The same formula holds if the parabola opens
2 a 2a
downward.
To find the y-value, we put the x-value of 4 in the function 3x2 − 24x + 10:
y-value = 3 × 42 − 24 × 4 + 10 = −38
We can use our knowledge about the vertex of a parabola to solve some classic
max-min problems.
y = length
= 100 – 2x
x = width x = width
Barn
202 CHAPTER 3
If the discriminant is 0, then the quadratic equation gives exactly one solution
−b
x=
2a
In this instance the graph touches the horizontal exactly once, at its vertex.
When the discriminant is positive, the quadratic formula gives two solutions, so
that the graph crosses the horizontal axis twice.
Further questions of this type may be found in the What Do You Think section of
exercises.
204 CHAPTER 3
10--13. A cannonball. If air resistance is ignored, a projectile 18. A garden. A farmer wants to enclose a rectangular garden
such as a rifle bullet or a cannonball follows a parabolic path. with an area of 1000 square feet. She wants the plot to be di-
Suppose a cannonball fired from the origin follows the graph vided into 3 sections as shown in Figure 3.44. If she uses 200
feet of fencing, what should the dimensions of the garden be?
y = x − 0.0005x2 Is there more than one answer? Round your results in feet to
one decimal place. Suggestion: Refer to Figure 3.44, where x is
where both x and y are measured in feet. This information is
in feet and the formula for the length uses the fact that the area
used in Exercises 10 through 13.
is 1000 square feet. Use the dimensions shown in the figure to
10. The 20-foot-high wall of the enemy fort is located 1970 calculate the total amount of fence used, and set that equal to
feet downrange. Will the cannonball clear the wall? 200 feet.
11. How far downrange will the cannonball hit the ground?
12. How far downrange will the cannonball reach its maxi-
mum height? 1000/x
13. What is the maximum height the cannonball reaches?
14. Women employed. The number of women employed out- Width x x x
side the home reached a maximum during World War II. Let =x
W denote the number, in millions, of women employed out- Length = 1000/x
side the home t years after 1940. A quadratic model is W =
−0.74t 2 + 3.1t + 16. Find the value of t for which the number
of women employed is the greatest. Round your answer to the
nearest year. As a date, what year is this? FIGURE 3.44
15. Speed and safety. For commercial vehicles driving at night
on city streets, the accident rate A (per 100,000,000 vehicle- 19. Three sides. A landscaper wants to enclose just 3 sides of a
miles) depends on the speed s. A quadratic model is A = rectangular garden with an area of 1000 square feet. He wants
7.8s 2 − 514s + 8734. Find the speed at which the accident rate the plot to be divided into 4 sections as shown in Figure 3.45. If
is at a minimum. Round your answer in miles per hour to the he uses 200 feet of fencing, what should the dimensions of the
nearest whole number. garden be? Is there more than one answer? Round your results
in feet to one decimal place. Suggestion: Refer to Figure 3.45,
16. Parking. The number N, in thousands, of vehicles parked where x is in feet and the formula for the length uses the fact
downtown in a city depends on the time h in hours since 9:00 that the area is 1000 square feet. Use the dimensions shown
A.M. A quadratic model is N = −0.16h2 + h + 6. Find the value in the figure to calculate the total amount of fence used, and
of h for which the number of vehicles parked is the greatest. set that equal to 200 feet. Keep in mind that only three sides
Round your answer to the nearest hour. What time of day is are enclosed.
this?
17. A box. You want to make an open box (it has no lid) with a
square bottom, a height of 4 inches, and a volume of 324 cubic
inches. To do this, you take a square piece of cardboard, cut 4-
inch-long squares from each corner and fold up the edges. What
Width x x x x
should the dimensions of the cardboard be before you cut it?
=x
Suggestion: Refer to Figures 3.42 and 3.43, where x is in inches.
Remember that for a box, Volume = Length×Width×Height. Length = 1000/x
Use the dimensions shown in Figure 3.43 to calculate the vol-
ume of the box in terms of x, and set that equal to 324 cubic
inches.
FIGURE 3.45
x
44 4
4 20. Fencing. A farmer wants to enclose a rectangular garden
x–8 with an area of 180 square feet. Two opposite sides of the gar-
den require fencing material that costs $4 per linear foot. The
other two opposite sides of the garden require fencing mate-
x–8 x Height = 4 rial that costs $7 per linear foot. She has a total of $300 to
spend. How many feet of the $4 fencing does she need, and
Length Width how many feet of the $7 fencing does she need? Is there more
=x–8 =x–8 than one answer? Round your results in feet to one decimal
place. Suggestion: Refer to Figure 3.46, where the expression
180/x depends on the fact that the area is 180 square feet. Use
FIGURE 3.42 Cutting squares FIGURE 3.43 Cardboard the information in the figure to calculate the cost of the fence,
from a piece of cardboard. folded to make a box. and set that cost equal to $300.
Crauder-4100237 QL November 6, 2014 15:45
the figure to calculate the area, and set that equal to 25 square
yards.
180/x feet @ $7 per foot
x feet @ x feet @
$4 per foot $4 per foot
Width x x x
=x x
Length = (200 – 4x)/2
Length 4
= 4 + 2x
x 3
x x
FIGURE 3.47 Width = 3 + 2x
Width Length
x x x x x 14
=x = 14 + 2x x 10
x x
Width = 10 + 2x
FIGURE 3.48
FIGURE 3.51
23. Carpet. A carpet is 4 yards longer than it is wide. If it has
an area of 25 square yards, what are its dimensions? Round 26. Circular path. You measured the area of a circular garden
your results in yards to one decimal place. Suggestion: Refer to (in square meters) and then measured its circumference (in me-
Figure 3.49, where x is in yards. Use the dimensions shown in ters). You discovered that the numerical value of the area was
Crauder-4100237 QL November 6, 2014 15:45
206 CHAPTER 3
29. Correct method? A friend tells you that you can solve the
equation
(x − 1)(x − 2) = 12
by saying x − 1 = 12, so x = 13, or x − 2 = 12, so x = 14. Is
x your friend correct?
7
30. Taking square roots. An important step in deriving the
quadratic formula is to take square roots when the left-hand
side of the equation is a perfect square. Consider the equation
x–2 x2 − 4x + 4 = 7
Write the left-hand side as a square, then take square roots of
FIGURE 3.52 both sides to solve the equation. Give exact answers, not nu-
merical approximations.
31. Completing the square. A key step in deriving the quadratic
28. Ladder II. One end of a 13-foot-long ladder is resting on formula is to complete the square. If we have x2 + bx, what do
the top of a vertical wall. The distance from the foot of the lad- we need to add to make a perfect square? The answer turns out
der to the base of the wall is 7 feet less than the height of the b2 42
to be . For example, if we add = 4 to x2 + 4x, we get
wall. How high is the wall? Suggestion: Recall the Pythagorean 4 4
theorem, which says for legs a and b and hypotenuse c of a right x2 + 4x + 4, which is the perfect square (x + 2)2 . What do you
triangle that a 2 + b2 = c2 . Apply this theorem to Figure 3.53, need to add to x2 +8x in order to make it a perfect square? Add
where x is measured in feet. the number you find, and write the result as a perfect square.
W-6. Max-min for linear or exponential functions? In this section we solve max-
min problems for phenomena modeled by quadratic functions. Does it make sense
to solve max-min problems for phenomena modeled by linear or exponential
functions?
W-7. Dish TV antennas. Below is a photo of Dish TV antennas. Discuss what you
think they have in common with reflecting telescopes and solar collectors.
Edgars Dubrovskis/Shutterstock
W-8. Vomit Comet. Part of the training of astronauts is to take them on an airplane
that flies a parabolic path. The result is that the passengers feel weightless. (The plane
is affectionately called the Vomit Comet because the passengers usually get sick.)
Read this article: http://www.avweb.com/news/airman/184318-1.html?redirected=1
and write a brief report on what you learned.
W-9. Braking distance. The term braking distance refers to the distance your car
travels after you have hit your brakes. Obviously your braking distance increases
as your speed increases. Suppose you double your speed, say from 30 to 60 mph.
What effect do you think it will have on your braking distance? Will it double? It
may surprise you to learn that your braking distance increases by a factor of four!
In general, your braking distance is proportional to the square of your speed (a
quadratic relationship). That means, for example, that if you triple your speed, your
braking distance increases by a factor of nine! The Nevada State Driver’s Hand-
book (http://www.dmvnv.com/pdfforms/dlbook.pdf) has a discussion of this. Exam-
ine what it has to say and check your own state’s manual to see what it says about
this issue.
Crauder-4100237 QL November 6, 2014 15:45
208 CHAPTER 3
CHAPTER SUMMARY
This chapter discusses four basic types of functions: linear, exponential, logarithmic,
and quadratic. All four are important and occur naturally.
Change in function
Slope = Growth rate =
Change in independent variable
In practical settings, the slope always has an important meaning, and understanding
that meaning is often the key to analyzing linear functions. When a linear function
is given in a real-world context, proper use of the units of the slope (such as inches
per hour) can help us determine the meaning of the slope.
The graph of a linear function is always a straight line. When data points almost
fall on a straight line, it may be appropriate to approximate the data with a linear
function. We use the terms regression line or trend line for graphs of these linear
approximations.
500,000 2200
450,000 2000
400,000 1800
350,000 1600
1400
300,000
1200
250,000
1000
200,000
800
150,000 600
100,000 400
50,000 200
0 0
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
25
20
Decibel reading
15
10
0
0 10 20 30 40 50 60 70 80 90 100
Relative intensity
Logarithms can be used to solve exponential equations. This solution yields exact
formulas for the doubling time, tripling time, etc., of exponential growth.
210 CHAPTER 3
Graphs of quadratic functions are parabolas, which have shapes as shown in Fig-
ure 3.38 or Figure 3.39. The vertex of a parabola is the minimum point if the parabola
opens upward (and the maximum point if it opens downward). The x-value of the
−b
vertex occurs at x = . Quadratic functions arise in some max-min problems, in
2a
which the vertex corresponds to a solution.
KEY TERMS
linear, p. 139 common logarithm, p. 178 quadratic equation, p.195
regression line, p. 150 relative intensity, p. 180 quadratic formula, p. 196
trend line, p. 150 magnitude, p. 180 parabola, p. 199
exponential function, p. 159 decibel, p. 183 vertex, p. 199
half-life, p. 169 quadratic function, p.195
CHAPTER QUIZ
1. For the following scenarios, explain the practical meaning 4. My salary in 2005 was $48,000. I got a 3% raise each year
of the growth rate, and determine which functions are linear. for eight years. Find a formula that gives my salary as a func-
a. A gym charges a $100 initiation fee and $25 per month. tion of the time since 2005. What was my salary in 2013?
Is the total amount spent on the gym membership a linear Answer If we use S for my salary (in dollars) and t for the time
function of the number of months you go? in years since 2005, S = 48,000×1.03t . In 2013 my salary was
b. Another gym charges a $50 initiation fee plus $30 per $60,804.96.
month for the first 6 months and $25 per month for If you had difficulty with this problem, see Example 3.11.
each month over 6. Is the total amount spent on the gym
membership a linear function of the number of months 5. Uranium-235, which can be used to make atomic bombs,
you go? has a half-life of 713 million years. Suppose 2 grams of
uranium-235 were placed in a “safe storage location” 150 mil-
Answer In both cases, the growth rate is the extra cost for lion years ago. Find an exponential formula that gives the
another month. The first scenario determines a linear func- amount of uranium-235 remaining after h half-lives. How
tion because that monthly cost is constant. The second sce- many half-lives is 150 million years? (Round your answer to
nario does not determine a linear function because the monthly two decimal places.) How much uranium-235 still remains in
cost is not constant: it varies depending on how many months “safe storage” today?
you go.
Answer If A is the amount (in grams) remaining, A = 2 ×
If you had difficulty with this problem, see Example 3.1. (1/2)h . 150 million years is 0.21 half-life. The amount remain-
2. It begins raining at 10:00 A.M. and continues raining at ing is about 1.7 grams.
a constant rate all day. The rain gauge already contains 2 If you had difficulty with this problem, see Example 3.14.
inches of water when the rain begins. Between 12:00 P.M. and
6. One earthquake has a magnitude of 3.5, and another
2:00 P.M., the level of the rain gauge increases by 1 inch. Find
has a magnitude of 6.5. How do their relative intensities
a linear formula that gives the height of water in the gauge as
compare?
a function of the time since 10:00 A.M.
Answer The larger earthquake is 1000 times as intense as the
Answer If H denotes the height (in inches) of water in the smaller quake.
gauge t hours after 10:00 A.M., H = 2 + 0.5t.
If you had difficulty with this problem, see Example 3.17.
If you had difficulty with this problem, see Example 3.6.
7. An idling bulldozer makes a sound of about 85 decibels.
3. Data for the yearly best 100-meter dash time for a small What would be the decibel reading of two idling bulldozers
college from 2005 through 2010 were collected and fit with a side-by-side?
trend line. The equation for the trend line is D = 12 − 0.03t,
where t is the time in years since 2005 and D is the best time (in
Answer 88 decibels
seconds) for that year. Explain in practical terms the meaning If you had difficulty with this problem, see Example 3.19.
of the slope of the trend line. 8. My salary is increasing by 4% each year. How long will it
Answer The slope is −0.03 second per year. It means that the take for my salary to double?
best time was decreasing by about 0.03 second each year. Answer About 17.7 years.
If you had difficulty with this problem, see Example 3.7. If you had difficulty with this problem, see Example 3.21.
Crauder-4100237 QL November 6, 2014 15:45
Patricia Beck/KRT/Newscom
Crauder-4100237 QL November 6, 2014 16:14
4 PERSONAL FINANCE
T
he following article appeared at NerdScholar, a financial literacy site for students.
This article makes clear that borrowing and saving money are a big part of our
lives and that understanding these transactions can make a huge difference to us. In
this chapter, we explore basic financial terminology and mechanisms. In Section 4.1,
we examine the basics of compound interest and savings, and in Section 4.2, we look
at borrowing. In Section 4.3, we consider long-term savings plans such as retirement
funds. In Section 4.4, we focus on credit cards, and in Section 4.5, we discuss financial
terms heard in the daily news.
213
Crauder-4100237 QL November 6, 2014 16:14
214 CHAPTER 4
(annual percentage rate) Home TV & Video U. S. World Politics Entertainment Tech Travel
and APY (annual
percentage yield). Americans Still Aren’t Saving Enough
NEW YORK (CNNMoney)
MELANIE HICKEN @MELHICKEN February 26, 2013
M any Americans are still failing to sock away enough money to pay for
retirement—or emergencies for that matter, according to a survey of more
than 1,000 people released Monday.
Only half of respondents reported good savings habits, including having a
spending and/or saving plan in place, according to the Consumer Federation
of America and the American Savings Education Council which conducted the
survey.
While more than half of all Baby Boomers and Gen-Xers will be able to retire
with enough money to cover basic retirement needs, including health care costs,
a significant number are at risk of running short, according to projections by the
Employee Benefit Research Institute.
The survey also found that only 49% of non-retired respondents feel they are
saving enough to achieve a “desirable standard of living” in retirement.
When it came to overall household savings, the survey found there was little
improvement compared to last year.
“Clearly the great recession has had a lingering effect on many Americans,”
said Stephen Brobeck, executive director of the Consumer Federation of Amer-
ica. “Millions of families have been unable to make progress in rebuilding their
finances, especially their savings.”
Money management begins with saving, and (as the preceding article notes)
Americans generally don’t save enough. In this section, we see how to measure the
growth of savings accounts.
Before we delve into the issue of saving money, we should say a little bit about
interest rates. As you can see from the chart in Figure 4.1, interest rates on certificates
of deposit (CDs) have fluctuated over the years, and since 2010 they have been at
historic lows. As of March 2014, a one-year CD was paying about 0.25% and a five-
year CD 0.8%—mere fractions of a percent. While we want to acknowledge realistic
18
16
14
12
Percent
10
8
6
4
2
0
1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
Years
rates, many of the examples and exercises in this chapter sometimes use rates like
6%, 10%, or 12% to help make concepts easy to understand by avoiding complex
and distracting calculations.
Some readers may wish to take advantage of the following Quick Review of linear
and exponential functions before proceeding.
$4500 $4500
4000 4000
3500 3500
Balance (dollars)
Balance (dollars)
3000 3000
2500 2500
2000 2000
1500 1500
1000 1000
500 500
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (years) Time (years)
FIGURE 4.2 The graph of a linear function is FIGURE 4.3 The graph of an exponential function (with a
a straight line. base greater than 1) gets steeper as we move to the right.
216 CHAPTER 4
Investments such as savings accounts earn interest over time. Interest can be
earned in different ways, and that affects the growth in value of an investment.
Simple interest
The easiest type of interest to understand and calculate is simple interest.
cartoonstock.com
Suppose, for example, that we invest $1000 in an account that earns simple in-
terest at a rate of 10% per year. Then we earn 10% of $1000 or $100 in interest
each year. If we hold the money for six years, we get $100 in interest each year for
six years. That comes to $600 interest. If we hold it for only six months, we get a
half-year’s interest or $50.
Compound interest
Situations involving simple interest are fairly rare and usually occur when money
is loaned or borrowed for a short period of time. More often, interest payments are
made in periodic installments during the life of the investment. The interest payments
are credited to the account periodically, and future interest is earned not only on
Crauder-4100237 QL November 6, 2014 16:14
the original principal but also on the interest earned to date. This type of interest
calculation is referred to as compounding.
To see how compound interest works, let’s return to the $1000 investment earn-
ing 10% per year we looked at earlier, but this time let’s assume that the interest is
compounded annually (at the end of each year). At the end of the first year, we earn
© Karenr/Dreamstime.com
10% of $1000 or $100—the same as with simple interest. When interest is com-
pounded, we add this amount to the balance, giving a new balance of $1100. At the
end of the second year, we earn 10% interest on the $1100 balance:
This amount is added to the balance, so after two years the balance is
For comparison, we can use the simple interest formula to find out the simple
interest earned after two years:
Simple interest after two years = Principal × Yearly interest rate × Time in years
An illustration of growth due to
compound interest. = $1000 × 0.10/year × 2 years = $200
The following table compares simple interest and annual compounding over var-
ious periods. It uses $1000 for the principal and 10% for the annual rate. This table
shows why compounding is so important for long-term savings.
218 CHAPTER 4
$4500
4000
r est
3500 i nte
nd
ou
Balance (dollars)
3000 mp
Co
2500 interest
Simple
2000
1500
1000
500
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (years)
This interest is added to the principal to give an account balance at the end of the
first year of $530.00. We use this figure to calculate the second year’s interest:
We add this to the balance to find the balance at the end of two years:
$7000
6000
5000
Dollars
4000
3000
2000
1000
0
0 1 2 3 4 5 6 7 8 9 10
Years
Compounding frequency
Continuously Monthly Quarterly Yearly
The following formula shows how the APR is used in calculations to determine
interest rates.
APR
Period interest rate =
Number of periods in a year
Suppose, for example, that we invest $500 in a savings account that has an APR
of 6% and compounds interest monthly. Then there are 12 compounding periods
each year. We find the monthly interest rate using
APR 6%
Monthly interest rate = = = 0.5%
12 12
Crauder-4100237 QL November 6, 2014 16:14
220 CHAPTER 4
Morris/cartoonstock.com
Each month we add 0.5% interest to the current balance. The following table
shows how the account balance grows over the first few months:
Here is an explanation for the formula: Over each compounding period, the bal-
ance grows by the same percentage, so the balance is an exponential function of t.
That percentage growth is r as a decimal, and the initial value is the principal. Using
these values in the standard exponential formula
We can use this formula to find the balance of the account after five years. Five years
is 60 months, so we use t = 60 in the formula:
The APR by itself does not determine how much interest an account earns. The
number of compounding periods also plays a role, as the next example illustrates.
b. Again, we use the compound interest formula. To find the quarterly rate, we
divide the APR by 4. The APR as a decimal is 0.06, so as a decimal, the
quarterly rate is
APR 0.06
r = Quarterly rate = = = 0.015
4 4
Thus, 1 + r = 1.015. Also five years is 20 quarters, so we use t = 20 in the
compound interest formula:
c. This time we want the monthly rate, so we divide the APR by 12:
APR 0.06
r = Monthly rate = = = 0.005
12 12
Crauder-4100237 QL November 6, 2014 16:14
222 CHAPTER 4
d. We assume that there are 365 days in each year, so as a decimal, the daily
rate is
APR 0.06
r = Daily rate = =
365 365
This table shows that increasing the number of compounding periods increases
the interest earned even though the APR and the number of years stay the
same.
224 CHAPTER 4
cartoonstock.com
Here is a formula for the APY.
A derivation of this formula is shown in Algebraic Spotlight 4.1. Let’s apply this
formula in the preceding example with 10% APR compounded semi-annually. Com-
pounding is semi-annual, so the number of periods is n = 2, and the APR as a decimal
is 0.10:
APR n
APY = 1 + −1
n
2
0.10
= 1+ −1
2
= 0.1025
APY = 1 + −1
n
0.10 12
= 1+ −1
12
To four decimal places, this is 0.1047. Thus, the APY is about 10.47%.
CD rates
30-Day APR APY
$1000 – $99,999.99 3.21% 3.25%
$100,000+ 3.26% 3.30%
90-Day APR APY
$1000 − $9999.99 3.22% 3.25%
$10,000 − $99,999.99 3.26% 3.30%
$100,000+ 3.35% 3.40%
1-Year APR APY
$1000 − $9999.99 3.31% 3.35%
$10,000 − $99,999.99 3.35% 3.40%
$100,000+ 3.55% 3.60%
18-Month APR APY
$1000 − $9999.99 3.56% 3.60%
$10,000 − $99,999.99 3.59% 3.65%
$100,000+ 3.74% 3.80%
2-Year APR APY
$1000 − $9999.99 3.80% 3.85%
$10,000 − $99,999.99 3.84% 3.90%
$100,000+ 3.98% 4.05%
Crauder-4100237 QL November 6, 2014 16:14
226 CHAPTER 4
The APY can be used directly to calculate a year’s worth of interest. For example,
suppose we purchase a one-year $100,000 CD from First Command Bank. Table 4.1
gives the APY for this CD as 3.60%. So after one year the interest we will earn is
b. The APY for this CD is 1.05%, meaning the CD earns 1.05% interest over
the period of one year. Therefore,
We can also use the APY rather than the APR to calculate the balance over several
years if we wish. The APY tells us the actual percentage growth per year, including
interest we earned during the year due to periodic compounding.1 Thus, we can think
of compounding annually (regardless of the actual compounding period) using the
APY as the annual interest rate. Once again, the balance is an exponential function.
In this formula, the APY is in decimal form.
1 The idea of an APY normally applies only to compound interest. It is not used for simple interest.
Crauder-4100237 QL November 6, 2014 16:14
228 CHAPTER 4
If the account grows only by compounding each period at a constant interest rate
after we make an initial investment, then the present value is the principal, and the
future value is the balance given by the compound interest formula (Formula 4.3):
Balance after t periods = Principal × (1 + r )t
so
Future value = Present value × (1 + r )t
We can rearrange this formula to obtain
Future value
Present value =
(1 + r )t
In these formulas, t is the total number of compounding periods, and r is the period
interest rate expressed as a decimal.
Which of these two formulas we should use depends on what question we are
trying to answer. By the way, if we make regular deposits into the account, then the
formulas are much more complicated. We will examine that situation in Section 4.3.
EXAMPLE 4.7 Calculating present and future value: Investing for a car
You would like to have $10,000 to buy a car three years from now. How much would
you need to invest now in a savings account that pays an APR of 9% compounded
monthly?
SOLUTION
In this problem, we know the future value ($10,000) and would like to know the
present value. The monthly rate is r = 0.09/12 = 0.0075 as a decimal, and the
number of compounding periods is t = 36 months. Thus,
Future value
Present value =
(1 + r )t
$10,000
=
1.007536
= $7641.49
Therefore, you should invest $7641.49 now.
log 2
Number of periods to double =
log(1 + r )
log 2
=
log(1 + 0.02)
The result is about 35.0. Therefore, the actual doubling time is 35.0 quarters, or
eight years and nine months. Our estimate of nine years was three months too high.
230 CHAPTER 4
Questions of this sort may invite many correct responses. We present one possibility:
The APY is indeed a better indication of the return you can expect on an investment
than is the APR, just as it is a better indication of the interest you will pay on a
consumer loan. The APR does not take into account the number of compounding
periods, so this figure alone cannot tell you the interest that will be earned. The
APY uses both the APR and the number of compounding periods to give the actual
percentage earnings for an investment. The APY is always at least as large as the
APR.
Further questions of this type may be found in the What Do You Think section of
exercises.
Month Interest earned Balance 17. Retirement options. At age 25, you start work for a com-
pany and are offered two retirement options.
$4000.00
Retirement option 1: When you retire, you will receive a lump
1 $ $ sum of $30,000 for each year of service.
2 $ $ Retirement option 2: When you start to work, the company
3 $ $ deposits $15,000 into an account that pays a monthly inter-
est rate of 1%, and interest is compounded monthly. When
9–11. Compound interest calculated. Assume that we in- you retire, you get the balance of the account.
vest $2000 for one year in a savings account that pays an APR Which option is better if you retire at age 65? Which is better
of 10% compounded quarterly. Exercises 9 through 11 refer to if you retire at age 55?
this account.
18. Compound interest. Assume that an 18-month CD pur-
9. Make a table to show how much is in the account at the end chased for $7000 pays an APR of 6% compounded monthly.
of each quarter. (Calculate the interest and compound it each What is the APY? Would the APY change if the investment were
quarter rather than using the compound interest formula.) $11,000 for 30 months with the same APR and with monthly
10. Use your answer to Exercise 9 to determine how much to- compounding?
tal interest the account has earned after one year. 19. More compound interest. Assume that a 24-month CD
11. Compare the earnings to what simple interest or semi- purchased for $7000 pays an APY of 4.25% (and of course
annual compounding would yield. interest is compounded). How much do you have at maturity?
12. Using the compound interest formula. This is a continua- 20. Interest and APY. Assume that a one-year CD pur-
tion of Exercise 9. In Exercise 9, we invested $2000 for one year chased for $2000 pays an APR of 8% that is compounded
in a savings account with an APR of 10% compounded quar- semi-annually. How much is in the account at the end of
terly. Apply the compound interest formula (Formula 4.3) to each compounding period? (Calculate the interest and com-
see whether it gives the answer for the final balance obtained in pound it each period rather than using the compound inter-
Exercise 9. est formula.) How much total interest does it earn? What is
the APY?
13. The difference between simple and compound interest.
Suppose you invest $1000 in a savings account that pays an 21. More interest and APY. Assume that a one-year CD pur-
APR of 0.6%, which is a realistic rate for the year 2014. If the chased for $2000 pays an APR of 8% that is compounded
account pays simple interest, what is the balance in the account quarterly. How much is in the account at the end of each com-
after 20 years? If interest is compounded monthly, what is the pounding period? (Calculate the interest and compound it each
balance in the account after 20 years? period rather than using the compound interest formula.) How
much total interest does it earn? What is the APY?
14–15. Calculating interest. Assume that an investment
of $7000 earns an APR of 6% compounded monthly for 22–24. A CD from First Command Bank. Suppose you
18 months. Exercises 14 and 15 refer to this investment. bought a two-year CD for $10,000 from First Command Bank
back when the rates were as shown in Table 4.1. Exercises 22
14. How much money is in your account after 18 months? through 24 refer to this CD.
15. How much interest has been earned? 22. Use the APY from Table 4.1 to determine how much inter-
16. Getting rich. The following Dilbert cartoon jokes about est it would have earned for you at maturity.
making a million dollars by investing $100 at an APR of 23. Assume monthly compounding and use the APY formula
5% and waiting for 190 years. Assuming that interest is in part 2 of Summary 4.2 to find the APY from the APR. Com-
compounded annually, how much does this investment really pare this to the APY in Table 4.1.
produce?
DILBERT © 2008 Scott Adams. Used By permission of UNIVERSAL
UCLICK. All rights reserved.
Crauder-4100237 QL November 6, 2014 16:14
232 CHAPTER 4
24. Assume monthly compounding. Use the APR in Table 4.1 37. Future value. What is the future value of a 10-year invest-
and the compound interest formula to determine how much ment of $1000 at an APR of 9% compounded monthly? Ex-
interest the CD would have earned for you at maturity. plain what your answer means.
25. Some interest and APY calculations. Parts b and c refer 38. Present value. What is the present value of an investment
to the 2008–2009 rates at First Command Bank shown in Ta- that will be worth $2000 at the end of five years? Assume an
ble 4.1. APR of 6% compounded monthly. Explain what your answer
a. Assume that a one-year CD for $5000 pays an APR of means.
8% that is compounded quarterly. How much total interest 39. Doubling again. You have invested $2500 at an APR of
does it earn? What is the APY? 9%. Use the Rule of 72 to estimate how long it will be until
b. If you purchased a one-year CD for $150,000 from First your investment reaches $5000, and how long it will be until
Command Bank, how much interest would you have re- your investment reaches $10,000.
ceived at maturity? Was compounding taking place? Ex- 40. Getting APR from doubling rate. A friend tells you that
plain. her savings account doubled in 12 years. Use the Rule of 72 to
c. If you purchased a two-year CD for $150,000 from First estimate what the APR of her account was.
Command Bank, the APY (4.05%) was greater than the 41. Find the doubling time. Consider an investment of $3000
APR (3.98%) because compounding was taking place. We at an APR of 6% compounded monthly. Use the formula that
are not told, however, what the compounding period was. gives the exact doubling time to determine exactly how long
Use the APR to calculate what the APY would be with it will take for the investment to double. (See the first part of
monthly compounding. How does your answer compare Summary 4.3. Be sure to use the monthly rate for r .) Express
to the APY in the table? your answer in years and months. Compare this result with the
26–31. True or false. In Exercises 26 through 31, answer estimate obtained from the Rule of 72.
“True” if the statement is always true and “False” otherwise. The following exercises are designed to be solved using tech-
If it is false, explain why. nology such as calculators or computer spreadsheets.
26. Principal is the amount you have in your account. 42. Find the rate. A news article published in 2004 said that the
27. If the APY is greater than the APR, the cause must be com- U.S. House of Representatives passed a bill to distribute funds
pounding. to members of the Western Shoshone tribe. Here is an excerpt
28. The APY can be less than the APR over a short period of from that article:
time. The Indian Claims Commission decided the Western
29. If the interest earned by a savings account is compounded Shoshone lost much of their land to gradual encroach-
semi-annually, equal amounts of money are deposited twice ment. The tribe was awarded $26 million in 1977. That
during the year. has grown to about $145 million through compound
interest, but the tribe never took the money.
30. If the interest earned by a savings account paying an APR
of 8% is compounded quarterly, 2% of the current balance is Assume monthly compounding and determine the APR that
deposited four times during the year. would give this growth in the award over the 27 years from
1977 to 2004. Note: This can also be solved without technol-
31. If a savings account pays an APR of 8% compounded
ogy, using algebra.
monthly, 8% of the balance is deposited each month during
the year. 43. Solve for the APR. Suppose a CD advertises an APY of
8.5%. Assuming that the APY was a result of monthly com-
32. Interest and APR. Assume that a two-year CD for $4000
pounding, solve the equation
pays an APY of 8%. How much interest will it earn? Can you
determine the APR? 12
APR
33. Find the APR. Sue bought a six-month CD for $3000. She 0.085 = 1+ −1
12
said that at maturity it paid $112.50 in interest. Assume that
this was simple interest, and determine the APR.
to find the APR. Note: This can also be solved without tech-
34–36. First Command Bank. Exercises 34 through 36 refer nology, using algebra.
to the rates at First Command Bank shown in Table 4.1.
44. Find the compounding period. Suppose a CD advertises an
34. Larry invested $99,999 and Sue invested $100,000, each APR of 5.10% and an APY of 5.20%. Solve the equation
for one year. How much more did Sue earn than Larry?
n
35. Compute the amount earned on $100,000 invested at 0.051
0.052 = 1+ −1
3.26% APR in one year if compounding is taking place daily. n
36. Considering the results of Exercise 35 and the APY given
in Table 4.1, how often do you think First Command Bank was for n to determine how frequently interest is compounded.
compounding interest on 30-day $100,000 CDs?
Crauder-4100237 QL November 6, 2014 16:14
W-2. Simple formula? You and a friend are discussing the return on an account with
an initial balance of $100 and an APY of 5%. You want to use the APY balance
formula (Formula 4.5). Your friend says that he has a simpler approach: Each year
add 5% of $100 or $5 to the balance. Is your friend’s approach valid? What term
is used to describe the interest calculation employed by your friend? Under what
conditions will your friend’s approach give a reasonably good solution?
W-3. Simple versus compound interest. What features would you expect to see if
you graph the balance of an account earning simple interest? What features would
you expect to see if the account earns compound interest?
Exercise 13 provides additional information relevant to this topic.
W-4. Which would you choose? You have the option of borrowing money from one
source that charges simple interest or from another source that charges the same APR
but compounds the interest monthly. Which would you choose, and why?
Exercise 13 provides additional information relevant to this topic.
W-5. Which would you choose? You have the option of loaning money to one friend
who promises to pay simple interest or to another friend who promises to pay the
same APR but compound the interest. Which would you choose, and why?
Exercise 13 provides additional information relevant to this topic.
W-6. Given APR. You are considering two banks. Both banks offer savings accounts
with an APR of 2%. What other information would you want to know so that you
can choose between the two banks?
W-7. Doubling the number of compounding periods. If a bank doubles the number
of times per year that it compounds interest, does that double the amount you earn
in a year on your savings account? Explain.
W-8. Doubling principal invested. If you double the principal you invest in your
savings account, does that double the interest you earn in a year? Explain.
W-9. APR = APY. A bank tells you that its APR and APY are the same. What does
that tell you about compounding?
Crauder-4100237 QL November 6, 2014 16:14
234 CHAPTER 4
TAKE AWAY
FROM THIS
SECTION IN THE NEWS 4.3 Search
Be able to Home TV & Video U. S. World Politics Entertainment Tech Travel
calculate the
monthly payment Paying in Installments
on a loan.
I t can be challenging to pay for a whole semester of college in one lump sum.
So many colleges offer tuition payment plans that allow you to pay your bill
over time. Here’s how they work.
Like a No-Interest Loan
Deferred tuition payment plans, also known as tuition installment plans, are a
convenience to help you manage college expenses, particularly if you have trouble
paying a whole year’s tuition at once. Tuition paid in installments must normally
be paid off by the end of the relevant academic period, such as a semester or
academic year, and most such plans do not charge interest if you pay by check or
direct deposit. However, like loans, you must have a good credit history to qualify
for them.
The Tuition Deferment Process
Don’t show up on the first day of classes expecting to automatically qualify for
such a plan. You should discuss any payment plans beforehand with your college.
The plan may be handled by a private company or by the college. In either case
you will usually pay a relatively small service fee and, if appropriate, late fees. The
first payment of an installment plan is normally due at registration and may be
the largest. If you enroll at a college that does not offer such a plan, its financial
aid office may be able to refer you to a private commercial tuition management
company that does.
Beware Hidden Costs
Installment plans can end up costing you more in the long run if you drop out
or transfer to another college. Because the college expects costs to go up each
year, payments in the first two years of college may be larger than they would
be otherwise. An additional cost may also result if you pay by credit card. Some
plans charge a fee for this service, as high as nearly three percent. To determine
your college’s policy, check with its bursar’s office.
Tuition payment plans are not for everybody, but it’s good to know they’re
available if you need them. It’s one more way colleges try to help you pay for
your education.
Installment payments are a part of virtually everyone’s life. The goal of this sec-
tion is to explain how such payments are calculated and to see the implications of
installment plans.
2 Interest rates for mortgage and various other loans change over time. Thus, in many of the examples and exercises in
this section we use interest rates that may not be current, just as we did with CD interest rates in Section 4.1.
Crauder-4100237 QL November 6, 2014 16:14
This would be two monthly payments of approximately $55.13. This amount over-
estimates your monthly payment, however: in the second payment, you shouldn’t
have to pay interest on the amount you have already repaid.
We will see shortly that the correct payment is $53.78 for each of the two months.
This is not an obvious amount, but it is reasonable—it lies between the two extremes
of $50 and $55.13.
When we take out an installment loan, the amount of the payment depends on
three things: the amount of money we borrow (sometimes called the principal), the
interest rate (or APR), and the term of the loan.
Each payment reduces the balance owed, but at the same time interest is accruing
on the outstanding balance. This makes the calculation of the monthly payment fairly
complicated, as you might surmise from the following formula.3
3 This formula is built in as a feature on some hand-held calculators and most computer spreadsheets. Loan calculators
are also available on the Web.
Crauder-4100237 QL November 6, 2014 16:14
236 CHAPTER 4
To give a sense of these monthly payments, here are two examples that might
arise when getting a mortgage:
• If you borrowed $100,000 at 6% APR for 30 years, your monthly payment
would be approximately $600.
• If you borrowed $100,000 at 3% APR for 15 years, your monthly payment
would be approximately $690.
If you want to borrow money, the monthly payment formula allows you to de-
termine in advance whether you can afford that car or home you want to buy. The
formula also lets you check the accuracy of any figure that a potential lender quotes.
Let’s return to the earlier example of a $100 loan to buy a calculator. We have a
monthly rate of 5%. Expressed as a decimal, 5% is 0.05, so r = 0.05 and 1 + r =
1.05. Because we pay off the loan in two months, we use t = 2 in the monthly
payment formula (Formula 4.6):
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
$100 × 0.05 × 1.052
=
(1.052 − 1)
= $53.78
SOLUTION
a. The monthly rate as a decimal is
APR 0.06
r = Monthly rate = = = 0.005
12 12
This gives 1 + r = 1.005. We want to pay off the loan in three years or 36
months, so we use a term of t = 36 in the monthly payment formula:
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
$5000 × 0.005 × 1.00536
=
Benny Evans
(1.00536 − 1)
= $152.11
4 In March 2014, federally subsidized student loans were around 3.9% APR, and some private loans were advertised at
about 6.7% APR.
Crauder-4100237 QL November 6, 2014 16:14
Suppose you can afford a certain monthly payment and you’d like to know how
much you can borrow to stay within that budget. The monthly payment formula can
be rearranged to answer that question.
238 CHAPTER 4
and
APR (12y)
Monthly payment × 1+ −1
12
Amount borrowed =
APR APR (12y)
× 1+
12 12
SOLUTION
a. The monthly rate as a decimal is
APR 0.09
r = Monthly rate = = = 0.0075
12 12
Crauder-4100237 QL November 6, 2014 16:14
SOLUTION
a. In this case, the principal is $5000, the monthly interest rate r as a decimal is
0.075/12 = 0.00625, and the number t of payments is 12. We use the monthly
payment formula (Formula 4.6):
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
$5000 × 0.00625 × 1.0062512
=
(1.0062512 − 1)
= $433.79
We will make 12 payments of $433.79 for a total of 12 × $433.79 = $5205.48.
Because the amount we borrowed is $5000, the total amount of interest paid is
$205.48.
Crauder-4100237 QL November 6, 2014 16:14
240 CHAPTER 4
b. To calculate the interest earned on the CD, we use the compound interest
formula (Formula 4.3 from the preceding section). The monthly rate is the same
as in part a. Therefore,
Balance = Principal × (1 + r )t
= $5000 × 1.0062512
= $5388.16
This means that the total amount of interest we earned is $388.16.
c. The interest earned on the $5000 CD is $182.68 more than the interest paid
on the $5000 loan in part a.
Here is the explanation for this difference: When we save money, the
financial institution credits our account with an interest payment—in this case,
every month. We continue to earn interest on the full $5000 and on those
interest payments for the entire year. When we borrow money, however, we
repay the loan monthly, thus decreasing the balance owed. We are being
charged interest only on the balance owed, not on the full $5000 we borrowed.
One common piece of financial advice is to reduce the amount you borrow by
making a down payment. Exercises 17 through 19 explore the effect of making a
down payment on the monthly payment and the total interest paid.
5 If the term is at most five years and the APR is less than 7.5%, the actual payment is within 20% of the ratio. If the
term is at most two years and the APR is less than 9%, the actual payment is within 10% of the ratio.
Crauder-4100237 QL November 6, 2014 16:14
For the loan in the preceding example, our rule of thumb says that the monthly
payment is at least $833.33. Remember that this amount does not include the inter-
est payments. If the loan has an APR of 6.6%, for example, the actual payment is
$891.83.
Once again, a rule of thumb gives an estimate—not the exact answer. It can at
least tell us quickly whether we should be shopping on the BMW car lot.
SOLUTION
The monthly rate is 12%/12 = 1%. As a decimal, this is r = 0.01. The monthly
payment formula with t = 12 gives
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
$1000 × 0.01 × 1.0112
=
(1.0112 − 1)
= $88.85
Because the monthly rate is 1%, each month we pay 1% of the outstanding bal-
ance in interest. When we make our first payment, the outstanding balance is $1000,
so we pay 1% of $1000, or $10.00, in interest. Thus, $10 of our $88.85 goes toward
interest, and the remainder, $78.85, goes toward the outstanding balance. So after
the first payment, we owe:
Balance owed after 1 payment = $1000.00 − $78.85 = $921.15
When we make a second payment, the outstanding balance is $921.15. We pay
1% of $921.15 or $9.21 in interest, so $88.85 − $9.21 = $79.64 goes toward the
balance due. This gives the balance owed after the second payment:
Balance owed after 2 payments = $921.15 − $79.64 = $841.51
Crauder-4100237 QL November 6, 2014 16:14
242 CHAPTER 4
The loan in Example 4.14 is used to buy a computer. The amount you have paid
toward the actual cost of the computer (the principal) at a given time is referred to
as your equity in the computer. For example, the preceding table tells us that after
four payments we still owe $679.84. This means we have paid a total of $1000.00 −
$679.84 = $320.16 toward the principal. That is our equity in the computer.
b. For the first month, the interest we pay is 0.5% of the outstanding balance of
$150,000:
Now $750 of the $1665.31 payment goes to interest and the remainder,
$1665.31 − $750 = $915.31, goes toward reducing the principal. The balance
owed after one month is
This gives the first row of the table. The completed table is shown below.
c. The table from part b tells us that after four payments we still owe
$146,311.20. So our equity is
The graph in Figure 4.6 shows the percentage of each payment from Exam-
ple 4.15 that goes toward interest, and Figure 4.7 shows how equity is built. Note
that in the early months, a large percentage of the payment goes toward interest. For
long-term loans, an even larger percentage of the payment goes toward interest early
on. This means that equity is built slowly at first. The rate of growth of equity in-
creases over the life of the loan. Note in Figure 4.7 that when you have made half of
the payments (60 payments), you have built an equity of just over $60,000—much
less than half of the purchase price.
50 $160,000
45 140,000
Percentage to interest
40
120,000
Equity (dollars)
35
30 100,000
25 80,000
20 60,000
15
40,000
10
5 20,000
0 0
0 10 20 30 40 50 60 70 80 90 100 110 120 0 10 20 30 40 50 60 70 80 90 100 110 120
Payment number Payment number
FIGURE 4.6 Percentage of payment that FIGURE 4.7 Equity built: 10-year loan.
goes to interest: 10-year loan.
Crauder-4100237 QL November 6, 2014 16:14
244 CHAPTER 4
Home mortgages
Figure 4.8 shows that, like other interest rates, mortgage rates have fluctuated over
the years, and have recently been at historic lows. As of March 2014, the average
rate for a 30-year mortgage was 4.32%.
20.0
17.5
15.0
12.5
Percent
10.0
7.5
5.0
2.5
1980 1990 2000 2010
Source: Freddie Mac
FIGURE 4.8 A graph with data from Freddie Mac of 30-year fixed rate mortgage
averages in the United States, from 1970 to the present.
A home mortgage is a loan for the purchase of a home. It is very common for a
mortgage to last as long as 30 years. The early mortgage payments go almost entirely
toward interest, with a small part going to reduce the principal (see Figure 4.9).
As a consequence, your home equity grows very slowly. For a 30-year mortgage of
$150,000 at an APR of 6%, Figure 4.10 shows the percentage of each payment that
goes to interest, and Figure 4.11 shows the equity built. Your home equity is very
important because it tells you how much money you can actually keep if you sell
your house, and it can also be used as collateral to borrow money.
FIGURE 4.9 A typical mortgage statement. Note the small amount of the
payment going to reduce the principal.
Crauder-4100237 QL November 6, 2014 16:14
90 $160,000
80 140,000
Percentage to interest
70
120,000
Equity (dollars)
60
100,000
50
80,000
40
60,000
30
20 40,000
10 20,000
0 0
0 40 80 120 160 200 240 280 320 360 0 40 80 120 160 200 240 280 320 360
Payment number Payment number
FIGURE 4.10 Percentage of payment that FIGURE 4.11 Equity built: 30-year mortgage.
goes to interest: 30-year mortgage.
APR 0.09
r = Monthly rate = = = 0.0075
12 12
Because the loan is for 30 years, we use t = 30 × 12 = 360 months in the monthly
payment formula:
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
$300,000 × 0.0075 × 1.0075360
=
(1.0075360 − 1)
= $2413.87
The interest paid is the excess over $300,000, or $568,993.20. Your neighbor
paid almost twice as much in interest as she did for the home.
Now we see the effect of varying the term of the mortgage on the monthly
payment.
Crauder-4100237 QL November 6, 2014 16:14
246 CHAPTER 4
SOLUTION
Option 1: The monthly rate as a decimal is
APR 0.084
r = Monthly rate = = = 0.007
12 12
We use the monthly payment formula with t = 360 months:
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
$250,000 × 0.007 × 1.007360
=
(1.007360 − 1)
= $1904.59
Option 2: The APR for a 20-year loan is lower. (It is common for loans with shorter
terms to have lower interest rates.) An APR of 7.2% is a monthly rate of r =
0.072/12 = 0.006 as a decimal. We use the monthly payment formula with t =
20 × 12 = 240 months:
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
$250,000 × 0.006 × 1.006240
=
(1.006240 − 1)
= $1968.37
The monthly payment is about $60 higher than for the 30-year mortgage, but you
pay the loan off in 20 years rather than 30 years.
Option 3: Adding 4% to the amount we borrow gives a new loan amount of $260,000.
As with Option 2, the APR of 7.2% gives a monthly rate of r = 0.006. The term of
30 years means that we put t = 360 months into the monthly payment formula:
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
$260,000 × 0.006 × 1.006360
=
(1.006360 − 1)
= $1764.85
Getting the lower interest rate makes a big difference in the monthly payment, even
with the 4% added to the original loan balance. This is clearly a better choice than
Option 1 if we consider only the monthly payment. Option 3 does require borrowing
more money than Option 1 and could have negative consequences if you need to sell
Crauder-4100237 QL November 6, 2014 16:14
the home early. Further, comparing the amount of interest paid shows that Option 2
is the best choice from that point of view—if you can afford the monthly payment.
Randy Glasbergen
Adjustable-rate mortgages
In the summer of 2007, a credit crisis involving home mortgages had dramatic effects
on many people and ultimately on the global economy. Several major financial insti-
tutions collapsed in the autumn of 2008, and more than 3 million people lost their
homes through foreclosure.6 There were many causes of the crisis, but one promi-
nent cause was the increase in risky mortgage lending, a large percentage of which
involved adjustable-rate mortgages or ARMs.
One advantage of an ARM is that the initial rate is often lower than the rate
for a comparable fixed-rate mortgage. One disadvantage of an ARM is that a rising
prime interest rate may cause significant increases in the monthly payment. That is
what happened in 2007–2008. When housing prices began to fall and interest rates
for adjustable-rate mortgages began to grow, people had higher monthly payments
© Oersin/Dreamstime.com
while the value of their homes fell. Thus, mortgage delinquencies increased. This
contributed to financial markets becoming concerned about the soundness of U.S.
credit, which led to slowing economic growth in the United States and Europe.
6 http://usatoday30.usatoday.com/money/economy/housing/2009-01-14-foreclosure-record-filings N.htm
Crauder-4100237 QL November 6, 2014 16:14
248 CHAPTER 4
SOLUTION
Both loans have a principal of $200,000 and a term of t = 360 months. For the
fixed-rate mortgage, the monthly rate in decimal form is r = 0.066/12 = 0.0055.
The monthly payment formula gives
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
$200,000 × 0.0055 × 1.0055360
=
(1.0055360 − 1)
= $1277.32
For the ARM, the initial monthly rate in decimal form is r = 0.06/12 = 0.005.
The monthly payment formula gives
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
$200,000 × 0.005 × 1.005360
=
(1.005360 − 1)
= $1199.10
The initial monthly payment for the ARM is almost $80 less than the payment for
the fixed-rate mortgage—but the payment for the ARM could change at any time.
Example 4.18 illustrates why an ARM may seem attractive. The following ex-
ample illustrates one potential danger of an ARM. It is typical of what happened in
2007 to many families who took out loans in 2006 when interest rates fell to historic
lows.
The family is now paying about $1517.51 − $1266.71 or about $250 more per
month, and the fraction of monthly income used for the mortgage payment is
about 1517.51/5000 or 0.30. The family is now paying 30% of its monthly
income for the mortgage payment.
In the example, the extra burden on family finances caused by the increase in the
interest rate could lead to serious problems. Further rate adjustments could easily
result in the loss of the family home.
Let’s see what this rule of thumb would estimate for the payment on a mortgage
of $100,000 at an APR of 7.2%. The monthly rate is 7.2%/12 = 0.6%. The monthly
interest on the original balance is 0.6% of $100,000:
If this is a 30-year mortgage, the monthly payment formula gives the value $678.79
for the actual payment. This is about 13% higher than the estimate.
We should note that most home mortgage payments also include taxes and in-
surance. These vary from location to location but can be significant.8 For many, a
home mortgage is the most significant investment they will ever make. It is crucial to
understand clearly both the benefits and the costs of such an investment.
7 If the term is at least 30 years and the APR is at least 6%, then the actual payment is within 20% of the estimate. If
the term is at least 30 years and the APR is greater than 8%, then the actual payment is within 10% of the estimate.
8 See the group of exercises on affordability beginning with Exercise 41.
Crauder-4100237 QL November 6, 2014 16:14
250 CHAPTER 4
252 CHAPTER 4
7. Affording a car. You can get a car loan with a term of three Payment Applied to Applied to Outstanding
years at an APR of 5%. If you can afford a monthly payment number Payment interest balance owed balance
of $450, how much can you borrow?
$1500.00
8. Affording a home. You find that the going rate for a 1
home mortgage with a term of 30 years is 6.5% APR. 2
The lending agency says that based on your income, your 3
monthly payment can be at most $750. How much can you
borrow?
21. Another amortization table. Suppose we borrow $100
9. No interest. A car dealer offers you a loan with no interest
at 5% APR and pay it off in 12 monthly payments. Make
charged for a term of two years. If you need to borrow $18,000,
an amortization table showing payments over the first three
what will your monthly payment be? Which rule of thumb is
months.
relevant here?
10–13. Interest paid. For Exercises 10 through 13, assume
that you take out a $2000 loan for 30 months at 8.5% APR. Payment Applied to Applied to Outstanding
number Payment interest balance owed balance
10. What is the monthly payment?
$100.00
11. How much of the first month’s payment is interest? 1
2
12. What percentage of the first month’s payment is interest?
3
(Round your answer to two decimal places as a percentage.)
13. How much total interest will you have paid at the end of
the 30 months? 22–27. Term of mortgage. There are two common choices
for the term of a home mortgage: 15 years or 30 years. Sup-
14–16. More saving and borrowing. In Exercises 14 through
pose you need to borrow $90,000 at an APR of 6.75% to buy
16, we compare saving and borrowing as in Example 4.12.
a home. Exercises 22 through 27 explore mortgage options for
14. Suppose you borrow $10,000 for two years at an APR of the term.
8.75%. What will your monthly payment be? How much in-
terest will you have paid by the end of the loan? 22. What will your monthly payment be if you opt for a 15-
year mortgage?
15. Suppose it were possible to invest in a two-year $10,000
CD at an APR of 8.75% compounded monthly. How much 23. What percentage of your first month’s payment will be in-
interest would you be paid by the end of the period? terest if you opt for a 15-year mortgage? (Round your answer
to two decimal places as a percentage.)
16. In Exercise 14, the financial institution loaned you $10,000
for two years, but in Exercise 15, you loaned the financial in- 24. How much interest will you have paid by the end of the
stitution $10,000 for two years. What is the difference in the 15-year loan?
amount of interest paid? 25. What will your monthly payment be if you opt for a 30-
17–19. Down payment. In Exercises 17 through 19, we ex- year mortgage?
amine the benefits of making a down payment. 26. What percentage of your first month’s payment will be in-
17. You want to buy a car. Suppose you borrow $15,000 for terest if you opt for a 30-year mortgage? (Round your answer
two years at an APR of 6%. What will your monthly payment to two decimal places as a percentage.)
be? How much interest will you have paid by the end of the 27. How much interest will you have paid by the end of the
loan? 30-year loan? Is it twice as much as for a 15-year mortgage?
18. Suppose that in the situation of Exercise 17, you make a 28–29. Formula for equity. Here is a formula for the equity
down payment of $2000. This means that you borrow only built up after k monthly payments:
$13,000. Assume that the term is still two years at an APR of
6%. What will your monthly payment be? How much interest
will you have paid by the end of the loan? Amount borrowed × ((1 + r )k − 1)
Equity =
((1 + r )t − 1)
19. Compare your answers to Exercises 17 and 18. What are
the advantages of making a down payment? Why would a
borrower not make a down payment? Note: One other fac- where r is the monthly interest rate as a decimal and t is the
tor to consider is the interest one would have earned on the term in months. Exercises 28 and 29 use this formula for a
down payment if it had been invested. Typically, the inter- mortgage of $100,000 at an APR of 7.2% with two different
est rate earned on investments is lower than that charged for terms.
loans.
28. Assume that the term of the mortgage is 30 years. How
20. Amortization table. Suppose we borrow $1500 at 4% APR much equity will you have halfway through the term of the
and pay it off in 24 monthly payments. Make an amortization loan? What percentage of the principal is this? (Round your
table showing payments over the first three months. answer to one decimal place as a percentage.)
Crauder-4100237 QL November 6, 2014 16:14
29. Suppose now that instead of a 30-year mortgage, you have of $200. Is this too good to be true? Be sure to explain your
a 15-year mortgage. Find your equity halfway through the term answer.
of the loan, and find what percentage of the principal that is. 38. Is this reasonable? A lending agency advertises in the pa-
(Round your answer to one decimal place as a percentage.) per an APR of 12% on a home mortgage with a term of 30
Compare this with the percentage you found in Exercise 28. years. The ad claims that the monthly payment on a principal
Why should the percentage for the 15-year term be larger than of $100,000 will be $10,290. Is this claim reasonable? What
the percentage for the 30-year term? should the ad have said the payment would be (to the nearest
30–33. Car totaled. Exercises 30 through 33 refer to the dollar)? What do you think happened here?
following situation. In order to buy a new car, you finance 39. Question for thought. Why would it be unwise (even if it
$20,000 with no down payment for a term of five years at an were allowed) to charge the purchase of a home to a credit
APR of 6%. After you have the car for one year, you are in an card? Go beyond the credit limit to consider factors such as
accident. No one is injured, but the car is totaled. The insur- interest rates, budgeting, etc.
ance company says that before the accident, the value of the
car had decreased by 25% over the time you owned it, and the 40. Microloans and flat interest. The 2006 Nobel Peace Prize
company pays you that depreciated amount after subtracting was awarded to Muhammad Yunus and the Grameen Bank he
your $500 deductible. founded. The announcement of the award noted the develop-
ment of microloans for encouraging the poor to become en-
30. What is your monthly payment for this loan? trepreneurs. Microloans involve small amounts of money but
31. How much equity have you built up after one year? Sug- fairly high interest rates. In contrast to the installment method
gestion: Use the formula for equity stated in connection with discussed in this section, for microloans typically interest is
Exercises 28 and 29. paid at a flat rate: The amount of interest paid is not reduced as
the principal is paid off. More recently such loans have come
32. How much money does the insurance company pay you? under some criticism. Investigate flat-rate loans, the reasons
(Don’t forget to subtract the deductible.) why microloans are often structured this way, and any down-
sides to them.
33. Can you pay off the loan using the insurance payment,
or do you still need to make payments on a car you no Exercises 41 through 46 are suitable for group work.
longer have? If you still need to make payments, how much
41–46. Affordability. Over the past 40 years, interest rates
do you still owe? (Subtract the payment from the insurance
have varied widely. The rate for a 30-year mortgage reached a
company.)
high of 14.75% in July 1984, and it reached 4.64% in October
34–36. Rebates. When interest rates are low, some automo- 2010. A significant impact of lower interest rates on society is
bile dealers offer loans at 0% APR, as indicated in a 2014 ad- that they enable more people to afford the purchase of a home.
vertisement by a prominent car dealership, offering zero per- In the following exercises, we consider the purchase of a home
cent financing or cash back deals on some 2013 models. that sells for $125,000. Assume that we can make a down pay-
“Zero percent financing” means the obvious thing—that no ment of $25,000, so we need to borrow $100,000. We assume
interest is being charged on the loan. So if we borrow $1200 at that our annual income is $40,000 and that we have no other
0% interest and pay it off over 12 months, our monthly pay- debt. In Exercises 41 through 46, we determine whether we can
ment will be $1200/12 = $100. afford to buy the home at the high and low rates mentioned at
Suppose you are buying a new truck at a price of $20,000. the beginning of this paragraph.
You plan to finance your purchase with a loan that you will re- 41. What is our monthly income?
pay over two years. The dealer offers two options: either dealer
financing with 0% interest, or a $2000 rebate on the purchase 42. Lending agencies usually require that no more than 28%
price. If you take the rebate, you will have to go to the local of the borrower’s monthly income be spent on housing. How
bank for a loan (of $18,000) at an APR of 6.5%. Exercises 34 much does that represent in our case?
through 36 refer to these options. 43. The amount we will spend on housing consists of our
monthly mortgage payment plus property taxes and hazard in-
34. What would your monthly payment be if you used dealer
surance. Assume that property taxes plus insurance total $250
financing?
per month, and subtract this from the answer to Exercise 42 to
35. What would your monthly payment be if you took the re- determine what monthly payment we can afford.
bate? 44. Use your answer to Exercise 43 to determine how much
36. Should you take the dealer financing or the rebate? How we can borrow if the term is 30 years and the interest rate is at
much would you save over the life of the loan by taking the the historic high of 14.75%. Can we afford the home?
option you chose? 45. Use your answer to Exercise 43 to determine how much
we can borrow if the term is 30 years and the interest rate is
37. Too good to be true? A friend claims to have found a re-
4.64%. Can we afford the home now?
ally great deal at a local loan agency not listed in the phone
book: The agency claims that its rates are so low you can bor- 46. What is the difference in the amount we can borrow
row $10,000 with a term of three years for a monthly payment between the rates used in Exercises 44 and 45?
Crauder-4100237 QL November 6, 2014 16:14
254 CHAPTER 4
47. Some history. In the United States before the 1930s, 52. Finding the term. You want to borrow $15,000 at an APR
home ownership was not standard—most people rented. In of 7% to buy a car, and you can afford a monthly payment of
part, this was because home loans were structured differently: $500. To minimize the amount of interest paid, you want to
A large down payment was required, the term of the loan was take the shortest term you can. What is the shortest term you
five years or less, the regular payments went toward interest, can afford? Note: Your answer should be a whole number of
and the principal was paid off in a lump sum at the end of years. Rule of Thumb 4.2 should give you a rough idea of what
the term. Home mortgages as we know them came into being the term will be.
through the influence of the Federal Housing Administration,
established by Congress in the National Housing Act in 1934,
which provided insurance to lending agencies. Find more de-
tails on the history of home mortgages, and discuss why the
earlier structure of loans would discourage home ownership. IN THE NEWS 4.4 Search
W-2. Is this true? You are considering buying a home, but a friend tells you that by
the time you finish paying for the home, the amount you pay in interest to the bank
will exceed the listed price for the home. Is this an unusual situation? What factors
contribute to determining the part of your total payments that is interest?
Exercise 13 provides additional information relevant to this topic.
W-3. Smart shopping. List the factors that determine the monthly payment for a car
loan. Which of these can you control? Experiment by changing each factor to see
the effect. For example, if the amount you borrow is doubled, what happens to the
monthly payment? Use the information you gather in this way to sketch how you
would make a decision about a loan.
Exercise 17 provides additional information relevant to this topic.
W-4. Amortization table. Explain what an amortization table is. Provide an example.
Exercise 20 provides additional information relevant to this topic.
W-5. What part of it is mine? You are making monthly installment payments on a
car. Explain what is meant by your equity in the car.
Exercise 28 provides additional information relevant to this topic.
W-7. Is that the right payment? You are negotiating a home mortgage that has a term
of 30 years. The interest rate is moderate. You are presented with a monthly payment
figure that does not include taxes and insurance. Explain how you can quickly check
whether the figure the bank officer gave you is reasonable.
Exercise 38 provides additional information relevant to this topic.
W-8. Research on taxes and insurance. When you take out a mortgage to buy a home,
the lending institution will require you to have insurance and to pay your property
taxes. The amount of these costs varies depending on your location. Fees to cover
these costs are normally added to your monthly payment. Investigate the property
tax rate in your area, and determine the amount of tax you would pay on a home
that is assessed at $100,000.
Exercise 43 provides additional information relevant to this topic.
W-9. Research on closing costs. When you secure a home mortgage, the lending in-
stitution will charge you closing costs. These are fees in addition to the price of the
home, interest on the loan, taxes, and insurance. These fees fall into several cat-
egories, and they can be substantial. Use the Internet to investigate closing costs.
Report your findings.
W-10. Research on Fannie and Freddie. Fannie Mae and Freddie Mac have been
featured in a number of recent news articles. Use the Internet to determine what
these entities are and what role they play in home mortgages. Report your findings.
Crauder-4100237 QL November 6, 2014 16:14
256 CHAPTER 4
R ecent studies have shown that a large percentage of young adults under
35 are declining enrollment in employer-provided retirement plans. Consid-
ering the economy, it’s understandable that many young adults are finding it
impossible to focus on things any further off than paying student loan debt, pur-
chasing a home or vehicle, and getting started in their careers.
Many won’t be financially able to leave their parents’ home until well into
their twenties. Even if they’re interested, today’s young adults find it hard
to sacrifice a portion of their often meager wages for retirement savings.
Especially if they’re barely getting by as it is.
Considering this trend in the next generation, some companies that em-
ploy mostly younger adults aren’t bothering to offer retirement plan options and
are choosing instead to pay higher wages. While this may be saving companies
money and allowing Generation Y or Millennial adults to channel their money into
immediate debt-payoff or other savings goals, it’s not encouraging the younger
generations to think about and plan for the future.
Market research shows that those who are actively saving and investing for
their retirement are in the 35–65 age bracket. But, according to many financial
experts, this is alarmingly late in the game. The obvious dilemma is that young
adults need to start saving for retirement as soon as possible—while not
floundering under other financial obligations. It’s a challenge, but it can be
done.
Here are four retirement tips that young adults can start implementing now.
1. Invest early, no matter how small
Time is on your side. Because of the wonder of compound interest, as little as
$200 a month invested in your twenties can build to millions by the time you re-
tire. But realistically, how many young adults in our economy have $200 a month
to put into retirement savings?
The recommended allocation for retirement savings is 15% of your income,
but this is often an unrealistic number. The key is to stop worrying about
percentages and focus on saving something. Even small contributions to a
401K or Roth IRA when you’re young will put you much further ahead. You can
always invest more heavily once you’re in a better financial situation.
2. If your employer doesn’t offer a 401K, open an IRA
If your employer offers a 401K, you should undoubtedly participate. Most em-
ployers match your contribution, sometimes up to 100%. Investments are
before-tax and non-taxable, unless you withdraw them early.
If they don’t offer a 401K, consider opening a Roth IRA. Although invest-
ment companies usually require a minimum amount to start an account, many
are catching on to the needs of young adults and waving the minimum as long
as you agree to regular contributions. Again, the amount isn’t as important as
the fact that you’re saving something.
(Continued )
Crauder-4100237 QL November 6, 2014 16:14
4.3 Saving for the long term: Build that nest egg 257
258 CHAPTER 4
4.3 Saving for the long term: Build that nest egg 259
In Table 4.2, the balance after 10 months is $1046.22. There were 10 deposits
made, for a total of $1000. The remaining $46.22 comes from interest.
Suppose we had deposited that entire $1000 at the beginning of the 10-month
period. We use the compound interest formula (Formula 4.3 from Section 4.1) to
find the balance at the end of the period. The monthly rate of 1% in decimal form is
r = 0.01, so we find that our balance would have been
Balance after 10 months = Principal × (1 + r )t
= $1000 × (1 + 0.01)10
= $1104.62
That number is larger than the balance for monthly deposits because it includes in-
terest earned on the full $1000 over the entire 10 months. Of course, for most of us
it’s easier to make monthly deposits of $100 than it is to come up with a lump sum
of $1000 to invest.
These observations provide a helpful rule of thumb even though we don’t yet
have a formula for the balance.
Deposit × ((1 + r )t − 1)
Balance after t monthly deposits =
r
where t is the number of deposits, and r is the monthly interest rate APR/12
expressed as a decimal.
Alternatively, we can write the formula as
APR (12×y)
Deposit × 1+ −1
12
Balance after deposits for y years =
APR
12
where deposits are made monthly and y is the number of years. This form is
APR
equivalent since r = and the number of deposits is t = 12 × y.
12
9 See Exercises 27 and 28 for the adjustment necessary when deposits are made at the beginning of the month.
Crauder-4100237 QL November 6, 2014 16:14
260 CHAPTER 4
The ending balance is often called the future value for this savings arrangement.
A derivation of this formula is given in Algebraic Spotlight 4.4 at the end of this
section. Throughout this section, we assume that interest is compounded monthly.
Let’s check that this formula agrees with the balance at the end of the tenth month
as found in Table 4.2. The deposit is $100 each month, and the monthly rate as a
decimal is r = 0.01. We want the balance after t = 10 deposits. Using Formula 4.8,
we find
Deposit × ((1 + r )t − 1)
Balance after 10 deposits =
r
$100 × (1.0110 − 1)
=
0.01
= $1046.22
This is the same as the answer we obtained in Table 4.2.
Deposit × ((1 + r )t − 1)
Balance after 60 deposits =
r
$20 × ((1 + 0.07/12)60 − 1)
=
(0.07/12)
= $1431.86
The future value is $1431.86.
4.3 Saving for the long term: Build that nest egg 261
For example, if your goal is $20,000 to buy a car in two years, and if the APR
is 6%, you can use this formula to find how much you need to deposit each month.
The monthly rate as a decimal is
APR 0.06
r= = = 0.005
12 12
We use t = 2 × 12 = 24 deposits in the formula:
Goal × r
Needed deposit =
((1 + r )t − 1)
$20,000 × 0.005
=
(1.00524 − 1)
= $786.41
You need to deposit $786.41 each month so you can buy that $20,000 car when you
graduate.
APR 0.072
r= = = 0.006
12 12
and the number of deposits is t = 5 × 12 = 60:
Goal × r
Needed deposit =
((1 + r )t − 1)
$10,000 × 0.006
=
(1.00660 − 1)
= $138.96
He needs to deposit $138.96 each month.
Crauder-4100237 QL November 6, 2014 16:14
262 CHAPTER 4
4.3 Saving for the long term: Build that nest egg 263
SOLUTION
a. We have a goal of $500,000, so we use Formula 4.9. The monthly rate as a
decimal is
APR 0.09
r = Monthly rate = = = 0.0075
12 12
The number of deposits is t = 40 × 12 = 480, so the needed deposit is
Goal × r
Needed deposit =
((1 + r )t − 1)
$500,000 × 0.0075
=
(1.0075480 − 1)
= $106.81
b. The computation is the same as in part a except that the new monthly rate as
a decimal is
0.06
r= = 0.005
12
We have
Goal × r
Needed deposit =
((1 + r )t − 1)
$500,000 × 0.005
=
(1.005480 − 1)
= $251.07
Note that the decrease in the interest rate from 9% to 6% requires that you
more than double your monthly deposit if you are to reach the same goal. The
effect of this possible variation in interest rates is one factor that makes financial
planning for retirement complicated.
264 CHAPTER 4
An annuity works just like the installment loans we considered in Section 4.2,
only someone is paying us rather than the other way around.10 In fact, the formula
for the monthly payment applies in this situation, too. We can think of the institution
that holds our account at retirement as having borrowed our nest egg; it will pay us
back in monthly installments over the term of the annuity.
Nest egg × r × (1 + r )t
Monthly annuity yield =
((1 + r )t − 1)
where r is the monthly rate (as a decimal), and t is the term (the number of
months the annuity will last).
Alternatively, we can write the formula as
APR APR (12y)
Nest egg × × 1+
12 12
Monthly annuity yield =
APR (12y)
1+ −1
12
where y is the term of the annuity in years.
cartoonstock.com
10 We assume that the payments are made at the end of the month.
Crauder-4100237 QL November 6, 2014 16:14
4.3 Saving for the long term: Build that nest egg 265
Nest egg × r × (1 + r )t
Monthly annuity yield =
((1 + r )t − 1)
$800,000 × 0.005 × 1.005240
=
(1.005240 − 1)
= $5731.45
How large a nest egg is needed to achieve a desired annuity yield? We can answer
this question by rearranging the annuity yield formula (Formula 4.10).
Here, r is the monthly rate (as a decimal), and t is the term (in months) of the
annuity.
The following example shows how we use this formula.
EXAMPLE 4.25 Finding nest egg needed for annuity: Retiring on a 20-
year annuity
Suppose our retirement account pays 5% APR compounded monthly. What size nest
Benny Evans
egg do we need in order to retire with a 20-year annuity that yields $4000 per month?
SOLUTION
We want to achieve an annuity goal, so we use Formula 4.11. The monthly rate as a
decimal is r = 0.05/12, and the term is t = 20 × 12 = 240 months:
Annuity yield goal × ((1 + r )t − 1)
Nest egg needed =
(r × (1 + r )t )
= $606,101.25
Crauder-4100237 QL November 6, 2014 16:14
266 CHAPTER 4
The balance at retirement (the nest egg) is called the present value of the annuity.
Future value and present value depend on perspective. When you started saving, the
balance at retirement was the future value. When you actually retire, it becomes the
present value.
The obvious question is, how many years should you plan for the annuity to
last? If you set it up to last until you’re 80 and then you live until you’re 85, you’re in
trouble. What a retiree often wants is to have the monthly annuity payment continue
for as long as he or she lives. Insurance companies offer such an arrangement, and
it’s called a life annuity. How does it work?
The insurance company makes a statistical estimate of the life expectancy of a
customer, which is used in determining how much the company will probably have
to pay out.11 Some customers will live longer than the estimate (and the company
may lose money on them) but some will not (and the company will make money on
them). The monthly income for a given principal is determined from this estimate
using the formula for the present value of a fixed-term annuity.
4.3 Saving for the long term: Build that nest egg 267
268 CHAPTER 4
8. Fill in the following table. (Don’t use the regular deposits 20. Suppose you are allowed to contribute $100 each month
balance formula.) in addition to your employer’s contribution. What will be the
size of your nest egg at age 65? Compare this with your answer
At end of Interest paid on to Exercise 19.
month number prior balance Deposit Balance 21–22. A different nest egg. You begin working at age 25,
1 and your employer deposits $250 each month into a retirement
2 account that pays an APR of 6% compounded monthly. You
3 expect to retire at age 65. Use this information for Exercises 21
4 and 22.
21. What will be the size of your nest egg when you retire?
9. Use the regular deposits balance formula (Formula 4.8) to
determine the balance in the account at the end of four months. 22. Suppose instead that you arranged to start the regular de-
Compare this to the final balance in the table from Exercise 8. posits two years earlier, at age 23. What will be the size of your
nest egg when you retire? Compare this with your answer from
10. Use the regular deposits balance formula to determine the
Exercise 21.
balance in the account at the end of four years.
Crauder-4100237 QL November 6, 2014 16:14
4.3 Saving for the long term: Build that nest egg 269
23–24. Planning to retire on an annuity. You plan to work nest egg) of $650,000. If the APR is 5% compounded monthly,
for 40 years and then retire using a 25-year annuity. You want what is your monthly income?
to arrange a retirement income of $4500 per month. You have 31. Another perpetuity. Refer to Exercise 29 for background
access to an account that pays an APR of 7.2% compounded on perpetuities. You have a perpetuity with a present value of
monthly. Use this information for Exercises 23 and 24. $900,000. If the APR is 4% compounded monthly, what is
23. What size nest egg do you need to achieve the desired your monthly income?
monthly yield? 32. Comparing annuities and perpetuities. Refer to Exercise 29
24. What monthly deposits are required to achieve the desired for background on perpetuities. For 40 years, you invest $200
monthly yield at retirement? per month at an APR of 4.8% compounded monthly, then you
25–26. Retiring. You want to have a monthly income of retire and plan to live on your retirement nest egg.
$2000 from a fixed-term annuity when you retire. Take the a. How much is in your account on retirement?
term of the annuity to be 20 years, and assume an APR of 6% b. Suppose you set up your account as a perpetuity on re-
over the period of investment covered in Exercises 25 and 26. tirement. What will your monthly income be? (Assume that
25. How large will your nest egg have to be at retirement to the APR remains at 4.8% compounded monthly.)
guarantee the income described above? c. Suppose now you use the balance in your account for a
26. You plan to make regular deposits for 40 years to build up life annuity instead of a perpetuity. If your life expectancy
your savings to the level you determined in Exercise 25. How is 21 years, what will your monthly income be? (Again, as-
large must your monthly deposit be? sume that the APR remains at 4.8% compounded monthly.)
27–28. Deposits at the beginning. In Section 4.3, we consid- d. Compare the total amount you invested with your to-
ered the case of regular deposits at the end of each month. If tal return from part c. Assume that you live 21 years after
deposits are made at the beginning of each month, then the retirement.
formula is a bit different. The adjusted formula is 33. Perpetuity goal: How much do I need to retire? Refer to
((1 + r )t − 1) Exercise 29 for background on perpetuities. Here is the for-
Balance after t deposits = Deposit × (1 + r ) × mula for the nest egg needed for a desired monthly yield on a
r
perpetuity:
Here, r is the monthly interest rate (as a decimal), and t is the Desired monthly yield
number of deposits. The extra factor of 1 + r accounts for the Nest egg needed =
Monthly interest rate
interest earned on the deposit over the first month after it’s
made. In this formula, the monthly interest rate is expressed as a
Suppose you deposit $200 at the beginning of each month decimal.
for five years. Take the APR to be 7.2% for Exercises 27 and 28. If your retirement account pays 5% APR with monthly
compounding, what present value (that is, nest egg) is required
27. What is the future value? In other words, what will your for you to retire on a perpetuity that pays $4000 per month?
account balance be at the end of the period?
34. Desired perpetuity. Refer to the formula in Exercise 33.
28. What would be the future value if you had made the de- You want a perpetuity with a monthly income of $3000. If the
posits at the end of each month rather than at the beginning? APR is 7%, what does the present value have to be?
Explain why it is reasonable that your answer here is smaller
than that from Exercise 27. 35–36. Planning to retire on a perpetuity. You plan to work
for 40 years and then retire using a perpetuity. You want to
29. Retirement income: perpetuities. If a retirement fund is set arrange to have a retirement income of $4500 per month.
up as a perpetuity, one withdraws each month only the inter- You have access to an account that pays an APR of 7.2%
est accrued over that month; the principal remains the same. compounded monthly. Use this information for Exercises 35
For example, suppose you have accumulated $500,000 in an and 36.
account with a monthly interest rate of 0.5%. Each month,
you can withdraw $500,000 × 0.005 = $2500 in interest, 35. Refer to the formula in Exercise 33. What size nest egg do
and the nest egg will always remain at $500,000. That is, the you need to achieve the desired monthly yield?
$500,000 perpetuity has a monthly yield of $2500. In general, 36. What monthly deposits are required to achieve the desired
the monthly yield for a perpetuity is given by the formula monthly yield at retirement?
Monthly perpetuity yield = Nest egg × Monthly interest rate Exercises 37 through 40 are suitable for group work.
In this formula, the monthly interest rate is expressed as a 37–40. Starting early, starting late. In Exercises 37 through
decimal. 40, we consider the effects of starting early or late to save for
Suppose we have a perpetuity paying an APR of 6% com- retirement. Assume that each account considered has an APR
pounded monthly. If the value of our nest egg (that is, the of 6% compounded monthly.
present value) is $800,000, find the amount we can withdraw 37. At age 20, you realize that even a modest start on saving for
each month. Note: First find the monthly interest rate. retirement is important. You begin depositing $50 each month
30. Perpetuity yield. Refer to Exercise 29 for background on into an account. What will be the value of your nest egg when
perpetuities. You have a perpetuity with a present value (that is, you retire at age 65?
Crauder-4100237 QL November 6, 2014 16:14
270 CHAPTER 4
38. Against expert advice, you begin your retirement program and write a report on it. Include the use of annuities in Rome,
at age 40. You plan to retire at age 65. What monthly con- in Europe during the seventeenth century, in colonial America,
tributions do you need to make to match the nest egg from and in modern American society.
Exercise 37? 43. History of actuaries. Look up the history of the actuarial
39. Compare your answer to Exercise 38 with the monthly de- profession and write a report on it. Be sure to discuss what
posit of $50 from Exercise 37. Also compare the total amount mathematics are required to become an actuary.
deposited in each case. The following exercises are designed to be solved using tech-
40. Let’s return to the situation in Exercise 37: At age 20, you nology such as calculators or computer spreadsheets.
begin depositing $50 each month into an account. Now sup- 44. How much? You begin saving for retirement at age 25,
pose that at age 40, you finally get a job where your employer and you plan to retire at age 65. You want to deposit a certain
puts $400 per month into an account. You continue your $50 amount each month into an account that pays an APR of 6%
deposits, so from age 40 on, you have two separate accounts compounded monthly. Make a table that shows the amount
working for you. What will be the total value of your nest egg you must deposit each month in terms of the nest egg you desire
when you retire at age 65? to have when you retire. Include nest egg sizes from $100,000
41. Retiring without interest. Suppose we lived in a society to $1,000,000 in increments of $100,000.
without interest. At age 25, you begin putting $250 per month 45. How long? You begin working at age 25, and your em-
into a cookie jar until you retire at age 65. At age 65, you begin ployer deposits $350 each month into a retirement account that
to withdraw $2500 per month from the cookie jar. How long pays an APR of 6% compounded monthly. Make a table that
will your retirement fund last? shows the size of your nest egg in terms of the age at which you
42. History of annuities. The origins of annuities can be traced retire. Include retirement ages from 60 to 70.
back to the ancient Romans. Look up the history of annuities
W-7. Annuities, plus and minus. Explain some of the advantages and disadvantages
of a fixed-term annuity.
W-8. Perpetuities, plus and minus. See Exercise 29 for background on perpetuities.
Explain some advantages and disadvantages of a perpetuity.
Exercise 32 provides additional information relevant to this topic.
W-9. Fixed vs. life. Explain in your own words the difference between a fixed-term
annuity and a life annuity.
TAKE AWAY
FROM THIS
SECTION
IN THE NEWS 4.6 Search
Be savvy in paying
Home TV & Video U. S. World Politics Entertainment Tech Travel
off credit cards.
The Case for College Students to Have Credit Cards
ERIN LOWRY July 11, 2014
A credit card is a simple tool to both establish credit history and begin working
toward a good credit score.
Student loans help establish credit and can positively impact a credit score.
However, part of the FICO scoring model depends on diversity of credit. And
responsible use of a credit card—making on-time payments, for example—is a
simple way to increase a credit score.
