SPA 3e_ Teachers Edition _ Ch 6
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
6<br />
Sampling Distributions<br />
Please read the Introduction to the Teacher’s <strong>Edition</strong>.<br />
It will help prepare you for teaching this course, as it<br />
includes a lot of helpful information and advice.<br />
The Big Picture<br />
This chapter focuses on sampling distributions. A sampling<br />
distribution describes the possible values of a statistic such<br />
as the sample mean x or the sample proportion p^ and how<br />
often they occur. Three characteristics of sampling distributions<br />
will be examined in detail: center, variability, and shape.<br />
These are the same three characteristics used in <strong>Ch</strong>apter 1<br />
to describe distributions of quantitative data. The mean and<br />
standard deviation measure the center and variability of<br />
sampling distributions. The shapes of sampling distributions<br />
will be described with the same terms in use since <strong>Ch</strong>apter 1:<br />
skewed left, skewed right, symmetric, mound-shaped, and a<br />
new term from <strong>Ch</strong>apter 5: approximately normal.<br />
The variables examined in this chapter are examples of<br />
random variables, which were introduced in <strong>Ch</strong>apter 5.<br />
The sample count X is a binomial random variable and<br />
is therefore discrete. The sample proportion p^ is closely<br />
related to the sample count X. Finally, the sample mean x<br />
is a continuous random variable.<br />
The process of making a conclusion about a population<br />
based on the data in a sample is called statistical inference.<br />
<strong>Ch</strong>apter 6 lays the foundation for the statistical inference<br />
techniques learned in <strong>Ch</strong>apters 7–10. <strong>Ch</strong>apter 6 describes<br />
the sampling distribution of a sample statistic when certain<br />
characteristics are known about a population. In future<br />
chapters, we will test claims and estimate population<br />
parameters using what we have learned about these sampling<br />
distributions, even when population characteristics<br />
are unknown.<br />
The sampling distribution of p^ , the sample proportion,<br />
and x, the sample mean, are of particular importance for<br />
<strong>Ch</strong>apters 7–9. In those chapters, the values of unknown<br />
population proportions and population means will be estimated,<br />
and claims about them will be tested. Furthermore,<br />
estimates will be made and claims will be tested about the<br />
difference in two proportions and the difference in two<br />
means.<br />
Pacing and Assignment Guide<br />
Day Lesson Learning Targets/Classroom Activities Suggested Assignment<br />
1 <strong>Ch</strong>. 6 Introduction Lesson 6.1 Activity: A penny for your thoughts? None<br />
2 6.1 What Is a Sampling<br />
Distribution?<br />
• Distinguish between a parameter and a statistic.<br />
• Create a sampling distribution using all possible samples from a<br />
small population.<br />
• Use the sampling distribution of a statistic to evaluate a claim<br />
about a parameter.<br />
1–15 odd, 19<br />
3 6.2 Introduction Lesson 6.2 Activity: How many craft sticks are in the bag? None<br />
4 6.2 Sampling<br />
Distributions: Center<br />
and Variability<br />
5 6.3 Sampling<br />
Distribution of a Sample<br />
Count (The Normal<br />
Approximation to the<br />
Binomial Distribution)<br />
• Determine if a statistic is an unbiased estimator of a population<br />
parameter.<br />
• Describe the relationship between sample size and the variability<br />
of a statistic.<br />
• Calculate the mean and the standard deviation of the sampling<br />
distribution of a sample count and interpret the standard<br />
deviation.<br />
• Determine if the sampling distribution of a sample count is<br />
approximately normal.<br />
• If appropriate, use the normal approximation to the binomial<br />
distribution to calculate probabilities involving a sample count.<br />
1–15 odd, 19<br />
1–15 odd, 19<br />
6-2<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 2<br />
11/01/17 3:51 PM
6 Flex Day Consider giving Quiz 6A: Lessons 6.1–6.3, showing one or more of<br />
the online videos listed in the Additional <strong>Ch</strong>apter 6 Resources, or<br />
re-teaching (if needed).<br />
Optional assignment:<br />
6.1 Ex12, 6.2 Ex10, 6.2 Ex14,<br />
6.3 Ex14<br />
7 6.4 The Sampling<br />
Distribution of the<br />
Sample Proportion<br />
• Calculate the mean and standard deviation of the sampling<br />
distribution of a sample proportion p^ and interpret the<br />
standard deviation.<br />
• Determine if the sampling distribution of p^ is approximately<br />
normal.<br />
• If appropriate, use a normal distribution to calculate probabilities<br />
involving p^ .<br />
1–15 odd, 19<br />
8 6.5 The Sampling<br />
Distribution of the<br />
Sample Mean<br />
• Find the mean and standard deviation of the sampling distribution<br />
of a sample mean x and interpret the standard deviation.<br />
• Use a normal distribution to calculate probabilities involving x<br />
when sampling from a normal population.<br />
1–15 odd, 19<br />
9 6.6 The Central Limit<br />
Theorem<br />
• Determine if the sampling distribution of x is approximately<br />
normal when sampling from a non-normal population.<br />
• If appropriate, use a normal distribution to calculate probabilities<br />
involving x.<br />
1–15 odd, 19<br />
10 Flex Day This is a great day to have students work in groups on the STATS<br />
applied! at the end of Lesson 6.6. Also consider showing one<br />
or more of the online videos listed in the Additional <strong>Ch</strong>apter 6<br />
Resources, giving Quiz 6B: Lessons 6.4–6.6, or re-teaching (if<br />
needed).<br />
Optional assignment:<br />
6.4 Ex14, 6.5 Ex10, 6.5 Ex12,<br />
6.6 Ex10, 6.6 Ex12<br />
11 <strong>Ch</strong>. 6 Review <strong>Ch</strong>. 6 Practice Test <strong>Ch</strong>. 6 Review Exercises 1–6<br />
12 <strong>Ch</strong>. 6 Test <strong>Ch</strong>. 6 Test<br />
To save time, it is possible to skip Lessons 6.2 and 6.3<br />
without losing much continuity with future chapters. However,<br />
the other lessons in this chapter are crucial to understanding<br />
much of the remainder of the course.<br />
Four kinds of exercises end each lesson: Mastering Concepts<br />
and Skills, Applying the Concepts, Extending the Concepts,<br />
and Recycle and Review. They have been written so<br />
that assigning the odd-numbered exercises provides appropriate<br />
practice. Mastering Concepts and Skills exercises<br />
address a single learning target, while Applying the Concepts<br />
exercises address two or more learning targets. Recycle<br />
and Review exercises reinforce concepts learned earlier. For<br />
exceptional or motivated students, Extending the Concepts<br />
exercises are a good way to differentiate instruction.<br />
The even-numbered exercises form a pair with the preceding<br />
odd exercise in the Mastering Concepts and Skills<br />
and Applying the Concepts sections so that another full<br />
assignment can be created from the even-numbered exercises<br />
of these types. Consider using the even-numbered<br />
exercises to spiral into future lessons, for re-teaching, or<br />
as additional practice for students. The answers to the<br />
odd-numbered exercises appear in the back of the student<br />
textbook, while the answers to the even-numbered exercises<br />
do not. The answers to all of the <strong>Ch</strong>apter Review<br />
Exercises and the <strong>Ch</strong>apter Test are in the back of the<br />
student textbook so that students can check their own<br />
progress.<br />
C H A P T E R 6 • Sampling Distributions 6-3<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 3<br />
11/01/17 3:51 PM
Promoting Good Habits and Skills<br />
<strong>Ch</strong>apter 6 is about the characteristics of sampling distributions.<br />
The sampling distribution of the sample proportion<br />
and sample mean will play a key role in <strong>Ch</strong>apters 7–9, so<br />
understanding them is very important for future success.<br />
Here are some important habits to develop in your students<br />
as you teach <strong>Ch</strong>apter 6:<br />
1. Pay attention to the vocabulary: Understanding the differences<br />
among population, parameter, sample, and statistic is<br />
absolutely vital to understanding this and future chapters.<br />
Also, make sure your students understand the difference<br />
between the distribution of the population, the distribution<br />
of a single sample, and the sampling distribution. They are<br />
not the same!<br />
2. Emphasize symbols, not formulas: The symbols used in<br />
this chapter are quite standard in all of statistics. While we<br />
don’t want the symbols to overwhelm students, they are<br />
an integral part of basic statistical practice. The symbols<br />
n, p, p^ , m, s, and x will be used extensively in <strong>Ch</strong>apters<br />
7–9. Being comfortable with them now will be of great<br />
help later. On the other hand, don’t have students memorize<br />
the formulas for the mean and standard deviations of<br />
the sampling distributions in this chapter. Having students<br />
understand the symbols is far more important than having<br />
them memorize the formulas.<br />
3. Look for the underlying variable: If your students can<br />
recognize the underlying variable and classify it as categorical<br />
or quantitative, they can tell which sampling distribution<br />
is appropriate. Categorical variables (like color of<br />
a Reese’s Pieces ® candy) lead to sample counts and sample<br />
proportions. Quantitative variables (like the year a penny<br />
was manufactured) lead to sample means. Developing this<br />
skill now will pay dividends in future chapters as well!<br />
4. Think back to the simulations: Because sampling distributions<br />
are very abstract, simulating the sampling process can<br />
provide insight for students. If students have trouble understanding<br />
the different distributions in sampling situations,<br />
have them think back to concrete simulations like the<br />
“A penny for your thoughts?” activity in Lesson 6.1 and<br />
computer simulations with software like the <strong>SPA</strong> applets.<br />
Physical simulations are so important for student understanding<br />
that if you were to do only one activity in the entire<br />
chapter, it should be the penny activity.<br />
5. Watch the conditions: Students will often pay little attention<br />
to the conditions about sampling distributions. Have<br />
them focus on the Large Counts condition and the Normal/<br />
Large Sample condition because these concepts will be used<br />
repeatedly in future chapters. If students don’t pay attention<br />
to them now, they will have a more difficult time in the<br />
future.<br />
Lesson-by-Lesson Content<br />
Overview<br />
Lesson 6.1 What Is a Sampling Distribution?<br />
A large collection of individuals is called a population.<br />
A subset of that population is called a sample. A number<br />
that measures some characteristic of a population is<br />
called a parameter, while a numerical measure of some<br />
characteristic of a sample is called a statistic. Statistics<br />
vary from sample to sample. The distribution of values<br />
taken on by a statistic from every possible sample of a<br />
given size is called the sampling distribution of that statistic.<br />
We can evaluate claims about a parameter by calculating<br />
probabilities from the sampling distribution of<br />
the corresponding statistic.<br />
6-4<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 4<br />
11/01/17 3:51 PM
Lesson 6.2 Sampling Distributions:<br />
Center and Variability<br />
If the mean of the sampling distribution of a statistic is equal<br />
to the corresponding population parameter, the statistic is<br />
said to be an unbiased estimator of the population parameter.<br />
Otherwise, the statistic is a biased estimator of the population<br />
parameter. The standard deviation of a sampling distribution<br />
of a statistic measures the variability of a statistic. The smaller<br />
the standard deviation, the more precise the estimate of the<br />
parameter. The variability of the sampling distribution of a<br />
statistic will decrease as sample size increases.<br />
Lesson 6.3 The Sampling Distribution of a Sample<br />
Count (The Normal Approximation to<br />
the Binomial)<br />
Let the random variable X be the count of successes in a sample<br />
of size n, where p is the probability of a success on a single<br />
trial. The sampling distribution of X will have mean m X = np<br />
and standard deviation s X = "np(1 − p). The Large Counts<br />
condition states that the sampling distribution of X will have<br />
an approximately normal distribution whenever np ≥ 10<br />
and n(1 2 p) ≥ 10. When the sampling distribution of X<br />
is approximately normal, probabilities involving the sample<br />
count X may be approximated by a normal distribution.<br />
Lesson 6.4 The Sampling Distribution of a Sample<br />
Proportion<br />
Let the random variable p^ be the proportion of successes in<br />
a sample size of size n, where p is the proportion of successes<br />
in the population. The sampling distribution of p^ will have<br />
p(1 − p)<br />
mean m p^ = p and standard deviation s p^ = . The<br />
Å n<br />
Large Counts condition states that the sampling distribution<br />
of p^ will have an approximately normal distribution whenever<br />
np ≥ 10 and n(1 2 p) ≥ 10. When the sampling distribution<br />
of p^ is approximately normal, probabilities involving<br />
the sample proportion p^ may be approximated by a normal<br />
distribution.<br />
Lesson 6.5 The Sampling Distribution of a<br />
Sample Mean<br />
Let the random variable x be the sample mean in a sample<br />
of size n from a population with mean m and standard<br />
deviation s. The sampling distribution of x will have mean<br />
m x = m and standard deviation s x = s . If the population<br />
"n<br />
is normal, then the sampling distribution of x will be normal.<br />
When the sampling distribution of x is exactly normal,<br />
probabilities involving the sample mean x may be calculated<br />
using a normal distribution.<br />
Lesson 6.6 The Central Limit Theorem<br />
The Central Limit Theorem states that when sampling from<br />
a non-normal population, the sampling distribution of x is<br />
approximately normal when the sample size is large. As a rule<br />
of thumb, when sampling from non-normal populations, we<br />
will consider the sampling distribution of x to be approximately<br />
normal when n ≥ 30. When the sampling distribution<br />
of x is approximately normal, probabilities involving the sample<br />
mean x may be approximated by a normal distribution.<br />
C H A P T E R 6 • Sampling Distributions 6-5<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 5<br />
11/01/17 3:51 PM
<strong>Ch</strong>apter 6 Resources<br />
<strong>SPA</strong> Applets<br />
highschool.bfwpub.com/spa<strong>3e</strong><br />
• The Normal Approximation to the Binomial applet<br />
allows students to view a binomial probability<br />
distribution with parameters n and p and a normal<br />
probability distribution with the same mean and<br />
standard deviation superimposed on the binomial.<br />
Sliders allow students to easily change the values of<br />
n and p.<br />
• The Probability applet computes probabilities for normal<br />
distributions, which are used frequently in <strong>Ch</strong>apter 6.<br />
Teacher’s Resource Materials<br />
The following resources can be found by clicking on the<br />
links in the Teacher’s e-Book (TE-book), logging into<br />
LaunchPad (password required) highschool.bfwpub.com<br />
/launchpad/spa<strong>3e</strong>, or opening the Teacher’s Resource Flash<br />
Drive (TRFD).<br />
• <strong>Ch</strong>apter Videos<br />
• <strong>Ch</strong>apter 6 Overview video (for teachers)<br />
• Lesson Overview videos for Lessons 6.1–6.3 and<br />
Lessons 6.4–6.6 (for teachers but fine to share with<br />
students, if desired)<br />
• Worked Example videos for every example<br />
(for students and teachers)<br />
• <strong>Ch</strong>apter 6 Review Exercise videos (for students and<br />
teachers)<br />
• Alternate Examples<br />
All of the <strong>Ch</strong>apter 6 Alternate Examples are provided<br />
in a Word document. Use these as additional examples<br />
in class, as the basis for assessments, or as additional<br />
practice for students.<br />
• Lesson App Handout<br />
All of the <strong>Ch</strong>apter 6 Lesson Apps are provided in PDF<br />
format. The Lesson Apps assess each learning target<br />
in the lesson. Print these for use as exit tickets or as a<br />
performance task for individuals or groups of students.<br />
Each Lesson App can also be used as formative assessment.<br />
• Teacher’s Resource Material Documents<br />
• Lesson 6.1 Activity Overview for teachers<br />
• Lesson 6.1 Activity Handout [Use with Lesson 6.1.]<br />
• Lesson 6.2 Activity Fathom file [Use with Lesson 6.2.]<br />
• <strong>Ch</strong>apter 6 Activity: Sampling Movies [Use after<br />
Lesson 6.4.]<br />
• Sampling Distributions Summary <strong>Ch</strong>art [Use with<br />
Lesson 6.5.]<br />
• <strong>Ch</strong>apter 6 Activity: Sampling Movies (The Sequel)<br />
[Use after Lesson 6.6.]<br />
• <strong>Ch</strong>apter 6 Learning Targets Grid<br />
• Lecture Presentation Slides—one prepared PowerPoint<br />
presentation per lesson (for teachers)<br />
• <strong>Ch</strong>apter Quizzes and Tests<br />
• Quiz 6A: Lessons 6.1–6.3<br />
• Quiz 6B: Lessons 6.4–6.6<br />
• <strong>Ch</strong>apter 6 Test<br />
• <strong>Ch</strong>apter 6 Answers to Quizzes and Tests<br />
• Full Solutions to Exercises—the worked solutions file for<br />
each lesson and end-of-chapter exercises and test are<br />
provided.<br />
• <strong>Ch</strong>apter Data Files<br />
• Additional <strong>Ch</strong>apter Resources<br />
We have created a list of third-party videos and other<br />
resources to support the content in this chapter. The<br />
Word document includes clickable URLs to help you<br />
access this external content. (Note: All of the URLs<br />
were live when this book was published.)<br />
6-6<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 6<br />
11/01/17 3:51 PM
Notes<br />
C H A P T E R 6 • Sampling Distributions 6-7<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 7<br />
11/01/17 3:51 PM
PD <strong>Ch</strong>apter 6 Overview<br />
Watch the chapter overview video for<br />
guidance from the authors on teaching<br />
the content in this chapter. Find it in the<br />
Teacher’s Resource Materials by clicking<br />
on the link in the TE-book, logging into<br />
the Teacher’s Resource site highschool<br />
.bfwpub.com/launchpad/spa<strong>3e</strong>,<br />
or accessing it on the TRFD.<br />
6<br />
Sampling<br />
Distributions<br />
Lesson 6.1 What Is a Sampling Distribution? 400<br />
Lesson 6.2 Sampling Distributions: Center and Variability 409<br />
Lesson 6.3 The Sampling Distribution of a Sample Count<br />
(The Normal Approximation to the Binomial) 417<br />
Lesson 6.4 The Sampling Distribution of a Sample Proportion 424<br />
Lesson 6.5 The Sampling Distribution of a Sample Mean 432<br />
Lesson 6.6 The Central Limit Theorem 439<br />
<strong>Ch</strong>apter 6 Main Points 445<br />
<strong>Ch</strong>apter 6 Review Exercises 447<br />
<strong>Ch</strong>apter 6 Practice Test 448<br />
398<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 398<br />
18/08/16 4:57 PMStarnes_<strong>3e</strong>_CH0<br />
398<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 398<br />
11/01/17 3:52 PM
Steve Gorton and Gary Ombler/Getty Images<br />
STATS applied!<br />
How can we build “greener” batteries?<br />
Kids love getting toys for their birthdays, especially electronic ones that have flashing lights<br />
and make loud noises. But these devices require lots of power and can drain batteries quickly.<br />
Battery manufacturers are constantly searching for ways to build longer-lasting batteries.<br />
When the manufacturing process is working correctly, AA batteries from a particular<br />
company should last an average of 17 hours, with a standard deviation of 0.8 hours. Also,<br />
at least 73% of the batteries should last 16.5 hours or more.<br />
Quality-control inspectors select a random sample of 50 batteries during each hour of<br />
production and then drain them under conditions that mimic normal use. The graph and<br />
summary statistics describe the distribution of the lifetimes (in hours) of the batteries from<br />
one sample of 50 AA batteries.<br />
Frequency<br />
12<br />
10<br />
8<br />
6<br />
4<br />
2<br />
15.0 15.5 16.0<br />
16.5 17.0 17.5 18.0 18.5<br />
Lifetime (h)<br />
n Mean SD min Q 1 med Q 3 max<br />
50 16.718 0.66 15.46 16.31 16.7 17.28 17.98<br />
Do these data suggest that the production process isn’t working properly? Or is it safe<br />
for plant managers to send out all the batteries produced in this hour for sale?<br />
Teaching Tip:<br />
STATS applied!<br />
The STATS applied! feature is designed<br />
to appeal to students and preview<br />
interesting questions that statistics<br />
can answer. To answer this STATS<br />