Credit cards also offer students without loans the opportunity to establish
their credit and spend four years proving they are responsible borrowers be-
fore graduation. Lest we forget, recent grads who don’t want to return to their
parents’ basement will need a credit score to get their own apartment or house.
To avoid credit card debt and protect their credit scores, young adults need
to be vigilant about paying their bills on time and in full. Credit card users should
understand the ramifications of carrying a balance and paying interest on their
debt. A credit card shouldn’t be used any differently than a debit card, except
you have to pay your bill at the end of the month instead of automatically seeing
the money get deducted from your account.
The decision whether or not to apply for a credit card should be made based
on self-awareness and fiscal responsibility. It takes a basic level of financial edu-
cation to understand how credit cards work, the interest rates that get charged
on debt and how easy it is to get trapped in a cycle of throwing money toward
interest. But for responsible, independent college students, a credit card can help
establish their financial footing.
Credit cards are convenient and useful. They allow us to travel without carrying
large sums of cash, and they sometimes allow us to defer cash payments, even interest-
free, for a short time. In fact, owning a credit card can be a necessity: Most hotels
and rental car companies require customers to have a credit card. Based on March
2013 credit card bills, the average balance on a card that usually carries a balance
was $8220.12
12 See http://www.foxbusiness.com/personal-finance/2013/07/08/average-credit-card-debt-take-your-pick/
As that article discusses, averages can be a bit misleading. We will say more about this in Chapter 6.
Crauder-4100237 QL November 6, 2014 16:14
272 CHAPTER 4
The convenience comes at a cost, however. For example, some credit cards carry
an APR much higher than other kinds of consumer loans. The APR depends on
various factors, including the customer’s credit rating. At the time of this writing,
the site www.indexcreditcards.com/lowaprcreditcards.html reveals cards with APRs
ranging from 7.25% to 22.99%.
In this section, we explore credit cards in some detail. In particular, we look at
how payments are calculated, the terminology used by credit card companies, and
the implications of making only the minimum required payments.
www.glasbergen.com
© Randy Glasbergen/
FIGURE 4.12 A typical credit card statement. Look at the fine print!
Each month your credit card statement shows your previous month’s balance,
payments and purchases you’ve made, finance charges (or interest), and the new
balance. The new balance is calculated as follows:
New balance = Previous balance − Payments + Purchases + Finance charge
The finance charge is calculated as follows: Start with the balance shown in the state-
ment from the previous month, subtract payments you’ve made since the previous
Crauder-4100237 QL November 6, 2014 16:14
statement, and add any new purchases. That is the amount that is subject to finance
charges.13
The finance charge is calculated by applying the monthly interest rate (the APR
divided by 12) to this amount. Thus,
APR
Finance charge = × (Previous balance − Payments + Purchases)
12
Let’s look at a card that charges 21.6% APR. Suppose your previous statement
showed a balance of $300, you made a payment of $100 in response to your previous
statement, and you have new purchases of $50.
Because the APR is 21.6%, we find the monthly interest rate using
APR 21.6%
Monthly interest rate = = = 1.8%
12 12
Now we calculate the finance charge:
Finance charge = 0.018 × (Previous balance − Payments + Purchases)
= 0.018 × ($300 − $100 + $50)
= 0.018 × $250 = $4.50
This makes your new balance
New balance = Previous balance − Payments + Purchases + Finance charge
= $300 − $100 + $50 + $4.50 = $254.50
Your new balance is $254.50.
SOLUTION
We start by entering the information given.
The APR is 22.8%, and we divide by 12 to get a monthly rate of 1.9%. In decimal
form, this calculation is 0.228/12 = 0.019. Therefore,
Finance charge = 0.019 × (Previous balance − Payments + Purchases)
= 0.019 × ($500 − $200 + $400)
= 0.019 × $700 = $13.30
13 The most common method of calculation actually uses the average daily balance. That is the average of the daily
balances over the payment period (usually one month). This means that a pair of jeans purchased early in the month
will incur more finance charges than the same pair purchased late in the month. Most experts consider this a fair way
to do the calculation. See Exercises 33 and 34.
Crauder-4100237 QL November 6, 2014 16:14
274 CHAPTER 4
Let’s pursue the situation described in the previous example. The following table
covers the first four payments if we continue making the minimum 5% payment each
month:
The table shows that the $400 is not being paid off very quickly. In fact, after
four payments, the decrease in the balance is only about $47 (from $400 to $352.65).
If we look closely at the table, we will see that the balances follow a pattern. To find
the pattern, let’s look at how the balance changes in terms of percentages:
Month 1: The new balance of $387.60 is 96.9% of the initial balance of $400.
Month 2: The new balance of $375.58 is 96.9% of the Month 1 new balance of $387.60.
Month 3: The new balance of $363.94 is 96.9% of the Month 2 new balance of $375.58.
Month 4: The new balance of $352.65 is 96.9% of the Month 3 new balance of $363.94.
Crauder-4100237 QL November 6, 2014 16:14
276 CHAPTER 4
$450
400
The balance exhibits a constant percentage change. This makes sense: Each month,
the balance is decreased by a constant percentage due to the minimum payment and
increased by a constant percentage due to the finance charge. This pattern indicates
that the balance is a decreasing exponential function. That conclusion is supported
by the graph of the balance shown in Figure 4.13, which has the classic shape of
exponential decay.
Because of this exponential pattern, we can find a formula for the balance on
our credit card in the situation where we stop charging and pay off the balance by
making the minimum payment each month.
SOLUTION
The APR in decimal form is 0.2, so the monthly interest rate in decimal form is
r = 0.2/12. To avoid rounding, we leave r in this form. The minimum payment is
4% of the new balance, so we use m = 0.04. The initial balance is $250. The number
of payments for two years is t = 24. Using the minimum payment balance formula
Crauder-4100237 QL November 6, 2014 16:14
We already noted that the credit card balance is not paid off quickly when we
make only the minimum payment each month. The reason for this is now clear: The
balance is a decreasing exponential function, and such functions typically decrease
very slowly in the long run. The next example illustrates the dangers of making only
the minimum monthly payment.
EXAMPLE 4.29 Paying off your credit card balance: Long repayment
Suppose you have a balance of $10,000 on your Visa card, which has an APR of 24%.
The card requires a minimum payment of 5% of the balance. You stop charging and
begin making only the minimum payment until your balance is below $100.
a. Find a formula that gives your balance after t monthly payments.
b. Find your balance after five years of payments.
c. Determine how long it will take to get your balance under $100.14
Wise credit card use avoids debt. d. Suppose that instead of the minimum payment, you want to make a fixed
monthly payment so that your debt is clear in two years. How much do you pay
each month?
SOLUTION
a. The minimum payment as a decimal is m = 0.05, and the monthly rate as a
decimal is r = 0.24/12 = 0.02. The initial balance is $10,000. Using the
minimum payment balance formula (Formula 4.12), we find
b. Now five years is 60 months, so we put t = 60 into the formula from part a:
14 It is worth noting that paying in this fashion will never actually get your balance to exactly 0. But doing so will
eventually make the balance small enough that you could pay it off in a lump sum. Normally, credit cards take care of
this with a requirement such as “The minimum payment is 5% or $20, whichever is larger.”
Crauder-4100237 QL November 6, 2014 16:14
278 CHAPTER 4
c. We will show two ways to determine how long it takes to get the balance
down to $100.
Method 1: Using a logarithm: We need to solve for t the equation
100 10,000
= × 0.969t
10,000 10,000
0.01 = 0.969t
log 0.01
t= months
log 0.969
This is about 146.2 months. Hence, the balance will be under $100 after
147 monthly payments, or more than 12 years of payments.
Method 2: Trial and error: If you want to avoid logarithms, you can solve this
problem using trial and error with a calculator. The information in part b
indicates that it will take some time for the balance to drop below $100. So we
might try 10 years, or 120 months. Computation using the formula from part a
shows that after 10 years, the balance is still over $200, so we should try a
larger number of months. If we continue in this way, we find the same answer
as that obtained from Method 1: The balance drops below $100 at payment
147, which represents over 12 years of payments. (Spreadsheets and many
calculators will create tables of values that make problems of this sort easy to
solve.)
d. Making fixed monthly payments to clear your debt is like considering your
debt as an installment loan: Just find the monthly payment if you borrow
$10,000 to buy (say) a car at an APR of 24% and pay the loan off over 24
months. We use the monthly payment formula from Section 4.2:
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
Recall that t is the number of months taken to pay off the loan, in this case
24, and that r is the monthly rate as a decimal, which in this case is 0.02.
Hence,
So a payment of $528.71 each month will clear the debt in two years.
Crauder-4100237 QL November 6, 2014 16:14
280 CHAPTER 4
sometimes raise your credit limit, if you have always made required payments by its
due dates.
In addition to all these complications, there are often devious hidden fees and
charges. President Barack Obama signed into law the Credit Card Accountability
Responsibility and Disclosure (CARD) Act of 2009. Among other things, the law
requires that credit card companies inform cardholders how long it will take to
pay off the balance if they make only the minimum payment. The following ex-
cerpt is from an article at the Web site of the federal Consumer Financial Protection
Bureau.
T his factsheet was published in February 2011, as part of the Bureau’s initial
report on the CARD Act. The Bureau has since assumed responsibility for ad-
ministering the CARD Act. See our homepage for the latest from the CFPB.
One year ago today, on February 22, 2010, many of the provisions of the
Credit Card Accountability, Responsibility, and Disclosure Act (CARD Act) took
effect.
When the CARD Act was signed into law in May 2009, it was clear the credit
card market was in need of serious reform. Congress concluded that certain prac-
tices in the credit card industry were not fair and transparent to consumers, and
the CARD Act passed with very strong bipartisan support in both the House
and the Senate. As President Obama said when he signed the CARD Act into
law, the Act was intended to uphold “basic standards of fairness, transparency,
and accountability.”
Later this summer, the new Consumer Financial Protection Bureau (CFPB) will
assume responsibility for administering the CARD Act. To help guide it in perform-
ing these responsibilities, the CFPB has convened a conference, The CARD Act:
One Year Later. To provide a factual basis for the discussion, the CFPB conducted a
voluntary survey of the nine largest card issuers, who together represent approx-
imately 90 percent of the credit card marketplace, about their current practices
and future plans. Several other studies were also developed for this conference,
including a report by the Office of the Comptroller of the Currency (OCC) that
looks specifically at changes in the pricing practices of card issuers. To explore the
impact of the CARD Act’s new disclosure requirements on the consumer experi-
ence, the CFPB also commissioned a survey of cardholders.
These studies demonstrated that one year after the effective date of many
provisions of the CARD Act, industry practices have changed in four significant
ways.
• The long-standing practice of hiking interest rates on existing cardholder
accounts has been dramatically curtailed.
• The amount of late fees consumers are paying has been substantially
reduced.
• Overlimit fees have virtually disappeared in the credit card industry.
• Consumers report that their credit card costs are clearer, but significant
confusion remains.
Crauder-4100237 QL November 6, 2014 16:14
Suppose a credit card has an initial balance. Assume that we incur no further
charges and make only the minimum payment each month. Suppose r is the
monthly interest rate and m is the minimum monthly payment as a percentage
of the new balance. Both r and m are in decimal form. Our goal is to derive the
formula
Here is the derivation. Assume that we have made a series of payments, and let
B denote the balance remaining. We need to calculate the new balance. First
we find the minimum payment on the balance B. To do so, we multiply B by
m:
Minimum payment = mB
282 CHAPTER 4
Questions of this sort may invite many correct responses. We present one possibility:
You certainly want to know the interest rate, but you also need to know how finance
charges are calculated. The method of “average daily balance” is thought by many
to be the fairest method of arriving at the balance on which charges are calculated.
You also need to understand penalties charged by the card. They can be substantial
and result in long-lasting problems. Credit cards are best used if the balance is paid in
full each month, but in case that isn’t possible, you need to know how the card deals
with minimum payments. There are many more credit card issues that are important,
including rewards programs. Recent laws are designed to limit predatory practices
of credit card companies, but in the end it is up to you to “read the fine print.” If
you don’t understand how a credit card works, don’t get it.
Further questions of this type may be found in the What Do You Think section of
exercises.
Try It Yourself 4.27: Finding next month’s minimum payment: One payment
New balance: $375.58; minimum payment: $18.78.
Try It Yourself 4.28: Using the minimum payment balance formula: Balance after
three years $227.59.
7. Calculating balances. You have a credit card with an APR 11. Find a formula for the balance after t monthly payments.
of 22.8%. You begin with a balance of $1000, in response to 12. What will the balance be after 30 months?
which you make a payment of $200. The first month you make
charges amounting to $500. You make a payment of $200 13. What will the balance be after 10 years?
to reduce the new balance, and the second month you charge 14–17. Paying tuition on your American Express card at
$600. Complete the following table: the maximum interest rate. Use the following information
for Exercises 14–17. You have a balance of $10,000 for your
tuition on your American Express credit card. Assume that you
Previous Finance New
make no more charges on the card. Also assume that American
balance Payments Purchases charge balance
Express charges 24% APR and that each month you make only
Month 1 the minimum payment of 2% of the balance.
Month 2
14. Find a formula for the balance after t monthly payments.
15. How much will you owe after 10 years of payments?
8. Calculating balances. You have a credit card with an APR of
12%. You begin with a balance of $200, in response to which 16. How much would you owe if you made 100 years of
you make a payment of $75. The first month you make charges payments?
amounting to $50. You make a payment of $75 to reduce the 17. Find when the balance would be less than $50.
new balance, and the second month you charge $60. Complete
18–19. Paying tuition on your American Express card. Use
the following table:
the following information for Exercises 18 and 19. You have a
balance of $10,000 for your tuition on your American Express
Previous Finance New credit card. Assume that you make no more charges on the
balance Payments Purchases charge balance card. Also assume that American Express charges 12% APR
and that each month you make only the minimum payment of
Month 1 4% of the balance.
Month 2
18. Find a formula for the balance after t monthly payments.
9. A balance statement. Assume that you start with a balance 19. How long will it take to get the balance below $50?
of $4500 on your Visa credit card. During the first month, you 20–21. Paying off a Visa card. Use the following informa-
charge $500, and during the second month, you charge $300. tion for Exercises 20 and 21. You have a balance of $1000 on
Assume that Visa has finance charges of 24% APR and that your Visa credit card. Assume that you make no more charges
each month you make only the minimum payment of 2.5% of on the card and that the card charges 9.9% APR and requires
the balance. Complete the following table: a minimum payment of 3% of the balance. Assume also that
you make only the minimum payments.
Previous Minimum Finance New 20. Find a formula for the balance after t monthly payments.
Month balance payment Purchases charge balance 21. Find how many months it takes to bring the balance below
1 $4500.00 $50.
2 22. Balance below $200. You have a balance of $4000 on your
credit card, and you make no more charges. Assume that the
card requires a minimum payment of 5% and carries an APR
10. A balance statement. Assume that you start with a balance
of 22.8%. Assume also that you make only the minimum pay-
of $4500 on your MasterCard. Assume that MasterCard has
ments. Determine when the balance drops below $200.
finance charges of 12% APR and that each month you make
only the minimum payment of 3% of the balance. During the 23–27. New balances. Use the following information for
first month, you charge $300, and during the second month, Exercises 23 through 27. Assume that you have a balance
you charge $600. Complete the following table: of $3000 on your MasterCard and that you make no more
charges. Assume that MasterCard charges 12% APR and that
each month you make only the minimum payment of 5% of the
Previous Minimum Finance New balance.
Month balance payment Purchases charge balance
23. Find a formula for the balance after t monthly payments.
1 $4500.00
2 24. What will the balance be after 30 months?
25. What will the balance be after 10 years?
11–13. New balances. Assume that you have a balance of 26. At what balance do you begin making payments of $20 or
$4500 on your Discover credit card and that you make no more less?
charges. Assume that Discover charges 21% APR and that each 27. Find how many months it will take to bring the remaining
month you make only the minimum payment of 2.5% of the balance down to the value from Exercise 26.
balance. This is the setting for Exercises 11 through 13.
Crauder-4100237 QL November 6, 2014 16:14
284 CHAPTER 4
28–30. Paying off a Discover card. Assume that you have a begin with a balance of $1000. On day 1, you charge $200.
balance of $3000 on your Discover credit card and that you On day 3, you charge $500. On day 6, you make a payment of
make no more charges. Assume that Discover charges 21% $400 and you charge $100.
APR and that each month you make only the minimum pay- 35. Assume that finance charges are calculated using the sim-
ment of 2% of the balance. This is the setting for Exercises 28 plified method shown in this section. Find the account balance
through 30. at the end of the week.
28. Find a formula for the remaining balance after t monthly 36. Assume that finance charges are calculated using the aver-
payments. age daily balance. Find the account balance at the end of the
29. On what balance do you begin making payments of $50 week.
or less? 37. Finance charges versus minimum payments. In all of the
30. Find how many months it will take to bring the remaining examples in this section, the monthly finance charge is always
balance down to the value from Exercise 29. less than the minimum payment. In fact, this is always the case.
31. Monthly payment. You have a balance of $400 on your Explain what would happen if the minimum payment were less
credit card and make no more charges. Assume that the card than the monthly finance charge.
carries an APR of 18%. Suppose you wish to pay off the card 38. A bit of history. Credit cards are relatively new. Write a
in six months by making equal payments each month. What is brief report on the introduction of credit cards into the U.S.
your monthly payment? economy.
32. What can you afford to charge? Suppose you have a new The following exercises are designed to be solved using tech-
credit card with 0% APR for a limited period. The card requires nology such as calculators or computer spreadsheets.
a minimum payment of 5% of the balance. You feel you can 39. Paying off a Visa card----in detail. Assume that you have a
afford to pay no more than $250 each month. How much can balance of $1000 on your Visa credit card and that you make
you afford to charge? How much could you afford to charge if no more charges on the card. Assume that Visa charges 12%
the minimum payment were 2% instead of 5%? APR and that the minimum payment is 5% of the balance each
33–34. Average daily balance. Use the following information month. Assume also that you make only the minimum pay-
for Exercises 33 and 34. The most common way of calculating ments. Make a spreadsheet listing the items below for each
finance charges is not the simplified one we used in this sec- month until the payment falls below $20.
tion but rather the average daily balance. With this method,
we calculate the account balance at the end of each day of the Previous Minimum Finance New
month and take the average. That average is the amount sub- Month balance payment Purchases charge balance
ject to finance charges. To simplify things, we assume that the
1 $1000.00
billing period is one week rather than one month. Assume that 2
the weekly rate is 1%. You begin with a balance of $500. On and so on
day 1, you charge $75. On day 3, you make a payment of $200
and charge $100. On day 6, you charge $200. 40. Paying off an American Express card. Assume that you
33. Assume that finance charges are calculated using the sim- have a balance of $3000 on your American Express credit card
plified method shown in this section. Find the account balance and that you make no more charges on the card. Assume that
at the end of the week. American Express charges 20.5% APR and that the minimum
payment is 2% of the balance each month. Assume also that
34. Assume that finance charges are calculated using the aver-
you make only the minimum payments. Make a spreadsheet
age daily balance. Find the account balance at the end of the
listing the items below for each month until the payment drops
week.
below $50.
35–36. More on average daily balance. Use the following in-
formation for Exercises 35 and 36. The method for calculating Previous Minimum Finance New
finance charges based on the average daily balance is explained Month balance payment Purchases charge balance
in the setting for Exercises 33 and 34. As in those exercises, to 1 $3000.00
simplify things, we assume that the billing period is one week 2
rather than one month and that the weekly rate is 1%. You and so on
Crauder-4100237 QL November 6, 2014 16:14
286 CHAPTER 4
Fed’s extraordinary measures to revive the economy, even though interest rates
are already close to zero.
But as chairwoman, Ms. Yellen will face the arduous task of overseeing the
gradual unwinding of those measures, despite an uncomfortably high unemploy-
ment rate of 7 percent and subdued inflation.
During the confirmation process, senators from both sides of the aisle crit-
icized the Fed for not doing enough to aid the economy and help middle-class
Americans, and for trying to do too much, thus distorting the markets and risking
new bubbles.
In a speech at the University of California Berkeley in November 2012, Ms.
Yellen spoke of pairing the Fed’s inflation goal with a specific employment goal.
“I fear that they are already in way too deep,” said Senator Charles E. Grass-
ley, an Iowa Republican, on the Senate floor, before the confirmation vote. Mr.
Grassley questioned how the Fed would pull back on its recent campaign of large-
scale asset purchases “without spooking investors,” and whether that might
stoke inflation.
Janet L. Yellen faces the delicate task of tapping the brakes on the Fed’s huge
stimulus, but keeping markets from overheating . . .
Despite those objections, Ms. Yellen won confirmation easily. Ms. Yellen was
Janet Yellen, chair of the U.S. approved 56 to 26, with many senators kept away from the Capitol by inclement
Federal Reserve, speaks during a weather.
news conference in Washington,
D.C., on March 19, 2014. The
Federal Reserve said it will look
at a wide range of data in
determining when to raise its
benchmark interest rate from
zero, dropping a pledge tying
borrowing costs to a 6.5 percent This article refers to financial issues such as inflation, financial markets, and a
unemployment rate.
stimulus package that involved tax rebates. We will explore such issues in this section.
288 CHAPTER 4
For example, as of 2013, there were about 57.5 million Social Security benefi-
ciaries and 47.5 million SNAP (food stamp) recipients who are affected by the CPI
because their benefits are adjusted periodically to compensate for inflation. In addi-
tion, millions of military and federal civil service retirees and subsidized lunches at
schools are affected.
Table 4.3 shows the annual change in prices in the United States over a 65-year
period. For example, from December 1945 to December 1946, the CPI changed from
18.2 to 21.5, an increase of 21.5 − 18.2 = 3.3. Now we find the percentage change:
Change in CPI 3.3
Percentage change = × 100% = × 100%
Previous CPI 18.2
This is about 18.1%, and that is the inflation rate for this period shown in the table.
Usually, we round the inflation rate as a percentage to one decimal place.
Crauder-4100237 QL November 6, 2014 16:14
Table 4.4 lists four countries and their estimated annual rates of inflation for
2013. A very high rate of inflation, as in countries like Venezuela and Syria, is referred
to as hyperinflation.
If the rate of inflation is 10%, one may think that the buying power of a dollar
has decreased by 10%, but that is not the case. Inflation is the percentage change in
prices, and that is not the same as the percentage change in the value of a dollar. To
see the difference, let’s imagine a frightening inflation rate of 100% this year. With
such a rate, an item that costs $200 this year will cost $400 next year. This means
that my money can buy only half as much next year as it can this year. So the buying
power of a dollar would decrease by 50%, not by 100%.
The following formula tells us how much the buying power of currency decreases
for a given inflation rate.
100 i
Percent decrease in buying power =
100 + i
Here i is the inflation rate expressed as a percent, not a decimal. Usually we
round the decrease in buying power as a percentage to one decimal place.
Crauder-4100237 QL November 6, 2014 16:14
290 CHAPTER 4
The buying power formula is derived in Algebraic Spotlight 4.6 at the end of this
section.
100 i
Percent decrease in buying power =
100 + i
100 × 5
=
100 + 5
A companion formula to the buying power formula (Formula 4.13) gives the
inflation rate in terms of the percent decrease in buying power of currency.
100B
Percent rate of inflation =
100 − B
In this formula, B is the decrease in buying power expressed as a percent,
not as a decimal.
100B
Percent rate of inflation =
100 − B
100 × 2.5
=
100 − 2.5
The next example covers all of the concepts we have considered so far in this
section.
100 i
Percent decrease in buying power =
100 + i
100 × 42
=
100 + 42
or about 29.6%. The buying power of the Iranian rial decreased by 29.6%.
Crauder-4100237 QL November 6, 2014 16:14
292 CHAPTER 4
e. We want to find the inflation rate for Ethiopia. We know that the reduction
in buying power was 23%, so we use the inflation formula (Formula 4.14) with
B = 23%:
100B
Percent rate of inflation =
100 − B
100 × 23
=
100 − 23
or about 29.9%. Therefore, the rate of inflation in Ethiopia in 2012 was about
29.9%.
Income taxes
Consider the following tax tables for the year 2013 from the Internal Revenue Ser-
vice. Table 4.5 shows tax rates for single people, and Table 4.6 shows tax rates for
married couples filing jointly. Note that the tax rates are applied to taxable income.
The percentages in the tables are called marginal rates, and they apply only to earn-
ings in excess of a certain amount. With a marginal tax rate of 15%, for example, the
tax owed increases by $0.15 for every $1 increase in taxable income. This makes the
tax owed a linear function of the taxable income within a given range of incomes.
The slope is the marginal tax rate (as a decimal).
TABLE 4.6 2013 Tax Table for Married Couples Filing Jointly
Note that these marginal rates increase as you earn more and so move from one
tax bracket to another. A system of taxation in which the marginal tax rates increase
for higher incomes is referred to as a progressive tax.15
15 In 1996 and 2000, magazine publisher Steve Forbes ran for president on a platform that included a 17% “flat tax.”
(The flat-tax concept is explored a bit more in the exercises at the end of this section.) In that system everyone would
pay 17% of their taxable income no matter what that income is. In a speech Forbes said, “When we’re through with
Washington, the initials of the IRS will be RIP.” Forbes did not win the Republican nomination.
Crauder-4100237 QL November 6, 2014 16:14
294 CHAPTER 4
b. In the year 2013, Alice and Bob were married. Together they had a taxable
income of $150,000 per year and filed jointly. How much income tax did they owe?
c. Explain what you notice about the amount of tax paid in the two situations in
parts a and b.
SOLUTION
a. According to the tax table for singles in Table 4.5, Ann owed $4991.25 plus
25% of everything over $36,250, which is $14,678.75 in income tax.
b. According to the tax table for married couples shown in Table 4.6, Alice and
Bob owe $28,457.50 plus 28% of everything over $146,400, which comes to
$29,465.50.
c. If Alice and Bob each had a taxable income of $75,000 and could file as singles,
then, using the solution to part a, they would owe 2 × $14,678.75 = $29,357.50.
But because they were married and filing jointly, they had to pay $29,465.50,
which is an extra $108 in taxes.
Marty Bucella/cartoonstock.com
SOLUTION
The taxable income of Betty is $75,000 − $10,000 = $65,000. According to the tax
table in Table 4.5, Betty owes $4991.25 plus 25% of the excess taxable income over
$36,250. That tax is
$4991.25 + 0.25 × ($65,000 − $36,250) = $12,178.75
Betty has no tax credits, so the tax she owes is $12,178.75.
The taxable income of Carol is $75,000 − $9000 = $66,000. According to the
tax table in Table 4.5, before applying tax credits, Carol owes $4991.25 plus 25%
of the excess taxable income over $36,250. That tax is
$4991.25 + 0.25 × ($66,000 − $36,250) = $12,428.75
Carol has a tax credit of $1000, so the tax she owes is
$12,428.75 − $1000 = $11,428.75
Betty owes
$12,178.75 − $11,428.75 = $750.00
more tax than Carol.
In Example 4.36, the effect of replacing a $1000 deduction by a $1000 credit was
to reduce the tax owed by $690. This is a significant reduction in taxes and highlights
the benefits of tax credits.
The Dow
In the late nineteenth century, tips and gossip caused stock prices to move because
solid information was hard to come by. This prompted Charles H. Dow to introduce
the Dow Jones Industrial Average (DJIA) in May 1896 as a benchmark to gauge the
state of the market. The original DJIA was simply the average price of 12 stocks
that Mr. Dow picked himself. Today the Dow, as it is often called, consists of 30
“blue-chip” U.S. stocks picked by the editors of the Wall Street Journal. For example,
ZAK BRIAN/SIPA/AP Photo
296 CHAPTER 4
in June 2009 General Motors was removed from the list as it entered bankruptcy
protection. Here is the list as of June 2014.
As we said earlier, the original DJIA was a true average—that is, you simply
added up the stock prices of the 12 companies and divided by 12. In 1928 a divi-
sor of 16.67 was used to adjust for mergers, takeovers, bankruptcies, stock splits,
and company substitutions. Today, they add up the 30 stock prices and divide by
0.15571590501117 (the divisor), or equivalently, multiply by 1/0.15571590501117
or about 6.42. This means that for every $1 move in any Dow company’s stock price,
the average changes by about 6.42 points. (The DJIA is usually reported using two
decimal places.)
The graph in Figure 4.14 shows how the Dow has moved over the last several
decades. In Exercises 36 through 39, we explore a few of the more common types of
stock transactions.
Crauder-4100237 QL November 13, 2014 11:47
16,000
14,000
12,000
10,000
8,000
6,000
4,000
2,000
0
1930 1934 1938 1942 1946 1950 1954 1958 1962 1966 1970 1974 1978 1982 1986 1990 1994 1998 2002 2006 2010 2014
FIGURE 4.14 This graph shows the movement of the Dow from October 1, 1928
through December 26, 2013.
Suppose the inflation rate is i% per year. We want to derive the buying power
formula:
100 i
Percent decrease in buying power =
100 + i
Suppose we could buy a commodity, say, one pound of flour, for one dollar a
year ago. An inflation rate of i% tells us that today that same pound of flour
would cost 1 + i/100 dollars. To find the new buying power of the dollar, we
need to know how much flour we could buy today for one dollar. Because one
pound of flour costs 1 + i/100 dollars, one dollar will buy
1 100
= pounds of flour
1 + i/100 100 + i
This quantity represents a decrease of
100 i
1− = pounds
100 + i 100 + i
from the one pound we could buy with one dollar a year ago. The percentage
decrease in the amount of flour we can buy for one dollar is
i
Decrease 100+i 100 i
Percentage decrease = × 100% = × 100% =
Previous value 1 100 + i
This is the desired formula for the percentage decrease in buying power.
298 CHAPTER 4
Questions of this sort may invite many correct responses. We present one possibil-
ity: Inflation causes the cost of typical consumer items to rise. If income remains
constant while inflation drives up prices, many who depend on Social Security will
suffer. Increasing benefits to match inflation means that Social Security maintains a
constant value to recipients. Other federal programs with benefits tied to inflation in-
clude military retirement, Veterans Disability Compensation, food stamps, Medicaid
eligibility, and many more.
Further questions of this type may be found in the What Do You Think section of
exercises.
300 CHAPTER 4
34. Average price. What is the average price of a share of stock 39. Selling short. Selling short is the selling of stock you do
in the Dow list when the DJIA is 10,000? not actually own but promise to deliver. Suppose you place an
35. Dow highlights. Use Figure 4.14 to determine in approxi- order to sell short 100 shares of stock X at $35. Eventually,
mately what year the DJIA first reached 5000. About when did your order must be covered. That is, you must sell 100 shares
it first reach 10,000? of stock X at a value of $35 per share. If when the order is cov-
ered, the value of the stock is less than $35 per share, then you
Exercises 36 through 39 are suitable for group work. make money; otherwise, you lose money. Suppose that on the
36–39. Stock market transactions. There are any number of day you must cover the sell short order, the price of stock X is
ways to make (or lose) money with stock market transactions. $50 per share. How much money did you lose?
In Exercises 36 through 39, we will explore a few of the more 40. History: More on selling short. In 1992 George Soros
common types of transactions. (Fees for such transactions will “broke the Bank of England” by selling short the British pound.
be ignored.) Write a brief report on his profit and exactly how he managed
36. Market order. The simplest way to buy stock is the mar- to make it.
ket order. Through your broker or online, you ask to buy 100 41. History: The Knights Templar. The Knights Templar was
shares of stock X at market price. As soon as a seller is located, a monastic order of knights founded in 1112 to protect pil-
the transaction is completed at the prevailing price. That price grims traveling to the Holy Land. Recent popular novels have
will normally be very close to the latest quote, but the prevailing revived interest in them. Some have characterized the Knights
price may be different if the price fluctuates between the time Templar as the first true international bankers. Report on the
you place the order and the time the transaction is completed. international aspects of their early banking activities.
Suppose you place a market order for 100 shares of stock X
and the transaction is completed at $44 per share. Two weeks 42. History: The stock market. The Dow Jones Industrial Av-
later the stock value is $58 per share and you sell. What is your erage normally fluctuates, but over the last half-century it has
net profit? generally increased. Dramatic drops in the Dow (stock market
crashes) can have serious effects on the economy. Report on
37. Limit orders. If you want to insist on a fixed price for a some of the most famous of these. Be sure to include the crash
transaction, you place a limit order. That is, you offer to buy of 1929.
(or sell) stock X at a certain price. If the stock can be purchased
for that price, the transaction is completed. If not, no transac- 43. History: The Federal Reserve Bank. The Federal Reserve
tion occurs. Often limit orders have a certain expiration date. Bank is an independent agency that regulates various aspects
Suppose you place a limit order to buy 100 shares of stock X of American currency. Write a report on the Federal Reserve
for $40 per share. When the stock purchase is completed, you Bank. Your report should include the circumstances of its
immediately place a limit order to sell stock X at $52 per share. creation.
The following table shows the value of stock X. On which days 44. History: The SEC. The Securities and Exchange Commis-
are these two transactions completed, and what is your profit? sion regulates stock market trading in the United States. Write
a report on the creation and function of the SEC.
Date Day 1 Day 2 Day 3 Day 4 Day 5 Day 6
Market price $44 $45 $40 $48 $52 $50 The following exercises are designed to be solved using tech-
nology such as calculators or computer spreadsheets.
38. Stop loss and trailing stops. If you own a stock, a stop loss 45–48. Mortgage interest deduction. Use the following in-
order protects you from large losses. For example, if you own formation for Exercises 45 through 48. Interest paid on a home
100 shares of stock X, bought at $45 per share, you might place mortgage is normally tax deductible. That is, you can subtract
a stop-loss order for $40 per share. This order automatically the total mortgage interest paid over the year in determining
sells your stock if the price drops to $40. No matter what hap- your taxable income. This is one advantage of buying a home.
pens, you can’t lose more than $5 per share. A similar type of Suppose you take out a 30-year home mortgage for $250,000
order that protects profits is the trailing stop. The trailing-stop at an APR of 8% compounded monthly.
order sells your stock if the value goes below a certain percent-
age of the market price. If the market price remains the same or 45. Determine your monthly payment using the monthly pay-
drops, the trailing stop doesn’t change and acts like a stop-loss ment formula in Section 4.2.
order. If, however, the market price goes up, the trailing stop 46. Make a spreadsheet showing, for each month of the first
follows it so that it protects profits. Suppose, for example, that year of your payment, the amount that represents interest, the
you own 100 shares of stock X, that the market price is $40 amount toward the principal, and the balance owed.
per share, and that you place a trailing-loss order of 5%. If the
47. Use the results of Exercise 46 to find the total interest paid
price drops by $2 (5% of $40), the stock is sold. If the market
over the first year. (Round your answer to the nearest dollar.)
value increases to $44, then you will sell the stock when it de-
That is what you get to deduct from your taxable income.
clines by 5% of $44. That is, you sell when the market price
drops to $41.80. Consider the table in Exercise 37, and sup- 48. Suppose that your marginal tax rate is 27%. What is your
pose that you purchase 100 shares of the stock and place a 5% actual tax savings due to mortgage payments? Does this make
trailing-stop order on day 1. On which day (if any) will your the $250,000 home seem a bit less expensive?
stock be sold?
Crauder-4100237 QL November 6, 2014 16:14
CHAPTER SUMMARY
This chapter is concerned with financial transactions of two basic kinds: saving and
borrowing. We also consider important financial issues related to inflation, taxes,
and the stock market.
Simple interest earned = Principal × Yearly interest rate (as a decimal) × Time in years
Financial institutions normally compound interest and advertise the annual percent-
age rate or APR. The interest rate for a given compounding period is calculated using
APR
Period interest rate =
Number of periods in a year
Crauder-4100237 QL November 6, 2014 16:14
302 CHAPTER 4
We can calculate the account balance after t periods using the compound interest
formula:
Balance after t periods = Principal × (1 + r )t
Here, r is the period interest rate expressed as a decimal, and it should not be rounded.
In fact, it is best to do all the calculations and then round.
The annual percentage yield or APY is the actual percentage return in a year. It
takes into account compounding of interest and is always at least as large as the APR.
If n is the number of compounding periods per year,
APR n
APY = 1 + −1
n
Here, both the APR and the APY are in decimal form. The APY can be used to
calculate the account balance after t years:
Balance after t years = Principal × (1 + APY)t
Here, the APY is in decimal form.
The present value of an investment is the amount we initially invest. The future
value is the value of that investment at some specified time in the future. If the in-
vestment grows by compounding of interest, these two quantities are related by the
compound interest formula. We can rearrange that formula to give the present value
we need for a desired future value:
Future value
Present value =
(1 + r )t
In this formula, t is the total number of compounding periods, and r is the period
interest rate expressed as a decimal.
The Rule of 72 can be used to estimate how long it will take for an account
growing by compounding of interest to double in size. It says that the doubling time
in years can be approximated by dividing 72 by the APR, where the APR is expressed
as a percentage, not as a decimal. The exact doubling time can be found using the
formula:
log 2
Number of periods to double =
log(1 + r )
Here, r is the period interest rate as a decimal.
is at least as large as the principal times the monthly interest rate as a decimal. This
number can be used to estimate the monthly payment for a long-term loan with a
moderate or high interest rate.
A record of the repayment of a loan is kept in an amortization table. In the case
of buying a home, an important thing for the borrower to know is how much equity
he or she has in the home. The equity is the total amount that has been paid toward
the principal, and an amortization table keeps track of this amount.
Some home loans are in the form of an adjustable-rate mortgage or ARM, where
the interest rate may vary over the life of the loan. For an ARM, the initial rate is
often lower than the rate for a comparable fixed-rate mortgage, but rising rates may
cause significant increases in the monthly payment.
304 CHAPTER 4
the balance is a decreasing exponential function, the balance decreases very slowly
in the long run.
KEY TERMS
principal, p. 216 Rule of 72, p. 228 prime interest rate, p. 247
simple interest, p. 216 installment loan, p. 235 nest egg, p. 263
compound interest, p. 217 term, p. 235 monthly yield, p. 263
period interest rate, p. 219 amortization table (or amortization annuity, p. 264
annual percentage rate (APR), p. 219 schedule), p. 241 Consumer Prime Index (CPI), p. 287
annual percentage yield (APY), equity, p. 242 inflation, p. 287
p. 223 fixed-rate mortgage, p. 247 rate of inflation, p. 287
present value, p. 227 adjustable-rate mortgage (ARM), deflation, p. 287
future value, p. 227 p. 247
CHAPTER QUIZ
1. We invest $2400 in an account that pays simple interest of 3. We have an account that pays an APR of 9.75%. If interest
8% each year. Find the interest earned after five years. is compounded quarterly, find the APY. Round your answer as
Answer $960 a percentage to two decimal places.
Answer 10.11%
If you had difficulty with this problem, see Example 4.1.
2. Suppose it were possible to invest $8000 in a four-year CD If you had difficulty with this problem, see Example 4.4.
that pays an APR of 5.5%. 4. How much would you need to invest now in a savings ac-
a. What is the value of the mature CD if interest is compounded count that pays an APR of 8% compounded monthly in order
annually? to have a future value of $6000 in a year and a half?
b. What is the value of the mature CD if interest is com- Answer $5323.64
pounded monthly?
If you had difficulty with this problem, see Example 4.7.
Answer a. $9910.60, b. $9963.60
5. Suppose an account earns an APR of 5.5% compounded
If you had difficulty with this problem, see Example 4.3. monthly. Estimate the doubling time using the Rule of 72, and
Crauder-4100237 QL November 6, 2014 16:14
then calculate the exact doubling time. Round your answers to 11. Suppose your MasterCard calculates finance charges using
one decimal place. an APR of 16.5%. Your previous statement showed a balance
Answer Rule of 72: 13.1 years; exact method: 151.6 months of $400, toward which you made a payment of $100. You then
(about 12 years and 8 months) bought $200 worth of clothes, which you charged to your card.
Complete the following table:
If you had difficulty with this problem, see Example 4.8.
6. You need to borrow $6000 to buy a car. The dealer offers Previous Finance New
an APR of 9.25% to be paid off in monthly installments over balance Payments Purchases charge balance
2 12 years.
a. What is your monthly payment? Month 1
Nikulin Vyacheslav/Itar-Tass/Landov
Crauder-4100237 QL November 6, 2014 12:59
5 INTRODUCTION TO
PROBABILITY
T
he following excerpt is from an article that appeared at the Web site of the
American Association for the Advancement of Science (AAAS). It refers to the
Chelyabinsk meteor, an asteroid that exploded near Chelyabinsk, Russia, on
5.1 Calculating February 15, 2013, and released energy about 20 times that released by the first
probabilities: How likely atomic bomb.
is it?
5.2 Medical testing and
conditional probability: IN THE NEWS 5.1 Search
Ill or not? Home TV & Video U. S. World Politics Entertainment Tech Travel
This article both warns and reassures us. The article states that there is a pos-
sibility that an object will strike Earth sometime in the distant future but that no
such event has a significant probability within the next century. All of NASA’s highly
trained scientists, mathematicians, and engineers with access to NASA’s state-of-the-
art equipment cannot tell us for sure whether such an event will occur. What are we
to make of such information? Probabilities of this sort apply not only to asteroids
but also to hurricanes, rainstorms, traffic accidents, sports contests, and winning the
lottery. In this chapter, we examine the idea of probability, what it means, and how
probabilities are calculated.
307
Crauder-4100237 QL November 6, 2014 12:59
308 CHAPTER 5
I n April 2008, scientists and engineers released a new earthquake forecast for
the State of California called the Uniform California Earthquake Rupture Fore-
cast (UCERF). Compiled by USGS, Southern California Earthquake Center (SCEC),
and the California Geological Survey (CGS), with support from the California
Earthquake Authority, it updates the earthquake forecast made for the greater
San Francisco Bay Area by the 2002 Working Group for California Earthquake
Probabilities.
1 A 6.0 magnitude earthquake did in fact occur in Napa, California, on August 24, 2014.
Crauder-4100237 QL November 6, 2014 12:59
Theoretical probability
To discuss theoretical probability, we need to make clear what we mean by an event.
An event is a collection of specified outcomes of an experiment.2 For example, sup-
pose we roll a standard die (with faces numbered 1 through 6). One event is that
the number showing is even. This event consists of three outcomes: a 2 shows, a 4
shows, or a 6 shows.
To define the probability of an event, it is common to describe the outcomes
corresponding to that event as favorable and all other outcomes as unfavorable. If
the event is that an even number shows when we toss a die, there are three favorable
outcomes (2, 4, or 6) and three unfavorable outcomes (1, 3, or 5).
2 The situations we will consider all have a finite number of outcomes, so we are considering only discrete probabilities.
Crauder-4100237 QL November 6, 2014 12:59
310 CHAPTER 5
SOLUTION
There are 52 different cards that I might
draw, and these outcomes are equally likely.
There are four aces, so four outcomes are fa-
vorable. Thus,
Favorable outcomes 4 1
P(Ace) = = =
Total outcomes 52 13
Distinguishing outcomes
One needs to be careful in determining what constitutes an outcome. To illustrate
this point, consider a committee consisting of three members. They need to send one
member as a representative to a meeting. They decide to choose the representative
by writing their three names on slips of paper and drawing a slip at random from a
basket.
Suppose first that the names of the members are Alan, Benny, and Jerry. What is
the probability that Alan is chosen? There are three equally likely outcomes: Alan,
Benny, and Jerry. Only one of these outcomes is favorable (Alan). Thus,
Favorable outcomes 1
P(Alan) = =
Total outcomes 3
Now suppose that Jerry is replaced by another member whose name is also Alan.
Thus, two of the three committee members are now named Alan, and one is named
Benny. What is the probability that someone named Alan is chosen in this case?
To understand this situation, we imagine that members put their surnames on
the slips, say, Alan Jones and Alan Smith. Now it is easy to see that there are three
312 CHAPTER 5
equally likely outcomes: Alan Jones, Alan Smith, and Benny. Two of these outcomes
are favorable, so the probability of selecting a member named Alan is 2/3.
This example warns of a common error in calculating probabilities: If I choose
a name from the basket, I will get either Alan or Benny. The temptation is to say
that the probability of getting Alan is 1/2. The error here is that although one can
think of having two outcomes, Alan and Benny, they are not equally likely. In fact,
because there are two Alans in the hat and only one Benny, we are twice as likely to
draw Alan as Benny. Often this sort of error is easy to spot, but we must be careful to
avoid it. We can often avoid difficulties by distinguishing outcomes just as we used
surnames, Jones and Smith, to distinguish the two Alans.
So let’s imagine that one of the coins is a nickel and the other is a dime.
First we must figure out how many equally likely outcomes there are. Either coin
can come up as a head (H) or a tail (T). All possibilities are listed in the table below.
Nickel Dime
H H
H T
T H
T T
There are four equally likely outcomes: HH, HT, TH, TT. Only one of them gives
two heads, so the probability of getting two heads is
Favorable outcomes 1
P(HH) = =
Total outcomes 4
These examples show how distinguishing things (Alan Jones and Alan Smith, or
nickel and dime) that appear to be the same can ensure that all outcomes are equally
likely.
To work To home
Stop Stop
Stop Don’t stop
Don’t stop Stop
Don’t stop Don’t stop
This means that there are four equally likely outcomes: Stop–Stop, Stop–Don’t
stop, Don’t stop–Stop, Don’t stop–Don’t stop. Three of these yield the result of
stopping at least once, so the desired probability is
Favorable outcomes 3
=
Total outcomes 4
b. One of the four possible outcomes (Don’t stop–Don’t stop) corresponds to
not having to stop at all, so the desired probability is
Favorable outcomes 1
=
Total outcomes 4
Probability of non-occurrence
The calculations in Example 5.3 lead us to something important. There we found
3
P(Stopping at least once) =
4
1
P(Not stopping at all) =
4
Crauder-4100237 QL November 6, 2014 12:59
314 CHAPTER 5
That is, the probability of not stopping at all is 1 minus the probability of stopping
at least once. This is no accident, as the following observations show: Recall that
Favorable outcomes
Probability of event occurring =
Total outcomes
Then
Unfavorable outcomes
Probability of event not occurring =
Total outcomes
If we add the favorable and unfavorable outcomes, we get all outcomes. Hence,
Total outcomes
Probability of event occurring + Probability of event not occurring = =1
Total outcomes
Rearranging this equation yields the formula in Summary 5.2.
Now we use the idea of distinguishing items in pairs with the formula for non-
occurrence.
Now we need to know how many outcomes yield a 7 (what we would call a
favorable outcome). We list the possibilities in a table.
Note that if you tried to compute this probability directly, you would have to
count all the ways of getting a sum different from 7, which would be very tedious.
Probability of disjunction
Our first calculation in this section concerned the probability that when we roll a
six-sided die, we will get a 1 or a 6. This is an example of finding the probability that
either one event or another occurs. In accordance with the terminology of formal
logic (see Section 1.3), we call this the probability of the disjunction of the two events.
It turns out that there is a convenient formula for finding this probability.
4 In Section 5.3, we explain in more detail how to find the total number of outcomes in such situations.
Crauder-4100237 QL November 6, 2014 12:59
316 CHAPTER 5
The rationale for this formula comes from the fact that if you add the probability
of event A and the probability of event B, you are counting the event “A and B”
twice. (Recall that in mathematics the disjunction allows for the possibility that both
events occur.) Therefore, subtracting P(A and B) from the sum gives the correct
answer.
asteroids striking Earth. Such estimates relate probability to area. To explain the
connection, we examine the probability related to a man made object striking Earth.
In 1979, Skylab, the first space station of the United States, reentered Earth’s
atmosphere. At the time there was widespread concern that the debris would fall in a
heavily populated area, but in the end the area most affected was a sparsely populated
region in western Australia. The area was roughly rectangular in shape. If pieces of
Skylab fell at random over this rectangle, what is the probability that a given region
would be affected? To see how to answer this question, we first consider the simpler
situation in which the rectangle is divided into three regions, designated by color as
follows: 1/4 of the area is red, 1/4 is blue, and 1/2 is white. See Figure 5.1. If a given
small piece of Skylab fell at random onto this rectangle, what is the probability that
it would land on the color blue?
NASA
There are three possible outcomes: The piece lands on red, blue, or white. These
outcomes are not equally likely, however. Because a lot more of the target is white,
one would expect that there is a greater likelihood of the piece landing on white
than on either of the other two colors. One way of looking at this problem is to
cut the white part into two sections, White 1 and White 2, as shown in Figure 5.2.
Now because each of the four sections is the same size, it should be equally likely
that the piece lands in any one of them. Because there are now four equally likely
FIGURE 5.1 Rectangle divided into FIGURE 5.2 Cutting white into two
three regions. sections.
Crauder-4100237 QL November 6, 2014 12:59
318 CHAPTER 5
outcomes and only one of them is blue, we see that the probability of landing on blue
is 1/4 = 0.25 = 25%. Observe that this number equals the percentage of the area
that was labeled blue.
This observation holds in general. The probability of landing on a given color is
the fraction of the area covered by that color. In this case, that means the probability
of landing on white is 1/2, the probability of landing on red is 1/4, and the probabil-
ity of landing on blue is 1/4. As a practical matter, this means that an object falling to
Earth at random is likely to strike an unpopulated region: Water covers about 70%
of Earth’s surface.
SOLUTION
The total area covered by North and South America is 9.37 + 6.88 = 16.25 million
square miles. As a fraction of the surface area of Earth, that is 16.25/197 or about
0.082. Hence, the probability that the meteor strikes North or South America is
0.082 or 8.2%.
The idea seems simple enough, but the important question about an empirical
probability is, how meaningful is it? For example, if I select three jelly beans from a
jar and two of them are red while one is not, I surely can’t deduce from this single
experiment that exactly 2/3 of the jelly beans in the jar are red. However, suppose
we perform this experiment numerous times and in 95% of them about 2/3 of the
5 These numbers are far from acceptable for any reputable producer, but we use them for illustration.
Crauder-4100237 QL November 6, 2014 12:59
320 CHAPTER 5
jelly beans we selected were red. Then, we might start believing that about 2/3 of
jelly beans in the jar really are red.
So one question is, how many samples should we take before we have some con-
fidence in our empirical probability? And is there some way to measure a level of
confidence? As it turns out, these questions are at the heart of what professional
polling organizations do. We will return to these questions in Section 3 of Chapter 6.
1432
16,652
Try It Yourself 5.7: Finding probability using area: Meteor 0.913 or 91.3%.
322 CHAPTER 5
24–34. Jelly beans. Exercises 24 through 34 refer to exper- 43–44. A box. Suppose the inside bottom of a box is painted
iments of selecting jelly beans from groups of varying sizes. with three colors: 1/3 of the bottom area is red, 1/6 is blue,
For exercises 24 and 25, suppose there are five jelly beans in and 1/2 is white. You toss a tiny pebble into the box without
a box—two red and three green. aiming. Exercises 43 and 44 refer to this.
24. If a jelly bean is selected at random, what is the probability 43. What is the probability that the pebble lands on the color
that it is red? white?
25. A friend claims that the answer to Exercise 24 is 1/2 be- 44. What is the probability that the pebble does not land on
cause there are two outcomes, red and green. Explain what’s the color blue?
wrong with that argument. 45. Heart disease. In one study, researchers at the Cleveland
For Exercises 26 through 30, use the following information. Clinic looked at more than 120,000 patients enrolled in 14 in-
Suppose there are 15 jelly beans in a box—2 red, 3 blue, 4 ternational studies in the past 10 years. Among the patients
white, and 6 green. A jelly bean is selected at random. with heart problems, researchers found at least one risk factor
26. What is the probability that the jelly bean is red? in 84.6% of the women and 80.6% of the men.
If a woman is selected at random from the women with
27. What is the probability that the jelly bean is not red? heart problems, what is the probability she has none of the
28. What is the probability that the jelly bean is blue? risk factors? If a man is selected at random from the men with
heart problems, what is the probability he has none of the risk
29. What is the probability that the jelly bean is red or blue?
factors? Write both of your answers as percentages.
30. What is the probability that the jelly bean is neither red nor
46–48. Health. Exercises 46 through 48 refer to the accom-
blue?
panying table. The percentages are probabilities of 15-year sur-
For Exercises 31 through 34, use the following information. vival without coronary heart disease, stroke, or diabetes in men
Suppose I pick a jelly bean at random from a box containing aged 50 years with selected combinations of risk factors. Here
five red and seven blue ones. I record the color and put the jelly “BMI” stands for body mass index, an index of weight adjusted
bean back in the box. for height. The World Health Organization gives the following
31. What is the probability of getting a red jelly bean both classifications: “normal” 18.5 to 24.9; “grade 1 overweight”
times if I do this twice? 25.0 to 29.9; “grade 2 overweight” 30.0 to 39.9; “grade 3
overweight” 40.0 or higher.
32. What is the probability of getting a blue jelly bean both
times if I do this twice?
BMI 20−24 BMI 25−30 BMI 30+
33. If I do this three times, what is the probability of getting a normal somewhat very
red jelly bean each time? weight overweight overweight
34. If I do this three times, what is the probability of getting a Active Inactive Active Inactive Active Inactive
blue jelly bean each time?
Never 89% 83% 85% 77% 78% 67%
35–39. Tossing three coins. Exercises 35 through 39 refer to smoked
the experiment of tossing three coins. The answers to some of Former 86% 78% 81% 71% 72% 59%
them can be easily obtained from answers to earlier exercises. smokers
35. What is the total possible number of outcomes? Current 77% 67% 70% 56% 58% 42%
smokers
36. What is the probability of getting three heads?
37. What is the probability of getting one head and two tails? 46. What is the probability that a 50-year-old man will not
38. What is the probability of getting anything but one head survive to age 65 without coronary heart disease, stroke, or
and two tails? diabetes if he smokes, is somewhat overweight, but is active?
Write your answer as a percentage.
39. What is the probability of getting at least one tail?
47. A 50-year-old man who has never smoked, has normal
40. Tossing three dice. You toss three dice and add up the faces
weight, and is active has a much higher probability of being
that show. What is the probability that the sum will be 3, 4,
healthy than a man who smokes, is very overweight, and is in-
or 5?
active. Is the probability twice as high, less than twice as high,
41. Passing. The probability of passing the math class of or more than twice as high?
Professor Jones is 62%, the probability of passing Profes- 48. A man who used to smoke and who is very overweight
sor Smith’s physics class 34%, and the probability of passing and inactive is considering either starting an exercise program
both is 30%. What is the probability of passing one or the or dieting to lose enough weight to be somewhat overweight—
other? but not both. According to the table, which choice would be
42. The team. The probability of making the basketball team better?
is 32%, the probability of making the softball team is 34%, 49. Skiing. Records kept by the National Ski Patrol and the
and the probability of making one or the other is 50%. What National Safety Council show that, over a period of 14 years,
is the probability of making both teams? skiers made 52.25 million visits to the slopes annually, with
Crauder-4100237 QL November 6, 2014 12:59
an average of 34 deaths each year. Assuming that these figures 52. Terrorism. The accompanying excerpt is from an article at
are current, what is the probability that a skier will die in a the Web site of the Daily Mail. Based on it, would you say that
skiing accident in a given year? Is your answer empirical or real probabilities are being used or just educated opinions? Ex-
theoretical? plain your reasoning.
50. Weather. What does a meteorologist mean when he or she
says there’s a 30% chance of rain tomorrow? It means that the
meteorologist is predicting that about 30% of the land area
covered by the forecast (say, northeast Texas) will receive some
rain during the time period of the forecast (which is usually IN THE NEWS 5.4 Search
324 CHAPTER 5
favor of drawing an ace? What are the odds against drawing you point toward a door,’’ the host says, “then I’ll open
an ace? one of the other doors to reveal a goat. After I’ve shown
55. If we flip a pair of fair coins, what is the probability of you the goat, you may decide whether to stick with your
getting one head and one tail? What are the odds in favor of initial choice of doors, or switch to the other remain-
getting one head and one tail? ing door. You win whatever is behind the door [you
choose].”6
56. Alaska comprises about 1/5 of the total surface area of the
United States. If a meteor lands at some random point in the The question is, should you switch doors or stay with your
United States, what is the probability that it will land in Alaska? original choice? Or does it matter? The fun thing about this
What are the odds in favor of it landing in Alaska? problem is that a number of PhD mathematicians got the wrong
answer!
57. The weatherman says that there is a 30% chance of rain in
We will present two analyses below. One is right and the
your area. What are the odds against your getting rain?
other is wrong. Which is right? Explain why you chose the an-
58. If an event has a probability of 10%, what are the odds in swer you did. If you can’t decide which is right and which is
favor of it, and what are the odds against it? wrong, you should try the experiment suggested in the next
59–61. More odds. Use the following information for Ex- exercise.
ercises 59 through 61. We can calculate odds if we know the Analysis number 1: It doesn’t matter whether you switch
probability. But we can also go the other way. Suppose the odds The probability that I guessed right to begin with is 1/3. Either
in favor of an event are 3 to 2. We can interpret this as saying of the two remaining doors is equally likely to be a winning
that there are three (equally likely) favorable outcomes and two door or a losing door. One of them is certain to be a loser, and
unfavorable ones. That is a total of five outcomes. So the prob- by telling me that one of the doors is a loser, the host has given
ability of the event occurring is 3/5. In general, if the odds in me no new information on which to base my decision. There-
favor are p to q, then the probability is p/( p + q). fore, it makes no difference whether I change or not. In either
case, my probability of winning is 1/3.
59. Suppose the odds in favor of an event are 5 to 2. What is Analysis number 2: It does matter. You should switch doors
the probability that the event will occur? What is the probabil- If I have initially chosen the right door, then changing will cause
ity that the event will not occur? me to lose. This happens one time out of three.
60. Suppose the odds against an event are 2 to 1. What is the Suppose I have chosen one of the losing doors and the host
probability that the event will occur? What is the probability shows me the other losing door. Then if I change, I win. This
that the event will not occur? happens two times out of three.
So the strategy of changing allows me to win two times out
61. Odds in a horse race are, in fact, a reflection of money bet
of three. My probability of winning is increased from 1/3 to
on the various horses rather than a statement of probability.
2/3.
But suppose for the moment that the money bet does indeed
indicate the true mathematical odds. If Whirlaway is a 6 to 5 63. Experimenting with the Monty Hall problem. This exer-
favorite to win, what is the probability that Whirlaway actually cise is a continuation of Exercise 62. With the help of one
wins the race? other person, it is not difficult to simulate the Monty Hall prob-
lem described earlier. Let one person act as host and hide pic-
Exercises 62 through 64 involve hands-on experiments and are
tures of one car and two goats under sheets of paper numbered
suitable for group work.
1 through 3. You be the contestant. Make your guesses and
62. The Monty Hall problem. Monty Hall was the host of record your wins and losses if you never change and if you al-
the popular game show Let’s Make a Deal. At the end of ways change. If you repeat the game 100 times, you should
each show, some lucky contestant was given the opportunity to have fairly good empirical estimates of the probability of win-
choose one of three doors. Behind one door was a nice prize, ning using either strategy. Do the results make you change your
and behind the other two were lesser prizes or even worthless mind about your answer to Exercise 62?
items. The contestant had no information about which door
64. An experiment---Buffon’s needle problem. The French nat-
might hide the prize, so his or her probability of getting the
uralist Buffon posed the following problem—which he sub-
nice prize was one-third.
sequently solved, and which bears his name today. Suppose
An interesting variant of this scenario gained notoriety
a floor is marked by parallel lines, each a distance D apart.
when columnist Marilyn vos Savant posed a related question
(Think of a hardwood floor.) Suppose we toss a needle of length
in her Parade Magazine article of September 9, 1990. Here is
L to the floor. (We assume that the length L of the needle
the question as stated by Mueser and Granberg in 1999.
is smaller than the distance D between the parallel lines on
the floor.) What is the probability that the needle will land
A thoroughly honest game-show host has placed a car
touching one of the parallel lines? It turns out that the correct
behind one of three doors. There is a goat behind each
answer is
of the other two doors. You have no prior knowledge
that allows you to distinguish among the doors. “First 2L
Probability =
πD
6 Copyright © 1990 Marilyn vos Savant. Initially published in Parade Magazine. All rights reserved. Reprinted by permission.
Crauder-4100237 QL November 6, 2014 12:59
One may be surprised that the number π turns up in the an- 66. History. The formulation of the modern abstract theory of
swer, but this fact leads us to an interesting idea that can (at probability is usually attributed to the Russian mathematician
least theoretically) give a novel method of approximating the A. N. Kolmogorov (1903–1987). He lived during the reign of
decimal value of π . Get on a nice, level hardwood floor and the brutal dictator Joseph Stalin and once wrote a paper contra-
toss a needle onto the floor 100 times. Record the number n of dicting the academician Lysenko, who was popular with Stalin.
times the needle touches one of the parallel lines. Then n/100 This was a very brave act indeed. Write a paper about the life
is an empirical estimate of the probability that the needle will and work of Kolmogorov.
touch a line. This gives The following exercises are designed to be solved using tech-
n 2L nology such as calculators or computer spreadsheets.
≈
100 πD 67. Random numbers. Simulate flipping a coin in the following
(Here, we use the symbol ≈ to indicate approximate equality.) way: Use a random-number generator to obtain a random se-
Rearranging gives quence of 0’s and 1’s, with, say, 100 numbers altogether. Count
how many 0’s there are altogether. Is the result close to 50%?
200L
π≈ Is it exactly 50%? Does the result give a theoretical or an em-
nD pirical probability? Note: Computer programs do not actually
Perform this experiment to get an estimate for the value of π. produce random numbers, in spite of the name, but they are
(Do not be disappointed if your answer is not very close to the close enough for our purposes.
correct value. The method is correct, but in practice, it takes a
68. More random numbers. Think of the real numbers be-
lot of tosses to get anywhere close to π .)
tween 0 and 10 as divided into two colors: red for those at least
65. History. Pierre de Fermat and Blaise Pascal were French 0 and at most 6, blue for those greater than 6 and at most 10.
mathematicians and philosophers who lived in the seventeenth Use a random-number generator to obtain a random sequence
century. They are credited with some of the early development of real numbers between 0 and 10, with, say, 50 numbers alto-
of probability theory. Pascal was an interesting figure who was gether. (Be sure to get a random sequence of real numbers, not
also very much involved with religion. Write a paper about just a sequence of whole numbers.) Count how many “red”
Pascal’s life and beliefs. (Exercise 65 in Section 5.3 describes numbers there are altogether. What percentage of red numbers
a specific question about probability that attracted the interest would you expect to get? Is the result of the experiment close to
of Pascal, and Exercise 50 in Section 5.5 is concerned with his this?
use of a concept from probability in a religious context.)
W-2. UFO? People sometimes attach a probability to an event that has never oc-
curred. An example might be, “There’s a 20% chance that a UFO will land in the
next year.” Can you say what kind of probability this is: empirical, theoretical, or
someone’s opinion? Explain your answer. Find additional examples of probabilities
of this same type.
Exercise 7 provides additional information relevant to this topic.
W-3. The euro. A news article from June 2012 begins, “There is a one-in-three
chance Greece will leave the euro currency union in the months ahead, according
to Standard & Poor’s.”7 What kind of probability is this: empirical, theoretical, or
someone’s opinion? Explain your answer. Find additional examples of probabilities
of this same type.
Exercise 9 provides additional information relevant to this topic.
7 Ben Rooney, S&P: 1-in-3 chance Greece leaves euro. CNNMoneyInvest, June 4, 2012.
Crauder-4100237 QL November 6, 2014 12:59
326 CHAPTER 5
W-4. Large percentage. A gloomy friend says that there is a 110% chance it will rain
on his wedding day. Does this statement make sense? What do you think he means?
Exercise 16 provides additional information relevant to this topic.
W-5. Making the team. At a certain school the probability of making the soccer
team is 50%, and the probability of making the basketball team is 50%. A student
concludes that it’s a sure bet to make one of those two teams because 50% + 50% =
100%. Do you agree with this reasoning? Explain your answer.
Exercise 42 provides additional information relevant to this topic.
W-6. Horses. On my farm there are 10 brown horses and 8 mares. Can you conclude
that there are 18 horses on my farm? Explain. How does this example relate to
calculating the probability of a disjunction?
Exercise 41 provides additional information relevant to this topic.
W-8. The weather. The morning weather forecast is for a 70% chance of rain today.
You pack your umbrella, but it does not rain. Was the forecast wrong? Explain what
was meant by the rain prediction.
Exercise 50 provides additional information relevant to this topic.
W-9. Where to live. If you choose to live in California, there is a certain probability
that you will experience an earthquake. In Hawaii, you may see a tidal wave. In
Oklahoma, there is a significant probability that you will witness a tornado. Have
such probabilities affected your chosen living area in the past? Do you anticipate that
they will in the future?
W-10. Numbers. The study of probability depends on the use of numbers to express
the likelihood of an event. What is the value of assigning numbers in this way? Why
can’t we just say that a certain event is either likely or not likely?
Crauder-4100237 QL November 6, 2014 12:59
T ALLAHASSEE—A federal judge ruled today that a 2011 law requiring welfare
applicants to undergo drug tests is unconstitutional, striking a blow to Gov.
Rick Scott’s administration over the controversial tests.
Scott quickly said he would appeal U.S. District Judge Mary Scriven’s ruling,
the latest defeat for the governor in a drawn-out battle over drug testing some
of the state’s poorest residents.
Scriven ruled that the urine tests violate the Fourth Amendment’s protections
against unreasonable searches and seizures by the government.
In a harshly worded, 30-page opinion, Scriven concluded that “there is no
set of circumstances under which the warrantless, suspicionless drug testing at
issue in this case could be constitutionally applied.”
Scott, who used the mandatory drug tests as a campaign issue, insists that the
urine tests are needed to make sure poor children don’t grow up in drug-riddled
households.
“Any illegal drug use in a family is harmful and even abusive to a child. We
should have a zero tolerance policy for illegal drug use in families—especially
those families who struggle to make ends meet and need welfare assistance to
provide for their children. We will continue to fight for Florida children who de-
serve to live in drug-free homes by appealing this judge’s decision to the U.S.
Court of Appeals,” Scott said in a statement after today’s ruling.
At Scott’s urging in 2011, the Legislature passed the law requiring all ap-
plicants seeking Temporary Assistance for Needy Families—the “poorest of the
poor”—-to undergo the urine tests. Applicants had to pay for the tests, which
cost about $35, up front and were to have been reimbursed if they did not test
Gary Hershorn/Reuters/Landov
positive.
Whether in the context of welfare applicants being tested for drug use, Olympic
athletes being tested for use of banned substances, or women being screened for
breast cancer, medical tests are pervasive in modern society, and they have serious
consequences. In 2012, American athlete Crystal Cox was stripped of the gold medal
she won in the 4 × 400 meter relay at the 2004 Athens Olympics. In 2010, Cox ad-
mitted to using anabolic steroids from 2001 to 2004. Some may be surprised to learn
that medical tests are not as definitive as we might hope. In fact, they typically yield
only probabilistic results. A positive result of a medical test for a disease does not tell
us for certain that the patient actually has the disease—rather, it gives a probability
Crystal Cox at the 2004 Athens that the disease is present. Similarly, a negative result does not guarantee that the
Olympics disease is not present.
Crauder-4100237 QL November 6, 2014 12:59
328 CHAPTER 5
What does it really mean to say that medical tests provide probabilistic results?
A patient who has tested positive for a disease would certainly like to know what the
test results really mean. That is the issue that we examine in this section.
8 Although we refer to diseases, our discussion also applies to tests for banned substances, in which case having the
disease means use of the substance.
Crauder-4100237 QL November 6, 2014 12:59
Use these data to estimate the sensitivity and specificity of this test.
SOLUTION
A total of 571 + 259 = 830 people in this group have TB. Of these, 571 test positive,
so there are 571 true positives. This indicates a sensitivity of
True positives 571
Sensitivity = =
All who have the disease 830
This is about 0.688 or 68.8%. Hence, about 69% of the people who have TB will
test positive, so about 31% of the people who have TB will test negative.
A total of 8 + 162 = 170 do not have TB. Among these 162 test negative, there
are 162 true negatives. This indicates a specificity of
True negatives 162
Specificity = =
All who do not have the disease 170
This is about 0.953 or 95.3%. Thus, about 95% of the people who do not have TB
will test negative, so about 5% of the people who do not have TB will test positive.
Use these data to estimate the sensitivity and specificity of this test.
The answer is provided at the end of this section.
Crauder-4100237 QL November 6, 2014 12:59
330 CHAPTER 5
It is important to note that sensitivity and specificity depend only on the chemical
and physical properties of a test and not on the population being tested. Hence,
the preceding calculations give empirical estimates for theoretical probabilities. To
explain this point, we consider lightbulbs of a certain brand. There is a theoretical
probability that lightbulbs of this brand will burn out in the first month of use. To
estimate this probability, we might observe the lights in a large building for a month
to see how many burn out. That is, we can estimate the theoretical probability by
calculating an empirical probability.
The next example shows how we can recover information by using the sensitivity
and specificity.
The calculation of sensitivity and specificity for a medical test requires that we
know whether a positive result (for example) represents a true positive or a false
positive. This means that we need to be able to determine independently who in the
test group is healthy and who is not. For this determination, we need an independent
test that can be presumed to be accurate. Such a test is called a gold standard. A gold
standard is very often expensive or difficult to administer. For example, in testing for
HIV (a virus that causes AIDS), the gold standard would include a careful clinical
examination. Researchers have developed less expensive tests for HIV, and the sen-
sitivity and specificity provide a way to measure their accuracy in comparison with
the gold standard.
The prevalence of a disease depends on the population under consideration. For ex-
ample, the prevalence of HIV in Botswana in 2008 was almost 24%, but in the United
States it was less than 1%.
A patient who has tested positive for a disease wants to know whether the dis-
ease is really present or not. That is, he or she wants to know the probability that the
disease is actually present, given that test results were positive. But this probability
depends on the prevalence of the disease in the population being tested. A positive
result for an HIV test administered among the residents of Botswana, where preva-
lence is high, is more likely to be accurate than is a positive result for the same test
administered to residents of the United States, where prevalence is low.
We capture this notion of accuracy using the following terms: positive predictive
value and negative predictive value.
332 CHAPTER 5
SOLUTION
For the general population, there are 9 true positives and 99 false positives, for a
total of 9 + 99 = 108 positive results. Therefore,
True positives
PPV =
All positives
9
=
108
This is about 0.083, so only 8.3% of those who test positive from the general popu-
lation actually have hepatitis C. Applying this test to the general population would
be of little value.
For the population of older and former injection drug users, there are 720 true
positives and 20 false positives, for a total of 720 + 20 = 740 positive results.
Therefore,
True positives
PPV =
All positives
720
=
740
This is about 0.973, so 97.3% of those who test positive from the population of
older and former injection drug users do, in fact, have hepatitis C.
afflicts many Americans. Here are some data from a study of a new screening test for
Crohn’s disease.9
We can use these data to estimate the accuracy of this test in terms of sensitivity
and specificity. The number of true positive tests in the study is 36, and the total
number of patients tested who actually have the disease is 45 (36 true positives plus
9 false negatives). Therefore, we estimate that the sensitivity is 36/45 = 0.8 or 80%.
A sensitivity of 80% tells us that 80% of those people with Crohn’s disease will test
positive.
The number of true negative tests in the study is 230, and the total number of
patients tested who do not have the disease is 255 (230 true negatives plus 25 false
positives). Therefore, we estimate that the specificity is 230/255 or about 90.2%. A
specificity of 90.2% tells us that 90.2% of those people without Crohn’s disease will
test negative.
Consider a hypothetical patient who has tested positive for Crohn’s disease. The
patient wants to know whether the disease is actually present or not. That is, the
patient wants to know the positive predictive value of the test. But, as we noted,
the PPV depends on the population. To explore this point, we first consider applying
the test to the entire population of the United States.
c. Use the information from part b to calculate the PPV of the Crohn’s test for
the general U.S. population.
SOLUTION
a. Because the prevalence is 0.18%, to find the number of Americans suffering
from Crohn’s, we multiply the total population of 320 million by 0.0018:
Number with Crohn’s = 0.0018 × 320
Thus, about 0.58 million people in the United States suffer from Crohn’s.
b. Now 0.58 million people in the United States have Crohn’s disease. Because
the sensitivity of the test is 80%, to calculate the number of true positives, we
9 These data are adapted from the illustration of sensitivity and specificity at www.fpnotebook.com/Prevent/Epi.
Crauder-4100237 QL November 6, 2014 12:59
334 CHAPTER 5
or about 0.46 million. This is the number of people who have Crohn’s and test
positive, so 0.58 − 0.46 = 0.12 million Crohn’s sufferers will test negative. This
is the number of false negatives.
Because 0.58 million have Crohn’s disease, 320 − 0.58 = 319.42 million do
not. Because the specificity of the test is 90.2%, to find the number of true
negatives, we multiply 319.42 million by 0.902:
or about 288.12 million. This is the number of people who do not have Crohn’s
and test negative, so 319.42 − 288.12 = 31.30 million disease-free individuals
will test positive. This is the number of false positives. The completed table is
shown below.
c. The total number of positive results is 0.46 + 31.30 = 31.76 million. Because
0.46 million of these are true positives, we calculate the PPV as follows:
True positives
PPV =
All positives
0.46
=
31.76
This is about 0.014, so the PPV for this population is 1.4%. That is, if everyone
in the United States were tested for Crohn’s, only 1.4% of people who test
positive would actually have the disease.
The results of Example 5.12 seem to indicate that the Crohn’s test is so ineffective
that it is useless. This raises a big question: Why even bother with such a test if it
is so ineffective in predicting the presence of the disease? Should a person who tests
positive even be concerned?
The key is the population we regarded as the patient’s peers—the entire popula-
tion of the United States. Because the prevalence of Crohn’s in the entire population
is so low, we should not be surprised that even an accurate test has a hard time in
picking out those with the disease. But because Crohn’s disease is so rare, a doctor
would never prescribe a test for it unless there was some reason to do so. Surely, it
must be that the patient’s medical profile (e.g., his or her symptoms, age, or family
medical history) suggested the possible presence of the disease.
The point is that we should not think of the patient’s peer group as everyone
in the United States. Rather, we should restrict attention to the roughly 3.3 million
Americans who have a medical profile suggesting the possibility of Crohn’s disease.
Surely among this group, a positive test will be a more reliable indicator of whether
a person has Crohn’s. In fact, among this group the screening test has a PPV of
63.0%. (See Exercises 16 and 17.) Thus, the probability that a patient in this group
has Crohn’s given that he or she tested positive is 63.0%. Comparing the PPV of
1.4% for the general population with the PPV of 63.0% for those with symptoms
shows that this test is much more reliable when we choose an appropriate population
of peers.
Crauder-4100237 QL November 6, 2014 12:59
Conditional probability
Positive and negative predictive values (as well as the sensitivity and specificity) are
examples of conditional probabilities.
Calculate the conditional probability that a person tests positive given that the
person has TB.
SOLUTION
There are 446 + 216 = 662 people who have TB, and 446 of this group test positive.
Hence, the probability that a person tests positive given that the person has TB is
True positives
P(Positive test given TB is present) =
All who have TB
446
=
662
This is about 0.674 or 67.4%.
You may have noticed that the conditional probability in Example 5.13 is the
same as the sensitivity: Both of them can be written as
True positives
All who have the disease
Crauder-4100237 QL November 6, 2014 12:59
336 CHAPTER 5
This connection holds in general: The sensitivity is the conditional probability that a
person tests positive given that the person has the disease. Similar reasoning shows
that for the PPV the order is reversed: The PPV is the conditional probability that a
person has the disease given that the person tests positive.
3. Use the preceding data to calculate the positive predictive 8. For what percentage of the individuals in the study did the
value of the test. test return an incorrect result?
Crauder-4100237 QL November 6, 2014 12:59
9. Lupus. You have seen in this chapter that there are limita- 17, we calculate the PPV of this test within a population of in-
tions to using diagnostic tests. For example, the ANA (Anti- dividuals who show symptoms of the disease. Assume that this
Nuclear Antibody) test has been useful in diagnosing Systemic test is applied to the group of approximately 3.3 million Amer-
Lupus Erythematosus (SLE) in patients who show multiple icans who suffer from symptoms that are very similar to those
signs of the auto-immune disease, including chronic inflamma- of Crohn’s. Suppose also that all of the 0.58 million Crohn’s
tion of various tissues in the body. But many physicians use the sufferers belong to this group.
ANA test on patients who do not show multiple signs of the 16. Fill in the following table based on this group of 3.3 mil-
disease (sometimes called a “shotgun” approach to diagnosis). lion. (Write your answers in millions and round to two decimal
In these cases, the test results in many more false positives than places.)
true positives, meaning an overwhelming number of patients
who test positive for SLE don’t actually have the disease.10
Has disease Does not have disease Totals
Explain the above in terms of sensitivity, specificity, and
positive predictive value. Test positive
(millions)
10. Sensitivity and specificity. The accompanying table gives
Test negative
the results of a screening test for a disease. Estimate the sensi- (millions)
tivity and specificity of the test. Totals
Has disease Does not have disease 17. Use the table in Exercise 16 to show that the PPV of the
Test positive 285 280 test within this population of individuals showing symptoms
Test negative 15 420 of the disease is about 63.0%.
18–19. Another disease. Complete Exercises 18 and 19, in
11. Sensitivity and specificity. The accompanying table gives order, using the following information. Suppose a test for a
the results of a screening test for a disease. Estimate the sensi- disease has a sensitivity of 75% and a specificity of 85%. Fur-
tivity and specificity of the test. ther suppose that in a certain country with a population of
1,000,000, 20% of the population has the disease. Exercises
Has disease Does not have disease 18 and 19 refer to this situation.
15. What are the PPV and NPV of this test for this country? Drug user Not a drug user Totals
16–17. More on screening test. Recall the screening test for Test positive
Crohn’s that has a sensitivity of 80% and a specificity of 90.2% Test negative
(see the discussion beginning on page 333). In Exercises 16 and Totals
338 CHAPTER 5
23. Suppose a randomly chosen person from the population in 33. What does the figure 69% specificity in the above study
question tests positive. How likely is the individual to be a drug mean?
user? That is, what is the PPV of this test for this population? 34. Assume that 10% of the male population over the age of
24. If you found out a student was expelled because she tested 50 has prostate cancer. What are the PPV and NPV of this test
positive for drug use with this test, would you feel that this for this population? Suggestion: First make a table as in Exer-
action was justified? Explain. cise 14. Because the PPV and NPV don’t depend on the size of
25. Suppose a randomly chosen person from the population the population, you may assume any size you wish. A choice
in question tests negative. How likely is the individual not to of 10,000 for the population size works well.
be a drug user? That is, what is the NPV of this test for this 35. Assume that 40% of the male population over the age of
population? 70 has prostate cancer. What are the PPV and NPV of this test
26. If you found out that a student who had been accused of us- for this population?
ing drugs was exonerated because she tested negative for drug 36–39. HIV. Complete Exercises 36 through 39, in order,
use with this test, would you feel that this action was justified? using the following information. Suppose that a certain HIV
Explain. test has both a sensitivity and specificity of 99.9%. This test is
27. Testing for Drug Abuse II. Repeat Exercises 22 through 26, applied to a population of 1,000,000 people.
this time assuming that the individual being tested is chosen 36. Suppose that 1% of the population is actually infected with
from a population of which 20% are drug users. (This might HIV.
be the case if the test is being applied to a prison population a. Calculate the PPV. Suggestion: First make a table as in
that has a known history of drug use.) Exercise 14.
28. Conclusions. In Exercises 22 through 26, it was assumed b. Calculate the NPV.
that a drug test was applied to a population with 5% users. In
Exercise 27, it was assumed that this drug test was applied to a c. How many people will test positive who are, in fact,
population with 20% users. What do the results suggest about disease-free?
how the drug test could be used most effectively? 37. Suppose that the population comes from the blood donor
29–31. Genetic testing. Complete Exercises 29 through 31, pool that has already been screened. In this population, 0.1%
in order, using the following information. A certain genetic con- of the population is actually infected with HIV.
dition affects 5% of the population in a city of 10,000. Suppose a. Calculate the PPV.
there is a test for the condition that has an error rate of 1% b. Calculate the NPV.
(i.e., 1% false negatives and 1% false positives). Exercises 29
c. How many people will test positive who are, in fact,
through 31 refer to this genetic condition.
disease-free?
29. Fill in the table below.
38. Consider a population of drug rehabilitation patients for
Has condition Does not have condition Totals which 10% of the population is actually infected with HIV.
Test positive a. Calculate the PPV.
Test negative b. Calculate the NPV.
Totals
c. How many people will test positive who are, in fact,
30. What is the probability (as a percentage) that a person has disease-free?
the condition if he or she tests positive? 39. Compare the values for the PPV you found in part a of Ex-
31. What is the probability (as a percentage) that a person does ercises 36 through 38, and explain what this says about testing
not have the condition if he or she tests negative? for HIV.
32–35. PSA test. Complete Exercises 32 through 35, in order, 40–41. Conditional probability and medical testing. Exer-
using the following information. The following is paraphrased cises 40 and 41 refer to the table in Example 5.13 on page 335.
from a Web site that discusses the sensitivity and specificity of 40. Find the probability that a person tests positive given that
the PSA (Prostate Specific Antigen) test for prostate cancer.12 In the person does not have the disease.
a larger study (243 subjects), Ward and colleagues used a cutoff
ratio of 0.15 to demonstrate 78% sensitivity and 69% speci- 41. Find the probability that a person does not have the disease
ficity in the diagnosis of prostate cancer. In contrast, tPSA mea- given that the person tests negative. Which of the four basic
surements alone yielded equivalent sensitivity but only 33% quantities (sensitivity, specificity, PPV, NPV) does this number
specificity. These reports are encouraging, but larger clinical represent?
studies are required for determination of the optimal fPSA:tPSA 42. Conditional probability. You roll a fair six-sided die and
cutoff ratio. don’t look at it. What is the probability that it is a 5 given that
32. What does the figure 78% sensitivity in the above study your friend looks and tells you that it is greater than 2? Leave
mean? your answer as a fraction.
12 www.ivdtechnology.com/article/equimolar-psa-assays.
Crauder-4100237 QL November 6, 2014 12:59
43. More conditional probability. In a standard deck of cards, butter cookies. The 35 cookies are put together and offered on
the jack, queen, and king are “face cards.” You draw a card a single plate. I pick a cookie at random from the plate.
from a standard deck. Your friend peeks and lets you know a. What is the probability that the cookie is chocolate chip?
that your card is a face card. What is the probability that it is b. What is the probability that Mary baked the cookie?
a queen given that it is a face card? Leave your answer as a
fraction. c. What is the probability that the cookie is chocolate chip
given that Mary baked it?
44. Shampoo. Your wife has asked you to pick up a bottle of
her favorite shampoo. When you get to the store, you can’t d. Use Bayes’s theorem to calculate the probability that the
remember which brand she uses. There are nine shampoo cookie was baked by Mary given that it is chocolate chip.
brands on the shelf. If you select one, what is the probability In this situation, Bayes’s theorem tells us that
that you get the right brand of shampoo? Of the nine bottles,
P(Mary given Chocolate)
four are blue, three are white, and two are red. You remem-
ber that your wife’s shampoo comes in a blue bottle. What is P(Chocolate given Mary) × P(Mary)
=
the probability that you get the right bottle with this additional P(Chocolate)
information? Leave your answers as fractions.
e. Calculate without using Bayes’s theorem the probability
45. Conditional probability. The total area of the United States that the cookie was baked by Mary given that it is chocolate
is 3.79 million square miles, and Texas covers 0.27 million chip. Suggestion: How many of the chocolate chip cookies
square miles. A meteor falls from the sky and strikes Earth. were baked by Mary?
What is the probability that it strikes Texas given that it strikes
49. Cookies again. As in Exercise 48, Mary bakes 10 choco-
the United States?
late chip cookies and 10 peanut butter cookies. Bill bakes 5
46. This one is tricky. A family that you haven’t yet met moves chocolate chip cookies and 10 peanut butter cookies. But now,
in next door. You know the couple has two children, and you Mary’s cookies are put on one plate and Bill’s on another. I pick
discover that one of them is a girl. What is the probability that a plate at random and then pick a cookie from that plate. Do
the other child is a boy? That is, what is the probability that you think the probability that cookie is chocolate chip is the
one child is male given that the other child is female? Leave same as in the scenario from Exercise 48? To help answer that
your answer as a fraction. question, consider the extreme case in which Mary bakes 99
47. Formula for conditional probability. We express the prob- chocolate chip cookies and no peanut butter cookies, while Bill
ability that event A occurs given that event B occurs as bakes no chocolate chip cookies and 1 peanut butter cookie.
Explain why you give the answer you do.
P(A given B) 50. Car lots. Yesterday car lot Alpha had two Toyotas and one
[In many books, the notation used is P(A|B).] It turns out Chevrolet for sale. Car lot Beta had three Toyotas and five
that the following basic formula for conditional probabilities Chevrolets for sale. This morning Alan bought a car, choos-
holds: ing one of the two lots at random and then choosing a car at
random from that lot. Leave your answers as fractions.
P(A and B) = P(A given B) × P(B) a. What is the probability that Alan bought from Alpha?
In this exercise, we will verify this formula in the case that A b. If Alan bought from Alpha, what is the probability he
is the event that a card drawn from a standard deck is a heart bought a Toyota?
and B is the event that a card drawn from a standard deck is c. If Alan bought from Beta, what is the probability he
an ace. Leave your answers as fractions. bought a Toyota?
a. Calculate P(A and B). (This is the probability that the
d. It can be shown that the probability that Alan bought
card is the ace of hearts.)
a Toyota is 25/48. What is the probability Alan bought
b. Calculate P(A given B). (This is the probability that the from Alpha given that he bought a Toyota? Give your an-
card is a heart given that it is an ace.) swer both as a fraction and in decimal form rounded to two
c. Calculate P(B). (This is the probability that the card is places. Note: In this situation, Bayes’s theorem says
an ace.)
P(Alpha given Toyota)
d. Use parts a, b, and c to verify the formula above in this P(Toyota given Alpha) × P(Alpha)
case. =
P(Toyota)
48. Bayes’s theorem. Bayes’s theorem on conditional probabil-
ities states that, if P(B) is not 0, then 51. Bowls of cookies. There are two bowls of cookies. Bowl Al-
pha has 5 chocolate chip cookies and 10 peanut butter cookies.
P(B given A) × P(A) Bowl Beta has 10 chocolate chip cookies and 10 peanut butter
P(A given B) =
P(B) cookies. Jerry selects a bowl at random and chooses a cookie
Also note that Exercises 50 and 51 can be solved using this at random from that bowl. Leave your answers as fractions.
formula. Leave your answers as fractions. a. What is the probability that he chooses bowl Alpha?
Mary bakes 10 chocolate chip cookies and 10 peanut but- b. What is the probability that he gets a chocolate chip
ter cookies. Bill bakes 5 chocolate chip cookies and 10 peanut cookie given that he chooses from bowl Alpha?
Crauder-4100237 QL November 6, 2014 12:59
340 CHAPTER 5
c. It can be shown that the probability that he gets a choco- century, but the concepts in probability behind them are often
late chip cookie is 5/12. What is the probability that he traced to Thomas Bayes, who lived in Britain in the eighteenth
chose from bowl Alpha given that he got a chocolate chip century. Research the life and work of Bayes and the ideas in
cookie? probability and statistics currently associated with his name.
Note: In this situation, Bayes’s theorem says The following exercise is designed to be solved using technol-
ogy such as a calculator or computer spreadsheet.
P(Alpha given Chocolate)
P(Chocolate given Alpha) × P(Alpha) 53. Spreadsheet. Devise a spreadsheet that will allow the user
= to input the prevalence of a disease in a population along
P(Chocolate)
with the sensitivity and specificity of a test for the disease. The
52. History. The tables used in this section are examples of spreadsheet should then complete a table similar to that in Ex-
contingency tables. These tables came into use in the twentieth ample 5.9 on page 329 and report the PPV and NPV.
W-2. Lost needle. Sometimes we use the phrase “looking for a needle in a haystack”
to refer to the difficulty of locating one object in the midst of a lot of others. Explain
how this phrase is relevant to the connection between the prevalence of a disease and
the usefulness of testing the general population.
Exercise 19 provides additional information relevant to this topic.
W-3. Test my child? A community activist who supports zero tolerance advocates
testing every child in high school for drugs even though there is no evidence of
widespread drug abuse in that school. Is this a good idea? Explain your reasoning.
Exercise 23 provides additional information relevant to this topic.
W-4. Calculating sensitivity. In some exercises in this section, you are asked to use
given data to estimate the sensitivity of a test. Why is the instruction to estimate
rather than to calculate exactly? What information is needed in order to calculate
exactly the sensitivity?
Exercise 10 provides additional information relevant to this topic.
W-5. Specificity. Is it necessarily true that a large specificity (nearly equal to 100%)
indicates a relatively small number of false positives? Explain your answer.
Exercise 11 provides additional information relevant to this topic.
W-6. Which causes concern? If you test positive for a disease, would you be more
concerned if the test is known to have a high sensitivity or a high specificity? Explain
your reasoning.
W-7. What do they mean? Use your own words to explain the meaning of the terms
positive predictive value and negative predictive value, and give some good examples.
Exercise 36 provides additional information relevant to this topic.
W-8. Conditional probability. Use your own words to explain the meaning of condi-
tional probability, and give some examples.
Exercise 40 provides additional information relevant to this topic.
W-9. Connection. Explain the connection between conditional probability and the
terms such as sensitivity used in medical testing.
Crauder-4100237 QL November 6, 2014 12:59
O n March 10, 2003, police arrived on the scene of a grisly murder in a dead-
end neighborhood. . . .
Later, police found a fingerprint thought to be made in blood on [a] cigarette
box’s cellophane wrapping. The police used the print to place a then 29-year-old
named Chad Enderle, who said he sometimes bought drugs from the victim, at
the scene of the crime.
At trial, authorities highlighted seven identifying characteristics—or points
of minutiae—to link the cigarette print to Enderle’s right ring finger. But seven
points isn’t usually enough to make a match, according to several fingerprint
experts who say the industry typically calls for more. The Medill Justice Project
interviewed several fingerprint experts who offered conflicting conclusions.
Since Enderle’s sentencing [for murder], studies, federal agencies on foren-
sic sciences and some judges across the country have urged greater scrutiny of
fingerprint analysis. . .
Tomas Bravo Reuters/Landov
Industry challenges
Cedric Neumann, an assistant statistics professor at South Dakota State Univer-
sity with a Ph.D. in forensic science, said differences in the understanding of the
logic behind fingerprint examination resulting from examiners’ varied field and
apprenticeship experiences are like “night and day.”
Neumann, in the February 2012 issue of Significance, a statistical magazine,
described a statistical model he developed for quantifying fingerprint character-
istics.
Neumann said that although it is difficult to summarize fingerprints into sets
of data that can be processed using statistics, he compared his model to that
used to back DNA identification. Humans have 46 chromosomes and each chro-
mosome has thousands of genes, yet DNA identification relies on a minimum of
16 of those genes. In Neumann’s model, still being tested, the most prominent
minutiae in a suspect’s print would be connected to form a polygonal figure.
Qualities of that polygon would be assigned numerical values that would render
the probability that the suspect is the source of the crime-scene mark.
This article describes the use of probability calculations to aid in examining fin-
gerprints. To calculate the necessary probabilities, we need to be able to count the
number of ways in which an event can happen. In this section, we will examine some
fundamental counting methods.
Simple counting
Counting the number of outcomes in the case of flipping two coins, as we did in
Section 5.1, is not difficult. We distinguish the two coins (by thinking of a nickel
Crauder-4100237 QL November 6, 2014 12:59
342 CHAPTER 5
and a dime, for example) and then simply list the four possible outcomes: HH, HT,
TH, TT.
If we are flipping three coins, the number of possible outcomes is much larger.
Distinguishing the coins as a penny, nickel, and dime, we find the possibilities listed
below.
Favorable outcomes
P(10 heads) =
Total outcomes
1
=
1024
Another way to think about this principle is as follows: Suppose there are two
blanks
that we wish to fill, and we have N choices for the first blank and M choices for the
second blank:
NM
Then the total number of ways we can fill the two blanks is N × M. This extends to
any number of blanks.
As a simple application of the Counting Principle, let’s look at product labeling.
Suppose a company labels its products with a letter of the alphabet followed by a
numeral (0 through 9), for example, model A4 or model D9. How many different
labels can the company make? Let’s think of this as filling in two blanks:
Crauder-4100237 QL November 6, 2014 12:59
344 CHAPTER 5
A letter goes in the first blank and a numeral goes in the second blank. There are 26
letters that can go in the first blank and 10 numerals that can go in the second blank:
26 10
According to the Counting Principle, the total number of labels is the number of
outcomes for the first blank times the number of outcomes for the second blank.
This is 26 × 10 = 260.
There are 5 letters that can go in the first blank and 26 in each of the other two:
5 26 26
Applying the Counting Principle, we find a total of 5 × 26 × 26 = 3380 codes.
The Counting Principle allows us to calculate the total number of outcomes for
fairly complex events. Suppose, for example, that we toss five dice and want to know
the total number of outcomes. We think of tossing the dice one at a time. Applying
the Counting Principle, we see that the total number of outcomes is
Outcomes for die #1 × Outcomes for die #2 × · · · × Outcomes for die #5
Because each die has six possible outcomes, we are multiplying five 6’s together,
which gives the total number of outcomes as
6 × 6 × 6 × 6 × 6 = 65 = 7776
We can also think of this in terms of five blanks, one for each die. Each blank has
six possible outcomes:
66666
© FORRAY Didier/SAGAPHOTO.COM/Alamy
b. How many such plates are possible if we insist that on each plate no numeral
can appear more than once and no letter can appear more than once?
SOLUTION
a. Let’s think of a blank license tag with three slots for numbers and three slots
for letters:
numbers letters
We need to figure out how many ways we can fill in the slots. For each number
slot, there are 10 possible numerals (0 through 9) to use: 10 for the first blank,
10 for the second, and 10 for the third:
10 10 10
numbers letters
For each letter slot, there are 26 possible letters of the alphabet: 26 for the first
blank, 26 for the second, and 26 for the third:
10 10 10 26 26 26
numbers letters
Applying the Counting Principle gives
10 × 10 × 10 × 26 × 26 × 26 = 17,576,000
346 CHAPTER 5
a first card from a deck and, without putting it back, draw a second card. In this
situation, drawing the 2 of spades for the first card and the 3 of spades for the second
is different from drawing the 3 of spades for the first card and the 2 of spades for the
second. We can think of filling in two blanks. There are 52 possibilities for the first,
but only 51 for the second:
52 51
The counting principle gives 52 × 51 = 2652 possible outcomes. These kinds of
calculations will be helpful in calculating probabilities for cards.
Suppose you draw a card, put it back in the deck, and draw another. What is the
probability that the first card is an ace and the second one is a jack? (Assume that
the deck is shuffled after the first card is returned to the deck.)
SOLUTION
To find the probability, we have two numbers to calculate: The number of ways we
can draw two cards (the total number of possible outcomes), and the number of ways
we can draw an ace followed by a jack (the number of favorable outcomes).
First we count the total number of ways to draw two cards in order from a deck.
Think of filling two blanks:
The first blank is filled by a card drawn from the full deck, and the second blank is
also filled by a card drawn from a full deck (because we put the first card back in the
deck). There are 52 possible cards to place in the first blank, and for each of these,
there are 52 possible cards to place in the second blank:
52 52
Using the Counting Principle, we find 52 × 52 = 2704 ways to fill the two blanks
with two cards from the deck. This is the total number of possible outcomes.
Now we ask how many ways there are to fill these two blanks with an ace fol-
lowed by a jack. There are four aces in the deck, so there are four choices to fill the
first blank:
4
After this has been done, there are four jacks with which to fill the second blank:
44
Again, using the Counting Principle, we find 4 × 4 = 16 ways to fill the two blanks
with an ace followed by a jack. This is the number of favorable outcomes. So the
probability of drawing an ace followed by a jack is
Favorable outcomes
P(Ace followed by jack) =
Total outcomes
16
=
2704
This is about 0.006 or 0.6%.
Rex May/Balooscartoon.com
The Counting Principle applies to elections just as it does to cards.
SOLUTION
a. Imagine that we have three blanks corresponding to president, vice president,
and treasurer:
P VP T
There are 10 possibilities for the president blank. When that is filled, there are
nine names left, so there are nine possibilities for the vice president blank. When
that is filled, there are eight possibilities for the third blank:
10 9 8
P VP T
The Counting Principle gives
Number of outcomes = 10 × 9 × 8 = 720
b. We must count the number of ways John can be vice president. In this case,
John is not president, so there are nine possible names for the president blank.
There is just one possibility, John, for the vice president blank, and that leaves
eight possibilities for the treasurer blank:
9 1 8
P VP T
Crauder-4100237 QL November 6, 2014 12:59
348 CHAPTER 5
Hence,
Number of ways John is vice president = 9 × 1 × 8 = 72
This is the number of favorable outcomes. Using these numbers, we find the
probability we want:
Favorable outcomes
Probability John is vice president =
Total outcomes
72
=
720
= 0.1
Thus, there is a 10% chance of John being selected as vice president.
Alternatively, we can think along the following lines. All we are interested
in knowing is whose name goes in the vice president blank. There are 10 equally
likely possibilities. Exactly one of them is John. Thus, the probability that he is
vice president is 1/10 = 0.1.
c. We want to arrange the three names John, Jim, and Mary in the three
office-holder blanks. There are three possible names for the first, two for the
second, and one for the third:
3 2 1
P VP T
Thus, the total number of ways these three candidates can all win office is
3 × 2 × 1 = 6. These are the favorable outcomes. We calculated in part a that
there are 720 possible outcomes. Using these values, we calculate
Favorable outcomes
Probability all three selected =
Total outcomes
6
=
720
This is about 0.008 and represents a 0.8% chance that all three are selected.
d. If these three names don’t appear in the officer blanks, then there are seven
possibilities for the first, six for the second, and five for the third:
7 6 5
P VP T
That is a total of 7 × 6 × 5 = 210 favorable outcomes. Thus, the probability
that none will be selected is
Favorable outcomes
Probability none selected =
Total outcomes
210
=
720
This is about 0.292, and it represents a 29.2% chance that none will be selected.
e. We found in part c that the probability that all three are selected is 6/720.
The probability that at least one will not be selected is the probability that the
event of all three being selected does not occur. This is 1 minus the probability
that all three are selected, so the probability that at least one will not be
selected is
6
1−
720
or about 0.992. Hence, there is a 99.2% chance that at least one will not be
selected.
Crauder-4100237 QL November 6, 2014 12:59
Independent events
Suppose we pick a person at random and consider the probability that the person
is female and the probability that the person’s last name is Smith. Knowing that the
person is female does not affect the probability that the person’s last name is Smith.
These two events (picking a female and picking someone whose last name is Smith)
are independent.
In contrast, suppose we pick a person at random and look at the probability that
the person is female and the probability that the person’s first name is Mary. If we
know that the person is female, the probability that the name will be Mary may
be quite small. But if we don’t know whether the person is female, the probability
that the name will be Mary is much smaller. These two events (picking a female and
picking someone named Mary) are not independent.
are independent. For example, the events of being a smoker and having heart dis-
ease are not independent. The establishment of the link between smoking and heart
disease was a milestone in public health policy.
There is an important formula involving probabilities of independent events. We
look at an example to illustrate it. One of my favorite stores is having a sale. When
I buy something, I get a ticket at the checkout stand that gives me either 10% off,
20% off, 30% off, 40% off, or 50% off. (We suppose that each of these tickets is
equally likely.) I buy a pair of jeans in the morning and a shirt in the afternoon. What
is the probability that I get at least 30% off in the morning and at least 40% off in
the afternoon? These are independent events because winning one discount does not
give any information about the other.
First we count the total number of possible outcomes. Because there are five pos-
sibilities for the morning and five for the afternoon, the Counting Principle tells us
that there are 5 × 5 = 25 possible outcomes. Now let’s count favorable outcomes.
For the morning, 30%, 40%, and 50% are favorable, and for the afternoon, only
40% and 50% are favorable. With three favorable outcomes for the morning and
two for the afternoon, the Counting Principle gives 3 × 2 = 6 favorable outcomes.
So the probability we seek is
Favorable outcomes
P(At least 30% and At least 40%) =
Total outcomes
6
=
25
Now we calculate the two probabilities individually. For the morning, there are
three favorable outcomes out of a total of five. So the probability of getting 30% or
greater in the morning is 3/5. For the afternoon, there are two favorable outcomes
out of a total of five. Thus, the probability of getting 40% or more in the afternoon
is 2/5.
If we multiply the individual probabilities, we find the probability of 30% or
more in the morning and 40% or more in the afternoon:
3 2 6
× =
5 5 25
Crauder-4100237 QL November 6, 2014 12:59
350 CHAPTER 5
so
P(At least 30%) × P(At least 40%) = P(At least 30% and At least 40%)
This result is no accident. Given two independent events, the probability that
both occur is the product of the separate probabilities for the two events.13 Here,
each probability must be written as a decimal or a fraction, not as a percentage.
Let’s look further at the previous example. Suppose I got home and found that my
digital camera was defective. I might reason as follows: “The probability of both the
camera and the printer being defective is so small that perhaps I can be assured that
the printer is fine.” This is a bogus argument. The probability of both being defective
is indeed quite small. Nonetheless, the two events are independent, so knowing that
the camera is defective does not change the probability that the printer is defective.
13 In fact, our initial description of independent events is useful to build intuition, but the mathematical definition of
independent events is that the product formula applies.
Crauder-4100237 QL November 6, 2014 12:59
4. Playing two games.You play two unrelated games, one after 9. Until 1996 all toll-free numbers started with the area code
the other. There is a 30% chance of winning the first game, and 800. Assume that all seven digits for the rest of a number were
there is a 40% chance of winning the second game. Use the possible, and determine how many toll-free 800 numbers there
product formula for independent events to find the probability were.
that you win both games. Express your answer as a percentage. 10. Today there are a total of six toll-free area codes: 800, 844,
5. Why use letters? Automobile license plates often use a mix 855, 866, 877, and 888. Assume that all seven digits for the rest
of numerals and letters of the alphabet. Why do you suppose of a number are possible, and determine how many toll-free
this is done? Why not use only numerals? numbers there are now.
6. Bicycle speeds. A customer in a bicycle shop is looking at a 11. I won the lottery. I just won the Powerball lottery and
bicycle with three gears on the front (technically, three chain- wanted to keep it a secret. But I told two close friends, who
rings) and seven gears on the rear (technically, seven sprockets). each told five other different people, and those in turn each told
He decides to look at a different bike because he wants more four other different people. How many people (other than me)
than 10 speeds. Has he counted correctly? heard the story of my lottery win? Be careful how you count.
7. Shampoo. A company wants to market different types of 12. Voting. I am voting in a state election. There are three can-
shampoo using color-coded bottles, lids, and label print. It has didates for governor, four candidates for lieutenant governor,
three candidates for the state house of representatives, and four
Crauder-4100237 QL November 6, 2014 12:59
352 CHAPTER 5
candidates for the senate. In how many different ways can I fill 14. How many different faces can be formed by the children’s
out a ballot if I vote once for each office? toy?
13. My bicycle lock. My combination bicycle lock has four 15. If a child makes one change each minute and spends 48
wheels, each of which has the numbers 0 through 20. Exactly hours a week changing faces, how many years will it take to
one combination of numbers (in order) on the wheels (2-17-0- see every possible face?
12, e.g.) will open the lock. 16–21. Independent events. In each of Exercises 16 through
21, determine whether you think the given pair of events is in-
dependent. Explain your answer.
© Michael Spring/Dreamstime.com
16. You collect Social Security benefits. You are over age 65.
17. You are in good physical condition. You work out
regularly.
18. You suffer from diabetes. You are a registered Democrat.
19. You are over 6 feet tall. You prefer light-colored attire.
20. You frequent casinos. You have lost money gambling.
a. How many possible combinations of numbers are there 21. You are male. You are a smoker.
on my bicycle lock? 22. An old joke. An old joke goes like this: A traveler reads in
b. Suppose that I have forgotten my combination. At ran- the newspaper that the probability of a passenger having a gun
dom I try 1-2-3-4. What is the probability that the lock will on an airplane is only one one-thousandth (0.001). He asked
open with this combination? his math major friend what the probability would be of two
c. Out of frustration I try at random 50 different combina- passengers having guns on an airplane, and she replied that it
tions. What is the probability that 1 of these 50 efforts will would be the product of the probabilities or one one-millionth
open the lock? (0.001 × 0.001 = 0.000001). The traveler says, “Great, then
I’m going to carry a gun next time I fly—that’ll reduce the odds
14–15. Counting faces. Complete Exercises 14 and 15, in that someone else will have a gun!” Critique the traveler’s logic.
order, using the following information. A classic children’s toy
consists of 11 different blocks that combine to form a rectangle 23. Coins. We toss a coin twice. Use the product formula for
showing a man’s face.14 Each block has four sides with alter- independent events to calculate the probability that we get a
natives for that part of the face; for example, the block for the tail followed by a head.
right eye might have a closed eye on one side, a partially open 24. A coin and a die. We toss a coin and then roll a die. Use the
eye on another, and so on. A child can turn any block and so product formula for independent events to calculate the prob-
change that part of the face, and in this way many different ability that we get a tail followed by a 6.
faces can be formed. 25–26. Jelly beans. Assume that a box contains four red
jelly beans and two green ones. We consider the event that a
Courtesy of www.continentalhobby.com
30. What is the probability that the total number of dots ap- a given day is 11.7%. Suppose a car is picked at random at 5th
pearing on top is not 7? and Main on a given day and another at random at 7th and
31–32. Marbles. Suppose an opaque jar contains 5 red mar- Polk. Write each probability as a percentage rounded to one
bles and 10 green marbles. Exercises 31 and 32 refer to the ex- decimal place.
periment of picking two marbles from the jar without replacing 43. What is the probability that the car will not run a red light
the first one. at 5th and Main?
31. What is the probability of getting a green marble first and 44. What is the probability that the car will not run a red light
a red marble second? at 7th and Polk?
32. What is the probability of getting a green marble and a red 45. What is the probability that neither car runs a red light?
marble? (How is this exercise different from Exercise 31?) 46–49. A box. Exercises 46 through 49 refer to the following
33–34. More jelly beans. Suppose I pick (without looking) a situation. Suppose that the inside bottom of a box is painted
jelly bean from a box containing five red beans and seven blue with three colors: 1/3 of the bottom area is red, 1/6 is blue,
beans. I record the color and put the bean back in the box. Then and 1/2 is white. You toss a tiny pebble into the box without
I choose a bean again. Exercises 33 and 34 refer to probabilities aiming and note the color on which the pebble lands. Then you
for this situation. toss another tiny pebble into the box without aiming and note
the color on which that pebble lands.
33. What is the probability of getting a red bean both times?
46. Are the two trials independent?
34. What is the probability of getting a blue bean both times?
47. What is the probability that both pebbles land on the color
35–40. Two cards. Exercises 35 through 40 refer to choos- blue?
ing two cards from a thoroughly shuffled deck. Assume that
the deck is shuffled after a card is returned to the deck. 48. What is the probability that the first pebble lands on the
color blue and the second pebble lands on red?
35. If you put the first card back in the deck before you draw
the next, what is the probability that the first card is a 10 and 49. What is the probability that one of the pebbles lands on
the second card is a jack? the color blue and the other pebble lands on red? (How is this
exercise different from Exercise 48?)
36. If you do not put the first card back in the deck before you
draw the next, what is the probability that the first card is a 10 50–53. Terrorism. The next four exercises refer to the follow-
and the second card is a jack? ing two excerpts from an article on math applied to the risk of
terrorism.
37. If you put the first card back in the deck before you draw
the next, what is the probability that the first card is a club and
the second card is a diamond?
38. If you do not put the first card back in the deck before you IN THE NEWS 5.7 Search
354 CHAPTER 5
63. The answer to Exercise 62 is very small. To see how small 65. Double sixes. Chevalier de Méré was a gambler who lived
it is, assume that everyone in the United States filled out many in the seventeenth century. He asked whether or not one should
brackets and that all of the brackets produced in this way were bet even money on the occurrence of at least one “double six”
different. (In fact, the contest was limited to 15 million entries during 24 throws of a pair of dice. This question was a fac-
altogether.) How many brackets would each person have to fill tor that motivated Fermat and Pascal to study probability (see
out to make sure that altogether all of the possible outcomes Exercise 65 in Section 5.1). Answer de Méré’s question: What
of the 63 games were represented? Use the fact that the pop- is the probability of at least one “double six” in 24 consecu-
ulation of the United States at this time was 317 million, and tive rolls of a pair of dice? Write your answer in decimal form
round your answer to the nearest billion. Note: You will need rounded to two places. Hint: First find the probability of 24
a calculator that can handle large numbers. consecutive rolls of a pair of dice without a “double six.”
64. History. The Counting Principle has surely been known for The following exercises are designed to be solved using tech-
a long time. In fact, it may be involved in a curious problem nology such as calculators or computer spreadsheets.
found in one of the oldest surviving mathematical texts, the 66. Factorials. Suppose you want to arrange five people in
Rhind papyrus from ancient Egypt. The papyrus itself dates order in a waiting line. There are five choices for the per-
from about 1650 B.C., but it may be a copy of a source 150 years son who goes first, four for the person who goes second, and
older than that. The curious problem from the Rhind papyrus so on. Thus, the number of ways to arrange five people is
has the following data (where a hekat is a measure of volume): 5 × 4 × 3 × 2 × 1 = 120. This sort of product arises often
enough that it has a special notation:16 We write n! (read “n
Houses 7 factorial”) for n × (n − 1) × · · · × 1. For example, 5! = 120
Cats 49
according to the previous expansion. Use technology to deter-
Mice 343
Heads of wheat 2401 mine how many ways there are to arrange 12 people in order
Hekat measures 16,807 in a waiting line.
67. More on factorials. This is a continuation of Exercise 66.
The papyrus correctly gives 19,607 for the total of the numer- Use technology to determine how many ways there are to ar-
ical column. One historian has suggested that the context of range 50 people in order in a waiting line. Can you find the
the table is as follows: In an estate there are seven houses, each answer for 200 people?
house had seven cats, each cat could eat seven mice, each mouse
could eat seven heads of wheat, and each head of wheat could 68. Listing cases. Devise a computer program that will list all
produce seven hekat measures of grain.15 Use the Counting possible outcomes of tossing n coins. Warning: If you actually
Principle to explain how this interpretation is consistent with run your program for even moderate values of n, you will get
the numbers in the table. a very long output.
15 See Howard Eves, An Introduction to the History of Mathematics, 6th ed. (Philadelphia: Saunders, 1990).
16 The idea of factorials is discussed more fully in Section 5.4.
Crauder-4100237 QL November 6, 2014 12:59
356 CHAPTER 5
Which is right? What does this have to do with the concepts of “independent events”
and “empirical probability”?
Exercise 43 provides additional information relevant to this topic.
W-6. Going to St. Ives. Here is a simplified version of an old riddle. I saw 7 men
going to St. Ives. Each man had 7 sacks, and each sack had 7 cats. Each cat had 7
kits. How many people were going to St. Ives? Of course the answer is 7, but what
counting idea would you use if you were asked how many items were traveling to St.
Ives? Can you think of other traditional riddles or fairy tales that involve this same
counting idea?
Exercise 64 provides additional information relevant to this topic.
W-7. Justice. Discuss how probability plays a role in our judicial system. Hint: Con-
sider DNA testing, among other things.
W-8. Outcomes. In the statement of the Counting Principle, we assume that for each
outcome of the first experiment there are the same number of outcomes of the sec-
ond experiment. Would the Counting Principle apply without this assumption? Hint:
Consider, for example, a very simple menu with two entrées. There is one side dish
available for one entrée, and there are two side dishes available for the other. How
many possible combinations of an entrée with a side dish are there?
W-9. Self-independence. Use the product formula to determine under what conditions
an event A is independent of itself. Does your result make sense in terms of the initial
informal description of independence in the text? Reminder: The solutions of the
equation x = x2 are x = 0 and x = 1.
O n April 14, 2003 the National Human Genome Research Institute (NHGRI),
the Department of Energy (DOE) and their partners in the International Hu-
man Genome Sequencing Consortium announced the successful completion of
the Human Genome Project.
What is a genome?
© Rolffimages/Dreamstime.com
One major goal of understanding the sequence in a DNA molecule is the identi-
fication of genes that give rise to cancer. Understanding the genetics of cancer comes
down to determining how four letters, A, C, G, T, go together in long strings. The
number of possibilities is staggering, but there are mathematical tools to help in the
enumeration. These tools, which ultimately rely on the Counting Principle, are con-
sidered in this section.
Permutations
Suppose we arrange a group of people into a line. The first in line gets first prize,
the second in line gets second prize, and so on. Your place in line determines the
prize you get. So the order in which people are arranged in the line is important. An
arrangement where order is important is called a permutation.
Suppose, for example, that we use the letters A, B, and C to make product codes
where each letter is used exactly once. How many such codes can we make? That
is, how many permutations are there of the letters A, B, and C? We can list them
all to get the answer. The permutations of the letters A, B, C, are ABC, ACB, BAC,
BCA, CAB, CBA—six in all. When the number of items is large, it is not efficient to
list all of the permutations. Instead, we need a faster way of calculating how many
there are.
Think of making a permutation (or arrangement) by filling three blanks with the
three letters without repetition, that is, using each letter once:
There are three choices for the first blank. When that blank is filled, we have only
two letters left, so there are two choices for the next blank. Now there is only one
letter left, so we have just one choice for the last blank:
321
The Counting Principle says that the total number of ways to fill all the blanks is
3 × 2 × 1 = 6. This agrees with the number we found by making a list.
This kind of calculation occurs so often that it is given a special name and sym-
bol. For a positive whole number n, the product of all the whole numbers from n
down through 1 is denoted by n with an exclamation point: n!. This is pronounced
“n factorial.”
Although it may seem strange, zero factorial (0!) is always defined to be the num-
ber 1. One way to justify this is by asserting that there’s just one way to arrange no
items. We’ll see another reason why later. Here are some factorials:
0! = 1
1! = 1
2! = 2 × 1 = 2
3! = 3 × 2 × 1 = 6
4! = 4 × 3 × 2 × 1 = 24
The method we used to find the number of arrangements of the letters A, B, C works
in general. For example, suppose there are 10 people waiting to buy tickets to some
event. In how many ways can we arrange the 10 people in a line at the ticket booth?
Thinking again of filling in slots, we see that there are 10 people who could be at the
head of the line. That leaves nine who might be second, eight in third place, etc. So
Crauder-4100237 QL November 6, 2014 12:59
358 CHAPTER 5
the total number of ways to arrange the 10 people is 10!, which equals 3,628,800.
That’s a pretty big number.
Factorials get big fast—as we just saw, 10! = 3,628,800. To find large factorials
on a calculator using the definition, you would have to do a lot of multiplications,
which could be quite tedious. The accompanying tip should help.
More on arrangements
Let’s revisit the idea of putting people in order and awarding prizes. But this time, let’s
do it under the more realistic assumption that not everyone gets a prize. We think
of selecting 3 lucky winners from a group of 10 people. The first person selected
gets first prize, the second gets second prize, and the third gets third prize. Such an
arrangement is called a permutation of 10 people taken 3 at a time.
Crauder-4100237 QL November 6, 2014 12:59
There are 10 names available for the first slot. But because nobody gets more than
one prize, there are only 9 available for the second and then 8 available for the third:
10 9 8
The Counting Principle shows that the number of ways to select the three winners is
10 × 9 × 8 = 720.
Now we use our ideas about counting permutations to assign poll watchers.
360 CHAPTER 5
Frances M. Roberts/Newscom
SOLUTION
a. As we noted earlier, the number of ways to arrange five items is
5! = 5 × 4 × 3 × 2 × 1 = 120
In terms of assigning poll watchers, there are five choices for the first polling
place, followed by four choices for the next place, and so on.
b. From part a there are 120 outcomes, and only one of them has each of the
watchers assigned to his or her home precinct. So the probability that each one
will be assigned to the polling place for the precinct where he or she lives is
1/120 or about 0.008.
c. There are five choices to watch the first polling place and four choices to
watch the second polling place. That makes 5 × 4 = 20 different arrangements.
d. As we have noted, the number of arrangements of 20 items is 20!, which is
(get ready!) 2,432,902,008,176,640,000. Yikes! This number is too big even for
most calculators, which will report at best an estimate. Generally speaking,
specialized mathematical software is needed to handle numbers of this size, and
hand calculation is out of the question.
as
20 × 19 × 18 × 17 × 16 × 15 × 14 × 13 × 12 × (11!)
11!
20 × 19 × 18 × 17 × 16 × 15 × 14 × 13 × 12 × (11 × 10 × 9 × 8 × 7 × 6 × 5 × 4 × 3 × 2 × 1) 20!
=
11! 11!
20! 20!
=
11! (20 − 9)!
So we can divide two factorials on our calculator to find the number of permutations:
60,949,324,800.
This method is often useful, and we call attention to it in the accompanying
summary.
25!
Permutations of 25 items taken 7 at a time =
(25 − 7)!
25!
=
18!
362 CHAPTER 5
Combinations
A permutation of a set of items is an arrangement of those items in a certain order.
By contrast, a combination of a set of items is a collection that does not take into
account the order in which they happen to appear.
count combinations. We find the number of permutations, then look to see how many
times each different combination is listed, and divide. Expressed as a formula, this is
Permutations
Combinations =
Repeated listings
In the case of permutations of k items, the number of times each combination is
listed is k!. (Why?) This observation yields the formula in the following summary.
364 CHAPTER 5
9 × /8 × 7 × 6
=
4/ × 3 × 2/ × 1
3
9/ ×7 × 6
=
3/ × 1
= 3×7×6
= 126
throughout the circuit and meet only rarely. Assuming there are 28 judges, there are
more than 3000 possible combinations of three-judge panels.”
a. Is the article correct in stating that there are more than 3000 possible
combinations consisting of three-judge panels? Exactly how many three-judge
panels can be formed from the 28-judge court?
b. The article says that the judges meet rarely. Assume that there are 28 judges.
How many 28-judge panels are there?
SOLUTION
a. We want to find the number of ways to choose 3 items from 28 items. We use
the combinations formula with n = 28 and k = 3:
28!
Combinations of 28 items taken 3 at a time =
3!(28 − 3)!
28!
=
3!25!
This fraction can be simplified by cancellation:
28! 28 × 27 × 26 × 25 × 24 × 23 × · · · × 1/
=
3!25! (3 × 2 × 1) × (25 × 24 × 23 × · · · × 1)
/
28 × 27 × 26
=
3×2×1
9 13
28× 27 × 26
=
3/ × 2/ × 1
= 3276
The article is correct in stating that there are more than 3000 possible
three-judge panels. There are, in fact, 3276 such combinations.
Crauder-4100237 QL November 6, 2014 12:59
366 CHAPTER 5
b. We are really being asked, “In how many ways can we choose 28 judges
from 28 judges to form a panel?” There is obviously only one way: All 28
judges go on the panel.
Let’s see what happens if we apply the combinations formula to this simple
case. We use the combinations formula with n = 28 and k = 28:
28!
Combinations of 28 items taken 28 at a time =
28!(28 − 28)!
28!
=
28!0!
=1
This is the same as the answer we found before. Remember that 0! is taken to
be 1, and this formula is one reason why.
b. What is the probability that exactly two of the cards are kings?
c. What is the probability that all of them are hearts?
SOLUTION
a. The total number of hands is the number of combinations of 52 items taken
5 at a time:
52!
Combinations of 52 items taken 5 at a time =
5!(52 − 5)!
52!
=
5!47!
= 2,598,960
17 Although the chart gives a probability of 0.00% for a Royal Flush, the probability is actually a very small positive
number.
Crauder-4100237 QL November 6, 2014 12:59
368 CHAPTER 5
© Cloki/Dreamstime.com
Various poker hands.
(Thus, there are more than 2.5 million different five-card poker hands.) This is
the number we will use in the denominator of the remaining calculations.
b. We must count how many ways we can choose five cards from the deck so
that exactly two are kings. To do this, we must choose two kings and three
others. There are four kings, from which we choose two. The combinations
formula gives
4!
Combinations of 4 items taken 2 at a time =
2!(4 − 2)!
4!
=
2!2!
=6
There are 48 cards that are not kings, and we must choose 3 from these. That is,
we need the number of combinations of 48 items taken 3 at a time:
48!
Combinations of 48 items taken 3 at a time =
3!(48 − 3)!
48!
=
3!45!
= 17,296
The Counting Principle says that there are 6 × 17,296 = 103,776 ways to do
both. Therefore, the probability we seek is 103,776/2,598,960. This is about
0.040 or 4.0%.
c. We need to count how many ways we can choose five cards from the deck so
that all of them are hearts. There are 13 hearts, and the number of ways to
choose 5 of them is the number of combinations of 13 items taken 5 at a time:
13!
Combinations of 13 items taken 5 at a time =
5!(13 − 5)!
13!
=
5!8!
= 1287
So the probability is 1287/2,598,960. This is about 0.0005 or 0.05%.
Crauder-4100237 QL November 6, 2014 12:59
370 CHAPTER 5
12. We select three people for a committee. The first chosen rotors. Four of the rotors must be initially set to A, but the
will hold the office of president, the second vice president, and remaining rotors can be set to any letter from B to Z. How
the third secretary. many initial settings are there for this new Enigma machine?
13. We go to the deli and buy three kinds of cheeses for next Suggestion: First select four rotors that will be set to A. (In
week’s lunches. how many ways can you do that?) Now determine how many
settings there are for the remaining six rotors. Finally, use the
14. We select four men and six women to serve on a committee. Counting Principle to put the two together.
15. We put five people on a basketball team. They get the po-
19. Permutations and passcodes. The following excerpt is from
sitions of point guard, shooting guard, small forward, power
a post that appeared at the Zscaler Security Cloud blog.
forward, and center.
16. Signals. The king has 10 (different) colored flags that he
uses to send coded messages to his general in the field. For
example, red-blue-green might mean “attack at dawn” and IN THE NEWS 5.10 Search
first number drawn was 257, corresponding to September 14, 30. How many committees can we make consisting of two
and young men born on that day were the first group to be women and three men?
called up. 31. How many committees can we make with Jack as commit-
tee chair? (Assume that only one member has that name!)
AP Photo
32. How many committees can we make with fewer men than
women? Suggestion: There could be three women and two
men, four women and one man, or five women.
33. Choosing 200 from 500. Is it true that the number of ways
of choosing 200 items (not considering order) from 500 items
is the same as the number of ways of choosing 300 items from
500 items? Explain how you arrived at your answer.
34. Mom’s Country Kitchen. A dinner order at Mom’s Coun-
try Kitchen consists of a salad with dressing, an entrée, and
three sides. The menu offers a salad with your choice of ranch,
bleu cheese, French, or Italian dressing; a choice of entrée
from chicken-fried steak, meat loaf, fried chicken, catfish, roast
beef, or pork chops; and, as sides, mashed potatoes, French
fries, fried okra, black-eyed peas, green beans, corn, rice, pinto
beans, or squash. How many possible dinner orders are avail-
Military draft lottery in 1969. able at Mom’s?
35–38. Marbles. In Exercises 35 through 38, we have a box
20. Assume that capsules were drawn at random. (This point with eight green marbles, five red marbles, and seven blue mar-
has been the subject of much discussion.) What is the proba- bles. We choose three marbles from the box at random with-
bility that a given young man would be in the first group to be out looking. Write each probability in decimal form rounded
called up? to three places.
21. How many different outcomes are possible for such a 35. What is the probability that they will all be green?
drawing? Leave your answer in terms of factorials.
36. What is the probability that they will all be red?
22. Young men born on the same day of the year were distin-
37. What is the probability that they will all be blue?
guished according to the first letter of their last, first, and mid-
dle names. This was done by randomly selecting the 26 letters 38. What is the probability that they will all be the same color?
of the alphabet. How many permutations of those 26 letters 39. Money. You have the following bills: a one, a two, a five,
are there? Leave your answer in terms of factorials. a ten, a twenty, and a hundred. In how many ways can you
23. Candy. On Halloween, a man presents a child with a distribute three bills to one person and two bills to a second
bowl containing eight different pieces of candy. He tells her person?
that she may have three pieces. How many choices does she 40. Poll watchers. A certain political organization has 50 poll
have? watchers. Of these, 15 are to be sent to Ohio, 15 to Penn-
24. More candy. On Halloween, a man presents a child with a sylvania, 15 to Florida, and the remaining 5 to Iowa. In how
bowl containing eight different pieces of candy and six differ- many ways can we divide the poll watchers into groups to be
ent pieces of gum. He tells her that she may have four of each. sent to the different states? First write your answer in terms
How many choices does she have? of factorials, and then give an estimate such as “about 1020 .”
Suggestion: Choose 15 to go to Ohio, then choose 15 from the
25–28. Committees. In Exercises 25 through 28, suppose remaining 35 to go to Pennsylvania, and so on.
there is a group of seven people from which we will make a
41–43. Cards. In Exercises 41 through 43, we draw three
committee.
cards from a full deck. Write each probability in decimal form
25. In how many ways can we pick a three-person committee? rounded to four places.
26. In how many ways can we pick a four-person committee? 41. What is the probability that they are all red?
Compare your answer with what you found in Exercise 25. 42. What is the probability that they are all hearts?
27. In how many ways can we pick a five-person committee? 43. What is the probability that they are all aces?
28. In how many ways can we pick a two-person committee?
44–46. More cards. In Exercises 44 through 46, we draw
Compare your answer with what you found in Exercise 27.
five cards from a full deck. Write each probability in decimal
29–32. Committees of men and women. We want to select form rounded to four places.
a committee of five members from a group of eight women 44. What is the probability that you have at least one ace?
and six men. The order of selection is irrelevant. Exercises 29 Suggestion: First find the probability that you get no aces.
through 32 refer to this information.
45. What is the probability that they are all face cards (jacks,
29. What is the total number of committees we can make? queens, or kings)?
Crauder-4100237 QL November 6, 2014 12:59
372 CHAPTER 5
46. What is the probability that exactly three of them are Exercises 50 through 54 refer to this article. In Exercises 51,
jacks? Suggestion: Think of first picking three jacks. Then pick 52, and 54, you do not need to simplify your answers; just
the remaining two cards at random. leave your answers in terms of factorials. In one of the tech-
47–48. The Ninth Circuit Court. Exercises 47 and 48 re- nology exercises, you will be asked to simplify those answers
fer to the article on the Ninth Circuit Court summarized in (see Exercise 57).
Example 5.26. 50. The article gives an example of a worst-case scenario. Is it
47. Suppose a three-judge panel has a designated chief justice really a worst-case scenario? Can you think of one that might
and that we count two panels with the same members as being be worse than this?
different if they have different chief justices. That is, we count 51. Suppose that a trailer park has 50 sets of 10 trailers, each
the panel Adams, Jones, Smith, with Smith as chief, as a differ- set keyed the same. (This is what the news article seems to
ent panel than Adams, Jones, Smith, with Jones as chief. How suggest would be a “worst case.”) If you have one of the 50
many such panels can be formed from the 28-judge court? keys and try unlocking 10 trailers at random from the 500,
48. Can you think of a reason why a large number of three- what is the probability that you can open at least one of them?
judge panels could have an adverse effect on the court’s Suggestion: First calculate the probability that you can open
performance? none of them.
49. Birthdays. Consider the following problem: Assume that 52. Suppose that a trailer park has 50 sets of 10 trailers, each
Jack and Jane were both born in August and that they were set keyed the same. If you have one of the 50 keys and you
born on different days. Find the probability that both were try unlocking 34 trailers at random from the 500, what is the
born in the first week of August. probability that you can open at least one of them? Suggestion:
Here is a purported solution: The number of ways of choos- First calculate the probability that you can open none of them.
7!
ing 2 days from the first 7 is = 21. Next we get the number 53. Now let’s assume that each of the 50 keys fits the same
5!2! number of trailers and that there are 118,000 trailers. Then
of ways of picking 2 birthdays out of 31 (because August has
each key would fit 118,000/50 = 2360 trailers. If we select
31 days). There are 31 days for the first birthday and 30 for
two trailers at random from the 118,000 trailers, what is the
the second. That is 30 × 31 = 930 ways to choose 2 birthdays
probability that the same key will fit them both? That is, what
from 31. So the probability of both birthdays occurring in the
21 is the probability that the two trailers are keyed the same?
first week is . Write the answer as a simple fraction and in decimal form
930
What is wrong with this solution? What is the correct an- rounded to two places.
swer? Suggestion: Remember that counting using permutations 54. Let’s make the same assumptions as in Exercise 53. If we
shouldn’t be mixed with counting using combinations. select 500 trailers at random from the 118,000, what is the
Exercises 50 through 54 are more challenging and are suitable probability that we get 50 sets (one for each key) of 10 trailers
for group work. See also Exercise 57. each that are keyed the same? (This is what the news article
seems to suggest would be a “worst case.”)
50–54. Changing locks. The following article is concerned
with the continuing plight of evacuees from Hurricanes Katrina 55. History. The earliest detailed work on combinations was
and Rita in 2005. apparently done by Rabbi Abraham ben Meir ibn Ezra in the
twelfth century in a text on astrology. Research the problems
on combinations of interest to ibn Ezra, and determine some
of his other scholarly pursuits.
IN THE NEWS 5.11 Search
Home TV & Video U. S. World Politics Entertainment Tech Travel 56. History. Exercise 17 provides a very brief introduction to
the German Enigma machine used in World War II. Investi-
FEMA Changing Locks on Trailers gate the Enigma code, including its eventual cracking by the
LARA JAKES JORDAN August 15, 2006
code breakers at Bletchley Park, England, and report on your
finding.
W ASHINGTON—FEMA will replace locks on as
many as 118,000 trailers used by Gulf Coast
hurricane victims after discovering the same key could
The following exercise is designed to be solved using technol-
ogy such as a calculator or computer spreadsheet.
open many of the mobile homes.
One locksmith cut only 50 different kinds of keys
57. More on changing locks. Simplify the answers to Exer-
for the trailers sold to FEMA, officials said Monday. That cises 51, 52, and 54. Write the probability in decimal form
means, in an example of a worst-case scenario, one key rounded to two places. Warning: For Exercise 54, this may
could be used to unlock up to 10 mobile homes in a require more sophisticated technology.
park of 500 trailers.
FEMA officials said such a situation was unlikely,
but they still moved to warn storm evacuees living in
Louisiana and Mississippi trailer parks of the security
risk.
Crauder-4100237 QL November 6, 2014 12:59
374 CHAPTER 5
T oday the vast majority of Americans either gamble recreationally and experi-
ence no measurable side effects related to their gambling, or they choose not
to gamble at all. Regrettably, some of them gamble in ways that harm themselves,
their families, and their communities. This Commission’s research suggests that
86 percent of Americans report having gambled at least once during their lives.
Sixty-eight percent of Americans report having gambled at least once in the past
year. In 1998, people gambling in this country lost $50 billion in legal wagering,
a figure that has increased every year for over two decades, and often at double-
digit rates. And there is no end in sight: Every prediction that the gambling market
was becoming saturated has proven to be premature.
Once exotic, gambling has quickly taken its place in mainstream culture: Tele-
vised megabucks drawings; senior citizens’ day-trips to nearby casinos; and the
transformation of Las Vegas into family friendly theme resorts, in which gam-
bling is but one of a menu of attractions, have become familiar backdrops to
daily life.
Some form of legal gambling has spread to all but two states.18 Examples in-
clude casinos, state lotteries, and parimutuel wagering on horse or dog racing. With
gambling now so prevalent, an educated person should understand how it works.
Gambling establishments have one thing in common: The odds are stacked against
the player. In fact, a good title for our discussion could be, “Why you shouldn’t
gamble.”
The mathematical concept of expected value is key to understanding games of
chance. Theoretical expected values are connected to empirical observations through
the law of large numbers. In this section, we will discuss these concepts and give
applications to gambling, the insurance industry, and other areas. To introduce the
concept of expected value, we take a quick look at lotteries.
5.5 Expected value and the law of large numbers: Don’t bet on it 375
Mike Baldwin/cartoonstock.com
The Iowa State Lottery illustrates the notion of expected value. It’s not whether
you win or how many times you win, it’s the overall net payback you can expect
that’s important. Expected value has a precise mathematical description that we will
introduce shortly. But for the moment, an informal notion of expected value will
suffice.
376 CHAPTER 5
Our informal definition of the expected value is that it tells the average amount you
can expect to win or lose per play in the long run. But there is a theoretical expected
value that has meaning for a single play of a game. Later we will give a formula
for calculating the theoretical expected value, but first we explain the idea using slot
machines.
Slot machines are designed so that each outcome has a certain probability and a
certain payout. According to the manufacturers, the machines are designed so that
every time you pull the handle, the result is independent of whatever has happened
before. The overall result is that each machine has a theoretical expected value that
can be set by the manufacturer.
According to the Nevada Gaming Control Board, the payback on all slot ma-
chines in a recent year was 94.93%, which means that casinos kept, on average,
about 5% of all money put into them. This suggests that the theoretical expected
value for the casino on slot machines is about 5% (and therefore the expected value
for the player is −5%). A return of 5% might not seem enough to support a lavish
casino, but it can because of the huge amount of money involved.
Crauder-4100237 QL November 6, 2014 12:59
5.5 Expected value and the law of large numbers: Don’t bet on it 377
The connection between the theoretical expected value and the effect of that ex-
pected value in the long run is expressed by the law of large numbers, which applies to
games for which repeated trials are independent. (Independent events were discussed
in Section 5.3.)
To illustrate the calculation, we look at a simple game. A fair die is rolled. If the
result is a 1 or 2, I win $9. If any other number comes up, I lose $3. To calculate the
expected value for me, we must first find the probability for a win and for a loss. For
Crauder-4100237 QL November 6, 2014 12:59
378 CHAPTER 5
a win, there are two favorable outcomes (1 and 2) out of six total outcomes. Hence,
the probability of a win is
Favorable outcomes
P(Win) =
Total outcomes
2 1
= =
6 3
We find the probability of a loss in a similar fashion: There are 4 losing numbers out
of 6, so the probability of a loss is 4/6 = 2/3. We use these probabilities to calculate
the expected value for me:
Expected value = P(Win) × Profit − P(Lose) × Loss
1 2
= × $9 − × $3
Matytsin Valery/ITARTASS/Corbis
3 3
= $1
My expected value for this game is $1 per play. So if I play this game many times,
I can expect to win an average of $1 for each time I play. This is a very attractive
game. Typical casino games are not as lucrative for the player, as is illustrated by the
following example.
A standard roulette wheel has numbers 1 through 36 alternately colored red and
black. It also has a green 0 and a green 00 called “double zero.”19 The wheel is
spun with a small white ball inside. While the wheel is spinning, the ball falls into
a numbered slot, which determines the winners. Successive spins of a wheel yield
independent results.
There are many wagers you can place at the roulette table. For example, you may
bet on the color (red or black), or on whether the number is odd or even, or on a
single number. Note that, even though 0 is in fact an even number, in roulette both
0 and 00 count as neither odd nor even. These green numbers give the edge to the
house.
5.5 Expected value and the law of large numbers: Don’t bet on it 379
In addition to slot machines and roulette, casinos offer a variety of games, among
which are blackjack, keno, craps, and various kinds of poker. But they all have one
thing in common: The theoretical expected value for the player is negative. And no
system of placing wagers can change this fact.20
Many people enjoy buying lottery tickets or going to casinos even though they
assume they will probably lose some money. Some view it as buying entertainment,
just as they might pay to go to a movie. A few will get lucky once in a while and win
money, but the law of large numbers ensures that the vast majority will eventually
lose and that casinos and state lotteries will be profitable for the proprietors.
Fair games
In probability theory, a game is called fair if its expected value is zero. Intuitively then,
a fair game is one in which no participant has an advantage over others. Clearly, we
would not call the game of roulette fair, because it gives the edge to the casino.
For example, suppose Sue and Alice toss a coin. Sue wins $2 from Alice if the coin
comes up heads and loses $2 to Alice if it comes up tails. Let’s determine whether
this is a fair game.
There are two outcomes: heads (Sue wins) and tails (Sue loses). The probability
that Sue wins is 1/2, in which case she wins $2. The probability that Sue loses is also
1/2, and then she loses $2. Sue’s expected value is
Because the expected value is 0, this is a fair game. If Sue plays this game many
times, she can expect to lose about as much as she wins. This does not mean that she
is certain to come out even in the end, but it would be unusual for her either to win
or to lose a large amount of money over a long stretch.
20 “Card counting” at blackjack can produce a very small positive expected value for the player. But most casinos do
not allow the practice.
Crauder-4100237 QL November 6, 2014 12:59
380 CHAPTER 5
SOLUTION
The probability of a win is 1/6, and the probability of a loss is 5/6. Thus, we find
that my expected value is
When you take the SAT, winning corresponds to giving a correct answer, with a profit
of 1 point, and losing corresponds to giving an incorrect answer, with a loss of 1/4
point.
Now we find the probabilities. There are five possible answers one can fill in,
and one of them is correct. The probability of getting a correct answer by random
guessing is thus 1/5, so the probability of an incorrect answer is 4/5.
Thus, the expected value of the score is
This means that random guessing gives no advantage in the long run—which is
the purpose of subtracting 1/4 point for incorrect answers. Because the expected
value is 0, the awarding of points on the SAT exam is fair in the sense that we have
been discussing.
21 In March 2014, the College Board announced that in 2016 it would eliminate the penalty for wrong answers on the
SAT.
Crauder-4100237 QL November 6, 2014 12:59
5.5 Expected value and the law of large numbers: Don’t bet on it 381
Risk assessment
So far we have discussed the concept of expected value primarily in the context of
games of chance. But this concept is relevant whenever we weigh risks and benefits
by considering the likelihood of various outcomes. Applications of expected values
are found in various contexts, including manufacturing, finance, economics, and the
insurance industry. Blaise Pascal, a famous mathematician and religious philosopher
of the seventeenth century, applied the concept of expected value to belief in the
existence of God. See Exercise 52.
The concept of expected value plays a key role in the insurance industry. Com-
panies that insure property, for example, have to pay for catastrophic damages oc-
casionally, such as those caused by a hurricane. The companies must set the costs
of premiums to compensate for such losses but also provide a profit (i.e., a positive
expected value) in the long run. The experts whom they employ to figure this out are
called actuaries.
Bill Koplitz/FEMA NewsPhoto
Storm damage.
Crauder-4100237 QL November 6, 2014 12:59
382 CHAPTER 5
This means that, on average, the insurance company earns a profit of $131.20
on each policy it writes. If the insurance company issues 1000 such policies, it can
expect an annual profit of $131.20 × 1000 = $131,200.
5.5 Expected value and the law of large numbers: Don’t bet on it 383
384 CHAPTER 5
expected losses. If you play the lottery many times, the law of large numbers virtually
assures that you will lose the amount of money predicted by the expected value. One
may play the lottery for fun or as a way of contributing to your state. If one does it
expecting to win, disappointment is highly likely.
Further questions of this type may be found in the What Do You Think section of
exercises.
5.5 Expected value and the law of large numbers: Don’t bet on it 385
16. Gambler’s fallacy. One sports fan, in describing how his 25. Expected value II. Consider the following game: The prob-
mediocre team went from a five-game winning streak to a ability of winning a profit of twice your wager is 1/7, the prob-
five-game losing streak, had this summary: “The law of aver- ability of winning a profit equal to your wager is 2/7, and all
ages was bound to catch up with a team like this.” Is this state- other probabilities of winning are 0. What is your expected
ment an application of the law of large numbers, an example value for this game?
of the gambler’s fallacy, or neither? Explain your answer. 26. Expected value III. What is your expected value for a game
17. Martingale method. Suppose a game pays even money. whose probabilities and outcomes are given in the following
That is, if you win, you get whatever you bet, and if you lose, table?
you lose that same amount. Here’s a strategy called the Mar-
tingale method. Probability Profit
If you win, bet the same amount again. But every time you 0.2 You win 1.5 times your wager.
lose, you play again and double your previous bet. For exam- 0.4 You win your wager.
ple, suppose you bet a dollar in such a game. If you win, you 0.2 You lose 2 times your wager.
bet $1 again. If you lose, then you play again but this time 0.2 You win nothing.
you bet $2. If you win, you’re back to even and you bet $1
next time. If you lose, you play again and bet $4. If you win, 27. Expected value IV. What is your expected value for a game
you’re back to even, and so forth. This strategy seems appeal- whose probabilities and outcomes are given in the following
ing because no matter how many times you lose, eventually table?
you’re bound to win one time, and that puts you back to even.
If you win twice in a row, you have a profit. So in theory you
Probability Profit
can’t lose in the end. But this strategy is flawed. Can you find
the flaw? 0.1 You win your wager.
0.4 You win half your wager.
18. A coin game. In a game you bet $1 and toss two coins. If 0.2 You lose 3 times your wager.
two heads come up, you win a profit of 75 cents. If two tails 0.3 You win nothing.
come up, no one wins anything. Otherwise, you lose your dol-
lar. What is your expected value for this game? Don’t round
your answer. 28. Exam score I. Each question on a multiple-choice exam has
four choices. One of the choices is the correct answer, worth
19. How much will we win? We decide to open a casino where 5 points, another choice is wrong but still carries partial credit
we offer only one game. We win a dollar 60% of the time, and of 1 point, and the other two choices are worth 0 points. If a
our customer wins a dollar 40% of the time. How much profit student picks answers at random, what is the expected value of
would we expect to have if 5000 games are played? his or her score for a problem? If the exam has 30 questions,
20. Make it fair. You play a gambling game with your friend what is his or her expected score?
in which you win 30% of the time and lose 70% of the time. 29. Exam score II. Each question on a multiple-choice exam
When you lose, you lose $1. What profit should you earn when has five choices. One of the choices is the correct answer,
you win in order for the game to be fair? worth 5 points. There are two other choices that are wrong
21. Voting. A poll taken in Massachusetts in October 2012 but still carry partial credit of 2 points each. The other two
showed Obama leading Romney by 60% to 40% in the U.S. choices are worth 0 points. If a student picks answers at ran-
presidential race. Let’s assume that these numbers represent dom, what is the expected value of his or her score for a prob-
probabilities. I wished to make a bet with my friend on who lem? If the exam has 40 questions, what is his or her expected
would win Massachusetts, and we agreed to base our bet on score?
that poll. I predicted a victory for Obama and would win a 30. Insurance I. Suppose a company charges an annual pre-
profit of $1 if I were right. Assuming we wished this to be a mium of $450 for a fire insurance policy. In case of a fire claim,
fair game, how much should I have agreed to pay my friend if the company will pay out an average of $100,000. Based on ac-
he was right and Massachusetts went to Romney? tuarial studies, it determines that the probability of a fire claim
22. Card game. A card game goes like this: You draw a card in a year is 0.004. What is the expected annual profit of a fire
from a 52-card deck. If it is a face card (jack, queen or king), insurance policy for the company? What annual profit can the
you win $5; otherwise, you lose $2. What is your expected company expect if it issues 1000 policies?
value for this game? 31. Insurance II. Suppose a company charges an annual pre-
23. Modified card game. Exercise 22 described a card game mium of $100 for an insurance policy for minor injuries. Actu-
in which you draw a card from a 52-card deck. Suppose that arial studies show that in case of an injury claim, the company
this game is modified so that if you draw an ace, then no will pay out an average of $900 for outpatient care and an av-
money changes hands. What is your expected value for this erage of $3000 for an overnight stay in the hospital. They also
game? determine that, on average, each year there are five claims made
that result in outpatient care for every 1000 policies and three
24. Expected value I. What is your expected value for a game
claims made that result in an overnight stay out of every 1000
where your probability of winning is 1/6 and your profit is four
policies. What is the expected annual profit of an insurance
times your wager?
policy for the company?
Crauder-4100237 QL November 6, 2014 12:59
386 CHAPTER 5
32. Tires. A company manufactures tires at a cost of $60 each. 40. Even or odd. In roulette, you may bet whether the num-
The following are probabilities of defective tires in a given pro- ber that comes up is odd (or even). (Remember that 0 does
duction run: probability of no defective tires: 10%; probability not count as either odd or even.) The wager pays even money,
of 1% defective tires: 30%; probability of 2% defective tires: that is, if you win, you get whatever you bet, and if you lose,
40%; probability of 3% defective tires: 20%. When a defec- you lose that same amount. What is your expected value for
tive tire is detected, it must be removed from the assembly line this bet?
and replaced. This process adds an extra $5 to the cost of the 41–44. Roulette and gambler’s fallacy. Roulette wheels in
replaced tire. What is the expected cost to the company of a Monte Carlo have no double zero, so there are 37 numbers on
batch of 3000 tires? a wheel. Exercises 41 through 44 use this fact.
33. Chairs. A company manufactures chairs at a cost of $30 41. What is the probability of getting an even number on one
each. The following are probabilities of defective chairs in spin of a roulette wheel in Monte Carlo? Express your an-
a given production run: probability of 1% defective chairs: swer as a fraction and as a percentage rounded to one decimal
40%; probability of 2% defective chairs: 35%; probability place.
of 3% defective chairs: 25%. When a defective chair is de-
tected, it must be removed from the assembly line and replaced. 42. You spin a roulette wheel in Monte Carlo 26 times. What
This process adds an extra $10 to the cost of the replaced is the probability of getting an even number on all 26 spins?
chair. What is the expected cost to the company of a batch of Express your answer as a fraction, then express the chances
500 chairs? using a phrase such as “1 in 5 billion.”
43. In 1913, at one fair roulette table in Monte Carlo, an even
34–35. Expected rate of return. Exercises 34 and 35 refer to
number came up on 26 spins in a row. What was the probabil-
the following concept. In finance, the notion of expected value
ity of getting an even number on the 27th spin?
is used to analyze investments for which the investor has an
estimate of the chances associated with various returns (and 44. Explain how the historical event described in Exercise 43
losses). For example, suppose you have the following infor- is consistent with the law of large numbers.
mation about one of your investments: With a probability of 45–48. Powerball. Exercises 45 through 48 refer to the fol-
0.9, the investment will return 20 cents for every dollar you lowing description of Powerball, a lottery in which most U.S.
invest, and with a probability of 0.1, the investment will lose states participate. To play, you pay $2 and choose five white
50 cents for every dollar you invest. The expected rate of re- numbers between 1 and 59. Then you choose a single red ball,
turn for this investment is calculated the way we calculate the the Powerball, from a second set of numbers that range from
expected value of a game: Multiply the probability of each out- 1 to 35. To win the Powerball jackpot, you must match all five
come by the amount you earn (or by minus the amount if you white numbers, plus the Powerball number. Like other lottery
lose) and add up these numbers. games, varying amounts of money can also be won by match-
34. Calculate the expected rate of return for the investment ing fewer numbers, allowing for nine different ways to win a
described above. Powerball prize. Here they are:
35. Here is information about another investment: With a • 5 white + Powerball = Powerball jackpot prize,
probability of 0.5, it will return 20 cents for every dollar you $40 million minimum (Probability of winning: 1 in
invest; with a probability of 0.3, it will return 10 cents for every 175 million)
dollar you invest; and otherwise, it will lose 90 cents for every • 5 white = $1 million prize (Probability of winning: 1 in
dollar you invest. Calculate the expected rate of return for this 5.2 million)
investment. Is this a good investment? • 4 white + Powerball = $10,000 prize (Probability of
winning: 1 in 650,000)
36. Betting on 11. In a dice game, you win if the two dice come • 4 white = $100 prize (Probability of winning: 1 in
up 11. Otherwise, you lose $1. What should be the profit for 19,000)
winning to make this game fair? • 3 white + Powerball = $100 prize (Probability of
37. A fair game. You roll a pair of dice. You bet $1 that you winning: 1 in 12,000)
will get a 2, 3, or 12. If you win, you are paid $9 (a profit of • 3 white = $7 prize (Probability of winning: 1 in 360)
$8). Show that this is a fair game. • 2 white + Powerball = $7 prize (Probability of
winning: 1 in 706)
38. The real game. In Exercise 37, you rolled a pair of dice, • 1 white + Powerball = $4 prize (Probability of
and you bet $1 that you will get a 2, 3, or 12. In the casino winning: 1 in 111)
game called craps, the casino pays you $8 (a profit of $7) if • Powerball only = $4 prize (Probability of winning:
these numbers come up on your “opening roll.” What is your 1 in 55)
expected value for this bet expressed as a percent?
The Powerball jackpot starts at $40 million and increases after
39. Betting on red. Suppose you bet $1 on red at the roulette each drawing with no Powerball jackpot hit. The largest Power-
wheel. There are 18 red numbers and 20 nonred numbers (in- ball jackpot on record (as of 2014) is $590.5 million, won in
cluding the green 0 and 00). You win $1 if a red number comes May 2013, by a ticket sold in Florida.
up, and you lose $1 if any other number comes up. What is
your expected value for this wager? 45. Powerball players may choose to have the numbers gener-
ated randomly by the computer—called a Quick Pick—rather
Crauder-4100237 QL November 6, 2014 12:59
5.5 Expected value and the law of large numbers: Don’t bet on it 387
than choosing the numbers themselves. Do you think it mat- in Section 5.3 about “double sixes.” Write a paper about this
ters? Why? historical connection.
46. How many possible combinations of five numbers chosen 50. History. Blaise Pascal was an eminent scientist, mathemati-
from 1 through 59 are there? cian, and religious philosopher. His work led to the develop-
47. How many possible combinations of six Powerball num- ment of probability theory and the notion of expected value.
bers are there, where five white numbers are chosen from In a section of his writings, he discusses belief in the existence
1 through 59 and one red Powerball number is chosen from of God in terms of expected value. Write a paper about Pascal’s
1 through 35? wager.
48. Suppose several people chose the same winning numbers 51. History. Choose one of the games discussed in this section
for the jackpot. Wouldn’t that cause the state to lose money? and write a paper about its history and evolution.
(This may require some investigation.) 52. History. If your state or a neighboring state has a lottery,
49. History. Historically, many of the ideas from probability write a paper about its history, the revenue it generates, and
were developed to understand games of chance. See Exercise 65 how it operates.
W-2. Lie? A friend said he was told that the expected value of a game was $1.00, but
he played twice and lost $1.50. He said he must have been told a lie. How would
you respond?
Exercise 11 provides additional information relevant to this topic.
W-3. Fair game. What does it mean to say that a game is fair? Does it mean that you
can never lose money?
Exercise 20 provides additional information relevant to this topic.
W-4. Insurance. Investigate the role of expected value in the insurance industry. Re-
port on your findings.
Exercise 30 provides additional information relevant to this topic.
W-5. Expected value and manufacturing. We have seen applications of expected value
to manufacturing. After a manufacturer calculated the expected cost of a production
batch, what could the manufacturer do with this information?
Exercise 32 provides additional information relevant to this topic.
W-6. Roulette. Explain how the standard roulette wheel gives the edge to the casino.
Exercise 39 provides additional information relevant to this topic.
W-7. Gambler’s fallacy—or not. If we flip a coin 10 times and each time it comes up
heads, then the gambler’s fallacy would lead us to bet tails for the next toss because
“tails are due.” For independent events, the gambler’s fallacy is indeed a fallacy.
But suppose you flip a coin 20 times, and each time it comes up heads. Would you
conclude that heads and tails are equally likely on the next toss, or would the first 20
tosses make you suspect that the coin is in fact not a fair coin? Discuss the conditions
under which you might question a game that purports to be fair.
Exercise 43 provides additional information relevant to this topic.
Crauder-4100237 QL November 6, 2014 12:59
388 CHAPTER 5
W-8. Law of large numbers. Explain in your own words why the law of large numbers
does not support the gambler’s fallacy.
W-9. Adjusting expected value. The text states that the expected value for the Iowa
State Lottery is about −46% for the player. What could the state do to increase (or
decrease) this expected value?
CHAPTER SUMMARY
The mathematical study of probability is an attempt to model and understand un-
certainty. We examine how a probability is obtained, what it means, and where it is
useful in the real world.
For example, suppose we want to find the theoretical probability that I roll a 3 with
two dice. We need to count the number of possible outcomes for the two dice, which
is 6 × 6 = 36. The number of favorable outcomes, that is, those that give a 3, is
2 (either one and two or two and one). Then the probability is 2/36 = 1/18. In
cases involving identical items, such as coins or dice, it may help to label the items
to distinguish them.
Here are two useful formulas:
and
and
A conditional probability is the probability that one event occurs given that another
has occurred. Positive and negative predictive values, as well as sensitivity and speci-
ficity, are examples of conditional probabilities.
n!
(n − k)!
The number of ways to select k items from n items when order makes no differ-
ence is the number of combinations of n objects taken k at a time. The formula for
calculating this is
n!
k!(n − k)!
390 CHAPTER 5
KEY TERMS
probability, p. 310 negative predictive value (NPV), combination of n items taken k at a
disjunction, p. 316 p. 331 time, p. 361
empirical probability, p. 319 conditional probability, p. 335 combination, p. 362
sensitivity, p. 328 Counting Principle, the, p. 343 expected value, p. 375
specificity, p. 329 independent, p. 349 law of large numbers, p. 377
prevalence, p. 330 permutation, p. 357 gambler’s fallacy, p. 377
positive predictive value (PPV), permutations of n items taken k at a fair, p. 379
p. 331 time, p. 359
CHAPTER QUIZ
1. If I draw a card at random from a standard deck of 52 cards, Answer PPV 25.0%; NPV 88.9%
what is the probability that I draw a red queen? Write your an-
If you had difficulty with this problem, see Example 5.11.
swer as a fraction.
Answer 2/52 = 1/26 6. How many four-letter codes can we make if the first letter
is A, B, or C and the remaining letters are D or E? Letters may
If you had difficulty with this problem, see Example 5.1. be repeated. One such code is ADDE.
2. Suppose I flip two identical coins. What is the probability Answer 24
that I get a head and a tail? Write your answer as a fraction.
If you had difficulty with this problem, see Example 5.15.
Answer 1/2
7. It has been found that only 1 in 100 bags of M&M’s have
If you had difficulty with this problem, see Example 5.2. five more than the average number of M&Ms. Only 1 in
3. Suppose a softball team has 25 players. Assume that 13 bat 20 boxes of Cracker Jacks has a really good prize. I buy a box
over .300, 15 are infielders, and 5 are both. What is the prob- of Cracker Jacks and a bag of M&Ms. What is the probability
ability that a player selected at random is either batting over that I get a really good prize and five more than the average
.300 or an infielder? Write your answer as a fraction. number of M&Ms? Write your answer as a fraction.
Answer 23/25 Answer 1/2000
If you had difficulty with this problem, see Example 5.6. If you had difficulty with this problem, see Example 5.19.
4. A new medical test is given to a population whose disease 8. In how many ways can I choose 4 people from a group of
status is known. The results are in the table below. 10 and make them president, vice president, secretary, and trea-
surer? Write your answer in terms of factorials.
12. A company manufactures coats at a cost of $20 each. The extra $10 to the cost of the replaced coat. What is the expected
following are probabilities of defective coats in a given produc- cost to the company of a batch of 4000 coats?
tion run: probability of no defective coats: 5%; probability of Answer $80,460
1% defective coats: 75%; probability of 2% defective coats:
20%. When a defective coat is detected, it must be removed If you had difficulty with this problem, see Example 5.34.
from the assembly line and replaced. This process adds an
Crauder-4100237 QL November 10, 2014 11:34
6 STATISTICS
T
he following excerpt is from an article that appeared at the Web site of POLICE
magazine.
down the numbers Home TV & Video U. S. World Politics Entertainment Tech Travel
This article refers to “statistics” and claims that law enforcement agencies mis-
report statistics about crime. Statistics is the science of gathering and analyzing data
393
Crauder-4100237 QL November 10, 2014 11:34
394 CHAPTER 6
with the objective of either describing the data or drawing conclusions from them.
Descriptive statistics refers to organizing data in a form that can be more easily un-
derstood. Inferential statistics refers to the process of drawing conclusions from the
data, along with some estimate of how accurate those conclusions are. In this chap-
ter, we will examine both descriptive and inferential statistics in various contexts,
including polling and medical research.
6.1 Data summary and presentation: Boiling down the numbers 395
When there are many data points, they may be presented using a frequency table,
which shows how often each data point occurs. Frequency tables can be used to
calculate the mean, median, and mode, as the next example illustrates.
Find the mean, median, and mode for the number of goals scored per game.
Round the mean to one decimal place.
Paul Thomas/Getty Images
SOLUTION
To find the mean, we add the data values (the total number of goals scored) and
divide by the number of data points. To find the total number of goals scored, for
each entry we multiply the goals scored by the corresponding number of games. Then
we add. For example, there were 14 games in which exactly 2 goals were scored, so
these games contribute 14 × 2 = 28 goals. The total number of goals scored is
5 × 0 + 10 × 1 + 14 × 2 + 14 × 3 + 10 × 4 + 2 × 5 + 1 × 6 + 1 × 7 = 143
Now we find the number of data points, which is the total number of games:
5 + 10 + 14 + 14 + 10 + 2 + 1 + 1 = 57
The mean is the total number of goals scored divided by the number of games
played, so
143
Mean =
57
or about 2.5.
Now we find the median. The number of games is 57, which is odd, so we count
from the bottom to find the 29th lowest total goal score. This is 2, so
Median = 2
Crauder-4100237 QL November 10, 2014 11:34
396 CHAPTER 6
Because 2 and 3 occur equally often (14 times each) and more frequently than
any other, this data set is bimodal. We have
Modes = 2 and 3
Thus, on average, the teams combined to score 2.5 goals per game. Half of the
games had goals totaling 2 or more (almost half had goals more than 2), and the
most common numbers of goals scored in a Chelsea FC game were 2 and 3.
The list includes the price of one luxury home. Calculate the mean and median of
this data set. (Round the mean to one decimal place.) Which of the two is more
appropriate for describing the housing market?
SOLUTION
To calculate the mean we add the data values and divide by the number of data
points:
80 + 120 + 125 + 140 + 180 + 190 + 820
Benny Evans
1655
Mean = =
7 7
or about 236.4 thousand dollars. This is the average price of a home.
The list of seven prices is arranged in order, so the median is the fourth value,
140 thousand dollars.
Note that the mean is higher than the cost of every home on the market except for
one—the luxury home. The median of 140 thousand dollars is more representative
of the market.
An outlier is a data point that is significantly different from most of the others.
For example, the luxury home in Example 6.2 represents an outlier.
Crauder-4100237 QL November 10, 2014 11:34
6.1 Data summary and presentation: Boiling down the numbers 397
T he median price for newly built single-family homes and condos in San Diego
County shot up to a record high in December, while the number of new homes
sold hit a record low for a December.
Last month, 284 new residences sold at a median $699,000, a 48 percent
hike over the $473,250 median new home price in December 2012, when 419
new homes sold. The previous record for a median new home price was $553,000
in February 2008, when 330 units closed escrow, real estate tracker DataQuick
reported Tuesday.
The new home sales may have skewed the data up due to a number of high-
end closings scheduled to be on the books by the end of 2013.
398 CHAPTER 6
We note that in everyday usage we may speak of a number that is in the first
quartile. We mean that it is less than the first quartile value, which means that it is in
the lowest one-quarter of the data. We say that a number is in the second quartile if it
is between the first quartile value and the median—in the next-to-lowest one-quarter
of the data. Similarly, being in the third quartile means being in the next-to-highest
one-quarter of the data, and being in the fourth quartile means being in the highest
one-quarter of the data.
We use quartiles to make the five-number summary for data.