applied!, students must make inferences<br />
about a population from sample data.<br />
Knowledge of the sampling distribution<br />
of the sample proportion p^ and the<br />
sampling distribution of the sample<br />
mean x are needed to answer this<br />
question, although students won’t<br />
understand why until the end of the<br />
chapter. However, students should<br />
understand that this question is<br />
about quality control, an important<br />
statistical application for manufacturing<br />
businesses.<br />
We’ll revisit STATS applied! at the end of the chapter, so you can use what you have learned to help<br />
answer these questions.<br />
399<br />
18/08/16 4:57 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 399<br />
31/08/16 4:32 PM<br />
C H A P T E R 6 • Sampling Distributions<br />
399<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 399<br />
11/01/17 3:52 PM
PD LESSONS 6.1–6.3 Overview<br />
Watch the Lessons 6.1–6.3 overview<br />
video for guidance on teaching the<br />
content in these lessons. Find it in the<br />
Teacher’s Resource Materials by clicking<br />
on the link in the TE-book, logging into<br />
the Teacher’s Resource site, or accessing<br />
it on the TRFD.<br />
Lesson 6.1<br />
What is a Sampling<br />
Distribution?<br />
L e A r n i n g T A r g e T S<br />
d Distinguish between a parameter and a statistic.<br />
d Create a sampling distribution using all possible samples from a small<br />
population.<br />
d Use the sampling distribution of a statistic to evaluate a claim about a<br />
parameter.<br />
Learning Target Key<br />
The problems in the test bank are<br />
keyed to the learning targets using<br />
these numbers:<br />
d 6.1.1<br />
d 6.1.2<br />
d 6.1.3<br />
BELL RINGER<br />
What is the difference between random<br />
sampling and random assignment when<br />
collecting data? What inferences can be<br />
made in each case? Discuss your answers<br />
with a partner.<br />
AcT iviT y<br />
A penny for your thoughts?<br />
In this activity, your class will investigate how the<br />
mean year x and the proportion of pennies from the<br />
2000s p^ vary from sample to sample, using a large<br />
population of pennies of various ages. 1<br />
1. Have each member of the class randomly select 1<br />
penny from the population and record the year of<br />
the penny with an “X” on the dotplot provided by<br />
your teacher. Return the penny to the population.<br />
Repeat this process until at least 100 pennies have<br />
been selected and recorded. This graph gives you<br />
an idea of what the population distribution of<br />
penny years looks like.<br />
2. Have each member of the class take an SRS of 5<br />
pennies from the population and note the year on<br />
each penny.<br />
• Record the average year of these 5 pennies<br />
with an “x” on a new class dotplot. Make<br />
sure this dotplot is on the same scale as the<br />
dotplot in Step 1 above.<br />
• Record the proportion of pennies from<br />
the 2000s with a “p^ ” on a different dotplot<br />
provided by your teacher.<br />
Return the pennies to the population. Repeat<br />
this process until there are at least 100 x's and<br />
100 p^'s.<br />
3. Repeat Step 2 with SRSs of size n 5 20. Make sure<br />
these dotplots are on the same scale as the corresponding<br />
dotplots from Step 2 above.<br />
4. Compare the distribution of X (year of penny)<br />
with the two distributions of x (mean year).<br />
How are the distributions similar? How are they<br />
different? What effect does sample size seem to<br />
have on the shape, center, and variability of the<br />
distribution of x ?<br />
5. Compare the two distributions of p^ . How are the<br />
distributions similar? How are they different? What<br />
effect does sample size seem to have on the shape,<br />
center, and variability of the distribution of p^ ?<br />
Teaching Tip<br />
This activity is the most important in<br />
the whole chapter and worth the time<br />
it takes because it introduces students<br />
to sampling distributions by simulating<br />
repeated sampling from a population.<br />
We recommend spending an entire class<br />
period on this activity as the introduction<br />
to this chapter. As an alternative,<br />
consider starting it in <strong>Ch</strong>apter 4 or<br />
<strong>Ch</strong>apter 5 and do a little each day, as<br />
explained in the Activity Overview<br />
document.<br />
400<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 400<br />
Activity Overview<br />
Time: 40–50 minutes<br />
Materials: Large chart paper and markers,<br />
dot stickers, or bingo daubers to make a<br />
dotplot. Alternatively, you can make a class<br />
dotplot on the whiteboard. You will also need<br />
a population of pennies. You need a minimum<br />
of 600 pennies, but 1000 or more is ideal.<br />
Teaching Advice: See the Lesson 6.1 Activity<br />
overview and activity handout.<br />
To estimate the mean income of U.S. residents with a college degree, the Current<br />
Population Survey (CPS) selected a random sample of more than 60,000 people with<br />
at least a bachelor’s degree. The mean income in the sample was $69,609. 2 How close<br />
is this estimate to the mean income for all members of the population? To find out<br />
how an estimate varies from sample to sample, we want to gain some understanding<br />
of sampling distributions.<br />
TRM Lesson 6.1 Activity Overview<br />
for <strong>Teachers</strong><br />
TRM Lesson 6.1 Activity Handout<br />
A detailed Activity Overview document with<br />
sample graphs for teachers, as well as an<br />
Activity Handout for students, is available for<br />
this important activity. Consider giving the<br />
handout to your students so they don’t look<br />
ahead in their books for ideas and hints. You<br />
can find these resources by clicking on the<br />
link in the TE-book, logging into the Teacher’s<br />
Resource site, or accessing them on the TRFD.<br />
18/08/16 4:57 PMStarnes_<strong>3e</strong>_CH0<br />
400<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 400<br />
11/01/17 3:53 PM
L E S S O N 6.1 • What Is a Sampling Distribution? 401<br />
Parameters and Statistics<br />
For the sample of college graduates contacted by the CPS, the mean income was<br />
x 5 $69,609. The number $69,609 is a statistic because it describes this one sample.<br />
The population that the researchers want to draw conclusions about is all U.S. college<br />
graduates. In this case, the parameter of interest is the mean income m of the population<br />
of all college graduates.<br />
DEFINITION Statistic, Parameter<br />
A statistic is a number that describes some characteristic of a sample.<br />
A parameter is a number that describes some characteristic of the population.<br />
Because we can’t examine the entire population, the value of a parameter is usually<br />
unknown. To estimate the value of the parameter, we use a statistic calculated using<br />
data from a random sample of the population.<br />
Remember s and p: statistics come from samples, and parameters come from<br />
populations. The notation we use should reflect this distinction. For example, we<br />
write m (the Greek letter mu) for the population mean and x for the sample mean.<br />
The table lists some additional examples of statistics and their corresponding<br />
parameters.<br />
Teaching Tip<br />
Remind students that we have seen a<br />
“hat” before in this course. The estimated<br />
value of y is denoted y^ . Likewise, the<br />
estimated value of p is denoted p^ .<br />
TRM chapter 6 Alternate Examples<br />
You can find the Alternate Examples for<br />
this chapter in Microsoft Word format by<br />
clicking the link in the TE-book, logging<br />
into the Teacher’s Resource site, or<br />
accessing it on the TRFD.<br />
Alternate Example<br />
Lesson 6.1<br />
Sample statistic<br />
Population parameter<br />
x (the sample mean) estimates m (the population mean)<br />
p^ (the sample proportion) estimates p (the population proportion)<br />
s (the sample SD) estimates s (the population SD)<br />
How are teens different from turkeys?<br />
Parameters and statistics<br />
PROBLEM: Identify the population, the parameter,<br />
the sample, and the statistic in each of the following<br />
settings:<br />
(a) A Pew Research Center poll asked 1102 12- to<br />
17-year-olds in the United States if they have a cell<br />
phone. Of the respondents, 71% said “Yes.” 3<br />
(b) Tom is roasting a large turkey breast for a holiday<br />
meal. He wants to be sure that the turkey is<br />
safe to eat, which requires a minimum internal<br />
temperature of 165°F. Tom uses a thermometer<br />
to measure the temperature of the turkey breast<br />
at four randomly chosen points. The minimum<br />
reading he gets is 170°F.<br />
e XAMPLe<br />
SOLUTION:<br />
(a) Population: all 12- to 17-year-olds in the United<br />
States. Parameter: p 5 the proportion of all 12- to<br />
17-year-olds with cell phones. Sample: the 1102<br />
12- to 17-year-olds contacted. Statistic: the sample<br />
proportion with a cell phone, p^ 5 0.71.<br />
(b) Population: all possible locations in the turkey breast.<br />
Parameter: the true minimum temperature in all possible<br />
locations. Sample: the four randomly chosen locations.<br />
Statistic: the sample minimum, 170°F.<br />
FOR PRACTICE TRY EXERCISE 1.<br />
Pictures of coworkers?<br />
Parameters and statistics<br />
PROBLEM: Identify the population,<br />
parameter, sample, and statistic in each<br />
of the following settings:<br />
(a) A professional photographer is<br />
interested in the average number of<br />
photographs she took per day last year.<br />
She randomly selected 10 days from the<br />
year and recorded the number of photographs<br />
she took on each of the 10 days.<br />
The average number of photographs she<br />
took on those 10 days is 831.2 photos.<br />
(b) A Pew Research Center Poll asked a<br />
random sample of U.S. adults 18 or older<br />
whether they prefer to have a male<br />
coworker, a female coworker, or<br />
whether it doesn’t matter. Of the 2002<br />
respondents, 77% said it “doesn’t matter.”<br />
SOLUTION:<br />
18/08/16 4:57 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 401<br />
Teaching Tip<br />
18/08/16 4:58 PM<br />
Point out the phrase “who would say” in the<br />
solution to part (b) of the alternate example.<br />
It is not correct to say that the parameter is<br />
“the true proportion of all U.S. adults 18 or<br />
older who said it ‘doesn’t matter.’” The true<br />
proportion who said it doesn’t matter is 77%,<br />
which is the statistic (the sample proportion).<br />
Tell students to be careful about using the<br />
past tense to describe parameters!<br />
(a) Population: all days last year.<br />
Parameter: m, the true average<br />
number of photographs per day the<br />
photographer took over all days last year.<br />
Sample: the 10 randomly chosen days.<br />
Statistic: the sample mean number of<br />
photographs per day, x 5 831.2 photos.<br />
(b) Population: all U.S. adults 18 or older.<br />
Parameter: p 5 the true proportion of all<br />
U.S. adults 18 or older who would say it<br />
“doesn’t matter.”<br />
Sample: the 2002 U.S. adults 18 or older<br />
who participated in the survey.<br />
Statistic: the sample proportion who said<br />
it “doesn’t matter,” p^ 5 0.77.<br />
L E S S O N 6.1 • What Is a Sampling Distribution? 401<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 401<br />
11/01/17 3:53 PM
402<br />
C H A P T E R 6 • Sampling Distributions<br />
Common Error<br />
The phrase “sampling distribution”<br />
sounds similar to “distribution of a<br />
sample,” but they mean very different<br />
things. In the “A penny for your<br />
thoughts?” activity, the distribution of<br />
a sample is distribution of year for the<br />
5 (or 20) pennies in a student’s hand.<br />
The dotplots of the sample means and<br />
sample proportions created by the class<br />
are examples of sampling distributions.<br />
While some parameters and statistics have special symbols (such as p for the population<br />
proportion and p^ for the sample proportion), many parameters and statistics<br />
do not have their own symbol. To distinguish between a parameter and statistic, use<br />
descriptors such as “true” minimum and “sample” minimum as we did in the turkey<br />
example.<br />
Sampling Distributions<br />
In the Penny for Your Thoughts Activity, you encountered sampling variability—<br />
meaning that different random samples of the same size from the same population<br />
produce different values of a statistic. The statistics that come from these samples<br />
form a sampling distribution.<br />
DEFINITION Sampling distribution<br />
The sampling distribution of a statistic is the distribution of values taken by the statistic<br />
in all possible samples of the same size from the same population.<br />
Alternate Example<br />
Disproportionate males?<br />
Sampling distributions<br />
PROBLEM: There are six employees in<br />
a small company, Atsuko, Bernadette,<br />
Carlos, Dandre, Easton, and Freddie.<br />
Atsuko and Bernadette are female<br />
and the others are male. List all 15<br />
possible SRSs of size n 5 4, calculate the<br />
proportion of males for each sample, and<br />
display the sampling distribution of the<br />
sample proportion on a dotplot.<br />
SOLUTION:<br />
Sample 1: A, B, C, D p^ 5 0.50<br />
Sample 2: A, B, C, E p^ 5 0.50<br />
Sample 3: A, B, C, F p^ 5 0.50<br />
Sample 4: A, B, D, E p^ 5 0.50<br />
Sample 5: A, B, D, F p^ 5 0.50<br />
Sample 6: A, B, E, F p^ 5 0.50<br />
Sample 7: A, C, D, E p^ 5 0.75<br />
Sample 8: A, C, D, F p^ 5 0.75<br />
Sample 9: A, C, E, F p^ 5 0.75<br />
Sample 10: A, D, E, F p^ 5 0.75<br />
Sample 11: B, C, D, E p^ 5 0.75<br />
Sample 12: B, C, D, F p^ 5 0.75<br />
Sample 13: B, C, E, F p^ 5 0.75<br />
Sample 14: B, D, E, F p^ 5 0.75<br />
Sample 15: C, D, E, F p^ 5 1.00<br />
a<br />
e XAMPLe<br />
Just how tall are their sons?<br />
Sampling distributions<br />
Remember that a distribution describes the possible values of a variable and how<br />
often these values occur. The easiest way to picture a distribution is with a graph, such<br />
as a dotplot or histogram.<br />
PROBLEM: John and Carol have four grown sons. Their heights (in inches) are 71, 75, 72, and<br />
68. List all 6 possible SRSs of size n 5 2, calculate the mean height for each sample, and display<br />
the sampling distribution of the sample mean on a dotplot.<br />
SOLUTION:<br />
Sample 1: 71, 75 x 5 73 Sample 4: 75, 72 x 5 73.5<br />
Sample 2: 71, 72 x 5 71.5 Sample 5: 75, 68 x 5 71.5<br />
Sample 3: 71, 68 x 5 69.5 Sample 6: 72, 68 x 5 70<br />
FigUre 6.1 Dotplot<br />
showing the sampling<br />
distribution of the<br />
sample range of height<br />
for SRSs of size n 5 2.<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 402<br />
69 70 71 72 73 74<br />
Sample mean height (in.)<br />
FOR PRACTICE TRY EXERCISE 5.<br />
Every statistic has its own sampling distribution. For example, Figure 6.1 shows<br />
the sampling distribution of the sample range of height for SRSs of size n 5 2 from<br />
John and Carol’s four sons.<br />
Sample 1: 71, 75 sample range 5 4 Sample 4: 75, 72 sample range 5 3<br />
Sample 2: 71, 72 sample range 5 1 Sample 5: 75, 68 sample range 5 7<br />
Sample 3: 71, 68 sample range 5 3 Sample 6: 72, 68 sample range 5 4<br />
d d<br />
d d d d<br />
0 1 2 3 4 5 6 7 8<br />
Sample range of height (in.)<br />
18/08/16 4:58 PMStarnes_<strong>3e</strong>_CH0<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
0.5 0.6 0.7 0.8 0.9 1.0<br />
Sample proportion of men<br />
402<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 402<br />
11/01/17 3:53 PM
L E S S O N 6.1 • What Is a Sampling Distribution? 403<br />
Be specific when you use the word “distribution.” There are three different types of distributions<br />
in this setting:<br />
1. The distribution of height in the population (the four heights):<br />
d d d d<br />
67 68 69 70 71 72 73 74 75 76<br />
Height (in.)<br />
2. The distribution of height in a particular sample (two of the heights):<br />
d<br />
67 68 69 70 71 72 73 74 75 76<br />
Height (in.)<br />
3. The sampling distribution of the sample range for all possible samples (the six<br />
sample ranges):<br />
d d<br />
d d d d<br />
0 1 2 3 4 5 6 7 8<br />
Sample range of height (in.)<br />
Notice that the first two distributions consist of heights (data values), while the third<br />
distribution consists of ranges (statistics). Lesson: Always use “the distribution of __”<br />
and never just “the distribution.”<br />
d<br />
cAutIOn<br />
!<br />
Common Error<br />
Emphasize the difference in the three<br />
distributions shown here. It will be<br />
difficult, but important, for students to<br />
do this. Ask students to describe what<br />
the leftmost dot represents in each<br />
graph. In graph 1, it represents the<br />
height of a son (in the population of<br />
four sons) who is 68 inches tall. In graph<br />
2, it represents the height of a son (in<br />
a sample of two sons) who is 71 inches<br />
tall. In graph 3, it represents the sample<br />
range of heights for the sample of the<br />
two sons who are 71 and 72 inches<br />
tall. Each dot in dotplot 3 represents a<br />
statistic from a sample, not a value from<br />
a single individual.<br />
Lesson 6.1<br />
Using Sampling Distributions to Evaluate Claims<br />
Sampling distributions are the foundation for the methods of statistical inference you<br />
will learn about in <strong>Ch</strong>apters 7–10. Knowing the sampling distribution of a statistic<br />
will help us know how much the statistic tends to vary from its corresponding parameter<br />
and what values of the statistic should be considered unusual.<br />
How long will we bead doing the homework?<br />
Evaluating a claim<br />
PROBLEM: At the beginning of class, Mrs. <strong>Ch</strong>auvet shows her<br />
class a box filled with black and white beads. She claims that<br />
the proportion of black beads in the box is p 5 0.50. To determine<br />
the number of homework exercises she will assign that<br />
evening, she invites a student to select an SRS of n 5 30 beads<br />
from the box. The number of black beads selected will be the<br />
number of homework exercises assigned. When the student<br />
selects 19 black beads (p^ 5 19/30 5 0.63), the students groan<br />
and suggest that Mrs. <strong>Ch</strong>auvet included more than 50% black<br />
beads in the box.<br />
To determine if a sample proportion of p^ 5 0.63 provides convincing evidence that Mrs. <strong>Ch</strong>auvet<br />
cheated, the class simulated 100 SRSs of size n 5 30, assuming that she was telling the truth. That is,<br />
they sampled from a population with 50% black beads. For each sample, they recorded the sample<br />
proportion of black beads. The results of the simulation are shown on the next page.<br />
e XAMPLe<br />
© Monalyn Gracia/Corbis<br />
18/08/16 4:58 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 403<br />
Alternate Example<br />
What’s in the box?<br />
Evaluating a claim<br />
PROBLEM: At the end of class, Mr. Osters<br />
allows one student to select a ticket from<br />
a shoebox without looking. The tickets are<br />
labeled either “Homework pass” or “Try<br />
again.” Once a ticket is drawn, it is replaced<br />
for the next drawing and the tickets are<br />
mixed thoroughly. Mr. Osters claims that<br />
the proportion of homework passes in<br />
the shoebox is p 5 0.25. At the end of the<br />
first quarter, one student noted that only<br />
6 students won in 50 drawings ( p^ 5 0.12).<br />
The students were suspicious that less than<br />
25% of the tickets in the box are homework<br />
passes.<br />
18/08/16 4:58 PM<br />
To determine if a sample proportion of<br />
p^ 5 0.12 provides convincing evidence that<br />
the true proportion of homework passes is<br />
less than 25%, the class simulated 100 SRSs of<br />
size n 5 50, assuming that 25% of the tickets<br />
were homework passes. For each sample, they<br />
recorded the sample proportion of homework<br />
passes. Here are the results of the simulation:<br />
d d dddddd d<br />
d d d d d d<br />
d d d d d d<br />
d<br />
d d d d d d d<br />
d d d d d d d<br />
d d d d d d d<br />
d d d d d d d d<br />
d d d d d d d d d d<br />
d d d d d d d d d d d<br />
d d d d d d d d d d d<br />
d<br />
d d d d d d d d d d d d<br />
d d d d<br />
0.10 0.15 0.20 0.25 0.30 0.35 0.40<br />
Sample proportion of<br />
homework passes<br />
(a) There is one dot on the graph at p^ 5 0.38<br />
Explain what this dot represents.<br />
(b) Would it be unusual to get a sample<br />
proportion of 0.12 or less in a sample of<br />
size 50 when p 5 0.25? Explain.<br />
(c) Based on your answer to part (b), is<br />
there convincing evidence that Mr. Osters<br />
lied about the contents of the shoebox?<br />
SOLUTION:<br />
(a) In one SRS of size n 5 50, 38% of the<br />
tickets were homework passes.<br />
(b) Yes; in the 100 trials of the<br />
simulation, only 2 of the SRSs included<br />
12% or fewer homework passes.<br />
(c) Yes; because the probability from part<br />
(b) is small—only 0.02—it is not plausible<br />
that the proportion of homework passes in<br />