The five-number summary does a much better job of describing data than any
single number could. If the five numbers are relatively close together, the summary
tells us that the data are fairly closely bunched about the median. If, in contrast, the
five numbers are widely separated, we know that the data are spread out as well.
Each year Forbes magazine publishes a list it calls the Celebrity 100, which “takes
a stab at measuring that most ephemeral of riches: fame.” The accompanying table
shows the top nine names on the list for 2013, ordered according to the ranking of
Forbes. The table also gives the incomes of the celebrities between June 2012 and
June 2013.
6.1 Data summary and presentation: Boiling down the numbers 399
The median of this upper half is 90, so the third quartile of incomes is $90 million.
Thus, the five-number summary is
Minimum = $53 million
First quartile = $56.5 million
Median = $77 million
Third quartile = $90 million
Maximum = $125 million
What does the five-number summary in Example 6.3 show? All of the incomes
are between $53 million and $125 million. The median income is $77 million, so as
many of the incomes are above that level as are below. The lowest one-quarter of all
incomes are between $53 million and $56.5 million, the next one-quarter are between
$56.5 million and the median of $77 million, the next one-quarter are between the
median of $77 million and $90 million, and the highest one-quarter are between $90
million and $125 million. The summary provided by this breakdown gives a general
sense of how the incomes are spread out.
In our example, the number of data points was small, but if we were working
with 100,000 data points, for example, the five-number summary would be a big
help.
Boxplots
There is a commonly used pictorial display of the five-number summary known
as a boxplot (also called a box and whisker diagram). Figure 6.1 shows the basic
geometric figure used in a boxplot. Note that the bottom of the figure is the mini-
mum and the top is the maximum. The box area represents the region from the first to
the third quartile. (The width of the box has no significance.) The median is marked
with a horizontal line inside the box. A greater length between the top and bottom
“whiskers” indicates a greater spread from the maximum to the minimum, and a
Maximum
Third quartile
Median
First quartile
Minimum
400 CHAPTER 6
longer box indicates a greater spread from the first to third quartiles. The location
of the median within the box gives information about how the data between the first
quartile and the third quartile are distributed about the median.
City Highway
Model mileage (mpg) mileage (mpg)
Toyota Prius C 53 46
Toyota Prius Plug-In Hybrid 51 49
Toyota Prius 51 48
Lexus CT 200H 51 48
Honda Civic Hybrid 44 47
Volkswagen Jetta Hybrid 42 48
Honda Insight 41 44
Mitsubishi Mirage 37 44
Mercedes-Benz Smart
ForTwo Convertible/Coupe 34 38
Honda Civic Natural Gas 27 38
42 + 44
Median = = 43 mpg
2
The lower half of the list is 27, 34, 37, 41, 42, and the median of this half is 37.
Thus, the first quartile is 37 mpg. The upper half of the list is 44, 51, 51, 51, 53,
and the median of this half is 51. Thus, the third quartile is 51 mpg.
b. The corresponding boxplot appears in Figure 6.2. The vertical axis is the
mileage measured in miles per gallon.
c. Referring to the boxplot, we note that the maximum is not far above the
third quartile. This is because the top four cars (in terms of city mileage) are not
far apart. At the lower end, note that the minimum of 27 mpg is for the Honda
Civic using natural gas. The relatively large distance between the first quartile
and the minimum should be interpreted in light of this fact.
Crauder-4100237 QL November 10, 2014 11:34
6.1 Data summary and presentation: Boiling down the numbers 401
45
Honda Civic Hybrid (44)
Volkswagen Jetta Hybrid (42)
Honda Insight (41)
40
Mitsubishi Mirage (37)
35
Mercedes-Benz Smart ForTwo Convertible/Coupe (34)
30
Honda Civic Natural Gas (27)
25
20
Standard deviation
As we have noted, data sets can be clustered together (such as 10, 11, 12, 12, 13) or
spread out (such as 1, 10, 20, 100, 300). One measure of how much data are spread
out is the five-number summary. Another way to measure the spread of data is the
standard deviation. This number is commonly denoted by the lowercase Greek letter
σ (pronounced “sigma”).
For example, a data set of the yearly rainfall amounts in the Gobi Desert would
show values that deviate very little from the mean because there is not a significant
variation in rainfall from one year to the next. In contrast, a data set of prices for
new automobiles would have a wide range because high-end sports cars cost a great
deal more than the average family automobile. For this data set, we would have a
larger standard deviation. The standard deviation is used in many settings, and, like
the mean, it is sensitive to outliers.
The unit of measurement for the standard deviation is the same as the unit for the
original data. For example, if the data show car prices in dollars, then the standard
deviation is measured in dollars.
Deviation from the mean refers to the difference between data and the mean of
the data. For example, let’s say the mean price for a new automobile is $30,000.
If the price of a Mercedes is $48,000, we say that the price deviates by $18,000
from the mean because $48,000 − $30,000 = $18,000. If the price for a Hyundai
Elantra is $17,000, we say that the price deviates by −$13,000 from the mean be-
cause $17,000 − $30,000 = −$13,000.
The standard deviation is used in finance to measure the risk of an investment.
An investment with a high standard deviation for its returns is a volatile investment.
The formula for standard deviation looks intimidating. But, as we shall see, the
actual calculation is not very difficult. Suppose the data points are
x1 , x2 , x3 , . . . , xn
Crauder-4100237 QL November 10, 2014 11:34
402 CHAPTER 6
The template provided in the following Calculation Tip makes the process of calcu-
lating the standard deviation manageable.
1 There are actually two common definitions of the standard deviation of a data set. One is the population standard
deviation, denoted by σ and defined as we have it here. The other is the sample standard deviation, denoted by s and
defined in the same way except that the denominator n is replaced by n − 1. The sample standard deviation s is used
when the data consist of a sample from a larger population. The population standard deviation σ is used when the data
are not a sample but consist of the entire “population” of data. For large data sets, their ratio is very close to 1 (so they
are nearly the same). The reason statisticians have different ways of defining these is due to a technical point about
using a sample standard deviation as an “unbiased estimator” for the population standard deviation.
Crauder-4100237 QL November 10, 2014 11:34
6.1 Data summary and presentation: Boiling down the numbers 403
(earned run average—the lower the number, the better) histories are given in the
following table.
Calculate the mean and the standard deviation for Kershaw’s ERA history. It
turns out that the mean and standard deviation for Hernandez’s ERA history are
μ = 2.87 and σ = 0.43. What comparisons between Kershaw and Hernandez can
you make based on these numbers?
SOLUTION
The mean for Kershaw is
We make the calculation of the standard deviation using Calculation Tip 6.1.
The first column shows the data points. We find the second column by subtracting
the mean μ = 2.47 to find the deviations. We calculate the third column by squaring
the second column to find the squares of the deviations, which we round to three
decimal places.
We conclude that the mean and the standard deviation for Kershaw’s ERA history
are μ = 2.47 and σ = 0.39.
Because Kershaw’s mean is smaller than Hernandez’s mean of μ = 2.87, over this
period Kershaw had a better pitching record. Also, Kershaw’s ERA had a smaller
standard deviation than that of Hernandez (who had σ = 0.43), so Kershaw was
more consistent—his numbers are not spread as far from the mean.
In the next example, we use the concepts from this section to compare two data
sets.
Crauder-4100237 QL November 10, 2014 11:34
404 CHAPTER 6
What do these values for the mean and standard deviation tell us about free-
throws shot at home compared with free-throws shot away from home? Does com-
parison of the minimum and maximum of each of the data sets support your
conclusions?
David E. Klutho/Sports Illustrated/Getty Images
SOLUTION
The means for free-throw percentages are 75.51 at home and 74.88 away, so on
average the teams do somewhat better at home than on the road. (This is not the case
for every team.) The standard deviation of 3.17 percentage points at home, which
is larger than the standard deviation of 2.95 percentage points away from home,
means that the free-throw percentages at home vary from the mean more than the
free-throw percentages away.
Crauder-4100237 QL November 10, 2014 11:34
6.1 Data summary and presentation: Boiling down the numbers 405
μ = 74.88
μ – = 72.34 μ = 75.51 μ + = 78.68 μ – = 71.93 μ + = 77.83
66 68 70 72 74 76 78 80 66 68 70 72 74 76 78 80
Free-throw percentages at home Free-throw percentages away
FIGURE 6.3 Eastern Conference NBA FIGURE 6.4 Eastern Conference NBA
free-throw percentages at home. free-throw percentages away.
The difference between the maximum and minimum percentages shows the same
thing: The free-throw percentages at home range from 67.2% to 79.0%, and the
free-throw percentages away range from 66.9% to 77.7%. The plots of the data in
Figures 6.3 and 6.4 provide a visual verification that the data for home free-throws
are more broadly dispersed than the data for away free-throws.
Histograms
In addition to numerical measures of distributions of data, such as the five-number
summary and the standard deviation, it is often very useful to visualize a distribution
of data, as we did with the boxplot. In cases where we are dealing with frequencies,
that is, how many times an item occurs, bar graphs as described in Section 2.1 are
useful representations.
Consider again the data in Example 6.1 showing goals scored in the games played
by Chelsea FC between August 2013 and May 2014:
The two bar graphs in Figures 6.5 and 6.6 show two ways to represent the data
on frequencies of goals scored. The first graph uses the raw data. For the second
25 25
20 20
Number of games
Number of games
15 15
10 10
5 5
0 0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 to 7
Total goals scored Total goals scored
FIGURE 6.5 Goals scored, raw data. FIGURE 6.6 Goals scored, grouped data.
Crauder-4100237 QL November 10, 2014 11:34
406 CHAPTER 6
graph, because there are relatively few games with a high number of goals scored,
we group the highest score totals. In the category of 5 to 7 goals scored, there are
2 + 1 + 1 = 4 games altogether.
Bar graphs with grouped data are especially useful in visualizing distributions of
large data sets. These graphs are called histograms.
Often we use histograms to reduce data to manageable chunks. Suppose, for ex-
ample, we toss 1000 coins and write down the number of heads we got. We do this
1000-coin toss experiment a total of 1000 times. The accompanying table shows one
part of the results from doing these experiments using a computer simulation.
Number of heads 451 457 458 459 461 462 463 464 465 467
Number of 1000-coin tosses 2 2 1 3 3 2 1 3 1 1
The first entry shows that twice we got 451 heads, twice we got 457 heads, once
we got 458 heads, and so on. Note that the entire table would include all possibilities
from 0 through 1000. The raw data are a bit hard to digest because there are so
many data points. The five-number summary provides one way to analyze the data.
An alternative way to get a grasp of the data is to arrange them in groups—rather
than to look at all the individual data points—and then draw a histogram.
For example, it turns out that the number of 1000-coin tosses yielding fewer
than 470 heads is 23. Because 470 out of 1000 is 47%, this means that 23 tosses
yielded less than 47% heads. We find the accompanying table by dividing the data
into groups this way. (Here, “47% to 48%” means 470 through 479 heads, “48%
to 49%” means 480 through 489 heads, and so on.)
Percent heads Less than 47% 47% to 48% 48% to 49% 49% to 50%
Number of 1000-coin tosses 23 75 140 234
Percent heads 50% to 51% 51% to 52% 52% to 53% At least 53%
Number of 1000-coin tosses 250 157 94 27
Figure 6.7 shows a histogram for this grouping of the data. This visual presen-
tation makes the situation even clearer. We can clearly see that the vast majority of
© Eyebyte/Alamy
300
250
1000-coin tosses
Number of
200
150
100
50
0
Less 47% 48% 49% 50% 51% 52% At
than to to to to to to least
47% 48% 49% 50% 51% 52% 53% 53%
Percent heads
6.1 Data summary and presentation: Boiling down the numbers 407
the 1000-coin tosses were between 47% and 53% heads. Note also how the number
of heads increases near the 50% mark. This is what we would expect from tosses of
fair coins.
408 CHAPTER 6
150
$80,000 $120,000 $120,000 $120,000 $122,000
130
$135,000 $135,000 $150,000 $160,000
110
Find the mean, median, and mode of these home prices (to the 90
nearest thousand). 70
5. Sales. An auto dealer’s sales numbers are shown in the table 50
below. Find for each month the mean, median, and mode prices
of the cars she sold. Round your answers to the nearest dollar.
FIGURE 6.8 Boxplot.
Number sold 9–10. Math scores. Use the following information to answer
exercises 9 through 10. At one point, the San Diego Achieve-
Price May June July
ment Forum reported that during the four prior years, the per-
$20,000 22 25 24 centage of students scoring in the lowest quartile in mathemat-
$15,000 49 24 24 ics had declined by 11 percentile points, while the percentage
$12,500 25 49 49 of students in the top quartile had risen from 24 percent to 31
percent. (Presumably, this referred to quartiles in a larger group
of test-takers and not just the San Diego schools.) This report
6. Land areas
seemed to suggest that things were going well. The next two ex-
a. Calculate the five-number summary of the land areas of
ercises refer to this suggestion. In these exercises, we write T1
the states in the U.S. Midwest.
for the percentage of students in the lowest quartile (i.e., in the
first quartile), T2 for the percentage of students in the second
Area Area quartile, T3 for the percentage of students in the third quartile,
State (sq. miles) State (sq. miles) and T4 for the percentage of students in the top quartile (i.e.,
in the fourth quartile).
Illinois 55,584 Missouri 68,886
Indiana 35,867 Nebraska 76,872 9. Suppose that the original numbers were T1 = 25%, T2 =
Iowa 55,869 North Dakota 68,976 25%, T3 = 26%, and T4 = 24%, and that four years later the
Kansas 81,815 Ohio 40,948 numbers were T1 = 14%, T2 = 5%, T3 = 50%, and T4 =
Michigan 56,804 South Dakota 75,885 31%. Is this an improvement? Is there a possible downside to
Minnesota 79,610 Wisconsin 54,310 the new numbers in comparison to the original ones?
10. Suppose that the original numbers were T1 = 25%, T2 =
b. Explain what the five-number summary in part a tells us 25%, T3 = 26%, and T4 = 24%, and that four years later the
about the land areas of the states in the Midwest. numbers were T1 = 14%, T2 = 50%, T3 = 5%, and T4 =
31%. Is this an improvement? Is there a possible downside to
c. Calculate the five-number summary of the land areas of
the new numbers in comparison to the original ones?
the states in the U.S. Northeast.
11. Mathematics faculty salaries. The first three columns of
the following table are salary figures taken from an Ameri-
Area Area can Mathematical Society (AMS) survey of mathematics fac-
State (sq. miles) State (sq. miles) ulty from Ph.D.-granting public institutions with large PhD
Connecticut 4845 New York 47,214 programs. The last two columns in the table (marked with an
Maine 30,862 Pennsylvania 44,817 asterisk) were not reported by the AMS, so for the purposes of
Massachusetts 7840 Rhode Island 1045 this exercise, we have just made up the (plausible) numbers in
New Hampshire 8968 Vermont 9250 these columns.
New Jersey 7417
Academic First Third
rank quartile Median quartile Minimum* Maximum*
d. Explain what the five-number summary in part c tells us
about the land areas of the states in the Northeast. Assistant $80,400 $83,200 $86,800 $55,000 $115,000
professor
e. Contrast the results from parts b and d. Associate $82,200 $88,200 $94,400 $55,000 $155,000
7. Boxplot. A boxplot for a data set is shown in Figure 6.8. Es- professor
timate the minimum, first quartile, median, third quartile, and Professor $107,100 $126,900 $153,600 $75,000 $225,000
maximum of the data set.
8. Highway mileage. Find the five-number summary for the a. Make a boxplot for each of the three categories.
data on highway mileage in Example 6.4, and draw the cor- b. Use the boxplots you made in part a to compare the dis-
responding boxplot. persal of salaries among the three categories.
Crauder-4100237 QL November 10, 2014 11:34
6.1 Data summary and presentation: Boiling down the numbers 409
12. Employee absence. A certain company recorded the num- 21. An application. We have 1200 lightbulbs in our building.
ber of employee absences each week over a period of 10 weeks. Over a 10-month period, we record the number of bulbs that
The result is the data list 3, 5, 1, 2, 2, 4, 7, 4, 5, 5. Find the burn out each month. The result is the data list 23, 25, 21, 33,
mean and standard deviation of the number of absences per 17, 39, 26, 24, 31, 22.
week. Round the standard deviation to two decimal places. a. What is the average number of bulbs that burn out each
13. Calculation. Find the mean and standard deviation of the month?
following list of quiz scores: 75, 88, 65, 90. Round the standard b. What is the standard deviation of these data? Round the
deviation to two decimal places. standard deviation to one decimal place.
14. Free-throws. In the situation of Example 6.6, verify that c. Use Chebyshev’s inequality and your answer to parts a
the means for free-throws at home and away are 75.51% and and b to estimate how many replacement bulbs you should
74.88%, respectively. Verify that the standard deviations for keep on hand so that for at least 75% of the months you
free-throws at home and away are 3.17 percentage points and don’t have to acquire additional replacement bulbs.
2.95 percentage points, respectively. 22. Histogram. Consider an experiment in which we toss
15. No calculation. Suppose every student in your class gets 20 coins. We perform this experiment 100 times and record
a score of 85% on an exam. Find the mean and standard de- the number of experiments in which we got a given number of
viation of those exam scores by inspection without doing any heads. The results are given in the following table:
calculations.
16. A challenge problem. A class has an equal number of boys
Number of heads 5 6 7 8 9 10 11 12 13 14 15
and girls. The boys all got 78% on a test and the girls all got
Number of 3 5 8 11 16 18 16 13 6 2 2
84%. What are the mean and standard deviation of the test experiments
scores for the entire class?
17. Another challenge problem. A class has 9 boys and 21 girls.
The class as a whole has a GPA (grade point average) of 2.96, Arrange the data in the following groups: less than 35% heads,
and the boys have a GPA of 2.40. What is the GPA of the girls? at least 35% but less than 45% heads, at least 45% but less
than 55% heads, at least 55% but less than 65% heads, at
18. Tossing a die least 65% heads. (Here, of course, 35% refers to 35% of the
a. Perform the following experiment: Roll a fair die six 20 coins, etc.) Make a histogram showing the grouped data.
times, and record a 1 when the die comes up 6 and a 0 oth-
erwise. Fill in the blanks in the following table with your Exercises 23 and 24 are suitable for group work.
results: 23. Your class. Collect the heights of each of your classmates.
Roll 1 2 3 4 5 6 Find the mean and standard deviation of these data and make
Data a five-number summary and box plot of it.
24. Experiment. Do the following experiment: Toss 20 coins.
b. Calculate the mean and standard deviation of the data Perform this experiment 25 times and record in the table below
list you made in part a. Leave the mean as a fraction, and the number of experiments in which you got a given number
round the standard deviation in decimal form to two deci- of heads.
mal places.
19–21. Spread. We have seen that the standard deviation
σ measures the spread of a data set about the mean μ. Number of heads 0 1 2 3 4 5 6 7 8 9 10
Chebyshev’s inequality gives an estimate of how well the stan- Number of
experiments
dard deviation measures that spread. One consequence of this
inequality is that for every data set at least 75% of the data Number of heads 11 12 13 14 15 16 17 18 19 20
points lie within two standard deviations of the mean, that Number of
experiments
is, between μ − 2σ and μ + 2σ (inclusive). For example, if
μ = 20 and σ = 5, then at least 75% of the data are at least
20 − 2 × 5 = 10 and at most 20 + 2 × 5 = 30. Exercises 19 Arrange the data in the following groups: less than 35%
through 21 illustrate this result.2 heads, at least 35% but less than 45% heads, at least 45% but
19. Consider the following data: 5, 10, 10, 10, 10, 10, 10, 15. less than 55% heads, at least 55% but less than 65% heads, at
Find the mean and the standard deviation. How many data least 65% heads. (Here, of course, 35% refers to 35% of the
points does the Chebyshev inequality promise will lie within 20 coins, etc.) Make a histogram showing the grouped data.
two standard deviations of the mean? 25. History. The term standard deviation was coined by Karl
20. Our statement of this result says “at least 75%,” not “ex- Pearson (1857–1936), who was the founder of the world’s first
actly 75%.” What percent of the data points in Exercise 19 university statistics department. Write a brief biography of this
actually lies within two standard deviations of the mean? colorful and controversial figure.
2 In the next section, we will see similar, but more precise, statements for special data sets, namely those that are normally distributed.
Crauder-4100237 QL November 10, 2014 11:34
410 CHAPTER 6
26. History. Pafnuty Lvovich Chebyshev (1821–1894) and record the percentage of heads for each experiment. Use
was a Russian mathematician. Write a brief biography, a computer simulation to carry out this procedure. (This was
and investigate Chebyshev’s inequality, sometimes called done to get the data on coin tosses given on page 406 in connec-
Chebyshev’s theorem, which gives a more precise statement tion with the topic of histograms.) Report your results as a his-
of how well the standard deviation measures the spread of a togram. How does your histogram compare with that shown
data set. (See Exercises 19 through 21.) in Figure 6.7?
The following exercises are designed to be solved using tech- 29. Five-number summary. This is a continuation of Exer-
nology such as calculators or computer spreadsheets. cise 28. Consider the data on coin tosses you found in Exercise
27. Standard deviations in Excel© . In this section, we discussed 28. Use technology to find the five-number summary for the
the fact that there are two versions of the standard deviation, percentage of heads obtained for each experiment.
one referred to as the “sample standard deviation” and the 30. Another computer simulation. Consider an experiment in
other as the “population standard deviation.” In the Excel© which we roll 1000 standard dice. We perform the experi-
spreadsheet, locate the formulas for these and use them to find ment 1000 times and record for each experiment the percent-
both versions of the standard deviations for this list of home age of dice coming up 6. Use a computer simulation to carry
prices: $80,000 $120,000 $120,000 $120,000 $122,000 out this procedure. Report your results as a histogram. How
$135,000 $135,000 $150,000 $160,000. does the shape of your histogram compare with that shown in
28. Computer simulation. Consider an experiment in which Figure 6.7?
we toss 1000 coins. We perform the experiment 1000 times
TAKE AWAY
FROM THIS
SECTION
Understand why IN THE NEWS 6.3 Search
the normal Home TV & Video U. S. World Politics Entertainment Tech Travel
distribution is so
important. Tumbling Dice & Birthdays
Michelle Paret Product Marketing Manager, Minitab Inc.
Eston Martz Senior Creative Services Specialist, Minitab Inc.
M ark Twain famously quipped that there were three ways to avoid telling the
truth: lies, damned lies, and statistics. The joke works because statistics fre-
quently seems like a black box—it can be difficult to understand how statistical
theorems make it possible to draw conclusions from data that, on their face, defy
easy analysis. But because data analysis plays a critical role in everything from jet
engine reliability to determining the shows we see on television, it’s important to
acquire at least a basic understanding of statistics. And one of the most impor-
tant concepts to understand is the central limit theorem. In this article, we will
explain the central limit theorem and show how to demonstrate it using common
examples, including the roll of a die and the birthdays of Major League Baseball
players. [The Central Limit Theorem says that] if we repeatedly take independent
random samples of size n from any population, then when n is large, the dis-
tribution of the sample means will approach a normal distribution. How large is
large enough? Generally speaking, a sample size of 30 or more is considered to
be large enough for the central limit theorem to take effect.
Dice are ideal for illustrating the central limit theorem. If you roll a six-sided
die, the probability of rolling a one is 1/6th, a two is 1/6th, a three is also 1/6th,
etc. The probability of the die landing on any one side is equal to the probability
of landing on any of the other five sides. To get an accurate representation of the
population distribution, let’s roll the die 500 times. When we use a histogram to
graph the data, we see that—as expected—the distribution looks fairly flat. It’s
definitely not a normal distribution (Figure 6.9).
Let’s take more samples and see what happens to the histogram of the aver-
ages of those samples. This time we will roll the die twice, and repeat this process
500 times. We can then create a histogram of these averages to view the shape of
their distribution (Figure 6.10). Although the blue normal curve does not accu-
rately represent the histogram, the profile of the bars is looking more bell-shaped.
Now let’s roll the die five times and compute the average of the five rolls, again
repeated 500 times. Then, let’s repeat the process rolling the die 10 times, then
30 times.
The histograms for each set of averages (Figure 6.11) show that as the sam-
ple size, or number of rolls, increases, the distribution of the averages comes
closer to resembling a normal distribution.
Now let’s demonstrate the central limit theorem using birthdays. You’ll recall
that the sides of dice have an equal probability. Contrary to popular belief, there is
not necessarily an equal chance of being born on a Sunday instead of a Monday or
any other day of the week. Currently, the most popular day for babies to be born
in the United States is Wednesday—there are 15.4% more births on Wednesday
Crauder-4100237 QL November 10, 2014 11:34
412 CHAPTER 6
80 60 0
Frequency
Average of 10 rolls
50 80
60
40
40
40 30
0
20 Average of 5 rolls
20
10 50
0 0 25
1 2 3 4 5 6 0.8 1.6 2.4 3.2 4.0 4.8 5.6 6.4
0
Value on die Average of 2 rolls 1.6 2.4 3.2 4.0 4.8 5.6
FIGURE 6.9 Because the odds of FIGURE 6.10 The distribution of 500 FIGURE 6.11 As the number of rolls of the
landing on all sides of a six- sided averages for two rolls of a die begins to die increases, the distribution of averages
die are equal, the distribution of resemble the familiar bell shape of a approaches a normal distribution.
500 die rolls is flat. normal distribution.
Histograms of averages
Average of 30 birth days
16
8
0
20
10
0
Average of 5 birth days
20
10
0
The distribution of a set of data measures the frequency with which each data
point occurs. In the preceding section, we introduced the histogram as a way of
visualizing how data are distributed in terms of frequency. Note that each of the
graphs in Figure 6.9 through Figure 6.12 is a histogram.
In the News 6.3 refers to a very important distribution, the normal distribution.
Results of many games of chance, as illustrated in Figure 6.11, and most common
birthday, as illustrated in Figure 6.12, provide examples of (approximately) normally
distributed data, but, as the article suggests, many other common data sets are nor-
mally distributed as well. In this section, we see the importance of the normal distri-
bution for both descriptive and inferential statistics, as defined at the beginning of this
chapter: The normal distribution provides a very efficient way of organizing many
types of data, and the Central Limit Theorem concerning the normal distribution is
a powerful tool for drawing conclusions from data based on sampling.
14
12
Adult males (%)
10
0
50 55 60 65 70 75 80 85
Height (inches)
Following are some important features of normally distributed data. These fea-
tures are expressed in terms of the mean and median, which we studied in the pre-
ceding section.
The mean and median are the same: For normally distributed data, the mean and
median are the same. Figure 6.13 indicates that the median height of adult males is
69.1 inches, or about 5 feet 9 inches. That means about half of males are taller than
69.1 inches and about half are shorter. Because the mean and median are the same,
the average height of adult males is also 69.1 inches.
Most data are clustered about the mean: This is reflected in the fact that the vast
majority of adult males are within a few inches of the mean. In fact, about 95%
of adult males are within 5 inches of the mean—between 5 feet 4 inches and 6 feet
2 inches. This is shown in Figure 6.14. Only about 5% are taller or shorter than this.
This is illustrated in Figure 6.15.
The bell curve is symmetric about the mean: This is reflected in the fact that the curve
to the left of the mean is a mirror image of the curve to the right of the mean. In terms
of heights, it means that there are about the same number of men 2 inches taller than
the mean as there are men 2 inches shorter than the mean, and there are about the
Crauder-4100237 QL November 10, 2014 11:34
414 CHAPTER 6
FIGURE 6.14 95% of adult males FIGURE 6.15 Relatively few men are FIGURE 6.16 The bell curve is symmetric
are within 5 inches of the median. very tall or very short. about the mean.
same number of men 6 feet 4 inches in height (7 inches above the mean) as men 5 feet
2 inches in height (7 inches below the mean). This is illustrated in Figure 6.16.
3.0 14
12
2.0
8
1.5 6
1.0 4
2
0.5
0
<$10
10–20
20–30
30–40
40–50
50–60
60–70
70–80
80–90
90–100
100–150
150–200
0
60 70 80 90 100 110 120 130 140
IQ Income (thousands of dollars)
FIGURE 6.17 Percentage of population with FIGURE 6.18 Percentage of families with given
given IQ score. income.
35 9
Number of births (thousands)
8
30
7
Population (%)
25 6
5
20
4
15 3
10 2
1
5
0
0–5
5–9
10–14
15–19
20–24
25–29
30–34
35–39
40–44
45–49
50–54
55–59
60–64
65–69
70–74
75–79
80–84
85–89
90–94
95
0
0 1 2 3 4 5 6
Birth weight (kg) Age (years)
FIGURE 6.19 Number of births of a given weight in FIGURE 6.20 Percentage of Americans of a given
Norway. age.
416 CHAPTER 6
Mean 500
0.008 0.008
0.006 0.006
Standard
deviation
Mean 500 of 50
0.004 0.004
Standard
deviation
of 100
0.002 0.002
0 0
200 300 400 500 600 700 800 200 300 400 500 600 700 800
FIGURE 6.21 Normal curve with mean 500 and FIGURE 6.22 Normal curve with mean 500 and
standard deviation 100. standard deviation 50.
The following rule of thumb gives more information on the role of the standard
deviation in the normal distribution.
1 standard 1 standard
deviation deviation
2 standard 2 standard
deviations deviations
68% of data 95% of data 99.7% of data
3 standard 3 standard
deviations deviations
FIGURE 6.23 68% of data lie within FIGURE 6.24 95% of data lie within FIGURE 6.25 99.7% of the data lie within
one standard deviation of the mean. two standard deviations of the mean. three standard deviations of the mean.
20% 20%
15 15
Students (%)
10 Students (%) 10
5 5
0 0
200 300 400 500 600 700 800 200 300 400 500 600 700 800
SAT score SAT score
FIGURE 6.26 SAT histogram. FIGURE 6.27 SAT bell curve plot.
In Figure 6.26, we see a histogram of these data. The bell curve associated with
the data is shown in Figure 6.27. These figures suggest that SAT verbal scores are
normally distributed. This is indeed the case, and according to Table 6.1, the mean
verbal score on the SAT is 507, and the standard deviation is 111. Applying the
68-95-99.7% rule to SAT verbal scores gives the following information:
• 68% of SAT scores are within one standard deviation of the mean. This means
that 68% of scores are between 507 − 111 = 396 and 507 + 111 = 618. Because of
the symmetry of the bell curve, about half of these, 34% of the total, are above the
mean (between 507 and 618), and about 34% of the scores are below the mean
(between 396 and 507).
• 95% of SAT scores are within two standard deviations of the mean. This means
that 95% of scores are between 507 − 2 × 111 = 285 and 507 + 2 × 111 = 729.
• 99.7% of SAT scores are within three standard deviations of the mean. This
means that 99.7% of scores are between 507 − 3 × 111 = 174 and
507 + 3 × 111 = 840.3
3 Because the number of scores above 800 is so small, the College Board lumps together all scores 800 or higher and
reports them as 800. Similarly, the lowest reported score is 200.
Crauder-4100237 QL November 10, 2014 11:34
418 CHAPTER 6
The next example illustrates how to apply the 68-95-99.7% rule to the problem
of counting the number of data points within a given range.
4% 4%
Mean 200 g
3 3
Apples (%)
Apples (%)
2 2
2 standard 2 standard
deviations deviations
1 1
0 0
150 175 200 225 250 150 175 200 225 250
Weight (grams) Weight (grams)
FIGURE 6.28 Apples with mean 200 grams FIGURE 6.29 Apples between 176 and 224
and standard deviation 12 grams. grams.
Crauder-4100237 QL November 10, 2014 11:34
z-scores
The standard deviation is such a useful tool when dealing with normal distributions
that it is often used as a unit of measurement. Instead of saying that a SAT score of
618 is 111 points above the mean of 507, often one says that the score is one standard
deviation above the mean. A score of 452 is 55 points or approximately one-half of
a standard deviation below the mean. When used as a unit of measurement in this
way, the standard deviation is often referred to as a z-score.
420 CHAPTER 6
The standard deviation is about 530 grams. Calculate the z-score for a newborn
weighing 3700 grams. (That is just over 8 pounds 2 ounces.)
SOLUTION
A 3700-gram newborn is 3700 − 3332 = 368 grams above the mean weight of
3332 grams. We divide by the number of grams in one standard deviation to find the
z-score:
368
z-score for 3700 grams =
530
or about 0.7.
Percentile scores
Recall that in Section 6.1 we discussed quartiles. There is a more general term, per-
centile, that tells how a data point is positioned in relation to other data points in a
normal distribution.
Median 40
30
Students (%)
Score of 660
20
FIGURE 6.30 The percentile for the FIGURE 6.31 91.6% of SAT verbal scores
median is 50%. were 660 or lower.
Crauder-4100237 QL November 10, 2014 11:34
Table 6.2 gives the information needed to calculate the percentile from the
z-score.
For example, a z-score of 0.5 corresponds to a percentile score of 69.15. A per-
centile score of 93.32 corresponds to a z-score of 1.5.
Let’s see how this table helps in understanding SAT verbal scores. Suppose you
scored 650 on the verbal section of the SAT. How does your test score compare with
scores of other students? We calculated earlier that the z-score for a SAT score of 650
is 1.3. Table 6.2 shows a percentile of 90.32 for a z-score of 1.3. We conclude that
about 90.3% of all SAT verbal scores were 650 or less.
We also found that the z-score for a SAT score of 320 is −1.7. The table shows a
percentile of 4.46 for a z-score of −1.7. Thus, about 4.5% of all SAT verbal scores
were 320 or less. Expressed differently, this says that about 95.5% of all SAT verbal
scores were higher than 320.
422 CHAPTER 6
SOLUTION
Ten days is 2 days above the mean of 8 days. This gives a z-score of 2/0.9 or about
2.2. Table 6.2 gives a percentile of about 98.6% for this z-score. This means that
about 98.6% of patients will recover in 10 days or less. Thus, only about 100% −
98.6% = 1.4% will be ill for more than 10 days.
304.4
z-score =
530
or about 0.6. Consulting Table 6.2, we find that this represents a percentile of
about 72.6%. This means that about 72.6% of newborns weigh 8 pounds or
less. So, about 100% − 72.6% = 27.4% of newborns weigh more than 8
pounds.
b. Again, we calculate the z-score: 2500 grams is 3332 − 2500 = 832 grams
below the mean of 3332 grams. This is 832/530 or about 1.6 standard
deviations below the mean. That is a z-score of −1.6. Table 6.2 shows a
percentile of about 5.5% for a z-score of −1.6. Hence, about 5.5% of newborns
are classified as low-birthweight babies.
To see an application of the Central Limit Theorem, suppose that 60% of the
registered voters in the state of Oklahoma in October 2012 intended to vote for Mitt
Romney for president. Suppose we took a poll of 100 (randomly selected) registered
Oklahoma voters and calculated the percentage of those polled who intended to vote
for Romney. The Central Limit Theorem tells us that the results of such a poll are
(approximately) normally distributed with mean p = 60% because that is the mean
for this population.4 The sample size is n = 100, so the standard deviation is
p(100 − p)
σ =
n
60(100 − 60)
=
100
4 Strictly speaking, we should say that if many such polls, all of the same size, were taken from the same population,
then the results would be approximately normally distributed.
Crauder-4100237 QL November 10, 2014 11:34
424 CHAPTER 6
SOLUTION
Benny Evans
a. The mean is the percentage of untreated patients who will improve within a
week, which is p = 30%. The sample size is n = 50. This gives a standard
deviation of
p(100 − p)
σ =
n
30 × 70
=
50
SOLUTION
a. The mean is the percentage of people from the general population who suffer
from the allergy. Therefore, the mean is p = 20%.
b. For a sample size of 100,000, we find
p(100 − p)
σ =
n
20 × 80
=
100,000
or about 0.13 percentage point.
c. Our sample of 21% is one percentage point larger than the mean of 20%.
That gives
1
z-score =
0.13
or about 7.7. This score is far larger than any z-score in Table 6.2. The chance
of getting a random sample 7.7 standard deviations above the mean is virtually
zero. There is almost no chance that in a randomly chosen sample of this size, at
least 21% will suffer from this allergy. Thus, this is a truly anomalous sample:
This town is not representative of the total population of Americans. Its allergy
rate is highly unusual.
To summarize: It may seem counterintuitive, but even though 21% seems
to be only a little larger than 20%, a percentage that high is, in fact, a very
improbable result for a truly random sample of 100,000 people. Note that it is
the large size of the sample that makes the standard deviation small and thus
makes such a small percentage variation unlikely.
5 http://www.cdc.gov/nceh/clusters/fallon/.
Crauder-4100237 QL November 10, 2014 11:34
426 CHAPTER 6
r Fallon
r s o n Rive
Old River
Reservoir
Las Vegas
Ca
50
Fallon
95
Fallon
Naval
Air
Station
N
Examples like this illustrate how vital mathematics and statistics are to our well-
being.
Try It Yourself 6.13: Using the Central Limit Theorem: Education majors
a. The mean is 20%. The standard deviation is 4.6 percentage points.
b. 1.4%.
Number of occurrences
Mean = 40.2° F
16
normal, explain why. Minimum = 24° F
14 Maximum = 55° F
12 Standard
deviation = 6.8° F
10
8
6
4
2
0
0)
0)
5)
0)
5)
5
25
5)
>5
5,3
5,4
0,4
5,5
0,5
0,3
<=
(2
(3
(4
(4
(5
(3
Annual average temperature
428 CHAPTER 6
6. Use the 68-95-99.7% rule to determine what percentage of period, birthweights (in grams) of newborns in Norway were
the test group had a diastolic pressure between 72 millimeters distributed normally (see Figure 6.37), with a mean of 3668
and 92 millimeters of mercury. grams and standard deviation of 511 grams. Exercises 19
7. Use your answer to Exercise 6 to determine how many men through 22 refer to these data.
in the test group had a diastolic pressure between 72 millime-
ters and 92 millimeters of mercury.
Birthweight parameters
8. Calculating the z-score. It is found that the number of raisins
in a box of a popular cereal is normally distributed, with a 30,000 Predominant
mean of 133 raisins per box and standard deviation of 10 Residual
25,000 Observed
raisins. My cereal box has 157 raisins. What is the z-score for
this box of cereal? Round your answer to one decimal place. 20,000
Frequency
Residual = 28% Predominant
9. More on z-scores. The number of nuts in a can of mixed nuts 15,000 mean = 3668 g
is found to be normally distributed, with a mean of 500 nuts 10,000 SD = 511 g
and standard deviation of 20 nuts. My can of mixed nuts has
5,000
only 475 nuts. What is the z-score for this can of nuts? Round
0
your answer to one decimal place. 1,000
10–13. Standardized exam scores. Exercises 10 through 13 500
refer to scores on standardized exams with results that are nor- 0
mally distributed. Round your answers to one decimal place. 150 650 1150 1650 2150 2650 3150 3650 4150 4650 5150 5650
10. Suppose you have a score that puts you 1.5 standard devi- Weight (gm)
Optimum Truncation Point = 2600 gm
ations below the mean. What is your percentile score?
11. Suppose you have a score that puts you in the 60th per-
centile. What is your z-score? FIGURE 6.37 Birthweights in Norway.
12. Suppose you have a score of 360 on the verbal section of
the SAT Reasoning Test. What is your z-score? (See Table 6.1 19. What is the z-score for a newborn weighing 4000 grams?
on page 417.) What is your percentile score? Round your answer to one decimal place.
13. Suppose you have a score of 720 on the verbal section of 20. What is the z-score for a newborn weighing 3000 grams?
the SAT Reasoning Test. What is your z-score? (See Table 6.1 Round your answer to one decimal place.
on page 417.) What is your percentile score?
21. Newborns weighing less than 2500 grams are classified as
14–15. IQ. A person’s IQ (Intelligence Quotient) is supposed having a low birthweight. What percentage of newborns suf-
to be a measure of his or her “intelligence.” The IQ test scores fered from low birthweight? Round your answer as a percent-
are normally distributed and are scaled to a mean of 100 and a age to the nearest whole number.
standard deviation of 15. This information is used in Exercises
22. Would you expect the distribution to remain normal if
14 and 15.
birthweights were reported in pounds?
14. If your IQ is 115, what is your percentile score? Round
23–24. Campinas, Brazil. Use the following information for
your answer to one decimal place.
Exercises 23 and 24. The average maximum monthly tempera-
15. It is often said that a genius is someone with an IQ of 140 ture in Campinas, Brazil, is 29.9 degrees Celsius. The standard
or more. What is the z-score for a 140 IQ? (Round your an- deviation in maximum monthly temperature is 2.31 degrees.
swer to one decimal place.) What percentage of the population Assume that maximum monthly temperatures in Campinas are
scores as a genius? Round your answer as a percentage to one normally distributed.
decimal place. Note: The actual number is 0.4%, but you may
23. Use Rule of Thumb 6.1 to fill in the following two blanks:
get a slightly different value due to rounding.
“95% of the time the maximum monthly temperature is be-
16–18. SAT scores. In Exercises 16 through 18, we suppose tween and .” Round your answers to one
that you had a score of 670 on the verbal section of the SAT decimal place.
Reasoning Test (see Table 6.1 on page 417). Round your an-
24. What percentage of months would have a maximum tem-
swers to one decimal place. Exercises 16 through 18 should be
perature of 35 degrees or higher? Round your answer as a per-
done in that order.
centage to one decimal place.
16. What is the z-score for this SAT score?
25–28. Male and female height in America. The heights of
17. What is your percentile score? adult men in America are normally distributed, with a mean of
18. Explain in practical terms the meaning of your percentile 69.1 inches and standard deviation of 2.65 inches. Exercises 25
score. through 28 use this information. Round your answers as per-
centages to the nearest whole number.
19–22. Birthweights in Norway. The National Institute of
Environmental Health Sciences reports that over a six-year 25. Tall. What percentage of adult males in America is over 6
feet tall?
Crauder-4100237 QL November 19, 2014 12:38
26. Short. What percentage of adult males in America is under b. In what percentage of such groups (of 400) would 275
5 feet 7 inches tall? or fewer improve? Round your answer to one decimal place
27. Heights of women. The heights of American women be- as a percentage.
tween the ages of 18 and 24 are approximately normally dis- 32–35. Lots of coins. Suppose we toss 100 fair coins and
tributed. The mean is 64.1 inches, and the standard deviation record the percentage of heads. Exercises 32 through 35 refer
is 2.5 inches. What percentage of such women is over 5 feet 9 to this situation and should be done in order.
inches tall? 32. According to the Central Limit Theorem, the percentages
28. Dancing. It is sometimes said that women don’t like to of heads resulting from such experiments are approximately
dance with shorter men. To get a dancing partner, a 6-foot- normally distributed. Find the mean and standard deviation of
tall woman draws from a hat the name of an adult American this normal distribution.
male. What is the probability that he will be as tall as she? 33. What is the z-score for 55 heads? How about 60 heads?
Give your answer as a percentage rounded to the nearest whole
number. 34. What percentage of tosses of 100 fair coins would show
more than 55 heads? Round your answer to one decimal place
29. Curving grades. When exam scores are low, students of- as a percentage.
ten ask the teacher whether he or she is going to “curve” the
grades. The hope is that by curving a low score on the exam, the 35. What percentage of tosses of 100 fair coins would show
students will wind up getting a higher letter grade than might more than 60 heads? Round your answer to one decimal place
otherwise be expected. The term curving grades, or grading on as a percentage.
a curve, comes from the bell curve of the normal distribution. 36–37. Contracting diseases. It is known that under ordinary
If we assume that scores for a large number of students are dis- conditions, 15% of people will not contract a certain disease.
tributed normally (as with SAT scores) and we also assume that Consider the situation in which test groups of 500 were selected
the class average should be a “C,” then a teacher might award and the percentages of those who did not contract the disease
grades as listed in Table 6.3. were recorded. Exercises 36 through 37 use this context.
36. According to the Central Limit Theorem, percentages of
groups of 500 who don’t contract the disease are approxi-
TABLE 6.3 Curving Grades mately normally distributed. Find the mean and standard devi-
ation of this normal distribution. Round the standard deviation
A 1.5 standard deviations above the mean or higher to one decimal place.
B 0.5 to 1.5 standard deviations above the mean
37. For what percentage of such groups (of 500) will 80 or
C within 0.5 standard deviation of the mean
fewer not contract the disease? Round your answer to one dec-
D 0.5 to 1.5 standard deviations below the mean
imal place as a percentage.
F 1.5 standard deviations below the mean or lower
38. Marbles in a jar. A large jar is filled with marbles. Of these
marbles, 25% are red and the rest are blue. We randomly draw
Suppose a teacher curved grades using the bell curve as in Ta- 100 marbles.
ble 6.3 and that the grades were indeed normally distributed. a. According to the Central Limit theorem, the percentages
What percent of students would get a grade of “A”? What per- of red marbles in such samples are approximately normally
cent of students would get a grade of “B”? Round each answer distributed. Find the mean and standard deviation of this
as a percentage to one decimal place. Suggestion: To find the normal distribution. Round the standard deviation to one
percentage of students getting a grade of “B,” subtract the per- decimal place.
centage of students 0.5 standard deviation or less above the b. Our friend pulls 100 marbles from the jar, supposedly at
mean from the percentage of students 1.5 standard deviations random, and we find that 20 of them are red. Do we have
or less above the mean. reason to suspect our friend of cheating? Explain why you
30. More on curving grades. Suppose an exam had an average answered as you did.
(mean) score of 55% and a standard deviation of 15%. If the
39. Voting for president. In a certain presidential election,
teacher curved grades using the bell curve as in Table 6.3 from
38% of voting-age Americans actually voted. In our town of
Exercise 29, what score would be necessary to receive an “A”?
200,000, 40% of voting-age citizens voted. What should we
How about a “B”? A “C”? Round each answer as a percentage
conclude regarding our town?
to one decimal place.
40. Obese Americans. About 35% of Americans over age 20
31. A medical test. It is known that in the absence of treat-
are obese.
ment, 70% of the patients with a certain illness will improve.
The Central Limit Theorem tells us that the percentages of pa- a. If we take random samples of 500 Americans over age
tients in groups of 400 that improve in the absence of treatment 20 and record the percent of people in the sample who are
are approximately normally distributed. obese, we expect to get data that are approximately nor-
a. Find the mean and standard deviation of the normal dis- mally distributed. Find the mean and standard deviation of
tribution given by the Central Limit Theorem. Round the this normal distribution. Round the standard deviation to
standard deviation to one decimal place. one decimal place.
Crauder-4100237 QL November 10, 2014 11:34
430 CHAPTER 6
b. Would it be unusual to find a random sample of 500 Does it appear that these data are normally distributed? Ex-
Americans over 20 in which at least 38% were obese? plain why you gave the answer you did.
41. Allergies. About 20% of a certain population suffer from 45. History. Write a paper on the history of the bell curve
some form of allergies. and the normal distribution. Some names to consider are de
Moivre, Gauss, Jouffret, and Laplace.
a. The Central Limit Theorem tells us that percentages of
allergy sufferers in random samples of 1000 people are 46. IQ and the bell curve. There are controversies associated
approximately normally distributed. Find the mean and with the connection between the bell curve and the measure-
standard deviation of this normal distribution. Round the ment of intelligence. The book The Bell Curve: Intelligence and
standard deviation to one decimal place. Class Structure in American Life caused quite a stir when it was
published.6 Summarize and discuss these controversies.
b. How unusual would it be to find a random sample of
1000 people in which at least 25% suffered from allergies? 47. History. Proofs of the Central Limit Theorem were given
early in the twentieth century. The ideas involved are much
older because certain other distributions have been studied for
42. Diabetes. About 6.9% of Americans suffer from diabetes. a long time. Summarize the results of de Moivre and Laplace
This exercise examines the percentage of people in samples of about the binomial distribution and its relationship to the nor-
2000 who have diabetes. mal distribution.
a. The Central Limit Theorem tells us that these per- 48. History. Write a paper about the men who are credited with
centages are approximately normally distributed. Find the proving the Central Limit Theorem: Lyapunov and Lindeberg.
mean and standard deviation. Round the standard devia-
tion to two decimal places. The following exercises are designed to be solved using tech-
nology such as calculators or computer spreadsheets.
b. How unusual would it be for such a sample to show at
least 7.6% suffering from diabetes? 49. A computer simulation. Consider an experiment in which
we toss 500 coins. We repeat the experiment 500 times and
43. Tall people. Take the average height for adult males in record the percentage of heads for each experiment. Use a com-
America to be 5 feet 9 inches. Suppose for the moment that puter simulation to carry out this procedure, and record your
half are taller and half are shorter. Let’s take samples of 1000 results as a histogram. Does it appear that your data are ap-
men and record the percentage of tall (over 5 feet 9 inches) proximately normally distributed? Explain why you gave the
men. Would it be unusual to find such a sample with 45% short answer you did.
people? Would it be unusual to find that the percentage of tall
people in a sample of 1000 college basketball players is over 50. Formula for a bell. Most people know about the number π,
80%? which is approximately 3.14159. But there is another very im-
portant number in mathematics, denoted by the letter e, which
Exercise 44 is suitable for group work. 1
is approximately 2.71828. Plot the graph of e−x /2 for x
2
W-2. Symmetry. The bell curve is symmetric about the mean. Explain what this fact
tells you about normally distributed data.
W-3. More on the shape of the bell curve. Find practical examples of bell curves that
differ in their height and narrowness. In each case explain in practical terms what
the shape indicates about the data.
6 By Richard Herrnstein and Charles Murray (New York: The Free Press, 1994).
Crauder-4100237 QL November 10, 2014 11:34
W-4. Finding examples. What does the 68-95-99.7% rule tell you about normally
distributed data? Find several examples of normally distributed data, and for each
explain in practical terms the information this rule provided.
Exercise 5 provides additional information relevant to this topic.
W-5. Percentile score. Explain what the percentile score on a standardized exam such
as the SAT tells you.
Exercise 17 provides additional information relevant to this topic.
W-6. Central Limit Theorem. A friend says the Central Limit Theorem seems to imply
that all data are normally distributed. Is he correct? Explain.
Exercise 31 provides additional information relevant to this topic.
W-7. z-score. Would the use of z-scores make it possible to compare students’ scores
on the SAT with scores on another standardized test, such as the ACT? Explain why
or why not.
W-8. Mean. If the mean for a set of normally distributed data is 100, is it the case
that half of the data have the value 100? If so, explain why. If not, explain what can
be said.
W-9. Mode. What is the mode for a set of normally distributed data?
432 CHAPTER 6
V irginians strongly support mental health and ethics reforms, but are divided
on Medicaid expansion, according to The Roanoke College Poll. In a very early
test of a possible November match-up, Senator Mark Warner leads Republican
Ed Gillespie by almost 30 points (50%–21%), but Gillespie is unfamiliar to most
Virginians.
The Roanoke College Poll interviewed 633 residents in Virginia between Jan-
uary 13 and January 17 and has a margin of error of ±3.9 percent. (The election
question was only asked of registered voters—N = 553, margin of error = 4.2%.)
There is a consensus among most Virginians that gifts to elected officials and
their families should be regulated more strictly. Sixty percent of the respondents
agree that gifts to elected officials should be limited to $250 in value, while an-
other 20 percent think that limit is too high. Two-thirds (66%) think that gifts to
families should also be limited, and 70 percent agree that those gifts should be
reported.
Election polling
Professional polling organizations have a good record of predicting the outcomes
of elections. Of course, even these professionals sometimes get things wrong. In the
2000 U.S. presidential election, early projections by some news organizations led
them to claim that Al Gore had won Florida, which would have secured his election.
Many pre-election polls underestimated the popular vote for Barack Obama in the
2012 U.S. presidential election. Nevertheless, polls are remarkably good at getting it
right. How are they able to do this?
There are two important aspects of polling, one of which involves mathematical
and statistical considerations. The preceding article told us how many people were
interviewed and also mentioned something they called a “margin of error.” How
were the numbers in the article determined, what do they really mean, and how
much faith should we put in them?
To answer these questions, we need to understand the basics of statistical infer-
ence. In Section 6.2, we saw how the Central Limit Theorem concerning the normal
distribution allows us to draw inferences about a large population from samples of
the population. That discussion is the background for our study of polls in this sec-
tion. Here, we highlight the problem of estimating the accuracy of those inferences.
A second aspect of polling involves “methodology,” that is, the way the polls are
conducted, which includes how the sample is chosen. This is at least as important as
the mathematical aspects.
we want to know what percentage of the marbles are red. The only way to know this
for certain is to actually count all the marbles to determine how many are red and
how many are blue. But this may be impractical if there are hundreds or millions of
them. Is there a way we might make an educated guess and, if so, can we tell how
good that guess might be?
One way is to count the number of red marbles in a small sample. For example,
suppose we shake the jar and then select a few marbles at random from the jar and
find that 20% of them are red. Of course, we cannot conclude that exactly 20% of
the marbles in the jar are red, but based on the sample we have chosen, that would
certainly be a reasonable guess. The big question is, how confident can we be in the
accuracy of this guess?
There are two issues to resolve:
1. How large a sample size should we use?
2. How do we quantify the level of confidence we have in our answer?
Let’s suppose that, in fact, 25% of the marbles in the jar are red, but let’s also
assume we don’t know that. If we draw 100 marbles from the jar, we can use the
ideas developed in Section 5.4 to show that the probability there will be exactly 25
red ones is very small, only about 0.09. But the same ideas show that the probability
a sample of 100 will contain between 23% and 27% (inclusive) is 0.44 or 44%. This
is the key.
The probability that a sample exactly reflects the population may be quite small,
but the probability that it approximately reflects the population is much greater.
We can predict with some confidence that the number of red marbles will be in a
certain range, say, 23% to 27%, even though we can’t be very confident of the exact
number.
Now suppose we chose a marble sample of a different size than 100. What ef-
fect would that have on the probability? The following table gives the approximate
probabilities that between 23% and 27% of the marbles in our sample are red for
various sample sizes, denoted by n. Remember that we’re assuming the jar has a vast
number of marbles and that 25% of the marbles in the jar are red.
Note that the larger the sample size, the greater the probability that our results
fall within the 23% to 27% range. It makes sense that a larger sample is more rep-
resentative of the whole. We saw the same phenomenon in our discussion of the
Central Limit Theorem in the preceding section, where we noted that large sample
sizes make even small deviations from the mean unlikely.
Let’s consider a sample size of 2000. In that case, the probability is 0.96 or 96%
that we will get between 23% and 27% red marbles. This means that if we were to
do many such drawings, 96% of the time we would expect them to be between 23%
and 27% red. That is, for a sample size of 2000, we can expect that 96% of the time
we would get a result within two percentage points of the correct percentage of red
marbles, 25%.
The number 2% in this example is the margin of error. This number is usually
reported as “plus or minus 2%” or as ±2%. We will simply say 2%, with the “plus
or minus” understood.
What does a margin of error of 2% mean in terms of polls? Suppose that a polling
organization samples 2000 marbles and finds that 24% are red. Their report might
be: “One can say with 96% confidence that the percentage of red marbles is 24%,
with a margin of error of two percentage points.” In other words, with 96% con-
fidence the true percentage of red marbles is between 22% and 26%. The interval
22% to 26% is the confidence interval, and the number 96% is the confidence level.
Crauder-4100237 QL November 10, 2014 11:34
434 CHAPTER 6
The meaning of the confidence level is that about 96 out of 100 such samples will
produce a confidence interval that includes the true percentage (25%) of red marbles.
of differentiating on the basis of marble color, we ask people what opinion they hold
on a certain question.
As an example, suppose a poll reports with 95% confidence that 45% of the
American people support foreign aid and that the margin of error is two percentage
points. (Note that 95% is a common confidence level for polls conducted by Gallup,
Harris, Zogby, and other professional polling organizations.) Let’s recall what this
means: If we conducted this poll thousands of times, and each time we wrote down
the percentage of the people in the sample who support foreign aid, we would expect
about 95% of those results to be within two percentage points of the true value. In
brief, we can be fairly confident that the support of the entire population for foreign
aid lies in the confidence interval between 43% and 47%.
We reiterate that the result from a single poll does not determine for certain the
truth about the population, not even within the range given by the margin of error.
In the poll about foreign aid, we cannot be certain that between 43% and 47% of the
entire population really do support aid. In fact, the theory says that we should not
be surprised if about 5% of the time our sample will not be within two percentage
points of the true value. Indeed, it is entirely possible that 70% support aid, although
the chances are very remote.
Also, opinion polls are often taken periodically over time and usually by differ-
ent organizations, so you don’t have to trust just one poll. Your confidence should
increase if several polls report similar results.
436 CHAPTER 6
to decide how many people should be questioned (the sample size) and how they
should be chosen. But there is a simple rule of thumb that gives a reasonably good
estimate for a confidence
√ level of 95%: For a sample size of n, the margin of error is
approximately 100/ n%.
Normally, when we use this estimate, we round the margin of error to one decimal
KIyoshi Ota/EPA/Landov
place. This estimate for the margin of error is a consequence of the discussion in the
preceding section. The key points are that about 95% of normal data lie within two
standard deviations of the mean, and the Central Limit Theorem gives a formula for
the standard deviation that depends on the sample size n.
To show how to use the formula, suppose we choose 1100 Americans at random
and obtain their opinion on a certain issue.
For a confidence level of 95% we get
a margin of error of approximately 100/ 1100%, or about 3.0%. It is perhaps
surprising that we can, with 95% confidence, know the opinions of more than 280
million people within three percentage points by sampling only 1100 people.
with n = 500:
100
Margin of error ≈ %
500
Our margin of error is about 4.5%. We can be 95% confident that our poll result is
within 4.5 percentage points of the true value.
7 http://www.alafista.com/2013/05/13/young-females-view-kyary-pamyu-pamyu-as-fashion-leader/
Crauder-4100237 QL November 10, 2014 11:34
The next example looks at a poll concerning recovery efforts after a natural
disaster.
438 CHAPTER 6
a. The poll surveyed 683 people. What is the approximate margin of error for a
95% confidence interval?
b. The poll of 683 people found that 61% of respondents were dissatisfied with
New Jersey’s recovery efforts so far. Can we conclude with certainty that no
more than 66% of displaced residents were dissatisfied?
c. Suppose instead that the poll of 683 people had found that 53% were
dissatisfied. Explain what we could conclude from this result. Could we assert
with confidence that a majority of displaced residents were dissatisfied with
recovery efforts?
d. Suppose we wish to have a margin of error of two percentage points.
Approximately how many people should we interview?
SOLUTION
a. To find the margin of error, we use the formula
100
Margin of error ≈ √ %
n
with n = 683:
100
Margin of error ≈ %
683
or about 3.8%.
b. Our answer to part a tells us that we can be 95% confident that the poll
number of 61% is within 3.8 percentage points of the true percentage of all
displaced residents who were dissatisfied. This means that if we conducted this
poll 100 times, then about 95 of the results would be within 3.8 percentage
points of the true value. Thus, it is very likely that the true value is between
61 − 3.8 = 57.2% and 61 + 3.8 = 64.8%. Because the whole interval is below
66%, we can be quite confident (at a 95% level) that no more than 66% of
displaced residents were dissatisfied. However, we cannot make this conclusion
with absolute certainty.
c. We can be 95% confident that the poll number of 53% is within 3.8
percentage points of the true percentage of all displaced residents who were
dissatisfied. This means that if we conducted this poll 100 times, about 95 of the
results would be within 3.8 percentage points of the true value. Thus, it is very
likely that the true value is between 53 − 3.8 = 49.2% and 53 + 3.8 = 56.8%.
Most of this interval falls above 50%, so we continue to have good reason to
think that a majority of displaced residents were dissatisfied. But, because a
portion of the interval falls below 50%, we should be more cautious in drawing
conclusions.
d. To calculate how large our sample should be, we use the formula
100
Margin of error ≈ √ %
n
and substitute 2% for the margin of error. Thus, we want to find n so that
100
2= √
n
Crauder-4100237 QL November 10, 2014 11:34
Hence,
n = 502 = 2500
We should interview about 2500 people. (Note that one Harris Poll with a
95% confidence level and a margin of error of 2% surveyed 2415 people—very
close to the 2500 given by the formula.)
The procedure in part d of Example 6.18 can be used to estimate the sample size
needed to get a given margin of error: To get a margin of error of m (as a percentage),
the sample size must be about (100/m)2 . This assumes a 95% level of confidence. In
part d of the example, we had m = 2%, and this formula gives us
100 2
= 502 = 2500
2
440 CHAPTER 6
Barack Obama
51
48 48 48 Mitt Romney
47 47
46 47 46 45
43 42
Swing states include Colorado, Florida, Iowa, Michigan, Nevada, New Hampshire, New Mexico,
North Carolina, Ohio, Pennsylvania, Virginia and Wisconsin.
With six months to go before Election Day, a new poll finds Mitt Romney in a
dead heat with President Barack Obama in 12 key political battleground states.
A USA Today/Gallup poll of 12 expected swing states—Colorado, Florida,
Iowa, Michigan, Nevada, New Hampshire, New Mexico, North Carolina, Ohio,
Pennsylvania, Virginia and Wisconsin—finds Obama leading Romney by just two
points among registered voters, 47 percent to 45 percent. The results are well
within the poll’s plus or minus 4 percent margin of error.
What does this article mean when it says that the race is a “dead heat”? First
of all, “dead heat” is a synonym for “tie.” So this article seems to be saying that
the race is, statistically speaking, a tie. But the poll has Obama ahead 47% to
45%, so why does the article say that the race is a statistical tie? What is going on
here?
Assume a 95% confidence level. The ±4% margin of error in this poll means
that we can say with 95% confidence that Obama’s poll number is within 4 per-
centage points of his actual percentage—so it could be as low as 47 − 4 = 43%.
We can say with 95% confidence that Romney’s poll number is within 4 percentage
points of his actual percentage, so it could be as high as 45 + 4 = 49%. Because the
confidence intervals overlap, Romney could actually be ahead of Obama. In other
words, the difference of two percentage points between them is “within the margin of
error.” This fact is what the reporter means by a “dead heat.” The fact that the differ-
ence of two percentage points between them is within the margin of error does not,
Crauder-4100237 QL November 10, 2014 11:34
however, mean that they are statistically tied. Probably Obama does have a lead, but
the level of confidence in that statement is smaller than the 95% level we have been
considering.
Sidney Harris
A good explanation may be found in the following excerpt from an earlier on-
line article written by Humphrey Taylor, chairman of the Harris Poll at the time. In
discussing the 2000 presidential election, it clarifies very well the idea of a statistical
dead heat:
If one well-conducted poll shows a candidate with a two-point lead, there is
a significant possibility that there is no such lead. However, even a two-point
lead should never be described as a “statistical dead heat” as the probability
is better than 50% that the lead is real. For this reason, the words “statistical
dead heat” should only be used when the candidates are actually tied. Better
still, avoid them altogether.
When, as they do now, one poll after another shows the vice president with
a lead, even a small one, the chances statistically that this is due to sampling
error are so small as to be virtually non-existent.9
Methodology of polling
In our initial discussion of polling, we assumed that we could obtain information
from a random sample of the American public in the same way we could determine
the color of randomly selected marbles. In fact, it is much harder to “randomly select”
an American than it is to randomly select a marble, and it is far easier to determine a
marble’s color than it is to determine an individual’s opinion on a given issue. In an
idealized poll, all that is needed to get a more accurate poll is to increase the sample
size. In practice, things are not so easy. Here is another quote from the article by
Humphrey Taylor cited earlier: “The idea that the sample size is the main source of
error, and that the bigger the sample the smaller the error, is dangerously misleading.”
The professional polling organizations themselves are good sources of informa-
tion on the difficulties of accurate polling. The following excerpt is from the Web site
of the Harris Poll. It is the methodology section of a polling report.
9 “Contrary to what much of the media have been saying, Gore has a statistically significant, if not huge, lead over
Bush,” September 18, 2000, http://www.harrisinteractive.com/vault/Harris-Interactive-Poll-Research-
CONTRARY-TO-WHAT-MUCH-OF-THE-MEDIA-HAVE-BEEN-SAYIN-2000-09.pdf
Crauder-4100237 QL November 10, 2014 11:34
442 CHAPTER 6
T he Harris Poll was conducted online within the United States between De-
cember 12 and 18, 2006 among 2,309 adults (aged 18 and over). Figures for
age, sex, race, education, region and household income were weighted when
necessary to bring them into line with their actual proportions in the population.
Propensity score weighting was also used to adjust for respondents’ propensity
to be online. . . .
With pure probability samples, with 100 percent response rate, it is possible
to calculate the probability that the sampling error (but not other sources of error)
is not greater than some number. With a pure probability sample of 2,309 adults
one could say with a 95 percent probability that the overall results would have a
sampling error of +/− 2 percentage points. However that does not take other
sources of error into account. This online survey is not based on a probability
sample and therefore no theoretical sampling error can be calculated.
© Wally McNamee/CORBIS
The preceding Harris Poll methodology statement says, “This online sample is not
based on a probability sample.” A probability sample is one in which the participants
are selected randomly, which is crucial to the mathematical basis of any poll. This
particular Harris poll was conducted online, so the participants were not randomly
selected. They were “self-selected,” which means that they are people who happened
to find the Web site and chose to respond. By contrast, one Zogby America poll states
that the poll “involved 1013 likely voters selected randomly from throughout the
48 contiguous states using listed residential telephone numbers.” This is arguably a
probability sample, although it misses people without landline phones, for example.
Newspapers and TV shows often invite their audiences to respond to questions.
It is usually pointed out that this is not a “scientific poll,” which means, among
other things, this is not a probability sample—the respondents are not randomly
chosen and may not represent a true cross section of the whole country. For example,
Crauder-4100237 QL November 10, 2014 11:34
444 CHAPTER 6
value would lie in the confidence interval corresponding to the result. Therefore, we
are 95% confident that the actual percentage lies between 62.1% and 69.9%.
Try It Yourself 6.19: Estimating sample size: 5% margin of error 400.
17. Increasing sample size. If you double the sample size in a 24. Biased questions? Suppose you conduct a poll asking about
poll, what effect does that have on the margin of error? By what foreign policy. You conceive of two possible ways to word the
factor should the sample size in a poll be increased in order to question you wish to ask:
cut the margin of error in half? First question: Do you approve of current U.S. policies to-
18. Proportionality. One quantity is proportional to another if ward Iran?
it is a nonzero, constant multiple of the other. Thus, the state- Second question: Do you approve of the president’s policies
ment “y is proportional to x” can be expressed using the for- toward Iran?
mula y = cx, where c is a nonzero constant.
Do you think you would get the same result using either ques-
Determine which of the following is correct, and explain
tion? Explain why or why not.
your answer. For a sample size of n, the margin of error is ap-
proximately proportional to: 25. Legalizing marijuana. Suppose you supported legalizing
√ marijuana. Make up a question asking whether people sup-
a. n c. n
√ ported legalizing marijuana that will bias the answer your way.
b. 1/n d. 1/ n Now do this supposing you did not support legalizing mari-
19–22. 90% confidence levels. There are estimates of the juana.
margin of error for confidence levels other than 95%. For a 26. Tax cuts. Suppose you are in favor of a tax cut. Make up a
90% confidence level and√a sample size of n, the margin of er- question asking whether people support a tax cut that will bias
ror is approximately 82/ n%. the answer your way. Now do this supposing you are against a
19. If we survey 1000 people, what is the approximate margin tax cut.
of error for a 90% confidence level? Round your answer as a
27. Polling your campus. If you poll students on your campus
percentage to one decimal place.
regarding how they intend to vote in an upcoming state elec-
20. Suppose we conduct a poll using a representative sample tion, do you believe the results would give a valid picture of
of 1000 people to find whether Americans like salad. We find what might happen in the statewide election? Explain why or
that 53% say they do. For a 90% confidence level, explain why not.
in plain English what we can conclude from this survey. Can
28–33. Simulating a poll. Exercises 28 through 33 refer to
you assert with confidence that a majority of Americans like
the following; simulations of repeated polling. We toss a mil-
salad?
lion or so fair coins on the floor and think of them as the voting
21. Suppose our salad survey of 1000 people found that only population. Coins that show a head intend to vote Democratic.
48% like salad. For a 90% confidence level, explain in plain Coins that show a tail intend to vote Republican. We want to
English what we can conclude from this survey. Can you as- predict the upcoming election, so we do what the Harris Poll
sert with confidence that less than a majority of Americans like would do. We draw a random sample of 1000 coins from the
salad? floor and tabulate how they intend to vote.
22. Suppose we wish to have a margin of error of 2% with a But let’s do more. Let’s repeat this poll simulation 1000
90% confidence level. Approximately how many people should times and tabulate the results. With the aid of computer sim-
we interview? ulation, we did exactly that. The number of polls showing the
given percentage range of people intending to vote Democratic
23. Sample size and margin of error. The margin of error for
is shown in the accompanying table. (Here, “47% to 48%”
a given sample size presented in Rule of Thumb 6.2 is a good
means 470 through 479 people intending to vote Democratic,
estimate. But it is not exactly right. To get the exact number,
and so on.) Exercises 28 through 33 should be done in that
we need to make a complicated calculation using the methods
order.
developed in Section 5.4. Suppose 25% of Americans actually
believe Congress is doing a good job, but we don’t know this.
So we select n Americans at random and ask their opinion Percent voting Less than 47% to 48% to 49% to
about Congress. The probability that our poll will show be- Democratic 47% 48% 49% 50%
Number of polls 23 75 140 234
tween 23% and 27% (for a margin of error of 2%) depends
on the sample size n and is given in the table below. (This table Percent voting 50% to 51% to 52% to At least
is the same as the one on page 433.) Democratic 51% 52% 53% 53%
Number of polls 250 157 94 27
Number 100 500 1000 1500 2000 2500
n selected 28. Recall that these are fair coins. What is an accurate de-
Probability 0.44 0.72 0.87 0.93 0.96 0.98 scription of the voting population (in terms of percentage vot-
23% to 27% ing Democratic)? (For this question, the table with the polling
data is irrelevant.)
a. For a 2% margin of error, what confidence level will a 29. Use Rule of Thumb 6.2 to estimate the margin of er-
sample size of 1000 give? ror for each of these polls for a confidence level of 95%.
Round your answer as a percentage to the nearest whole
b. We wish to have a margin of error of 2%. What sample
number.
size should we choose for a 72% confidence level?
Crauder-4100237 QL November 10, 2014 11:34
446 CHAPTER 6
30. What percentage of polls gave predictions that within three idential elections in 1936 and 1948. Do research to discover
percentage points accurately describe the voting population, as who the candidates were, what happened in the elections, and
given in the answer to Exercise 28? Round your answer as a what the polling organizations did wrong.
percentage to the nearest whole number. The following exercises are designed to be solved using tech-
31. Explain how your answers in Exercises 28, 29, and 30 are nology such as calculators or computer spreadsheets.
consistent with each other. 35. Margin of error. Make a table showing the approximate
32. What percentage of polls gave predictions that within two margin of error for sample sizes ranging from 1000 to 2000 in
percentage points accurately describe the voting population, as steps of 100. Assume that the level of confidence is 95%, and
given in the answer to Exercise 28? Round your answer as a round each percentage to two decimal places.
percentage to the nearest whole number. 36. Sample size. Make a table showing the sample size needed
33. Make a histogram summarizing the data in the table. (See to get a given margin of error ranging from 2% to 3% in steps
Section 6.1.) of 0.1 percentage point. Assume that the level of confidence is
95%, and round to the nearest whole number.
34. History. Two famous examples of polls predicting the
wrong outcome of an election are given by the American pres-
W-7. Push polls. Reputable polling organizations work very hard to produce unbi-
ased, accurate results. But sometimes polls are conducted by organizations with a
different goal. As an extreme example, an automobile company might ask, “Do you
prefer our high-quality cars or the inferior cars the competition sells?” Then they
might report the results as, “70% of people polled prefer our cars.” Polls that are
designed to push a certain point of view rather than gain information are called push
polls. Can you find actual examples of push polls?
Crauder-4100237 QL November 10, 2014 11:34
W-8. Try it yourself. Conduct a poll of students on your campus regarding an issue
relevant to current campus life. Report the results, including margin of error and
confidence level. What issues did you encounter, and how did you deal with them?
You might consider such things as question structure and how to make your sample
as nearly random as possible.
W-9. Polling works. Often people question how polls can be accurate when they
sample relatively few people. Use the concepts of this section and the preceding one
to explain how polling a relatively small sample can give fairly accurate results.
448 CHAPTER 6
Experimental design
There are many factors involved in the setup of an experiment or clinical trial. In
this case, the purpose of the trial is to determine the efficacy (or lack of efficacy) of a
drug.
In a typical trial, one group of subjects is given the actual drug or treatment and
another group is given a placebo. A placebo is a benign substance that contains no
medication and that is outwardly indistinguishable from the real drug. One kind of
placebo might be a simple sugar pill. Of course, the subjects do not know which they
are getting.10
cartoonstock.com
In some studies, the second group will be given a standard treatment rather than
a placebo to compare the effect of a known drug with a proposed new drug. In these
cases, the known drug will be disguised to look like the proposed drug.
Here are the typical key ingredients when planning a clinical trial:
1. Researchers first decide how much of a difference between treatment groups is
medically important.
2. Then they decide how many subjects should be enrolled in the trial. The sample
should include enough participants for the researchers to have confidence in the
results.
3. Next, part of the group is given the drug or treatment to be tested, and the other
subjects are given the placebo. The group given the placebo is usually called the
control group.
4. Finally, the researchers compare the subjects in the two groups to see whether
there is a significant difference between their conditions after the trial. The key
word here is “significant.” That’s where statistics comes in.
So far so good, but just as in polling, there are many possible pitfalls in such a
study. One such pitfall is that the observer or the subject can inadvertently affect the
results. For example, suppose a researcher has worked so hard to develop a drug that
he or she has come to believe in it, and just really wants it to work. If the researcher
knows which subjects have been given the drug and which the placebo, the chance
10 It’s interesting that taking a placebo can create an expectation of improved health and thus result in positive medical
benefits. This is called the “placebo effect.”
Crauder-4100237 QL November 10, 2014 11:34
exists that he or she may inadvertently interpret the condition of the subjects in favor
of the drug.
That is why clinical trials are usually double blind, an experiment in which neither
the subjects nor the researchers know who is getting the drug and who is getting the
placebo. A third party is the only one who knows which patient is receiving what.
In this Key Concept, we have not said what we mean by “unlikely.” We will
explain this point shortly. First we present examples related to statistical significance.
450 CHAPTER 6
subtle. For example, consider the following table showing the results of a hypothet-
ical trial:
Certainly, the treated group improved a little more than the control group, but
is this difference statistically significant? The answer is not obvious. Statisticians in-
volved with clinical trials determine the answer by calculating something called the
p-value. The p-value measures how likely it is that the difference between the two
groups in a clinical trial would occur by chance alone if the treatment had no effect.
The p-value is a way to quantify the word “unlikely” that appeared in our explana-
tion of the term statistically significant.
with a p-value of 0.002. Explain the meaning of the p-value in this test. Would the
result of this study normally be accepted as statistically significant?
SOLUTION
If Vioxx and Aleve had the same effects, the probability that the difference in results
between the Vioxx and the Aleve patients would occur by chance alone is 0.002 or
2/10th of 1%. Because this probability is much less than the accepted threshold of
0.05, this result is certainly statistically significant.
As the example indicates, in practice, one computes a p-value that measures the
probability that the difference between the subjects in the control and test groups
11 Mukherjee, D., Nissen, S.E., and Topol, E.J., “Risk of Cardiovascular Events Associated With Selective COX-2
Inhibitors,” Journal of American Medical Association (2001), 954–959.
Crauder-4100237 QL November 10, 2014 11:34
would occur by pure chance if the treatment had no effect. To explain in detail how
this is done would take us too far afield, so we will use a simplified example to give
the basic idea.
Idealized trials
Suppose we test a drug that we believe may prevent colds. There is no control group
in this case, but suppose we know that, on average, 50% of people will get colds this
winter. Suppose we treat 100 patients with this drug and find that 55 did not get a
cold. Would we conclude that the drug is effective in preventing colds? Or did the
fact that 55 patients did not get a cold just happen by chance?
Here’s the idea: Without treatment, 50% of people will get colds, and from this
fact it follows that the probability of 55 or more out of 100 subjects not getting a
cold is 0.184 or 18.4%.12 The number 0.184 is the p-value for this experiment. It
says that in 18.4% of observations of 100 untreated subjects, we expect that 55 or
more of the subjects will not get a cold. Therefore, having 55 or more stay healthy
is not a highly unlikely event—we expect it to occur almost 20% of the time even
without treatment. The benefits of the test drug would be in doubt if only 55 out of
100 patients we actually treated did not get a cold.
Suppose, however, that 65 of the patients taking the test drug did not get a cold.
The p-value for this outcome turns out to be 0.002 = 0.2%. This is the probability
that 65 or more patients will not get a cold if the drug had no effect. We’d be very
surprised to see that many patients not get a cold merely by chance. If 65 of the pa-
tients taking the test drug did not get a cold, it is reasonable to believe that something
other than chance is at work.
Having a small p-value associated with a study suggests that the outcome is due to
something other than chance alone, but it doesn’t necessarily say what that something
is. For this reason, it’s important for the researchers to take care to avoid factors,
other than the drug or procedure being tested, that might cause the two groups to
differ. If this is done, then having a small p-value suggests that the drug or procedure
was probably responsible.
In the end, no matter how good the study, there is always a probability, small
though it may be, that the conclusion suggested by the results is incorrect. That is
one reason why independent researchers might repeat a clinical trial. If a small p-
value is obtained in several different trials, then we have much greater confidence in
the result. This is important—after all, these results decide which drugs get on the
market and which do not.
12 This number can be found using the ideas developed in Section 5.4.
Crauder-4100237 QL November 10, 2014 11:34
452 CHAPTER 6
For example, the time required for a trip and the speed you drive are negatively
correlated, because an increase in travel time accompanies a decrease in speed. By
contrast, the price of a pizza and its diameter are positively correlated because an
increase in price usually accompanies an increase in diameter.
In these examples, the correlations are the result of real relationships between
the variables, but some correlations are the result of chance or coincidence, and we
have to be careful not to draw unwarranted conclusions. Furthermore, even when
two variables are connected in some way, we should not necessarily assume that one
causes the other. For example, there is a positive correlation between ice cream sales
and the number of shark attacks on swimmers. Both of these tend to rise during warm
weather when more people swim and buy ice cream, but certainly neither causes the
other.
There are statistical methods that quantify the extent to which two variables are,
in fact, correlated. In particular, the method called linear regression quantifies the
extent to which two variables are linearly correlated. We have already introduced
linear regression in the context of trend lines in Section 3.1.
The following data show weekly expenditures in British pounds of households in
the given region:13
We want to determine whether there is a linear correlation between the two vari-
ables, spending on alcohol and spending on tobacco. First we make a scatterplot of
the data, as shown in Figure 6.38. The plot suggests a positive correlation between
13 Taken from Family Expenditure Survey, Department of Employment (British Official Statistics). We have omitted the
data for Northern Ireland.
Crauder-4100237 QL November 10, 2014 11:34
5 5
y = 0.6115x + 0.1082
4 4
Tobacco
Tobacco
3 3
2 2
1 1
4 5 6 7 4 5 6 7
Alcohol Alcohol
FIGURE 6.38 Tobacco versus alcohol FIGURE 6.39 Trend line added.
spending, in British pounds.
the two variables: For the most part, an increase in one accompanies an increase
in the other. To determine whether this is a linear correlation, we add the trend (or
regression) line, as seen in Figure 6.39. Note that the points seem roughly to fall
along this line. This fact leads us to suspect that there is a linear relation between the
two variables. In summary, we have evidence that a positive linear correlation exists
between spending on alcohol in a region and spending on tobacco in that region.
Figure 6.39 also shows the equation of the trend line. We round the coefficients
to get y = 0.612x + 0.108. Here, x represents spending on alcohol and y represents
spending on tobacco, both in British pounds. This equation lets us do more than
just establish a link between the variables. It quantifies, approximately, how the two
variables are related. For example, if in a given region, households spend an average
of 4 pounds per week on alcohol, we would expect 0.612 × 4 + 0.108 = 2.556, or
about 2.56 pounds per week to be spent on tobacco. Nevertheless, the data do not
tell us that the consumption of alcohol causes tobacco usage, or vice versa.
Figure 6.40 shows a plot of the data along with the trend line. Do these data show
a correlation between people visiting the dentist within the past year and the elderly
having all their teeth extracted? If so, what type of correlation exists?
SOLUTION
In general, as the percent visiting the dentist increases, the percent having had all
their teeth extracted decreases. Therefore, the two variables are correlated, and the
correlation is negative. The graph suggests that the correlation is linear.
Crauder-4100237 QL November 10, 2014 11:34
454 CHAPTER 6
Goran Bogicevic/Shutterstock
30
25
15
y = -0.6745x + 65.996
10
0
55 65 75 85
Dental visits
Figure 6.41 shows a plot of the data along with the trend line. Are length and
speed correlated? If so, what type of correlation is there?
100
y = 2.0298x + 5.0873
(feet per second)
80
Running speed
60
40
20
0
0 10 20 30 40 50
Length (inches)
Exactly how strong is the evidence of a linear correlation between alcohol and
tobacco expenditures or between seeing the dentist regularly and having had all teeth
extracted? For alcohol and tobacco purchases, consider again Figure 6.39. If the
points were to lie exactly on the line, anyone would be convinced of a strong linear
Crauder-4100237 QL November 10, 2014 11:34
relationship. In fact, the points roughly follow the trend line, but most of them do
not fall on the line. How do we decide how strong the correlation is?
Statisticians attempt to quantify the degree of linear correlation by using a number
known as the correlation coefficient. The formula for calculating the correlation co-
efficient is complicated, and the job is best left to your calculator or computer. In the
case of Figure 6.39, the correlation coefficient is 0.78. What does this number mean?
4 4
3 3
2 2
1 1
0 0
0 1 2 3 4 5 0 1 2 3 4 5
4 4
3 3
2 2
1 1
0 0
0 1 2 3 4 5 0 1 2 3 4 5
456 CHAPTER 6
One calculation shows that the correlation coefficient between the prices of U.S.
stocks in the S&P 500 Index and stocks in the London stock exchange called the
FTSE 100 averaged 0.848. Does this correlation coefficient indicate that there is a
great advantage in adding a London stock index fund to your portfolio?
SOLUTION
A correlation coefficient of 0.848 indicates a strong positive linear correlation. This
says that a London index fund’s prices will rise and fall, more or less, in a similar
way to a U.S. index fund, so there is no great advantage in adding this investment.
Causation
Now we return to the subject of causation and its relation to correlation. Consider
the excerpt from a news article.
Crauder-4100237 QL November 10, 2014 11:34
This article suggests that there is a link (a positive correlation) between boredom
and early death. By noting that boredom is probably only a symptom associated with
other risk factors, the article warns us against concluding that boredom is a cause of
early death.
We have already noted that determining whether one event causes another event
is much more difficult than determining mere correlation. Two phenomena can be
correlated without one of them causing the other. When two quantities are correlated,
there are three possibilities: One causes the other, both are caused by something else,
or the correlation may simply be coincidence—just pure chance.
Here is an example: Suppose a mother notices that from time to time her baby
gets a runny nose and a cough. The cough seems to come first most of the time, so she
concludes that the cough causes the runny nose. In fact, the most likely explanation is
that both are caused by a third factor: The baby has a cold. That they occur together,
and thus are correlated, does not mean that one must cause the other.
For another example, consider again the data on tobacco and alcohol consump-
tion in British households for various regions of Britain (page 452). Consumption of
tobacco and consumption of alcohol in a region are certainly positively correlated
because, in general, an increase in one corresponds to an increase in the other. (In
fact, the correlation coefficient of 0.78 confirms the positive correlation.) It’s possible
that one of these somehow causes the other. But it’s also possible that people who
have a tendency to use one of these products also have a tendency to use the other.
A classic example of an argument about correlation versus causation is the con-
nection between smoking and lung cancer. Many studies have shown that smoking
and lung cancer are positively correlated. Although it is indisputable that smokers
have a much higher rate of lung cancer than non-smokers, the tobacco industry has
argued in the past that this does not necessarily tell us that smoking causes lung can-
cer. They have suggested that it may be like the baby with a cough and runny nose.
The conditions may occur together, but how can we say one causes the other? Maybe,
they argue, there is a gene or some unknown behavior that increases the propensity
for both smoking and lung cancer.
Showing causation requires much more evidence than showing correlation, but
enough research has been done to convince the medical community that smoking
does, in fact, cause lung cancer.
Crauder-4100237 QL November 10, 2014 11:34
458 CHAPTER 6
14 Many such totally ethical studies have been done, and the results surprise no one.
Crauder-4100237 QL November 10, 2014 11:34
what this means. Home TV & Video U. S. World Politics Entertainment Tech Travel
5. An interesting drug. Does anything appear to be wrong NIH Study Finds Regular Aspirin Use
in the following news release? Explain. A drug company an- May Reduce Ovarian Cancer Risk
nounced the results of a clinical trial for a certain drug for hang-
nails. Patients showed statistically significant improvement ( p-
value > 0.05) in the treatment group as compared to the con-
trol group.
W omen who take aspirin daily may reduce their
risk of ovarian cancer by 20 percent, according
to a study by scientists at the National Cancer Insti-
6. What caused it? In a clinical trial for patients with hang- tute (NCI), part of the National Institutes of Health.
nails, the test group soaked their hands in warm water and were However, further research is needed before clinical rec-
given an experimental drug. They showed statistically signifi- ommendations can be made. The study was published
cant improvement ( p-value < 0.03) as compared to the control February 6, 2014, in the Journal of the National Cancer
group, which did nothing to treat their hangnails. The presi- Institute.
dent of the drug company sponsoring this trial asserted that It is estimated that over 20,000 women in the
this p-value shows that the new drug had a 97% probabil- United States will be diagnosed with ovarian cancer in
ity of helping hangnails. Discuss the validity of the president’s 2014, and more than 14,000 will die from the disease.
comment. Early stage ovarian cancer may be successfully treated.
However, symptoms associated with this disease can
7. Sunburn. The CEO of a certain drug company announced mimic more common conditions, such as digestive and
that the results of a clinical trial for an experimental drug in bladder disorders, so for this reason and others, it is of-
patients with sunburn showed statistically significant improve- ten not diagnosed until it has reached advanced stages.
ment. His justification for this claim was the following fact: Late stage ovarian cancer leaves women with limited
55% of the subjects treated with the drug got better in two treatment options and poor prognoses, making pre-
days, but it is well known that, on average, only 50% of sun- ventive strategies potentially important for controlling
burn victims get better in two days. Discuss the validity of the this disease.
CEO’s comment. Suggestion: To find the p-value, see the dis- ...
cussion on page 451 of a test drug for preventing colds.
Among study participants who reported whether or
8. What’s the cause? A certain drug company announced that
not they used aspirin regularly: 18 percent used aspirin,
the results of a clinical trial for a new drug in patients with sun-
24 percent used non-aspirin NSAIDs [non-steroidal
burn showed statistically significant improvement ( p-value =
anti-inflammatory drugs], and 16 percent used ac-
0.04). The president of the company asserted that this shows
etaminophen. The researchers determined that partici-
that 96% of the subjects in the trial were helped by the drug.
pants who reported daily aspirin use had a 20 percent
Discuss the validity of the president’s comment. lower risk of ovarian cancer than those who used as-
9. Reducing ovarian cancer risk. This exercise refers to In the pirin less than once per week. For non-aspirin NSAIDs,
News 6.9, which is an excerpt from the Web site of the National which include a wide variety of drugs, the picture was
Cancer Institute. less clear: the scientists observed a 10 percent lower
According to the report, women who take aspirin daily may ovarian cancer risk among women who used NSAIDs
reduce their risk of ovarian cancer. But the report states that the at least once per week compared with those who used
finding for NSAIDs “did not fall in a range that was significant NSAIDs less frequently. However, this finding did not
statistically.” Explain in your own words what this statement fall in a range that was significant statistically. In con-
means and what implications it has. trast to the findings for aspirin and NSAIDs, use of ac-
etaminophen, which is not an anti-inflammatory agent,
was not associated with reduced ovarian cancer risk.
Crauder-4100237 QL November 10, 2014 11:34
460 CHAPTER 6
10–11. Smarter people. Exercises 10 and 11 refer to the c. Damage in a city at the epicenter of an earthquake and
following information. A 2011 article available from the Na- the Richter scale reading of the earthquake
tional Institutes of Health discusses the benefits of breastfeed- d. A golfer’s tournament ranking and her score
ing during the first year of an infant’s life. The article cites
controversial studies of the effects of breastfeeding on cogni- 16–17. Intensity of light. If one moves away from a light
tive development and IQ. Part of the controversy is due to the source, the intensity of light striking one’s eye decreases. Fig-
methodologies used in the studies, and the real possibility that ure 6.46 shows a graph of the relationship between the intensity
women who do breastfeed are inherently different, in terms of of light and distance from the source. (We have not included the
their own habits and lifestyles, from those who do not breast- units because they are not relevant here.)
feed. For instance, in one key study of almost 500 infants of
non-smoking women who gave their children only breast milk,
tests showed between 2.1 and 3.8 higher IQ points in these
children than those who were not breastfed exclusively (they
were fed breast milk plus infant formula). In this and similar
Intensity
studies, the overall IQ advantage from breastfeeding appears
to be small, but the effect is highly significant from a public
health perspective. The controversy surrounding the benefits
of breastfeeding requires one to critically consider the cause-
and-effect relationships between breastfeeding and cognitive
outcomes. Distance from source
10. One paragraph in the article states, “the above arguments FIGURE 6.46 Intensity versus distance.
require one to critically consider the cause-and-effect relation-
ships between breastfeeding and cognitive outcomes.” Use the
concepts developed in this section to explain what this para-
16. Is there a correlation between the intensity of light and
graph means.
distance from the source? If so, is the correlation positive or
11. Using your own words, explain what the next-to-last sen- negative?
tence from the summary above means.
17. Does the relationship shown by the graph appear to be
12–14. Calculating p-values. Suppose that for a certain ill- linear?
ness, the probability is 45% that a given patient will improve
18. Correlation coefficient. Assume that Figures 6.47 through
without treatment. Then the probability that at least n out of
6.50 show data sets with correlation coefficients 0.99, 0.58,
20 patients will improve without treatment is given in the fol-
−0.10, and −0.63. Which figure shows data with which corre-
lowing table:
lation coefficient?
n 10 11 12 13 14 15 16
Probability 14
at least 0.41 0.25 0.13 0.058 0.021 0.0064 0.0015 12
n improve 10
8
6
4
2
We give an experimental drug to 20 patients who have this ill- 0
ness. Exercises 12 through 14 use this information. –2 2 4 6 8
–4
12. Suppose 11 patients show improvement. What is the –6
–8
p-value?
13. Suppose we count the test significant if the p-value is 0.05 FIGURE 6.47
or less. How many patients must show improvement in order
to make the test statistically significant?
14. Suppose 16 of the 20 subjects showed improvement. In 14
light of the design of the test, would it be appropriate to rec- 12
ommend this drug as a “cure” for this disease? 10
15. Correlation as an idea. For each pair of variables, deter- 8
mine whether they are positively correlated, negatively corre- 6
lated, or uncorrelated. 4
a. Brightness of an oncoming car’s headlights and your dis- 2
tance from the oncoming car 0
0 2 4 6 8
b. The number of home runs hit by Barry Bonds in a year
and my water bill for that year FIGURE 6.48
Crauder-4100237 QL November 10, 2014 11:34
21. Use the regression line in Figure 6.52 to estimate how many
14 4
millions of people Ford should have expected to remember one
12 2
of its ads. (See Exercise 20.) Round your answer to the nearest
10 0
2 4 6 8 10 million of those remembering.
8 –2
6 –4 22. The correlation coefficient was reported as 0.65. Comment
4 –6 on the hypothesis that there is a straight-line relation between
2 –8 advertising spending and effectiveness.
0 –10
0 2 4 6 8 23. A study? Many times you hear testimonials regarding the
health benefits of certain activities, foods, or herbal supple-
FIGURE 6.49 FIGURE 6.50 ments. Suppose your trusted friend Jane tells you that in spite
of a good diet and exercise and treatment from her physi-
cian, her blood pressure remained high. Then one year ago,
she started a daily intake of 2 ounces of dried lemon peel. Her
19–22. Advertising expense. Figure 6.51 shows the effective- blood pressure returned to normal within 3 months and has
ness of spending on advertising by major American retail com- remained so.
panies. Exercises 19 through 22 refer to this graph. Discuss the pros and cons of Jane’s testimonial regarding
the curative properties of dried lemon peel in light of the con-
cepts presented in this section.
24. Vegetarians. The following excerpt comes from an article
120 published by the Vegetarian Society of the United Kingdom.
Millions remembering
100
80
60
40
20
0
IN THE NEWS 6.10 Search
Your good friend Larry tells you that this article has con-
vinced him to raise his child as a vegetarian. Discuss what you
120 would say to Larry in light of the concepts presented in this
Millions remembering
y = 0.3632x + 22.163
100 section.
80 25. Potatoes contribute to athletics. Suppose a study showed
60 that more than 90% of gold medal winners from the 2010
40 Olympic Games in Vancouver eat potatoes as a regular part
20 of their diet. In light of the concepts presented in this section,
0 discuss whether it’s proper to conclude that eating lots of pota-
0 50 100 150 200 toes will improve one’s athletic abilities.
Millions spent
26. Coed housing. The following excerpt comes from an arti-
cle that appeared in the Oklahoma State University newspaper,
FIGURE 6.52 Regression line added. The Daily O’Collegian.
Crauder-4100237 QL November 10, 2014 11:34
462 CHAPTER 6
CHAPTER SUMMARY
Descriptive statistics refers to organizing data in a form that can be clearly under-
stood and that describes what the data show. Inferential statistics refers to the attempt
to draw valid conclusions from the data, along with some estimate of how accurate
those conclusions are. We study both in this chapter.
464 CHAPTER 6
standard deviation σ of a list of numbers measures how much the data are spread
out from the mean. It is calculated as follows: Subtract the mean from each number
in the list, square these differences, add all these squares, divide by the number of
entries in the list, and then take the square root. As a formula, this is
(x1 − μ)2 + (x2 − μ)2 + . . . + (xn − μ)2
σ =
n
where the data points are x1 , x2 , x3 , . . . , xn and μ is the mean.
A common method of describing data is the five-number summary. It consists of
the minimum, the first quartile, the median, the third quartile, and the maximum.
The first quartile is the median of the lower half of the list of numbers, and the
third quartile is the median of the upper half. A boxplot is a pictorial display of a
five-number summary.
A histogram is a bar graph that shows frequencies with which certain data appear.
Histograms can be especially useful when we are dealing with large data sets.
0.008 0.008
0.006 0.006
0.004 0.004
0.002 0.002
0 0
200 300 400 500 600 700 800 200 300 400 500 600 700 800
FIGURE 6.53 Normal distribution with FIGURE 6.54 Normal distribution with
mean 500 and standard deviation 100. mean 500 and standard deviation 50.
data point lies above or below the mean. The z-score, together with a standard table
of z-scores and percentiles, allows easy and complete analysis of normally distributed
data.
One of the most striking facts about normal distributions is that, according to
the Central Limit Theorem, data that consist of percentages of population samples
(all of the same size) are approximately normally distributed. This allows us to use
the normal distribution to analyze virtually any type of data from repeated samples.
KEY TERMS
mean (or average), p. 394 five-number summary, p. 398 confidence level, p. 434
median, p. 394 standard deviation, p. 401 statistically significant, p. 449
mode, p. 394 histogram, p. 406 p-value, p. 450
bimodal, p. 394 normally distributed, p. 414 positively correlated, p. 451
multimodal, p. 394 z-score (or standard score), p. 419 negatively correlated, p. 451
outlier, p. 397 percentile, p. 420 uncorrelated, p. 451
first quartile, p. 397 Central Limit Theorem, p. 423 correlation coefficient, p. 455
second quartile, p. 397 margin of error, p. 434
third quartile, p. 397 confidence interval, p. 434
Crauder-4100237 QL November 10, 2014 11:34
466 CHAPTER 6
CHAPTER QUIZ
1. The goals scored (by either team) in the games played by a 6. For a certain disease, 40% of untreated patients can be ex-
soccer team are listed below according to the total number of pected to improve within a week. We observe a population of
goals scored in each game. Find the mean, median, and mode 80 untreated patients and record the percentage who improve
for these data. Round the mean to one decimal place. within one week. According to the Central Limit Theorem,
such percentages are approximately normally distributed. Find
Goals scored 0 1 2 3 4 5 6 7 8 the mean and standard deviation of this normal distribution.
by either team Round the standard deviation to two decimal places.
Number of games 7 14 11 12 3 2 1 2 2
Answer
The mean is 40%. The standard deviation is
Answer Mean: 2.4; median: 2; mode: 1. (40 × 60) /80 or about 5.48 percentage points.
If you had difficulty with this problem, see Example 6.1. If you had difficulty with this problem, see Example 6.13.
2. Calculate the five-number summary for this list of automo-
bile prices: 7. Use the mean and standard deviation calculated in the pre-
ceding exercise to find the percentage of test groups of 80 un-
$27,000 $19,500 $24,500 $25,600 $17,000 treated patients in which 50% or less improve within a week.
$32,700 $18,000 $27,800 $29,000 $30,200 Round your answer as a percentage to one decimal place.
Answer z-score = 10/5.48 or about 1.8 standard deviations
Answer The minimum is $17,000; the first quartile is
above the mean. Table 6.2 (page 421) gives a corresponding
$19,500; the median is $26,300; the third quartile is $29,000;
percentage of about 96.4%.
the maximum is $32,700.
If you had difficulty with this problem, see Example 6.13.
If you had difficulty with this problem, see Example 6.3.
8. Explain the meaning of a poll that says 55% of Americans
3. Calculate the mean and standard deviation for this list of
approve of the job the president is doing, with a margin of error
automobile prices:
of 3% and a confidence level of 95%.
$27,000 $19,500 $24,500 $25,600 $17,000
$32,700 $18,000 $27,800 $29,000 $30,200
Answer In 95% of such polls, the reported approval of the
president will be within three percentage points of the true ap-
proval level. Thus, we can be 95% confident that the true level
Answer Mean: $25,130; standard deviation: $5078.
is between 52% and 58%.
If you had difficulty with this problem, see Example 6.5.
If you had difficulty with this problem, see Example 6.15.
4. The weights of oranges in a harvest are normally distributed,
9. A poll asked 800 people their choice for mayor. What is
with a mean weight of 150 grams and standard deviation of 10
the approximate margin of error for a 95% confidence level?
grams. In a supply of 1000 oranges, how many will weigh be-
Round your answer as a percentage to one decimal place.
tween 140 and 160 grams? How many will weigh between 130
and 170 grams?
Answer 3.5%.
Answer 680 oranges between 140 and 160 grams; 950 or- If you had difficulty with this problem, see Example 6.16.
anges between 130 and 170 grams.
If you had difficulty with this problem, see Example 6.9. 10. What sample size is needed to give a margin of error of
2.5% with a 95% confidence level?
5. The weights of oranges in a harvest are normally distributed,
with a mean weight of 150 grams and standard deviation of
Answer 1600.
10 grams. Calculate the z-score for an orange weighing If you had difficulty with this problem, see Example 6.19.
a. 160 grams
b. 135 grams 11. For each of these situations, determine whether the results
Round your answers to one decimal place. described appear to be statistically significant.
a. You are walking down the street and encounter three people
Answer in a row with blond hair.
a. 1 b. You cut a deck of playing cards 30 times and get an ace every
b. −1.5 time.
Answer 13. Decide whether the following variables are positively cor-
a. Not statistically significant related, negatively correlated, or uncorrelated.
b. Statistically significant a. Your grades and the number of hours per week you study
b. Your grades and the number of hours per week you party
If you had difficulty with this problem, see Example 6.20.
c. Your grades and the number of cookies per week you eat
12. A clinical trial for an experimental drug finds that the drug d. The daily average temperature and sales of cell phones
is effective with a p-value of 0.08. Explain the meaning of the
e. The daily average temperature and sales of overcoats
p-value in this case. Would this result normally be accepted as
statistically significant? f. The daily average temperature and sales of shorts
Answer The probability that the outcome of the clinical trial Answer Positively correlated: a and f; negatively correlated:
would occur by chance alone if the drug had no effect is 0.08 b and e; uncorrelated: c and d.
(8%). This result would not normally be accepted as statisti- If you had difficulty with this problem, see Example 6.22.
cally significant because the p-value is larger than 0.05.
7 GRAPH THEORY
T
he following appeared in a report published in 2010 by The National Security
Agency’s Review of Emerging Technologies.
7.2 Hamilton circuits and Home TV & Video U. S. World Politics Entertainment Tech Travel
traveling salesmen:
Revealing Social Networks of Spammers
Efficient routes
7.3 Trees: Viral e-mails S pam doesn’t really need an introduction—anyone who owns an email ad-
dress likely receives spam emails every day. However, spam is much more
than just an annoyance. Spam’s hidden economic cost for companies in wasted
and spell checkers storage, bandwidth, technical support, and most important, the loss of employee
productivity, is astronomical.
Understanding the behavior of spammers is one of the benefits of a global
approach to fighting spam. But what do the social networks of spammers look
like? In particular, how well organized are spammers? Do they operate alone, or
in groups?
The social network of spammer can be represented as a graph consisting of
nodes and edges, as shown in the figure.
The nodes correspond to spammers, and an edge between two nodes cor-
responds to a social relationship between the corresponding spammers. A social
relationship can be inferred by the use of common resources or by similar behavior
patterns over time.
Communities in a social network emerge by partitioning the graph into groups
of nodes. Sets of nodes in the same group are highly similar and sets of nodes in
different groups are not similar. The main challenge in constructing the graph is
choosing the edges because we can not observe relationships among spammers.
This problem does not arise in most other community detection studies. For ex-
ample, in a friendship or collaboration networks, users willingly participate in the
study, and information on relationships among members of the network is readily
available.
469
Crauder-4100237 QL November 10, 2014 12:9
470 CHAPTER 7
This article shows one application of the mathematical concept of a graph, which
in the simplest terms tells whether certain objects are somehow connected to each
other.1 For example, we can model Facebook using a graph if we say that two users
are connected when they are friends. Graphs can be represented pictorially using dots
and segments, as in the background of the image in Figure 7.1. The dots are called
vertices, and the segments connecting them are called edges.
Graph theory has important real-world applications, and in this chapter we will
examine some of the basic ideas behind them. Serious applications of graph theory
involve highly complex situations, such as fast routing of dynamic content on the
World Wide Web, with huge numbers of vertices and edges that can be dealt with
only by very powerful computers. We will focus on everyday applications of graphs
but also will explain their importance to business, government, and industry.
1 In earlier chapters, we talked about a graph as a pictorial representation of data or a formula. In this chapter, we are
going to use the term in a completely different way.
Crauder-4100237 QL November 10, 2014 12:9
Satellites will take the place of 1960s-era radars for navigation and surveillance,
increasing the accuracy of location information, enabling planes to safely fly closer
together and providing pilots a complete, real-time picture of all air traffic. The
satellite technology will also help airport flow, allowing planes to land and depart
faster, quieter and greener. Satellite navigation, moreover, will enable pilots to
choose more direct routes, no longer limiting them to zigzagging over ground-
based radar stations. The results? Safer and more efficient flights, fewer weather
delays and reduced emissions and noise.
Efficient routes for airplanes save money and get passengers to their destinations
in a timely manner. Efficient routing is important for many business applications,
from trucking to mail delivery. Typical options may consist of cities to pass through
or highway routes to utilize. In this section, we study how to use graphs to find
efficient routes.
Before we begin this study, let’s agree that the graphs we will discuss have two
properties: First, an edge cannot start and end at the same vertex. (See Figure 7.2.)
Allowable edges
472 CHAPTER 7
Second, the graphs are connected, that is, each pair of vertices can be joined by a
sequential collection of edges, a path. The graph in Figure 7.3 is connected, but the
one in Figure 7.4 is not connected—it comes in two pieces.
E E
D D
F F
C C
B B
A A
The first step in our study of efficient routes is learning how to use graphs to make
models of real situations.
Tom
Wichita
Tulsa
John Jerry
Perry
Stillwater
Jill
Oklahoma
City Mary Bill
Sherman
Dallas
Bob
SOLUTION
The appropriate graph is shown in Figure 7.7. Note, for example, that the edge join-
ing the Cubs and Giants indicates they have a game scheduled. There is no edge
joining the Tigers with the Yankees because they do not have a game scheduled.
Suzanne Tucker/Shutterstock
Cubs
Red Sox
Lions
Yankees
Bears
Giants
Tigers
474 CHAPTER 7
Euler circuits
We noted at the beginning of this section that graphs can be used to find efficient
routes. Let’s consider the example of mail delivery. The edges in Figure 7.8 represent
streets along which mail must be delivered, and the vertices represent the post office
and street intersections. We have labeled one vertex as “post office” and the others
as A through F . An efficient route for the mail carrier starts at the post office, travels
along each street exactly once, and ends up back at the post office. Such a route is
shown in Figure 7.9. We indicate the route by listing the vertices as we encounter
them along the route:
The path for this postal route starts and ends at the same vertex and traverses each
edge exactly once. Such a path is known as an Euler circuit or Euler cycle after the
eighteenth-century Swiss mathematician Leonhard Euler (pronounced “oiler”). Euler
is considered to be the father of modern graph theory. Euler circuits represent efficient
paths for postal delivery, street sweepers, canvassers, and lots of other situations.
Such efficient routes are not always possible. For example, trial and error can
E D C
quickly convince us that there is no Euler circuit for the graph in Figure 7.10. Any
FIGURE 7.10 No Euler circuit that starts and ends at the post office must either omit a street or traverse some
circuit. street more than once.
Degree 2
> KEY CONCEPT
Degree 3 The degree of a vertex is the number of edges that touch that vertex. Some
FIGURE 7.11 The numbers in
texts use valence instead of degree.
this figure indicate the degree
of each vertex. The numbers in Figure 7.11 indicate the degree of each vertex.
Crauder-4100237 QL November 13, 2014 11:53
Miami
Chicago
Seattle New York
Washington
San Francisco
Jacksonville
Los San New
Angeles Antonio Orleans
476 CHAPTER 7
There is a crucial relationship between Euler circuits and the degrees of vertices.
An Euler circuit goes into and then out of each vertex and uses each edge exactly
once, as is illustrated in Figure 7.14. Because the edges meeting a vertex come in
pairs (in-edges paired with out-edges), the number of edges at each vertex must be a
multiple of 2. That is, the degree of each vertex is even.
In the eighteenth century, Leonhard Euler discovered that this degree condition
completely characterizes when such circuits exist.
FIGURE 7.14 Edges of an Theorem2 (Euler, 1736) A connected graph has an Euler circuit precisely when
Euler circuit come in pairs. every vertex has even degree.
The two mail routes shown in Figures 7.8 and 7.10 provide evidence for this
theorem: In Figure 7.8, each vertex has even degree, and there is an Euler circuit. But
in Figure 7.10, vertex A has degree 3, which is odd. Thus, there is no Euler circuit
for Figure 7.10.
Miami
2 “Solutio Problematis ad Geometriam Situs Pertinentis (The solution of a problem relating to the geometry of
position),” Commentarii Academiae Scientiarum Imperialis Petropolitanae 8 (1736; published 1741), 128–140.
Crauder-4100237 QL November 13, 2014 11:53
Degree 3
Degree 5 Degree 3
Degree 3
478 CHAPTER 7
Cycle 2
B A
Cycle 1
We note that the two circuits in Figure 7.19 also meet at vertex B. We can find a
different Euler circuit by breaking the circuits and rejoining at B rather than A. This
is shown in Figure 7.22. When circuits meet at several vertices, we can break and
reroute the circuits at any of these vertices.
A second method for making Euler circuits is known as Fleury’s algorithm. For
this method, we remove edges from the graph one at a time and add them to a path
that will grow to become an Euler circuit. To lengthen this path, we add edges by
repeatedly applying the following two-step procedure:
Crauder-4100237 QL November 10, 2014 12:9
A Cycle 2 Cycle 2
Two circuits B A B A
Two circuits
joined Cycle 1 Cycle 1
FIGURE 7.20 FIGURE 7.21 The resulting Euler FIGURE 7.22 Breaking and rerouting
Breaking and circuit. at B.
rerouting at A.
Step 1: Remove an edge from the graph and add it to the path so that the path is
lengthened. But do not choose an edge whose removal will cut the graph into two
pieces unless there is no other choice.
Step 2: Delete from the graph any isolated vertices resulting from the first step.
Repeat these two steps until an Euler circuit is constructed. Let’s apply Fleury’s
algorithm to the graph in Figure 7.23. Starting at the vertex A, we have selected the
three red edges to begin Fleury’s algorithm. In Figure 7.24, we have used these edges
to start an Euler circuit.
We have arrived at vertex D, and from here we might select any of the yellow
edges in Figure 7.25. But removal of the edge DC would result in the disconnected
graph shown in Figure 7.26.
D C D C
B B
E E
F F
A A
D C D C
B B
E E
F F
A A
480 CHAPTER 7
So we can choose either of the edges DF or DE, but not DC. In Figure 7.27, we
have selected the edge DE and added that edge to the path in Figure 7.28.
We have arrived at the vertex E. Removal of the edge EF, as shown in Figure 7.29,
will result in a disconnected graph, but there is no other choice, so we do it. We add
this edge to the path, as shown in Figure 7.30. Note that this leaves the vertex E with
no edges attached. We delete this isolated vertex and continue.
From this point on, we have no decisions to make, and the completed Euler circuit
is shown in Figure 7.31.
D C D C
B B
E E
F F
A A
FIGURE 7.27 The edge DE is FIGURE 7.28 The edge DE is added to the
selected. path.
D C D C
B B
E E
F F
A A
D C
B
E
© David Grossman/Alamy
57 St & 57 St &
7 Ave 5 Ave
42 St & Rockefeller
Port 7 Ave Center Grand
Authority Central
Station
14 St & Union
8 Ave 14 St & 14 St & Square
7 Ave 6 Ave
Greenwich Washington
Village Square Park
SOLUTION
Each of the vertices has degree 2 or 4. Because each vertex has even degree, we are
guaranteed an Euler circuit. One could find an Euler circuit by trial and error, but we
apply the method of breaking and rerouting. In Figure 7.33, we have marked two
circuits that share no common edges.
Circuit 1: 57 St & 7 Ave - 57 St & 5 Ave - Rockefeller Center - 14 St & 6 Ave -
Washington Square Park - Greenwich Village - 14 St & 7 Ave - 42 St & 7 Ave - 57
St & 7 Ave
Circuit 2: Port Authority - 42 St & 7 Ave - Rockefeller Center - Grand Central Station
- Union Square - 14 St & 6 Ave - 14 St & 7 Ave - 14 St & 8 Ave - Port Authority
In Figure 7.34, we have broken the circuits at Rockefeller Center and rerouted to
combine the two circuits into one. The result is the required subway tour:
Euler circuit: 57 St & 7 Ave - 57 St & 5 Ave - Rockefeller Center - Grand Central
Station - Union Square - 14 St & 6 Ave - 14 St & 7 Ave - 14 St & 8 Ave - Port
Authority - 42 St & 7 Ave - Rockefeller Center - 14 St & 6 Ave - Washington Square
Park - Greenwich Village - 14 St & 7 Ave - 42 St & 7 Ave - 57 St & 7 Ave
You should verify that Fleury’s algorithm can also be used to produce an Euler
circuit.
482 CHAPTER 7
Eulerizing graphs
We have seen that in many practical settings, a most efficient route is an Euler circuit—
if there is one. If there is no Euler circuit, any circuit traversing each edge must in-
clude some backtracking of edges, that is, going back along some paths we have
already taken. We still want a most efficient route possible in the sense that we have
to backtrack as little as possible. We do this by Eulerizing graphs, temporarily adding
duplicate edges.
We know that if all vertices have even degree, then we can find an Euler circuit.
We will show how to find an efficient path when there are exactly two odd-degree
vertices. The graph in Figure 7.35 shows a part of the shuttle route in Yosemite
National Park. Note that the vertices at Yosemite Creek and the Ansel Adams Gallery
have degree 3 and all other vertices have even degree. Because there are odd-degree
vertices, there is no Euler circuit: It is not possible to start and end at Yosemite Creek
and travel each shuttle route without backtracking. We want to find a route that
involves the least possible backtracking. The first step in our search for an efficient
route is to pick a shortest3 (in terms of the number of edges) path joining the odd-
degree vertices, in this case Yosemite Creek and Ansel Adams Gallery. There is more
than one choice for this, and any of them will serve. We have selected one shortest
path from Yosemite Creek to the Ansel Adams Gallery and highlighted it in orange
in Figure 7.35. The second step in our search is to add duplicate edges, as shown in
Figure 7.36. Observe that after the addition of the duplicate edges, all vertices have
even degree. Hence, there is an Euler circuit, as shown in Figure 7.36. We complete
the process by collapsing the duplicate edges back to their originals. The result is
a circuit starting and ending at Yosemite Creek that backtracks as little as possible
(twice in this case), as shown in Figure 7.37.
Museum
Ansel Museum Ansel Museum Ansel
Yosemite Yosemite Yosemite
Creek Adams Adams Adams
Gallery Creek Gallery Creek Gallery
FIGURE 7.35 Step 1: Find shortest FIGURE 7.36 Step 2: Add duplicate FIGURE 7.37 Final step: Collapse
path between odd-degree vertices. edges. Step 3: Find an Euler circuit. the duplicate edges.
Best route: Yosemite Creek - Visitor center - Ansel Adams Gallery - Village store -
Yosemite Creek - Museum - Ansel Adams Gallery - Visitor center - Yosemite Creek
3 In large graphs, even the seemingly simple problem of finding a shortest path can be problematic.
Crauder-4100237 QL November 10, 2014 12:9
A E A E
F F
B B
D D
C C
A E
484 CHAPTER 7
Le Havre
Caen
Cherbourg
St. Malo
Billboa
Try It Yourself 7.2: Finding the degree of a vertex: Rail service Chicago: de-
gree 7; Jacksonville: degree 2; Los Angeles: degree 3; New Orleans: degree 4; New
York: degree 2; San Antonio: degree 3; San Francisco: degree 3; Seattle: degree 2;
Washington: degree 4.
Try It Yourself 7.3: Applying Euler’s theorem: A rail service map Some of the
vertices have an odd degree, so it is not possible to take the desired trip.
Try It Yourself 7.5: Finding Euler circuits: A subway tour The required Euler cir-
cuit is
57 St & 7 Ave 57 St & 5 Ave
14 St & 7 Ave
Try It Yourself 7.6: Eulerizing graphs: Streets in our town An Eulerized graph is
A E
D
C
Bears play Red Sox and Yankees. Central security connects to all computers.
Tigers play Red Sox, Yankees, and Giants. The food court computer is connected to central security
Make a graph to model this season’s Little League schedule. and custodial services.
4. More Little League. The Little League in our town consists The custodial services computer is connected to central se-
of the Red Division—the Cubs, Lions, Bears, and Tigers—and curity, food court, and Toys “R” Us.
the Blue Division—the Red Sox, Yankees, and Giants. This sea- The Toys “R” Us computer is connected to central security
son each team is to play exactly three teams from the other di- and custodial services.
vision. Can you make a graph to model this schedule? Is such Make a graph modeling the mall computer network.
a schedule possible?
9. Degrees of vertices. Find the degrees of each of the vertices
5. A computer network. Each of the offices 501, 502, 503, 504, in Figure 7.43.
505, 506, and 507 has a computer. Some of these are connected E
to others:
The computer in 501 connects to computers in 502, 503, D
505, and 507. A
The computer in 502 connects to the computer in 501, 503,
and 504.
The computer in 503 connects to all other computers.
Make a graph that shows the office computer network. C
B
6. Bridges. Figure 7.42 shows islands in a river and bridges
connecting them. Make a graph model of walking paths us- FIGURE 7.43
ing vertices to indicate land masses and edges to indicate paths 10. More degrees of vertices. Find the degrees of each of the
across bridges. vertices in Figure 7.44.
E
A
C A D
B
D
E C
B
486 CHAPTER 7
15. Refer to the graph in Figure 7.46. 19. Paving streets. The edges in Figure 7.50 represent unpaved
roads, and the vertices represent intersections. Either find a
E D route that will not require the paving machine to move along
F already paved streets or show that no such route exists. That
is, either find an Euler circuit or explain why no such circuit
exists.
C
A A
B F
FIGURE 7.46
B E
16. Refer to the graph in Figure 7.47.
A
C D
B FIGURE 7.50
C
D G
20. Utility ditches. The edges in Figure 7.51 represent ditches
F that must be dug for utilities, and the vertices represent homes.
Either find a route that will not require the ditch digger to travel
E
along ditches that have already been dug or show that no such
FIGURE 7.47 route exists. That is, either find an Euler circuit or explain why
no such circuit exists.
17–27. Euler circuits in practical settings. Exercises 17
through 27 involve Euler circuits in practical settings. A G
F
17. Mail routes. The edges in Figure 7.48 represent streets B
along which there are mail boxes, and the vertices represent
intersections. Either find a mail route that will not require any
E
retracing by the mail carrier or explain why no such route ex-
C
ists. That is, either find an Euler circuit or explain why no such D
circuit exists.
FIGURE 7.51
F
Post office
21. Street inspections. The edges in Figure 7.52 represent
B streets that must be periodically inspected, and the vertices
C D represent intersections. Either find a route that the inspectors
can take to avoid traveling over previously inspected streets or
show that no such route exists. That is, either find an Euler
E circuit or explain why no such circuit exists.
FIGURE 7.48
A
F
18. Snow plows. The edges in Figure 7.49 represent streets that
have to be cleared of snow, and the vertices represent intersec-
B
tions. Either find a route for the snow plow that will not require E
it to travel over any streets that have already been cleared of
snow or explain why no such route exists. That is, either find
C D
an Euler circuit or explain why no such circuit exists.
A FIGURE 7.52
B
22. Repairing hiking paths. The edges in Figure 7.53 represent
F hiking paths that must be repaired, and the vertices represent
C intersections. Either find a route that the crews can follow to
E
avoid traveling over already repaired trails or show that no
D
such route exists. That is, either find an Euler circuit or explain
FIGURE 7.49 why no such circuit exists.
Crauder-4100237 QL November 10, 2014 12:9
A G FIGURE 7.58
F
B
28. Making Euler circuits. The graph in Figure 7.59 represents
streets and intersections. We have marked three circuits, labeled
cycle 1, cycle 2, and cycle 3. Patch these cycles together at ver-
E
tices E and G to make an Euler circuit.
C D
C Cycle 1 D
FIGURE 7.55
A B E F
25. Fire inspections. The edges in Figure 7.56 represent streets
Cycle 2
A J Cycle 3 I
F
FIGURE 7.59
B
E 29–32. Efficient circuits. The graphs in Exercises 29 through
32 each have exactly two vertices of odd degree, namely A and
B. Eulerize these graphs to find a most efficient path that starts
C D and ends at the same vertex and traverses each edge at least
once.
FIGURE 7.56
Crauder-4100237 QL November 10, 2014 12:9
488 CHAPTER 7
⎛A B C⎞
A B A − 0 1
B ⎝0 − 1⎠
C 1 1 −
A
33. Euler paths. In some graphs where no Euler circuit exists,
there is instead an Euler path. That is a path that starts at one
vertex, traverses each edge exactly once, and ends at another
vertex. Euler showed in 1736 that such a graph has exactly two
vertices of odd degree (the beginning and ending points of the
Euler path). It turns out that Fleury’s algorithm produces an
Euler path in this situation. In this exercise, we will show an- B
other way to find Euler paths when they exist. We will use the C
example of the graph in Figure 7.60.
35. Make the graph represented by the following incidence
matrix:
⎛A B C D E F⎞
A − 0 1 1 0 1
B ⎜0 − 0 1 1 0⎟
⎜ ⎟
C ⎜1 0 − 1 1 1⎟
⎜1 − 1 0⎟
⎟
D ⎜ 1 1
E ⎝0 1 1 1 − 1⎠
F 1 0 1 0 1 −
FIGURE 7.60 A graph with an Euler path.
Crauder-4100237 QL November 10, 2014 12:9
36. Make the incidence matrix that describes the graph shown
below. ⎛A B C D E F⎞
A − 0 1 1 0 1
A B ⎜0 − 0 1 1 0⎟
⎜ ⎟
C ⎜1 0 − 1 1 1⎟
⎜1 − 1 0⎟
⎟
D ⎜ 1 1
B E ⎝0 1 1 1 − 1⎠
C
F 1 1 1 1 1 −
D G
38. Considering the results of the previous exercise, can you
F
E suggest a property that all incidence matrices of graphs must
have?
37. Can you make a graph with the following incidence ma- 39. History. Write a paper on the life and work of Leonhard
trix? Explain any difficulties you encounter. Euler.
W-2. Three dimensions. It’s easy to imagine a graph in three dimensions. Such a
graph would just be a collection of dots connected by segments. Can you think of a
situation in which a three-dimensional graph would be a natural model?
W-3. Graphs and biology. Use the Internet to find an application of graph theory to
biology, and write a brief report on your findings.
W-4. Vertices and edges. Give several examples of graphs, and for each graph make
two calculations:
Calculation 1: Add up the degrees of all the vertices in the graph.
Calculation 2: Count the number of edges in the graph.
Do your calculations lead you to any ideas regarding graphs? Can you show that
your idea is always correct?
W-5. A road map. You are looking at a local road map. In what situations might
you be interested in an Euler circuit for the map? What businesses might have need
of such a circuit?
Exercise 17 provides additional information relevant to this topic.
W-6. The best you can do. If there is no Euler circuit, we have shown how to make
efficient routes when there are exactly two odd-degree vertices. Use the Web to in-
vestigate ways of making efficient routes when there are more than two odd-degree
vertices.
Exercise 29 provides additional information relevant to this topic.
W-7. Social media. The text pointed out that graphs can be used to model connec-
tions between users of Facebook. Can you think of other social media that can be
modeled using graphs?
W-8. Web links. Many Web sites have hyperlinks, also called links, to other sites.
Consider a graph in which the vertices are Web sites and the edges indicate that one
Web site has a link to another. Could you tell by looking at the graph which site
is the source (that is, which one originates the link) for an edge? If not, how could
you mark the graph to indicate which is the source? Use the Internet to investigate
directed graphs.
Crauder-4100237 QL November 10, 2014 12:9
490 CHAPTER 7
W-9. More on Web links. Consider the graph in which the vertices are all of the sites
on the Web and the edges indicate that one Web site has a hyperlink to another. Does
that graph have any circuits? Is that graph connected, in the sense that each pair of
vertices can be joined by a path? Does that graph have an edge that starts and ends
at the same vertex?
In the preceding section, we studied Euler circuits for graphs. The path for the
traveling salesman problem described in the article above is a different type of circuit,
called a Hamilton circuit. With an Euler circuit, our goal is to traverse efficiently the
edges by using each edge exactly once. For a Hamilton circuit, we want to traverse
efficiently the vertices; that is, we want a path starting and ending at the same place
that visits each vertex exactly once (though it may miss some edges altogether).
To compare the two types of circuits, we consider a highway map. An Euler circuit
would be appropriate if you wish to travel along each road exactly once, as shown
in Figure 7.61. A Hamilton circuit would be appropriate if you wish to visit each city
exactly once, as shown in Figure 7.62. Note that in a Hamilton circuit, some edges
may not be used at all.
Oklahoma Oklahoma
Albuquerque City Albuquerque City
Los Los
Angeles Angeles
El Paso El Paso
A graph may have many Hamilton circuits, just as it may have many Euler cir-
cuits. For example, Figures 7.63 and 7.64 show two different Hamilton circuits for
the same graph. Thus, when you complete an exercise or follow one of our examples,
you may well find a correct answer that is different from the circuit that we found.
Some graphs have no Hamilton circuit. A little trial and error will quickly con-
vince you that the graph in Figure 7.65 has no Hamilton circuit: Any circuit contain-
ing all the vertices must visit the middle vertex twice.
B A
C E
D
FIGURE 7.65 A graph with no Hamilton
circuit.
Hamilton circuits are very important in routing problems that arise with airlines,
delivery services, Internet routing, etc. For example, suppose a delivery driver has
dozens of packages to be delivered to various destinations in a day. She would want
to travel to each destination exactly once and not go to (or pass) one destination
many times. If the vertices are the destinations, along with headquarters, and the
Crauder-4100237 QL November 10, 2014 12:9
492 CHAPTER 7
edges are the road routes between them, the desired efficient route is a Hamilton
circuit from headquarters and back.
Let’s see how this observation can help us make Hamilton circuits. Suppose we
need an efficient route that allows us to make deliveries from our terminal to the
bank, the hardware store, Starbucks, Walmart, the Shoe Box, and Old Navy. That
is, we want to find a Hamilton circuit for the graph in Figure 7.66. The vertices at
the hardware store and at Walmart have degree 2. Hence, any Hamilton circuit must
use the four roads highlighted in Figure 7.67.
It is now easy to see how to complete the Hamilton circuit, as shown in Fig-
ure 7.68.
It is helpful to keep in mind that as we try to extend a path to a Hamilton cir-
cuit, we cannot return to any vertex already used (except at the last step, when the
path returns to its starting point). Therefore, no smaller circuit can be part of any
Hamilton circuit.
FIGURE 7.66 A delivery route FIGURE 7.67 We must use edges FIGURE 7.68 Completing the
problem. that meet vertices of degree 2. delivery route.
SOLUTION
Note that the vertices Grizzly Lake, Mammoth, Norris, and Lewis Lake all have
degree 2. Thus, in any Hamilton circuit, we must use the hiking paths that touch
these vertices. The corresponding edges are highlighted in Figure 7.70. The result is
that we already have a circuit at the top of Figure 7.70. Such a circuit cannot be part
of any larger Hamilton circuit, and we conclude that it is not possible to take the
desired walking path. We will have to settle for a path that visits at least one point
of interest more than once.
Santa Store
Staples Kay
Radio Jewelers
Shack
Linens’n
Outback Things
Goody’s
494 CHAPTER 7
13
33
using a shortest route. This is an example of the famous traveling salesman problem,
928 1425
which seeks to find a shortest route that visits each city once and then returns home.
In other words, the problem is to find a “shortest” Hamilton circuit. The traveling
865 salesman problem has many important applications.
New Orleans Miami We can solve the traveling salesman problem in this case by listing the six possible
routes and calculating the mileage for each:
FIGURE 7.72 A simple map. • New York - Miami - Chicago - New Orleans - New York: Distance 5153 miles.
• New York - New Orleans - Chicago - Miami - New York: Distance 5153 miles.
(Note: This is the reverse of the preceding trip.)
• New York - Miami - New Orleans - Chicago - New York: Distance 3938 miles.
• New York - Chicago - New Orleans - Miami - New York: Distance 3938 miles.
(Note: This is the reverse of the preceding trip.)
• New York - Chicago - Miami - New Orleans - New York: Distance 4385 miles.
• New York - New Orleans - Miami - Chicago - New York: Distance 4385 miles.
(Note: This is the reverse of the preceding trip.)
The shortest route is then New York - Miami - New Orleans - Chicago - New
York or its reverse. It is worth noting that the mileage saved over alternative routes
is significant.
solution of the traveling salesman problem for this map. Calculate the mileage for
38
54
1313 12
each possible route to find the solution.
736
SOLUTION
1177
Phoenix
There are six possible routes altogether, but we need to list only three because we
gain no new information by looking at the reverse of a route:
Houston
• Kansas City - Houston - Phoenix - Portland - Kansas City: Distance 5179 miles.
FIGURE 7.73 Delivery map.
• Kansas City - Phoenix - Houston - Portland - Kansas City: Distance 6722 miles.
• Kansas City - Phoenix - Portland - Houston - Kansas City: Distance 5641 miles.
Bakery 4 Coffee shop
The shortest route of 5179 miles is the first one listed.
8 3
9 6 TRY IT YOURSELF 7.8
For the graph in Figure 7.74, find a shortest delivery route that starts and ends at the
Restaurant 5 Convenience bakery and makes deliveries to the coffee shop, the restaurant, and the convenience
store
store. That is, solve the traveling salesman problem for this graph.
FIGURE 7.74 Delivery map.
The answer is provided at the end of this section.
Crauder-4100237 QL November 10, 2014 12:9
An example of the traveling salesman problem for five cities is shown in Fig-
ure 7.77. If we start at Seattle, it turns out that we need to check 12 separate routes
(Hamilton circuits) to solve the problem for this map.
Minneapolis
9 96
5
161
Seattle 2619 Buffalo
241
949
3 925
81
1263
19
03
26
1819
Memphis
San Diego
496 CHAPTER 7
The number of Hamilton circuits for the traveling salesman problem starting at
a given city, not counting reverse routes, is shown in the table below.4
Indeed, on a complete graph with n vertices, there are (n−1)!/2 Hamilton circuits
starting at a given vertex, not counting reverse routes. (Recall that k! = k × (k − 1) ×
. . . × 1. For example, 3! = 3 × 2 × 1 = 6.)
This table means, for example, that if we deliver to 11 or more locations, we
have to check more than a million routes to find a shortest one. For 100 cities, there
are about 10158 possible tours. As a consequence, even myriads of the very fastest
computers could not solve the traveling salesman problem for 100 cities within a
reasonable time frame. The result is that solving the traveling salesman problem for
a large number of cities by listing each possible route is totally impractical.5
In 1962, Procter & Gamble held a contest to find a shortest route connecting 33
U.S. cities. There are 131,565,418,466,846,765,083,609,006,080,000,000 possible
routes for 33 cities, but amazingly, the contest did have some winners!
Nearest-neighbor algorithm
Solving the traveling salesman problem by listing each route is not practical if the
number of cities is large. To get around this difficulty, researchers have developed a
number of clever algorithms that find Hamilton circuits that are “nearly optimal.”
One of these is the nearest-neighbor algorithm, which starts at a vertex and makes
a Hamilton circuit. At each step in the construction, we travel to the nearest vertex
that has not already been used. If there are two or more vertices equally nearby, we
just pick any one of those nearest vertices.
Minneapolis Minneapolis
96 96
1 619 5
Buffalo 1619 5
Buffalo
Seattle Seattle
2619 2619
241 241
949
949
3 925 3 925
81
81
1263 1263
19
19
03 03
26 26
1819
Memphis 1819
Memphis
5 2 5 2 5 2 5 2
Kinder- Primary Kinder- Primary Kinder- Primary Kinder- Primary
garten school garten school garten school garten school
6 6 6 6
7 2 7 2 7 2 7 2
7 1 8 7 1 8 7 1 8 7 1 8
4 4 4 4
FIGURE 7.81 22 miles. FIGURE 7.82 20 miles. FIGURE 7.83 18 miles. FIGURE 7.84 28 miles.
(This is one of two routes
that are found by the
nearest-neighbor algorithm
in part b.)
Crauder-4100237 QL November 10, 2014 12:9
498 CHAPTER 7
5 2 5 2 5 2 5 2
Kinder- Primary Kinder- Primary Kinder- Primary Kinder- Primary
garten school garten school garten school garten school
6 6 6 6
7 2 7 2 7 2 7 2
7 1 8 7 1 8 7 1 8 7 1 8
4 4 4 4
FIGURE 7.85 17 miles. FIGURE 7.86 29 miles. FIGURE 7.87 28 miles. FIGURE 7.88 27 miles.
(This is one of two routes
that are found by the
nearest-neighbor algorithm
in part b.)
5 2 5 2 5 2 5 2
Kinder- Primary Kinder- Primary Kinder- Primary Kinder- Primary
garten school garten school garten school garten school
6 6 6 6
7 2 7 2 7 2 7 2
7 1 8 7 1 8 7 1 8 7 1 8
4 4 4 4
FIGURE 7.89 25 miles. FIGURE 7.90 16 miles. FIGURE 7.91 17 miles. FIGURE 7.92 23 miles.
high school. We complete the route by traveling to the kindergarten and then back
to the office. The resulting route, which is shown in Figure 7.83, is Office - Primary
school - Middle school - High school - Kindergarten - Office. This route is 18 miles
long.
The reader can easily verify that if we had chosen to go first to the high school
rather than the primary school, we would have gotten the route shown in
Figure 7.85. That route is 17 miles long. We note that both of these routes are
longer than the shortest route of 16 miles found in part a.
This example shows that the nearest-neighbor algorithm can fail to find a shortest
route.
primary school to the middle school. This is shown in Figure 7.93. Next there are
two edges of length 2, and we can choose both of them without violating either Rule
1 or Rule 2. The pieces thus far assembled are shown in Figure 7.94.
The next shortest edge is the 3-mile route joining the middle school to the high
school. But we can’t use this edge because it would make the circuit shown in Fig-
ure 7.95. Instead, we choose the 4-mile route from the middle school to the kinder-
garten, as shown in Figure 7.96.
Office Office
5 2 5 2
Kinder- 6 Primary Kinder- 6 Primary
garten school garten 2 school
7 2 7
7 1 8 7 1 8
4 4
3 High 3 High
school school
Middle school Middle school
Office Office
5 2 5 2
The next shortest edge is the 5-mile route from the kindergarten to the office. But
Figure 7.97 shows that this choice violates both Rule 1 and Rule 2. The only other
choice is the 7-mile route from the kindergarten to the high school, which results in
the Hamilton circuit shown in Figure 7.98. We note that this is the shortest possible
route, as we found in Example 7.9. But the cheapest link algorithm does not always
produce the shortest path, as the next example shows.
Office Office
5 2 5 2
500 CHAPTER 7
Minneapolis Minneapolis
9 96 9 96
5 5
161 Buffalo 161 Buffalo
Seattle Seattle
2619 2619
949
241 241
949
3 925 3 925
81
81
1263 1263
19
19
03 03
26 26
Memphis Memphis
1819 1819
SOLUTION
We begin by choosing the two shortest edges in the graph, as shown in Figure 7.100.
The next shortest edge is the 965-mile route from Minneapolis to Buffalo. But we
can’t use this route because it would result in the circuit shown in Figure 7.101. We
choose instead the next shortest edge, the 1263-mile route from San Diego to Seattle.
The edges from Seattle to Minneapolis and San Diego to Buffalo complete the route,
as shown in Figure 7.102.
Minneapolis Minneapolis
96 9 96
161
9 5 161 5
Seattle Buffalo Seattle Buffalo
2619 2619
949
949
241 241
3 925 3 925
81
81
1263 1263
19
19
26
03 03
26
Memphis Memphis
1819 1819
FIGURE 7.101 Adding the edge from FIGURE 7.102 The completed
Minneapolis to Buffalo makes a circuit. route.
This route has a total length of 925 + 949 + 1263 + 1619 + 2603 = 7359 miles.
This is longer than the shortest route of 6591 miles that we found earlier using the
nearest-neighbor algorithm (see Figure 7.79).
Crauder-4100237 QL November 10, 2014 12:9
700
1449
13 939
3 735
188
00
Phoenix 1127
Atlanta
101
4
242
Houston
The two algorithms that we have presented have the advantage of being simple,
and they provide a flavor of how algorithms work. But, in fact, they are of question-
able effectiveness in real situations. Much better (but more complex) algorithms are
available that provide practical approximations to the optimal solution.
502 CHAPTER 7
The figure on the left has no Euler circuit because it has vertices of odd degree.
The outer circuit is a Hamilton circuit. The “bow tie” in the figure on the right has
an obvious Euler circuit, but every circuit must visit the center vertex twice. Hence
there is no Hamilton circuit.
Further questions of this type may be found in the What Do You Think section of
exercises.
Santa Store
Staples Kay
Jewelers
Radio
Shack
Outback Linens’n
Things
Goody’s
Try It Yourself 7.8: Finding by hand a shortest route: Delivery truck The route
Bakery - Coffee shop - Convenience store - Restaurant - Bakery (or its reverse) has
the minimal length of 20 miles.
Try It Yourself 7.10: Applying the cheapest link algorithm: Air freight deliveries
The route below has a length of 4275 miles.
Chicago 436 Toronto
700
1449
13 939
3 735
188
00
Phoenix 1127
Atlanta
101
4
242
Houston
E
D
Park entrance
Quaker
Meeting Franklin
House Court
Geyser Liberty
Bell
Independence
Mountain Hall City
Carpenters Town
Hall
Falls Washington Square
FIGURE 7.106 A park. 9. Internet connections. The edges in Figure 7.110 represent
fiber-optic lines. Vertex A is an Internet service provider. The
6. A bakery. Figure 7.107 shows a bakery and several sites remaining vertices are homes that must be wired for an Inter-
where baked goods are to be delivered. Either find a route that net connection. Either find a route that begins and ends at the
begins and ends at the bakery and passes each store exactly provider location and visits each home exactly once, or show
once, or show that no such route exists. That is, either find a that no such route exists. That is, either find a Hamilton circuit
Hamilton circuit or explain why no such circuit exists. or explain why no such circuit exists.
Bakery
A
Office E
Coffee shop B
Mall Convenience
Sweet store
shop C D
504 CHAPTER 7
10. Tour of western states. The vertices in Figure 7.111 repre- 13. Holiday shopping. Figure 7.114 shows a shopping center.
sent western states. Vertices are connected by an edge if the two The vertices are stores, and the edges are sidewalks. We wish
states share a common border. We wish to take a scenic tour to visit each store for holiday shopping. Either find a route that
of the western United States. Either find a path that begins and begins and ends at T.J. Maxx and visits each store once or show
ends at Washington and visits each state exactly once, or show that no such route exists. That is, either find a Hamilton circuit
that no such path exists. That is, either find a Hamilton circuit or explain why no such circuit exists.
or explain why no such circuit exists.
T. J. Maxx Old Navy
Home Depot
MT ND Toys
WA ID “R” Us Music City
SD Dillards
OR WY
NE
NV CO Sears Pets “R” Us
KS Circuit
UT City
CA Radio Hastings
OK Target
NM Shack
AZ
TX FIGURE 7.114 Holiday shopping.
FIGURE 7.111 Tour of western states. 14. Checking in-home care patients. The edges in Figure 7.115
represent roads. The vertex marked “office” is the headquarters
11. Tour of southern states. The vertices in Figure 7.112 rep- of an in-home care service. The remaining vertices are homes
resent southern states. Vertices are connected by an edge if the where patients must be visited each day. Either find a route be-
two states share a common border. We wish to take a scenic ginning and ending at headquarters and visiting each patient
tour of the southern United States. Either find a path that be- home exactly once, or show that no such route exists. That is,
gins and ends at Missouri and visits each state exactly once, or either find a Hamilton circuit or explain why no such circuit
show that no such path exists. That is, either find a Hamilton exists.
circuit or explain why no such circuit exists. Office
KY
MO WV
Home Home
VA
TN
AR NC
Home
SC
LA
MS AL GA Home
Home
FL
FIGURE 7.115 Checking in-home
FIGURE 7.112 Tour of southern states. care patients.
12. Viewing holiday decorations. The edges in Figure 7.113 15. Meals on Wheels. The edges in Figure 7.116 represent
represent streets. Vertex A is your home, and the remaining roads. Meals on Wheels is an organization that provides meals
vertices are neighborhood homes that are displaying holiday for the elderly. The remaining vertices are homes of elderly peo-
decorations. Either find a path that begins and ends at your ple to which hot meals are to be delivered. Either find a path
own home and passes each neighborhood home exactly once, that begins and ends at the kitchen and passes each home ex-
or show that no such path exists. That is, either find a Hamilton actly once, or show that no such path exists. That is, either find
circuit or explain why no such circuit exists. a Hamilton circuit or explain why no such circuit exists.
A Meals on Wheels
F
Home
Home
Home
Home
B E
Home
C D Home
16. Routing a message. The vertices in Figure 7.117 represent 20. Either find a Hamilton circuit or show that none exists for
computers, and the edges represent connections. To check the the graph in Figure 7.121.
network setup, we want to send a test message that starts at
computer A, passes through each other computer exactly once,
and successfully returns to computer A. Either find a workable
path or show that no such path exists. That is, either find a
Hamilton circuit or explain why no such circuit exists.
A
F
B
E
FIGURE 7.121
8
B
34
Louisville 562
E Richmond
FIGURE 7.122
C D
FIGURE 7.118 22. Solve the traveling salesman problem for the map in Fig-
ure 7.123 by calculating the mileage for each possible route.
18. Either find a Hamilton circuit or show that none exists for Use New York as a starting point.
the graph in Figure 7.119.
A New York London
3459
F
2032
66 78 6003
E 54 01
B
C D Tucson 511
506 CHAPTER 7
24–29. Nearest-neighbor algorithm. In Exercises 24 through 28. Use the nearest-neighbor algorithm starting at vertex A of
29, use the nearest-neighbor algorithm. the map in Figure 7.129 to find an approximate solution of the
24. Use the nearest-neighbor algorithm starting at vertex A of traveling salesman problem.
the map in Figure 7.125 to find an approximate solution of the A
traveling salesman problem. 7
10
A 5 B
E 8
4 9
5
4 7 4 6 7
B
E 3
2
6 5 4 C
3 D
2
FIGURE 7.129
D 6 C
29. Use the nearest-neighbor algorithm starting at vertex A of
FIGURE 7.125 the map in Figure 7.130 to find an approximate solution of the
traveling salesman problem.
25. Use the nearest-neighbor algorithm starting at vertex A of
A
the map in Figure 7.126 to find an approximate solution of the
traveling salesman problem. 4
5
4 B
A E
2 3
4
5 2 5 3
6
4 B
E 2 3
D 8 C
2 2 5 5
FIGURE 7.130
6 C
D 30–35. Cheapest link algorithm. In Exercises 30 through 35,
FIGURE 7.126 use the cheapest link algorithm.
30. Use the cheapest link algorithm to find an approximate
26. Use the nearest-neighbor algorithm starting at vertex A of solution of the traveling salesman problem for the map in
the map in Figure 7.127 to find an approximate solution of the Figure 7.125.
traveling salesman problem.
31. Use the cheapest link algorithm to find an approximate
A solution of the traveling salesman problem for the map in
Figure 7.126.
4
5 32. Use the cheapest link algorithm to find an approximate so-
4 B
E
lution of the traveling salesman problem for the map in Fig-
7 1
ure 7.127.
5 5 3
6 33. Use the cheapest link algorithm to find an approximate
solution of the traveling salesman problem for the map in
D 4 C Figure 7.128.
34. Use the cheapest link algorithm to find an approximate
FIGURE 7.127
solution of the traveling salesman problem for the map in
Figure 7.129.
27. Use the nearest-neighbor algorithm starting at vertex A of
the map in Figure 7.128 to find an approximate solution of the 35. Use the cheapest link algorithm to find an approximate
traveling salesman problem. solution of the traveling salesman problem for the map in
Figure 7.130.
A 36. Nearest-neighbor failure. This exercise provides an exam-
5
4 ple like Example 7.9 where the nearest-neighbor algorithm
4 fails to find a shortest route. Consider the map shown in Fig-
B
E 3 ure 7.131.
2
a. Use the nearest-neighbor algorithm starting at A to find
2 5 3
6 an approximate shortest route. How long is the route you
found?
D 6 C b. By trial and error, find a shortest route. How does its
length compare with that in part a?
FIGURE 7.128
Crauder-4100237 QL November 10, 2014 12:9
508 CHAPTER 7
come in to play. What are some of these factors? Contact a local trucking company
and find out how they use efficient routes.
Exercise 10 provides additional information relevant to this topic.
W-6. Trial and error. A fellow student says that there’s no need to learn one of the
algorithms for the traveling salesman problem—trial and error will always work.
How would you respond?
W-7. Nearest-neighbor algorithm. What are the choices one makes in implementing
the nearest-neighbor algorithm? Use your answer to explain why this algorithm can
fail to find a shortest path—or even to give a unique result.
Exercise 43 provides additional information relevant to this topic.
W-8. Cheapest link algorithm. What are the choices one makes in implementing the
cheapest link algorithm? Use your answer to explain why this algorithm can fail to
find a shortest path—or even to give a unique result.
W-9. Complete graphs. The text states that the graphs to which the traveling sales-
man problem applies are complete graphs. Why do you think that is the case?
TAKE AWAY
FROM THIS IN THE NEWS 7.4 Search
E veryone who has an email account has probably received a forwarded chain
letter promising good luck if the message is forwarded on to others—or ter-
rible misfortune if it isn’t. The sheer volume of forwarded messages such as chain
letters, online petitions, jokes and other materials leads to a simple question—
Jason Koski/Cornell University Photography
Tomás Tomás
Luis Jane Phil Eduardo Mike Luis Jane Phil Eduardo Mike
The path typically taken by a chain letter or chain e-mail is a special kind of
graph known as a tree. Suppose, for example, that you send an e-mail to some of
your friends, who in turn forward it to some of their friends, and so on. The resulting
path of the chain e-mail might look like the picture in Figure 7.132.
Informally, such a graph resembles an upside-down tree, in the sense that it has
a starting vertex and the edges make branches that don’t grow back together. To
say that the branches don’t grow back together is to say that the graph contains no
circuits. If, for example, both Pat and Maria send the message to Eduardo, then the
resulting graph is not a tree because it contains the circuit shown in heavy edges in
Figure 7.133.
For the trees we consider, we always choose a starting vertex, typically drawn at
the top.
The following terminology is a whimsical mixture of botanical and familial terms
that are standard (we did not invent them). The vertex at the top of the tree is called
the root. For example, Tomás is the root of the e-mail tree, as shown in Figure 7.134
on the following page. There is a path leading from the root to any other vertex.
The vertex in the path just before a given vertex is called its parent, and any vertex
immediately after is its child. We see in Figure 7.134 that Maria is the parent of
Eduardo and Mike, so Eduardo and Mike are children of Maria. Vertices that are
not parents are leaves. These are typically at the bottom of the tree, as shown in
Figure 7.134.
Every vertex of a tree is either a parent or a leaf, as shown in Figure 7.135. This
gives us our first important formula for trees:
A tree also has levels, which are shown in Figure 7.136. The root is level 0. The
children of the root are level 1. The grandchildren are level 2, and so on. The largest
level of the tree is the height. Thus, the height of the tree in Figure 7.136 is 3.
Crauder-4100237 QL November 10, 2014 12:9
510 CHAPTER 7
FIGURE 7.134 Root, parents, and FIGURE 7.135 Vertices separated FIGURE 7.136 Levels of a tree.
children in trees. into parents and leaves.
Binary trees
Perhaps one of the most common trees we see in everyday life, and certainly the
one with the most hype, is the tree formed by the NCAA basketball tournament
brackets.6 A piece of the 2014 tree is shown in Figure 7.137. This particular type of
tree is known as a binary tree because each parent vertex has exactly two children.7
Dayton UCONN
Michigan UCONN
State
UCONN
UCONN
Arizona Wisconsin
Wisconsin
Kentucky
Michigan Kentucky
Kentucky
Because each parent in a binary tree has exactly two children, the number of
children is twice the number of parents. But the only vertex in a tree that is not a
child is the root. So, in a binary tree
6 In our discussion of this tournament, we ignore the “First Four” play-in games.
7 In some texts, a binary tree is one where parents have at most two children. If each parent has exactly two children,
those texts call the tree a full binary tree.
Crauder-4100237 QL November 10, 2014 12:9
SOLUTION
We represent the phone calls with a binary tree. Each vertex represents a PTA mem-
ber, and edges indicate phone calls. A part of this tree is shown in Figure 7.138.
The parent vertices in the tree are the PTA members who make phone calls. The
total number of vertices is the number of PTA members, 127. We can use this infor-
Member mation to find the number of parent vertices:
A
> KEY CONCEPT
A complete binary tree has all the leaves at the highest level.
In a complete binary tree, each parent has exactly two children, and the only
FIGURE 7.139 This binary
tree is not complete.
leaves are at the highest level. Hence, going from one level to the next doubles the
number of vertices. At level 0 there is 1 vertex, the root. There are 2 vertices at level
1, 4 at level 2, 8 at level 3, and so on (see Figure 7.140 on the following page). In
general, there are 2 L vertices at level L. Note that in a tournament bracket like the one
in Figure 7.141 on the following page, the leaves are the entrants in the tournament,
and the parent vertices represent games played.
Thinking of binary trees as tournaments is often useful. For example, there are
64 teams in the NCAA basketball tournament. Each game produces exactly one loser,
and every team loses except for the tournament champion. Hence, there are 63 games
played altogether. Because each game played represents a parent in the tournament
tree, there are 63 parents in the tree.
This reasoning allows us to count vertices in any complete binary tree. In a com-
plete binary tree of height H, there are 2 H vertices at the highest level. These are
the leaves. We think of the leaves as tournament entrants and the parents as games
Crauder-4100237 QL November 10, 2014 12:9
512 CHAPTER 7
1 = 20 at level 0 Parents:
Games
played
2 = 21 at level 1
4 = 22 at level 2
8 = 23 at level 3
Leaves: Tournament entrants
played. Each game produces one loser, and every entrant except one loses a game.
Hence,
Number of parents = Number of games played
= Number of losers
= Number of entrants − 1
Thus,
Number of parents = 2 H − 1
Now we can find the total number of vertices in a complete binary tree of
height H:
Number of vertices = 2 × Number of parents + 1
= 2(2 H − 1) + 1
= 2 × 2H − 2 + 1
Hence,
Number of vertices = 2 H+1 − 1
Spell checkers
Modern word processors come equipped with spell checkers that include a dictionary
of as many as 100,000 or so words, and most cell phones have a spell checker built
into “autocorrect.” When we type a word, the spell checker instantly compares that
Crauder-4100237 QL November 10, 2014 12:9
514 CHAPTER 7
word with the words in its dictionary and highlights the word if there is no match. Did
the spell checker instantly make 100,000 comparisons? No, even very fast computers
aren’t that fast. At the heart of the process is a binary tree. To illustrate the process,
let’s adopt a very simple dictionary containing only the following list of 13 one-letter
“words”:
A, C, E, G, I, K, M , O, Q, S, U, W, Y
We have put a box around M because it is the middle “word” in the list, and
it will be the root of our tree—the place where we start our search. If we type in
a letter, we will first compare it with M. If the letter we type happens to be M, we
have a valid word. If it precedes M in the alphabet, we move to the left in our list;
otherwise, we move to the right. These two possibilities correspond to two branches
from the root.
Now we consider the two branches. Again, we go to the middle to make our
comparison:
A, C, E, G, I, K, M , O, Q, S, U, W, Y
In the left branch A, C, E, G, I, K, there are two letters, E and G, that have equal
claim to the middle position. We adopt the convention that if there are two middle
Cartoonstock.com
Crauder-4100237 QL November 10, 2014 12:9
words, we will take the one on the left. Thus, we chose E to make our spell-checking
tree. In the right branch, there are also two letters that can claim the middle, and we
chose the one on the left. Thus, we use S. We go to the middle of each remaining
piece to make the next comparison.
The completed spell-checking tree is shown in Figure 7.142. At each stage, we
move to the right if the letter we typed is “bigger” and to the left if it is “smaller,”
and we report a valid word if it is the same. Note that the tree is binary except at A
and O. The path we would take if we typed in J, which is not in our list and so not
a valid word, is highlighted in Figure 7.143. Note that we need to make a total of
only four comparisons (not 13) to find out that J is not in our dictionary.
M M
J<M
E S E S
J>E
A I O W A I O W
J>I
C G K Q U Y C G K Q U Y
J<K
apple, boy, cat, dog, ear, fox, girl, hat, jet, kit, love, mom, net, out, pet
How many checks (or comparisons) are needed to test whether “cow” is a valid
word? Show the path you take on the spell-checking tree to check whether “cow” is
a valid word.
SOLUTION
We go to the middle of the list, which is “hat”:
apple, boy, cat, dog, ear, fox, girl, hat , jet, kit, love, mom, net, out, pet
apple, boy, cat, dog, ear, fox, girl, hat , jet, kit, love, mom, net, out, pet
We get “dog” and “mom.” These go on the first level of our tree. We continue to go
to the middle of each piece to get the next comparison word. The completed tree is
shown in Figure 7.144 on the following page.
Four checks are needed to spell-check “cow,” and because it matches no word in
the tree, we conclude that “cow” is not a valid word in our limited dictionary. The
path of the check is shown in Figure 7.145 on the following page.
Crauder-4100237 QL November 10, 2014 12:9
516 CHAPTER 7
hat hat
cow < hat
217 − 1 = 131,071
This means that we can check a word against a 100,000-word dictionary easily—we
need at most 17 checks. With only 3 more checks, we can accommodate a 1-million-
word dictionary. That is why spell checkers work so fast. We note that this kind of
binary branching is a formalized version of what we all do when we look up a word
in the dictionary.
a. You want to use a dictionary so that you have to check any word at most five
times. How big a dictionary can you accommodate?
b. Estimate the number of checks needed if you have a 10,000-word dictionary.
Crauder-4100237 QL November 10, 2014 12:9
SOLUTION
a. We know that with at most c checks allowed, a dictionary of 2c − 1 words can
be accommodated. We want 5 checks, so we use c = 5 in this formula. We calculate
that 25 − 1 equals 31. This means that we can accommodate a dictionary with 31
words.
b. We want to choose the smallest c so that 2c − 1 is at least 10,000. We proceed
by trial and error (with the help of a calculator) and find that 214 − 1 is greater than
10,000 but 213 − 1 is not. Thus at most, 14 checks are needed.
hit
dip man
518 CHAPTER 7
dip man
it < man
it < joy
each time two silver dollars came out. This pattern continued We make a typing tree associated with this list as shown in the
through the fifth night, after which I took my winnings and figure below.
went home.
a. How much money did I take home? e
520 CHAPTER 7
A A
B C B C
D E F D E F
This search tree will either produce a tree including all the ver-
We want to determine whether a graph represented by an in-
tices (as it did here) or terminate without including all the ver-
cidence matrix is connected, but we don’t want to reproduce
tices. In the first case, the graph is connected, and in the second
the entire graph. The method we present is known as a tree
case, it is not.
search because it employs the construction of a tree. We will
For the following incidence matrix, make a search tree
show how it works using the preceding example. Begin with
starting at the vertex A to determine whether the graph rep-
any vertex you like, say, A, and start the construction of a tree
resented is connected.
by listing the vertices adjacent to A. We get the list B, C. The
partial tree associated with this list is shown below. A B C D E F G H I J
⎛ ⎞
A − 1 1 0 0 0 0 0 0 0
A B ⎜1 − 1 1 1 1 0 0 0 0⎟
⎜1 1 − 0⎟
C ⎜ 1 0 0 1 0 0 ⎟
D ⎜0 1 1 − 0 1 0 0 0 0⎟
⎜ ⎟
B C E ⎜0 1 0 0 − 0 0 0 0 0⎟
⎜ ⎟
F ⎜0 1 0 1 0 − 1 0 0 0⎟
⎜ ⎟
Note that we do not include the edge joining B to C. We are G ⎜0 0 1 0 0 1 − 0 0 0⎟
⎜ ⎟
interested only in connecting to new vertices, not what connec- H ⎜0 0 0 0 0 0 0 − 1 1⎟
⎝ ⎠
tions may or may not exist among vertices already listed. Next I 0 0 0 0 0 0 0 1 − 1
we go to the first vertex of this list. We look only for new (other J 0 0 0 0 0 0 0 1 1 −
than A, B, or C) vertices adjacent to B. We find the new vertices
D and E and thus obtain the following partial tree: 27. History. Minimal-cost spanning trees are very important in
many applications. A number of efficient algorithms have been
A
developed to find them. Report on the development of methods
for solving this problem.
B C The following exercise is designed to be solved using technol-
ogy such as a calculator or computer spreadsheet.
28. A very large dictionary. What size dictionary can be accom-
modated by a spell checker that uses 33 checks? What size dic-
D E tionary can be accommodated by a spell checker that uses 100
checks? We know names for numbers like millions, billions, or
Finally, we look for new vertices adjacent to C, and we find trillions. Do you have a name for the number of words you
only F . Our final tree is shown below. found for 100 checks?
W-5. Social Security. In a recent U.S. presidential primary, some candidates referred
to Social Security as a Ponzi scheme. Is this an accurate criticism of Social Security?
You may need to investigate carefully how Social Security works.
W-6. NCAA basketball tournament. Some complain that too many teams are left
out of the NCAA basketball tournament. How would the number of teams included
change if there were one additional round of games? What if there were two addi-
tional rounds? Three?
Exercise 21 provides additional information relevant to this topic.
W-7. Number of paths. How many different paths are there from the root of a tree
to a given vertex? Suggestion: Look again at the definition of a tree.
W-8. Complete binary trees. Is it easier to count the number of leaves in a complete
binary tree than in a binary tree that is not complete? Why or why not?
W-9. Root. Consider a graph that contains no circuits. Could any vertex be used as
the root for the corresponding tree? Explain why in some situations one choice of
root might make more sense than another.
CHAPTER SUMMARY
In this chapter, a graph is understood to be a collection of points joined by a number
of line segments. The points are called vertices, and the segments connecting them
are called edges. A highway map is a typical example of a graph.
522 CHAPTER 7
method used to find approximate solutions of the traveling salesman problem is the
cheapest link algorithm. Algorithms that are much better than these two are avail-
able, but the better algorithms are more complex.
KEY TERMS
circuit (cycle), p. 474 complete graph, p. 495 binary tree, p. 510
Euler circuit (Euler cycle), p. 474 traveling salesman problem, p. 495 complete binary tree, p. 511
degree (valence), p. 474 nearest-neighbor algorithm, p. 496
Hamilton circuit, p. 490 tree, p. 509
CHAPTER QUIZ
1. For the graph in Figure 7.147, find the degree of each vertex. Answer No Euler circuit exists because some of the vertices
have odd order.
A
F If you had difficulty with this problem, see Example 7.3.
3. The edges in Figure 7.149 represent streets that must be
swept, and the vertices represent intersections. Is there a route
B E
for the street sweeper that will not require traversing already
clean streets? In other words, is there an Euler circuit for the
graph in Figure 7.149? If an Euler circuit exists, find one.
B
C D
C F
Answer A: 3; B: 2; C: 3; D: 3; E: 2; F : 3.
If you had difficulty with this problem, see Example 7.2.
B C F
F
C A
E D
D
FIGURE 7.148 Inspecting electrical lines. If you had difficulty with this problem, see Example 7.5.
Crauder-4100237 QL November 10, 2014 12:9
4. For the graph in Figure 7.150, either find a Hamilton circuit Answer Oklahoma City - Ft. Smith - Little Rock - Dallas -
or explain why no such circuit exists. Oklahoma City; distance: 869 miles.
If you had difficulty with this problem, see Example 7.8.
6. There are 31 members of an organization. The plan for dis-
seminating news requires one member to contact two others
with the information. Each member who receives the news con-
tacts two members who have not yet gotten the news. This
continues until all members have the information. Use the
formula
Number of vertices = 2 × Number of parents + 1
FIGURE 7.150 Hamilton circuit? to determine how many members contact others.
Answer 15.
Answer One Hamilton circuit is If you had difficulty with this problem, see Example 7.11.
7. Make a spell-checking tree for the following dictionary:
aardvark, bear, cat, deer, emu, fox, giraffe, hound, impala,
javelina, krill
How many checks are needed to test whether “dove” is a
valid word?