the shoebox is p 5 0.25 and the students<br />
got a sample proportion of p^ 5 0.12 by<br />
chance alone.<br />
L E S S O N 6.1 • What Is a Sampling Distribution? 403<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 403<br />
11/01/17 3:53 PM
404<br />
C H A P T E R 6 • Sampling Distributions<br />
FYI<br />
Sampling distributions of statistics from<br />
most real populations are extremely<br />
large and therefore difficult to imagine.<br />
For example, if we were to sample 6 U.S.<br />
senators from the population of 100<br />
current senators, there are 1,192,052,400<br />
possible samples. In this case, a sampling<br />
distribution would have 1,192,052,400<br />
different dots in its dotplot! When<br />
sampling distributions are very large,<br />
we use simulations to create a good<br />
approximation.<br />
SOLUTION:<br />
(a)<br />
(b)<br />
(c)<br />
(a) There is one dot on the graph at p^ 5 0.77. Explain<br />
what this dot represents.<br />
(b) Would it be unusual to get a sample proportion<br />
of 0.63 or higher in a sample of size 30 when<br />
d d dddddddddd d ddddddd d dddddddddddd p 5 0.50? Explain.<br />
d<br />
d<br />
d<br />
d<br />
d d d d d d d d d<br />
d d d d d d d d d<br />
d d d d d d d d d<br />
d d d d d d d d d d<br />
d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d<br />
0.2 0.3 0.4 0.5 0.6 0.7 0.8<br />
Sample proportion of black beads<br />
In one simulated SRS of size n 5 30, 77% of the beads<br />
were black.<br />
No. In the 100 simulated samples, 9 of the SRSs included<br />
at least 63% black beads.<br />
No. Because the probability from part (b) isn’t that small,<br />
it is plausible that the proportion of black beads in the box<br />
is p 5 0.50 and the student got a sample proportion of<br />
(c) Based on your answer to part (b), is there convincing<br />
evidence that Mrs. <strong>Ch</strong>auvet lied about the<br />
contents of the box?<br />
Notice that 9 of the<br />
100 simulated SRSs<br />
d d dd resulted in a sample<br />
d d<br />
d d proportion of 0.63<br />
d d d<br />
d d d<br />
or higher.<br />
d d d<br />
d d d<br />
d d d d<br />
d d d d<br />
d d d d<br />
d d d d d<br />
d d d d d d d d d<br />
d d d d d d d d d<br />
d d d d d d d d d<br />
d d d d d d d d d d<br />
d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d<br />
p^ 5 0.63 by chance alone. 0.2 0.3 0.4 0.5 0.6 0.7 0.8<br />
Sample proportion of black beads<br />
FOR PRACTICE TRY EXERCISE 9.<br />
We used 100 simulated samples to produce the dotplot of sample proportions in<br />
this example. Because it doesn’t include all possible samples of size 30, it is only an<br />
approximation of the actual sampling distribution of p^ . Thankfully, the simulated<br />
sampling distribution should be a good approximation as long as we use a large number<br />
of samples in the simulation.<br />
L e SSon APP 6. 1<br />
How cold is it inside the cabin?<br />
During the winter months, outside temperatures at the Starneses' cabin in<br />
Colorado can stay well below freezing (32°F, or 0°C) for weeks at a time. To<br />
prevent the pipes from freezing, Mrs. Starnes sets the thermostat at 50°F. The<br />
manufacturer claims that the thermostat allows variation in home temperature<br />
that follows a normal distribution with s 5 3°F. To test this claim, Mrs. Starnes<br />
programs her digital thermostat to take an SRS of n 5 10 readings during a<br />
24-hour period. The standard deviation of the results is s x 5 5°F.<br />
Quasarphoto/Getty Images<br />
TRM <strong>Ch</strong>apter 6 Lesson App Handout<br />
All of the <strong>Ch</strong>apter 6 Lesson Apps can<br />
be found by clicking on the link in the<br />
TE-book, logging into the Teacher’s<br />
Resource site, or accessing this resource<br />
on the TRFD. The Lesson Apps assess all<br />
learning targets in the lesson, so they<br />
are excellent resources to gauge student<br />
understanding. Use them as a formative<br />
evaluation at the end of each lesson to<br />
help you and your students understand<br />
exactly which learning targets are<br />
challenging and which are not.<br />
1. Identify the population, the parameter, the sample, and the statistic in this context.<br />
Suppose the thermostat is working properly and that the temperatures in the cabin vary according to a normal<br />
distribution with mean m 5 50°F and standard deviation s 5 3°F. The dotplot shows the distribution of the sample<br />
standard deviation in 100 simulated SRSs of size n 5 10 from this distribution.<br />
Lesson App<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 404<br />
Answers<br />
1. Population: All possible times during<br />
the 24-hour period. Parameter: s 5 the<br />
true standard deviation of temperature<br />
readings at all possible times during the<br />
24-hour period. Sample: The SRS of 10 times.<br />
Statistic: The sample standard deviation of<br />
temperature readings, S x 5 5°F.<br />
2. Yes; in the 100 simulated samples, 0 of<br />
the SRSs had a sample standard deviation<br />
of 5°F or higher. Based on the simulation,<br />
P(s x > 5) = 0∙100 = 0.<br />
3. Yes; because the probability from Question<br />
2 is small, it is not plausible that the true<br />
standard deviation is s 5 3°F and<br />
Mrs. Starnes got the sample standard<br />
deviation of S x 5 5°F by chance alone.<br />
18/08/16 4:59 PMStarnes_<strong>3e</strong>_CH0<br />
404<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 404<br />
11/01/17 3:53 PM
L E S S O N 6.1 • What Is a Sampling Distribution? 405<br />
18/08/16 4:59 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 405<br />
d<br />
2. Would it be unusual to get a sample standard<br />
d d ddd deviation of s x 5 5°F or higher in a sample of size<br />
n 5 10 when s 5 3°F? Explain.<br />
d dddd d d<br />
dd dddd ddddd d d d<br />
d dd ddddddddddddd dd d<br />
d dd ddddddddddddddddd dd<br />
ddddddddddddddddddddddddddddddddd<br />
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0<br />
Sample standard deviation of temperature (°F)<br />
Lesson 6.1<br />
3. Based on your answer to Question 2, is there convincing<br />
evidence that the thermometer is more<br />
variable than the manufacturer claims? Explain.<br />
WhAT DiD y o U LeA rn?<br />
LEARNINg TARgET EXAMPLES EXERCISES<br />
Distinguish between a parameter and a statistic. p. 401 1–4<br />
Create a sampling distribution using all possible samples from a small<br />
population.<br />
Use the sampling distribution of a statistic to evaluate a claim about a<br />
parameter.<br />
Exercises<br />
Mastering Concepts and Skills<br />
For Exercises 1–4, identify the population, the parameter,<br />
the sample, and the statistic in each setting.<br />
1. Smoking and height<br />
(a) From a large group of people who signed a card<br />
saying they intended to quit smoking, a random<br />
sample of 1000 people was selected. It turned out<br />
that 210 (21%) of the sampled individuals had not<br />
smoked over the past 6 months.<br />
(b) A pediatrician wants to know the 75th percentile<br />
for the distribution of heights of 10-year-old boys,<br />
so she selects a sample of 50 10-year-old male<br />
patients and calculates that the 75th percentile in<br />
the sample is 56 inches.<br />
2. Unemployment and gas prices<br />
(a) Each month, the Current Population Survey<br />
interviews a random sample of individuals in<br />
about 60,000 U.S. households. One of its goals<br />
is to estimate the national unemployment rate. In<br />
January 2015, 5.7% of those interviewed were<br />
unemployed.<br />
(b) How much do gasoline prices vary in a large<br />
city? To find out, a reporter records the price per<br />
gallon of regular unleaded gasoline at a random<br />
sample of 10 gas stations in the city on the same day.<br />
pg 401<br />
Lesson 6.1<br />
p. 402 5–8<br />
p. 403 9–12<br />
The range (Maximum – Minimum) of the prices in<br />
the sample is 25 cents.<br />
3. Tea and screening<br />
(a) On Tuesday, the bottles of iced tea filled in a plant were<br />
supposed to contain an average of 20 ounces of iced<br />
tea. Quality-control inspectors sampled 50 bottles at<br />
random from the day’s production. These bottles contained<br />
an average of 19.6 ounces of iced tea.<br />
(b) On a New York–Denver flight, 8% of the 125 passengers<br />
were selected for random security screening<br />
before boarding. According to the Transportation<br />
Security Administration, 10% of passengers at this airport<br />
are supposed to be chosen for random screening.<br />
4. Bearings and thermostats<br />
(a) A production run of ball bearings is supposed to<br />
have a mean diameter of 2.5000 centimeters. An<br />
inspector chooses a random sample of 100 bearings<br />
from the container and calculates a mean diameter<br />
of 2.5009 centimeters.<br />
(b) During the winter months, Mrs. Starnes sets the thermostat<br />
at 50°F to prevent the pipes from freezing in<br />
her cabin. She wants to know how low the interior<br />
temperature gets. A digital thermometer records the<br />
indoor temperature at 20 randomly chosen times<br />
during a given day. The minimum reading is 38°F.<br />
Teaching Tip<br />
18/08/16 4:59 PM<br />
In part (a) of Exercise 3, the value of the<br />
parameter is not necessarily 20. It is a target<br />
the company is hoping to achieve, but it may<br />
not be the actual average number of ounces<br />
in the bottles. Likewise, in part (b), 10% may<br />
not be the true proportion of all passengers<br />
selected for a security screening, so it is not<br />
necessarily the actual parameter value.<br />
TRM Full Solutions to Lesson 6.1<br />
Exercises<br />
You can find the full solutions for this<br />
lesson by clicking on the link in the<br />
TE-book, logging into the Teacher’s<br />
Resource site, or accessing this resource<br />
on the TRFD.<br />
Answers to Lesson 6.1 Exercises<br />
1. (a) Population: All people who<br />
signed a card saying that they intend to<br />
quit smoking. Parameter: p 5 the true<br />
proportion of the population who quit<br />
smoking. Sample: A random sample<br />
of 1000 people who signed the cards.<br />
Statistic: The proportion of the sample<br />
who quit smoking; p^ 5 0.21.<br />
(b) Population: All 10-year-old boys.<br />
Parameter: The true 75th percentile of all<br />
10-year-old boys. Sample: Sample of 50<br />
patients. Statistic: The 75th percentile of<br />
the sample, 56 inches.<br />
2. (a) Population: Individuals in<br />
U.S. households. Parameter: p 5 true<br />
proportion of the U.S. population who<br />
are unemployed. Sample: A random<br />
sample of individuals from 60,000 U.S.<br />
households. Statistic: The proportion of<br />
the sample who were unemployed;<br />
p^ 5 0.057.<br />
(b) Population: All gasoline stations in<br />
a large city. Parameter: True range of<br />
gas prices at all gasoline stations in the<br />
city. Sample: A random sample of 10 gas<br />
stations in the city. Statistic: The range<br />
of prices in the sample; sample range 5<br />
25 cents.<br />
3. (a) Population: All bottles of iced tea<br />
filled in a plant on Tuesday. Parameter:<br />
m 5 the true mean amount of tea in the<br />
population. Sample: A random sample of<br />
50 bottles. Statistic: The mean amount of<br />
tea in the sample; x 5 19.6 ounces.<br />
(b) Population: All passengers in the<br />
airport. Parameter: p 5 the true proportion<br />
of the population who are chosen for<br />
random screening. Sample: The 125<br />
passengers on a New York-to-Denver flight.<br />
Statistic: The proportion of the sample<br />
selected for security screening; p^ 5 0.08.<br />
4. (a) Population: All ball bearings in<br />
the production run. Parameter: m 5 the<br />
true mean diameter in the population.<br />
Sample: A random sample of 100<br />
bearings. Statistic: The mean diameter in<br />
the sample; x 5 2.5009 cm.<br />
(b) Population: All possible times during<br />
the given day. Parameter: The true minimum<br />
temperature during the 24-hour period.<br />
Sample: The 20 randomly chosen times<br />
during the day. Statistic: The minimum<br />
temperature in the sample 5 38°F.<br />
Lesson 6.1<br />
L E S S O N 6.1 • What Is a Sampling Distribution? 405<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 405<br />
11/01/17 3:53 PM
406<br />
C H A P T E R 6 • Sampling Distributions<br />
Exercises 5–8 refer to the following population of 2<br />
Teaching Tip<br />
male students and 3 female students, along with their<br />
quiz scores:<br />
Exercises 5–8 are very important because<br />
Abigail 10 Bobby 5 Carlos 10 DeAnna 7 Emily 9<br />
students can create (and see) an entire<br />
5. Sample means List all 10 possible SRSs of size<br />
sampling distribution. Don’t skip these<br />
pg 402 n 5 2, calculate the mean quiz score for each sample,<br />
and display the sampling distribution of the sample<br />
exercises!<br />
mean on a dotplot.<br />
6. Sample ranges List all 10 possible SRSs of size<br />
5.<br />
n 5 3, calculate the range of quiz scores for each<br />
sample, and display the sampling distribution of<br />
Sample #1: Abigail (10), Bobby (5) x 5 7.5<br />
the sample range on a dotplot.<br />
Sample #2: Abigail (10), Carlos (10) x 5 10<br />
7. Sample proportions List all 10 possible SRSs of size<br />
n 5 2, calculate the proportion of females for each<br />
Sample #3: Abigail (10), DeAnna (7) x 5 8.5<br />
sample, and display the sampling distribution of<br />
Sample #4: Abigail (10), Emily (9) x 5 9.5<br />
the sample proportion on a dotplot.<br />
8. Sample medians List all 10 possible SRSs of size<br />
Sample #5: Bobby (5), Carlos (10) x 5 7.5<br />
n 5 3, calculate the median quiz score for each<br />
Sample #6: Bobby (5), DeAnna (7) x 5 6<br />
sample, and display the sampling distribution of<br />
the sample median on a dotplot.<br />
Sample #7: Bobby (5), Emily (9) x 5 7<br />
9. Who does their homework? A school newspaper<br />
Sample #8: Carlos (10), DeAnna (7) x 5 8.5<br />
pg 403 article claims that 60% of the students at a large<br />
high school completed their assigned homework<br />
Sample #9: Carlos (10), Emily (9) x 5 9.5<br />
last week. Some statistics students want to investigate<br />
if this claim is true, so they choose an SRS of<br />
Sample #10: DeAnna (7), Emily (9) x 5 8<br />
100 students from the school to interview. When<br />
d d d<br />
they found that only 45 of the 100 students completed<br />
their assigned homework last week, they<br />
d d d d d d d<br />
6 6.5 7 7.5 8 8.5 9 9.5 10<br />
suspected that the proportion of all students who<br />
Sample mean quiz score<br />
completed their assigned homework last week is<br />
less than the 60% claimed by the newspaper.<br />
6.<br />
To determine if a sample proportion of p^ 5 0.45<br />
provides convincing evidence that the true proportion<br />
is less than p 5 0.60, the class simulated 250<br />
Sample #1: Abigail (10), range 5 5<br />
Bobby (5), Carlos (10)<br />
SRSs of size n 5 100 from a population in which<br />
p 5 0.60. Here are the results of the simulation.<br />
Sample #2: Abigail (10), range 5 5<br />
d<br />
d<br />
Bobby (5), DeAnna (7)<br />
d<br />
Sample #3: Abigail (10), range 5 5<br />
d d d d<br />
d d<br />
d d d<br />
Bobby (5), Emily (9)<br />
dddd d d d d<br />
d d d d<br />
ddd<br />
d d d d d<br />
d d d d d d<br />
d d d d d d d d<br />
d d d d d d d d<br />
d d d<br />
Sample #4: Abigail (10), range 5 3<br />
d d d<br />
d d d<br />
Carlos (10), DeAnna (7)<br />
d d d<br />
d d d dddd d d d d d<br />
d d d<br />
d d<br />
d<br />
d d d d<br />
d d d d d d<br />
d d d d d d d<br />
d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d<br />
Sample #5: Abigail (10), range 5 1<br />
d d d d d d d d d d<br />
d d d d d d d d d<br />
d d d d d d d d d<br />
Carlos (10), Emily (9)<br />
dd dddddddddd dddddddddddddddd d dddddd d ddddddd<br />
d d d d d<br />
d d d d d d d d<br />
d d d d d d d d<br />
d dddddd d d dd d d d<br />
Sample #6: Abigail (10), range 5 3<br />
0.45 0.50 0.55 0.60 0.65 0.70 0.75<br />
DeAnna (7), Emily (9)<br />
Sample proportion of students who<br />
completed homework<br />
Sample #7: Bobby (5), range 5 5<br />
Carlos (10), DeAnna (7)<br />
(a) There is one dot on the graph at 0.73. Explain what<br />
this dot represents.<br />
Sample #8: Bobby (5), range 5 5<br />
(b) Would it be surprising to get a sample proportion<br />
of 0.45 or less in an SRS of size 100 when<br />
Carlos (10), Emily (9)<br />
Sample #9: Bobby (5), range 5 4<br />
p 5 0.60? Explain.<br />
DeAnna (7), Emily (9)<br />
Sample #10: Carlos (10), range 5 3<br />
DeAnna (7), Emily (9)<br />
d<br />
d dd<br />
d d d d<br />
1 1.5 2 2.5 3 3.5 4 4.5 5<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 406<br />
Sample range of quiz score<br />
8.<br />
Sample #1: Abigail (10), median 5 10<br />
7.<br />
Bobby (5), Carlos (10)<br />
Sample #1: Abigail, Bobby p^ 5 0.50 Sample #2: Abigail (10), median 5 7<br />
Sample #2: Abigail, Carlos<br />
Bobby (5), DeAnna (7)<br />
p^ 5 0.50<br />
Sample #3: Abigail (10), median 5 9<br />
Sample #3: Abigail, DeAnna p^ 5 1<br />
Bobby (5), Emily (9)<br />
Sample #4: Abigail, Emily p^ 5 1<br />
Sample #4: Abigail (10), median 5 10<br />
Sample #5: Bobby, Carlos p^ 5 0<br />
Carlos (10), DeAnna (7)<br />
Sample #5: Abigail (10), median 5 10<br />
Sample #6: Bobby, DeAnna p^ 5 0.50 Carlos (10), Emily (9)<br />
Sample #7: Bobby, Emily p^ 5 0.50 Sample #6: Abigail (10), median 5 9<br />
Sample #8: Carlos, DeAnna p^ 5 0.50 DeAnna (7), Emily (9)<br />
Sample #9: Carlos, Emily Sample #7: Bobby (5), median 5 7<br />
p^ 5 0.50<br />
Carlos (10), DeAnna (7)<br />
Sample #10: DeAnna, Emily p^ 5 1<br />
Sample #8: Bobby (5), median 5 9<br />
Carlos (10), Emily (9)<br />
Sample #9: Bobby (5), median 5 7<br />
d d ddddd d dd DeAnna (7), Emily (9)<br />
0 0.5 1<br />
Sample #10: Carlos (10), median 5 9<br />
Sample proportion of females<br />
DeAnna (7), Emily (9)<br />
(c) Based on your answer to part (b), is there convincing<br />
evidence that the proportion of all students<br />
who completed their assigned homework last week<br />
is less than p 5 0.60? Explain.<br />
10. First-serve percentage One important aspect of<br />
a tennis player’s effectiveness is her first-serve<br />
percentage—the proportion of the time the first of<br />
her two attempts to serve the ball to her opponent<br />
is successful. For her first three years on the<br />
tennis team, Shruti’s first-serve percentage is 53%.<br />
Hoping to improve, Shruti works over the summer<br />
with a coach who specializes in serves. In her<br />
first match of the next season, Shruti’s first serve<br />
is successful 42 times in 60 attempts, a first-serve<br />
percentage of 70%.<br />
Suppose we treat Shruti’s first 60 attempts<br />
as an SRS of her serves after working with the<br />
new coach. To determine if a sample proportion<br />
of p^ 5 0.70 provides convincing evidence that<br />
the true proportion is greater than p 5 0.53, we<br />
simulate 200 SRSs of size n 5 60 from a population<br />
in which p 5 0.53. Here are the results of<br />
the simulation.<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d d<br />
d d<br />
d d d d<br />
d d d d<br />
d d d d<br />
d ddddddddd d d d d<br />
d d d d<br />
d d d d<br />
d d<br />
dd<br />
d d d d<br />
d d d d d d d<br />
d d d d d d d d<br />
d<br />
d d ddd d d d d d d d d<br />
d d d d d d d d d<br />
d d<br />
dd d d d d d d d d d d ddddddddddddddddd<br />
0.37 0.40 0.43 0.47 0.50 0.53 0.57 0.60 0.63 0.67 0.70<br />
Sample proportion of successful first serves<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
7 7.5 8 8.5 9 9.5 10<br />
Sample median of quiz scores<br />
9. (a) In one SRS of size n 5 100, 73% of the<br />
students did all their assigned homework.<br />
(b) Yes; in the 250 simulated samples, 0<br />
of the SRSs had a sample proportion of<br />
0.45 or lower. Based on the simulation,<br />
P( p^ ≤ 0.45) = 0∙250 = 0.<br />
(c) Yes; because the probability from part<br />
(b) is small, it is not plausible that the true<br />
proportion is p 5 0.60 and the statistics<br />
students got a sample proportion of p^ 5 0.45<br />
by chance alone.<br />
d d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d d d<br />
d d d d d<br />
(a) There is one dot on the graph at 0.37. Explain what<br />
this dot represents.<br />
(b) Would it be surprising to get a sample proportion<br />
of 0.70 or more in an SRS of size 60 when<br />
p 5 0.53? Explain.<br />
(c) Based on your answer to part (b), is there convincing<br />
evidence that Shruti’s first-serve percentage has<br />
improved since working with the new coach? Explain.<br />
11. Are we taller? According to the National Center<br />
for Health Statistics, the distribution of heights for<br />
16-year-old females is modeled well by a normal<br />
distribution with mean m 5 64 inches and standard<br />
deviation s 5 2.5 inches. To see if this distribution<br />
applies at their high school, a statistics class takes<br />
an SRS of 20 of the 300 16-year-old females at the<br />
school and measures their heights. When they calculate<br />
a sample mean of 64.7 inches, they wonder<br />
if the population of 16-year-old girls at their school<br />
has a mean height greater than 64 inches.<br />
To determine if a sample mean of x 5 64.7 inches<br />
provides convincing evidence that the average<br />