Answer For “dove”: 4 checks.
fox
impala
If you had difficulty with this problem, see Example 7.7. cat
WA
12 ME
VT 3 4
MT ND
3 3
OR MN
7 ID NH 4
10
4 WI NY MA 11
SD 29
10 MI
WY 3
16 RI 4
3 CT 7
IA PA
NV NE 6 20 NJ 14
6 OH
UT 5 IN DE 3
IL 18
6 11
CA CO 20 WV
MO 5 VA MD 10
55 9 KS KY
10 13 D.C. 3
6 8
NC
TN 15
OK 11
AZ AR SC
NM 7
11 6 9
5
AL GA
MS 9 16
6
TX LA
38 8
FL
Freedom’s Lighthouse
AK 29
3 270 Needed
HI
4 Obama – 332
Romney – 206
Crauder-4100237 QL November 10, 2014 13:45
V
oting rules and their consequences are not always as straightforward as one
might first imagine, especially when some participants may have more than one
vote. A prime example is the U.S. Electoral College. In electing a president of
the United States, each state receives a number of electoral votes equal to the number
of representatives and senators from that state. If a candidate wins the majority of the
8.1 Measuring voting popular vote in a state, the candidate normally receives all the electoral votes from
power: Does my vote that state (the current exceptions are Maine and Nebraska). The following article
count? from CBSNews.com concerns the controversy arising from this system.
F or the first time since the historic 2000 presidential election, a majority of
Republicans now favor adopting a popular vote system to elect the president
of the United States over the Electoral College, a new Gallup poll shows. Nearly
two-thirds of Americans overall would prefer the popular vote.
As many as 62 percent of Americans say they would amend the Constitution
to replace the Electoral College, according to the poll, conducted October 6–9.
Just 35 percent say they would keep the current system.
To win the presidency, a candidate must win 270 out of 538 Electoral College
votes. Each state is granted a number of Electoral College delegates that equals
its number of U.S. senators and representatives. States are allowed to determine
how their delegates will vote—typically, whichever candidate wins the popular
vote in a state gets all of that state’s electoral votes. The candidate who wins the
popular vote nationwide does not always win the Electoral College.
In the 2000 presidential election, Democrat Al Gore won the popular vote,
but Republican George W. Bush won the Electoral College. In mid-December of
that year, the Supreme Court ended a vote recount in Florida, handing Bush that
state’s 25 electoral votes and the presidency.
When Gallup surveyed Americans about their support for the Electoral Col-
lege after the Supreme Court decision, 75 percent of Democrats said they’d favor
a popular vote, but just 41 percent of Republicans favored the idea. A majority
of Republicans, 56 percent, wanted to keep the Electoral College.
Republicans have grown more interested in a popular vote over the past
decade, but this is the first year a majority have supported the idea.
Supporters of the popular vote have been working for years to make the
Electoral College system obsolete—even if that means bypassing Congress and
the amendment process. The National Popular Vote Campaign is coordinating
state lawmakers to promote the popular vote.
525
Crauder-4100237 QL November 10, 2014 13:45
526 CHAPTER 8
Questions of how a group should make decisions arise not only in weighty situa-
tions such as the election of a president but also in everyday situations, for example,
choosing which movie a group of friends will rent for the night. Other situations
in which making a decision is difficult come up in the context of equitable divi-
sion. Typical examples include inheritance and property settlements associated with
divorces. Determining how to divide assets fairly is often difficult and sometimes
contentious.
A particularly sticky division problem comes up when delegates to the House
of Representatives are assigned. The issue of how many representatives each state
should get was a problem when our country was founded, and it remains an area of
controversy today. We consider all of these topics in this chapter.
This article refers to political coalitions and to a “powerful voting bloc.” One
question that arises is how to measure the relative power of voters or voting blocs.
Voting alliances or coalitions play an important role here. It may seem that a voter’s
power should be proportional to the number of his or her votes, but that is not
the case. Just because one entity (a person, state, or coalition) has more votes than
Crauder-4100237 QL November 10, 2014 13:45
another does not mean that entity has more power. In this section, one way we will
measure voting power is in terms of coalitions of voters.
Here’s an example to illustrate the importance of voters joining together to form
an alliance or coalition. Suppose the board of directors of ACE Computer Corpora-
tion has three members. Ann controls 19 shares, Beth controls 18 shares, and Cate
controls only 3 shares. They want to elect a chairperson, and the election requires a
simple majority of the voting shares. Because there are 40 shares altogether, 21 shares
are required for a majority.
How powerful are these voters? It might seem on the surface that Ann is the most
powerful but that Cate doesn’t have much power at all because she has only 3 shares
out of a total of 40. A good case can be made, however, that all three members have
equal power. Let’s see how to make that case.
© Grantland Enterprises; www.grantland.net
None of the members has enough shares (21) to win. The only way for any mem-
ber to be elected is to convince one of the other members to throw support to her.
If Cate can convince Ann to support her, she will have 3 + 19 = 22 votes, so she
will be elected. If Cate can convince Beth to support her, she will have 3 + 18 = 21
votes, so she will be elected. Thus, if Cate can get the support of either Ann or Beth,
then she will be elected, even though she controls only 3 shares. Similarly, if Beth can
convince either Cate or Ann to support her, she will win. This observation suggests
that all three members have equal power.
This example illustrates that voting power is not simply proportional to the num-
ber of votes a person or bloc has. Coalitions play an important role in voting power,
and we now focus on the question of measuring voting power in terms of coalitions.
1 “Weighted Voting Doesn’t Work: A Mathematical Analysis,” Rutgers Law Review 19 (1965), 317–343.
2 In the real-life situation addressed by Banzhaf in his original paper, votes were allocated among six district
supervisors, but three of the supervisors had no power because their votes were never necessary to determine the
outcome. See Exercises 43 and 44 for the details.
Crauder-4100237 QL November 10, 2014 13:45
528 CHAPTER 8
In many cases, the quota is just a simple majority, which requires more than
half of the total number of votes cast. For instance, in the previous example there
were 40 votes altogether, so the quota for a simple majority is 21. In the case of
a cloture vote in the U.S. Senate, however, the quota is 60 votes out of 100. To
ratify a constitutional amendment in the United States requires (among other things)
approval by three-quarters of the states, which makes a quota of 38 states.
In this section, you should assume (unless you are instructed otherwise) that the
quota is a simple majority. The methods we use are unaffected by the choice of
quota. Changing the quota can, however, affect the power of voters. See Exercises
24 through 26. National Archives and Records Administration
Article V
The Congress, whenever two thirds of both Houses shall deem it necessary, shall propose Amendments to this
Constitution, or, on the Application of the Legislatures of two thirds of the several States, shall call a Convention
for proposing Amendments, which, in either Case, shall be valid to all Intents and Purposes, as Part of this
Constitution, when ratified by the Legislatures of three fourths of the several States, or by Conventions in three
fourths thereof, as the one or the other Mode of Ratification may be proposed by the Congress; Provided that no
Amendment which may be made prior to the Year One thousand eight hundred and eight shall in any Manner
affect the first and fourth Clauses in the Ninth Section of the first Article; and that no State, without its
Consent, shall be deprived of its equal Suffrage in the Senate.
Crauder-4100237 QL November 10, 2014 13:45
In the example involving the ACE Corporation, the winning coalitions consist of
any two voters or all three. Here is another example.
Number of votes
Total Winning
4 3 1 votes coalition?
Abe Ben Condi 8 Yes
Abe Ben 7 Yes
Abe Condi 5 Yes
Abe 4 No
Ben Condi 4 No
Ben 3 No
Condi 1 No
Now we focus attention on the voters who are crucial components of winning
coalitions.
Crauder-4100237 QL November 10, 2014 13:45
530 CHAPTER 8
Let’s determine the critical voters for each of the three winning coalitions in Ex-
ample 8.1. These winning coalitions appear in the top three rows of the table.
Number of votes
Total Winning
4 3 1 votes coalition?
Abe Ben Condi 8 Yes
Abe Ben 7 Yes
Abe Condi 5 Yes
Abe, Ben, Condi (8 total votes): If Abe is removed, the remaining coalition controls
only 4 votes—not a majority. So, removing Abe changes the coalition from a winning
one to a losing one. Therefore, Abe is a critical voter for this coalition.
Removing either Ben or Condi from the coalition leaves at least 5 votes controlled
by the remaining coalition, so the remaining coalition is still a winning coalition.
Therefore, neither Ben nor Condi is a critical voter in this coalition.
Abe, Ben (7 total votes): Removing either voter makes it a losing coalition, so both
Abe and Ben are critical for this coalition.
Abe, Condi (5 total votes): As with the preceding coalition, removing either voter
makes it a losing coalition. So, both Abe and Condi are critical for this coalition.
We summarize all of this information in a coalition table (see Table 8.1). It lists
the critical voters.
Number of votes
Total Winning Critical
4 3 1 votes coalition? voters
Abe Ben Condi 8 Yes Abe
Abe Ben 7 Yes Abe, Ben
Abe Condi 5 Yes Abe, Condi
Abe 4 No Not applicable
Ben Condi 4 No Not applicable
Ben 3 No Not applicable
Condi 1 No Not applicable
Counting coalitions
To make a coalition table, we had to list all the different coalitions of voters. Suppose
there are 7 voters. How many different coalitions are possible? Each of the 7 voters
is either in the coalition or not in it, so there are 27 possibilities for voters to be
either in or not in a coalition.3 This includes all the voters not being in, which isn’t
a coalition, so there are 27 − 1 coalitions possible. In general, if there are n voters,
then there are 2n − 1 possible coalitions. In the case of voters Abe, Ben, and Condi
that we considered in Example 8.1, the number of voters is n = 3, so there should be
23 − 1 = 7 coalitions—just as indicated in the coalition table, Table 8.1. Of course,
for large collections of voters, it is impractical to list all the possible coalitions.
We illustrate this definition by computing the Banzhaf index for each voter in the
situation we have been considering.
532 CHAPTER 8
Let’s make a couple of observations based on the preceding example. First, note
that even though Abe has only one more vote than Ben, he has three times the voting
power of Ben (as measured by the Banzhaf index). Also note that Ben has 3 votes to
Condi’s 1 vote, but they have the same voting power. Once again, this indicates that
simply knowing how many votes a voter has does not really tell us much about his
or her voting power. The Banzhaf index gives us a way to measure the real power of
voters.
Total Winning
666.25 201.75 195.75 90.25 votes coalition?
R S O G 1154 Yes
R S O 1063.75 Yes
R S G 958.25 Yes
R S 868 Yes
R O G 952.25 Yes
R O 862 Yes
R G 756.5 No
R 666.25 No
S O G 487.75 No
S O 397.5 No
S G 292 No
S 201.75 No
O G 286 No
O 195.75 No
G 90.25 No
c. The accompanying table lists only the winning coalitions along with the
critical voting blocs in each case.
d. To calculate the Banzhaf index of a voter, we divide the number of times that
voter is critical in some winning coalition by the total number of instances of
critical voters. In this case, there are 10 instances in which a voting bloc is
critical. This means that to find the Banzhaf index of a voting bloc, we first find
the number of times that bloc is critical and then divide by 10.
Let’s do the computation for each bloc:
Roosevelt is critical in 6 of these 10 instances. Therefore, this Banzhaf index is
6/10 = 60%.
Smith is critical in 2 of these 10 instances. Therefore, this Banzhaf index is
2/10 = 20%.
“Other” is critical in 2 of these 10 instances. Therefore, this Banzhaf index is also
2/10 = 20%.
Garner is not critical in any of these 10 instances. Therefore, this Banzhaf index is
0%.
Crauder-4100237 QL November 10, 2014 13:45
534 CHAPTER 8
The calculations in Example 8.3 show that, as measured by the Banzhaf index,
the most powerful candidate by far was Roosevelt. They also show that although
Smith and “Other” had different numbers of votes, they had the same power.
EXAMPLE 8.4 Finding the swing voter: Council of the European Union
Member states of the European Union have votes on the Council assigned by the
2001 Treaty of Nice. The number of votes is roughly determined by a country’s
population but progressively weighted in favor of smaller countries. Ireland has 7
votes, Cyprus has 4 votes, and Malta has 3 votes. Suppose that these three countries
serve as a committee where a simple majority is required to win, so the quota is 8
votes. Make a table listing all the permutations5 of the voters and the swing voter in
each case.
SOLUTION
There are n! different permutations6 of n objects, where n! = n × (n − 1) × · · · × 1.
Our three objects are Ireland, Cyprus, and Malta and 3! = 3 × 2 × 1 = 6, so there
Flags of the European Union. will be six rows in our table. We let the first row represent the voting order Ireland,
4 Shapley and Shubik are prominent game theorists who were portrayed in the book and movie A Beautiful Mind. See
their paper “A Method for Evaluating the Distribution of Power in a Committee System,” American Political Science
Review 48 (1954), 787–792.
5 Recall from Section 4 of Chapter 5 that a permutation of objects is just an arrangement of them in a certain order.
6 For more information on permutations, see Section 4 of Chapter 5.
Crauder-4100237 QL November 10, 2014 13:45
Cyprus, Malta. Then Ireland, with 7 votes, doesn’t have a majority, but the 4 votes
of Cyprus added to Ireland’s do give a majority. Therefore, Cyprus is the swing voter
for this order, and we indicate this in the last column of the first row.
If we continue in this way for each row, we obtain the following table:
Note that in Example 8.4, the second voter is not always the swing voter. The third
voter is the swing in the fourth and sixth rows of the table. Note, too, that Ireland is
the swing voter 4 out of 6 times. This proportion is the key to the measurement we
will now introduce.
536 CHAPTER 8
make 20, but Wisconsin’s 10 added to it exceeds 20. Therefore, Wisconsin is the
swing voter in this order. Looking at the fifth row, we see that Virginia’s 13
electoral votes plus Iowa’s 6 are not enough to make 20, but Wisconsin’s 10 added
to these exceeds 20. Therefore, Wisconsin is the swing voter in this order too.
States Swing
VA (13) WI (10) CO (9) IA (6) WI
VA (13) WI (10) IA (6) CO (9) WI
VA (13) CO (9) WI (10) IA (6) CO
VA (13) CO (9) IA (6) WI (10) CO
VA (13) IA (6) WI (10) CO (9) WI
VA (13) IA (6) CO (9) WI (10) CO
WI (10) VA (13) CO (9) IA (6) VA
WI (10) VA (13) IA (6) CO (9) VA
WI (10) CO (9) VA (13) IA (6) VA
WI (10) CO (9) IA (6) VA (13) IA
WI (10) IA (6) VA (13) CO (9) VA
WI (10) IA (6) CO (9) VA (13) CO
CO (9) VA (13) WI (10) IA (6) VA
CO (9) VA (13) IA (6) WI (10) VA
CO (9) WI (10) VA (13) IA (6) VA
CO (9) WI (10) IA (6) VA (13) IA
CO (9) IA (6) VA (13) WI (10) VA
CO (9) IA (6) WI (10) VA (13) WI
IA (6) VA (13) WI (10) CO (9) WI
IA (6) VA (13) CO (9) WI (10) CO
IA (6) WI (10) VA (13) CO (9) VA
IA (6) WI (10) CO (9) VA (13) CO
IA (6) CO (9) VA (13) WI (10) VA
IA (6) CO (9) WI (10) VA (13) WI
d. Virginia is the swing in 10 cases, Wisconsin and Colorado are the swing in 6
cases each, and Iowa is the swing in 2 cases. Therefore, the power indices are
Virginia: 10/24 or about 41.67%, Wisconsin: 6/24 or 25%, Colorado: 6/24 or
25%, and Iowa: 2/24 or about 8.33%.
It is interesting to note that in this example, Virginia has about 2.2 times the
number of votes that Iowa has, but it has 5 times the voting power as measured by
the Shapley–Shubik index.
538 CHAPTER 8
Number of votes
Total Winning
5 3 3 votes coalition?
Abe Ben Condi 11 Yes
Abe Ben 8 Yes
Abe Condi 8 Yes
Abe 5 No
Ben Condi 6 Yes
Ben 3 No
Condi 3 No
Crauder-4100237 QL November 10, 2014 13:45
Try It Yourself 8.2: Computing the Banzhaf index: ACE Computer Corporation
Number of votes
Total Winning Critical
19 18 3 votes coalition? voters
Ann Beth Cate 40 Yes None
Ann Beth 37 Yes Ann, Beth
Ann Cate 22 Yes Ann, Cate
Ann 19 No Not applicable
Beth Cate 21 Yes Beth, Cate
Beth 18 No Not applicable
Cate 3 No Not applicable
540 CHAPTER 8
9. Two voters again. Suppose there are only two voters. Anne 26. Refer to Exercises 24 and 25. Does Condi’s voting power
controls 152 votes, and the quota is 141 votes. What is Anne’s increase or decrease with the change from a simple majority to
Shapley–Shubik index? What is the Shapley–Shubik index of a two-thirds majority?
the other voter?
27–31. Five voters. Suppose there are five voters, A, B, C, D,
10. One person, one vote. In a group of 100 voters, each per- and E, with 6, 5, 4, 3, and 2 votes, respectively. Suppose that a
son has 1 vote, and the quota is a simple majority. What is the simple majority is required to win. This information is used in
Banzhaf index for each voter? Hint: Each voter should have the Exercises 27 through 31.
same index.
27. What is the quota?
11. One person, one vote again. In a group of 100 voters, each 28. How many permutations of five voters are there?
person has 1 vote, and the quota is a simple majority. What
is the Shapley–Shubik index for each voter? Hint: Each voter 29. How many coalitions are there?
should have the same index. 30. Fill out the accompanying coalition table.
12–16. Three delegates. Suppose there are three delegates to
a county convention. Abe has 100 votes, Ben has 91 votes, Votes
Total Winning Critical
and Condi has 10 votes. A simple majority wins. Exercises 12 6 5 4 3 2 votes coalition? voters
through 16 refer to this situation.
A B C D E
12. What is the quota? A B C D
A B C E
13. Make a table listing all of the coalitions and the critical A B C
voter in each case. A B D E
14. Find the Banzhaf index for each voter. A B D
A B E
15. Make a table listing all of the permutations and the swing A B
voter in each case. A C D E
A C D
16. Find the Shapley–Shubik index for each voter. A C E
A C
17–23. Four voters. Suppose there are four voters: A with 13 A D E
votes, B with 6 votes, C with 5 votes, and D with 2 votes. Sup- A D
pose that a simple majority is required to win. Use this infor- A E
mation for Exercises 17 through 23. A
B C D E
17. What is the quota? B C D
18. How many coalitions are there? B C E
B C
19. Make a coalition table. B D E
B D
20. Find the Banzhaf index for each voter. B E
B
21. How many permutations of four voters are there?
C D E
22. Make a table listing all of the permutations of the voters C D
and the swing voter in each case. C E
C
23. Find the Shapley–Shubik index for each voter. D E
D
24–26. Ace Solar. Suppose there are four stockholders in a E
meeting of the Ace Solar Corporation and that each person
gets as many votes as the number of his or her shares. Assume 31. Using the coalition table from Exercise 30, find the Banzhaf
that Abe has 49 shares, Ben has 48 shares, Condi has 4 shares, index for each voter.
and Doris has 3 shares. Exercises 24 through 26 refer to this
situation. 32. Four states. This exercise refers to Example 8.6, which dis-
cusses four swing states in the 2012 U.S. presidential election.
24. Assume that a simple majority is required to prevail in a Recall that Virginia had 13 electoral votes, Wisconsin had 10,
vote. Make a table listing all of the permutations of the voters Colorado had 9, and Iowa had 6. Assume again that a simple
and the swing voter in each case, and calculate the Shapley– majority vote from only these four states would determine the
Shubik index for each voter. election, so the quota is 20. Find the Banzhaf index for each of
these four states.
25. Assume now that a two-thirds majority is required to pre-
vail in a vote, so the quota is 70. Calculate the Shapley–Shubik 33. Party switch. In 2001, the U.S. Senate had 50 Republicans
index for each voter. and 50 Democrats. On May 24 of that year, James Jeffords,
Crauder-4100237 QL November 10, 2014 13:45
a Republican from Vermont, announced his intention to leave is, determine how many votes should be allotted to each of
the Republican Party and declare himself an Independent. That the remaining four permanent members and what the quota
made the Senate consist of 49 Republicans, 50 Democrats, and should be. (Assign 1 vote to each rotating member.) Note: The
1 Independent. For a simple majority vote, make a table listing Shapley–Shubik index for this weighted voting system can be
all of the permutations of the three voting blocs and the swing calculated using a computer program. The result is 23.25%
voter in each case. What is the Shapley–Shubik index for each for each permanent member and 0.70% for each rotating
voting bloc? member.
34. Party switch 2. Repeat Exercise 33, this time making a Exercises 41 and 42 may be suitable for group work.
coalition table. What is the Bahzhaf index for each voting bloc? 41. A research project. Voting coalitions are a vital part of par-
35–36. Electoral College. Use the following information liamentary systems. A nice example is the Israeli Knesset, where
for Exercises 35 and 36. Suppose that there are only three there are several parties and coalitions often are necessary for
hypothetical states with a distribution of popular and elec- forming a government. Write a report on the Knesset or other
toral votes as shown in the accompanying table. (The num- parliaments. Focus on coalitions.
ber of electoral votes for a state is the size of its congressional 42. A research project. At national party conventions, state
representation.) delegations usually vote for the presidential nominee in
alphabetical order. Find other situations in which votes are
taken in a particular order.
State A State B State C Total
43–44. History. Use the following information for Exercises
Population 100 200 400 700
43 and 44. When John Banzhaf introduced his index of voting
Senators 2 2 2 6
Representatives 1 2 4 7 power in 1965, he analyzed the voting structure of the Nassau
Electoral votes 3 4 6 13 County Board of Supervisors. (Nassau County is located in the
western part of Long Island, New York.) That board had di-
vided 115 votes among the districts based on population. The
35. Find the Shapley–Shubik index for each state using the following table shows the vote allocation based on the popu-
popular vote and then using the electoral vote. Assume that lation in 1964. In the table, Hempstead #1 represents the pre-
a simple majority is required in each case. siding supervisor, always from Hempstead, and Hempstead #2
36. Use your answers to Exercise 35 to explain why states with represents the second supervisor from Hempstead.
small populations resist changing the Electoral College system.
37–40. The Security Council. Use the following informa- District Number of votes
tion for Exercises 37 through 40. The United Nations Secu-
rity Council has five permanent members (China, France, Great Hempstead #1 31
Hempstead #2 31
Britain, Russia, and the United States) and 10 rotating mem-
Oyster Bay 28
bers. Each member has 1 vote, and 9 votes are required to pass North Hempstead 21
a measure. But each permanent member has a veto. That is, if Glen Cove 2
a permanent member votes “no,” the measure fails. Although Long Beach 2
this is not a true weighted voting system, it is equivalent to one.
37. Suppose that each rotating member has 1 vote and that
each permanent member has 7 votes, for a total of 45 votes. If 43. Assume that a simple majority is required to win a vote.
the quota is set at 39 votes, explain why a measure cannot pass Explain why North Hempstead, Glen Cove, and Long Beach
if a permanent member votes “no.” have zero power (in the sense that their votes are never neces-
sary to determine the outcome of an election). Do not answer
38. Consider the weighted system in Exercise 37. Explain why this question by calculating the Banzhaf index.
a measure will pass if nine members of the council, including
all of the permanent members, support it. 44. Even though the number of votes was allocated according
to the size of a district, clearly voting power was not allocated
39. Explain why the weighted voting system described in Ex- that way. This voting system was modified several times over
ercise 37 is equivalent to the actual United Nations Security the years because of legal challenges. Investigate this history
Council voting system. Note: The Shapley–Shubik index for and write a brief summary, including the final resolution.
this weighted voting system can be calculated using a computer
program. The result is 19.63% for each permanent member 45–46. History. Exercises 45 and 46 refer to the article by
and 0.19% for each rotating member. Shapley and Shubik in which they first defined their index.
40. In Exercises 37 through 39, we failed to mention some- 45. The following quotation is from the introductory portion
thing: In reality, a permanent member of the Security Coun- of the Shapley–Shubik article. “It is possible to buy votes in
cil may abstain, which does not count as a veto. Suppose most corporations by purchasing common stock. If their poli-
France abstains and no one else does. Then 9 votes, includ- cies are entirely controlled by simple majority votes, then there
ing those of all four of the remaining permanent members, are is no more power to be gained after one share more than 50%
required to pass a measure. Devise an equivalent weighted vot- has been acquired.” How does this observation support the im-
ing system for these 14 members, as was done previously. That portance of the concept of a swing voter?
Crauder-4100237 QL November 10, 2014 13:45
542 CHAPTER 8
46. Here is another quotation from the introductory portion c. Compare the results given by your spreadsheet with
of the Shapley–Shubik article. “Put in crude economic terms, those given by the software mentioned in Calculation
. . . if votes of senators were for sale, it might be worthwhile Tip 8.1.
buying forty-nine of them, but the market value of the fiftieth 50. Four voters.
(to the same customer) would be zero.” How do the numbers a. Produce a spreadsheet that computes the Banzhaf index
49 and 50 in the statement reflect the fact that the paper was in the case when there are four voters. The spreadsheet
published in 1954? How would you update the statement to be should allow the user to enter the number of votes con-
true now? trolled by each voter, as well as the quota.
47–48. History 2. Exercises 47 and 48 refer to the following b. Solve Example 8.3 (on the 1932 Democratic National
quotation from the article by Shapley and Shubik in which they Convention) using your spreadsheet.
first defined their index.
c. Experiment with different vote totals for the four blocs
To illustrate, take Congress without the provision for (Roosevelt, Smith, Garner, and “Other”) in Example 8.3
overriding the President’s veto by means of two-thirds and report how the power index varies.
majorities. This is now a pure tricameral system with
d. Compare the results given by your spreadsheet with
chamber sizes of 1, 97, and 435. The values come out
those given by the software mentioned in Calculation
to be slightly under 50% for the President, and approx-
Tip 8.1.
imately 25% each for the Senate and House, with the
House slightly less than the Senate. 51. Paradox. Attempting to measure voting power can give rise
to paradoxes.7 Two voting schemes, each with a total of 11
votes, are given. Assume that a two-thirds majority is required
47. Note that the figure of 97 refers to the Senate. Explain why
to win. Express each value of the Banzhaf index as a percentage
it was valid in 1954 (when the paper was published). Hint:
rounded to two decimal places.
Consider the role of the vice president.
48. Does it make sense that the president’s power index is a. How many votes are required to win?
nearly 50%? Explain. b. Suppose that A has 5 votes; B has 3 votes; and C, D,
The following exercises are designed to be solved using tech- and E have 1 vote each. Use a computer program to find
nology such as calculators or computer spreadsheets. the Banzhaf index for each voter.
49. Three voters. c. Suppose that, instead of the votes in part b, A gives away
a. Produce a spreadsheet that computes the Banzhaf index 1 vote to B. So A has 4 votes; B has 4 votes; and C, D, and
in the case when there are three voters and the quota is a E still have 1 vote each. Use a computer program to find
simple majority. The spreadsheet should allow the user to the Banzhaf index for each voter.
enter the number of votes controlled by each voter. Do the d. Did the Banzhaf index for A increase or decrease after
same for the Shapley–Shubik index. giving away 1 vote? Does this make sense?
b. Use your spreadsheet to check the computations of the e. Explain why C, D, and E have no voting power in the
Shapley–Shubik index for some of the previous examples voting scheme of part c. Use this observation to compute
and exercises. by hand your answers to part c.
W-2. Shapley–Shubik and order. A fellow student asserts, “The Shapley–Shubik in-
dex depends on the order in which the voters vote.” Is that an accurate statement?
Explain.
Exercise 23 provides additional information relevant to this topic.
7 See Dan S. Felsenthal and Moshe Machover, The Measurement of Voting Power: Theory and Practice, Problems and
Paradoxes (Cheltenham, UK: Edward Elgar, 1998).
Crauder-4100237 QL November 10, 2014 13:45
W-3. Secret ballot voting. The Shapley-Shubik index is defined under the assumption
that the voters vote in order. But if the vote is by secret ballot, then votes are neither
cast nor counted in any specific order. Does this fact affect the validity of the Shapley–
Shubik index?
Exercise 25 provides additional information relevant to this topic.
W-5. Weighted voting. Give some real-world examples of weighted voting, where
voters may have several votes each.
Exercise 38 provides additional information relevant to this topic.
W-6. Dummy. A voter whose votes have no effect on the outcome of an election is
sometimes called a dummy voter. Assume that a simple majority is required to win.
Ann has 4 votes, Beth has 1, and Cate has 1. Which of the three, if any, are dummy
voters?
W-7. More on the dummy. A voter whose votes have no effect on the outcome of an
election is sometimes called a dummy voter. What is the Banzhaf index for a dummy
voter? Is every voter with that index a dummy voter? Explain your answers.
Exercise 43 provides additional information relevant to this topic.
W-8. Bias. Explain how the use of the power indices described in this section can
reveal bias that may not be apparent in a voting system.
Exercise 44 provides additional information relevant to this topic.
W-9. Different answers. Can you identify practical examples where the Banzhaf index
and the Shapley–Shubik index are different?
Crauder-4100237 QL November 10, 2014 13:45
544 CHAPTER 8
candidates. There are Home TV & Video U. S. World Politics Entertainment Tech Travel
many different voting
systems in use. We Need Instant Runoff Voting: It’s Time for
Election Reform
JOSH WAUGH January 5, 2014
O nce upon a time, the United States was referred to as “a city upon a hill.”
A Puritan leader in New England during the 1630s by the name of John
Winthrop first invoked this religious metaphor to describe early North American
colonial society as a moral and democratic example for the rest of the world.
This term would be used and reused over the ages by figures touting American
exceptionalism as a way to emphasize a supposed ethical high ground.
Today, after a harrowing year that featured a federal government shutdown
on top of pre-existent budget cuts due to sequestration, our democracy is any-
thing but an example to the rest of the world. An example of deep partisan
divides, inefficiency, and stagnant political thought, maybe. But not of a well-
functioning democracy.
The need for reform—either at the local, state, federal, or constitutional
level—is obvious, and a possible solution sits right in front of our faces here at
the UW. We’ve been using Instant Runoff Voting (IRV) in ASUW elections since
2006 to great effect. It’s time we take our system to the state at large.
IRV allows for the selection of the overall best candidate by each voter ranking
all choices available for a position, from best to worst. This ranked-choice method
is more inclusive and allows each vote to say more than just a selection of option
A or option B.
For those who don’t know, IRV operates by eliminating the candidate with the
smallest number of first-choice votes and dispersing those votes to their second
choices, and then repeating until one candidate receives a majority. Theoretically,
this candidate-elect would have the greatest support from a wide range of voters.
In contrast, the current national system awards victory to the candidate who
simply has one more vote than the rest—in many cases, a candidate with less
than half of the votes will still win office. With a win like that, it is hard to say
that the elected candidate really has the will of the people or is accountable to
the majority of the population.
One of the hallmarks of a democracy is that its leaders are chosen by the “people,”
but there are many voting schemes by which this may be accomplished. This excerpt
shows how controversial such schemes can be. The opinion piece refers to instant
runoff voting and majority voting.
If a candidate receives a majority of the votes cast, many would agree that he or
she should be the winner. Even this is not as simple as it sounds, as the Electoral Col-
lege system for the U.S. president shows. And even if we agree that a candidate who
receives a majority of the votes should be the winner, it may be that no one achieves
that goal. When there are more than two candidates, the problem of determining the
“people’s choice” is difficult, to say the least. Indeed, a case can be made that for
Crauder-4100237 QL November 10, 2014 13:45
On the surface, this system seems quite reasonable, but there are difficulties. One
objection to plurality voting is that if there are a number of candidates, the winner
could be determined by only a small percentage of the electorate.
In Example 8.7, note that a candidate could receive only 26% of the votes cast,
far less than a majority, and still win a plurality of the votes.
With three or more candidates, there can be a spoiler.
Typically, a spoiler splits the vote of a more popular candidate who may otherwise
have won a majority. Many argue that this was the case with Ross Perot in the 1992
and 1996 U.S. presidential races and with Ralph Nader in the 2000 presidential race.
(Of course, the president is not determined by popular vote, but this example still
illustrates a point.)
Crauder-4100237 QL November 10, 2014 13:45
546 CHAPTER 8
Candidate Votes
Clinton 2,546,870
Dole 2,244,536
Perot 483,870
Others 28,518
Total votes cast 5,303,794
a. What percentage of all votes cast in Florida were for Clinton? Did any
candidate achieve a majority of votes cast?
b. Florida determines the winner of a presidential election by plurality, so
Clinton was declared the winner of all 25 of Florida’s electoral votes. Let’s
suppose that the vote was conducted with only the top three candidates and
that all their supporters continued to vote for them. Would it be possible for any
of the three to achieve a majority if all the votes cast for “Others” were
awarded to him?
c. It was speculated that most Perot voters would have voted for Dole if Perot
were not in the race. Let’s suppose that 83% of Perot voters would have voted
for Dole and 17% for Clinton. How many votes would Clinton have had and
how many would Dole have had? Who would have won the election? Would
you call Perot a spoiler in this case?
SOLUTION
a. Of the 5,303,794 votes cast, Clinton received 2,546,870 votes, which is
2,546,870/5,303,794 or about 48.0% of the votes. This is not a majority.
Clinton won more votes than any other candidate, so no candidate received a
majority of votes cast.
Crauder-4100237 QL November 10, 2014 13:45
b. Suppose that all of the 28,518 “Other” votes went to Clinton. Then he
would have received 2,575,388 votes. But half of 5,303,794 votes is 2,651,897,
so he still would not have had a majority. Certainly the same is true of the other
two candidates as well.
c. If 83% of Perot’s votes went to Dole, then Dole would have received 83% of
483,870 or an extra 401,612 votes, for a total of 2,646,148. Clinton would
have received 17% of 483,870 or an extra 82,258 votes, for a total of
2,629,128. Therefore Dole would have won with 17,020 more votes than
Clinton. If this were indeed true, then Perot would have been a spoiler.
548 CHAPTER 8
As an example of how to interpret this table, look at the row labeled “First
choice.” The first-choice votes were 4 for Alfred, 4 for Gabby, and 2 for Betty. The
fourth column, labeled “2 voters,” indicates that two voters chose Betty first, Gabby
second, and Alfred third.
Determine the winner under the elimination runoff system. (Because there are
only three candidates, this is equivalent to the top-two runoff system.)
SOLUTION
No candidate has a majority of the first-choice votes. Betty has the fewest first-choice
votes (2), so she is eliminated in the first round. For the second round, we first remove
Betty from the original table:
Now with Betty eliminated, for the four voters represented in the second column,
Alfred is the first choice and Gabby is the second choice. Similarly, for the four voters
represented in the third column, Gabby is the first choice and Alfred is the second
choice. For the two voters represented in the fourth column, Gabby is the first choice
and Alfred is the second choice among the remaining candidates. The ballot outcomes
for the second round are as follows:
Now the first-choice votes are 4 votes for Alfred and 4 + 2 = 6 votes for Gabby.
In this runoff, Gabby has a majority of the votes and is the winner.
There are other methods that use ranked ballots. In 1770, Jean-Charles de Borda
proposed a system of this type, and it bears his name today.
EXAMPLE 8.10 Using the Borda count: Our first choice loses
A group of five people needs to make a decision on which food to order. To make the
choice, everybody marks a ranked ballot for pizza, burgers, and tacos. The number
2 indicates a first-place vote, the number 1 a second-place vote, and so on. The table
below shows the five ballots cast.
We mentioned earlier that the voting for the Heisman Trophy uses a ranked ballot.
A first, second, and third choice is indicated on each ballot. In the next example, we
consider the race for the 2013 Heisman.
EXAMPLE 8.11 Finding the Borda count and Borda winner: Heisman
Trophy
Here are the vote counts of the top five finalists for the 2013 Heisman.
550 CHAPTER 8
To find the Borda count for Jameis Winston, we assign 2 points to each of his 668
first-place votes, 1 point to each of his 84 second-place votes, and 0 points to each
of his 33 third-place votes. Then we add to find the total number of points. Here are
the results:
Borda count for Jameis Winston = 668 × 2 + 84 × 1 + 33 × 0 = 1420
Borda count for AJ McCarron = 79 × 2 + 162 × 1 + 143 × 0 = 320
8 The actual system used to award the Heisman is a slight modification of the Borda count. See Exercise 36. The
outcome here is not affected.
Crauder-4100237 QL November 10, 2014 13:45
The next example shows that the four voting systems we have considered can
give four different outcomes for an election.
Candidate A B C D
Votes 28 25 24 23
Candidate A has the most first-choice votes, so A wins under plurality voting.
b. As we saw in part a, the top two vote-getters on an unranked ballot are A
with 28 votes and B with 25 votes. To determine the outcome of a runoff
Crauder-4100237 QL November 10, 2014 13:45
552 CHAPTER 8
between A and B, we use the preferences shown on the ballots, but we consider
only A and B. That is, we eliminate C and D from the original table showing the
voter preferences and look at the resulting preferences as expressed by the
ranking on the ballots.
Now the first-choice votes are 28 votes for A, 25 votes for B, and 47 votes for
C. Therefore, B (having the fewest votes) is eliminated. If we remove B from the
table, the combined ballot outcomes are
In this runoff, C has 72 votes and so is the winner under the elimination runoff
system.
d. The Borda count assigns numbers 3, 2, 1, and 0 to the choices. We
incorporate this information into the original table:
Example 8.12 is not just a curiosity. It shows some of the difficulties in determin-
ing exactly who is the “people’s choice.” All four of the candidates in this example
have a reasonable argument that they deserve to be named the winner.
Crauder-4100237 QL November 10, 2014 13:45
Condorcet’s paradox
Normally, an individual’s preferences are transitive. That is, if I prefer Pepsi over
Coke and I prefer Coke over Dr. Pepper, then it seems obvious that, if I am given a
choice between Pepsi and Dr. Pepper, I’ll choose Pepsi.
It may seem paradoxical, but this transitivity is not necessarily true if the pref-
erences are expressed by groups instead of individuals. We will illustrate this by the
following example.
Suppose there are three voters and three candidates, say, A, B, and C. The voters’
preferences are shown in the accompanying table.
554 CHAPTER 8
In 1785, the Marquis de Condorcet published an essay that described this para-
dox, which is now named after him. He referred to this phenomenon as the intran-
sitivity of majority preferences.
The existence of this paradox suggests that we focus on a candidate who would
win in any head-to-head contest with every other candidate. The name given to such
candidates honors Condorcet.
The Gentoo Foundation, which distributes the software Gentoo Linux, uses a Con-
dorcet voting process for elections to its board of trustees.9 Suppose in such an elec-
tion there are seven voters and three candidates, say, A, B, and C. The voters’ pref-
erences are shown in the accompanying table.
9 See www.gentoo.org/foundation/en/.
Crauder-4100237 QL November 10, 2014 13:45
Even if there is a Condorcet winner, that candidate may lose the election in all
the voting systems we have discussed. This is a further troubling aspect of common
voting schemes, and we illustrate it in the following example.
The 7 votes are distributed as follows: Candidate A has 3 votes, B has 1 vote, C
has 2 votes, and D has 1 vote. Candidate A has the most votes and wins in the
plurality system.
b. The first-place winner is A. Candidate C is second with 2 votes. In a runoff
between A and C, A wins 5 votes to 2. So, A wins the top-two runoff.
c. Because B and D get only 1 vote each, they are eliminated and A wins, just as
in part b.
d. We calculate the Borda count as follows:
Borda count for A = 3 × 3 + 1 × 2 + 3 × 1 = 14
Borda count for B = 1 × 3 + 4 × 2 + 2 × 1 = 13
Borda count for C = 2 × 3 + 2 × 1 = 8
Borda count for D = 1 × 3 + 2 × 2 = 7
So, A wins the Borda count.
e. There is a Condorcet winner, and it is B, as the following list of head-to-head
outcomes shows:
B beats A (4 to 3)
B beats C (5 to 2)
B beats D (4 to 3)
Crauder-4100237 QL November 10, 2014 13:45
556 CHAPTER 8
10 Alan Taylor’s book Mathematics and Politics: Strategy, Voting, Power and Proof (New York: Springer-Verlag, 1995)
gives a good overview of the many ideas in this fruitful area of debate.
Crauder-4100237 QL November 10, 2014 13:45
Suppose that there are 10 voters and that the winner is the one with a plurality of
votes. Assume that with candidates A and B in the race, the outcome is as follows:
Candidate A wins not only the plurality vote but also the majority vote by 6 to 4.
Now suppose that candidate C enters the race and that the new 10-vote tally is
This time, candidate B wins with a plurality of 4 votes. Note that none of the voters
in the elections changed their preference between A and B. Nonetheless, the intro-
duction of the irrelevant alternative of candidate C changed the election results from
A to B.
In this case, we see that the irrelevant alternative was nothing more than a spoiler.
So the criterion of independence of irrelevant alternatives perhaps seems a bit less re-
alistic than on first consideration. If the irrelevant alternative is, for example,
everyone’s first choice, then the irrelevant alternative would win under any system,
but that would not violate the independence of irrelevant alternatives. The presence
of a spoiler—as in the preceding example, changing the outcome from A to B—can
violate the independence of irrelevant alternatives.
The economist Kenneth Arrow studied the properties of voting systems. He was
seeking “fair” systems—in particular, systems that would have the reasonable Con-
dorcet winner criterion and also independence of irrelevant alternatives. These two
properties are quite different, but surely both seem fair. In point of fact, in 1950,
Arrow proved the following startling theorem.11
This is quite surprising for at least two reasons: First, the condition of indepen-
dence of irrelevant alternatives seems reasonable. Second, any voting system can be
adjusted to satisfy the Condorcet winner criterion by saying that, if there is a Con-
dorcet winner, then that candidate wins, and if there isn’t, then we carry on with the
voting system as originally designed.
Arrow’s theorem can be thought of as saying that there is no “best” voting sys-
tem for three or more candidates. Therefore, we must consider the strengths and
weaknesses of any voting system and decide what criteria seem the most fair to the
voters.
11 Kenneth Arrow, “A Difficulty in the Concept of Social Welfare,” Journal of Political Economy 58 (1950), 328–346.
Crauder-4100237 QL November 10, 2014 13:45
558 CHAPTER 8
a. Conventional wisdom is that Perot took more votes 9. Who wins the Borda count?
from Bush than from Clinton, but there is no real consen- 10. Is there a Condorcet winner? If so, who?
Crauder-4100237 QL November 10, 2014 13:45
11–15. Who wins? Use the following information for Exer- Rank 2 voters 2 voters 3 voters 2 voters 2 voters
cises 11 through 15. The vote tally of ranked ballots for a slate
of four candidates is as follows: First choice A C B A B
Second choice B A D C C
Third choice C B C B A
2 1 3 4 4 1 Fourth choice D D A D D
Rank voters voter voters voters voters voter
26. Who is the plurality winner?
First choice C A B D D B
Second choice B C C A B A 27. Who is the top-two runoff winner?
Third choice D B A B A D
28. Who is the elimination runoff winner?
Fourth choice A D D C C C
29. Who is the Borda winner?
11. Who is the plurality winner? 30. Is there a Condorcet winner? If so who?
12. Who is the top-two runoff winner? 31–35. Who wins? Use the following information for Exer-
cises 31 through 35. The vote tally of ranked ballots for a slate
13. Who is the elimination runoff winner? of four candidates is as follows:
14. Who is the Borda winner?
15. Is there a Condorcet winner? If so, who? Rank 4 voters 5 voters 3 voters 4 voters 6 voters
16–20. Who wins? Use the following information for Exer- First choice A C B A B
cises 16 through 20. The vote tally of ranked ballots for a slate Second choice B A D C C
of four candidates is as follows: Third choice C D C B A
Fourth choice D B A D D
560 CHAPTER 8
Assume that the quota is determined by a majority vote. Leave Exercises 43 and 44 are suitable for group work.
each index as a fraction. 43. The tyranny of the majority. There are critiques of majority
37. Banzhaf index. Find the Banzhaf index for each voting rule from Plato,12 de Tocqueville,13 religious communities,14
bloc. Suggestion: This will be easier if you first observe that and contemporary authors,15 each for quite different reasons.
the smallest bloc has no voting power. Winston Churchill is quoted as saying, “Democracy is an
38. Shapley--Shubik index. Find the Shapley–Shubik index for abysmal system of government. Its only advantage is that it
each bloc. is infinitely superior to anything else we’ve come up with so
far.” Report on the concerns of scholars regarding the concept
39. Interpretation. Interpret what your answers to Exercises 37 of majority rule.
and 38 say about the relative power of the four blocs. How is
this related to the discussion about Ross Perot in Example 8.8? 44. Constitutional limits. Majority rule does not mean that
the majority totally runs the show. For example, in the United
40–42. Bush versus Gore. Use the following information for States, it is not lawful for a majority to disenfranchise a mi-
Exercises 40 through 42. In the 2000 presidential election, the nority. There are many constitutional limits to what a majority
official popular vote tallies were as follows: may impose on a minority. Report on a few of these.
45. History. It has happened four times in U.S. history that a
presidential candidate lost the popular vote but won the presi-
Candidate Popular vote dency. Report on these four incidents.
Al Gore 51,003,926 46. History. The 1960 presidential race between Richard
George W. Bush 50,460,110 Nixon and John F. Kennedy was another very close race. In this
Others 3,898,559
election, Illinois was crucial. There were allegations of voter
fraud in Illinois that, if proven valid, could have changed the
results of the election. Report on the 1960 presidential race in
Gore won a plurality of the popular vote, but Bush won the Illinois.
electoral vote and so won the presidency. Let’s treat these re-
sults as voting blocs. Assume that the quota is determined by a 47. History. The year 1960 was the first time presidential
majority vote. This information is used in Exercises 40 through candidates faced each other in televised debates. The debates
42. Leave each index as a fraction. stirred controversy for any number of reasons. Report on the
1960 presidential debates.
40. Shapley--Shubik index. What is the Shapley–Shubik index
for each bloc? 48. History. In the United States, the winner of a presidential
race is the candidate who receives a majority of the electoral
41. Banzhaf index. What is the Banzhaf index for each bloc? vote. If no candidate achieves a majority, the House of Rep-
42. Interpretation. Interpret what your answers to Exercises 40 resentatives decides the winner of the election. This happened
and 41 say about the relative power of the three blocs. only once, in 1824. Report on the 1824 presidential election.
W-6. Plurality problems. What are some problems with plurality voting? Are these
problems addressed by the other systems we considered?
Exercise 11 provides additional information relevant to this topic.
W-7. Tie vote. Is it possible to have a vote for three candidates using a ranked ballot
that produces a tie in each of the following systems, all at the same time: plurality,
top-two runoff, elimination runoff, and Borda count?
W-9. Interpreting Arrow’s theorem. One could interpret Arrow’s impossibility the-
orem as an argument for choosing dictatorship as a form of government. Is this
interpretation reasonable? Explain your answer. Do you think dictatorship is a good
form of government?
N ewly divorced from Tiger Woods, Elin Nordegren is cutting back on the
help—even though she got a reported $110 million settlement from her
cheating ex, according to RadarOnline.com.
“She’s not cheap, but she’s just not throwing money around and wasting it
the way some people would,” a “friend” told the web site. “She’s not acting like
she hit the lottery.”
The Swedish siren still has a platoon of nannies to help her with the kids—
daughter Sam, 3, and son Charlie, 1.
“But she has cut back some of the hours some people work,” the friend said.
How else is Nordegren cutting back? She’s flying commercial—not on private
jets.
“She could afford the jet but didn’t even consider it,” the friend said.
There have been earlier reports that Woods was coughing up a sixth of his
fortune—around $100 million—as part of the divorce settlement, but Nordegren
has never confirmed how much she got.
When Nordegren finally ended nine months of silence in August about the
sex scandal that wrecked their marriage, she said she “felt stupid” that Woods
cheated on her with a parade of party girls and porn stars.
(Continued on next page)
Crauder-4100237 QL November 10, 2014 13:45
562 CHAPTER 8
There are many situations, such as divorces and inheritances, in which goods
have to be divided equitably. One way of accomplishing this is to liquidate all assets
and divide the resulting cash. But in many cases, liquidation is neither practical nor
desirable. For example, if a small business is part of an inheritance, one heir may
have an interest in taking over the business rather than selling it. There is also the
issue of value other than cash. For example, my father’s old guitar has little cash
value but holds great sentimental value to me. In this section, we will look at ways
to divide items so that, as nearly as possible, all parties see the division as fair. In the
process, we will see that the idea of fairness is difficult to define. See also Exercises
24 and 25, where the idea of envy-free procedures is explored.
Crauder-4100237 QL November 10, 2014 13:45
Divide-and-choose procedure
© cmbritclif/Dreamstime.com
Perhaps the simplest, and sometimes most practical, fair-division scheme is the divide-
and-choose procedure. Suppose you and I wish to share a chocolate chip cookie. I
could break the cookie into two pieces and give you one of the pieces, but maybe you
wouldn’t be happy with the outcome. Maybe you’re not happy because I gave you
the smaller piece, or maybe the pieces are of equal size but I gave you the half with
fewer chocolate chips. One simple solution is that I break the cookie but you choose
which piece you want (or vice versa—you break it and I choose). If I intentionally
break the cookie “unfairly,” I am setting myself up to get the less desirable piece.
> KEY CONCEPT
In the divide-and-choose procedure, one person divides the items into two
parts and the other person chooses which part he or she wants.
Here’s another example: Suppose my sister and I inherit a jewelry collection that
we want to divide. She could separate the jewelry into two piles and let me take the
pile I like best. While this may not result in an equal division, it is likely that both
of us will at least consider it fair. Perhaps my sister is particularly fond of a necklace
that Mom wore. When she divides the jewelry, she may intentionally put the necklace
in a small pile, knowing that I will choose the larger pile (which has the greater cash
value). Sister is happy because she gets to wear Mom’s necklace, and I am happy
because I got a larger cash value than I expected.
This division procedure can be extended easily to three people. The following
extension is the lone-divider method. Suppose now that my brother, my sister, and I
want to divide fairly a jewelry collection. I divide the jewelry into three piles that I
consider to be of equal value. There are piles 1, 2, and 3. My brother and my sister
each must surely consider that at least one pile has at least one-third of the total
value. Suppose my brother likes pile number 1 and my sister likes pile number 2.
Then brother and sister each take the pile they like, and I get pile number 3. Suppose,
however, that each of them likes the same pile, say, pile 1. Then I choose whichever of
piles 2 and 3 that I like. The remaining two piles are put back together, and brother
and sister use the divide-and-choose method for two persons to divide what’s left.
Just as in the case of Mom’s necklace, various strategies may be employed to get the
desired outcome.
In Exercises 21 and 22, you will be asked to extend this procedure to more than
three parties. Related procedures are the lone-chooser method (Exercise 19) and the
last-diminisher method (Exercise 20).
564 CHAPTER 8
Initially, each item goes to the person who bid the most for it. So, in the initial
round, I get the guitar and the library for a point total of 50. My sister gets the car
and the cash for a point total of 70. We give the tied item, the jewelry, to the point
leader—my sister. She now has a total of 80 points. This is not a fair division. In a
sense, I am only 50% satisfied, but my sister is 80% satisfied. We need to adjust to
even the totals.
To make the score come out even, my sister must give me some of her property.
Here is how we do it. For each item currently belonging to my sister (jewelry, car,
and cash), we calculate the following fraction:
Sister’s bid
My bid
Note that my sister’s bid goes on top because she is the leader (in terms of point
totals), and we are considering the items she owns. This fraction serves as a measure
of the relative value placed by each of us on the items. We calculate these fractions:
10
Ratio for jewelry = =1
10
40
Ratio for car = = 2.0
20
30
Ratio for cash = = 1.5
20
We arrange these ratios in increasing order:
Jewelry Cash Car
1 1.5 2
Now we begin transferring items from my sister to me in this order until we reach
an item that changes the point leader. We first transfer the jewelry. This gives me a
point total of 60 and my sister a point total of 70. Transferring the next item, the
cash, from my sister to me would make me the point leader. The cash is the critical
item, and we will transfer just enough of it from my sister to me to make the score
come out even. We want to take p percent of the cash from my sister and give it to
me. That will reduce her point score by p percent of her bid for the cash and increase
my score by p percent of my bid for the cash. We want the resulting point totals to
be the same. Expressed as an equation, this is
My score + p percent of my cash bid = Sister’s score − p percent of her cash bid
Writing p in decimal form gives the equation
60 + 20 p = 70 − 30 p
We solve this equation as follows:
60 + 20 p = 70 − 30 p Add 30 p to each side.
60 + 50 p = 70 Subtract 60 from each side.
50 p = 10 Divide both sides by 50.
p = 0.20
Crauder-4100237 QL November 10, 2014 13:45
So, in the end, I get the guitar, the jewelry, the library, and 20% of the cash. My
sister gets the car and the remaining 80% of the cash. Now my sister’s total points
are 40 + 30 × 0.80 = 64, and my total points are 35 + 10 + 15 + 20 × 0.20 = 64, so
each one of us is 64% satisfied with the outcome. This is a fair division, according
to each person’s valuation of the items.
SOLUTION
Initially, John gets the Hawk, the silver, and the cap and gown, and Faye gets the
condominium. John has a score of 50, and Faye has a score of 65. Faye is the leader.
We must use Faye’s only item, the condo, to even the score. (The condo is the critical
item.) We are going to take p percent of the condo from Faye and give it to John. We
find the decimal form of p by solving the equation from Step 5 of Summary 8.8:
566 CHAPTER 8
There are situations in which we transfer more than one item. We look at one of
these scenarios in the following example.
SOLUTION
Initially, Anne gets the laptop, the MP3 player, and the tablet for a point total of 87.
Becky gets the TV for a point total of 40. Anne is the leader, so she gets the tied item,
the DVDs. Anne’s new point total is 92. Because Anne has the lead, we look at ratios
of bids for items she owns, with Anne’s bid in the numerator:
50
Ratio for laptop = = 1.25
40
5
Ratio for DVDs = = 1
5
6
Ratio for MP3 player = = 1.2
5
31
Ratio for tablet = = 3.1
10
Crauder-4100237 QL November 10, 2014 13:45
In the final division, Becky gets the DVDs, the MP3 player, the TV, and 34% of
the laptop, and Anne gets the tablet and the remaining 66% of the laptop.
The adjusted winner procedure has several advantages. Under the assumption
that neither person has a change of heart about the original allocation of points, the
following can be shown to be true:
1. Both people receive the same number of points.
2. Neither person would be happier if the division were reversed.
3. No other division would make one person better off without making the other
worse off.
This means that the allocation is as fair as anyone knows how to make it. It may
or may not result in two happy campers, but from an objective point of view, it is a
fair division.
The adjusted winner procedure does have several drawbacks. One drawback is
that it is only for division between two people or parties. Another drawback is that
the procedure is useful only if there are two or more items to be divided. If there
is only one item to be divided, each person must bid all 100 points for the item
and, ultimately, each person will be given 50% ownership of that one item. A third
drawback is that the procedure may require that ownership of an item be divided in
some way between the people involved, or that some similar arrangement be made
for one of the items. Such an arrangement may be problematic if, for example, the
item has sentimental value and neither side wants to relinquish it.
Crauder-4100237 QL November 10, 2014 13:45
568 CHAPTER 8
To illustrate the procedure, let’s first assume that there is only one item to be
divided, an automobile. There are four people involved, Abe, Ben, Caleb, and Dan,
and each is entitled to one-fourth of the value (to them) of the car. Each bids the value
he places on the car, and each is going to wind up with a value of at least one-fourth
of his bid. Suppose the bids are as follows:
The highest bidder, Abe, gets the car. But that is the entire estate, and Abe is only
entitled to one-fourth of it. He corrects it with cash. The value of the car to him is
$30,000, and he is entitled to only one-fourth of that amount, $7500. So he puts the
remaining three-fourths, $22,500, into a kitty.
Now each of the other three withdraws from the kitty one-fourth of his
assigned value of the car. So, Ben takes one-fourth of $23,000, which is $5750.
Similarly, Caleb withdraws $7000, and Dan takes $6250. Note that the kitty is
not being divided evenly between Ben, Caleb, and Dan. The reason is that they did
not each assign the same value to the car. Each person is being given one-fourth of
the value he assigned to the car. The total withdrawals from the $22,500 kitty are
Crauder-4100237 QL November 10, 2014 13:45
$5750 + 7000 + 6250 = $19,000. That leaves a kitty of $3500 to be divided equally
among the four people, $875 for each. Each person is receiving one-fourth of the
value he assigned to the car, plus $875.
The following table summarizes the complete transaction:
One important result of this procedure is that each person gets at least a fourth
of his own evaluation of the worth of the car.
If there is more than one item to be divided, the Knaster procedure can be applied
to each item sequentially, as illustrated in Example 8.17.
Determine the fair division of the farm using the Knaster procedure.
SOLUTION
Each part of the farm is given to the highest bidder, so Julie will get the farmhouse,
Anne will get both the wheat fields and the cattle operation, and Steve will not get
any of the farm (although he will get cash). We begin with the farmhouse, which goes
to Julie. Because she valued the farmhouse at $120,000 but is entitled to only one-
third of that value ($40,000), she puts $80,000 in the kitty. Anne and Steve are each
entitled to one-third of their respective bids for the farmhouse, so Anne takes from
the kitty one-third of $60,000, or $20,000, and Steve takes one-third of $75,000, or
$25,000. That leaves $80,000 − 20,000 − 25,000 = $35,000 in the kitty, as shown
below.
570 CHAPTER 8
Similarly, Anne placed the highest bid for the wheat fields, which she gets. Be-
cause she valued the fields at $450,000 but is entitled to only one-third of that
value ($150,000), she puts $300,000 in the kitty. Julie and Steve are each entitled
to one-third of their respective bids for the wheat fields, so Julie takes from the
kitty one-third of $210,000, or $70,000, and Steve takes one-third of $390,000,
or $130,000. The net effect is to add $300,000 − 70,000 − 130,000 = $100,000 to
the $35,000 already in the kitty, as shown below.
Anne also had the highest bid for the cattle operation, which she gets. Because
she valued the cattle operation at $750,000 but is only entitled to one-third of that
value ($250,000), she puts $500,000 in the kitty. Julie and Steve are each entitled
to one-third of their respective bids for the cattle operation, so Julie takes from the
kitty one-third of $300,000, or $100,000, and Steve takes one-third of $600,000, or
$200,000. The net effect is to add $500,000 − 100,000 − 200,000 = $200,000 to
the $135,000 already in the kitty, as shown below. Finally, the balance of $335,000
left in the kitty is divided equally among the three heirs, so each gets approximately
$111,667, as shown below.
Summing the columns, we find that Julie gets the farmhouse and cash totaling
Anne gets the wheat fields, the cattle operation, and cash totaling
That is, Anne needs to pay $668,333. Steve gets no property, but he does get cash
totaling
$25,000 + 130,000 + 200,000 + 111,667 = $466,667
a. The item is given to the person with the highest bid, who is called
the winner. If two or more people make the same high bid, a
winner is chosen at random from the high bidders.
b. The winner contributes to the kitty the difference between that
person’s bid and his or her share of that item (which is his or her
bid divided by the number of bidders).
c. Those who don’t win the item receive from the kitty their share of
their bid for the item (which is their bid divided by the number of
bidders).
Step 3: The money left in the kitty is distributed equally to each bidder.
572 CHAPTER 8
10. What to do? How would you fairly divide things if the ini- 17. James and Gary want to divide three small businesses.
tial bids are as follows?
11. What to do? How would you fairly divide things if the ini- 18. A conglomerate splits into three companies and has to di-
tial bids are as follows? vide the distribution rights for a clothing line and a food line.
574 CHAPTER 8
22. Five people. Propose a divide-and-choose procedure for In the rest of the exercise, we assume the following: Abe,
five people. Ben, and Dan use the lone-divider method to divide the three
23. History. In 1945, the Polish mathematician Bronislaw cookies. Abe starts. The three cookies have equal value as far
Knaster introduced the procedure that bears his name. In fact, as he can see, so he just lays out the three cookies for Ben and
the study of fair division and the related mathematical area Dan. There are various possible outcomes after this.
called game theory developed rapidly in a period starting be- a. Let’s suppose that Ben and Dan both prefer cookie 1.
fore World War II and extending to the early stages of the Cold (This seems reasonable.) According to the procedure, Abe
War. Research the relevance of game theory to military strategy can take either cookie 2 or cookie 3. Suppose he takes
in general and in particular to the doctrine of mutually assured cookie 2. Then Ben and Dan divide cookie 1 and cookie
destruction. 3 between themselves using the divide-and-choose proce-
dure for two people. (The cookies can be broken!) Explain
Exercises 24 and 25 are suitable for group work. why either Ben or Dan will envy Abe.
24–25. Envy. A division procedure is envy-free if, when all is b. In part a, we sketched one possible outcome. There may
said and done, no one has reason to wish for someone else’s be reasons for Ben and Dan not to state a preference for
division. In Exercises 24 and 25, we investigate this idea. cookie 1. List all of the possible preferences the two of them
24. Explain why the divide-and-choose procedure for two can state. Show that if they state different preferences from
players is envy-free. each other, then one of them will envy the other. Also show
that if they state the same preference as each other, then
25. The point of this exercise is that the lone-divider method
Abe can settle on a choice that will make Ben or Dan envy
need not be envy-free. Abe, Ben, and Dan are dividing three
him.
chocolate chip cookies: cookie 1 with 10 chocolate chips,
cookie 2 with 9 chocolate chips, and cookie 3 with no chocolate The following exercise is designed to be solved using technol-
chips. Abe doesn’t care one way or the other about chocolate ogy such as a calculator or computer spreadsheet.
chips. As far as he is concerned, the three cookies are of equal 26. Knaster procedure. Make a spreadsheet that executes the
value. Ben and Dan care only about chocolate chips—they are Knaster procedure.
going to pick the chips out of the dough and eat only the
chocolate.
W-8. Try it yourself. To get a real feeling for how the adjusted winner procedure
works, it is a good idea to try it yourself. Make up a list of at least 10 items that
you and a friend will divide using the adjusted winner procedure. Can you devise a
strategy that will give you an advantage?
W-9. Try it yourself. To get a real feeling for how the Knaster procedure works, it is
a good idea to try it yourself. Make up a list of at least 10 items that you and several
friends will divide using the Knaster procedure. Can you devise a strategy that will
give you an advantage?
apportioned. What the U.S. Electoral Map Will Look Like in 2060
CHRIS CILLIZZA January 16, 2014
–1
–1 –1
–1
–1
+1 –1 –1
+1
+1
+1
+2 +1
576 CHAPTER 8
–1
–1 –1
+1 –1 –1 –1
–2 –1 –1
–3
–2
+2 –2 –1
+1 +1 +1
+2
+1 +1
–1 +1
+6 +1
+1
You may be surprised to learn that the U.S. Constitution lays out some basic
rules for the makeup of the House of Representatives, but leaves both the size of
the House and the exact number of representatives for each state to be determined
by Congress. This is the problem of apportionment and, as this article shows, the
issue remains important to this day. The Constitution gives these rules in Article 1,
Section 2: “Representatives and direct taxes shall be apportioned among the sev-
eral states which may be included within this union, according to their respective
numbers. . . .” The numbers—that is, the population of each state—are established
by the census: “The actual enumeration shall be made within three years after the
first meeting of the Congress of the United States, and within every subsequent term
of ten years, in such manner as they shall by law direct.” The history of the census is
Crauder-4100237 QL November 10, 2014 13:45
fascinating in itself. The nation has addressed questions such as how to count slaves,
people overseas, those in the military, native Americans, and noncitizens. In partic-
ular, before the Civil War, only three-fifths of the population of slaves was counted
for apportionment purposes.
© Albert Knapp/Alamy
16 It was temporarily increased to 437 in 1959 when Alaska and Hawaii were admitted to the Union.
17 Michel Balinski and H. Peyton Young, Fair Representation: Meeting the Ideal of One Man, One Vote (New Haven,
CT, and London: Yale University Press, 1982).
18 David C. Huckabee, The House of Representatives Apportionment Formula: An Analysis of Proposals for Change
and Their Impact on States, Congressional Research Service, 2001; http://usinfo.org/enus/government/branches/docs/
housefrm.pdf
Crauder-4100237 QL November 10, 2014 13:45
578 CHAPTER 8
181 members. To find the ideal size of a congressional district, we divide the U.S.
population by the size of the House. Then we divide the population of each state
by the size of a congressional district to determine that state’s proper share of the
House, called its quota.
For the population in 1810 with the House at 181, we find an ideal district size of
6,584,255
Ideal district size =
181
or about 36,377.099. For each state, we divide the population by the ideal district
size to calculate the quota. For example, Connecticut had a population of 261,818,
so its quota is
261,818
Quota for Connecticut =
36,377.099
or about 7.197. Here are the results of the 1810 census and the quota (to three
decimal places) for each state with the ideal district size of 36,377.099. We have
added a column giving the quota rounded to the nearest whole-number.
Crauder-4100237 QL November 10, 2014 13:45
The question is what to do with the fractional parts. For example, North
Carolina’s quota is 13.414. Should North Carolina get 13 representatives, or should
it get 14? This is no small issue—the number of representatives allocated to a state
makes a big difference in its congressional voting power.
The obvious solution is simply to round each quota to the nearest whole number,
as we have done in the last column of the table. One difficulty with this solution is
shown in the total. Rounding each state’s quota to the nearest whole number gives a
total of only 179 representatives. There are two representatives not assigned to any
state. This shows why the easy solution of rounding to the nearest whole number
by itself would not be a consistently reliable method for apportionment. In general
terms, the problem of apportionment is how to deal with the fractions.
Since 1790, several apportionment methods for the U.S. House of Representatives
have been developed and either used to allocate seats after various censuses or cited
in court cases, even as recently as 2001 (after the 2000 census). These methods fall
into two categories: those that rank the fractions, and those that round the fractions.
580 CHAPTER 8
Let’s see the results given by this method for the 1810 Congress.
Quota Fractional
rounded down: part Added
State Quota Initial seats of quota seats Final seats
Connecticut 7.197 7 0.197 7
Delaware 1.952 1 0.952 1 2
Georgia 5.782 5 0.782 1 6
Kentucky 10.289 10 0.289 10
Maryland 9.235 9 0.235 9
Massachusetts 19.263 19 0.263 19
New Hampshire 5.895 5 0.895 1 6
New Jersey 6.631 6 0.631 1 7
New York 26.199 26 0.199 26
North Carolina 13.414 13 0.414 1 14
Ohio 6.344 6 0.344 6
Pennsylvania 22.261 22 0.261 22
Rhode Island 2.114 2 0.114 2
South Carolina 9.252 9 0.252 9
Tennessee 6.705 6 0.705 1 7
Vermont 5.990 5 0.990 1 6
Virginia 22.476 22 0.476 1 23
Total 181 173 8 181
Using the whole-number part of the quota for initial allocation, we see that 173
seats are allocated. This leaves 181 − 173 = 8 leftover seats. These seats are given to
the 8 states for which the fractional parts of the quotas are the greatest. These states
correspond to the boldface numbers in the column giving the fractional parts. The
final allocations are shown in the last column of the table.
District Population
North 3900
East 4800
West 18,000
South 3300
Total 30,000
SOLUTION
We have 10 seats to fill and a total population of 30,000, so the ideal district size is
Total population
Ideal district size =
Number of seats
30,000
=
10
= 3000
We divide the population in each district by this district size, and the resulting quota
is shown in the second column of the accompanying table.
The third column of the table shows that the whole-number parts sum to 9, so
there is one leftover councilor to allocate. Because the quota of 1.6 has the largest
fractional part (0.6), the East district gets that councilor. The final result: North and
South should each get 1 councilor, East should get 2 councilors, and West should get
6 councilors.
The congressional bill of March 26, 1792, that allocated the U.S. House of Rep-
resentatives according to Hamilton’s method was vetoed by President George Wash-
ington. Indeed, this was the very first presidential veto. Some of the reasons for
Washington’s veto are investigated in Exercise 8. Some additional problems with
Hamilton’s method are investigated in Exercises 9 and 10. Hamilton’s method was
not used until it was revived for the 1850 census. Instead, a method devised by
Thomas Jefferson was used beginning with the 1790 census.
582 CHAPTER 8
one representative) using the whole-number part of the quota just as with the Hamil-
ton method. But rather than rank the fractional parts, Jefferson proposed that the
divisor be adjusted until the sum of the rounded-down quotas is the correct size—the
legislated size of the House of Representatives. There are complicated formulas that
can tell us how to adjust the divisor but, with the aid of a spreadsheet, trial-and-error
is a suitable method. Let’s consider again the allocation of seats for 1810. We start
with the divisor as the ideal district size of 36,377.099 and round each quota down
as our first allocation effort. As the fourth column of the accompanying table shows,
this allocates only 173 seats. In the fifth column, we decrease the divisor to 34,000
to increase the number of seats allocated. For example, with a divisor of 34,000,
Connecticut’s adjusted quota is 261,818/34,000 = 7.701 to three decimal places,
which rounds down to 7. The overall result is a total of 187 seats allocated—too
many. So, we increase the divisor to 35,000 and try again. The result is a sum of
181, as is shown in the last column of the table. Because this is the correct number
of seats, we have found the final allocation.
Comparing this apportionment with Hamilton’s, we see that they are very similar.
But Jefferson’s method gives one less delegate each to New Jersey, North Carolina,
and Tennessee and gives one additional delegate to each of Massachusetts, New York,
and Pennsylvania.
As we noted, trial-and-error (with the help of a spreadsheet) is an effective way
to adjust divisors to find an allocation that yields the correct House size. Finding the
proper divisor is usually not difficult because there is often a wide range of workable
answers. In the preceding table, any divisor between 34,856 and 35,037 will produce
the correct size.
District Population
North 3900
East 4800
West 18,000
South 3300
Total 30,000
SOLUTION
We have 10 councilors and a total population of 30,000, so our ideal district size is
30,000/10 = 3000. We use this as our first divisor. Dividing gives quotas of 1.3, 1.6,
6, and 1.1, which round down to 1, 1, 6, and 1 (see the accompanying table). These
sum to 9, so we need to adjust the divisor down.
Crauder-4100237 QL November 10, 2014 13:45
584 CHAPTER 8
Trying a divisor of 2500 gives quotas of 1.6, 1.9, 7.2, and 1.3, which round
down to 1, 1, 7, and 1. These sum to 10, so North, East, and South should each
get 1 councilor, and West should get 7 councilors. Note that this distribution differs
from that of Hamilton’s method in Example 8.18.
Jefferson’s method was adopted by Congress for the Third Congress (1793) and
used with each census through 1830. Even from the beginning, it was clear that small
states may well be disproportionately underrepresented by this method. For 1800,
Jefferson’s method awards Delaware only a single representative based on its appor-
tionment population21 of 61,812—far more than the ideal district size of 34,680 at
that time. As Jefferson’s method continued to be applied for the censuses through
1830, it became more and more evident that the method favored larger states over
smaller states. Note for the 1810 results that, compared with Hamilton’s method,
Jefferson’s method gives extra representatives to the larger states of Massachusetts,
New York, and Pennsylvania, and takes them away from the smaller New Jersey,
North Carolina, and Tennessee.
In addition to the problem of bias in favor of larger states, there was the question
of staying within the quota. In Exercise 32, you will be asked to verify that in the
1820 census New York had a quota of 32.503. We would thus expect New York
to get either 32 or 33 representatives. In fact, Jefferson’s method gives New York
34 representatives, thus violating quota. Exercise 23 gives a simple situation where
quota is violated by Jefferson’s method. By contrast, Hamilton’s method always stays
within quota.
21 The actual population of Delaware was enumerated in the census as 64,273; for apportionment purposes, however,
each of Delaware’s 6153 slaves was counted as three-fifths of a person, as noted at the beginning of this section.
Crauder-4100237 QL November 10, 2014 13:45
586 CHAPTER 8
faster than either state and Virginia had a larger population, after one year Virginia’s
fractional part of 0.599 would slip below that of Maine. Then Maine’s quota could
be rounded up and Virginia’s rounded down. This could happen despite the fact that
Virginia was growing faster than Maine.
A third problematic consequence of Hamilton’s method is the new states paradox.
When Oklahoma became a state in 1907, 5 seats were added to the House. This was
the number of seats to which Oklahoma would be entitled, based on its population
then. Oklahoma was indeed awarded 5 seats, and one would assume that the number
of representatives for the other states would be unchanged. If Hamilton’s method had
been used, however, it would have changed the number of representatives for Maine
and New York.
These paradoxes were sufficiently troubling to Congress that in time new methods
were sought that would avoid real and perceived problems with Hamilton’s method.
588 CHAPTER 8
1. We calculate ideal district size by dividing the total population by the size
of the House. This is used as the first divisor.
2. We calculate the quota for each state by dividing its population by the
divisor.
3. We round each quota to a whole number as follows:
Jefferson’s method: We round down (but not less than 1).
Adams’ method: We round up.
Webster’s method: We round to the nearest whole number (up if the
fractional part is 0.5 or greater and down otherwise, but not less than 1).
Then we sum the resulting rounded quotas.
4. If the sum from Step 3 is larger than the size of the House, we increase the
divisor and repeat Steps 2 and 3. If the sum is too small, we decrease the size
of the divisor and repeat Steps 2 and 3. We continue this process until a
divisor is found for which the sum of the rounded quotas is the House size.
This is the final apportionment.
State Population
Montana 994,416
Alaska 721,523
Florida 18,900,773
West Virginia 1,859,815
Total 22,476,527
SOLUTION
a. The ideal district size is the total population divided by the number of seats,
or 22,476,527/32 = 702,391.469. The quota for each state is calculated by
dividing the state’s population by this divisor:
Crauder-4100237 QL November 10, 2014 13:45
Quota Fractional
rounded down: part of Added Final
State Population Quota Initial seats quota seats seats
Montana 994,416 1.416 1 0.416 1
Alaska 721,523 1.027 1 0.027 1
Florida 18,900,773 26.909 26 0.909 1 27
West Virginia 1,859,815 2.648 2 0.648 1 3
Total 22,476,527 32 30 2 32
590 CHAPTER 8
Another try:
Quota First try: Increase divisor
using divisor Quota to 750,000,
State Population 702,391.469 rounded up round up
Montana 994,416 1.416 2 2
Alaska 721,523 1.027 2 1
Florida 18,900,773 26.909 27 26
West Virginia 1,859,815 2.648 3 3
Total 22,476,527 32 34 32
Note that the four methods used in Example 8.21 give three different answers.
SOLUTION
We begin by showing the quota using the ideal district size of 702,391.469 and the
corresponding geometric mean (from Table 8.3). We round down if the quota is less
than the geometric mean; otherwise, we round up.
In the final column of the table, we see that we have allocated a total of 33 seats—
too many. So we use a larger divisor of 710,000.
Second try:
Quota using First try: Quotient
divisor Geometric Rounded using divisor Rounded
State Population 702,391.469 mean quota of 710,000 quotient
Montana 994,416 1.416 1.414 2 1.401 1
Alaska 721,523 1.027 1.414 1 1.016 1
Florida 18,900,773 26.909 26.495 27 26.621 27
West Virginia 1,859,815 2.648 2.449 3 2.619 3
Total 22,476,527 32 33 32
The last column of the table shows that this allocates a total of 32 seats, which is the
required number. So Huntington–Hill awards 1 seat to Montana, 1 to Alaska, 27 to
Florida, and 3 to West Virginia.
592 CHAPTER 8
28
27 27 27
26 Hamilton method Jefferson method Adams method
3 3 3 3
2 2
1 1 1 1 1 1 1 1 1
Figure 8.1 illustrates the differing effects of the five methods from Examples 8.21
and 8.22 on allocation of seats to Montana, Alaska, Florida, and West Virginia. Since
each of these methods has historically been used at various times by Congress, these
differing effects raise the question, “Which is fair?”
Huntington–Hill method, despite lawsuits filed by states after both the 1990 and
2000 censuses. Although the Supreme Court has determined that Congress has the
discretion to choose an apportionment method, we can anticipate attempts to change
apportionment methods in the future.
594 CHAPTER 8
2. District size in 2010. In 2010, there were 435 members of c. Explain the difficulty Hamilton’s method has in allotting
the House of Representatives. The U.S. population in 2010 was the 10 House seats.
309,183,463. What was the ideal congressional district size in 10. Hamilton’s method and small states. When a state’s pop-
2010? Use three decimal places in your answer. ulation is too small, Hamilton’s method (if unaltered) may not
3. New York’s quota in 1790. The district size in 1790 was assign it any representatives. Suppose, for example, that the
34,437. The population of New York in 1790 was 331,589. country consists of two states: State 1 with a population of 9
What was New York’s quota in 1790? Use two decimal places and State 2 with a population of 91. There are 5 seats to be
in your answer. assigned. How does Hamilton’s method assign these 5 seats? It
should be noted that Hamilton’s method was later modified to
4. Virginia’s quota in 1790. The district size in 1790 was account for this difficulty.
34,437. The population of Virginia in 1790 was 630,560. What
was Virginia’s quota in 1790? Use two decimal places in your 11–16. A small town. Use the following information for Ex-
answer. ercises 11 through 16. A small town elects a total of 12 board
members from four districts: North, South, East, and West. The
5. California’s quota in 2010. In 2010, the U.S. population populations of each district are shown below.
was 309,183,463. There were 435 members of the House of
Representatives. The population of California in 2010 was District Population
37,341,989. What was California’s quota in 2010? Use two
decimal places for your answer. North 100
South 150
6. Minnesota’s quota in 2010. In 2010, the U.S. population East 200
was 309,183,463. There were 435 members of the House of West 700
Representatives. The population of Minnesota in 2010 was
5,314,879. What was Minnesota’s quota in 2010? Use two dec- 11. Find the ideal district size for this small town. Use two dec-
imal places for your answer. imal places in your answer.
7. An interesting House size. The U.S. Constitution states that 12. Find the distribution of the 12 board members using
the minimum size of a congressional district is 30,000 people. Hamilton’s method.
The U.S. population in 2013 was 316,128,839. If each district 13. Find the distribution of the 12 board members using
size were 30,000 people, how many House members would we Jefferson’s method.
have had in 2013? Round your answer to the nearest whole
14. Find the distribution of the 12 board members using
number.
Adams’s method.
8. Washington’s veto. President Washington vetoed the Hamil-
15. Find the distribution of the 12 board members using
ton apportionment bill. Part of his reasoning involved the state
Webster’s method.
of Delaware. Hamilton’s plan would have awarded Delaware
2 House seats. The population of Delaware in 1790 was 16. Find the distribution of the 12 board members using the
55,540. Huntington–Hill method.
a. With 2 House seats, what would have been the size of a 17–22. A larger town. Use the following information for Ex-
congressional district in Delaware? ercises 17 through 22. A town elects a total of 15 board mem-
bers from four districts: North, South, East, and West. The pop-
b. Compare your answer to part a with the constitutional
ulations of each district are shown below.
minimum for the number of people in a congressional
district.
9. Hamilton’s method and ties. One problem with Hamilton’s
District Population
method could arise if there are ties in the quotas. (This would North 145
be highly unlikely to happen in practice.) Suppose a country South 150
consisted of State 1 with population 140, State 2 also with pop- East 326
ulation 140, and State 3 with population 720. That is a total West 735
Crauder-4100237 QL November 10, 2014 13:45
17. What is the ideal district size for this town? Use one deci- a. Determine the distribution of board members among
mal place in your answer. the five states if the 136-member board is selected using
18. Find the distribution of the 15 board members using Hamilton’s method.
Hamilton’s b. Do Kansas and Arkansas have the same number of
method. board members?
19. Find the distribution of the 15 board members using c. Which of the four original states gained a board member
Jefferson’s method. with the addition of Kansas to the coalition?
20. Find the distribution of the 15 board members using 27. Dean’s method. Dean’s method is an adjusted divisor
Adams’s method. method. It follows the steps of the other adjusted divisor meth-
21. Find the distribution of the 15 board members using ods, but rounding is done using the harmonic mean: If n is the
Webster’s method. whole-number part of the quotient (initially the quota), the har-
n(n + 1)
22. Find the distribution of the 15 board members using the monic mean we use is . We round up if the quotient
n + 1/2
Huntington–Hill method. is at least this large; otherwise, we round down. The following
23. Violating quota. This exercise shows how Jefferson’s table provides a few helpful values for the harmonic mean:
method can violate quota. The following table shows the pop-
ulation for a nation consisting of State 1, State 2, and State 3.