height of 16-year-old girls at the school is taller<br />
Answers 10–11 are on page 407<br />
18/08/16 4:59 PMStarnes_<strong>3e</strong>_CH0<br />
406<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 406<br />
11/01/17 3:53 PM
L E S S O N 6.1 • What Is a Sampling Distribution? 407<br />
than 64 inches, the class simulated 200 SRSs of size<br />
n 5 20 from a normal population with mean<br />
m 5 64 inches and standard deviation s 5 2.5<br />
inches. Here are the results of the simulation.<br />
d<br />
d<br />
d<br />
d d<br />
d d<br />
d d<br />
d d d d<br />
d d d d<br />
d d d d<br />
d d d d<br />
d d d d d<br />
d d d d d d<br />
d d d d d d d d d<br />
d d d d d d d d d d<br />
d d d d d d d d d d d d<br />
d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d d d d d d d d dddd<br />
62.5 63.0 63.5 64.0 64.5 65.0 65.5 66.0<br />
Sample mean height (in.)<br />
(a) There is one dot on the graph at 62.5. Explain what<br />
this dot represents.<br />
(b) Would it be unusual to get a sample mean of 64.7<br />
or more in a sample of size 20 when m 5 64?<br />
Explain.<br />
(c) Based on your answer to part (b), is there convincing<br />
evidence that the mean height of the population<br />
of 16-year-old girls at this school is greater than 64<br />
inches? Explain.<br />
12. Relying on bathroom scales A manufacturer of<br />
bathroom scales says that when a 150-pound<br />
weight is placed on a scale produced in the factory,<br />
the weight indicated by the scale is normally<br />
distributed with a mean of 150 pounds and a<br />
standard deviation of 2 pounds. A consumeradvocacy<br />
group acquires an SRS of 12 scales from<br />
the manufacturer and places a 150-pound weight<br />
on each one. The group gets a mean weight of<br />
149.1 pounds, which makes them suspect that<br />
the scales underestimate the true weight. To test<br />
this, they use a computer to simulate 200 samples<br />
of 12 scales from a population with a mean of<br />
150 pounds and standard deviation of 2 pounds.<br />
Here is a dotplot of the means from these 200<br />
samples.<br />
d<br />
d<br />
d d<br />
d<br />
d<br />
d<br />
d<br />
d d<br />
d d d<br />
d d d d<br />
d d d d d d d<br />
d d d d d d d d<br />
d d d d d d d d d<br />
d d d d d d d d d d<br />
d d d d d d d d d d d d<br />
d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d d d d d d d d dd d d<br />
148.0 148.5 149.0 149.5 150.0 150.5 151.0 151.5<br />
Sample mean weight (lb)<br />
(a) There is one dot on the graph at 151.2. Explain<br />
what this dot represents.<br />
d<br />
(b) Would it be unusual to get a sample mean of 149.1 or<br />
less in a sample of size 12 when m 5 150? Explain.<br />
(c) Based on your answer to part (b), is there convincing<br />
evidence that the scales produced by this manufacturer<br />
underestimate true weight? Explain.<br />
Applying the Concepts<br />
13. Instant winners A fast-food restaurant promotes<br />
certain food items by giving a game piece with<br />
each item. Advertisements proclaim that “25% of<br />
the game pieces are Instant Winners!” To test this<br />
claim, a frequent diner collects 20 game pieces and<br />
gets only 3 instant winners.<br />
(a) Identify the population, the parameter, the sample,<br />
and the statistic in this context.<br />
Suppose the advertisements are correct and<br />
p 5 0.25. The dotplot shows the distribution of the<br />
sample proportion of instant winners in 100 simulated<br />
SRSs of size n 5 20.<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6<br />
Sample proportion of instant winners<br />
(b) Would it be unusual to get a sample proportion of<br />
p^ 5 3/20 5 0.15 or less in a sample of size n 5 20<br />
when p 5 0.25? Explain.<br />
(c) Based on your answer to part (b), is there convincing<br />
evidence that fewer than 25% of all game<br />
pieces are instant winners? Explain.<br />
14. Puny guppies? A large pet store that specializes<br />
in tropical fish has several thousand guppies. The<br />
store claims that the lengths of its guppies are<br />
approximately normally distributed with a mean of<br />
5 centimeters and a standard deviation of 0.5 centimeter.<br />
You come to the store and buy 10 randomly<br />
selected guppies and find that the mean length of<br />
your 10 guppies is only 4.8 centimeters.<br />
(a) Identify the population, the parameter, the sample,<br />
and the statistic in this context.<br />
Suppose the store’s description of the lengths of<br />
its guppies is true. The dotplot on the next page<br />
shows the distribution of sample means from 200<br />
simulated SRSs of size n 5 10 from a normally distributed<br />
population with m 5 5 centimeters and<br />
s 5 0.5 centimeter.<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
11. (a) In one SRS of size n 5 20, the<br />
mean height was 62.5 inches.<br />
(b) No, in the 200 simulated samples,<br />
23 of the SRSs had a mean of 64.7<br />
or more. Based on the simulation,<br />
P( x ≥ 64.7) = 23∙200 = 0.115.<br />
(c) No; because the probability from<br />
part (b) isn’t small, it is plausible that the<br />
true mean is m 5 64 and the class got<br />
a sample mean of x 5 64.7 by chance<br />
alone.<br />
12. (a) In one SRS of size n 5 12, the<br />
mean weight was 151.2 pounds.<br />
(b) No; in the 200 simulated samples,<br />
22 of the SRSs had a mean of<br />
149.1 or less. Based on the simulation,<br />
P( x ≤ 149.1) = 22∙200 = 0.11.<br />
(c) No; because the probability from<br />
part (b) isn’t small, it is plausible that the<br />
true mean is m 5 150 and the group got<br />
a sample mean of x 5 149.1 by chance<br />
alone.<br />
13. (a) Population: All game pieces.<br />
Parameter: p 5 the true proportion of<br />
the population that are instant winners.<br />
Sample: The 20 game pieces collected by<br />
the frequent diner. Statistic: The proportion<br />
of the sample that are instant winners;<br />
p^ 5 3/20 5 0.15.<br />
(b) No; in the 100 simulated samples,<br />
18 of the SRSs had a sample proportion<br />
of 0.15 or lower. Based on the simulation,<br />
P( p^ ≤ 0.15) = 18∙100 = 0.18.<br />
(c) No; because the probability from<br />
part (b) isn’t small, it is plausible that<br />
the true proportion is p 5 0.25 and the<br />
frequent diner got a sample proportion<br />
of p^ 5 0.15 by chance alone.<br />
Lesson 6.1<br />
18/08/16 4:59 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 407<br />
Answers continued<br />
10. (a) In one SRS of size n 5 60, 36.7% of<br />
the first serves were successful.<br />
(b) Yes; in the 200 simulated samples, only<br />
1 of the SRSs had a sample proportion of<br />
0.70 or higher. Based on the simulation,<br />
P( p^ ≥ 0.70) = 1∙200 = 0.005.<br />
(c) Yes; because the probability from part<br />
(b) is small, it is not plausible that the true<br />
proportion is still p 5 0.53 and the player got<br />
a sample proportion of p^ 5 0.70 by chance<br />
alone.<br />
18/08/16 4:59 PM<br />
L E S S O N 6.1 • What Is a Sampling Distribution? 407<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 407<br />
11/01/17 3:53 PM
408<br />
C H A P T E R 6 • Sampling Distributions<br />
14. (a) Population: All guppies at the<br />
pet store. Parameter: m 5 the true mean<br />
length of the population. Sample: A<br />
random sample of 10 guppies. Statistic: The<br />
mean length of the sample; x 5 4.8 cm.<br />
(b) No; in the 200 simulated<br />
samples, 21 of the SRSs had a mean<br />
of 4.8 or less. Based on the simulation,<br />
P( x ≤ 4.8) = 21∙200 = 0.105.<br />
(c) No; because the probability from<br />
part (b) isn’t small, it is plausible that the<br />
true mean is m 5 5 and I got a sample<br />
mean of x 5 4.8 by chance alone.<br />
15. (a) The distribution of heights for<br />
16-year-old females is approximately<br />
normal with a mean of m 5 64 inches<br />
and standard deviation of s 5 2.5 inches.<br />
56.5 59 61.5 64 66.5<br />
Height (in.)<br />
69 71.5<br />
(b) Answers will vary. This is the distribution<br />
of one possible sample.<br />
d d<br />
d ddd<br />
d d ddd ddddddd d d<br />
55 60 65 70<br />
Height (in.)<br />
16. (a) The distribution of measured<br />
weights for all scales is approximately<br />
normal with a mean of m 5 150 pounds<br />
and standard deviation of s 5 2 pounds.<br />
144 146 148 150 152<br />
Weight (lb)<br />
154 156<br />
(b) Answers will vary. This is the<br />
distribution of one possible sample.<br />
d d d d d dd dd d d d<br />
146 147 148 149 150 151 152 153 154 155<br />
Weight (lb)<br />
17. (a) In 10 cases of taking a random<br />
sample of size n 5 50 from each high<br />
school, the difference in proportions of<br />
students with Internet access at home<br />
is 0%. This means the proportion of<br />
students with Internet access was the<br />
same for each high school in 10 pairs of<br />
simulated samples<br />
(b) Yes; in the 100 pairs of simulated<br />
samples, 0 of the pairs had a<br />
difference in proportions of 0.20<br />
or higher. Based on the simulation,<br />
P(p^ N − p^ S ≥ 0.20) = 0∙100 = 0.<br />
(c) Yes; because the probability from<br />
part (b) is small, it is not plausible that<br />
the true difference in proportions is<br />
d<br />
d<br />
d<br />
d d<br />
d d<br />
d d d<br />
d d d<br />
d d d d<br />
d d d d d d d<br />
d d d d d d d d d d<br />
d d d d d d d d d d d d<br />
d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d d d d d d d d d d d d<br />
d d d d d d d d d d d d d d d d d d d d d d d d d d d d ddd<br />
d dd d<br />
4.6 4.8 5.0 5.2 5.4<br />
Sample mean length (cm)<br />
(b) Would it be unusual to get a sample mean of x = 4.8<br />
centimeters or less in a sample of size n 5 10 from<br />
this population? Explain.<br />
(c) Based on your answer to part (b), is there convincing<br />
evidence that the mean length of guppies at this<br />
store is less than 5 centimeters? Explain.<br />
15. More tall girls Refer to Exercise 11.<br />
(a) Make a graph of the population distribution of<br />
heights for 16-year-old females.<br />
(b) Sketch a possible dotplot of the distribution of sample<br />
data for an SRS of size 20 from this population.<br />
16. More bathroom scales Refer to Exercise 12.<br />
(a) Make a graph of the population distribution of<br />
weights, assuming the manufacturer’s claim is correct.<br />
(b) Sketch a possible dotplot of the distribution of sample<br />
data for an SRS of size 12 from this population.<br />
Extending the Concepts<br />
17. Difference of proportions A school superintendent<br />
believes that the proportion of North High School<br />
students with Internet access at home is greater<br />
than the proportion of South High School students<br />
with Internet access at home. To investigate, she<br />
selects SRSs of size n 5 50 from each school and<br />
finds p^ N 5 46/50 5 0.92 and p^ S 5 36/50 5 0.72.<br />
To determine if a difference in proportions of<br />
0.20 provides convincing evidence that North High<br />
School has a greater proportion of students with<br />
Internet access at home, we simulated two random<br />
samples of size n 5 50 from populations with the<br />
same proportion of students with Internet access.<br />
Then, we subtracted the sample proportions. Here<br />
are the results from repeating this process 100 times.<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 408<br />
p N 2 p S 5 0 and the superintendent<br />
got a sample difference of proportion of<br />
p^ N − p^ S = 0.20 by chance alone.<br />
18. (a) 300 C 25 = 1.95 × 10 36 different<br />
possible samples of 25 tomatoes.<br />
(b) Due to the extremely large number of<br />
possible samples, it is not practical to examine<br />
the complete sampling distribution of means<br />
for samples of size 25.<br />
19. (a) z = x − m<br />
s ;<br />
28,000 − 23,300<br />
0.67 = ;<br />
s<br />
s = 7014.93<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
–0.20 –0.10 0.00 0.10 0.20<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
Difference in proportion of students<br />
with Internet access at home<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
(a) There are ten dots at 0. Explain what these dots<br />
represent.<br />
(b) Would it be unusual to get a difference in sample<br />
proportions of at least 0.20 when there is no difference<br />
in the population proportions? Explain.<br />
(c) Based on your answer to part (b), is there convincing<br />
evidence that North High School has a greater<br />
proportion of students with Internet access at<br />
home? Explain.<br />
Recycle and Review<br />
18. Sampling tomatoes (4.8, 6.1) Zach runs a roadside<br />
stand during the summer, selling produce from his<br />
farm. On a single day in mid-August, he harvests<br />
300 tomatoes. Suppose Zach wants to take a simple<br />
random sample of 25 tomatoes from the day’s<br />
pick to estimate mean weight.<br />
(a) How many possible sets of 25 tomatoes could<br />
be sampled from the 300 tomatoes in the day’s<br />
crop?<br />
(b) What does this say about the practicality of examining<br />
the complete sampling distribution of the sample<br />
mean for samples of size 25 from this population?<br />
19. College debt (5.7) A report published by the Federal<br />
Reserve Bank of New York in 2012 reported the<br />
results of a nationwide study of college student<br />
debt. Researchers found that the average student<br />
loan balance per borrower is $23,300. They also<br />
reported that about one-quarter of borrowers owe<br />
more than $28,000. 4<br />
(a) Assuming that the distribution of student loan<br />
balances is approximately normal, estimate the<br />
standard deviation of the distribution of student<br />
loan balances.<br />
(b) Assuming that the distribution of student loan<br />
balances is approximately normal, use your answer<br />
to part (a) to estimate the proportion of borrowers<br />
who owe more than $54,000.<br />
(c) In fact, the report states that about 10% of borrowers<br />
owe more than $54,000. What does this<br />
fact indicate about the shape of the distribution of<br />
student loan balances?<br />
(d) The report also states that the median student loan<br />
balance is $12,800. Does this fact support your<br />
conclusion in part (c)? Explain.<br />
54,000 − 23,300<br />
(b) z =<br />
≈ 4.38;<br />
7014.93<br />
P(X ≥ 54,000)≈ P(Z ≥ 4.38) ≈ 0<br />
Using technology: Applet/normalcdf(lower:<br />
54000, upper:100000, mean:23300,<br />
SD:7014.93) 5 0.000006<br />
(c) If the distribution of loan balances is<br />
approximately normal, then we would expect<br />
almost no one to have a balance that large.<br />
Because 10% of borrowers owe more than<br />
$54,000, we can conclude that the distribution<br />
of loan balances isn’t normal and is rightskewed.<br />
(d) Yes; because the mean ($23,300) is so<br />
much larger than the median ($12,800), we can<br />
conclude that the distribution of loan balances<br />
is skewed to the right.<br />
18/08/16 4:59 PMStarnes_<strong>3e</strong>_CH0<br />
408<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 408<br />
11/01/17 3:54 PM
ddd<br />
ddd<br />
dddd<br />
ddd<br />
ddd<br />
dd<br />
ddd<br />
ddd<br />
ddd<br />
Lesson 6.2<br />
Sampling Distributions:<br />
center and variability<br />
L e A r n i n g T A r g e T S<br />
d Determine if a statistic is an unbiased estimator of a population parameter.<br />
d Describe the relationship between sample size and the variability of a<br />
statistic.<br />
Learning Target Key<br />
The problems in the test bank are<br />
keyed to the learning targets using<br />
these numbers:<br />
d 6.2.1<br />
d 6.2.2<br />
Lesson 6.2<br />
AcT iviT y<br />
How many craft sticks are in the bag?<br />
In this activity, you will create a statistic for estimating<br />
the total number of craft sticks in a bag (N). The<br />
sticks are numbered 1, 2, 3, . . . , N. Near the end of the<br />
activity, your teacher will select a random sample of<br />
n 5 7 sticks and read the number on each stick to the<br />
class. The team that has the best estimate for the total<br />
number of sticks will win a prize.<br />
1. Form teams of three or four students. As a team,<br />
spend about 10 minutes brainstorming different<br />
ways to estimate the total number of sticks. Try to<br />
come up with at least three different statistics.<br />
2. Before your teacher provides the sample of sticks,<br />
use simulation to investigate the sampling distribution<br />
of each statistic. For the simulation, assume<br />
that there are N 5 100 sticks in the bag and that<br />
you will be selecting samples of size n 5 7.<br />
j Using your TI-83/84 calculator, select an SRS of<br />
size 7 using the command RandIntNoRep(lower:<br />
1,upper:100,n:7). [With older OS, use the command<br />
RandInt(lower:1,upper:100,n:7) and verify<br />
that there are no repeated numbers. If there are<br />
repeats, press ENTER to get a new sample.]<br />
j For each sample, calculate the value of each of<br />
your three statistics.<br />
Unbiased Estimators<br />
j<br />
j<br />
graph these values on a set of dotplots like those<br />
shown here.<br />
Perform as many trials of the simulation as possible.<br />
Statistic 3 Statistic 2 Statistic 1<br />
60<br />
70 80 90 100 110 120 130 140<br />
Estimated total<br />
3. Based on the simulated sampling distributions,<br />
which of your statistics is likely to produce the<br />
best estimate? Discuss as a team.<br />
4. Your teacher will now draw a sample of n 5 7<br />
sticks from the bag. On a piece of paper, write<br />
the names of your group members, your group’s<br />
estimate for the number of sticks in the bag (a<br />
number), and the statistic you used to calculate<br />
your estimate (a formula).<br />
In the craft sticks activity, the goal was to estimate the maximum value in a population,<br />
with the assumption that the members of the population are numbered 1, 2, . . . , N.<br />
Two possible statistics that might be used to estimate N are the sample maximum (max)<br />
and twice the sample median (2 3 median).<br />
Assuming that the population has N 5 100 members and we use SRSs of size n 5 7,<br />
Figure 6.2 shows the simulated sampling distributions of the sample maximum and<br />
twice the sample median.<br />
409<br />
Bell Ringer<br />
Suppose you wish to estimate the<br />
average (mean) height of all students at<br />
your school by taking a random sample<br />
of students and calculating the average<br />
height of the sample. Would you expect<br />
the sample mean to be closer to the<br />
true average height from a sample of<br />
4 students or 40 students?<br />
The best statistics are centered at 100<br />
with low variability. A pre-made Fathom<br />
file is included in the Teacher’s Resource<br />
Materials. If you don’t have Fathom,<br />
the figure shows simulated sampling<br />
distributions for several commonly used<br />
statistics using 200 random samples. The<br />
statistics are<br />
• TwiceMean 5 2 · sample mean<br />
• TwiceMedian 5 2 · sample median<br />
• Max 5 sample maximum<br />
• MeanPlusMed 5 mean 1 median<br />
• SumQuartiles 5 Q 1 1 Q 3<br />
• TwiceIQR 5 2 · IQR<br />
• MeanPlus2SD 5 sample mean 1<br />
2 · sample standard deviation<br />
• Partition 5 (8/7) · sample maximum<br />
TwiceMean<br />
d<br />
d<br />
dd dd ddd<br />
d<br />
dd<br />
d d ddd dddd d<br />
dd d<br />
ddddddddd<br />
dddd d d<br />
dd ddddd d<br />
dd d dd d ddd<br />
d<br />
18/08/16 4:59 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 409<br />
Activity Overview<br />
Time: 40–50 minutes<br />
Materials: Graphing calculators or an Internetconnected<br />
device for each student or group of<br />
students, a prize for the winning team, and a<br />
population of craft sticks. We recommend using<br />
at least 100 sticks (but not exactly 100). Use an<br />
opaque container so students can’t use their<br />
eyes to estimate the total.<br />
Teaching Advice: The point of this activity is to<br />
illustrate the concepts of bias and variability of<br />
statistics. This is a long activity and should be<br />
done during an entire class period.<br />
Start by selecting one or two sticks to<br />
make the contents of the bag less abstract.<br />
Be patient in Step 1! Teams will struggle.<br />
18/08/16 4:59 PM<br />
Let them keep at it. Once one statistic is<br />
proposed, usually more will follow.<br />
During Step 1, rotate around the room to<br />
give suggestions to teams that are struggling.<br />
Consider giving away one or more of the<br />
methods used later in the lesson, such as the<br />
sample maximum, twice the sample median,<br />
or twice the sample mean. These are all<br />
decent statistics but not the best.<br />
During Step 2, students can use their<br />
calculators or a random generator on the<br />
Internet (like random.org) to generate 7<br />
integers without repeats. Groups should<br />
generate at least 20 samples. If you have the<br />
ability to simulate the sampling distributions<br />
for a number of statistics using Fathom or<br />
other software, consider offering another<br />
prize for the team with the best statistic.<br />
TwiceMedian<br />
Max<br />
MeanPlusMed<br />
SumQuartiles<br />
TwiceIQR<br />
MeanPlus2SD<br />
Partition<br />
d<br />
d<br />
dddd<br />
d d<br />
d ddd dd<br />
ddd ddd dd<br />
ddd ddd dddd dd d dd d d dddddddd ddd<br />
d dddd ddd d dd dd<br />
d<br />
dd<br />
dd<br />
d<br />
ddd<br />
d<br />
d<br />
dddd dd<br />
ddd<br />
ddddd d<br />
d d ddddddd ddddddddd d d<br />
dd ddddd<br />
d<br />
d d dd dd dddd<br />
dddd ddd<br />
dddd d ddd dd d dd d d<br />
d<br />
dd<br />
d d dd<br />
d<br />
ddddddd dd ddddddd d dd<br />
dd<br />
dd ddd<br />
dd d d ddd d dd d<br />
ddd<br />
dd<br />
d ddd d<br />
dd<br />
d dd<br />
dd<br />
ddd<br />
dddd<br />
ddddd<br />
d<br />
ddd dd d ddddd d ddd ddd dd d ddd<br />
d d dddd<br />
d<br />
d<br />
ddddd<br />
dddd d d dd ddd<br />
ddddddddd ddd d<br />
dd<br />
ddddd dddd ddddddddddd ddddd<br />
d 20 40 60 80 100 120 140 160 180 200<br />
ddd<br />
d dddd dd dddd dd dd<br />
dddd d<br />
ddddd d<br />
ddd dd<br />
ddd d<br />
dd ddd dddd<br />
dddddddd d ddddd dddd dddddddd<br />
d<br />
d dd ddd<br />
d<br />
dd d<br />
ddd<br />
dd d dd d<br />
Answers:<br />
1. Answers will vary by student group.<br />
2. Dotplots will vary.<br />
3. Answers will vary. The best statistics<br />
are centered at 100 with low<br />
variability.<br />
4. Answers will vary by student group.<br />
dd<br />
d<br />
d<br />
d<br />
ddd dd<br />
dd<br />
dd<br />
d<br />
dd<br />
dd d d<br />
d<br />
ddd<br />
L E S S O N 6.2 • Sampling Distributions: Center and Variability 409<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 409<br />
11/01/17 3:54 PM
410<br />
C H A P T E R 6 • Sampling Distributions<br />
TRM Lesson 6.2 Activity<br />
fAthom File<br />
If you are familiar with Fathom software,<br />
you can use the pre-made Fathom file to<br />
simulate statistics in Step 2 of the Lesson<br />
6.2 activity. Click on the link in the TE-book,<br />
log into the Teacher’s Resource site, or<br />
access this resource on the TRFD.<br />
Teaching Tip<br />
If you need to save time, Lessons 6.2<br />
and/or 6.3 can be skipped without losing<br />
much continuity in future chapters.<br />
However, the other lessons in this<br />
chapter are crucial to understanding<br />
much of the remainder of the course.<br />
FigUre 6.2 Simulated<br />
sampling distributions<br />
of the sample maximum<br />
and twice the sample<br />
median for samples of<br />
size n 5 7 from a population<br />
with N 5 100.<br />
Sample<br />
maximum<br />
twice sample<br />
median<br />
d<br />
d<br />
d d<br />
d d d d<br />
d d d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
ddd d d d d d d d d d<br />
d<br />
d d d<br />
d d d d d d<br />
d d d d d<br />
d d d d d d<br />
d d d d d dd<br />
0 20 40 60 80 100 120 140 160 180 200<br />
Estimated total<br />
These simulated sampling distributions look quite different. The sampling distribution<br />
of the sample maximum is skewed left, while the sampling distribution of<br />
twice the sample median is roughly symmetric.<br />
The values of the sample maximum are consistently less than the population maximum<br />
N. However, the values of twice the sample median aren’t consistently less than<br />
or consistently greater than the population maximum N. It appears that twice the<br />
sample median might be an unbiased estimator of the population maximum, while<br />
the sample maximum is clearly biased.<br />
DEFINITION unbiased estimator<br />
A statistic used to estimate a parameter is an unbiased estimator if the mean of its sampling<br />
distribution is equal to the value of the parameter being estimated.<br />
cAutIOn<br />
!<br />
The use of the word “bias” here is consistent with its use in <strong>Ch</strong>apter 3. The design<br />
of a statistical study shows bias if it would consistently underestimate or consistently<br />
overestimate the value you want to know when you repeat the study many times.<br />
Recall the Federalist Papers activity (page 188) in which the estimates were consistently<br />
too large when students were allowed to choose the words in the sample. Don’t<br />
trust an estimate that comes from a biased sampling method.<br />
Alternate Example<br />
What is the mean-ing of bias?<br />
Unbiased estimators<br />
PROBLEM: The dotplot displays<br />
simulated sampling distributions of two<br />
statistics that can be used to estimate<br />
the mean of a population distribution.<br />
The simulated sampling distributions are<br />
based on 1000 SRSs of size n 5 5, and<br />
the population mean m 5 40. The mean<br />
of each distribution is indicated by a blue<br />
line segment.<br />
a<br />
e XAMPLe<br />
Why do we divide by n 2 1?<br />
Unbiased estimators<br />
PROBLEM: In <strong>Ch</strong>apter 1, you learned to calculate the<br />
standard deviation of sample data using the formula<br />
∑ (x i − x ) 2<br />
s x = Å n − 1<br />
What if you divided by n instead of n 2 1? Let’s<br />
simulate the sampling distributions of two statistics<br />
that can be used to estimate the variance of a<br />
distribution, where the variance is the square of the<br />
standard deviation (variance 5 standard deviation 2 ).<br />
∑ (x i − x ) 2<br />
Statistic 1:<br />
Statistic 2: ∑ (x i − x ) 2<br />
n − 1<br />
n<br />
These simulated sampling distributions are based<br />
on 1000 SRSs of size n 5 3 from a population with<br />
variance 5 25. The mean of each distribution is<br />
indicated by a blue line segment.<br />
Is either of these statistics an unbiased estimator of<br />
the population variance? Explain your reasoning.<br />
Statistic 2 Statistic 1<br />
d<br />
d<br />
d d dd dd d<br />
dd<br />
d<br />
dddd dd ddddddd<br />
d ddd dd d<br />
ddd dddd d<br />
d d ddd d<br />
ddd<br />
ddd<br />
dd dddddddddd dd dd<br />
dd<br />
dd d<br />
dddd d<br />
dd<br />
d d<br />
d dd d dddddddddddddddddd d<br />
ddddddddddddddddddddddddd<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 410<br />
d dd<br />
d dddd d ddd<br />
dd dd ddddddddd<br />
dd d dddd dd<br />
d d d ddddddd dd d d dd d dd<br />
dddd dddddd<br />
dd ddd<br />
dddd d d<br />
dd<br />
d d d<br />
dddd dd dd ddd ddd d<br />
dd ddddddddddddd d dddddd d d d d d d d d<br />
0 20 40 60 80 100 120 140 160 180 200<br />
Estimated mean<br />
dd d<br />
dd d d d<br />
d<br />
Is either of these statistics an unbiased estimator of the population mean? Explain your<br />
reasoning.<br />
SOLUTION: Statistic 1 appears to be unbiased because the mean of its sampling<br />
distribution is very close to 40, the value of the population mean. Statistic 2 appears to be<br />
biased because the mean of its sampling distribution is about 44, which is clearly greater<br />
than 40, the value of the population mean.<br />
FYI<br />
For a sample size of 7, the best estimator of<br />
N in a population like the one in the activity<br />
is (8/7) · sample maximum 2 1. For any<br />
sample size n, the best estimator is<br />
(n 1 1)/n · sample maximum 2 1.<br />
FYI<br />
The definition of an unbiased estimator given<br />
here is based on the mean of a sampling<br />
distribution. If the median of the sampling<br />
distribution of a statistic is equal to the value<br />
of the parameter being estimated, it is also<br />
considered unbiased.<br />
18/08/16 4:59 PMStarnes_<strong>3e</strong>_CH0<br />
410<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 410<br />
11/01/17 3:54 PM
L E S S O N 6.2 • Sampling Distributions: Center and Variability 411<br />
Statistic 1<br />
Statistic 2<br />
0 20 40 60 80 100 120 140 160 180 200<br />
Estimated variance<br />
SOLUTION:<br />
Statistic 1 appears to be unbiased because the mean of<br />
its sampling distribution is very close to 25, the value of<br />
the population variance. Statistic 2 appears to be biased<br />
because the mean of its sampling distribution is clearly less<br />
than 25, the value of the population variance.<br />
FOR PRACTICE TRY EXERCISE 1.<br />
Teaching Tip:<br />
Differentiate<br />
Students who have trouble with<br />
mathematical notation might be<br />
intimidated by the formulas in the<br />
preceding example. Tell these students<br />
to ignore the formulas and focus on<br />
the big idea that there are two slightly<br />
different ways to calculate the standard<br />
deviation.<br />
Lesson 6.2<br />
We divide by n 2 1 when calculating the sample variance so it will be an unbiased<br />
estimator of the population variance. If we divided by n instead, our estimates would<br />
be consistently too small. Likewise, it is better to divide by n 2 1 instead of n when<br />
calculating the standard deviation for a distribution of sample data.<br />
Sampling Variability<br />
Another possible statistic that could be used in the craft sticks activity is twice the<br />
sample mean. Figure 6.3 shows the simulated sampling distributions of twice the<br />
sample mean and twice the sample median.<br />
Twice sample<br />
mean<br />
Twice sample<br />
median<br />
d<br />
dd<br />
d ddd dddd<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d d dd d d d d dd<br />
dd ddddddddd<br />
dd d d dddddd<br />
ddd d d d dddd<br />
0 20 40 60 80 100 120 140 160 180 200<br />
Estimated total<br />
Both statistics appear to be unbiased estimators because the mean of each sampling<br />
distribution is around 100. However, the sampling distribution of twice the<br />
sample mean (standard deviation ≈ 22) is less variable than the sampling distribution<br />
of twice the sample median (standard deviation ≈ 34). In general, we prefer statistics<br />
that are less variable because they produce estimates that tend to be closer to the value<br />
of the parameter.<br />
For some parameters, there is an obvious choice for a statistic. For example, to estimate<br />
the proportion of successes in a population p, we use the proportion of successes<br />
in the sample, p^ . Fortunately, p^ is an unbiased estimator of p. And as we learned in<br />
Lesson 3.3, we can reduce the variability of an estimate by increasing the sample size.<br />
Figure 6.4 on the next page shows the simulated sampling distributions for p^ 5<br />
the proportion of students in the sample who take the bus to school when taking SRSs<br />
of size n 5 10 and SRSs of size n 5 50 from a population in which the proportion of<br />
all students who take the bus to school is p 5 0.70.<br />
FigUre 6.3 Simulated<br />
sampling distributions of<br />
twice the sample mean<br />
and twice the sample<br />
median for samples<br />
of size n 5 7 from a<br />
population with N 5 100.<br />
Common Error<br />
Some students think that bias is about<br />
the shape of a sampling distribution.<br />
These students think that a statistic is<br />
unbiased if its sampling distribution<br />
is symmetric and/or mound-shaped.<br />
Remind them that bias is about the<br />
center of the sampling distribution.<br />
18/08/16 4:59 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 411<br />
18/08/16 5:00 PM<br />
L E S S O N 6.2 • Sampling Distributions: Center and Variability 411<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 411<br />
11/01/17 3:54 PM
dd<br />
dd d d dddddd<br />
412<br />
C H A P T E R 6 • Sampling Distributions<br />
Teaching Tip<br />
Have students look closely at the bottom<br />
dotplot in Figure 6.4 and the dotplot<br />
in the next example. Ask them if this<br />
simulated sampling distribution has<br />
a familiar shape. Later in this chapter,<br />
students will learn that certain statistics<br />
have approximately normal sampling<br />
distributions.<br />
Alternate Example<br />
Can you hand me that wrench?<br />
Sampling variability<br />
PROBLEM: A local auto parts store has<br />
records of the daily sales of hand tools<br />
(in dollars) over the last several years.<br />
To estimate the average daily sales,<br />
a manager selects 10 days and finds<br />
the sample mean daily sales. Here is a<br />
simulated sampling distribution of x,<br />
the sample mean daily sales of hand<br />
tools (in dollars) for 1000 samples of<br />
size n 5 10.<br />
a<br />
FigUre 6.4 Simulated<br />
sampling distributions of<br />
the sample proportion p^<br />
for samples of size n 5 10<br />
and samples of size<br />
n 5 50 from a population<br />
with p 5 0.7.<br />
e XAMPLe<br />
n = 10<br />
n = 50<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
ddddddddddddddddddd<br />
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0<br />
Sample proportion who take the bus<br />
As expected, both simulated sampling distributions have means near p 5 0.70.<br />
Also, the sampling distribution of p^ is much more variable when the sample size<br />
is n 5 10, compared with n 5 50. In a small sample, it is plausible that the sample<br />
proportion could be much smaller or much larger than the parameter, just by chance.<br />
However, when the sample size gets bigger, we expect the sample proportion to be<br />
fairly close to the value of the parameter.<br />
The lifetime of batteries: Hours or days?<br />
Sampling variability<br />
Decreasing sampling variability<br />
The sampling distribution of any statistic will have less variability when the sample size is larger.<br />
PROBLEM: For quality control, workers at a battery factory regularly<br />
select random samples of batteries to estimate the mean lifetime. Here is a<br />
simulated sampling distribution of x, the sample mean lifetime (in hours) for<br />
1000 random samples of size n 5 100 from a population of AAA batteries.<br />
(a) What would happen to the sampling distribution of the sample<br />
mean x if the sample size were n 5 50 instead? Justify.<br />
(b) What is the practical consequence of this change in sample size?<br />
d<br />
d<br />
d dddd<br />
d<br />
d<br />
ddd<br />
dd dd<br />
d ddddddd<br />
d ddd ddd<br />
dd<br />
d<br />
ddd d<br />
d d ddd<br />
ddd dd<br />
d d dd ddd dd<br />
dd d<br />
dd dd ddd ddd<br />
d d<br />
d<br />
dd d dd d dd<br />
d dd dd d dd d dd d d d dd dddd dd d dd<br />
d dddd dd<br />
d dd d ddd<br />
d d d d<br />
dd d ddd ddd d dddd dd<br />
ddd<br />
dddd d<br />
d d d<br />
d<br />
d d ddd d ddd d d d d dd dd ddd<br />
d dd ddd<br />
ddd<br />
d<br />
dd<br />
d dd d<br />
dd<br />
dd d d dd<br />
d dddd ddd dd dddddddddd ddddddddd<br />
d<br />
ddd<br />
ddd<br />
d dddd dddd d ddd<br />
dddd d dddd ddd<br />
ddd<br />
ddd ddd<br />
d dddd d d d dddd d dd dd<br />
d dddd d<br />
ddd ddd d ddd dddd dddd dddd ddd d<br />
ddd ddd<br />
dddd ddd dd dd<br />
d d d d ddd<br />
dddd dddddd ddd dddd d<br />
dd<br />
d ddd<br />
ddd d dddd d dddd ddd d<br />
ddd<br />
d d dddd dd<br />
dd ddd<br />
d dd<br />
ddd d d dd<br />
dd<br />
d d d ddd<br />
d d<br />
ddddd<br />
ddddd ddd d ddd<br />
d ddd ddd ddd ddd<br />
ddd<br />
dddd<br />
dd ddd<br />
ddd<br />
ddd dd dd ddd dddddd dddd dd<br />
dd ddd d d dd d<br />
dd dd<br />
d ddd<br />
ddd dd d dddddd d<br />
dd d ddddd<br />
d<br />
ddd dd dd<br />
dd<br />
ddd d<br />
ddd d dd<br />
d dd d<br />
ddd ddd<br />
d d<br />
ddd<br />
d ddd d<br />
d d<br />
ddd ddd ddd ddd d d dddddd ddd ddd ddd ddd ddd d d<br />
dd d<br />
ddd d<br />
dd dd dddd<br />
d<br />
d<br />
ddd d<br />
ddd ddd dd<br />
d<br />
d<br />
d d dd<br />
d<br />
d<br />
d<br />
d d d<br />
d<br />
d<br />
d dd ddd ddd d<br />
d d d dd dd<br />
100 120 140 160 180 200 220<br />
Sample mean daily sales of<br />
hand tools ($)<br />
(a) What would happen to the sampling<br />
distribution of the sample mean x if the<br />
sample size were n 5 30 instead? Justify<br />
your answer.<br />
(b) What is the practical consequence of<br />
this change in sample size?<br />
SOLUTION:<br />
(a) The sampling distribution of the<br />
sample mean x will be less variable<br />
because the sample size is larger.<br />
(b) The estimated mean daily sales of<br />
hand tools will typically be closer to the<br />
true mean daily sales of hand tools. In<br />
other words, the estimate will be more<br />
precise.<br />
d<br />
SOLUTION:<br />
(a) The sampling distribution of the sample mean x will be more variable because the<br />
sample size is smaller.<br />
44 46 48 50 52 54 56 58 60 62<br />
Sample mean lifetime (h)<br />
(b) The estimated mean lifetime will typically be farther away from the true mean lifetime. In other words, the<br />
estimate will be less precise.<br />
FOR PRACTICE TRY EXERCISE 5.<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 412<br />
Putting It All Together: Center and Variability<br />
We can think of the true value of the population parameter as the bullseye on a target<br />
and of the sample statistic as an arrow fired at the target. Both bias and variability<br />
describe what happens when we take many shots at the target.<br />
• Bias means that our aim is off and we consistently miss the bullseye in the same<br />
direction. That is, our sample values do not center on the population value.<br />
18/08/16 5:00 PMStarnes_<strong>3e</strong>_CH0<br />
412<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 412<br />
11/01/17 3:54 PM
L E S S O N 6.2 • Sampling Distributions: Center and Variability 413<br />
• High variability means that repeated shots are widely scattered on the target. In<br />
other words, repeated samples do not give very similar results.<br />
Figure 6.5 shows this target illustration of bias and variability. Notice that low variability<br />
(shots are close together) can accompany high bias (shots are consistently away<br />
from the bullseye in one direction). And low or no bias (shots center on the bullseye)<br />
can accompany high variability (shots are widely scattered). Ideally, we’d like our<br />
estimates to be accurate (unbiased) and precise (have low variability).<br />
d<br />
ddd dd dd<br />
High bias, low variability<br />
(a)<br />
d<br />
d<br />
d<br />
d d<br />
Low bias, high variability<br />
(b)<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
High bias, high variability<br />
(c)<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
FigUre 6.5 Bias and variability. (a) High bias, low variability. (b) Low bias, high variability.<br />
(c) High bias, high variability. (d) The ideal: no bias, low variability.<br />
L e SSon APP 6. 2<br />
How many tanks does the enemy have?<br />
During World War II, the Allies captured many german<br />
tanks. Each tank had a serial number on it. Allied<br />
commanders wanted to know how many tanks the<br />
germans had so that they could allocate their forces<br />
appropriately. They sent the serial numbers of the<br />
captured tanks to a group of mathematicians in<br />
Washington, D.C., and asked for an estimate of the<br />
total number of german tanks N.<br />
Here are simulated sampling distributions for three<br />
statistics that the mathematicians considered, using<br />
samples of size n 5 7. The blue line marks N, the total<br />
number of german tanks. The shorter red line segments<br />
mark the mean of each simulated sampling distribution.<br />
Statistic 3 Statistic 2 Statistic 1<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d dd<br />
d<br />
d<br />
dd<br />
d<br />
d ddd<br />
dd ddd d<br />
dd dd d d<br />
d<br />
d<br />
ddd<br />
d d<br />
d d d<br />
dd d<br />
ddd<br />
dd<br />
d dddddd d<br />
ddd<br />
d<br />
d<br />
dd<br />
d d<br />
ddd ddd d<br />
d<br />
d<br />
d<br />
N<br />
Estimated total<br />
dd d<br />
ddd<br />
The ideal: no bias, low variability<br />
(d)<br />
1. Do any of these statistics appear to be unbiased?<br />
Justify.<br />
2. Which of these statistics do you think is best?<br />
Explain your reasoning.<br />
3. Explain how the Allies could get a more precise<br />
estimate of the number of german tanks using the<br />
statistic you chose in Question 2.<br />
© Bettmann/Corbis<br />
Teaching Tip:<br />
Differentiate<br />
The target illustration in Figure 6.5<br />
will be the best way for some students<br />
to understand the ideas of bias and<br />
variability in sample statistics. Compare<br />
the statistics in the screenshot in the<br />
Activity Overview at the start of this<br />
lesson to the figure. The Max statistic<br />
corresponds to target (a), the TwiceIQR<br />
statistic corresponds to target (b), and<br />
the Partition statistic corresponds to<br />
target (d).<br />
There is no statistic from the<br />
screenshot in the Activity Overview at<br />
the start of this lesson that corresponds<br />
to target (c). <strong>Ch</strong>allenge your top students<br />
to draw a dotplot of a hypothetical<br />
statistic that would correspond to target<br />
(c). They should use the same scale as the<br />
one in the screenshot.<br />
Teaching Tip<br />
In the Lesson App, Statistic 1 is sample<br />
min 1 sample max, Statistic 2 is sample<br />
mean 1 3SD, Statistic 3 is sample max·<br />
(n 1 1)/n. The Allies used a statistic<br />
similar to Statistic 3 because it was<br />
unbiased and had low variability! Thus,<br />
their estimate of the number of tanks<br />
would be “on target” and was unlikely to<br />
be far from the true value.<br />
Lesson App<br />
Answers<br />
Lesson 6.2<br />
18/08/16 5:00 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 413<br />
18/08/16 5:00 PM<br />
1. Statistics 1 and 3 both appear to be<br />
unbiased because the mean of each<br />
sampling distribution is very close to N,<br />
the value of the population maximum.<br />
Statistic 2 appears to be biased because<br />
the mean of its sampling distribution<br />
is clearly more than N, the value of the<br />
population maximum.<br />
2. Statistic 3; while both Statistics 1 and<br />
3 are unbiased, Statistic 3 appears to<br />