The quota for a House size of 25 is also given. Whole number Harmonic mean
1 1.333
2 2.400
State Population Quota 26 26.491
State 1 200 3.571
State 2 200 3.571
State 3 1000 17.857
What allocation does Dean’s method give for the four states in
Total 1400 25 Example 8.21?
The following exercises are designed to be solved using tech-
nology such as calculators or computer spreadsheets.
a. Show that 52 works as a divisor for Jefferson’s method.
Exercises 28 through 32 are suitable for group work.
b. For which state is quota violated?
28–31. The 1790 Congress. The table below gives state pop-
24–26. The trouble with a new state. Use the following infor- ulations in 1790, when the size of the House was 105. The
mation for Exercises 24 through 26. Suppose that Oklahoma, ideal district size in 1790 was 34,437, and the table shows the
Texas, Arkansas, and New Mexico form a Board of Represen- quotas to three decimal places. This information is used in Ex-
tatives to attract business to the region. There are to be 125 ercises 28 through 31.
members of the board, and the membership from each state
was determined by its population in 2009. The populations of
these states are given in the table below.
State Population Quota
Connecticut 236,841 6.878
Delaware 55,540 1.613
State Population Georgia 70,835 2.057
Oklahoma 3,687,050 Kentucky 68,705 1.995
Texas 24,782,302 Maryland 278,514 8.088
Arkansas 2,889,450 Massachusetts 475,327 13.803
New Mexico 2,009,971 New Hampshire 141,822 4.118
New Jersey 179,570 5.214
Total 33,368,773 New York 331,589 9.629
North Carolina 353,523 10.266
Pennsylvania 432,879 12.570
Rhode Island 68,446 1.988
24. Find the ideal district size for the coalition. Use three deci- South Carolina 206,236 5.989
mal places in your answer. Vermont 85,533 2.484
25. Determine the distribution of board members among the Virginia 630,560 18.311
four states if Hamilton’s method is used. Total 3,615,920 105
26. Kansas, which has a population of 2,805,747, joins the
coalition. Because the population of Kansas is close to that of
Arkansas, it is decided that 11 new board members should be 28. Find each state’s representation for 1790 using Hamilton’s
added, for a new total membership of 136. apportionment method.
Crauder-4100237 QL November 10, 2014 13:45
596 CHAPTER 8
29. What allocation would Jefferson’s method give for 1790? State Population State Population
30. What allocation would Adams’s method give for 1790?
New York 1,368,775 Connecticut 275,208
31. What allocation would Webster’s method give for 1790? Pennsylvania 1,049,313 New Jersey 274,551
Virginia 895,303 New Hampshire 244,161
32. Jefferson’s method for 1820: violating quota. The popula-
Ohio 581,434 Vermont 235,764
tion figures for 1820 are given in the following table. North Carolina 556,821 Indiana 147,102
a. Use Jefferson’s method to calculate the apportionment Massachusetts 523,287 Louisiana 125,779
of 213 representatives for 1820. Kentucky 513,623 Alabama 111,147
South Carolina 399,351 Rhode Island 83,038
b. Verify that quota is violated for New York. Tennessee 390,769 Delaware 70,943
c. For what other states is quota violated? Maryland 364,389 Missouri 62,496
Maine 298,335 Mississippi 62,320
Georgia 281,126 Illinois 54,843
Total 8,969,878
W-2. Quota. How is the congressional quota for a state determined? If the ideal
district size and each state’s quota are not rounded, what is the sum of the state
quotas?
Exercise 3 provides additional information relevant to this topic.
W-3. Roots. Which apportionment method uses square roots? Explain in terms of
the rounding procedure why this method gives some advantage to smaller states.
W-4. Fair? List some factors you would consider in determining whether a method
for apportionment is fair.
W-5. Washington’s veto. The very first presidential veto was cast by George Washing-
ton when he vetoed the Hamilton apportionment bill. Do some research to find out
why Washington vetoed the bill. Some of the reasons are presented in this section.
Exercise 8 provides additional information relevant to this topic.
W-6. Paradox. Explain where the paradox lies in each of the three paradoxes men-
tioned in this section (Alabama, population, new states).
Exercise 26 provides additional information relevant to this topic.
W-7. Census. Sketch the history of the census, including the way slaves were counted.
W-8. Other applications. Do the ideas of this section apply to situations other than
choosing representatives? If so, give some practical examples.
W-9. How it’s done today. Explain in your own words how the U.S. House of Rep-
resentatives is apportioned today. Do you think the method is fair?
CHAPTER SUMMARY
This chapter revolves around questions of how a group should make decisions. How
to measure voting power and how to evaluate voting systems are two topics from
the political process. Other situations in which making a decision is difficult arise
in the context of equitable division, such as dividing an inheritance. One important
Crauder-4100237 QL November 10, 2014 13:45
Number of votes
Total Winning Critical
20 20 1 votes coalition? voters
A B C 41 Yes None
A B 40 Yes A, B
A C 21 Yes A, C
A 20 No Not applicable
B C 21 Yes B, C
B 20 No Not applicable
C 1 No Not applicable
The Banzhaf index of a voter is the number of times that voter is critical in a
winning coalition divided by the total number of instances of critical voters. So the
Banzhaf index of each of the three voters in the example is 2/6 = 1/3 or about
33.33%. This indicates that they, indeed, have equal voting power.
The Shapley--Shubik power index is another measure of voting power. If voters
cast their votes in order, then one specific voter, the swing voter, may cast a vote
that lifts the total to the quota. The Shapley–Shubik power index for a voter is the
fraction of all possible orderings in which that voter is the swing. For the preceding
example, we list the possible orderings of the three voters and the swing voter.
598 CHAPTER 8
Each voter is the swing voter in 2 of the 6 orderings, so the Shapley–Shubik index of
each is 2/6 = 1/3 or about 33.33%. This indicates once again that the three voters
have equal power.
The Banzhaf and Shapley–Shubik indices are not always equal. But each can point
out bias or complexity in a system.
For this ballot total, A wins the plurality, B wins the top-two runoff, C wins the
elimination runoff, and D wins the Borda count. Each of the four candidates has a
reasonable claim to be the “people’s choice.”
A Condorcet winner of an election is a candidate who would beat all others in
a head-to-head election. Some elections have a Condorcet winner, and some don’t.
Even if there is a Condorcet winner, none of the systems described in this chapter can
guarantee that the Condorcet winner will win the election.
If the voters prefer one candidate over another, the introduction of a third candi-
date should not reverse the result. That is the principle of independence of irrelevant
alternatives. Arrow’s impossibility theorem tells us that no voting system (other than
a dictatorship) that always selects the Condorcet winner can have this property. One
interpretation of this result is that no voting system is perfect, and we should be
aware that the voting system may play as large a role as the vote total in selecting
the winner.
among the other parties. The division of the cash takes into account each party’s
valuation. One advantage of this procedure is that each party gets at least a fixed
fraction of its valuation of each item. For example, if there are five parties, then each
party gets at least one-fifth of its valuation of each item. However, this procedure
may require each party to have relatively large amounts of cash on hand.
In general, fair division is a complex issue with no simple solutions that work in
all cases.
Apportionment: Am I represented?
The problem of apportionment is concerned with the allocation of representatives
for a governing body among several groups. The main focus of this section is on
historical approaches to apportionment of seats in the U.S. House of Representatives
among the states.
Congress sets the size of the House of Representatives. To find the ideal size of a
congressional district, we divide the U.S. population by the size of the House. Then
we divide the population of each state by the size of a congressional district to de-
termine that state’s proper share of the House, called its quota. In general, the quota
will not be a whole number, and the varying approaches to apportionment differ in
the way they treat the fractional parts.
Hamilton’s method initially gives each state the number of representatives corre-
sponding to the whole-number part of its quota. (In practice, this number is never
allowed to be 0.) If the sum of the resulting quotas is less than the House size, the
leftover seats are distributed, one per state, in order from the highest fractional part
down.
For Jefferson’s method of apportionment, we round each quota down (but not
less than 1). If the sum of the initial rounded quotas is larger than the size of the
House, we increase the divisor, recalculate the quotas, round, and check the sum
again. If the sum of the initial rounded quotas is too small, we decrease the divisor
and repeat those steps. We continue this process until a divisor is found for which
the sum of the rounded-down quotas (but not less than 1) is the House size.
Other methods of apportionment are similar to Jefferson’s in that they are ad-
justed divisor methods. They differ only in the way rounding is done. Adams’s method
rounds the quotas up, Webster’s method rounds to the nearest whole number, and the
Huntington–Hill method rounds in terms of the geometric mean. Currently, Congress
uses the Huntington–Hill method for apportionment.
There is no perfect method of apportionment. Hamilton’s method suffers from
the Alabama, population, and new states paradoxes. The final apportionment for a
state should be the quota either rounded down or rounded up, thus staying within
quota. But all adjusted divisor methods fail this test: They violate quota.
KEY TERMS
quota, p. 528 elimination runoff system, adjusted winner procedure,
winning coalition, p. 529 p. 547 p. 563
losing coalition, p. 529 preferential voting systems, p. 547 Knaster procedure (or method of
critical voter, p. 530 ranked ballot, p. 547 sealed bids), p. 568
Banzhaf power index, p. 531 Borda winner, p. 549 ideal district size, p. 578
swing voter, p. 534 Condorcet winner, p. 554 standard divisor, p. 578
Shapley–Shubik power index, Condorcet winner criterion, p. 556 state’s quota, p. 578
p. 535 independence of irrelevant stay within quota, p. 584
plurality, p. 545 alternatives, p. 556 violate quota, p. 584
spoiler, p. 545 divide-and-choose procedure,
top-two runoff system, p. 547 p. 563
Crauder-4100237 QL November 10, 2014 13:45
600 CHAPTER 8
CHAPTER QUIZ
1. Suppose there are three delegates to a small convention. Al- Abe, Ben, and Condi. The voters’ preferences are shown in the
fred has 10 votes, Betty has 7 votes, and Gabby has 3 votes. A accompanying table.
simple majority wins. Find the quota, make a table listing all
of the coalitions and the critical voter in each case, and find the
Banzhaf index for each voter.
Answer Quota: 11 votes. We use A for Alfred, B for Betty, and Preferences 7 voters 5 voters 3 voters
G for Gabby. First choice Abe Ben Condi
Second choice Ben Condi Abe
Number of votes Third choice Condi Abe Ben
Total Winning Critical
10 7 3 votes coalition? voters
A B G 20 Yes A
A B 17 Yes A, B Is there a Condorcet winner? If so, which candidate is it?
A G 13 Yes A, G
A 10 No Not applicable Answer No.
B G 10 No Not applicable
B 7 No Not applicable
If you had difficulty with this problem, see Example 8.13.
G 3 No Not applicable
5. Use the adjusted winner method to divide the listed items
Alfred: 3/5; Betty: 1/5; Gabby: 1/5. fairly when the following bids are made:
If you had difficulty with this problem, see Example 8.2.
2. There are three voters: A with 10 votes, B with 8 votes, and
C with 3 votes. A simple majority wins. Find the quota, make a Michael Item Juanita
table listing all of the permutations of the voters and the swing
voter in each case, and calculate the Shapley–Shubik index for 40 Lakeside cottage 30
each voter. 30 Penthouse apartment 25
30 Ferrari 45
Answer Quota: 11 votes.
3. Consider the following ranked ballot outcome for 14 Person Ann Beth Cate Dan
voters choosing among three candidates: Alfred, Betty, and
Gabby. Bid $1200 $2000 $1200 $1600
Rank 7 voters 4 voters 3 voters Answer Ann: $425; Beth: computer less $1375; Cate: $425;
First choice Alfred Betty Gabby Dan: $525.
Second choice Betty Gabby Alfred
Third choice Gabby Alfred Betty If you had difficulty with this problem, see Example 8.17.
Determine the winner under the elimination runoff system and 7. Our city has four districts: North, East, West, and South.
the winner using the Borda count. There are required to be 15 city councilors allocated according
to population, with at least one from each district. Given the
Answer Elimination runoff: Alfred; Borda count: Alfred. populations below, calculate the quota for each district. (You
need keep only one place beyond the decimal point.) Then use
If you had difficulty with this problem, see Example 8.9. Hamilton’s method to determine the number of councilors to
4. In an election, there are 15 voters and 3 candidates, say, represent each district.
Crauder-4100237 QL November 10, 2014 13:45
8. Our city has four districts: North, East, West, and South. North should get 2 councilors, East should get 6, West should
There are required to be 15 city councilors allocated ac- get 3, and South should get 4.
cording to population, with at least one from each district.
Given the populations below, use Jefferson’s method to de- If you had difficulty with this problem, see Example 8.19.
Crauder-4100237 QL November 10, 2014 14:8
© Dmitry Knorre/Dreamstime.com
Crauder-4100237 QL November 10, 2014 14:8
9 GEOMETRY
Y
ou might be surprised by the following article from the Proceedings of the Na-
tional Academy of Sciences. It indicates that all of us may have an innate knowl-
edge of geometry, even though we are not conscious of it.
even in the absence of formal mathematical education. Our tests probed intu-
itions of points, lines, and surfaces in participants from an indigene group in the
Amazon, the Mundurucu, as well as adults and age-matched children controls
from the United States and France and younger US children without education in
geometry. The responses of Mundurucu adults and children converged with that
of mathematically educated adults and children and revealed an intuitive under-
standing of essential properties of Euclidean geometry. For instance, on a surface
described to them as perfectly planar, the Mundurucu’s estimations of the inter-
nal angles of triangles added up to ∼180 degrees, and when asked explicitly, they
stated that there exists one single parallel line to any given line through a given
point. These intuitions were also partially in place in the group of younger US
participants. We conclude that, during childhood, humans develop geometrical
intuitions that spontaneously accord with the principles of Euclidean geometry,
even in the absence of training in mathematics.
In this chapter, we examine topics from geometry that are encountered in every-
day life. In the first section, we study the measurement of geometric figures. In the
next section, we examine how the notion of geometric similarity is related to the
idea of proportionality. In the last section, we consider symmetries of shapes and
the notion of tiling the plane.
TAKE AWAY
FROM THIS
SECTION
Calculate 9.1 Perimeter, area, and volume: How do I
perimeters, areas, measure?
volumes, and
surface areas of Geometric shapes do not concern only mathematicians. They are part of the lan-
familiar figures. guage of architects, engineers, and designers. Think of the Coliseum in Rome, the
spiral-shaped Hirshhorn Museum in Washington, DC, and all the baseball, soccer,
basketball, and football fields you visited or played on!
603
Crauder-4100237 QL November 13, 2014 12:18
604 CHAPTER 9
Diameter
Radius
Tomislav Forgo/Shutterstock
FIGURE 9.2 Circumference/Diameter is the same for Earth and a baseball.
Rectangles A rectangle is a figure with four right angles whose opposite sides are
equal. (See Figure 9.3.) Here are two basic formulas:
Triangles A triangle is a figure with three sides. To find the area of a triangle, we select
Length any one of the three sides and label it the base. Then we find the height by starting
at the vertex opposite the base and drawing a line segment that meets the base in a
FIGURE 9.3 Opposite sides right angle. In Figure 9.4, we have chosen the base to be AB, and the resulting height
of a rectangle are of equal is “inside” the triangle. In Figure 9.5, we have chosen the base to be the side AC,
length.
which results in a height “outside” the triangle. No matter which side we choose for
the base, the area is always given by the formula
1
Area = Base × Height
2
C B
Height Height
A Base B C Base A
606 CHAPTER 9
Right triangles A right triangle is a triangle with one 90-degree (or right) angle. The
Hy
po c
ten traditional names for sides of a right triangle are illustrated in Figure 9.6. One of the
Leg use
most familiar of all facts in geometry is the famous theorem of Pythagoras.
a
Pythagorean theorem For the right triangle shown in Figure 9.6,
Leg
b a 2 + b2 = c2
FIGURE 9.6 The sides of a In this formula, c always represents the hypotenuse, which is the side opposite the
right triangle. right angle. The other two sides of a right triangle are the legs. The converse of the
Pythagorean theorem is true: If a triangle has three sides of lengths a, b, and c such
that a 2 + b2 = c2 , then that triangle is a right triangle with hypotenuse c.
Cartoonstock.com
did the outside runner B go than the inside runner A? (Note that no information was
given about the diameter of the track.)
© Gan Hui/Dreamstime.com
2
r+
r
SOLUTION
Let r denote the radius of the inside lane, where A is running. Then r + 2 is the
radius of the outside lane, where B is running. The distance covered by runner A is
Therefore, the distances covered by the two runners differ by 4π or about 12.6
feet. At first glance, it may appear that a longer inside track would cause a greater
difference in the distance the runners travel. But the difference is the same, 12.6 feet,
whether the inside track is 100 yards in diameter or 100 miles in diameter.
Area = πr 2
Crauder-4100237 QL November 13, 2014 12:18
608 CHAPTER 9
© Iurii Osadchi/Dreamstime.com
Goal
a
5
b
8 A
7
B 3
SOLUTION
The distance for player A is marked a in the figure. It represents the hypotenuse of a
right triangle. Similarly, b is the distance for player B, and it represents the hypotenuse
Goal
6 a
b
9
A
7
3
B
Heron’s formula
We know how to find the area of a triangle if we know the lengths of the base and
height. Heron’s formula gives the area of a triangle if we know the lengths of all
three sides. Referring to the triangle shown in Figure 9.10 with sides a, b, and c, we
calculate one-half of the perimeter, the semi-perimeter, S = 12 (a + b + c). Heron’s
formula says that the area is given by
Area = S(S − a)(S − b)(S − c)
For the triangle in Figure 9.11, which has sides 5, 6, and 9, we find that
S = 12 (5 + 6 + 9) = 10. Thus, the area is
Area = 10(10 − 5)(10 − 6)(10 − 9) = 200
or about 14.1 square units.
c 6
a 5
b 9
SOLUTION
a. The lawn has an irregular shape, but by adding the line segments shown in
Figure 9.13, we can divide it into three pieces: a right triangle, a rectangle, and
an oblique triangle (a triangle with no right angle).
100 70 30
30 30
40 40 40
H
20 20
70 70
The length of fence we need is the perimeter. To find the perimeter, we need
to find the hypotenuse (marked H in Figure 9.13) of the right triangle. The legs
have length 30 feet and 40 feet. We find H using the Pythagorean theorem:
c2 = a 2 + b2
H2 = 302 + 402
H2 = 2500
H = 2500
H = 50 feet
b. Look again at Figure 9.13. The area of the lawn is the area of the rectangle
plus the area of the right triangle plus the area of the oblique triangle. For the
rectangle, we find
Three-dimensional objects
So far we have discussed geometric objects lying in a plane, but many objects that we
want to measure are not flat. Now we turn to the study of three-dimensional objects.
First we consider volume.
Height = 5
The simplest volume to compute is that of a box. Labeling the dimensions of a
box as length, width, and height as shown in Figure 9.14, we calculate the volume
Length = 3 using
Width = 2
Volume of a box = Length × Width × Height
FIGURE 9.14 Dimensions of For example, the volume of the box in Figure 9.14 is
a box.
2 units × 3 units × 5 units = 30 cubic units
Note that the volume of a box can also be expressed as the area of the base times
the height:
Volume = Area of base × Height
h = height This formula also holds for any three-dimensional object with uniform cross sec-
tions, such as a cylinder. A cylinder has uniform circular cross sections, and a box
has uniform rectangular cross sections.
For the cylinder shown in Figure 9.15, the radius is r , so the area of the base is
r = radius
πr 2 . Using h for the height, we obtain
FIGURE 9.15 A cylinder. Volume = Area of base × Height = πr 2 h
Crauder-4100237 QL November 10, 2014 14:8
612 CHAPTER 9
© Glenda Powers/Dreamstime.com
SOLUTION
Because the pool is 6 feet across, the diameter of the circular base is 6 feet. The
diameter is given in feet and the height is given in inches, so we need to convert units
for one of them. Because the final answer is supposed to be in cubic inches, we change
the diameter of 6 feet to 72 inches. Then the radius is r = 36 inches, and the height
is h = 15 inches. The volume of the pool is
4 3
The volume of a ball of radius r is πr . For example, Earth has a radius of
3
about 4000 miles. So its volume is
4 3
Volume of Earth = πr
3
4
= π (4000)3
3
or about 268,082,573,106 cubic miles. The volumes of other geometric objects are
addressed in the exercises.
The analogue of perimeter in three-dimensional objects is surface area. We get
the surface area of a box with a lid by summing the areas of the six sides of the box.
The surface area of a ball of radius r is given by 4πr 2 . Looking once again at Earth,
we find
Surface area of Earth = 4πr 2
= 4π (4000)2
or about 201,061,930 square miles.
To find the surface area of a cylinder of radius r and height h (excluding the
top and bottom), think of the cylinder as a can that we split lengthwise and roll
out flat. See Figure 9.16. This gives a rectangle with width h and length equal to
the circumference of the circular base. The circumference of the base is 2πr , so the
surface area of the cylinder (excluding the top and bottom) is
Surface area of cylinder = 2πr h
mference
Circu
Length = circumference
Width = height
Height
FIGURE 9.16 Splitting the cylindrical side of a can and rolling it out flat.
Crauder-4100237 QL November 10, 2014 14:8
614 CHAPTER 9
Questions of this sort may invite many correct responses. We present one possibility:
Perhaps the easiest way to solve this is to think backward. Let’s start with any fixed
volume. We are going to use the same enclosing material to make the volume as large
as possible. We assume the boundary is totally flexible but not expandable. We may
change the shape of the enclosure but not its volume. We blow in air to stretch the
enclosure as much as possible. This is similar to blowing up a balloon, and we get
the same shape: a sphere.
Further questions of this type may be found in the What Do You Think section of
exercises.
8 Basket
9 2
C
10
6
4 13 Free-throw
5 line
616 CHAPTER 9
4 ft
5 9
3 ft
A b C
FIGURE 9.21 A window.
FIGURE 9.19
18. Carpet for a ramp. A ramp is 3 feet wide. It begins at the
9. Area of a right triangle. Verify that the triangle with sides 3, floor and rises by 1 foot over a horizontal distance of 10 feet.
4, and 5 shown in Figure 9.20 is a right triangle, and find its (See Figure 9.22.) How much carpet is needed to cover the
area. This is the 3-4-5 right triangle. (top of the) ramp?
5
3
1
4 3
10
FIGURE 9.20 The 3-4-5 right
triangle. FIGURE 9.22 A ramp.
10. Whole-number sides. Exercise 9 introduced the 3-4-5 right 19. Isosceles triangles. An isosceles triangle has two equal
triangle. Can you find other right triangles with whole-number sides. In an isosceles triangle, consider the line segment starting
sides? Hint: You are looking for whole numbers a, b, and c so at the vertex that touches both of the equal sides and meeting
that a 2 + b2 = c2 . the opposite side in a right angle. It is a fact that this segment
cuts the opposite side (the base) in half. Use this fact to find
11. Finding the area. The hypotenuse of a right triangle has
the area and perimeter of the isosceles triangle in Figure 9.23.
length 11, and a leg has length 4. Find the area of the triangle.
(The base is 10 units, and the height is 20 units.) Suggestion:
12–15. Heron’s formula. Exercises 12 through 15 use First use the Pythagorean theorem to find the lengths of the
Heron’s formula. unknown sides of the triangle.
12. Find the area of a triangle with sides 7, 8, and 9. C
13. Find the area of a garden that is a triangle with sides 4 feet,
5 feet, and 7 feet.
14. Find the area of a triangle with sides 6 yards, 6 yards, and
7 yards.
15. One triangular plot has sides 4 yards, 5 yards, and 6 yards.
Another has sides 3 yards, 6 yards, and 7 yards. Which plot en-
closes the larger area? 20
16. Making drapes. A window in your home has the shape
shown in Figure 9.21. The top is a semi-circle (half of a circle).
To buy material for drapes, you need to know the area of the
window. What is it?
17. Trimming a window. How many feet of moulding do we A 10 B
need to make a border (excluding the diameter of the half- FIGURE 9.23 Isosceles
circle) for the window in Figure 9.21? triangle.
Crauder-4100237 QL November 10, 2014 14:8
2
A d
D
9 c
a B
FIGURE 9.27
24. Pitcher foam. A restaurant serves soda pop in cylindrical
4
pitchers that are 5 inches in diameter and 8 inches tall. If the
A 7
D pitcher has a 1-inch head of foam, how much soda is lost as a
B
result?
25. Soda and soup. This is a continuation of Exercise 24. A
FIGURE 9.24
typical soup can is about 2.5 inches in diameter and 3.5 inches
21. Refer to the triangle in Figure 9.25. tall. Compare the amount of soda lost in Exercise 24 with the
volume of a soup can.
C 26–29. A soup can. A soup can has a diameter of 2 58 inches
and a height of 3 34 inches. Exercises 26 through 29 refer to this
can.
26. Find the volume of the soup can.
c 27. How many square inches of paper are required to make
10
a the label on the soup can?
28. How many square inches of metal are required to make the
soup can?
2 29. When you open the soup can, how far does the can opener
A 5
D travel?
B 30. Distance from Dallas to Fort Smith. Dallas, Texas, is 298
miles due south of Oklahoma City, Oklahoma. Fort Smith,
FIGURE 9.25 Arkansas, is 176 miles due east of Oklahoma City. How many
miles long is a plane trip from Dallas directly to Fort Smith?
22. Refer to the triangle in Figure 9.26.
31. Houston to Denver. Houston, Texas, is 652 miles due
south of Kansas City, Missouri. Denver, Colorado, is 558 miles
C
due west of Kansas City. How many miles long is a plane trip
from Houston directly to Denver? Round your answer to the
nearest mile.
32. Big-screen TVs. The size of a television set is usually clas-
sified by measuring its diagonal. Suppose that one giant tele-
c b vision measures 5 feet long by 3 feet high, and another is a
6
4.5-foot square.
a. Which television has the larger diagonal?
2 b. Which television has the larger screen in terms of area?
A 5 33–35. Monitor sizes. The size of a computer monitor (also
D
called a display) is usually measured in terms of the diagonal of
B
the rectangle. Exercises 33 through 35 explore the implications
FIGURE 9.26 of this fact.
33. A monitor measures 12.6 inches by 16.8 inches. What is
23. Refer to the triangle in Figure 9.27. the length of the diagonal?
Crauder-4100237 QL November 10, 2014 14:8
618 CHAPTER 9
34. A monitor measures 10.3 inches by 18.3 inches. What is 39–43. Getting ready for spring. My lawn has the shape
the length of the diagonal? shown in Figure 9.30, where the dimensions are in feet. Ex-
35. Calculate the areas of the rectangles described in Exercises ercises 39 through 43 concern my lawn.
33 and 34. (Round your answers to the nearest whole number.)
In light of your answers to those exercises, does measurement 20 50
of a monitor in terms of the diagonal give complete information
about the size? p 16
36. Pizza coupon. At your local pizza place, a large pizza costs
$10.99. You have a coupon saying you can buy two medium
pizzas for $12.99. A medium pizza has a diameter of 12 inches,
and a large pizza has a diameter of 14 inches. Which is the bet-
ter value (measured in dollars per square inch), one large pizza 80
or two medium pizzas with the coupon? q
37–38. Taxes and square footage. Property taxes on a home
are often based on its valuation, determined by the number of
square feet of living space. Incorrect measurements of this area
can significantly affect your property taxes. Exercises 37 and
38 concern square footage of a home.
37. Suppose the floor plan for a home is shown in Figure 9.28.
(The curved part is a half-circle.) How many square feet of floor
space does this home have?
70 40
20
FIGURE 9.28 Floor plan. 42. Find the area of the triangle at the bottom of the figure.
(This is not a right triangle.)
38. The floor plan for a home is shown in Figure 9.29. How 43. I want to fertilize my lawn with bags of fertilizer that cover
many square feet of floor space does this home have? 500 square feet each. Use your answer to Exercise 42 to deter-
mine how many bags of fertilizer I need to buy.
44. Area of Oklahoma. Like many states, Oklahoma has an ir-
60 regular shape, as shown in Figure 9.31. This makes its area dif-
ficult to calculate exactly. The following table shows distances
between certain towns in Oklahoma.
© Michael Schmeling/Alamy
50
Kenton Buffalo Miami
40 40
Felt
50
60
Broken Bow
1 2
Volume = πr h
3
Needles, CA Surface area = πr r 2 + h2
620 CHAPTER 9
Note that the surface area does not include the top circle of the the three interior lines shown. You should have three pieces of a
cone. Exercises 59 through 62 use these formulas. puzzle. Reassemble the three pieces into the figure shown in Fig-
ure 9.35. Explain how this demonstrates why the Pythagorean
theorem is true.
Radius = r
c
Height = h a
b
c b c
a
FIGURE 9.33 A cone.
c
59. A cone for ice cream has a top diameter of 2.25 inches and
a height of 4.5 inches. How much ice cream will this cone hold? FIGURE 9.34 Trace and cut to make a
Pythagorean puzzle.
60. A (solid) conical spire is made of concrete. The spire has a
radius of 3 feet and a height of 10 feet. How much concrete is
needed to make the spire? b
61. How many square feet of paint are needed to paint a cone
with a top diameter of 5 feet and a height of 6 feet, assuming
that you will paint the entire cone including the top circle? a
b
62. A waffle cone has a top diameter of 2.25 inches and a
height of 4.5 inches. What is the area of the dough needed to a a
make the cone?
63. A demonstration of the Pythagorean theorem. This exer- b
cise uses a puzzle to demonstrate why the Pythagorean theo- a
rem is true. Trace the square shown in Figure 9.34 onto your FIGURE 9.35 Reassemble the puzzle to
own paper. Cut out the square and then cut the square along make this picture.
W-2. Rectangle. Does the perimeter of a rectangle determine its area? This question
can also be asked the other way around: Does the area of a rectangle determine its
perimeter? What if it is a square? Explain your answer and include examples.
W-3. A cube. Does the surface area of a box determine its area? What if it is a cube—
where all three dimensions (length, width, and height) are the same? Explain your
answer and include examples.
W-4. An interesting project. Many measurements are difficult to make because it is
not practical to stretch a tape between certain points. Here is an example. Figure out
Crauder-4100237 QL November 10, 2014 14:8
how far it is from a bottom corner of your classroom to the diagonally opposite cor-
ner at the ceiling. This is not a distance along a wall or floor, but such measurements
may be helpful to you. You may find the Pythagorean theorem to be useful. Once
you have completed this project, try to figure out the distance from a ground-level
corner of your classroom building to the diagonally opposite corner at the roof.
Exercise 5 provides additional information relevant to this topic.
W-5. Scarecrow. In the 1939 film The Wizard of Oz, the Scarecrow states the fol-
lowing as a fact about geometry: “The sum of the square roots of any two sides of
an isosceles triangle is equal to the square root of the remaining side.” An isosceles
triangle has two sides of equal length. To test the statement of the Scarecrow, con-
sider the case of a right triangle whose legs have equal length. What is the length of
the hypotenuse in this case? Is it true that the sum of the square roots of the two
legs is equal to the square root of the hypotenuse? If not, how can you adjust the
Scarecrow’s statement to make it true for isosceles right triangles? Does your ad-
justed statement hold for all isosceles triangles?
Exercise 19 provides additional information relevant to this topic.
W-7. A brain teaser. A long rope is stretched tightly around the equator of Earth. How
much additional rope would be needed if the rope ring is to be suspend a uniform
distance of 1 foot above the equator? How would things change if we were discussing
the equator of Mars rather than Earth? If you need help, look at Example 9.1.
W-8. Heron’s formula. One reason that Heron’s formula looks complicated is the
square root, but that part of the formula is suggested by analyzing the units: If the
sides a, b, c of the triangle are measured in feet, what are the units for the semi-
perimeter S? Then what are the units for S(S − a)(S − b)(S − c)? What should the
units for the area be? Use your answers to explain why the square root is reasonable.
Exercise 13 provides additional information relevant to this topic.
W-9. Cylinder. Recall that the volume of a cylinder of radius r and height h is πr 2 h.
Start with a cylinder having radius 1 foot and height 1 foot. If the radius is doubled
from 1 foot to 2 feet, but the height remains at 1 foot, what happens to the volume?
If the height is doubled from 1 foot to 2 feet, but the radius remains at 1 foot, what
happens to the volume? Which has a larger effect on the volume, doubling the radius
or doubling the height? Explain your answer by looking at the way r and h appear
in the formula for the volume.
Crauder-4100237 QL November 13, 2014 12:18
622 CHAPTER 9
N ews: construction has begun on the Lego visitor centre in Billund, Denmark,
designed by architect Bjarke Ingels, who told Dezeen that Legobricks pro-
vide the ultimate tool for testing architectural ideas (+ interview).
Six Lego-shaped foundation stones were laid yesterday to mark the ground-
breaking of the project, which is due for completion in 2016.
Bjarke Ingels, the founder of Danish firm BIG, used the basic two-by-four Lego
brick as the starting point for his design of the Lego House, which will provide a
visitor attraction in the town where Lego was invented back in 1932.
“Lego proportions are really the golden ratio of architecture,” Ingels told
Dezeen. “I think for testing ideas quickly it can be quite powerful.”
The opening article is concerned with ratio and proportion in a geometric setting.
Proportionality relationships are quite common in many settings, both numerical and
geometrical. For example, they arise when we convert from one unit of measurement
to another. In this section, our purpose is to examine proportionality and similarity
Cartoonstock.com
relationships that arise in geometry. In the process we will look more closely at the
golden ratio, which is at the heart of much controversy regarding the interplay of art
and mathematics.
1
Note that if A = cB, we can turn the relationship around and write B = A,
c
1
which says that B is proportional to A with constant of proportionality .
c
A
Another way to write the equation A = cB is = c. Expressed in this way, the
B
proportionality formula tells us that two quantities are proportional if their ratio is
constant.
For example, if you are driving at a constant speed of 60 miles per hour, then
the distance you travel is proportional to the time you spend driving. If distance is
measured in miles and time is in hours,
Distance = 60 × Time
The constant of proportionality is your speed, 60 miles per hour. Note that distance
traveled increases by the same amount, 60 miles, for each extra hour of travel. We
can also write the relationship as
Distance
= 60
Time
Note that distance traveled is proportional to time only when your speed is con-
stant. If your speed varies, distance is not proportional to time.
You can probably think of lots of quantities that are not proportional to each
other. One example would be your height and your age. Although your height in-
creases as you get older, it is not proportional to your age because you don’t grow
by the same amount each year.
One of the most important instances of proportionality in geometry is the familiar
formula for the circumference of a circle:
Circumference = π × Diameter
This formula tells us that a circle’s circumference is proportional to its diameter and
that the number π is the constant of proportionality. Expressed as a ratio, this rela-
tionship is
Circumference
=π
Diameter
It was pointed out in Section 9.1 that if we divide the circumference of a very
small circle by its diameter, we get exactly the same value, π , as we would if we
divided the length of the equator of Earth by its diameter. This is another way of
saying that the circumference is proportional to the diameter.
In contrast, the radius r and area A of a circle are related by
A = πr 2
Here, A is not a constant multiple of r , so the area is not proportional to the radius.
Because the diameter is twice the radius, the area of a circle is not proportional to its
diameter either.
624 CHAPTER 9
simply by running a tape measure around the tree. Is the diameter of a tree propor-
tional to its circumference? (Assume that the cross section is circular in shape, as in
Figure 9.36.) If so, what is the constant of proportionality?
ence = π x d
fer i
m
am
u
Circ
eter
Diameter
SOLUTION
We know that circumference is proportional to diameter, and the relationship is
Circumference = π × Diameter
1
Diameter = × Circumference
π
Thus, the diameter is proportional to the circumference, and the constant of propor-
tionality is 1/π.
Properties of proportionality
We can discover an important property of proportionality by further consideration
of tree measurement, as in the previous example. Foresters determine the diameter of
a tree by measuring its circumference. What happens to the circumference of a tree
if the diameter is doubled? We can find out by multiplying both sides of the equation
Circumference = π × Diameter
by 2. The result is
2 × Circumference = π × (2 × Diameter)
Crauder-4100237 QL November 10, 2014 14:8
Thus, doubling the diameter causes the circumference to double as well. Con-
versely, a tree with twice the circumference of another has twice the diameter. The
same is true for any multiple—tripling the diameter causes the circumference to triple,
and halving the diameter cuts the circumference in half. It turns out that this fact is
true for any proportionality relation, and it is another characterization of propor-
tionality.
Another example involves a square. Recall that the perimeter of a square of side S
is given by P = 4S. This formula says that the perimeter of a square is proportional
to the length of one side, with constant of proportionality 4. Assuming that the cost
of a square wooden picture frame is proportional to the length of wood used, we
can say that the cost of wood for framing the picture is proportional to the length of
a side.
Crauder-4100237 QL November 10, 2014 14:8
626 CHAPTER 9
PhotoDisc
This somewhat confusing statement can be restated to give a definition of the golden
ratio.
Longer Shorter φ 1
Whole φ+1
A formula for the golden ratio can be found with a bit of algebra. If we let the
shorter length be 1 unit and the longer be φ units, as shown in Figure 9.37 and
Figure 9.38, we find:
Longer Whole
=
Shorter Longer
φ φ+1
=
1 φ
φ2 = φ + 1
φ2 − φ − 1 = 0
We can use the quadratic formula from algebra to solve this equation for φ.
Recall that the quadratic formula says that if ax2 + bx + c = 0, then
−b ± b2 − 4ac
x= . Our equation is φ 2 − φ − 1 = 0, so in the quadratic
2a
formula we put φ in place of x and use a = 1, b = −1, and c = −1. The
quadratic formula allows us to conclude that
1 ± (−1)2 − 4 × 1 × (−1) 1± 5
φ= =
2×1 2
We discard the negative solution. The positive solution is
1+ 5
φ= ,
2
which is about 1.62.
The number φ arises in a number of settings, but perhaps its best-known occur-
rence is in the context of a golden rectangle.
Length
Width
Length
FIGURE 9.39 A golden rectangle: = φ.
Width
Crauder-4100237 QL November 10, 2014 14:8
628 CHAPTER 9
Length
=φ
Width
we have
Length = Width × φ
Because the width is 3 feet, the length is 3 × φ feet. This is about 3 × 1.62 or roughly
4.9 feet.
Similar triangles
Unlike golden rectangles, similar triangles are not likely to become part of the popular
culture. But they are examples of proportionality and form the basis of a great deal
of important mathematics, including trigonometry. They are also very important in
applications of mathematics.
Crauder-4100237 QL November 10, 2014 14:8
Figure 9.40 shows similar triangles. For comparison, Figure 9.41 shows triangles
that are not similar.
As is made clear from these figures, similar triangles have the same shape. The
most important feature of similar triangles is that the ratios of corresponding sides are
the same. That is, the lengths of corresponding sides are proportional. This key fact
is illustrated in Figure 9.42. In the notation of that figure, the similarity relationship
can be expressed using the equations
A B C
= =
a b c
A typical use of similarity can be illustrated using Figure 9.43. Let’s find the miss-
ing lengths B and c in the figure. First we find B:
A B
=
a b
12 B
=
3 2
B
4=
2
8=B
b c
c
B C b=2 B C = 16
a
a=3
A B C A = 12
= = A
a b c
630 CHAPTER 9
SOLUTION
We can represent the man and the building using vertical lines. Thus, the two triangles
in Figure 9.44 each have a right angle at their base. Because we are measuring the
shadows at the same time of day, the angles of sunlight with the vertical (marked α
Crauder-4100237 QL November 10, 2014 14:8
25
β
6
FIGURE 9.44 Using similar triangles: Shadows.
2
and β) are the same for both the man and the building. Because the two triangles have
two pairs of equal angles, the triangles are similar. Hence, the ratios of corresponding
sides are the same. In the figure, we have marked the height of the building as H.
Therefore,
H 25
=
6 2
25
H = 6×
2
H = 75
632 CHAPTER 9
Suppose we have several boxes of different sizes, each of which is in the shape of a
cube.
a. Is the volume of a box proportional to the length of one side of the box, the
square of the length, or the cube of the length? Find the constant of
proportionality.
b. One of our boxes has sides that are twice as long as those of another box.
How much more does the larger box hold than the smaller box?
c. Is the surface area of a box proportional to the length of one side of the box,
the square of the length, or the cube of the length?
d. The cost of wrapping paper for a box is proportional to the surface area.
What happens to the cost of wrapping paper if the sides of the cube are doubled
in length?
SOLUTION
a. Let x denote the length of one side of the cube, and let V denote the volume.
Because the box is cubical in shape, the length, width, and height are all the
same, x. Therefore,
V = Length × Width × Height
V = x×x×x
V = x3
Thus, the volume is proportional to the cube of the length, and the constant of
proportionality is 1.
b. As in part a, we denote the length of a side by x. If the sides are doubled, we
replace x by 2x. This gives a new volume of
(2x)3 = 23 x3 = 8x3
The other volume was x3 . Thus, the larger volume is 8 times the smaller volume.
c. The amount of paper needed is the surface area of the box. The surface of a
cube has six square “faces.” If x represents the length of a side, the area of one
face is x × x = x2 . Therefore, the area of the surface of the cube is 6x2 . This
means that the surface area is proportional to the square of the length, with
constant of proportionality 6.
d. We know that the cost is proportional to the surface area. To find the surface
area of the larger cube, we replace x by 2x in the formula we used in part c:
New surface area = 6 × (2x)2
= 4(6x2 )
The larger surface area is 4 times the other surface area. So 4 times as much
wrapping paper will be needed for the larger cube as for the smaller cube.
Crauder-4100237 QL November 10, 2014 14:8
Therefore, the cost of wrapping paper is multiplied by 4 when the sides are
doubled in length.
SOLUTION
Classic poster of King Kong with a. Let h denote the height of an ape and W the weight. Because the weight is
Fay Wray (from the original proportional to the cube of the height, we have the formula
movie released April 7, 1933).
W = ch3
Here, c is the constant of proportionality. If the height is multiplied by 100, we
replace h by 100h. This gives a new weight of
c(100h)3 = 1003 × ch3
The original weight was ch3 . Thus, the new weight is 1003 times the original
weight. In other words, the weight increases by a factor of 1003 = 1,000,000
(one million).
b. Let A denote the cross-sectional area of a limb and, as in part a, let h denote
the height. Because the cross-sectional area is proportional to the square of the
height, we have the formula
A = Ch2
Here, C is the constant of proportionality. If the height is multiplied by 100, we
replace h by 100h. This gives a new area of
C(100h)2 = 1002 × Ch2
The original area was Ch2 . Thus, the new area is 1002 times the original area.
In other words, the area increases by a factor of 1002 = 10,000 (ten thousand).
c. Because the pressure on a limb is the weight divided by the cross-sectional
area, the pressure will increase by a factor of
Factor by which weight increases 1003
= = 100
Factor by which area increases 1002
The pressure on a limb of the overgrown ape is 100 times the original pressure.
Crauder-4100237 QL November 10, 2014 14:8
634 CHAPTER 9
The tremendous increase in pressure on a limb means that the overgrown ape
would collapse under its own weight. Science fiction aside, such scaling arguments
are used by biologists to study the significance of the size and shape of organisms.
Exercise 53 provides one example.
3. Comparing areas. If the radius of a circle is tripled, what 6–15. Which are proportional? Exercises 6 through 15 ask
happens to the area? about proportionality relationships. For those that describe
proportionality relationships, give the constant of proportion-
4. The surface area of a sphere. The surface area of a sphere of ality. In some cases, your answer may depend on your interpre-
radius r is given by S = 4πr 2 . If the radius is multiplied by 2, tation of the situation.
how is the surface area affected?
Crauder-4100237 QL November 10, 2014 14:8
6. If each step in a staircase rises 8 inches, is the height of the titude to the ratio of PHI [sic, φ] to 1. . . . So the an-
staircase above the base proportional to the number of steps? cients assumed PHI must have been preordained by the
7. Suppose that hamburger meat costs $2.70 per pound. Is the Creator. . . .
cost of the hamburger meat proportional to its weight? Langdon grinned. “Ever study the relationship between
female and males in a honeybee community?”
8. Is the volume of a balloon proportional to the cube of its
“Sure. The female bees always outnumber the male
diameter?
bees.”
9. Is the volume of a balloon proportional to its diameter? “And did you know that if you divide the number of
10. Is the area of a pizza proportional to its diameter? female bees by the number of male bees in any beehive
in the world, you always get the same number?”
11. Is the volume of a pizza with diameter 12 inches propor-
“You do?”
tional to its thickness?
“Yup. PHI.”
12. Is the area of land in a square lot proportional to the length The girl gasped. “No way!”
of a side?
13. Is the area of a rectangular TV screen proportional to the There are various resources, including the Web, available
length of its diagonal? for doing research about bees. Does your research bear out the
claim that bees “adhered with eerie exactitude to the ratio of φ
14. Consider a box with a square base and a height of 2 feet.
to 1”?
Is the volume proportional to the area of the base?
15. The surface area S of a sphere of radius r is given by 26. Do some research on the use of the golden rectangle in art
S = 4πr 2 . Is the surface area of a balloon proportional to the and write a report.
square of its diameter? 27. Do some research on the use of the golden rectangle in ar-
16. Circle. Recall that the area of a circle of radius r is given by chitecture and write a report.
the formula A = πr 2 . Your friend says that this formula shows 28–32. Similar triangles. Exercises 28 through 32 deal with
that the area is proportional to the radius with proportional- similarity of triangles.
ity constant πr because the area is πr times r . Is your friend
28. The triangles in Figure 9.45 are similar. Find the sides la-
correct? Explain.
beled A and B.
17–27. The golden ratio. Exercises 17 through 27 concern
29. The triangles in Figure 9.46 are similar. Find the sides la-
the golden ratio φ.
beled A and b.
17. The shorter side of a golden rectangle is 5 inches. How long
is the longer side? Round your answer to one decimal place.
18. The longer side of a golden rectangle is 10 inches. How
long is the shorter side? Round your answer to one decimal
place. A
19. Measure your television screen. What is the ratio of the 3 7
longer side to the shorter? Is the screen a golden rectangle?
8
20. Measure your computer screen. What is the ratio of the
longer side to the shorter? Is the screen a golden rectangle? B
21. Is this textbook a golden rectangle?
22. It has been observed that the distance from a person’s
shoulder to the fingertips divided by the distance from the el- 2
bow to the fingertips is close to φ. Take these measurements
of your own arm and calculate the ratio. Is your answer close FIGURE 9.45
to φ?
23. It has been observed that a person’s height divided by the
distance from the person’s belly button to the floor is close to
φ. Take these measurements of your own body and calculate A
the ratio. Is your answer close to φ?
3 7
24. It has been observed that the distance from a person’s hip
to the floor divided by the distance from the knee to the floor
9
is close to φ. Take these measurements of your own leg and 4
calculate the ratio. Is your answer close to φ?
25. An excerpt from the 2003 best-selling novel The Da Vinci
Code goes as follows: b
Plants, animals and even human beings all possessed
dimensional properties that adhered with eerie exac- FIGURE 9.46
Crauder-4100237 QL November 10, 2014 14:8
636 CHAPTER 9
30. A pole casts a 4-foot shadow, and at the same time of day 43. If the cylinder is a can of tuna and the radius is held con-
a 6-foot-tall man casts a 1-foot shadow. How tall is the pole? stant, the area of the label is proportional to:
31. A 10-foot-tall tree casts a 4-foot shadow, and at the same a. r b. r 2 c. h d. π
time of day a girl casts a 2-foot shadow. How tall is the girl? What is the constant of proportionality?
32. A tennis player stands 39 feet behind the net, which is 3.5 44. Two tuna cans have a radius of 2 inches. One is twice as
feet tall. He extends his racket to a height of 9 feet and hits a tall as the other. How do the sizes of the labels on the cans
serve. The ball passes 6 inches above the net. How far beyond compare?
the net does the tennis ball strike the court? Suggestion: If d is 45. If the height is held constant, the surface area of a can is
the distance past the net, the distance from the server to where proportional to:
the ball strikes the court is 39 + d.
a. r b. r 2 c. h d. π
33. Doubling an equilateral triange. It is a fact that the area of What is the constant of proportionality?
an equilateral triangle is proportional to the square of a side. If
all sides of an equilateral triangle are doubled, what happens 46. Two soup cans have a height of 4 inches. One can has 3
to the area? times the radius of the other. How do their labels compare in
area?
34. Boxes of fixed height. Suppose a box has a fixed height
of 9 feet and a square base. Explain why the length of one of 47. If we include the top and bottom of a cylindrical can, the
the sides of the base is proportional to the square root of the total surface area is A = 2πr h + 2πr 2 . If the height is held
volume. What is the constant of proportionality? constant, the total surface area is proportional to:
a. r d. π
35–38. Volume of a sphere. The volume V of a sphere of
4 3 b. r 2 e. none of these
radius r is V = πr , and the surface area S is given by
3 c. h
S = 4πr . Exercises 35 through 38 make use of these formulas.
2
48–52. A cone. Suppose a cone has height h and the radius of
35. The amount of gas in a balloon of radius r is propor- 1 2
tional to: πr h.
its circular base is r . Then the volume is given by V =
3
a. r b. r 2 c. r 3 d. π The surface
area (not counting the circular base) is given by
What is the constant of proportionality? A = πr r 2 + h2 . Exercises 48 through 52 make use of these
36. A child wants a balloon of twice the radius of the offered formulas.
balloon. How does the increase in radius affect the amount of 48. Varying radius. If the height of a cone is held constant, the
gas in the balloon? volume is proportional to:
37. The amount of paint needed to color a styrofoam ball of a. r d. π
radius r is proportional to:
b. r 2 e. none of these
a. r b. r 2 c. r 3 d. π
What is the constant of proportionality? c. h
What is the constant of proportionality?
38. A softball has a radius 1.33 times that of a baseball. How
many times as much material is required to cover a softball 49. Two ice cream cones have the same height. One’s radius
than a baseball? is twice the radius of the other. Compare the amounts of ice
cream the cones can hold.
39–47. A cylinder. Recall that a cylinder with radius r and
height h has volume V = πr 2 h and surface area S = 2πr h (not 50. If the radius of a cone is held constant, the volume is pro-
counting the top and bottom). Exercises 39 through 47 make portional to:
use of these formulas. a. r d. π
39. If the height of a cylindrical can is held constant, the vol- b. r 2 e. none of these
ume is proportional to:
c. h
a. r b. r 2 c. h d. π
What is the constant of proportionality?
What is the constant of proportionality?
51. Two ice cream cones have the same radius. One has twice
40. Two soup cans have a height of 4 inches. One can is twice the height of the other. Compare the amounts of ice cream the
the radius of the other. How does the amount of soup in the cones can hold.
cans compare?
52. If the radius of the circular base is held constant, the sur-
41. Two soup cans have a radius of 2 inches. One can is twice face area is proportional to:
the height of the other. How does the amount of soup in the
a. r d. π
cans compare?
b. r 2 e. r 2 + h2
42. If the radius is held constant, the volume of a cylindrical
can is proportional to: c. h f. none of these
a. r b. r 2 c. h d. π
What is the constant of proportionality?
Crauder-4100237 QL November 10, 2014 14:8
53. Basal metabolic rate. The basal metabolic rate (BMR) of ences. Summarize the arguments used by Galileo and explain
an animal is a measure of the amount of energy it needs for how he applied them.
survival. An animal with a larger BMR will need more food to 56. A different point of view. In The News 9.3 is an
survive. One simple model of the BMR states that among ani- excerpt from the full article that appears at http://www
mals of a similar shape, the rate is proportional to the square of .maa.org/external archive/devlin/devlin 06 04.html. It refers
the length. (This model assumes that the BMR is proportional to the 2003 novel The Da Vinci Code, which was later made
to the surface area.) If one animal is 10 times as long as another, into a movie. Although the article was written in 2004, it
how do the BMR values compare? presents a skeptical view of some of the common stories about
54. History. The idea of proportionality has a long history in the golden ratio, a view that is still relevant.
geometry. Explain how the idea of proportion is defined by Eu- Try to find other sources that either perpetuate stories
doxus and used in Euclid’s Elements. about the golden ratio in art and architecture or that back up
55. History. Scaling arguments such as those used in Exam- the skepticism. Write a report on your findings.
ple 9.14 were given by Galileo in his 1638 book Two New Sci-
T he enormous success of Dan Brown’s novel The Da Vinci Code has introduced
the famous Golden Ratio . . . to a whole new audience. Regular readers of this
column will surely be familiar with the story. The ancient Greeks believed that
there is a rectangle that the human eye finds the most pleasing. . . .
Having found this number, the story continues, the Greeks then made exten-
sive use of the magic number in their architecture, including the famous Parthenon
building in Athens. Inspired by the Greeks, future generations of architects like-
wise based their designs of buildings on this wonderful ratio. Painters did not lag
far behind. The great Leonardo Da Vinci is said to have used the Golden Ratio to
proportion the human figures in his paintings—which is how the Golden Ratio
finds its way into Dan Brown’s potboiler.
It’s a great story that tends to get better every time it’s told. Unfortunately,
apart from the fact that Euclid did solve the line division problem in his book
Elements, there’s not a shred of evidence to support any of these claims, and
good reason to believe they are completely false. . . .
638 CHAPTER 9
questions. What happens if both are doubled? What happens if either the length or
the width is doubled but the other stays the same?
Exercise 12 provides additional information relevant to this topic.
W-3. How the volume of a box changes. How is the volume of a box affected if the
sides change by the same factor? First determine what happens if all the sides are
doubled. Then consider what happens if all the sides are multiplied by 3, 4, . . .
Exercise 14 provides additional information relevant to this topic.
W-4. The golden rectangle. Much has been written regarding the role of the golden
rectangle in science, art, and philosophy. There are lots of opinions, but there is little
agreement. Do your own research regarding the golden rectangle and report on your
findings.
Exercise 25 provides additional information relevant to this topic.
W-5. Similarity and surveying. Similarity of triangles plays an important role in sur-
veying. Give examples of how you might use similarity to produce measurements
that are difficult to make otherwise. For example, explain how to use a shadow to
determine the height of a building.
Exercise 30 provides additional information relevant to this topic.
W-6. Square. Show that the perimeter of an equilateral triangle is proportional to the
length of a side, and find the constant of proportionality. Show that the perimeter of a
square is proportional to the length of a side, and find the constant of proportionality.
Can you generalize these two results to other figures?
Exercise 33 provides additional information relevant to this topic.
640 CHAPTER 9
180˚
90˚
90˚
45˚
Now consider a rectangle that is not a square. This object does not have 90 de-
grees rotational symmetry because rotation by 90 degrees does not leave the rectangle
exactly the same as it was originally. It does have 180-degree rotational symmetry
about its center, however. These facts are illustrated in Figure 9.48.
Another example of rotational symmetry is provided by a circle. A circle has
complete rotational symmetry because if we rotate it any number of degrees about
the center of the circle, the result is exactly the same as the original circle. By way of
comparison, the triangle in Figure 9.49 has no rotational symmetries at all.
Many familiar objects have rotational symmetries, as the following photos show.
PhotoDisc
© Novadomus/Dreamstime.com
© Mableen/istockphoto/Getty Images
The pictured starfish has 60-degree rotational symmetry, the lunaria has 90-degree rotational symmetry, and the dahlia has
many more rotational symmetries.
Figure
L
L
Reflected
figure
FIGURE 9.53 This FIGURE 9.54 This star
FIGURE 9.52 Reflection is shape has reflectional has reflectional
the same as flipping the symmetry about the symmetry about five
plane about a line. line shown. different lines.
For example, the dart shape in Figure 9.53 has reflectional symmetry about the
line shown. Similarly, the triangle in Figure 9.49 has reflectional symmetry about a
horizontal line that bisects the angle at the left-hand side of the picture. Note that
neither of these figures shows any rotational symmetry. The five-pointed star in Fig-
ure 9.54 has reflectional symmetry about each of the five lines shown there.
Reflectional symmetries are shown in many familiar settings, including the faces
of animals.
Crauder-4100237 QL November 10, 2014 14:8
FourOaks/Getty Images
642 CHAPTER 9
A regular polygon is a polygon in which all sides are of equal length and all angles
have equal measure. Such polygons can be used to make a special kind of tiling.
α More than
α α 120˚
Less than
α α 120˚ left
α More than
120˚
To see which other regular polygons might be used to make a regular tiling of the
plane, let’s first list the sizes of angles in some regular polygons.
Regular polygons with more than six sides have angles larger than 120 degrees.
Hence, they cannot regularly tile the plane.
644 CHAPTER 9
As a contrast to the preceding example, you are asked in Exercise 16 to show that
there is no regular tiling of the plane by a regular pentagon.
Irregular tilings
Regular tilings must satisfy two conditions: They use a single regular polygon, and
the same configuration of edges must occur at each vertex. Tilings that are not regular
Marbury/iStock/360/Getty Images
fail to meet one of these two conditions. One classic example of a tiling that is not
regular is shown in the accompanying photo of floor tiles. This tiling does satisfy the
second condition, so it is semi-regular.
Figure 9.62 shows how to cut a regular hexagon into three (nonregular) pen-
tagons. We saw in Try It Yourself 9.17 that the plane can be tiled by regular hexagons.
Then such a tiling automatically gives an irregular tiling by pentagons. See Fig-
ure 9.63.
A semi-regular tiling.
The idea suggested by Figure 9.63—that we can make new tilings by subdividing
old ones—is fruitful. We will use it to show how to use any triangle to tile the plane.
We begin with the triangle shown in Figure 9.64. The first step is to note that we
can put together two copies of this triangle to make a parallelogram, as shown in
Figure 9.65. Now we use parallelograms to tile the plane as shown in Figure 9.66,
and this gives the required tiling by triangles.
EXAMPLE 9.18 Making new tilings from old: Three given pieces
Show that the three pieces in Figure 9.67 can be used to tile the plane.
SOLUTION
Note that the three pieces go together to make the regular hexagon shown in Fig-
ure 9.68. We know that regular hexagons tile the plane, and that gives a tiling using
the three pieces shown.
FIGURE 9.69
The answer is provided at the end of this section.
Escher tilings
Some of the best known tilings of the plane are those of M. C. Escher. Several of
his famous tilings are displayed here. They are interesting from both a mathematical
and an artistic point of view. They are much more elaborate than any tilings we have
produced, but they are constructed using similar ideas. Let’s show how we can make
tilings using unusual shapes.
MC Escher
MC Escher
646 CHAPTER 9
Let’s start with the regular tiling by squares shown on the left in Figure 9.70. Now
from each square, let’s remove a puzzle tab from the bottom and add it to the top of
the square. The resulting tiling piece is the irregular shape shown at the bottom of
Figure 9.70. We get the new tiling of the plane shown on the right in Figure 9.70.
A little thought can lead one to produce truly unusual tiling pieces. Many of
Escher’s prints use this idea as well as more sophisticated ones. Some start with
hexagons rather than squares.
Try It Yourself 9.18: Making new tilings from old: Three given pieces The three
pieces fit together to make a rectangle, which can tile the plane.
FIGURE 9.74
648 CHAPTER 9
1 2
21. Our goal here is to explain why the tiling in Figure 9.76 is
nonperiodic.
a. How many points of rotational symmetry does the tiling φ φ
in Figure 9.76 have? That is, how many points can we take φ φ
for the center of a rotation less than 360 degrees that leaves
the figure unchanged?
b. Can a tiling with exactly one point of rotational sym- 1 1 1 1
metry be periodic?
FIGURE 9.80 A dart with dots. FIGURE 9.81 A kite with dots.
c. Explain using parts a and b why the tiling in Figure 9.76
is not periodic.
a 72° 72°
36° 36° d
1 1 1 1
23. Penrose tiling: A periodic tiling with kites and darts. Show
that a kite and a dart can be fitted together to make a rhombus,
each side of which is φ. Explain why this gives a periodic tiling
of the plane using kites and darts.
24. Penrose tiling: Experimenting with kites and darts. Kites
and darts come with assembly rules, which are as follows:
Assembly Rule 1: Pieces must be assembled along matching
edges. FIGURE 9.84 A nonperiodic tiling using
kites and darts by Roger Penrose.
Crauder-4100237 QL November 10, 2014 14:8
W-2. Reflect. Can a figure have reflectional symmetry about a line that does not touch
the figure? Explain.
Exercise 6 provides additional information relevant to this topic.
W-3. Jigsaw. Do the pieces of a jigsaw puzzle make a regular tiling? Explain.
Exercise 16 provides additional information relevant to this topic.
W-4. Is symmetry beautiful? Some sources, such as the article from In The News 9.4,
suggest that part of the human perception of beauty is tied to symmetry. Do some
research on this topic and report on what you find. Is your own conception of beauty
related to symmetry?
W-5. Escher prints. Many prints by M. C. Escher are of interest from both math-
ematical and artistic perspectives. Find examples on the Web, and report on your
reaction to their artistic quality.
W-6. Do it yourself. A number of tilings of the plane are discussed in this section,
and many others are available through other sources. It can be fun to come up with
your own, and interesting tilings can be found by anyone. Make several examples
that you like.
Exercise 20 provides additional information relevant to this topic.
W-7. Roger Penrose. Sir Roger Penrose is famous not only for his nonperiodic tilings
of the plane but also for his work as a mathematical physicist. Read about him on
the Internet and report on his life and accomplishments.
Exercise 24 provides additional information relevant to this topic.
W-8. Reflection and rotation. A classmate says that if a shape has a reflectional sym-
metry, then it has a rotational symmetry of 180 degrees. Is that true? Explain your
answer.
W-9. Regular polygons. Look at the table on page 643 showing angle sizes for regular
polygons. Note that the sum of the angles for an equilateral triangle is 180 degrees
because there are three angles, each of size 60 degrees. Find the sum of the angles
for the square, the regular pentagon, and the regular hexagon. What pattern do you
see? Use this pattern to find the angle size for a regular heptagon (which has seven
sides).
CHAPTER SUMMARY
In this chapter, we consider geometric topics that are encountered in everyday life. We
study how to measure geometric objects and how the notion of geometric similarity
is related to the idea of proportionality. We also consider symmetries of shapes and
the notion of tiling the plane.
650 CHAPTER 9
objects, there are standard formulas for computing area and perimeter. The area and
circumference of a circle of radius r are given by the formulas
Area = πr 2
Circumference = 2πr
For a rectangle, the formulas are
Area = Length × Width
Perimeter = 2 × Length + 2 × Width
One way to find the area of a triangle is to pick one side as the base and then find
the height by starting at the opposite vertex and drawing a line segment that meets
the base in a right angle. The area is given by the formula
1
Area = Base × Height
2
Another way to find the area of a triangle is to use Heron’s formula:
Area = S(S − a)(S − b)(S − c)
Here, a, b, and c are the sides of the triangle, and S is the semi-perimeter, defined by
1
S = (a + b + c).
2
The Pythagorean theorem relates the legs a, b of a right triangle with the hy-
potenuse c. It states that
a 2 + b2 = c2
Conversely, if any triangle has three sides of lengths a, b, and c such that a 2 +b2 = c2 ,
then that triangle is a right triangle with hypotenuse c.
Three-dimensional geometric objects can be measured in terms of volume and
surface area. For a box, we have the formula
Volume = Length × Width × Height
For a cylinder with radius r and height h, the volume is
Volume = πr 2 h
and the surface area (excluding the top and bottom) is given by
Surface area = 2πr h
There is a general formula for the volume of any three-dimensional object with a
uniform cross section:
Volume = Area of base × Height
For a ball of radius r , the volume is
4 3
Volume = πr
3
and the surface area is given by
Surface area = 4πr 2
KEY TERMS
perimeter, p. 606 constant of proportionality, p. 622 reflectional symmetry about a
circumference, p. 606 golden ratio, p. 626 line, p. 641
area, p. 606 golden rectangle, p. 627 tiling or tessellation, p. 642
Pythagorean theorem, p. 606 similar, p. 629 regular tiling, p. 643
directly proportional, p. 622 rotational symmetry, p. 639
CHAPTER QUIZ
1. The net on a tennis court is 3.5 feet high. The baseline is 2. Use Heron’s formula to find the area of a triangle with sides
39 feet from the bottom of the net. How far is the baseline 11, 14, and 16.
from the top of the net? Answer 75.5 square units.
Answer 39.2 feet.
If you had difficulty with this problem, see Example 9.4.
If you had difficulty with this problem, see Example 9.3.
Crauder-4100237 QL November 10, 2014 14:8
652 CHAPTER 9
3. A trunk is 2 feet by 3 feet by 4 feet. What is the volume of 6. The volume of a cylindrical water tank of height 30 feet is
the trunk? proportional to the square of the diameter. Two tanks have
Answer 24 cubic feet. height 30 feet, and the diameter of one is 3 times the diame-
ter of the other. How do the volumes compare?
If you had difficulty with this problem, see Example 9.6.
Answer The larger tank has 9 times the volume of the smaller.
4. The volume of concrete needed to make a sidewalk is pro- If you had difficulty with this problem, see Example 9.12.
portional to the length of the sidewalk. (Assume that the side-
7. Find the rotational symmetries of an equilateral triangle.
walk has a fixed width and depth.) Six cubic yards of concrete
are required to make a sidewalk that is 100 feet long. How Answer It has 120-degree rotational symmetry (as well as
much concrete is needed to make a sidewalk that is 25 feet 240-degree).
long? If you had difficulty with this problem, see Example 9.15.
Answer 1.5 cubic yards. 8. Find the reflectional symmetries of an equilateral triangle.
If you had difficulty with this problem, see Example 9.9. Answer It has reflectional symmetry about three lines: the
5. A 5-foot-tall girl casts a 3-foot shadow at the same time that lines from a vertex to the midpoint of the opposite side.
a building casts a 12-foot shadow. How tall is the building? If you had difficulty with this problem, see Example 9.16.
Answer 20 feet tall.
If you had difficulty with this problem, see Example 9.11.
Crauder-4100237 QL_app November 10, 2014 16:39
In this appendix, we discuss conversion of physical units (e.g., feet to inches) and
currency conversion.
Physical units
The most common system of weights and measures in the United States is informally
called the “U.S. Customary System” or the “English system.” Most of the rest of the
world uses what we usually call the “metric system.” The official name of this system
of weights and measures is Le Système International d’Unités, or “SI” for short.
Some basic relationships for unit conversions are contained in Tables A1.1 and
A1.2.
Conversion from one metric unit to another is just a matter of multiplying by the
appropriate power of 10 indicated by the metric prefix. Common prefixes are shown
in Table A1.3.
TABLE A1.1
Conversion Among Familiar
English Units Conversion Factors
Length
1
1 foot 12 inches inches = 12 × feet feet = 12
× inches
1
1 yard 3 feet feet = 3 × yards yards = 3
× feet
1
1 (statute) mile 5280 feet feet = 5280 × miles miles = 5280
× feet
1
1 nautical mile 1.15 (statute) miles miles = 1.15 × nautical miles nautical miles = 1.15
× miles
1 lightyear 5.88 × 10 12
miles miles = 5.88 × 10 12
× lightyears lightyears = 1
5.88
× 10−12 × miles
1
1 parsec 3.26 lightyears lightyears = 3.26 × parsecs parsecs = 3.26
× lightyears
Area
1
1 acre 43,560 square feet square feet = 43,560 × acres acres = 43,560
× square feet
Volume
1
1 tablespoon 3 teaspoons teaspoons = 3 × tablespoons tablespoons = 3
× teaspoons
1
1 fluid ounce 2 tablespoons tablespoons = 2 × fluid ounces fluid ounces = 2
× tablespoons
1 cup 8 fluid ounces fluid ounces = 8 × cups cups = 18 × fluid ounces
1 pint 2 cups cups = 2 × pints pints = 12 × cups
1 quart 2 pints pints = 2 × quarts quarts = 12 × pints
1 gallon 4 quarts quarts = 4 × gallons gallons = 14 × quarts
Weight
1
1 pound 16 ounces ounces = 16 × pounds pounds = 16
× ounces
1
1 ton 2000 pounds pounds = 2000 × tons tons = 2000
× pounds
A-1
Crauder-4100237 QL_app November 10, 2014 16:39
TABLE A1.2
English/Metric Conversions Conversion Factors
Length
1
1 inch 2.54 centimeters centimeters = 2.54 × inches inches = 2.54
× centimeters
1
1 foot 30.48 centimeters centimeters = 30.48 × feet feet = 30.48
× centimeters
1
1 meter (100 centimeters) 3.28 feet feet = 3.28 × meters meters = 3.28
× feet
1
1 kilometer (1000 meters) 0.62 mile miles = 0.62 × kilometers kilometers = 0.62
× miles
Volume
1
1 liter 1.06 quarts quarts = 1.06 × liters liters = 1.06
× quarts
1
1 gallon 3.79 liters liters = 3.79 × gallons gallons = × liters
3.79
1 milliliter (mL) 0.001 liter liters = 0.001 × milliliters milliliters = 1000 × liters
1 cubic centimeter (cc) 0.001 liter liters = 0.001 × cubic cubic centimeters = 1000 × liters
or 1 milliliter centimeters
Weight/mass
1
1 ounce 28.35 grams grams = 28.35 × ounces ounces = 28.35
× grams
1
1 pound 453.59 grams grams = 453.59 × pounds pounds = 453.59 × grams
1
1 kilogram (1000 grams) 2.20 pounds pounds = 2.20 × kilograms kilograms = 2.20 × pounds
1 milligram (mg) 0.001 gram grams = 0.001 × milligrams milligrams = 1000 × grams
Other conversions (assuming standard temperature and pressure)
1 liter of water mass of 1 kilogram
1 cc of water mass of 1 gram
1 mL of water mass of 1 gram
TABLE A1.3
Metric Prefix Meaning Example
kilo 1000 1 kilogram is 1000 grams.
deci 1/10 1 deciliter is one-tenth of a liter.
centi 1/100 1 centimeter is one-hundredth of a meter.
milli 1/1000 1 milligram is one-thousandth of a gram.
1 milliliter is one-thousandth of a liter.
inches = 12 × feet
Informally, the conversion factor from unit A to unit B is the number of B’s in
a single A. For example, from Table A1.1 we find that there are 5280 feet in a mile.
Thus, 5280 feet per mile is the conversion factor for changing miles to feet. This is
Crauder-4100237 QL_app November 10, 2014 16:39
To convert miles to feet, we use the conversion factor of 5280 feet per mile from
Table A1.1:
feet
60 miles = 60 miles × 5280
mile
feet
= 60 × 5280 ——-
miles ×
mile
——
= 60 × 5280 feet
Note how, after we cancel, the final units come out to be “feet,” which is what we
intended. The fact that this happened confirms that we’re on the right track.
We proceed similarly to convert 1 hour to seconds:
minutes seconds
1 hour = 1 hour × 60 × 60
hour minute
minutes seconds
———-
= 1 × 60 × 60 ——
hour × ×
hour
—— minute
———-
= 60 × 60 seconds
Note again how, after we cancel “minutes” and “hours,” the final units come out to
be “seconds,” which is what we intended.
Therefore,
miles 60 × 5280 feet
60 =
hour 60 × 60 seconds
feet
= 88
second
Currency conversion
Unit conversion involving currency arises in connection with currency exchange. For
example, suppose that the value of one U.S. dollar is 0.7773 euro. Hence, the ex-
change rate for converting dollars to euros is 0.7773 euro per dollar. For example,
Crauder-4100237 QL_app November 10, 2014 16:39
Practice Exercises
In these exercises, round all answers to one decimal place unless “made the Kessel Run in less than 12 parsecs.” Do the units
you are instructed otherwise. here make sense?∗
1. Quarter pounders. How many grams of meat are in a “quar- 11. Cubic centimeters. The quantity of fluid administered in
ter pounder”? an inoculation is usually measured in cubic centimeters, abbre-
viated cc. One liter is 1000 cc, and 1 liter is about 33.8 fluid
2. Speed limit. In Canada the speed limit is measured in kilo-
ounces. How many cubic centimeters are in 1 fluid ounce?
meters per hour (abbreviated km/h), and it’s common to see
100 km/h speed limit signs in Canada. How fast is 100 km/h 12. Football field. How many meters long is a football field?
in miles per hour? (A football field is 100 yards long.)
3. Cola. You can get a quart bottle of cola for $1.50 or a liter 13. Meters and miles. Which is longer, a 1500-meter race or a
bottle for $1.55. Which is the better buy in terms of price per 1-mile race? By how much?
volume? 14. Meters and yards. Which is longer, a 100-meter dash or
4–5. Cans and bottles. Soft drinks are often sold in six-packs a 100-yard dash? By how much? Round your answer to the
of 12-ounce cans and in 2-liter bottles. A liter is about 33.8 fluid nearest yard.
ounces. 15. Records. In 1996 Ato Boldon of UCLA ran the 100-meter
dash in 9.92 seconds. In 1969 John Carlos of San Jose State
4. Which is the greater volume: a six-pack or 2 liters? ran the 100-yard dash in 9.1 seconds. Which runner had the
5. A store offers a 2-liter bottle of soft drink for $1.15 and a faster average speed? What was his speed in yards per second?
six-pack of 12-ounce cans for $1.20. Which is the better value Round your answer to two decimal places.
(based on the price per ounce)? 16. Gasoline. My car gets 26 miles per gallon, and gasoline
costs $3.17 per gallon. If we consider only the cost of gasoline,
6. Currency conversion. As of this writing, 1 U.S. dollar is
how much does it cost to drive each mile? Round your answer
worth 13.24 Mexican pesos. What is the exchange rate for
in dollars to two decimal places.
converting dollars to pesos? What is the exchange rate for
converting pesos to dollars? Round your answer to three deci- 17. Speed. A car is traveling at 70 miles per hour. What is its
mal places. speed in feet per second?
7. Currency conversion. In the fall of 2014, 1 U.S. dollar was 18. Gravity. In the metric system the acceleration due to grav-
worth 1.10 Canadian dollars. An automobile in Toronto sells ity near Earth’s surface is about 9.81 meters per second per
for 23,000 Canadian dollars. The same car sells in Detroit for second. What is the acceleration due to gravity as measured in
22,000 U.S. dollars. Which is the better buy? the English system, that is, in feet per second per second?
19. Speed of light. The speed of light is 300 million meters per
8. Conversion factors. What is the conversion factor for kilo-
second. What is the speed of light measured in miles per hour?
meters to miles? What is the conversion factor for miles to kilo-
(Round your answer to the nearest million miles per hour.)
meters? Round your answer to two decimal places.
How long does it take light to travel the distance of 93 mil-
9. Conversion factors. What is the conversion factor for par- lion miles from the sun to Earth? (Round your answer to the
secs to lightyears? What is the conversion factor for lightyears nearest minute.)
to parsecs? Round your answer to two decimal places.
20–21. Lightning The speed of sound in air on a certain
10. Star Wars. In the 1977 film Star Wars Episode IV: A New stormy day is about 1130 feet per second. This fact is used in
Hope, the character Han Solo says that a certain starship Exercises 20 and 21.
20. What is the speed of sound measured in miles per sec- This information is used in Exercises 22 through 24.
ond? In miles per hour? Round the former answer to three 22. How many pounds of cocaine were seized in this bust?
decimal places and the latter answer to the nearest whole
number. 23. What is the street value of a pound of cocaine? What is the
street value of an ounce of cocaine? In each case, round your
21. You see a flash of lightning striking some distance away, answer in dollars to the nearest whole number.
and five seconds later you hear the clap of thunder. How many
24. Assume that a typical dose contains half a gram of pure
miles away did the lightning strike? Round your answer to the
cocaine. How much would a dose cost? Round your answer to
nearest mile. Note: The speed of light is so much faster than
the nearest dollar.
the speed of sound that you can assume that the strike occurs
at the instant you see it. 25. A project, the Gimli glider. There are several examples of
errors involving unit conversion that had serious consequences.
22–24. A drug bust. The following information was reported One such story with a happy ending involved the Gimli glider,
in a 2004 article in a Virginia newspaper. A sheriff in Bedford a commercial aircraft that ran out of fuel at an altitude of about
County, Virginia, reported that investigators found 6.5 kilo- 28,000 feet. The crew glided the aircraft to a landing, and no
grams of cocaine in a car belonging to a man who lives in one was seriously injured. Do a research project on the Gimli
Roanoke, Virginia. This is a significant amount of cocaine, hav- glider. Pay special attention to how unit conversion played a
ing a street value of about $1.2 million dollars. part in the incident.
Exponents
Negative exponents: a −n is the reciprocal of a n .
Definition Example
1 1 1
a −n = n 2−3 = 3 =
a 2 8
Zero exponent: If a is not 0, then a 0 is 1.
Definition Example
a0 = 1 20 = 1
Basic properties of exponents:
Property Example
a a = ap+q
p q
2 × 2 = 22+3 = 25 = 32
2 3
ap 36
= a p−q = 36−4 = 32 = 9
aq 34
3 2
(a p )q = a pq 2 = 23 × 2 = 26 = 64
Powers of 10
Powers of 10 provide an easy way to represent both large and small numbers. If n
is a positive whole number, we find 10n by starting with 1 and moving the decimal
point n places to the right. For example,
3 tells how right shift
For 103 to move the 1.000 by 3 places 1000
decimal point results in
A negative power of 10 causes the decimal point to be moved to the left. If n is
a positive whole number, we find 10−n by starting with 1 and moving the decimal
point n places to the left. For example,
−3 tells how left shift
For 10−3 to move the 0001.0 by 3 places 0.001
decimal point results in
Scientific notation
When we multiply by a power of 10, the effect is to move the decimal point, just as
when powers of 10 stand alone. For example,
3 tells how right shift
For 3.45 × 103 to move the 3.450 by 3 places 3450
decimal point results in
−3 tells how left shift
For 3.45 × 10−3 to move the 003.45 by 3 places 0.00345
decimal point results in
This observation leads to the idea of expressing numbers in scientific notation.
A-7
Crauder-4100237 QL_app November 10, 2014 16:39
For example, to write 24,500 in scientific notation, we start with 2.45 because
that is between 1 and 10. To get 24,500 from 2.45, we need to move the decimal
place four places to the right. That is the power of 10 that we use. The result is
24,500 expressed in scientific notation is 2.54 × 104
To write 0.000245 in scientific notation, we start once more with 2.54. This time
we need to move the decimal point four places to the left. That means we use an
exponent of −4:
0.000245 expressed in scientific notation is 2.45 × 10−4
Scientific notation is useful in dealing with very large and very small numbers. For
example, according to the U.S. Treasury Department, the national debt as of January
22, 2009, was $10,618,718,703,374.78. This is over 10.6 trillion dollars. To write
10.6 trillion or 10,600,000,000,000 in scientific notation, we start with 1.06 because
that number is between 1 and 10. To get 10,600,000,000,000 from 1.06, we must
move the decimal point 13 places to the right. This is the power of 10 we use:
10.6 trillion = 10,600,000,000,000 = 1.06 × 1013
In chemistry there is a very large number called Avogadro’s number. It is the num-
ber of atoms in 12 grams of carbon-12 and is about 602,000,000,000,000,000,000,
000 (21 zeros in a row). To express this number in scientific notation, we start with
6.02 because it is between 1 and 10. We need to move the decimal point 23 places
to the right, so this is the power of 10 that we use. The result is 6.02 × 1023 .
As another example, the wavelength of blue light is 0.000000475 meters (6 zeros
after the decimal place). To express this number in scientific notation, we begin with
4.75 because it is between 1 and 10. We need to move the decimal seven places to
the left, so we use −7 for the power of 10. The result is 4.75 × 10−7 .
Scientific notation not only gives us an easy way to express very large and very
small numbers, but also makes it easier for us to do calculations based on the rules
of exponents. For example, light travels about 186 thousand miles per second. Let’s
figure out how far it will travel in a year (i.e., the number of miles in a lightyear).