have less variability.<br />
3. The Allies could get a more precise<br />
estimate of the number of German tanks<br />
by capturing more tanks (increasing the<br />
sample size). This way, the estimated<br />
number of tanks would typically be<br />
closer to the true number of tanks<br />
(more precise).<br />
L E S S O N 6.2 • Sampling Distributions: Center and Variability 413<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 413<br />
11/01/17 3:54 PM
414<br />
C H A P T E R 6 • Sampling Distributions<br />
Lesson 6.2<br />
WhAT DiD y o U LeA rn?<br />
LEARNINg TARgET EXAMPLES EXERCISES<br />
Determine if a statistic is an unbiased estimator of a population<br />
parameter.<br />
p. 410 1–4<br />
Describe the relationship between sample size and the variability of a<br />
statistic.<br />
p. 412 5–8<br />
TRM full Solutions to Lesson 6.2<br />
Exercises<br />
You can find the full solutions for this<br />
lesson by clicking on the link in the<br />
TE-book, logging into the Teacher’s<br />
Resource site, or accessing this resource<br />
on the TRFD.<br />
Answers to Lesson 6.2<br />
Exercises<br />
1. Yes; the mean of the sampling<br />
distribution is very close to 22.96,<br />
the value of the population median.<br />
2. No; the mean of the sampling<br />
distribution is clearly more than 0.20,<br />
the value of the population minimum.<br />
3. No; the mean of the sampling<br />
distribution is clearly less than 153.53, the<br />
value of the population range. Population<br />
range 5 max 2 min 5 153.73 2 0.20<br />
5 153.53.<br />
Exercises<br />
Mastering Concepts and Skills<br />
Exercises 1–4 refer to the following setting. The manager<br />
of a grocery store records the total amount spent<br />
(in dollars) for each customer who makes a purchase<br />
at his store during a week. The values in the table<br />
summarize the distribution of amount spent for this<br />
population:<br />
N mean SD Min Q 1 med Q 3 Max<br />
749 29.85 24.63 0.20 12.29 22.96 39.93 153.73<br />
1. Is the median unbiased? To investigate if the sample<br />
pg 410 median is an unbiased estimator of the population<br />
median, 1000 SRSs of size n 5 10 were selected from<br />
the population described. The sample median for<br />
each of these samples was recorded on the dotplot.<br />
The mean of the simulated sampling distribution is<br />
indicated by an orange line segment. Does the sample<br />
median appear to be an unbiased estimator of the<br />
population median? Explain your reasoning.<br />
d<br />
d d<br />
d<br />
d<br />
d<br />
d<br />
d d<br />
d<br />
d<br />
Lesson 6.2<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
ddddd<br />
dd d<br />
0 10 20 30 40 50 60 70<br />
Sample median<br />
2. Is the minimum unbiased? To investigate if the sample<br />
minimum is an unbiased estimator of the population<br />
minimum, 1000 SRSs of size n 5 10 were<br />
selected from the population described. The sample<br />
minimum for each of these samples was recorded<br />
on the dotplot. The mean of the simulated sampling<br />
distribution is indicated by an orange line segment.<br />
Does the sample minimum appear to be an unbiased<br />
estimator of the population minimum? Explain your<br />
reasoning.<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d d<br />
d d<br />
d<br />
d<br />
d d d d d d d<br />
0 5 10 15 20 25<br />
Sample minimum<br />
3. Is the range unbiased? To investigate if the sample<br />
range is an unbiased estimator of the population<br />
range, 1000 SRSs of size n 5 10 were selected from<br />
the population described. The sample range for<br />
each of these samples was recorded on the dotplot.<br />
The mean of the simulated sampling distribution<br />
is indicated by an orange line segment. Does the<br />
sample range appear to be an unbiased estimator<br />
of the population range? Explain your reasoning.<br />
d<br />
d d<br />
d<br />
d<br />
d<br />
dd<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d d<br />
d<br />
d<br />
d<br />
20 40 60 80 100 120 140 160<br />
Sample range<br />
d<br />
d<br />
d<br />
d<br />
d d<br />
d<br />
dd d d<br />
d<br />
d d d<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 414<br />
Teaching Tip<br />
Exercise 19 on p. 416 previews Lesson 6.3,<br />
which is based on binomial random variables.<br />
Make sure your students do this exercise.<br />
18/08/16 5:00 PMStarnes_<strong>3e</strong>_CH0<br />
414<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 414<br />
11/01/17 3:54 PM
L E S S O N 6.2 • Sampling Distributions: Center and Variability 415<br />
4. Is the IQR unbiased? To investigate if the sample<br />
IQR is an unbiased estimator of the population<br />
IQR, 1000 SRSs of size n 5 10 were selected from<br />
the population described. The sample IQR for each<br />
of these samples was recorded on the dotplot. The<br />
mean of the simulated sampling distribution is indicated<br />
by an orange line segment. Does the sample<br />
IQR appear to be an unbiased estimator of the<br />
population IQR? Explain your reasoning.<br />
dd<br />
0<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
20<br />
d<br />
d<br />
d<br />
d<br />
d<br />
dd<br />
d<br />
d<br />
d d<br />
d<br />
40 60<br />
Sample IQR<br />
d d d<br />
d<br />
d<br />
d d dd d d<br />
80<br />
100<br />
5. More about medians Refer to Exercise 1.<br />
(a) What would happen to the sampling distribution of<br />
pg 412 the sample median if the sample size were n 5 50<br />
instead? Justify.<br />
(b) What is the practical consequence of this change in<br />
sample size?<br />
6. More about minimums Refer to Exercise 2.<br />
(a) What would happen to the sampling distribution of<br />
the sample minimum if the sample size were n 5 50<br />
instead? Justify.<br />
(b) What is the practical consequence of this change in<br />
sample size?<br />
7. More about ranges Refer to Exercise 3.<br />
(a) What would happen to the sampling distribution<br />
of the sample range if the sample size were n 5 5<br />
instead? Justify.<br />
(b) What is the practical consequence of this change in<br />
sample size?<br />
8. More about IQRs Refer to Exercise 4.<br />
(a) What would happen to the sampling distribution<br />
of the sample IQR if the sample size were n 5 5<br />
instead? Justify.<br />
(b) What is the practical consequence of this change in<br />
sample size?<br />
Applying the Concepts<br />
9. <strong>Ch</strong>olesterol in teens A study of the health of teenagers<br />
plans to measure the blood cholesterol levels of<br />
an SRS of 13- to 16-year-olds. The researchers will<br />
report the mean x from their sample as an estimate<br />
of the mean cholesterol level m in this population.<br />
Explain to someone who knows little about statistics<br />
what it means to say that x is an unbiased<br />
estimator of m.<br />
10. Predict the election A polling organization plans to<br />
ask a random sample of likely voters who they will<br />
vote for in an upcoming election. The researchers<br />
will report the sample proportion p^ that favors the<br />
incumbent as an estimate of the population proportion<br />
p that favors the incumbent. Explain to<br />
someone who knows little about statistics what it<br />
means to say that p^ is an unbiased estimator of p.<br />
11. Sampling more teens Refer to Exercise 9. The sample<br />
mean x is an unbiased estimator of the population<br />
mean m no matter what size SRS the study chooses.<br />
Explain to someone who knows nothing about statistics<br />
why a large random sample will give more reliable<br />
results than a small random sample.<br />
12. Sampling more voters Refer to Exercise 10. The<br />
sample proportion p^ is an unbiased estimator of<br />
the population proportion p no matter what size<br />
random sample the polling organization chooses.<br />
Explain to someone who knows nothing about statistics<br />
why a large random sample will give more<br />
trustworthy results than a small random sample.<br />
13. Housing prices In a residential neighborhood, the<br />
median value of a house is $200,000. For which of<br />
the following sample sizes, n 5 10 or n 5 100, is<br />
the sample median most likely to be greater than<br />
$250,000? Explain.<br />
14. Houses with basements In a particular city, 74%<br />
of houses have basements. For which of the following<br />
sample sizes, n 5 10 or n 5 100, is the sample<br />
proportion of houses with a basement more likely<br />
to be greater than 0.70? Explain.<br />
15. Bias and variability The histograms show sampling<br />
distributions for four different statistics intended to<br />
estimate the same parameter.<br />
(i)<br />
(ii)<br />
(iii)<br />
(iv)<br />
Population parameter<br />
Population parameter<br />
Population parameter<br />
Population parameter<br />
10. If we chose many random samples<br />
and calculated the sample proportion<br />
p^ for each sample, the distribution of p^<br />
would be centered at the value of p. In<br />
other words, when we use p^ to estimate<br />
p, we will not consistently underestimate<br />
p or consistently overestimate p.<br />
11. A larger random sample will provide<br />
more information about the population<br />
and, therefore, more precise results.<br />
The variability of the distribution of x<br />
decreases as the sample size increases.<br />
12. A larger random sample will provide<br />
more information about the population<br />
and, therefore, more precise results.<br />
The variability of the distribution of p^<br />
decreases as the sample size increases.<br />
13. n 5 10; the sampling distribution<br />
of the sample median will be more<br />
variable with n 5 10 than with n 5 100.<br />
Because the distribution is more variable,<br />
it is more likely to get a sample median<br />
($250,000) that is far away from the true<br />
median ($200,000).<br />
14. n 5 100; the sampling distribution<br />
of the sample proportion will be less<br />
variable with n 5 100 than with n 5 10.<br />
Because the distribution is less variable,<br />
it is less likely to get a sample proportion<br />
that is far away (less than 0.70) from<br />
the true proportion (0.74). This makes it<br />
more likely for the sample proportion to<br />
be above 0.70 with n 5 100.<br />
Teaching Tip<br />
Don’t skip Exercise 15! It’s a wonderful<br />
way to assess the two learning targets<br />
from this Lesson. Consider having a<br />
short class discussion on it after students<br />
have had a chance to try the exercise for<br />
themselves.<br />
Lesson 6.2<br />
18/08/16 5:00 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 415<br />
4. Yes; the mean of the sampling distribution is<br />
very close to 27.64, the value of the population<br />
IQR. Population IQR 5 Q 3 2 Q 1 5 39.93 2 12.29<br />
5 27.64.<br />
5. (a) It will be less variable because the<br />
sample size is larger.<br />
(b) The estimated median amount spent will<br />
typically be closer to the true median amount<br />
spent. In other words, the estimate will be<br />
more precise.<br />
6. (a) It will be less variable because the<br />
sample size is larger.<br />
(b) The estimated minimum amount spent<br />
will typically be closer to the true minimum<br />
amount spent. In other words, the estimate<br />
will be more precise.<br />
18/08/16 5:00 PM<br />
7. (a) It will be more variable because the<br />
sample size is smaller.<br />
(b) The estimated range amount spent<br />
will typically be farther from the true range<br />
amount spent. In other words, the estimate<br />
will be less precise.<br />
8. (a) It will be more variable because the<br />
sample size is smaller.<br />
(b) The estimated IQR amount spent will typically<br />
be farther from the true IQR amount spent.<br />
In other words, the estimate will be less precise.<br />
9. If we chose many SRSs and calculated the<br />
sample mean x for each sample, the distribution<br />
of x would be centered at the value of m. In<br />
other words, when we use x to estimate m,<br />
we will not consistently underestimate m or<br />
consistently overestimate m.<br />
L E S S O N 6.2 • Sampling Distributions: Center and Variability 415<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 415<br />
11/01/17 3:54 PM
416<br />
C H A P T E R 6 • Sampling Distributions<br />
15. (a) Statistics (ii) and (iii) both appear<br />
to be unbiased because the mean of each<br />
sampling distribution is very close to the<br />
value of the population parameter.<br />
(b) Statistic (ii); while both statistics<br />
(ii) and (iii) are unbiased, statistic (ii) has<br />
lower variability.<br />
16. (a)<br />
10 + 5 + 10 + 7 + 9 41<br />
m =<br />
=<br />
5<br />
5 = 8.2<br />
(b)<br />
Sample #1: Abigail (10), x 5 7.5<br />
Bobby (5)<br />
Sample #2: Abigail (10), x 5 10<br />
Carlos (10)<br />
Sample #3: Abigail (10), x 5 8.5<br />
DeAnna (7)<br />
Sample #4: Abigail (10), x 5 9.5<br />
Emily (9)<br />
Sample #5: Bobby (5), x 5 7.5<br />
Carlos (10)<br />
Sample #6: Bobby (5), x 5 6<br />
DeAnna (7)<br />
Sample #7: Bobby (5), x 5 7<br />
Emily (9)<br />
Sample #8: Carlos (10), x 5 8.5<br />
DeAnna (7)<br />
Sample #9: Carlos (10), x 5 9.5<br />
Emily (9)<br />
Sample #10: DeAnna (7), x 5 8<br />
Emily (9)<br />
(c)<br />
d d d<br />
d<br />
d d d d d<br />
6 6.5 7 7.5 8 8.5 9 9.5 10<br />
Sample mean quiz score<br />
m x =<br />
7.5 + 10 + 8.5 + 9.5 + 7.5 +<br />
6 + 7 + 8.5 + 9.5 + 8<br />
10<br />
= 82<br />
10 = 8.2.<br />
Yes, the sample mean is an unbiased<br />
estimator of the population mean. The<br />
mean of the sampling distribution is equal<br />
to 8.2, which is the value of the population<br />
mean.<br />
(a) Which statistics are unbiased estimators? Justify<br />
your answer.<br />
(b) Which statistic does the best job of estimating the<br />
parameter? Explain.<br />
Extending the Concepts<br />
16. More about means In the Exercises for Lesson 6.1,<br />
you were introduced to the following population of<br />
2 male students and 3 female students, along with<br />
their quiz scores:<br />
Abigail 10 Bobby 5 Carlos 10 DeAnna 7 Emily 9<br />
(a) Calculate the mean quiz score for the entire population.<br />
(b) List all 10 possible SRSs of size n 5 2, calculate<br />
the mean quiz score for each sample, and display<br />
the sampling distribution of the sample mean in a<br />
dotplot.<br />
(c) Calculate the mean of the sampling distribution<br />
from part (b). Is the sample mean an unbiased estimator<br />
of the population mean? Explain.<br />
17. More about proportions In the Exercises for<br />
Lesson 6.1, you were introduced to the following<br />
population of 2 male students and 3 female<br />
students, along with their quiz scores:<br />
Abigail 10 Bobby 5 Carlos 10 DeAnna 7 Emily 9<br />
(a) Calculate the proportion of females in the entire<br />
population.<br />
(b) List all 10 possible SRSs of size n 5 2, calculate the<br />
proportion of females for each sample, and display<br />
the sampling distribution of the sample proportion<br />
in a dotplot.<br />
(c) Calculate the mean of the sampling distribution<br />
from part (b). Is the sample proportion an unbiased<br />
estimator of the population proportion? Explain.<br />
Recycle and Review<br />
18. Students and housing (4.3, 4.4) There are 104<br />
students in Professor Negroponte’s statistics class,<br />
49 males and 55 females. Sixty of the students live in<br />
the dorms and the rest live off campus. Twenty of the<br />
males live off-campus. <strong>Ch</strong>oose a student at random<br />
from this class. Let Event M 5 the student is male and<br />
Event D 5 the student lives in the dorms.<br />
(a) Construct a Venn diagram to represent the outcomes<br />
of this chance process using the events M<br />
and D.<br />
(b) Find each of the following probabilities and interpret<br />
them in context.<br />
(i) P(M c D) (ii) P(M C d D) (iii) P(D k M)<br />
19. Students and homework (5.3, 5.4) Refer to Exercise<br />
18. At the beginning of each day that Professor<br />
Negroponte’s class meets, he randomly selects a<br />
member of the class to present the solution to a homework<br />
problem. Suppose the class meets 40 times during<br />
the semester and the selections are made with<br />
replacement. Let X 5 the number of times a female<br />
student is selected to present a solution.<br />
(a) Is X a binomial random variable? Justify your<br />
answer.<br />
(b) Calculate the mean and standard deviation of X.<br />
(c) For the first 10 meetings of the class, Professor Negroponte<br />
selects only 1 female student to solve a problem.<br />
Is there convincing evidence that his selection<br />
process is not really random? Support your answer<br />
with an appropriate probability calculation.<br />
17. (a) p = 3 5 = 0.6<br />
(b)<br />
Sample #1: Abigail, Bobby p^ 5 0.5<br />
Sample #2: Abigail, Carlos p^ 5 0.5<br />
Sample #3: Abigail, DeAnna p^ 5 1<br />
Sample #4: Abigail, Emily p^ 5 1<br />
Sample #5: Bobby, Carlos p^ 5 0<br />
Sample #6: Bobby, DeAnna p^ 5 0.5<br />
Sample #7: Bobby, Emily p^ 5 0.5<br />
Sample #8: Carlos, DeAnna p^ 5 0.5<br />
Sample #9: Carlos, Emily p^ 5 0.5<br />
Sample #10: DeAnna, Emily p^ 5 1<br />
d<br />
d<br />
d d dd d d<br />
0 0.5 1<br />
Sample proportion of female<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 416<br />
(c)<br />
0.5 + 0.5 + 1 + 1 + 0 +<br />
0.5 + 0.5 + 0.5 + 0.5 + 1<br />
m p^ =<br />
= 6<br />
10<br />
10 = 0.6.<br />
Yes, the sample proportion is an unbiased<br />
estimator of the population proportion. The<br />
mean of the sampling distribution is equal<br />
to 0.6, which is the value of the population<br />
proportion.<br />
18. (a)<br />
Male 20 29 Dorms 31<br />
24<br />
20 + 29 + 31<br />
(b) (i) P(M c D) = P(M or D) =<br />
104<br />
= 80 = 0.769. There is about a 77% chance<br />
104<br />
that a randomly selected student is a male or<br />
lives in the dorm.<br />
(ii) P(M C d D) = P(M C 31<br />
and D)=<br />
104 = 0.298.<br />
There is about a 30% chance that a randomly<br />
selected student is not a male and lives in the<br />
dorm.<br />
P(D and M)<br />
(iii) P(D 0 M) = = 29∙104<br />
P(M) 49∙104 = 29<br />
49<br />
= 0.592. There is about a 59% chance that a<br />
randomly selected student lives in the dorm,<br />
given that the student is a male.<br />
Answer 19 is on page 417<br />
18/08/16 5:00 PMStarnes_<strong>3e</strong>_CH0<br />
416<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 416<br />
11/01/17 3:54 PM
18/08/16 5:00 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 417<br />
Lesson 6.3<br />
The Sampling Distribution<br />
of a Sample count<br />
(The normal Approximation<br />
to the Binomial)<br />
L e A r n i n g T A r g e T S<br />
d Calculate the mean and the standard deviation of the sampling distribution of<br />
a sample count and interpret the standard deviation.<br />
d Determine if the sampling distribution of a sample count is approximately<br />
normal.<br />
d If appropriate, use the normal approximation to the binomial distribution to<br />
calculate probabilities involving a sample count.<br />
In many cases, we are interested in the number of successes X in a random sample<br />
from some population. For example, X 5 the number of defective flash drives in a<br />
random sample of 10 flash drives or X 5 the number of Democrats in a random<br />
sample of 1000 registered voters. To do probability calculations involving X, we want<br />
an understanding of the sampling distribution of the sample count X.<br />
DEFINITION Sampling distribution of the sample count X<br />
The sampling distribution of the sample count X describes the distribution of values<br />
taken by the sample count X in all possible samples of the same size from the same<br />
population.<br />
The sampling distribution of X is closely related to the binomial distributions that<br />
you learned about in Lessons 5.3 and 5.4.<br />
Suppose that a supplier inspects an SRS of 10 flash drives from a shipment of<br />
10,000 flash drives in which 200 are defective. Let X 5 the number of bad flash drives<br />
in the sample. This is not quite a binomial setting. Because we are sampling without<br />
replacement, the independence condition is violated. The conditional probability that<br />
the second flash drive chosen is bad changes when we know whether the first is good<br />
or bad: P(second is bad | first is good) 5 200/9999 5 0.0200 but P(second is bad | first<br />
is bad) 5 199/9999 5 0.0199. These probabilities are very close because removing<br />
1 flash drive from a shipment of 10,000 changes the makeup of the remaining 9999<br />
flash drives very little. The distribution of X is very close to the binomial distribution<br />
with n 5 10 and p 5 0.02.<br />
Answers continued<br />
19. (a) Yes. Binary? “Success” 5 female is<br />
selected. “Failure” 5 male is selected.<br />
Independent? Knowing whether or not one<br />
randomly selected student is a female tells you<br />
nothing about whether or not another randomly<br />
selected student is a female. Number? n 5 40.<br />
Same probability? p = 55<br />
104 = 0.529<br />
(b) m X = np = 40(0.529) = 21.16<br />
s X = "np(1 − p) = "40(0.529)(1 − 0.529)<br />
= "40(0.529)(0.471) = "9.97 = 3.16<br />
(c) P(X = 0) = 10 C 0 (0.529) 0 (1 − 0.529) 10<br />
= 1(0.529) 0 (0.471) 10 = 0.0005<br />