In scientific notation, 186 thousand is 1.86 × 105 . There are 60 × 60 × 24 × 365 or
Crauder-4100237 QL_app November 10, 2014 16:39
about 31.5 million seconds in a year. That is 3.15 × 107 . So in a year, light will travel
about
1.86 × 105 × 3.15 × 107 = (1.86 × 3.15) × (105 × 107 )
= (1.86 × 3.15) × 105+7
= 5.859 × 1012 miles
That is almost 6 trillion miles.
Practice Exercises
1. Earth to moon. The distance from the center of Earth to the 7. 0.0003 (three ten-thousandths)
center of the moon is about 239,000 miles. Express this number 8. Fifteen millionths
in scientific notation.
9. One-hundred-and-eight billionths
2–9. Scientific notation. In Exercises 2 through 9, express the
given number in scientific notation. 10–12. World population. As of this writing, the world’s
population was about 7.2 billion, and the population of
2. Two hundred million Canada was 35 million. This information is used in Exercises
3. 953,000 10 through 12.
4. Twenty-eight billion 10. Write in scientific notation the world population.
5. Twenty and a half trillion 11. Write in scientific notation the population of Canada.
6. 0.12 (twelve hundredths) 12. What percentage of world population is the population of
Canada? Round your answer to one decimal place.
Parentheses
Difficulties with calculators often stem from improper use of parentheses. When they
occur in a formula, their use is essential, and sometimes additional parentheses must
be supplied.
For example, expressions like (1 + 0.06/12)36 arise in Chapter 4 when we
consider compound interest. In such expressions, the parentheses are crucial. If you
omit them, you may get the wrong answer. Let’s see what happens. The correct cal-
culator entry is
( 1 + .06 ÷ 12 ) ∧ 36
The calculator’s answer is 1.196680525. If the parentheses are omitted, the expres-
sion becomes 1 + 0.06/1236 , which we enter as
1 + .06 ÷ 12 ∧ 36
The calculator’s answer is 1. As we can see, the parentheses make a big difference.
Sometimes when parentheses do not appear, we must supply them. For example,
1.24 + 3.75
no parentheses appear in the expression . But when we enter this expres-
1.62
sion, we must enclose 1.24 + 3.75 in parentheses to tell the calculator that the entire
expression goes in the numerator. The correct entry is
2 ÷ ( 3.6 + 5.7 )
2 ∧ ( 1.7 × 3.9 )
A-10
Crauder-4100237 QL_app November 10, 2014 16:39
Rounding
It is often unnecessary to report all the digits given as an answer by a calculator, so
it is common practice to shorten the answer by rounding.
First we need to decide how many digits to keep when we round. There is no set
rule for determining this number. Generally, in this text we keep one digit beyond
the decimal point. In certain applications, though, it is appropriate to keep fewer or
more digits. For example, when we report the balance of an account in dollars, we
usually keep two digits beyond the decimal point, that is, we round to the nearest
cent. Another consideration in rounding is the accuracy of the data on which we base
our calculation. We should not report more digits in our final result than the number
of digits in the original data.
The usual method for rounding a number begins with the digit to be rounded.
That digit is increased by 1 if the digit to its right is 5 or greater and is left unchanged
if the digit to its right is less than 5. All the digits to the right of the rounded digit are
dropped.
For example, let’s round 23.684 to one decimal place. That means we are going
to keep only one digit beyond the decimal point. We focus on the 6. Because the next
digit to its right (8) is 5 or larger, we increase the 6 to 7 and drop all the digits to the
right of it. This results in 23.7.
If we choose to round 23.684 to two decimal places, we are going to keep two
digits beyond the decimal point. We focus on the 8. The next digit is a 4, which is
less than 5. So we leave the 8 unchanged and delete the remaining digits. This results
in 23.68.
Although it is often appropriate to round final answers, rounding in the middle
of a calculation can lead to errors. We were cautioned about this in Calculation Tip
4.1 in Section 4.1. Let’s look a bit more closely at this situation.
Here is a typical financial calculation. It gives the balance in dollars of an account
after 5 years of daily compounding:
1825
0.06
Balance = 10,000 1 +
365
The right way to calculate it: Use parentheses and enter the formula as it reads. Here
is the sequence of keystrokes:
Chain calculations
For complicated formulas, it is possible to make calculations in steps without in-
termediate rounding. To illustrate the procedure, let’s recall the monthly payment
formula from Section 4.2:
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
Here, t is the term in months and r = APR/12 is the monthly interest rate as a
decimal. Let’s do the calculation when the amount borrowed is $10,000, t = 30
months, and the monthly rate is r = 0.05/12. With these values, the formula gives
10000 × 0.05/12 × (1 + 0.05/12)30
Monthly payment =
30
(1 + 0.05/12) − 1
If we choose to enter the entire expression all at once, we must be very careful to
get all the parentheses right:
÷ ( ( 1 + .05 ÷ 12 ) ∧ 30 − 1 )
( 1 + .05 ÷ 12 ) ∧ 30
The calculator displays 1.132854218, and it automatically stores the result. We can
recall it using the keystrokes 2nd [ ans ] . So we can now enter the formula as
Practice Exercises
1–7. Using parentheses. Exercises 1 through 7 focus on the 7. Calculate the following four expressions. The only differ-
use of parentheses. In each case, show how parentheses should ence is placement of parentheses. Compare the results.
be used to calculate the given expression and then find the value a. 2 × 3 + 42
rounded to one decimal place. b. (2 × 3 + 4)2
3.1 + 5.8
1. c. 2 × (3 + 4)2
2.6
d. (2 × 3) + 42
3.1
2. 8. Round the number 14.2361172 to one decimal place.
5.8 − 2.6
3.1 − 1.3 9. Round the number 14.2361172 to two decimal places.
3.
5.8 + 2.6 10. Round the number 14.2361172 to three decimal places.
4. 31.1−2.2 11. Intermediate rounding. This exercise shows problems that
1.6 + 3.7 can arise using improper calculation techniques.
5. 3 −
4.3 a. Calculate $100,000(1 + 0.07/12)100 to the nearest cent.
6. Calculate the following three expressions. The only differ- b. Observe that 0.07/12 = 0.0058333333 · · ·. Round this
ence is placement of parentheses. Compare the results. to 0.006, and use that result to calculate $100,000(1 +
a. 2 × 3 + 4 0.07/12)100 to the nearest cent.
b. 2 × (3 + 4) c. What is the difference between your two answers?
c. (2 × 3) + 4
Crauder-4100237 QL_app November 10, 2014 16:39
1
In 0.1, the 1 is in the 10th position. Hence 0.1 =
10
12
In 0.12, the 2 is in the 100th position. Hence 0.12 =
100
123
In 0.123, the 3 is in the 1000th position. Hence 0.123 =
1000
1234
In 0.1234, the 4 is in the 10000th position. Hence 0.1234 =
10000
0.75
4 3.00
2.8
0.20
0.20
0
3
Hence = 0.75.
4
A-14
Crauder-4100237 QL_app November 10, 2014 16:39
b. In the decimal 3.625, the last digit, the 5, is in the 1000th position. Hence
3625
3.625 =
1000
We can simplify this fraction:
3625 29 × 125
=
1000 8 × 125
29 ×
125
=
8 ×
125
29
=
8
Using percentages
There is a close connection between percentages and decimals. If we want 25% of a
25
quantity, we multiply it by the decimal 0.25, which is . We can think of obtaining
100
0.25 from 25% by dropping the % symbol and moving the decimal point two places
to the left. If we want to express a decimal as a percentage we reverse this procedure:
To express 1.2 as a percentage we calculate 100 times 1.2 to obtain 120%.
Quick Review
r
If r is a number, then r % is a way of expressing the fraction . This fraction
100
is expressed in decimal form by moving the decimal point of r two places to
the left. We often refer to this associated decimal as the decimal form of the
percentage.
If we are given a number in decimal form and are asked to express it as a
percentage we reverse this procedure, multiplying the number by 100 followed
by a percent sign.
A Typical Example
In a drug trial involving 50 people, 74% showed positive effects from the drug. How
many people showed positive effects?
SOLUTION
The number of people who showed positive effects is 74% of 50. To find 74% of 50,
we multiply 50 by the decimal form of 74%, which is 0.74:
Number who showed positive effects = 50 × 0.74 = 37 people
A Typical Example
What percentage is value A of value B? To find the decimal form of the percentage,
we divide value A by value B.
A Typical Example
In a survey of 800 people, 644 approved of the current mayor. Judging from this
sample, what is the mayor’s percentage approval?
SOLUTION
We are being asked to find 644 as a percentage of 800, so we first divide 644 by 800
to find the decimal form:
644
= 0.805
800
To find the associated percentage, we multiply by 100 (or move the decimal point
two places to the right). The resulting answer is an 80.5% approval rating.
What is the percentage increase or decrease? If a quantity increases from one value
to another, then the change is the difference. This difference, expressed as a percent-
age of the original value, is the percentage increase. If a quantity decreases from one
value to another, then the difference, expressed as a percentage of the original value,
is the percentage decrease.
A Typical Example
SOLUTION
a. The actual increase is the difference 18,942 − 15,400 = 3542. Thus, to calculate
the percentage increase we need to know what percentage 3542 is of 15,400. As a
decimal this is the quotient
3542
= 0.23
15,400
This is the decimal form of 23%. Hence, the population increased by 23%.
b. The actual decrease is the difference 18,942 − 15,400 = 3542. Thus, to
calculate the percentage decrease we need to know what percentage 3542 is of
18,942. As a decimal this is the quotient
3542
18,942
or about 0.187. This is the decimal form of 18.7%. Hence, the population
decreased by 18.7%.
Practice Exercises
1–4. Fractions to decimals. 12. Find the percentage. Express these numbers as percentages:
4 0.1, 0.01, and 1.0.
1. Write as a decimal.
5 13. What percentage is it? Candidate 1 received 320 of 500
5 votes cast. What percentage of voters voted for Candidate 1?
2. Write as a decimal.
4 14. What percentage is it? Assume that 32 of 60 students in a
6 class are female. What percentage of students in the class are
3. Write as a decimal. Round your answer to two decimal female? Round your answer to the nearest tenth of a percent.
7
places. 15. What is the result of a percentage increase? A city of
7 433,000 experiences a population increase of 12%. What is
4. Write as a decimal. Round your answer to two decimal
3 the new population of the city?
places. 16. What is the result of a percentage decrease? An investment
5–6. Decimals to fractions. of $500 loses 15% of its value. What is the new value of the
5. Write 3.72 as a fraction in reduced form. investment?
6. Write 0.04 as a fraction in reduced form. 17. An item on sale. An appliance marked at $535 is on sale
for 15% off. What is the sale price of the item?
7–8. Percentages to decimals.
18. What is the percentage increase? A company buys items at
7. Write the decimal form of 34.5%. a wholesale cost of $35. The items are sold at the retail price
8. Write the decimal form of 1/2%. of $50. What is the percentage price mark-up for this item?
Round your answer to the nearest tenth of a percent.
9–11. Decimals to percentages.
19. What is the percentage decrease? An item that normally
9. The number 0.36 is the decimal form of what percentage?
costs $42.50 is on sale for $38.00. What is the percentage sav-
10. The number 0.055 is the decimal form of what percentage? ings? Round your answer to the nearest tenth of a percent.
11. The number 3.1 is the decimal form of what percentage?
Length
A-18
Crauder-4100237 QL_app November 10, 2014 16:39
Refer to Example 3.12, where we are asked to find an exponential formula that
describes the amount of amoxicillin in the bloodstream t hours after injection. Also,
we are asked to determine whether a second injection is necessary after five hours. If
we have properly followed the first principle, here is an orderly plan for finding an
exponential formula.
Step 1: Identify the initial value.
Step 2: Identify the base.
Step 3: Use the base and initial value to construct the exponential formula.
Once we have the formula, we can use it to find the level of amoxicillin in the
bloodstream after five hours.
Let’s look at Example 4.16: Your neighbor took out a 30-year mortgage for
$300,000 at a time when the APR was 9%. She says that she will wind up paying
more in interest than for the home (that is, the principal). Is that true?
Certainly the first principle applies. We will make no progress at all without
knowing what a mortgage is and how to use the APR. It may be helpful to restate
the problem: “Will she pay more than $300,000 in interest?”
Once we have a clear understanding of the problem, we need to devise a plan for
solving it. Here is a possible outline.
Step 1: Find the total amount that she paid.
Step 2: Calculate the difference between the total amount paid and the principal.
That difference is the interest paid.
Step 3: Compare the interest paid with the principal.
In this case it is probably helpful to enhance Step 1.
Enhancement of Step 1:
Step 1a: Calculate the monthly interest rate as a decimal.
Step 1b: Use the monthly payment formula to find the amount paid each month.
Step 1c: Multiply the monthly payment by the number of months to find the total
amount paid.
It is a good idea to actually write down the plan of attack.
There is no single strategy that will lead you to the solution of all problems, but
understanding the problem and devising a plan for solution are key ingredients in
the process.
complicated formula:
Amount borrowed × r (1 + r )t
Monthly payment =
((1 + r )t − 1)
$300,000 × 0.0075 × 1.0075360
=
(1.0075360 − 1)
We will have to evaluate this expression using a calculator or computer, and it is
very easy to enter this expression improperly. If you calculate a monthly payment of
47 cents or 47 thousand dollars, you can be pretty sure that something is wrong. The
point is that you should not give up. Try entering the formula again. If persistence
doesn’t result in a reasonable answer, it may be time to re-examine your plan of
attack. You are not alone when this happens. It happens to the authors of this text,
and it happens to the strongest mathematicians on the planet. In fact, this is a key to
how all problems in mathematics are solved. If your idea doesn’t work out, look for
your error and try again—and again.
BRIEF ANSWERS TO
ODD-NUMBERED EXERCISES
Exercise Set 1.1 Exercise Set 1.2
1.a. 1. False authority.
Local Statewide
3. The premises are that all dogs go to heaven and my
Low income 60.0% 61.0% terrier is a dog. The conclusion is that my terrier will go
High income 67.5% 68.0% to heaven.
Total 67.0% 66.6%
5. The premise is that the president was caught in a lie. The
conclusion is that nobody believes the president anymore.
b. Worse
c. Better 7. The premise is that people have freedom of choice. The
conclusion is that they will choose peace.
3. Fair condition
9. The premise is that bad times have a scientific value. The
% who lived
conclusion is that these are occasions a good learner
Mercy 98.3% would not miss.
County 96.6% 11. The argument is invalid.
13. The argument is invalid.
5. Answers will vary.
15. The argument is invalid.
7. 75%
17. Dismissal based on personal attack
9. 10%
19. False cause
11. Answers will vary.
21. False dilemma
13. Answers will vary.
23. Appeal to common practice
15. Below poverty line
25. Appeal to ignorance
% passing
27. Straw man
School A 52.5%
School B 36.2% 29. False dilemma
31. Circular reasoning
17. 90.5%
33. Straw man
19. 36.2%
35. Answers will vary.
21. Answers will vary.
37. Answers will vary.
23. All participants
39. The number of blocks is the cube (the third power) of the
Total % with insurance size of a side. 125,000 blocks fit in a cube 50 inches on
Not traced 1174 17% each side.
5-year group 416 11% 41. These are the positive odd integers in order.
43. Answers will vary.
25. White participants
Total % with insurance Exercise Set 1.3
Not traced 126 83% 1. (NOT p) → (NOT q)
5-year group 12 83%
3. (France OR England) AND (eighteenth OR nineteenth
century)
27. They are the same in both cases.
5. If I don’t buy this MP3 player, then I can’t listen to my
29. Answers will vary. favorite songs. No.
ANS-1
Crauder-4100237 QL_answers November 10, 2014 16:47
9. q → (NOT p) T T T T
T F T T
11. q → (NOT p)
F T T F
13. p: “We pass my bill”; r : “The economy will recover”; F F F F
p→r
71. Their truth values are both (F, T, T, T).
15. c: “I want cereal for breakfast”; e: “I want eggs for
breakfast”; c AND e
73. 0
17. c: “You clean your room”; f : “I’ll tell your father”;
75. 1
(NOT c) → f
77. Answers will vary.
19. d: “I’ll go downtown with you”; m: “We can go to a
movie”; r : “We can go to a restaurant”; (m OR r ) → d
21. “It is not the case that he’s American or Canadian.” Exercise Set 1.4
“He’s not American and he’s not Canadian.”
1. {5, 6, 7, 8, 9}
23. False 25. True 27. True
3. {b, c, d, f, g, h, j, k, l, m, n, p, q, r, s, t, v, w, x, y, z}
29. True 31. False 33. True
35. True 37. True 5. B and C are subsets of A. B and C are disjoint.
39. If you clean your room, I won’t tell your father. 7. A is a subset of B, A is a subset of C, and B is a subset
of C.
41. If you don’t agree to go to a movie or a restaurant, then I
won’t go downtown with you. 9.
N = Northern Hemisphere
43. If it is not a math course, then it is not important.
45. If you don’t take your medicine, you won’t get well.
United States
47. If I told your father, then you didn’t clean your Japan Brazil
room.
49. If I went downtown with you, then you agreed to go to a
movie or a restaurant. Australia A = Atlantic
nations
51. All important courses are math courses.
53. If you get well, then you took your medicine.
11.
55. If you have not chosen the side of the oppressor, you are
Y = Yellow fruit
not neutral in situations of injustice.
57. If you do not get arrested, then you did not drink
and drive.
lemon
banana orange
59. If I didn’t tell your father, then you didn’t clean your
room.
61. If I did not go downtown with you, then you did not cherry C = Citrus fruit
agree to go to a movie or a restaurant.
63. If you do not drive a Porsche, then you are not my 13. 68 15. 851 17. 991 19. 1989
friend.
21.
65. If radar does not see thunderstorms, then they are not H = History
approaching.
67.
p q ( p OR q) →
T T T T
T F T T
F T T T P = Poli Sci
F F F T
Crauder-4100237 QL_answers November 10, 2014 16:47
10 12 72
b.
30
Percentage increase
25
20
84 Blue
15
10
33. 5
S C
0
2010–11 2011–12 2012–13 2013–14
9
28 17
c. 2012–13
21
5 6 3. Cases per day; the number of new cases of the disease we
expect each day.
35. 24 7. 74.5%
37. 27 9. 9487
41. 140 13. About 0.4 percentage point per year. (Some variation in
answers is acceptable.)
43. 88
15. The percentage grew faster after 2012 than it did
45. Answers will vary. between 2005 and 2010.
Crauder-4100237 QL_answers November 10, 2014 16:47
Sales
23. 1996: 2.9%; 1997: 3.0%; 1998: 3.2%; 1999: 3.3% 60
(Some variation in answers is acceptable.) 40
25. $98 20
0
27. $454 per student Chains Independent Mass Super Mail
market order
29. −0.07 dollar per student for each additional student
enrolled
31. The cost per student decreases more at the low end of Line graph:
the scale, for consolidations of smaller schools.
33. Answers will vary. $120
100
35. Enrollment in French decreased from 1995 to 1998 and 80
increased after that.
Sales
60
37. 40
4 20
Average annual growth
(thousands of students
3 0
2 Chains Independent Mass Super Mail
per year)
1 market order
0
–1
–2 Answers will vary.
–3
1995–98 1998–02 2002–06 2006–09
7. One possibility is
39. 2010
1200
41. 33.5% 1000
Investment
800
60
$120 50
100 40
30
80 20
Sales
60 10
0
40 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
20 Age
0
Chains Independent Mass Super Mail
market order
Crauder-4100237 QL_answers November 10, 2014 16:47
Mortality rate
0 245
0 1 2 3 4 5 6 7 8
Day 230
215
13. 200
80
50–54 55–59 60–64 65–69 70–74 75–79 80–84 85–89 90–94 95–99 00–04 05–09
70
Years
60
50
40 5. The fact that nearly 60% of employees are in the un-
30
20
known category makes the graph of questionable value.
10
7. The lines connecting the dots do not represent any
0
–10
conceivable data. A table, or perhaps a bar graph or
0 1 2 3 4 5 6 7 8 9 scatterplot, would be better.
9. Answers will vary. Knowing the number of drivers in
15. One possibility is each category would be important.
90 11.
80 22.00
70 20.00
Number of bears
60 18.00
50 16.00
40 14.00
30 12.00
20 10.00
8.00
10
6.00
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 4.00
2.00
Years
0.00
Composite English Math Reading Science
17. Enrollments decreased from 1968 to about 1980 and
from about 1990 to about 1995. 13. a. From 1945 to 1990
19. 1960–1965; the slope then is larger. b. 0.29 thousand dollars per year, or 290 dollars per year
c. 1.14 thousand dollars per year, or 1140 dollars per year
21. Mystery curve 1: growth; Mystery curve 2: sales. d. Answers will vary.
Answers will vary.
15. A justification is difficult to find. Perhaps it seems natural
23. Answers will vary. to predict growth using a straight line.
17. Answers will vary.
Exercise Set 2.3
1. A narrow range on the vertical axis should be used. One 19.
1.2
possibility is
Attendance in millions
1
Graduation rate, percentage
89.0 0.8
88.5 0.6
88.0 0.4
87.5 0.2
87.0 0
1998 1999 2000 2001 2002
86.5
Year
86.0
1998 2002 2006
Answers will vary.
Year
Crauder-4100237 QL_answers November 10, 2014 16:47
21. 29. a.
47
Year Price (2010 dollars)
46.8
1970 $82.65
46.6 1980 $76.20
Age
600 300
2004 dollars, billions
500 200
400
100
300
0
200 1970 1980 1990 2000 2010
100 Year
0
2004 2005 2006 2007 2008 2009 2010 2011 2012
Year c. Tickets were about 6 times as expensive in 2010.
Answers will vary. 31. The vertical axis represents percentage of the total by
weight.
27. a.
1400
60
Price per ounce (dollars)
1200 50
1000 40
800 30
600 20
400 10
200 0
Flour Milk Sugar Eggs
0
1970 1980 1990 2000 2010
Year Eggs
10%
b. Sugar
Year 2010 dollars 15%
Flour
1970 $198.03 50%
1980 $1555.90
Milk
1990 $628.96
25%
2000 $351.68
2010 $1224.53
23. 1.16 7.
Quarter Interest Balance
25. 1000 times
1 $60.00 $3060.00
27. After 153 months, or 12 years and 9 months 2 $61.20 $3121.20
29. 2.32 half-lives; 69.6 years 3 $62.42 $3183.62
4 $63.67 $3247.29
31. C = 100 × 1.025t
33. 56 years 9.
Quarter Interest Balance
35. $348.68
1 $50.00 $2050.00
37. 23 trillion dollars 2 $51.25 $2101.25
39. Answers will vary. 3 $52.53 $2153.78
4 $53.84 $2207.62
41. Answers will vary.
43. Answers will vary. 11. Simple interest would yield $200 in interest. Semi-annual
45. Answers will vary. compounding would yield $205 in interest.
47. Answers will vary. 13. For simple interest, the balance is $1120.00. For monthly
compounding, the balance is $1127.46.
Exercise Set 3.4 15. $657.50
1. a. x = 1 or −2 17. Option 2 is better if you retire at age 65. Option 1 is
b. y = 6 or −3 better if you retire at age 55.
c. r = 5 or −2 19. $7607.64
d. x = 5/2 or −2
21. First quarter: $2040.00. Second quarter: $2080.80.
e. x = −2 ± 5
Third quarter: $2122.42. Fourth quarter: $2164.87.
3. x = (1 + 5)/2 Total interest $164.87. The APY is 8.24%.
31. 16; (x + 4)2 41. 11 years and 7 months. This is 5 months less than the
estimate of 12 years given by the Rule of 72.
Exercise Set 4.1 43. 8.19%
1. $1200.00
Exercise Set 4.2
3. a. $2382.03
b. $2393.36 1. $138.89
5. $50; $2050 3. $386.66; $3199.60
Crauder-4100237 QL_answers November 10, 2014 16:47
5. The rule of thumb says that the payment is at least Exercise Set 4.3
$1041.67. The monthly payment is $1096.78.
1. $5027.42
7. $15,014.57
3. $116.91
9. Rule of Thumb 4.2 gives the correct answer of $750.00
per month. 5. Both give $202.26.
13. $227.20 9. The formula gives $301.97; the table gives $301.98.
15. $1904.90 11. $31,733.38
17. $664.81; $955.44 13. The total amount is $132.07. Benny is wrong by virtue of
part 2 of Rule of Thumb 4.4.
19. Answers will vary.
15. $10,012.57
21. We omit the column showing the monthly payment
of $8.56. 17. $3295.02
19. $597,447.22
Applied Applied Outstanding
No. interest balance balance 21. $497,872.68
$15,000.00
1 $366.19 $100.00 $266.19 $14,733.81 Exercise Set 4.4
2 $366.19 $ 98.23 $267.96 $14,465.85
1. $11.55
Equity after 24 payments is about $6903. 3. $74.16; $68.75
Crauder-4100237 QL_answers November 10, 2014 16:47
5. We list only previous balance, finance charge, and new Exercise Set 4.5
balance.
1. 4.9%
53. 1/3; 9/19 49. The probabilities are not the same.
59. 88% 57. Exercise 51: 0.18; Exercise 52: 0.51; Exercise 54: 0.00
(less than 4.6 × 10−44 )
61. 14%
63. 29 billion
Exercise Set 5.5
65. Probability of at least one double six: 1 − (35/36)24
or 0.49 1. $81.60
11. The rule of thumb gives the same margin of error, 3.5%. 5. Small p-values, not large ones, indicate statistical
significance.
13. Answers will vary. About 816 people were sampled.
7. The p-value is 0.184, which is not normally considered
15. Answers will vary. statistically significant.
17. Doubling the sample size
multiplies the margin of 9. Answers will vary.
error by a factor of 1/ 2 or about 0.7. Cutting the
margin of error in half requires quadrupling the sample 11. Answers will vary.
size.
13. 14
19. 2.6%
15. a. These variables are negatively correlated: As distance
21. With 90% confidence, between 45.4% and 50.6% like increases, brightness decreases.
salad. The majority level is within the margin of error, so b. These variables are uncorrelated.
we cannot say with 90% confidence that a majority like
c. These variables are positively correlated: Earthquakes
salad.
with larger readings on the Richter scale generally
23. a. 87% cause more damage.
b. 500 d. These variables are negatively correlated: A golfer
25. Answers will vary. wins by getting the lowest score. So the lower a
golfer’s score, the higher her ranking.
27. Answers will vary.
17. No
29. 3%
19. Yes
31. 95% of the polls represented the voting population with
an error of at most 3 percentage points. 21. 80 million should remember
200
29. Answers will vary.
150
100 31. Answers will vary.
50
33. Answers will vary.
0
Less 47% 48% 49% 50% 51% 52% At
than to to to to to to least
47% 48% 49% 50% 51% 52% 53% 53%
Crauder-4100237 QL_answers November 10, 2014 16:47
Exercise Set 7.1 17. One Euler circuit is Post Office-F-B-E-D-Post Office-
C-B-Post Office.
F
1. Answers will vary. Post Office
B
3. C D
Red Division Blue Division
Cubs
Red Sox
E
Lions
Yankees 19. No Euler circuit exists.
Bears
Giants 21. No Euler circuit exists
Tigers 23. No Euler circuit exists.
25. No Euler circuit exists.
E
33. a. The vertices of odd degree are marked A and B.
A D A
C
B
F
B
E
B
B
A 1
4 C D
8
20
9 7
5
19 10 3 11. The highlighted edges show one Hamilton circuit.
6
18 11 17
16 2 KY
MO WV
14
13
15 B VA
12 TN
AR NC
d. Delete the first edge from the Euler circuit to get the
Euler path. SC
LA
35. A MS AL GA
C
FL
F
T. J. Maxx Old Navy
B Home Depot
Toys
E “R” Us Music City
Dillards
1. Euler circuit 15. The highlighted edges show one Hamilton circuit.
3. Yes
Meals on Wheels
5. The highlighted edges show one Hamilton circuit.
Home
Park entrance Home
Home
Home
Geyser
Home
Mountain
Home
Falls
Camp
Lake
Crauder-4100237 QL_answers November 10, 2014 16:47
17. The highlighted edges show one Hamilton circuit. 31. A-D-E-C-B-A has length 15.
G A
A
4
F 5
4 B
B E 2 3
2 2 5 5
E
D 6 C
C D
A
D 6 C
4
5
4 B
E 2 3
35. A-D-E-B-C-A has length 14.
2 5 5
2
A
D 6 C
4
5
4 B
27. A-D-E-B-C-A has length 14. E 3
2
2 6 5 3
A
4 D 8 C
5
4 B
E 3
2
2 5 3 37. a. A-B-C-D-E-A has length 205
6
b. B-A-D-E-C-B has length 8 or B-C-D-E-A-B has
length 205
D 6 C
c. C-B-A-D-E-C has length 8
d. D-E-C-B-A-D has length 8
e. E-D-C-B-A-E has length 205 or E-D-B-C-A-E has
29. A-D-E-B-C-A has length 14.
length 209 or E-D-B-A-C-E has length 12 or
E-D-A-B-C-E has length 8
A f. Different starting points may yield different answers.
4 39. 14
5
4 B
E 41. 18
2 3
2 5 3 43. Answers will vary.
6
45. Answers will vary.
D 8 C
Crauder-4100237 QL_answers November 10, 2014 16:47
co ca ch
1. One solution is
Votes
Total Winning
100 91 10 votes coalition? Critical voters
A B C 201 Yes None
A B 191 Yes A, B
5. No A C 110 Yes A, C
A 100 No not applicable
7. D, G, H, I, K B C 101 Yes B, C
B 91 No not applicable
9. kite
C 10 No not applicable
15.
Voters Swing
fax saw
Abe Ben Condi Ben
Abe Condi Ben Condi
can in never vow Ben Abe Condi Abe
Ben Condi Abe Condi
Condi Abe Ben Abe
and dig get it loud quiet top yes
Condi Ben Abe Ben
bake even hive is jump mate open rat under xray zebra
17. 14
19. We list only winning coalitions.
11. 7 checks
Coalition Votes Critical voters
13. 16 checks
ABCD 26 A
15. 501 ABC 24 A
17. 1024 leaves; 176 show at least 7 passes. ABD 21 A
AB 19 A, B
19. 7 games
ACD 20 A
21. 255 AC 18 A, C
AD 15 A, D
23. 240
Crauder-4100237 QL_answers November 10, 2014 16:47
21. 24 11. D
23. A: 18/24 = 3/4; B, C, D each 2/24 = 1/12 13. D
25. Abe: 1/2, Ben: 1/2, Condi: 0, Doris: 0 15. D
27. 11 17. D
29. 31 19. B
31. A: 1/3, B: 7/27, C: 5/27, D: 1/9, E: 1/9 21. C
23. C
33.
R D I D 25. C
R I D D
27. A
D R I R
D I R I 29. B
I R D D
31. B
I D R D
33. A
25. Oklahoma 14, Texas 93, Arkansas 11, New Mexico 7 19. Area is 100 square units. Perimeter is 51.2 units.
27. Montana should get 2 seats, Alaska 1 seat, Florida 26 21. Area is 7 75/2 orabout 30.3 square units.
seats, and West Virginia 3 seats, using Dean’s method. Perimeter is 17 + 79 or about 25.9 units.
29. 23. Area is 4(2 + 33)/2
or about 15.5 square units.
State Seats
Perimeter is 9 + 20 + 33 or about 19.2 units.
Connecticut 7 25. The can holds 17.2 cubic inches. Therefore, the amount
Delaware 1 of lost soda is more than that.
Georgia 2
Kentucky 2 27. Area: 30.9 square inches
Maryland 8 29. Circumference: 8.2 inches
Massachusetts 14
New Hampshire 4 31. 858 miles
New Jersey 5 33. 21 inches
New York 10
North Carolina 10 35. 212 square inches and 188 square inches; no
Pennsylvania 13 37. 2153.4 square feet
Rhode Island 2
39. 656 or about 25.6 feet
South Carolina 6
Vermont 2 41. 349.6 feet
Virginia 19
43. 11
Crauder-4100237 QL_answers November 10, 2014 16:47
45. Answers will vary. 49. One can hold 4 times that of the other.
47. Answers will vary. 51. One can hold twice as much as the other.
49. Answers will vary. 53. The BMR of the longer animal is 100 times that of the
shorter.
51. Answers will vary.
55. Answers will vary.
53. 920 cubic feet, 34.1 cubic yards
55. Volume: 12.8 cubic inches; Area: 26.4 square inches
57. Volume: 433.5 cubic inches; Area: 277.0 square inches Exercise Set 9.3
59. Volume: 6.0 cubic inches 1. Answers will vary.
ANS-22
Crauder-4100237 QL_Credits November 15, 2014 10:39
CREDITS
Ch. 1 p. 116: (In the News 2.5) From Beth Gollob, “State ACT Scores
Improve, Still Low.” Reprinted by permission of The Oklahoma
p. 3: (In the News 1.1) From Vicky Travis, “Reading Academy
Publishing Co.
teaches kids to think critically (Vanderbilt University models
program off of successful Saturday University).” Reprinted by p. 128: (Figure 2.62) Reproduced with permission of Yahoo.
permission of Vicky Travis and The Tennesean. c 2014 Yahoo. YAHOO! and the YAHOO! Logo are registered
trademarks of Yahoo.
p. 4: (In the News 1.2) From David Smith, “A great example of
Simpson’s Paradox: US median wage decline.” Reprinted by p. 130: (Figure 2.66) Data from OECD Health Data 2013 and
permission of the author. http://www.huffingtonpost.com/2013/10/03/health-care-costs-
n 3998425.html
p. 15: (In the News 1.3) From “Lhota Backs Off of Congress
Republicans During NYC Debate,” by Esmé E. Deprez and Henry
Goldman. Used with permission of Bloomberg L.P. Copyright Ch. 3
c 2014. All rights reserved.
p. 137: (In the News 3.1) Used by permission of United Nations
p. 28: (In the News 1.4) An excerpt of this article is reproduced Publications.
with permission from Chemistry World, Royal Society of
p. 138: (In the News 3.2 and Fig 3.1) Source: “Sled Dog Days of
Chemistry. The article was first published online on December 23,
Summer,” (Aug. 6, 2014). Boulder, Colorado USA: National Snow
2013, and it appeared in print in the February 2014 issue, p. 26.
and Ice Data Center.
p. 44: (In the News 1.5) From The Boston Globe, 5/2/04 c 2004
p. 158: (In the News 3.3) From David Wright “ASK A
Boston Globe. All rights reserved. Used by permission and
SCIENTIST: How Vulnerable are the Pools that Store Spent
protected by the Copyright Laws of the United States. The printing,
Nuclear Fuel at U.S. Reactors?” Reprintd by permission of The
copying, redistribution, or retransmission of this Content without
Union of Concerned Scientists, uscusa.org.
express written permission is prohibited.
p. 175: (In the News 3.4) From Salvador Rodriguez, “Pinterest,
p. 61: (In the News 1.6) From CNNMoney.com, 10/30/2013
Instagram continue meteoric growth,” The Los Angeles Times,
c 2013 Turner Broadcast Systems, Inc. All rights reserved.
August 29, 2012. Copyright c 2012. Los Angeles Times.
Used by permission and protected by the Copyright Laws of
Reprinted with Permission.
the United States. The printing, copying, redistribution, or
retransmission of this Content without express written p. 177: (In the News 3.5) Source: USGS-Oklahoma Geological
permission is prohibited. Survey Joint Statement on Oklahoma Earthquakes,
http://earthquake.usgs.gov/regional/ceus/products/newsrelease
Ch. 2 05022014.php.
p. 77: (In the News 2.1) From Lauren Hockenson and Rani Molla, p. 194: (In the News 3.6) Source: From Mark Crawford, “Solar
“Facebook grows daily active users by 25 percent, mobile users by Power Shines.” Used by permission of the American Society of
45 percent.” Reprinted by permission of Gigaom, Inc. Mechanical Engineers.
C-1
Crauder-4100237 QL_Credits November 18, 2014 9:20
C-2 Credits
MoneyNing.com. Reprinted by permission. http://moneyning p. 374: (In the News 5.12) Source: National Gambling Impact
.com/retirement/4-retirement-tips-young-adults-can-implement- Study via http://govinfo.library.unt.edu/ngisc/reports/1.pdf
now/.
p. 271: (In the News 4.6) Source: From “The Case For College Ch. 6
Students To Have Credit Cards.” Copyright c 2014 U.S. News p. 393: (In the News 6.1) From Dean Scoville, “What’s Really
and World Report LP. Reprinted by permission. Going On with Crime Rates: Crime stats are used to judge the
p. 280: (In the News 4.7) Source: http://www.consumerfinance effectiveness of law enforcement agencies, but hardly anyone is
.gov/credit-cards/credit-card-act/feb2011-factsheet/ questioning their validity or accuracy,” (October 09, 2013).
Reprinted by permission of POLICE Magazine and Bobit Business
p. 286: (In the News 4.8) From The New York Times, 01/07/2014 Media.
c 2014 The New York Times. All rights reserved. Used by
permission and protected by the Copyright Laws of the United p. 397: (In the News 6.2) From Jonathan Horn, “County new
States. The printing, copying, redistribution, or retransmission of home prices hit record high.” Reprinted by permission of The San
this Content without express written permission is prohibited. Diego Union Tribune.
p. 297: (Figure 4.14) Used by permission of p. 411: (In the News 6.3) Used by permission of Minitab Inc.
www.FedPrimeRate.com. p. 432: (In the News 6.4) Used by permission of Dr. Harry Wilson.
p. 299: (In the News 4.9) Source: http://www.ers.usda.gov/ p. 440: (In the News 6.5) Reproduced with permission of Yahoo.
data-products/food-price-outlook/summary-findings.aspx
c 2014 Yahoo. YAHOO! and the YAHOO! logo are registered
trademarks of Yahoo.
Ch. 5
p. 440: (Gallup Graph for In the News 6.5) From Lydia Saad,
p. 307: (In the News 5.1) From Ginger Pinholster, “Research to “Obama-Romney Race Competitive in 2012 Swing States”.
Address Near-Earth Objects Remains Critical, Experts Say,” Reprinted by permission of Gallup, Inc. All rights reserved.
(July 12, 2013). Copyright
c 2013. Reprinted with permission
from AAAS. p. 442: (In the News 6.6) Source: www.harrisinteractive.com/
vault/Harris-Interactive-Poll-Research-New-Congress-Priorities-
p. 308: (In the News 5.2) Source: p. C-2: United States Geological 2007-01.pdf.
Survey, “2008 Bay Area Earthquake Probabilities,” April, 2008,
http://earthquake.usgs.gov/regional/nca/ucerf. p. 447: (In the News 6.7) NIH Clinical Center. Article Source:
http://clinicalstudies.info.nih.gov/cgi/wais/bold032001.pl?A 14-H-
p. 323: (In the News 5.3) Reprinted by permission of Jeff Haby 0040.html@obesity@@@@#contacts
p. 323: (In the News 5.4) From ‘FIFTY PERCENT chance of p. 457: (In the News 6.8) From Maria Cheng, “Being Bored Could
another 9/11-style attack’ in the next ten years . . . But likelihood Be Bad for Your Health,” The Associated Press (February 10,
is MUCH higher if the world gets less stable.” Copyright c The 2010). Used with permission of The Associated Press Copyright
Daily Mail. Reprinted by permission of Associated Newspapers c 2014.
Limited.
p. 459: (In the News 6.9) Source: NCI Press Release
p. 324: (Exercise 62) Copyright c 1990 Marilyn vos savant. http://www.nih.gov/news/health/feb2014/nci-06.htm.
Initially published in Parade Magazine. All rights reserved.
Reprinted by permission. p. 461: (In the News 6.10) From “What Do You Do When Your
Child Wants to Stop Eating Meat?” The Guardian (February 12,
p. 327: (In the News 5.5) From Dana Kam, “Judge strikes down 2008). Copyright c Guardian News & Media Ltd 2008.
drug testing of Florida welfare recipients.” Used by permission of
The News Service of Florida. p. 462: (In the News 6.11) From Sean Harkin, “Coed Housing
Has Higher Rates of Sex, Binge Drinking,” ocolly.com (December
p. 341: (In the News 5.6) From Susan Du, Stephanie Haines, 2, 2009). Reprinted by permission of ocolly.com.
Gideon Resnick and Tori Simkovic, “How a Davenport murder
case turned on a fingerprint.” Used by permission of The Medill
Justice Project. Ch. 7
p. 353: (In the News 5.7) From Alex Fryer, “Terror Experts Doing p. 469: (In the News 7.1) Source: “Revealing Social Networks of
the Math to Assess Risks,” Seattle Times (August 4, 2003). Spammers,” The National Security Agency’s Review of Emerging
Reprinted by permission of The Seattle Times Company. Technologies (2010). http://www.nsa.gov/research/tnw/tnw183/
articles/pdfs/TNW 18 3 Web.pdf.
p. 354: (In the News 5.8) From Michael McNutt, “In Oklahoma,
Republicans continue to show net increase over Democrats in p. 470: (In the News 7.2) From NextGen: The Future of Flying.
registered voters Oklahoma Republicans had a 3-1 net increase Copyright c Aerospace Industries Association. Reprinted by
over Democrats in registered voters, registration records show.” permission.
Reprinted by permission of The Oklahoma Publishing Co.
p. 490: (In the News 7.3) From William J. Cook, “Traveling
p. 356: (In the News 5.9) Source: Salesman: A Seemingly Unsolvable Problem Offers a Glimpse of the
http://www.genome.gov/11006943. Limits of Computation,” Scientific American 306:6. Reproduced
with permission. Copyright c 2012 Scientific American, a division
p. 370: (In the News 5.10) Used by permission of Zscaler, Inc. of Nature America, Inc. All rights reserved.
p. 372: (In the News 5.11) From Lara Lakes Jordan, “FEMA p. 508: (In the News 7.4) http://www.nsf.gov/news/news summ
Changing Locks on Trailers,” Associated Press, August 15, 2006. .jsp?cntn id=111580&org=IIS.
Used with permission of The Associated Press Copyright c 2014.
Crauder-4100237 QL_Credits November 15, 2014 10:39
Credits C-3
Ch. 8 Ch. 9
p. 525: (In the News 8.1) From Stephanie Condon, “Most p. 603: (In the News 9.1) From Véronique Izard et al., “Flexible
Americans want popular presidential vote” by Stephanie Condon intuitions of Euclidean geometry in an Amazonian indigene
(10/24/11). Reprinted by permission from CBS Interactive Inc. All group,” Proceedings of the National Academy of Sciences of the
rights reserved. United States of America 108(24) 9782–9787. Reprinted by
permission of PNAS.
p. 526: (In the News 8.2) Used by permission of the Disability
Voting Coalition of Pennsylvania. p. 622: (In the News 9.2) Via Dezeen: http://www.dezeen.com/
2014/08/20/lego-golden-ration-of-architecture-big-lego-house-
p. 544: (In the News 8.3) From Josh Waugh, “We need instant denmark-bjarke-ingels-interview/.
runoff voting: It’s time for election reform” (January 5, 2014).
Reprinted by permission of The Daily of the University of p. 637: (In the News 9.3) From Keith Devlin, “Good Stories, Pity
Washington. They’re Not True.” Copyright c 2004 Mathematical Association
of America. Reprinted by permission.
p. 561: (In the News 8.4) From Corky Siemaszko, “Tiger Woods
divorce settlement: Elin Nordegren received $110M payout— p. 639: (In the News 9.4) J E Scheib, S W Gangestad, R Thornhill
report.” c Daily News, L.P (New York). Used with permission. Proc Biol Sci. 1999 September 22; 266(1431): 1913–1917
Copyright © 1999, The Royal Society.
p. 575: (In the News 8.5) From “In Reapportionment
Texas—Florida Gain Congressional Seats.” Copyright
c 2014 U.S.
News and World Report LP. Reprinted by permission.
Crauder-4100237 QL_Credits November 15, 2014 10:39
Crauder-4100237 QL_index November 13, 2014 11:26
INDEX
I-1
Crauder-4100237 QL_index November 13, 2014 11:26
I-2 Index
Index I-3
I-4 Index
Index I-5
I-6 Index
malnutrition, sustainable food systems monthly payments “or” principle, in mathematics and logic,
and, 137–138 chain calculations, A–12 32–33
Malthus, Thomas Robert, 137 college loans, 236–237 outcomes
marginal tax rate, 148–152, 292–295, fixed-rate and adjustable-rate determination, 311–313
304 mortgages, 247–249 election outcome counting and
margin of error installment loans, 234–236, 238–239, probability, 347–348
fashion survey estimation, 435–436 302–303 equally likely outcomes, 313
polling, 432–435 long-term loans, 249–251 simple counting, 341–342
marijuana legalization, 88–90 minimum payments, credit card debt, theoretical probability, 309–311, 388
“market basket” model of goods and 274–279, 303–304 voting systems and differences in,
services, 286–287 mortgages, 246–249 551–553
“marriage penalty” in taxes, 293–295 short-term loans, 240–241 outliers in data, 396–397
married couples, tax rates for, 292–295 monthly yield, retirement income, output value, 79
max-min problems, 201–202 263–267
mean mortgage rates, 244–249 P
bell curve, 413–414 multimodal data sets, 394 parabolas, 198–203, 209–210
calculation, 395–396 multiple transfers, adjusted winner parallelograms, 644–645
data summary and presentation, procedure, 566–567 parentheses
463–464 calculators, A–10
defined, 394 N logic, 36
home prices, 396 nanoparticles, 66 parent vertex
Huntington-Hill method of nanotechnology, 66 binary trees, 510–512
apportionment, geometric mean, natural logarithm, 194 trees, 509
590–592 NCAA graduation rates, measurement passing trucks, statistical significance,
normal distribution, 415–419 of, 78–79 449–451
median nearest-neighbor algorithm, 496–498, paths, graphing and Euler circuits,
bell curve, 413 521–522 472–474
calculation, 395–396 cheapest link algorithm, 498–501 pattern recognition, 22–26
data summary and presentation, negation, 27–28, 73 symmetries and, 638–642, 651
463–464 negative correlation, 451–456, 465 payday loans, 254
defined, 394 negative growth rate, 87–88, 104–105, payment estimates. See monthly
home prices, 396 139–141 payments
median wage decline, Simpson’s paradox negative predictive value, medical tests, pentagrams, 640
and, 3–6 331–332, 388–389 percentage change, 81–85, 131–132
medical tests nest egg, 263–267, 303 Central Limit Theorem, 422–425
conditional probability, 335–336, new states paradox, 586–587, 599 forecasts, 323
388–389 NextGen Air Transportation System, percentage growth and base, 163–165,
Crohn’s disease case study, 332–334 470–471 208
positive predictive value, 330–332 95% confidence level, polling and, percentage of payments, equity
probability, 327–336 435–439 calculations, 241–243
sensitivity and specificity, 328–330 n items, permutation, 358, 389 percentages, application, A–15 to A–17
Venn diagrams, 47–49 n items taken k at a time percentile scores, 420–422
menu options, disjunctions in, 33 combinations, 363 perimeter, 603, 605–606, 649–650
meteorites, area probabilities, 318–319 permutations formula, 359–361, 389 Heron’s formula, 609–611
microns (small numbers), 66 non-occurrence, probability of, 313–315 period interest rate, 219–220
minimum payments, credit card debt, normal distribution, 411–427, 464–465 permutation, 356–362, 366–369, 389
274–281 birthweight, 422 arrangements, 358–360
misleading graphs, 113–126, 133 Central Limit Theorem, 422–425 calculation, 362
axis scale choice, 114–116 health issues and, 425–427 defined, 357
data misrepresentation, 118–122 IQ scores and income, 414–415 factorials, 358
insufficient data, 122–123 mean and standard deviation, n items, 358
mobile phone users, 77–78 415–419 probability, 366–369
mode numbers, Venn diagrams and, 56–57 permutations, iPhone Touch ID, 370
calculation, 395–396 number sense, 2, 60–71, 74 perpetuity arrangements, 263–267, 303
data summary and presentation, personal finance
463–464 O financial literacy, 213
defined, 394 oil prices, graph representations of, 133 savings, 214–229
money management, 286–298 Olympic Games pictorial logic, 2, 44–46, 74
monthly compounding, annual athletic records, 143–144, 152 pictorial misrepresentation, 123–128
percentage yield, 225 drug testing, 327 pie charts, 123–128
monthly deposits, formulas for, 258–262 variables in medal awards, 80–81 Pinterest growth measurement, 175
Crauder-4100237 QL_index November 13, 2014 11:26
Index I-7
pizza measurement, area and diameter, presumption, fallacies of, 2, 15–22 quartile
607–608 prevalence defined, 397
placebo, clinical trials, 448–449 Crohn’s disease case study, percentile scores, 420–422
plane 333–334 quotas
reflectional symmetry, 641–642 disease, 330–332 apportionment and, 578–594
regular tilings, 642–647 positive and negative predictive violation of, 584, 593–594, 599
rotational symmetry in, 639–641 value, 332 voting power and, 527–528, 597
plurality voting, 545–547, 553, 598 prime interest rate, 247–249
poker hands, probability in, 367–368 principal R
political debates installment loans, 235–236 radioactive decay, 169–172, 189–190
conditional meaning in, 34–35 mortgage payments, 244–249 radiocarbon dating, 170–172,
logical fallacies in, 15–16 simple interest, 216 189–190
Venn diagrams and, 51–52 prize winners, counting permutations, radius (circle), 604
polling 359 proportionality and, 631–632
election outcomes, 432 probability ranked balloting, 547–549
margin of error, confidence interval, applications, 307 rate of inflation, 286–287, 304
confidence level, 432–435 area, 316–318 ratio, proportion and, 622–623
methodology, 441–444 calculation, 308–321, 388 recidivism rates, 124–125
permutations, 359–360 counting and, 345–348 rectangle
sample size, 435–439 definitions, 309 defined, 605
“statistical dead heat” and, 440–441 disjunction, 315–316 golden, 627–628
statistics, 431–444, 465 election outcomes, 347–348 Heron’s formula for area of,
polygon, 643 empirical probability, 309–311, 610–611
Ponzi schemes, 513 319–321 rotational symmetry in, 640–641
popular vote, 525–526 expected value and law of large tiling, 642–643
population. See also world population numbers and, 374–383 reflectional symmetry, 641–642, 651
apportionment and, 575–594 fair games and, 379–381 regression line, 150–152, 208
average growth rate, 86–88 independent events, 349–351 regular deposits balance formula,
constant percentage growth, 159–161 medical testing, 327–336 259–260, 267
exponential decay, 166–172 outcome determination, 311–313 regular polygon, 643
graphs, 107–109 permutations and combinations, regular tiling, 643, 651
linear vs. exponential growth, 137 366–369 relative change, 81–85
logarithmic function, 178, 186 p-value, 449–451 relative intensity
Malthusian theory on, 137 theoretical probability, 309–311 decibel ratings, 183–185
normal distribution and, 415 probability sample, polling earthquake, 179–182
percentage change in U.S., 83–85 methodology, 442–444 relevance, fallacies of, 16–19
quadratic formula, 198 problem solving, A–18 to A–20 representation, apportionment and,
in Russia, 88 product formula, independent events, 575–594
standard deviation, 402 349–351 retirement planning, 256–267
United States, table of, 79–80 progressive tax, 293 annuities, 263–266
world population, 85–86 proportionality, 622–634, 650–651 savings assessment, 262–267
population paradox, 586–587, 599 applications, 631–634 reversing conversions, A–3
positive correlation, 451–456, 465 golden ratio φ, 626–628 Richter, Charles, 179
positive growth rate, 104–105 properties, 624–625 Richter scale
positive predictive value (PPV) proportionality constant, 631–634 basic principles, 179–181, 209
Crohn’s disease case study, 333–334 PTA phone tree, 511 interpretation, 182
medical tests, 330–332, 388–389 public opinion polls, 434–435 right triangle, 606
post office routes, graph modeling of, Social Security reform, 442–444 risk assessment, expected value,
474 p-values, 449–451, 465 381–383
powers of 10, A–7 pyramids, golden ratio φ, 626–628 rocket, linear function formula,
preferential voting, 547–553 Pythagorean theorem, 606, 608–609, 143–144
premises, 73 650 rolling dice, probability of
logical argument, 15–16 non-occurrence and, 315
present value, 227–228, 302 Q root vertex, trees, 509
retirement savings, 266, 303 quadratic equations, 195–198 rotational symmetry, 639–641, 651
presidential elections quadratic formula, 196–198, 202 roulette, expected value calculation,
plurality voting and spoilers in, quadratic functions, 194–198, 209–210. 378–379
546–547 See also parabolas rounding
Shapley-Shubik index, 536–539 quantitative literacy, defined, 2 apportionment and, 579, 599
presidential promises, conditional quarterly compounding, doubling time, calculators, A–11
meaning of, 34–35 229–230 compound interest, 222–223
Crauder-4100237 QL_index November 13, 2014 11:26
I-8 Index
route mapping Severe Acute Respiratory Syndrome “statistical dead heat,” 440–441
cheapest link algorithm, 498–501 (SARS), 90–92 statistical inference, 465
degrees of vertices, 475–478 shadows, similar triangles and, drug trials, 447–458
Hamilton circuits, 490–493, 630–631 statistical significance, 449–451, 465
521–522 Shapley-Shubik power index, 534–539, statistics
nearest-neighbor algorithm, 497–498 597–598 boxplots, 399–401
Rule of 72, 228–230, 302 short-terms, payment estimation, Central Limit Theorem, 422–425
running speed calculations, 102–103, 240–241 data summary and presentation,
149–151 similarity, 650–651 394–407
circumference and diameter, 606–607 proportionality and, 622–634 defined, 393–394
runoff elections, 547–553 similar triangles, 628–631 health issues and, 425–427
simple counting, 341–342 histograms, 405–407
S simple interest, 216–218, 301–302 normal distribution, 411–427
salary analysis Simpson’s paradox, 2–10, 72–73 percentile scores, 420–422
linear functions, 140–141 singles, tax rates for, 292–295 polling, 431–444
percentage growth and base, 68-95-99.7% rule, 416–419 standard deviation, 401–405
163–165 Skylab area measurements, 316–318 z-scores, 419–420
sample size, polling and, 435–439 slap shot calculations, 608–609 staying withing quota, 593–594, 599
sample standard deviation, 402 slope stock prices, 295–297, 304
sampling, Central Limit Theorem, correlation coefficient, 455–456 index fund correlation, 456
424–425 interpretation, 144–148 straw man fallacy, 17–18, 21
SAT linear function, 142–144, 208 subsets, 45–46
guessing on, 380–381 tax data, 148–152 subway tour, Euler circuits, 480–482
mean and standard deviation of slot machines, expected value and law of surface area calculations, 614
scores, 416–419 large numbers, 374, 376–377 sustainable food systems, malnutrition
percentile scores, 420–422 smoothed line graphs, 100–103, and, 137–138
z-scores, 419 106–109, 132 swimming pool, volume calculations,
savings insufficient data in, 122–123 612–613
borrowing vs., 239–240 snow accumulation, growth rate, 144 swing voter, 534–535, 597–598
for college, 259–260 social media symmetry, 638–642, 651
compounding, 214–229, 301–302 graph modeling of, 472 reflectional symmetry, 641–642
linear functions, 215 growth measurement of, 175
long-term savings, 256–267, 303 spammer social networks, 469–470 T
monthly deposits, 258–259 Social Security reform, public opinion tables, 79–82
needs assessment, 259–260 on, 442–444 percentage change, 83–86
regular deposits balance formula, solar energy, parabolic measurements, tax tables, 292–295
259–260, 267 194–195 talent show auditions, permutations,
retirement, 262–267 sound measurement, decibel rating, 361–362
simple interest, 216 183–185 taxable income, 292–295
scales. See compressed scales specificity, medical tests, 328–330 tax bracket, 293
proportionality and similarity, speeding tickets, Venn diagrams of, tax credits, 294–295, 304
622–634, 650–651 50–51 tax tables, 292–295
scaling arguments, proportionality, speed of sound, slope calculation and television drug ad, misleading
633–634 interpretation, 146–148 representations in, 113–114
scatterplots, 99–103, 106–109, 132 spell checkers, 508, 513–518, 522 temperature
trend lines, 150–152 spent nuclear fuel, exponential growth conversion, slope interpretation,
school mapping, nearest-neighbor and decay and, 158–160 146–147
algorithm, 497 spherical measurements, 613–614 running speed and, 102–103
scientific notation, A–7 to A–9 spoilers, plurality voting and, 545–547 slope calculation and interpretation,
sealed bid method, fair division/fair sports participation, Venn diagrams of, 145–148
share, 568–571 54–56 term of loan, 235, 302–303
search engines, 29–31 squares mortgages, 246–249
secret bidding, fair division/fair share, rotational symmetry in, 640–641 terrorism, probability, 323, 353
568–571 tiling, 642–643 tessellation. See tilings
seisometers, 179 standard deviation, 401–405, theoretical probability, 309–311
semi-perimeter, 609–611 463–464 counting and, 341–351
semi-regular tiling, 644–645 normal distribution, 415–419 thirty-year mortgages, 245
sensitivity, medical tests, 328–330 z-scores, 419–420 three-dimensional objects, 650
sensitivity of 93%, 330 standard divisor, apportionment and, volume calculations, 611–614
sets, 2, 44–46, 74 578–579 three-judge panels, combinations
Venn diagrams, 54–56 standard score. See z-scores computation, 365–366
Crauder-4100237 QL_index November 13, 2014 11:26
Index I-9
three-person committee, combinations true positives, medical tests, 328–330, viral e-mails, 508–518, 522
formula, 363 388–389 volume
thrown rock problem, quadratic truth tables, 2, 27–43, 73 calculations, 611–614, 649–650
function, 197–198 complex statements, 36–38 estimation, 69–70
tilings, 642–647, 651 tuberculosis proportionality and comparisons
Escher tilings, 645–646 conditional probability and medical of, 625
top-two runoff system, 547, 553, 598 testing, 335–336 voter registration, probability, 354
towns and roads test sensitivity and specificity and, 329 voting
Eulerizing graphs of, 483–484 tuition cost apportionment, 575–594
graph modeling of, 472, 474 average growth rate, 88 Arrow’s impossibility theorem,
toxic waste management, graphic college loans, 236–237 556–558
analysis of, 107–108 tuition deferment process, 234 Banzhaf power index, 527, 531–534
traffic lights, outcomes and probability, Condorcet’s paradox, 553–556
312–313 U fair division, 561–571
transitive preferences, 553 plurality voting and spoilers,
uncorrelated variables, 451–456, 465
traveling salesman problem, 494–502, 545–547
undergraduate enrollment, pictorial
521–522 power of, 526–539, 597
representations of, 123–124
cheapest link algorithm, 498–501 preferential voting, 547–553
unfavorable outcome, theoretical
computational limits, 490–491 rules, 525–526
probability, 309–311, 388
nearest-neighbor algorithm, runoffs, 547–553
unit conversion, A–1 to A–5
496–498 swing voters, 534–535
untreated patients, Central Limit
trees, 508–513, 522 systems, 544–553
Theorem, 423–425
binary trees, 510–512 voting coalition, 526–527
formulas, 512–513
spell checkers, 513–518 V W
trend lines, 148–152, 208 valence. See degrees of vertices Webster, Daniel, 587–590, 599
correlation in, 452–456 valid argument, 15–16, 73 weights of apples, 68-95-99.7% rule and
trial and error calculations, credit card variables counting data for, 418–419
debt, 278 correlation, 451–456 width calculations, three-dimensional
triangle table of, 80–81 objects, 611–614
area, 650 vegetarianism, statistical research on, winning coalition, 527, 529–530, 597
defined, 605 461 world population, 85–86. See also
equilateral triangle, 643–644 Venn diagrams, 2, 44–57, 74 population
Heron’s formula for area vertex/vertices, 198–201, 210, 470
measurement, 609–611 degrees, 474–478 Y
irregular tilings, 644–645 Euler circuit construction, 478–482
Yellen, Janet, 286
right triangle, 606 graphing and Euler circuits, 471–474
similar triangles, 628–631 Hamilton circuits, 492–493
Z
trucking routes, cheapest link algorithm, nearest-neighbor algorithm, 496–498
trees, 508–518 zeros, quadratic functions, 195–198
500–501
violate quota, 584, 593–594, 599 z-scores, 419–420, 464–465
true negatives, medical tests, 328–330,
Vioxx trials, p-value, 450–451 percentile scores, 420–422
388–389
INDEX OF APPLICATIONS
Education Dow Jones Industrial Average (DJIA), pp. 128 (Exercise 16),
192 (Exercises 42–44), 296, 298 (Exercise 6), 299
Affect of poverty on student success rates, p. 12
(Exercise 33), 300 (Exercises 34–35)
(Exercises 14–22)
Equity, pp. 242, 253 (Exercises 28–29)
AP examinations, p. 96 (Exercises 38–39)
Estimating interest, p. 217
College expenditures, p. 95 (Exercises 16–19)
Estimating payments for long-term loans, p. 249
College freshmen grades, p. 94 (Exercises 12–15)
Estimating payments for short-term loans, pp. 240, 253
Enrollment in kindergarten, p. 94 (Exercises 7–8)
(Exercise 37), 255 (Exercise 52)
Enrollment at U.C. Davis, p. 123
Expected cost in manufacturing, p. 382, 386 (Exercises 32–33)
Exam scores, pp. 11 (Exercise 1), 429 (Exercises 29–30)
Expected profit in insurance, p. 382, 385 (Exercises 30–31)
High school costs, p. 95 (Exercises 27–31)
Federal receipts, p. 130 (Exercise 30)
High school graduation rates, p. 155 (Exercise 49)
Federal tax collected, p. 154 (Exercises 36–38)
IQ scores, p. 414, 428 (Exercises 14–15)
Finance charge, pp. 273, 282 (Exercises 1–2)
Math scores, pp. 11 (Exercise 9–10)
Flat tax, p. 299 (Exercises 21–30)
National ACT scores, p. 118
Flood insurance, p. 382
Oklahoma ACT scores, pp. 116, 127 (Exercises 10–12)
Future and present value, pp. 227, 232 (Exercises 37–38)
Oklahoma State University donations, p. 127 (Exercise 6)
Gasoline prices, pp. 68, 117, 120
SAT exam, p. 380
German gold bonds, p. 173 (Exercise 15)
SAT verbal scores, pp. 416, 420, 428 (Exercises 12–13, 16–18)
Gold prices, p. 67
Teaching financial literacy, p. 213
Gross Domestic Product (GDP), p. 175 (Exercises 48–49),
Tuition cost, p. 88
192 (Exercises 37–38)
Historical inflation, p. 288
Finance Home prices, p. 396, 408 (Exercise 4), 410 (Exercise 27)
Home mortgage, pp. 244–246, 253 (Exercises 22–27,
Adjustable rate mortgage (ARM), p. 247–248
38, 41–46)
Adjusting defense spending for inflation, p. 129 (Exercise 25)
Hyperinflation, pp. 289, 299 (Exercises 17–18)
Advertising expense, p. 461 (Exercises 19–22)
Income tax for married couples, pp. 148, 155 (Exercises 53–57),
Amazon’s Kindle price, p. 153 (Exercises 22–25)
293, 298 (Exercise 4)
Amortization table, pp. 241, 253 (Exercises 20–21),
Income tax for singles, pp. 153 (Exercise 19), 154
254 (Exercise 51)
(Exercises 41–44), 293–294, 299 (Exercise 32)
Annual percentage rate (APR), pp. 219, 230 (Exercises 4, 7–8),
Inflation, pp. 95 (Exercises 23–26), 129 (Exercises 26–29),
231 (Exercises 14–16), 232 (Exercise 32–33, 35, 42–44)
174 (Exercises 30–31), 192 (Exercises 31–33), 298
Annual percentage yield (APY), pp. 223, 231 (Exercises 20–25),
(Exercises 8–16), 299 (Exercises 19–20)
232 (Exercise 36)
Inheritance, pp. 565, 569
Annuities, pp. 264–265, 268 (Exercises 17–18)
Insurance payments, p. 253 (Exercises 30–33)
Average daily balance, p. 273
Investment growth, pp. 161, 163, 187–188, 192
Buying power, pp. 96 (Exercises 40–41), 290, 296 (Exercise 3)
(Exercises 35–36)
Certificate of deposit, p. 214
Limit order, p. 300 (Exercise 37)
Compound interest, pp. 216, 218, 220, 230 (Exercises 2–3),
Marginal tax rate, p. 154 (Exercise 40)
231 (Exercises 9–13, 18–19)
Market order, p. 300 (Exercise 36)
Consumer price index (CPI), pp. 155 (Exercises 50–52), 289, 298
Mathematics faculty salaries, p. 408 (Exercise 11)
(Exercises 1–2, 7)
Microloans, p. 253 (Exercise 40)
Credit card, pp. 277, 282 (Exercises 5–6), 283 (Exercises 7–10),
Minimum wage, p. 119
284 (Exercises 31–36)
Monthly payment formula, pp. 235, 237, 252–253
Credit card minimum payment, pp. 275, 282 (Exercises 3–4),
(Exercises 1–19)
283 (Exercises 11–30), 284 (Exercises 39–40)
Mortgage interest deduction, p. 300 (Exercises 45–48)
Currency in circulation, p. 118
National debt, p. 65
Currency conversion, p. 27 (Exercise 38)
New York City home prices, p. 352
Deflation, p. 174 (Exercises 32–33)
Newspaper/Magazine sales, p. 111 (Exercise 14, 21)
Diversifying index funds, p. 456
Oklahoma Teachers’ Retirement System, p. 153 (Exercise 18)
Doubling time for investment, pp. 228, 232 (Exercises 39–41)
Payday loans, p. 254 (Exercises 48–50) Medicine
Paying for college, pp. 236, 261, 268 (Exercise 3)
Age of first-time mothers, p. 92
Percentage change for an investment, p. 94 (Exercises 1–4)
Allergies, pp. 424, 430 (Exercise 41)
Perpetuity, p. 269 (Exercises 29–36)
Amoxicillin, p. 167
Ponzi scheme, pp. 175 (Exercise 43), 513
Antibiotics in the bloodstream, p. 167
Property tax and home size, p. 618 (Exercises 37–38)
Basal metabolic rate (BMR), p. 637 (Exercise 53)
Rate of return on investment, p. 386 (Exercises 34–35)
Benefits of breastfeeding, p. 460 (Exercises 10–11)
Rebates versus 0% financing, p. 253 (Exercises 34–36)
Blood pressure, p. 427 (Exercise 6–7)
Regular deposits, pp. 258, 260–261, 268 (Exercises 1–3, 15–16),
Body mass index (BMI), p. 322 (Exercises 46–48)
269 (Exercises 27–28)
Breast feeding and brain power, p. 460 (Exercises 10–11)
Retirement planning, pp. 174 (Exercise 37), 231 (Exercise 17),
Cancer mortality rates, pp. 103, 126 (Exercise 3)
261, 268 (Exercises 8–11, 19–26, 37–41), 270
Clinical drug trials, pp. 49, 459 (Exercises 6–8)
(Exercises 44–45)
Dental care, p. 453
Short sale, p. 300 (Exercise 39)
Diabetes, p. 430 (Exercise 42)
Salary growth, p. 164
DPT injections, p. 462 (Exercise 28)
Sales commission, pp. 153 (Exercise 20), 154 (Exercise 39)
Fallon leukemia cluster, p. 425
Savings accounts, p. 268 (Exercises 4–7, 12–14)
Fat epidemic, p. 427 (Exercise 2)
Small business expenses, p. 110 (Exercise 1)
Flu epidemic, pp. 110 (Exercise 11), 154 (Exercises 45–48),
Simple interest, pp. 216, 230 (Exercises 1, 5–6)
174 (Exercise 24)
Small business income, p. 105
Genetic testing, p. 338 (Exercises 29–31)
Stop loss, p. 300 (Exercise 38)
Heart health, p. 173 (Exercise 9), 322 (Exercise 45)
Tax bracket creep, p. 299 (Exercise 31)
Health care expenditures, pp. 130 (Exercises 32), 165
Tax cuts, p. 115, 445 (Exercise 26)
Heights of men/women, pp. 110 (Exercise 9), 413, 418, 428
Wage averages, p. 5
(Exercises 25–28), 430 (Exercise 43)
Yahoo stock, p. 192 (Exercise 41)
Hospital choice, p. 11 (Exercises 3–6)
Geography and Population Dynamics Length of illness, pp. 421, 423
Life expectancy, p. 156 (Exercises 58–61)
Areas of states, pp. 408 (Exercise 6), 618 (Exercise 44–51)
Lung cancer clinical trials, p. 447
Russia population, p. 87
Lupus, p. 337 (Exercise 9)
U.S. population, pp. 80, 83, 87, 96 (Exercises 32–34), 174
Medical test: Crohn’s disease, pp. 333, 336 (Exercises 6–8),
(Exercise 23)
337 (Exercises 16–17)
World population, pp. 72 (Exercises 24–26), 85, 173
Medical test: Hepatitis C, pp. 48, 331
(Exercise 22), 192 (Exercise 34)
Medical test: HIV, p. 338 (Exercises 36–39)
Yosemite National Park, p. 436
Medical test: PSA test, pp. 58 (Exercises 25–28),
Life Sciences 338 (Exercises 32–35)
Animal populations, pp. 110 (Exercise 10), 111 Medical test: TB, pp. 329, 335
(Exercises 15–16, 20) Misleading drug ads, p. 113
Apple weights, p. 418 Obesity, p. 429 (Exercise 40)
Arctic sea ice extent, p. 138 Prescription drug sales, p. 110 (Exercise 5)
Bacteria population growth and decay, pp. 158, 160, 166 Reducing ovarian cancer risk, p. 459 (Exercise 9)
Birds and graphs, p. 485 (Exercise 7) SARS, p. 90
Cricket chirp thermometer, p. 145 Sensitivity and specificity, pp. 330, 336 (Exercises 1–4), 337
Endangered species population, p. 110 (Exercise 3) (Exercises 10–11, 14–15, 18–19)
Fingerprints, p. 427 (Exercise 5) Smoking and heart disease, p. 349
Forestry, p. 623 Testing for drug abuse, p. 337 (Exercises 20–28)
George Reserve deer herd, p. 108 Vioxx, p. 450
Reflectional symmetry of faces, p. 639 Weights of newborns, pp. 419, 422, 428 (Exercises 19–22)
Running speed versus length, pp. 101, 149 Physical Sciences
Running speed versus temperature, p. 102
Speed of sound versus temperature, p. 146 Acid rain, p. 191 (Exercise 25)
Sunflower height versus age, p. 109 Bay Area earthquake probabilities, p. 308
Toxic waste disposal, p. 107 (continued in back of book)
INDEX OF APPLICATIONS
Beer-Lambert-Bouguer law, p. 168 1992 presidential election, p. 558 (Exercise 5), 559
Cesium-137, pp. 174 (Exercise 26), 192 (Exercises 28-29) (Exercises 37–39)
Climate change, p. 98 2000 presidential election Florida recount, p. 560
Cobalt-60, p. 192 (Exercise 30) (Exercises 40–42)
Comparison of important earthquakes, pp. 182, 190–191 2012 U.S. presidential election, pp. 536, 540 (Exercise 32)
(Exercises 11–14), 192 (Exercise 39) 2014 Wisconsin gubernatorial race, p. 444 (Exercise 11)
Construction/landscaping dimensions, p. 204–205 Affordable Care Act, p. 34
(Exercises 18–22, 25–26) Afghan war, p. 444 (Exercises 14–16)
Decibel scale, pp. 183, 185, 190 (Exercise 5), 191 Alabama paradox, p. 585
(Exercises 18–22) California Proposition 187, p. 37
Distance traveled by a rocket, p. 143 City council, pp. 580, 583, 594 (Exercises 11–22)
Earthquake magnitude compared with intensity, p. 179–182, Congressional approval, pp. 435, 445 (Exercise 23)
190 (Exercises 3–4) Electoral college, pp. 525, 541 (Exercises 35–36)
Energy of earthquakes, p. 191 (Exercises 15–17) European Union Council, pp. 534–535
Engine displacement, p. 70 2010 Congress, p. 588, 590, 594 (Exercises 2, 5–6)
Cleaning contaminated water, p. 175 (Exercises 38–40) George Washington’s first veto, pp. 581, 595 (Exercise 8)
Cleaning waste sites, p. 175 (Exercises 41–42) Government surveillance, p. 444 (Exercises 8–10)
Folding paper, p. 174 (Exercises 27–29) Hamilton’s apportionment method and small states, p. 594
Half-life, p. 169 (Exercise 10)
Light intensity, p. 460 (Exercises 16–17) Instant runoff voting, p. 544
Mailing boxes, p. 632 Legalizing marijuana, pp. 88, 445 (Exercise 25)
Mars rovers, p. 71 (Exercise 15) Ninth Circuit Court, pp. 365, 372 (Exercises 47–48)
Meteorology forecasts, p. 323 (Exercises 50–51) Party switch, p. 540 (Exercises 33–34)
Nanoparticles, p. 66 Political party preference, p. 354 (Exercises 59–61), 445
Newton’s law of heating, p. 173 (Exercises 17–18) (Exercises 28–33)
Old Faithful eruptions, p. 100 Poll watchers, p. 359, 371 (Exercise 40)
Paper airplanes, p. 110 (Exercise 8) Romney-Obama election, statistical dead heat, p. 440
Path of a cannonball, p. 204 (Exercises 10–13) Settlement with Shoshone tribe, p. 232 (Exercise 42)
pH scale, p. 191 (Exercise 26) Social Security reform, p. 442
Probability of meteor strike, p. 318, 324 (Exercise 56), United Nations Security Council, p. 541 (Exercises 37–40)
339 (Exercise 45) Violating quota, p. 595 (Exercise 23)
Radiocarbon dating, p. 170, p. 175 (Exercises 44–45, 46), 189,
190 (Exercise 8) Social Networking
Richter scale, pp. 179–182, 192 (Exercise 40)
Shadows, p. 630 Blogosphere growth, p. 174 (Exercises 34–36)
Speed of light, p. 71 (Exercise 16) Chain letters/emails, pp. 508, 518 (Exercise 15), 519 (Exercise 18)
Speed of sound versus ocean depth, p. 153 (Exercises 26–31) Facebook, pp. 77
Stellar magnitude, p. 191 (Exercises 23–24) Growth of social networks, p. 175 (Exercise 47)
Temperature conversion, p. 147 Web conferencing, p. 128 (Exercises 14–15)
Temperature in Campinas, p. 428 (Exercises 23–24)
Temperature in Madison, p. 427 (Exercises 3–4) Social Sciences
Tsunami waves in Crescent City, p. 173 (Exercises 19–20)
Abortion attitudes, p. 95 (Exercises 20–22)
Ages of federal employees, p. 129 (Exercises 21–24)
Political Science Berkeley gender discrimination case, p. 6
1790 Congress, pp. 594 (Exercises 1, 3–4, 8), 595 Car production, p. 122
(Exercises 28–31) Casino revenues, pp. 374, 384 (Exercises 7–8)
1810 apportionment, pp. 577, 582 Celebrity income, p. 398
1820 Congress, pp. 596 (Exercise 32) Coed housing, p. 461 (Exercise 26)
1932 Democratic convention, p. 532 Defective products, pp. 350, 382, 384 (Exercise 4), 386
1996 U.S. presidential election, p. 546, 560 (Exercise 39) (Exercises 32–33)
Defense expenditures, p. 114 Long jump record, p. 143
Divorce settlement, p. 561, 598 NCAA Basketball Tournament, pp. 81, 94 (Exercise 6), 354
Draft lottery, p. 370 (Exercises 20–22) (Exercises 62–63), 511, 519 (Exercises 21–24)
Enigma machine, p. 370 (Exercises 17–18) NCAA graduation rates, pp. 78, 110 (Exercise 6), 126
Enrollment in language courses, pp. 96 (Exercises 35–37), (Exercises 1–2)
111 (Exercises 17–19) Poker, p. 367
Fashion survey, p. 436 Tennis serve, p. 636 (Exercise 32)
FEMA trailers, p. 372 (Exercises 50–54) Winter Olympics, p. 80
Fertilizer for a lawn, p. 618 (Exercises 39–43) Female college coaches, p. 127 (Exercise 4)
Gambler’s fallacy, pp. 377, 385 (Exercise 16), 386 Skiing safety, p. 322 (Exercise 49)
(Exercises 41–44)
Golden rectangle, p. 628, 635 (Exercises 17–27), 637 Technology
(Exercise 56)
Autocomplete, p. 519 (Exercise 25)
Health insurance, p. 31
Computer memory, p. 64
Hiring practices, p. 12 (Exercises 7–13)
Computer/internet network connections, pp. 485
Historical sites in Philadelphia, p. 503 (Exercise 8)
(Exercises 5, 8), 503 (Exercises 7, 9)
Hurricane aftermath, pp. 372 (Exercises 50–54, 57), 437
The Gentoo Foundation, p. 554
Insured children in South Africa, p. 13 (Exercises 23–28)
Incidence matrix, pp. 488 (Exercises 35–38), 519 (Exercise 26)
Judicial probability, p. 365, 372 (Exercises 47–48)
Internet domain hosts, p. 173 (Exercises 12–14)
Kites and darts, p. 648 (Exercise 24)
iPhone TouchID, p. 370 (Exercise 19)
Königsberg bridges, p. 476
Logic gates, p. 40
License plates, p. 304
Monitor sizes, p. 617 (Exercises 33–35)
Meals on Wheels, p. 504 (Exercise 15)
Spell checker, pp. 515–516, 518 (Exercises 9–13),
Minorities employed at the federal level, p. 128 (Exercise 20)
520 (Exercise 28)
Monty Hall problem, p. 324 (Exercise 62–63)
Panels of judges, p. 324
Phone numbers, p. 351 (Exercises 8–10) Transportation
Spending on alcohol vs. tobacco, p. 452, 462 (Exercise 33) Airline customer complaints, p. 94 (Exercises 9–11)
State lottery, pp. 376, 384 (Exercise 1), 386 (Exercises 45–48) Airline routes, pp. 471, 475–47
Swimming pool, p. 612 Amtrak rail map, p. 475
Symmetry and art, p. 628 Automobile accident probability, p. 354 (Exercise 58)
Terrorist attack probability, pp. 323 (Exercise 52), 353 Automobile accidents in New York, pp. 94 (Exercise 4),
(Exercises 50–53) 127 (Exercise 9)
Tulsa State Fair, p. 128 (Exercises 17–19) Gas consumption, p. 144
Vegetarians, p. 461 (Exercise 24) Gas mileage, p. 400, 408 (Exercise 8)
Wedding costs, p. 128 (Exercise 13) Highway collisions, p. 54
Women employed during World War II, p. 204 (Exercise 14) Luggage size, p. 635 (Exercise 5)
New York City subway, p. 480
Sports Speed on a highway curve, p. 154 (Exercises 32–35)
2013 Heisman trophy voting, p. 549, 559 (Exercise 36) Speed on a straight road, p. 112 (Exercise 22)
Baseball and basketball sizes, p. 619 (Exercises 55–58) Traffic citations, p. 50
Batting averages, pp. 8, 11 (Exercise 2) Traffic headway, p. 173 (Exercise 16)
Chelsea Football Club, pp. 395, 405 Traffic signal probability, p. 312, 320, 321 (Exercise 7),
College Football Playoffs (CFP), p. 519 (Exercise 19) 353 (Exercises 41–45)
Division III playoff, p. 519 (Exercise 20) Traveling salesman problem, pp. 494, 505 (Exercises 21–23),
Earned run average (ERA), p. 402 506 (Exercises 30–35)
Free throw accuracy, p. 404, 409 (Exercise 14) Trucking routes, p. 500
High school sports survey, p. 54 UK ferry routes, p. 473
Hockey slap shot, p. 608 Vehicle accident rate, p. 204 (Exercise 15)
Home runs, pp. 94 (Exercise 5), 151 Vehicle parking times, p. 204 (Exercise 16)