P(X = 1) = 10C 1 (0.529) 1 (1 − 0.529) 9<br />
= 10(0.529) 1 (0.471) 9 = 0.006<br />
417<br />
18/08/16 5:00 PM<br />
P(X ≤ 1) = P(X = 0) + P(X = 1)<br />
= 0.0005 + 0.006 = 0.0065<br />
If the professor were to randomly choose<br />
students for the first 10 meetings, there is<br />
less than a 1% chance that he would select<br />
1 female or fewer purely by chance. Because<br />
this is unlikely, we have convincing evidence<br />
that his selection process is not really random.<br />
Teaching Tip<br />
To save time, Lessons 6.2 and/or 6.3<br />
can be skipped without losing much<br />
continuity in future chapters. However,<br />
the other lessons in this chapter are<br />
crucial to understanding much of the<br />
remainder of the course.<br />
Learning Target Key<br />
The problems in the test bank are<br />
keyed to the learning targets using<br />
these numbers:<br />
d 6.3.1<br />
d 6.3.2<br />
d 6.3.3<br />
BELL RINGER<br />
According to the manufacturer, the<br />
true proportion of blue M&M’S® milk<br />
chocolate candies is 0.24. If Mrs. Gallas<br />
takes a random sample of 50 candies,<br />
how many should she expect to be blue?<br />
If she repeatedly takes random samples<br />
of 50 candies, will she always get the<br />
same number of blue candies? Why or<br />
why not?<br />
FYI<br />
When sampling with replacement from<br />
a finite population or sampling from an<br />
infinite population, the distribution of<br />
the sample count is exactly binomial.<br />
However, these types of sampling<br />
methods are not common in practice. It<br />
is far more common to sample without<br />
replacement from a finite population.<br />
In this case, the sample count is usually<br />
approximately binomial.<br />
Lesson 6.3<br />
L E S S O N 6.3 • The Sampling Distribution of a Sample Count 417<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 417<br />
11/01/17 3:54 PM
418<br />
C H A P T E R 6 • Sampling Distributions<br />
Teaching Tip<br />
The rule of thumb mentioned here is<br />
sometimes called the 10% condition. As<br />
long as the sample size is less than 10%<br />
of the population size, the sample count<br />
will have a distribution that is close<br />
enough to a binomial distribution that<br />
binomial probabilities will be reasonably<br />
accurate.<br />
Teaching Tip<br />
Remind students that the mean is also<br />
called the expected value!<br />
In practice, we can ignore the violation of the independence condition caused by<br />
sampling without replacement whenever the sample size is relatively small compared<br />
to the population size. Specifically, we can assume that the sampling distribution of a<br />
sample count X is approximately binomial when the sample size is less than 10% of<br />
the population size.<br />
Center and Variability<br />
Because the sampling distribution of a sample count X is approximately binomial<br />
when the sample is a small fraction of the population, we can use the formulas from<br />
Lesson 5.4 to calculate the mean and standard deviation of X.<br />
How to Calculate μ x and σ x for a Binomial Distribution<br />
Suppose X is the number of successes in a random sample of size n from a large population<br />
with proportion of successes p. Then:<br />
• The mean of the sampling distribution of X is m X = np.<br />
• The standard deviation of the sampling distribution of X is s X = "np(1 − p).<br />
The formula for the mean is always correct, even if we are sampling without<br />
replacement. However, the formula for the standard deviation is not appropriate to<br />
use when the sample size is more than 10% of the population size.<br />
Alternate Example<br />
Rubber ducky, are you the one?<br />
Mean and SD of the sampling<br />
distribution of X<br />
PROBLEM: A popular carnival game<br />
has players choose a rubber duck from<br />
a small pool and look under the duck.<br />
If a special mark is on the duck, the<br />
player wins a prize. Suppose that a pool<br />
has 5000 ducks, 800 of which have the<br />
special mark. One generous father pays<br />
for his children to choose 20 rubber<br />
ducks. Let X 5 the number of ducks with<br />
the mark in the sample of 20.<br />
(a) Calculate the mean and standard<br />
deviation of the sampling distribution of X.<br />
(b) Interpret the standard deviation<br />
from part (a).<br />
a<br />
e XAMPLe<br />
How many flash drives are defective?<br />
Mean and SD of the sampling distribution of X<br />
PROBLEM: Two percent of the flash drives in a shipment of 10,000 flash drives are defective. An<br />
inspector randomly selects 10 flash drives from the shipment and records X 5 the number of<br />
defective flash drives in the sample.<br />
(a) Calculate the mean and standard deviation of the sampling distribution of X.<br />
(b) Interpret the standard deviation from part (a).<br />
SOLUTION:<br />
(a) m x = 10(0.02) = 0.2 flash drives<br />
s x = "10(0.02)(1 − 0.02) = 0.44 f lash drives<br />
(b) If the inspector took many samples of size 10, the number of<br />
defective flash drives would typically vary by about 0.44 from<br />
the mean of 0.2.<br />
Because the sample size is less than 10% of the<br />
population size, the distribution of X is approximately<br />
binomial with n 5 10 and p 5 0.2. The mean is m X = np<br />
and the standard deviation is s X = "np(1 − p).<br />
FOR PRACTICE TRY EXERCISE 1.<br />
SOLUTION:<br />
(a) m X = 20a 800 b = 20(0.16) = 3.2<br />
5000<br />
ducks with the mark<br />
s X = "20(0.16)(1 − 0.16) = 1.64 ducks<br />
with the mark<br />
(b) If the children took many samples<br />
of 20 ducks, the number of ducks with<br />
the mark would typically vary by about<br />
1.64 from the mean of 3.2.<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 418<br />
Teaching Tip<br />
The interpretation of the mean and standard<br />
deviation in the preceding example is the<br />
same as in past chapters. Connect the ideas of<br />
past lessons to the current one!<br />
18/08/16 5:00 PMStarnes_<strong>3e</strong>_CH0<br />
418<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 418<br />
11/01/17 3:55 PM
L E S S O N 6.3 • The Sampling Distribution of a Sample Count<br />
419<br />
Shape<br />
As you learned in Lesson 5.3, the shape of a binomial distribution can be skewed to<br />
the right, skewed to the left, or roughly symmetric. The histogram in Figure 6.6 shows<br />
the sampling distribution of X 5 the number of defective flash drives from the previous<br />
example. It is clearly skewed to the right.<br />
Probability<br />
.90<br />
.72<br />
.54<br />
.36<br />
FigUre 6.6 Probability<br />
histogram of X 5 the<br />
number of defective flash<br />
drives in a sample of size<br />
n 5 10 from a population<br />
in which p 5 0.02.<br />
Teaching Tip<br />
Remind students that the mean of a<br />
distribution is the balancing point of its<br />
histogram.<br />
Lesson 6.3<br />
18/08/16 5:00 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 419<br />
.18<br />
.00<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Number of defective flash drives<br />
The following activity explores the shape of the sampling distribution of a sample<br />
count for various combinations of n and p.<br />
AcT iviT y<br />
Simulating with the Normal Approximation to Binomial Distributions applet<br />
In this activity, you will explore the shape of the sampling distribution of a sample count X using an applet from<br />
the book’s website.<br />
1. Launch the Normal Approximation to the Binomial Distributions<br />
applet at highschool.bfwpub.com/spa<strong>3e</strong>. You<br />
.90<br />
will see a histogram with a normal curve overlaid.<br />
.72<br />
2. Using the sliders, set the number of trials to n 5 10 and the<br />
probability of success to p 5 0.02. Hint: You can also use<br />
.54<br />
the arrow keys on your computer’s keyboard to move the<br />
sliders. The normal curve has the same mean and standard<br />
deviation as the histogram, but it doesn’t model the<br />
.36<br />
histogram very well.<br />
.18<br />
3. Use the slider (or the arrow keys) to gradually<br />
change the probability from p 5 0.02 to p 5 1.00<br />
while keeping the number of trials the same.<br />
Does the normal curve fit well when p is close to<br />
0? Close to 0.5? Close to 1?<br />
4. Keep the number of trials set to n 5 10 and<br />
change the probability to p 5 0.1. Use the slider<br />
(or the arrow keys) to gradually increase the<br />
sample size from n 5 10 to n 5 100. Does the<br />
Probability<br />
.00<br />
0 1 2<br />
0.2<br />
3 4 5 6 7 8 9 10<br />
normal curve fit the histogram better when n is<br />
smaller or larger?<br />
5. Under what conditions will the distribution<br />
of X be approximately normal? Under what<br />
conditions will the distribution of X not be<br />
approximately normal?<br />
18/08/16 5:00 PM<br />
Activity Overview<br />
Time: 10–15 minutes<br />
Materials: An Internet-connected device<br />
for each student or group of students<br />
Teaching Advice: This activity<br />
helps students understand that the<br />
distribution of a sample count (a<br />
binomial distribution) is sometimes<br />
approximately normal, but not always<br />
approximately normal. The binomial<br />
distribution is never exactly a normal<br />
distribution.<br />
If you don’t have enough devices<br />
for each student, students can work in<br />
groups or you can demonstrate the applet<br />
to the entire class. Showing the applet<br />
as a demonstration also saves time, but<br />
doesn’t engage students as much.<br />
After students have had time to discuss<br />
their answers, be sure to emphasize the<br />
following points with the class:<br />
• The binomial distribution is never<br />
exactly normal.<br />
• The binomial distribution is perfectly<br />
symmetric when p 5 0.5.<br />
• The binomial distribution is more<br />
approximately normal as n increases.<br />
• Therefore, the criterion for deciding<br />
when a binomial distribution is “close”<br />
to normal should include the values<br />
of both n and p.<br />
Answers:<br />
1. Students should launch the applet.<br />
2. Students should input the specified<br />
values for n and p.<br />
3. The normal curve doesn’t<br />
approximate the binomial<br />
distribution very well.<br />
4. The normal curve fits the binomial<br />
best when p 5 0.5. It doesn’t fit well<br />
at all when p is close to 0 or 1.<br />
5. Student answers will vary. The normal<br />
curve fits better when for values of p<br />
near 0.5 and larger values of n.<br />
L E S S O N 6.3 • The Sampling Distribution of a Sample Count 419<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 419<br />
11/01/17 3:55 PM
420<br />
C H A P T E R 6 • Sampling Distributions<br />
As you learned in the activity, the shape of the sampling distribution of X will be<br />
approximately normal when the sample size is large enough. You also learned that<br />
“large enough” depends on the value of p. The farther p is from 0.5, the larger the<br />
sample size needs to be, as shown in Figure 6.7.<br />
0.330<br />
0.234<br />
0.148<br />
Probability<br />
0.165<br />
Probability<br />
0.117<br />
Probability<br />
0.074<br />
Teaching Tip<br />
Make sure students understand that<br />
both np and n(1 2 p) must be checked to<br />
see if they are at least 10.<br />
FYI<br />
The Large Counts condition presented<br />
here is one rule of thumb for<br />
ensuring that a binomial distribution<br />
is approximately normal. Another<br />
criterion is np ≥ 5 and n(1 2 p) ≥ 5. We<br />
recommend the Large Counts condition<br />
stated here.<br />
0.000<br />
0 8 10<br />
(a)<br />
n = 10, p = 0.8<br />
0.000<br />
0.000<br />
0 16 20<br />
0 40 50<br />
(b) n = 20, p = 0.8<br />
(c)<br />
n = 50, p = 0.8<br />
FigUre 6.7 Histograms of the sampling distribution of a sample count X with (a) n 5 10 and p 5 0.8,<br />
(b) n 5 20 and p 5 0.8, and (c) n 5 50 and p 5 0.8. As n increases, the shape of the sampling distribution<br />
gets closer and closer to normal.<br />
In practice, the sampling distribution of a sample count will have an approximately<br />
normal distribution when the Large Counts condition is met.<br />
DEFINITION the Large counts condition<br />
Suppose X is the number of successes in a random sample of size n from a population<br />
with proportion of successes p. The Large Counts condition says that the distribution of<br />
X will be approximately normal when<br />
np ≥ 10 and n(1 2 p) ≥10<br />
This condition is called “large counts” because np is the expected (mean) count of<br />
successes and n(1 2 p) is the expected (mean) count of failures.<br />
Alternate Example<br />
Spend or save?<br />
Shape of the sampling distribution of a<br />
sample count<br />
PROBLEM: Suppose that 24% of all<br />
Americans have more debt on their credit<br />
cards than they have money in their<br />
savings accounts. Let X 5 the number of<br />
Americans with more debt than savings<br />
in a random sample of 40 Americans.<br />
Would it be appropriate to use a normal<br />
distribution to model the sampling<br />
distribution of X ? Justify your answer.<br />
SOLUTION:<br />
a<br />
e XAMPLe<br />
How many teens have debit cards?<br />
Shape of the sampling distribution of a sample count<br />
PROBLEM: Suppose that 12% of teens in a large city<br />
have a debit card. Let X 5 the number of teens with a<br />
debit card in a random sample of 500 teens from this<br />
city. Would it be appropriate to use a normal distribution<br />
to model the sampling distribution of X? Justify<br />
your answer.<br />
SOLUTION:<br />
Because np 5 500(0.12) 5 60 ≥ 10 and n (1 2 p ) 5 500<br />
(1 2 0.12) 5 440 ≥ 10, the sampling distribution of X is<br />
approximately normal.<br />
<strong>Ch</strong>eck the Large Counts condition to determine if X will<br />
have an approximately normal distribution.<br />
In this context, 60 is the expected (mean) count of teens<br />
with a debit card and 440 is the expected (mean) count<br />
of teens without a debit card.<br />
FOR PRACTICE TRY EXERCISE 5.<br />
Because np 5 40(0.24) 5 9.6 < 10,<br />
the sampling distribution of X is not<br />
approximately normal. Although<br />
n(1 2 p) 5 40(1 2 0.24) 5 30.4 ≥ 10,<br />
we have still not met the Large Counts<br />
condition.<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 420<br />
18/08/16 5:01 PMStarnes_<strong>3e</strong>_CH0<br />
420<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 420<br />
11/01/17 3:55 PM
L E S S O N 6.3 • The Sampling Distribution of a Sample Count<br />
421<br />
Finding Probabilities Involving X<br />
When the Large Counts condition is met, we can use a normal distribution to calculate<br />
probabilities involving X 5 the number of successes in a random sample of<br />
size n.<br />
Is it fun to shop anymore?<br />
Probabilities involving X<br />
PROBLEM: Sample surveys show that fewer people<br />
enjoy shopping than in the past. A survey asked a<br />
nationwide random sample of 2500 adults if they<br />
agreed or disagreed with the statement “I like<br />
buying new clothes, but shopping is often frustrating<br />
and time-consuming.” 5 Suppose that exactly<br />
60% of all adult U.S. residents would say “Agree” if<br />
asked the same question. Calculate the probability<br />
that at least 1520 members of the sample would say<br />
“Agree.”<br />
SOLUTION:<br />
• Mean: m X 5 2500(0.60) 5 1500<br />
• SD: s X = "2500(0.6)(1 − 0.6) 5 24.49<br />
• Shape: Approximately normal because<br />
np 5 2500 (0.60) 5 1500 ≥ 10 and<br />
n (1 2 p ) 5 2500 (1 2 0.6) 5 1000 ≥ 10<br />
1426.53 1451.02 1475.51 1500 1524.49 1548.98 1573.47<br />
1520<br />
Sample count who would say “Agree”<br />
1520 − 1500<br />
Using Table A: Z = = 0.82<br />
24.49<br />
P (Z ≥ 0.82) 5 1 2 0.7939 5 0.2061<br />
Using technology : Applet/normalcdf (lower:1520,<br />
upper:100000, mean:1500, SD: 24.49) 5 0.2071<br />
e XAMPLe<br />
Let X 5 the number who would say “Agree.” The<br />
sampling distribution of X is approximately binomial<br />
with n 5 2500 and p 5 0.60.<br />
To use a normal approximation to calculate probabilities<br />
involving X, we need to know the mean, standard<br />
deviation, and shape of the sampling distribution of X.<br />
Recall that the mean is m X = np and the standard deviation<br />
is s X = "np(1 − p).<br />
1. Draw a normal distribution.<br />
2. Perform calculations.<br />
(i) Standardize the boundary value and use Table A to<br />
find the desired probability; or<br />
(ii) Use technology.<br />
FOR PRACTICE TRY EXERCISE 9.<br />
<strong>Ch</strong>ris Hondros/Getty Images<br />
Alternate Example<br />
The most romantic dinner?<br />
Probabilities involving X<br />
PROBLEM: Suppose that 23% of adult<br />
Americans would say the most romantic<br />
Valentine’s dinner option is preparing<br />
a home-cooked meal together. If you<br />
interviewed a random sample of 800 adult<br />
Americans, what is the probability that<br />
165 or fewer would say that this is the<br />
most romantic dinner option?<br />
SOLUTION:<br />
Let X 5 the count of adult Americans in<br />
the sample who would choose this option.<br />
• Mean: np 5 800(0.23) 5 184<br />
• SD: "np(1 − p)= "800(0.23)(1 − 0.23)<br />
= 11.90<br />
• Shape: Approximately normal because<br />
np 5 800(0.23) 5 184 ≥ 10 and<br />
n(1 2 p) 5 800(1 2 0.23) 5 616 ≥ 10.<br />
165<br />
Lesson 6.3<br />
18/08/16 5:01 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 421<br />
Teaching Tip<br />
In the “Is it fun to shop anymore?” example,<br />
the probability can be computed as a<br />
binomial distribution with the Probability<br />
applet or with a graphing calculator<br />
command: 1 2 binomcdf(2500, 0.6, 1519).<br />
The answer is 0.213, which is very close to the<br />
value using the normal approximation (0.207).<br />
Make sure your students understand that<br />
probabilities using the normal approximation<br />
will be close as long as the Large Counts<br />
condition is met.<br />
Teaching Tip<br />
18/08/16 5:01 PM<br />
Students might ask why we go to the<br />
trouble of using a normal approximation to<br />
calculate probabilities when one could just<br />
use the binomial distribution to calculate the<br />
probability. They’re not wrong! The reason is<br />
that using a single distribution (the normal<br />
distribution) in the following chapters will<br />
make our work much simpler.<br />
148.3 160.2 172.1 184.0 195.9 207.8 219.7<br />
Sample count who would<br />
choose this option<br />
165 − 184<br />
Using Table A: z = = −1.60<br />
11.9<br />
P(Z ≤ 21.60) 5 0.0548<br />
Using technology: Applet/normalcdf<br />
(lower:2100000, upper:165, mean:184,<br />
SD: 11.90) 5 0.0552<br />
L E S S O N 6.3 • The Sampling Distribution of a Sample Count 421<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 421<br />
11/01/17 3:55 PM
422<br />
C H A P T E R 6 • Sampling Distributions<br />
Lesson App<br />
Answers<br />
1. m X = np = 1500(0.12) = 180; s X =<br />
"np(1−p) ="1500(0.12)(1 − 0.12)<br />
= 12.59. If many samples of size 1500<br />
were taken, the number of American<br />
adults who identify themselves as black<br />
would typically vary by about 12.59 from<br />
the mean of 180.<br />
2. Because np = 1500(0.12) = 180 $ 10<br />
and n(1 − p) = 1500(1 − 0.12) = 1320<br />
$ 10, the sampling distribution of X is<br />
approximately normal.<br />
155 − 180<br />
3. z = ≈ −1.99;<br />
12.59<br />
205 − 180<br />
z = ≈ 1.99<br />
12.59<br />
P(155 ≤ X ≤ 205) ≈ P(−1.99 ≤ Z<br />
# 1.99) = 0.9767 − 0.0233 = 0.9543<br />
Using technology: Applet/normalcdf<br />
(lower:155, upper:205, mean:180,<br />
SD:12.59) 5 0.9529<br />
4. There’s about a 95% chance that the<br />
number of randomly selected American<br />
adults (out of an SRS of 1500) who will<br />
identify themselves as black is between<br />
155 and 205. So if our sample had<br />
fewer than 155 American adults who<br />
identify themselves as black, we would<br />
suspect that black Americans are being<br />
underrepresented in the sample. This<br />
same approach could be used to check<br />
for undercoverage using other variables<br />
known about the population.<br />
TRM Quiz 6A: Lessons 6.1–6.3<br />
You can find a prepared quiz for Lessons<br />
6.1–6.3 by clicking on the link in the<br />
TE-book, logging into the Teacher’s<br />
Resource site, or accessing this resource<br />
on the TRFD.<br />
L e SSon APP 6. 3<br />
How can we check for bias in a survey?<br />
One way of checking the effect of undercoverage,<br />
nonresponse, and other sources of bias in a sample<br />
survey is to compare the sample with known facts<br />
about the population. About 12% of American adults<br />
identify themselves as black. Suppose we take an SRS<br />
of 1500 American adults and let X be the number of<br />
blacks in the sample.<br />
1. Calculate the mean and standard deviation of the<br />
sampling distribution of X. Interpret the standard<br />
deviation.<br />
2. Justify that the sampling distribution of X is<br />
approximately normal.<br />
Exercises<br />
Lesson 6.3<br />
WhAT DiD y o U LeA rn?<br />
LEARNINg TARgET EXAMPLES EXERCISES<br />
Calculate the mean and the standard deviation of the sampling<br />
distribution of a sample count and interpret the standard deviation.<br />
Determine if the sampling distribution of a sample count is<br />
approximately normal.<br />
If appropriate, use the normal approximation to the binomial<br />
distribution to calculate probabilities involving a sample count.<br />
Mastering Concepts and Skills<br />
1. Lefties Eleven percent of students at a large high<br />
school are left-handed. A statistics teacher selects a<br />
random sample of 100 students and records X 5<br />
the number of left-handed students in the sample.<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of X.<br />
(b) Interpret the standard deviation from part (a).<br />
2. Hip dysplasia Dysplasia is a malformation of the<br />
hip socket that is very common in certain dog<br />
breeds and causes arthritis as a dog gets older.<br />
According to the Orthopedic Foundation for<br />
Animals, 11.6% of all Labrador retrievers have<br />
pg 418<br />
Lesson 6.3<br />
3. Calculate the probability that an SRS of 1500 American<br />
adults will contain between 155 and 205 blacks.<br />
4. Explain how a polling organization could use the<br />
results from the previous question to check for<br />
undercoverage and other sources of bias.<br />
p. 418 1–4<br />
p. 420 5–8<br />
p. 421 9–12<br />
hip dysplasia. 6 A veterinarian tests a random<br />
sample of 50 Labrador retrievers and records<br />
Y 5 the number of Labs with dysplasia in the<br />
sample.<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of Y.<br />
(b) Interpret the standard deviation from part (a).<br />
3. NASCAR cards and cereal boxes In an attempt to<br />
increase sales, a breakfast cereal company decides to<br />
offer a promotion. Each box of cereal will contain<br />
a collectible card featuring one NASCAR driver:<br />
Kyle Busch; Dale Earnhardt, Jr.; Kasey Kahne;<br />
Danica Patrick; or Jimmie Johnson. The company<br />
says that each of the 5 cards is equally likely to<br />
Rawpixel Ltd/Getty Images<br />
TRM full Solutions to Lesson 6.3<br />
Exercises<br />
You can find the full solutions for this<br />
lesson by clicking on the link in the TEbook,<br />
logging into the Teacher’s Resource<br />
site, or accessing this resource on the TRFD.<br />
Answers to Lesson 6.3<br />
Exercises<br />
1. (a) m X = np = 100(0.11) = 11;<br />
s X = "np(1−p) = "100(0.11)(1− 0.11)<br />
= 3.13<br />
(b) If many samples of size 100 were<br />
taken, the number of students who are<br />
left-handed would typically vary by<br />
about 3.13 from the mean of 11.<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 422<br />
2. (a) m Y = np = 50(0.116) = 5.8;<br />
s Y = "np(1− p) = "50(0.116)(1 − 0.116)<br />
= 2.26<br />
(b) If many samples of size 50 were taken, the<br />
number of Labs that have dysplasia would<br />
typically vary by about 2.26 from the mean<br />
of 5.8.<br />
3. (a) m X = np = 12a 1 5 b = 2.4;<br />
s X = "np(1 − p)<br />
= Å<br />
12a 1 5 b a1−1 5 b = 1.39<br />
(b) If many samples of size 12 were taken, the<br />
number of Kyle Busch cards would typically<br />
vary by about 1.39 from the mean of 2.4.<br />
4. (a) m Y = np = 50(0.20) = 10;<br />
s Y = "np(1− p) = "50(0.2)(1− 0.2) = 2.83<br />
(b) If many samples of size 50 were taken, the<br />
number of individuals who have never married<br />
would typically vary by about 2.83 from the<br />
mean of 10.<br />
5. Yes; because np = 100(0.11) = 11$ 10<br />
and n(1−p) = 100(1−0.11) = 89 ≥ 10, the<br />
sampling distribution of X is approximately<br />
normal.<br />
6. No; because np = 50(0.116)= 5.8 < 10,<br />
the sampling distribution of Y is not<br />
approximately normal.<br />
7. No; because np = 12a 1 b = 2.4 < 10,<br />
5<br />
the sampling distribution of X is not<br />
approximately normal.<br />
18/08/16 5:01 PMStarnes_<strong>3e</strong>_CH0<br />
422<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 422<br />
11/01/17 3:55 PM
L E S S O N 6.3 • The Sampling Distribution of a Sample Count<br />
423<br />
18/08/16 5:01 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 423<br />
appear in the 100,000 boxes of cereal that are part<br />
of this promotion. You buy 12 boxes and let X 5<br />
the number of Kyle Busch cards in the sample.<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of X.<br />
(b) Interpret the standard deviation from part (a).<br />
4. What, me marry? In the United States, 20% of<br />
adults ages 25 and older have never been married,<br />
more than double the figure recorded for 1960. 7<br />
Select a random sample of 50 U.S. adults ages 25<br />
and older and let Y 5 the number of individuals in<br />
the sample who have never married.<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of Y.<br />
(b) Interpret the standard deviation from part (a).<br />
5. Are lefties normal? Refer to Exercise 1. Would it be<br />
pg 420 appropriate to use a normal distribution to model<br />
the sampling distribution of X 5 the number of<br />
left-handed students in the sample? Justify your<br />
answer.<br />
6. Is hip dysplasia normal? Refer to Exercise 2. Would it<br />
be appropriate to use a normal distribution to model<br />
the sampling distribution of Y 5 the number of Labs<br />
with dysplasia in the sample? Justify your answer.<br />
7. Is NASCAR normal? Refer to Exercise 3. Would<br />
it be appropriate to use a normal distribution to<br />
model the sampling distribution of X 5 the number<br />
of Kyle Busch cards in the sample? Justify your<br />
answer.<br />
8. A normal marriage? Refer to Exercise 4. Would<br />
it be appropriate to use a normal distribution to<br />
model the sampling distribution of Y 5 the number<br />
of individuals in the sample who have never married?<br />
Justify your answer.<br />
9. Lefties are all right Refer to Exercises 1 and 5.<br />
pg 421 Calculate the probability that at least 15 of the<br />
members of the sample are left-handed.<br />
10. Never been married Refer to Exercises 4 and 8.<br />
Calculate the probability that at most 5 of the individuals<br />
in the sample have never been married.<br />
11. Public transportation In a large city, 34% of residents<br />
use public transportation at least once per<br />
week. If the mayor selects a random sample of 200<br />
residents, calculate the probability that at most 60<br />
residents in the sample use public transportation at<br />
least once per week.<br />
12. U.S. quarters According to www.usmint.gov, 54%<br />
of the quarters minted in 2014 were produced by<br />
the U.S. Mint in Denver, Colorado (the rest were<br />
produced in Philadelphia). In a random sample of<br />
200 quarters, what is the probability that at least<br />
115 of them were minted in Denver?<br />
8. Yes; because np = 50(0.20) = 10 $ 10<br />
and n(1− p) = 50(1− 0.20)= 40 ≥ 10, the<br />
sampling distribution of Y is approximately<br />
normal.<br />
9. From Exercises 1 and 5, X is approximately<br />
normal with a mean of 11 and standard<br />
deviation of 3.13.<br />
15 − 11<br />
z = ≈ 1.28; P(X ≥ 15)≈ P(Z ≥ 1.28)<br />
3.13<br />
= 1− 0.8997 = 0.1003<br />
Using technology: Applet/normalcdf(lower:15,<br />
upper:1000, mean:11, SD:3.13) 5 0.1006<br />
10. From Exercises 4 and 8, Y is<br />
approximately normal with a mean of 10 and<br />
standard deviation of 2.83.<br />
z = 5 − 10<br />
2.83 ≈ −1.77;<br />
P(Y ≤ 5) ≈ P(Z ≤ −1.77) = 0.0384<br />
Applying the Concepts<br />
13. Tasty chips For a statistics project, Zenon decided<br />
to investigate if students at his school prefer namebrand<br />
potato chips to store-brand potato chips. He<br />
prepared two identical bowls of chips, filling one<br />
with name-brand chips and the other with storebrand<br />
chips. Then, he selected a random sample<br />
of 30 students, had each student try both types of<br />
chips in random order, and recorded which type of<br />
chip each student preferred. Assume that 50% of<br />
students at Zenon’s school prefer the name-brand<br />
chips. Let X 5 the number of students in the sample<br />
that prefer the name-brand chips. 8<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of X. Interpret the standard<br />
deviation.<br />
(b) Justify that the distribution of X is approximately<br />
normal.<br />
(c) Calculate the probability that 19 or more of the<br />
students will prefer the name-brand chips.<br />
14. Blood types About 10% of people in the United<br />
States have type B blood. Suppose we take a random<br />
sample of 120 U.S. residents, and let X 5 the number<br />
of residents in the sample who have type B blood.<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of X. Interpret the standard<br />
deviation.<br />
(b) Justify that the distribution of X is approximately<br />
normal.<br />
(c) Calculate the probability that 16 or more individuals<br />
in the sample have type B blood.<br />
15. More chips! Refer to Exercise 13. In Zenon’s study,<br />
19 of the 30 students chose the name-brand chips.<br />
Based on your answer to Exercise 13(c), does this<br />
provide convincing evidence that more than half of<br />
the students at Zenon’s school prefer name-brand<br />
potato chips? Explain.<br />
16. More on blood type Refer to Exercise 14. Some people<br />
believe that one’s blood type has an impact on personality.<br />
For example, people with type B blood are<br />
supposed to be more creative, active, and passionate.<br />
To test this hypothesis, Jason selects a random sample<br />
of 120 art, music, and drama majors at his college<br />
and finds that 16 of them have type B blood. Based on<br />
your answer to Exercise 14(c), does this provide convincing<br />
evidence that art, music, and drama majors at<br />
Jason’s college are more likely than the general population<br />
to have type B blood? Explain.<br />
Extending the Concepts<br />
17. Binomial transportation Refer to Exercise 11. Use<br />
a binomial distribution to calculate the probability<br />
that at most 60 residents in the sample use public<br />
transportation at least once per week. Hint: See<br />
Lesson 5.4.<br />
18/08/16 5:01 PM<br />
Using technology: Applet/normalcdf(lower:<br />
−1000, upper:5, mean:10, SD:2.83) 5 0.0386<br />
11. X 5 the number of residents who use<br />
public transportation at least once per week.<br />
Mean: m X = np = 200(0.34) = 68;<br />
SD: s X = "np(1 − p)<br />
= "200(0.34)(1 − 0.34) = 6.70<br />
Shape: Approximately normal because<br />
np = 200(0.34) = 68 ≥ 10 and<br />
n(1−p) = 200(1− 0.34) = 132 ≥ 10.<br />
60 − 68<br />
z = ≈ −1.19;<br />
6.70<br />
P(X ≤ 60) ≈ P(Z ≤ −1.19) = 0.1170<br />
Using technology: Applet/normalcdf<br />
(lower:−1000, upper:60, mean:68,<br />
SD:6.70) 5 0.1162<br />
12. X 5 the number of quarters minted<br />
in Denver.<br />
Mean: m X = np = 200(0.54) = 108;<br />
SD: s X = "np(1 − p)<br />
= "200(0.54)(1 − 0.54) = 7.05<br />
Shape: Approximately normal because<br />
np = 200(0.54) = 108 ≥ 10 and<br />
n(1− p) = 200(1− 0.54) = 92 ≥ 10.<br />
115 − 108<br />
z = ≈ 0.99; P(X ≥ 115)<br />
7.05<br />
≈ P(Z ≥ 0.99) = 1− 0.8389 = 0.1611<br />
Using technology: Applet/normalcdf<br />
(lower:115, upper:1000, mean:108,<br />
SD:7.05) 5 0.1604<br />
13. (a) m X = np = 30(0.5) = 15<br />
students; s X = "np(1− p)<br />
= "30(0.5)(1 − 0.5) = 2.74 students.<br />
If many samples of size 30 were taken, the<br />
number of students who prefer namebrand<br />
chips would typically vary by about<br />
2.74 from the mean of 15.<br />
(b) Because np = 30(0.5) = 15 ≥ 10<br />
and n(1 − p) = 30(1 − 0.5) = 15 ≥ 10,<br />
the sampling distribution of X is<br />
approximately normal.<br />
19 − 15<br />
(c) z = ≈ 1.46; P(X ≥ 19) ≈<br />
2.74<br />
P(Z ≥ 1.46) = 1− 0.9279 = 0.0721<br />
Using technology: Applet/normalcdf<br />
(lower:19, upper:1000, mean:15,<br />
SD:2.74) 5 0.0722<br />
14. (a) m X = np = 120(0.1) = 12<br />
residents; s X = "np(1− p)<br />
= "120(0.1)(1 − 0.1) = 3.29 residents.<br />
If many samples of size 120 were<br />
taken, the number of residents who have<br />
type B blood would typically vary by<br />
about 3.29 from the mean of 12.<br />
(b) Because np = 120(0.1) = 12 $ 10<br />
and n(1− p) = 120(1 − 0.1) = 108 $ 10,<br />
the sampling distribution of X is<br />
approximately normal.<br />
16 − 12<br />
(c) z = ≈ 1.22; P(X ≥ 16) ≈<br />
3.29<br />
P(Z ≥ 1.22) = 1 − 0.8888 = 0.1112<br />
Using technology: Applet/normalcdf<br />
(lower:16, upper:1000, mean:12,<br />
SD:3.29) 5 0.112<br />
15. No; assuming that 50% of students<br />
prefer name-brand chips, there’s about a<br />
7% chance that the number of students<br />
who prefer name-brand chips (out of<br />
an SRS of 30) is 19 or more. The results<br />
from Zenon’s study could have happened<br />
purely by chance, so we do not have<br />
convincing evidence that more than half<br />
of the students prefer name-brand chips.<br />
Answers 16–17 are on page 424<br />
L E S S O N 6.3 • The Sampling Distribution of a Sample Count 423<br />
Lesson 6.3<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 423<br />
11/01/17 3:55 PM
Answers continued<br />
16. No; assuming that 10% of a group<br />
have type B blood, there’s about an 11%<br />
chance that the number who have type<br />
B blood (out of an SRS of 120) is 16 or<br />
more. The results from Jason’s study<br />
could have happened purely by chance,<br />
so we do not have convincing evidence<br />
that art, music, and drama majors at<br />
Jason’s college are more likely than the<br />
general population to have type B blood.<br />
17. X 5 the number of residents who<br />
use public transportation at least once<br />
per week. Using the applet (Probability,<br />
Binomial distribution, n 5 200,<br />
p 5 0.34, “at most 60 successes”) or<br />
binomcdf(n 5 200, p 5 0.34, X 5 60)<br />
gives P(X ≤ 60) = 0.131. (Compare to<br />
the answer of 0.1162 from Exercise 11.)<br />
18. (a) To provide a baseline for<br />
comparing the effects of the treatment.<br />
Otherwise, we wouldn’t be able to tell<br />
if the books or something else (e.g.,<br />
students maturing) caused an increase in<br />
reading ability.<br />
(b) The difference in the reading scores<br />
for the third-grade girls group was too<br />
large to be due only to chance variation<br />
in the random assignment to treatments.<br />
19.<br />
(a)<br />
Sister’s height (in.)<br />
70<br />
69<br />
68<br />
67<br />
66<br />
65<br />
64<br />
63<br />
62<br />
61<br />
60<br />
59<br />
58<br />
d<br />
65<br />
d<br />
d<br />
d d d<br />
66 67 68 69 70 71 72 73<br />
d<br />
d<br />
Brother’s height (in.)<br />
Direction: Positive. Form: Linear. Strength:<br />
Moderate. Outliers: No obvious outliers.<br />
(b) y^ = 27.635 + 0.527x, where x 5<br />
brother’s height and y 5 sister’s height.<br />
The slope of the line is 0.527, which tells<br />
us the predicted sister’s height increases<br />
by 0.527 inch for each additional increase<br />
of 1 inch in the brother’s height.<br />
(c) y^ = 27.635 + 0.527(70) = 64.525<br />
inches<br />
(d) The actual height of a sister is typically<br />
about 2.247 inches away from her<br />
predicted height using the least-squares<br />
regression line.<br />
d<br />
d<br />
424<br />
C H A P T E R 6 • Sampling Distributions<br />
Recycle and Review<br />
18. Summer reading (3.6, 3.8) A group of educational<br />
researchers studied the impact of summer reading<br />
with a randomized experiment involving secondand<br />
third-graders in North Carolina. Students<br />
were randomly assigned to either a group that was<br />
mailed one book a week for 10 weeks or a control<br />
group that was not mailed any books. Both groups<br />
were given a reading comprehension test at the<br />
start and end of the summer. Third-grade girls who<br />
were mailed books showed a statistically significant<br />
increase in reading ability, but third-grade boys and<br />
second-graders of both genders did not. 9<br />
(a) Explain the purpose of including a control group in<br />
this experiment.<br />
(b) Explain what is meant by “statistically significant<br />
increase” in the last sentence.<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 424<br />
PD LESSONS 6.4–6.6 Overview<br />
Watch the Lessons 6.4–6.6 overview video for<br />
guidance on teaching the content in these<br />
lessons. Find it in the Teacher’s Resource<br />
Materials by clicking on the link in the TEbook,<br />
logging into the Teacher’s Resource site,<br />
or accessing it on the TRFD.<br />
L e A r n i n g T A r g e T S<br />
19. Sisters and brothers (2.2, 2.5, 2.7) How strongly<br />
do physical characteristics of sisters and brothers<br />
correlate? Here are data on the heights (in inches)<br />
of 11 adult pairs: 10<br />
Brother 71 68 66 67 70 71 70 73 72 65 66<br />
Sister 69 64 65 63 65 62 65 64 66 59 62<br />
(a) Construct a scatterplot using brother’s height as the<br />
explanatory variable. Describe what you see.<br />
(b) Use technology to compute the least-squares<br />
regression line for predicting sister’s height from<br />
brother’s height. Interpret the slope in context.<br />
(c) Damien is 70 inches tall. Predict the height of his<br />
sister Tonya.<br />
(d) The standard deviation of residuals for this model<br />
is s 5 2.247. Interpret this value in context.<br />
Lesson 6.4<br />
The Sampling Distribution<br />
of a Sample Proportion<br />
d Calculate the mean and standard deviation of the sampling distribution of a<br />
sample proportion p^ and interpret the standard deviation.<br />
d Determine if the sampling distribution of p^ is approximately normal.<br />
d If appropriate, use a normal distribution to calculate probabilities involving p^ .<br />
What proportion of U.S. teens know that 1492 was the year in which Columbus<br />
“discovered” America? A Gallup poll found that 210 out of a random sample of 501<br />
American teens aged 13 to 17 knew this historically important date. 11 The sample<br />
proportion p^ 5 210/501 5 0.42 is the statistic that we use to estimate the unknown<br />
population proportion p. Because a random sample of 501 teens is unlikely to perfectly<br />
represent all teens, we can only say that “about” 42% of U.S. teenagers know<br />
that Columbus voyaged to America in 1492.<br />
To understand how much p^ varies from p and what values of p^ are likely to<br />
happen by chance, we want an understanding of the sampling distribution of the<br />
sample proportion p^ .<br />
Learning Target Key<br />
The problems in the test bank are keyed<br />
to the learning targets using these<br />
numbers:<br />
d 6.4.1<br />
d 6.4.2<br />
d 6.4.3<br />
18/08/16 5:01 PMStarnes_<strong>3e</strong>_CH0<br />
Teaching Tip<br />
Be picky with students about the correct<br />
symbols! The sample proportion is denoted p^<br />
and the population proportion is denoted p.<br />
BELL RINGER<br />
Mrs. Gallas loves blue M&M’S® milk chocolate<br />
candies. According to the manufacturer, the<br />
true proportion of blue M&M’S chocolate<br />
candies is 0.24. If Mrs. Gallas takes a random<br />
sample of 50 candies, what proportion of blue<br />
candies should she expect to have? Is she<br />
certain to get this proportion? Why or why not?<br />
424<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 424<br />
11/01/17 3:55 PM
L E S S O N 6.4 • The Sampling Distribution of a Sample Proportion 425<br />
DEFINITION Sampling distribution of the sample proportion p^<br />
The sampling distribution of the sample proportion p^ describes the distribution of<br />
values taken by the sample proportion p^ in all possible samples of the same size from the<br />
same population.<br />
When Mr. Ramirez’s class did the Penny for Your Thoughts activity at the beginning<br />
of the chapter, his students produced the “dotplot” in Figure 6.8 showing the<br />
simulated sampling distribution of p^ 5 the sample proportion of pennies from the<br />
2000s in 50 samples of size n 5 20.<br />
p<br />
p<br />
p p<br />
p p<br />
p p p<br />
p p p p<br />
p p p p p p<br />
p p p p p p<br />
p p p p p p p<br />
p p p p p p p p<br />
p p p p p p p p p<br />
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0<br />
Sample proportion of pennies from 2000s (n = 20)<br />
This distribution is roughly symmetric, with a mean of about 0.65 and a standard<br />
deviation of about 0.10. By the end of this lesson, you should be able to anticipate the<br />
shape, center, and variability of distributions like this one without getting your hands<br />
dirty in a jar of pennies.<br />
p<br />
FigUre 6.8 Simulated<br />
sampling distribution of<br />
the sample proportion of<br />
pennies in 50 samples of<br />
size n 5 20 from a population<br />
of pennies.<br />
Teaching Tip<br />
In Figure 6.8, ask students what the<br />
“dot” at 0.35 represents. It is the sample<br />
proportion of pennies from the 2000s<br />
for one sample of 20 pennies. That is,<br />
there were 7 pennies from the 2000s in<br />
one sample of 20 pennies, resulting in a<br />
sample proportion of 0.35.<br />
Teaching Tip<br />
This figure is a good opportunity to refer<br />
to the dotplots made by your students in<br />
the “A penny for your thoughts?” activity<br />
from Lesson 6.1. Compare the results<br />
from your class with those from<br />
Mr. Ramirez’s class.<br />
Lesson 6.4<br />
Center and Variability<br />
When we select random samples of size n from a population with proportion of<br />
successes p, the value of p^ will vary from sample to sample. As with the sampling<br />
distribution of the sample count X, there are formulas that describe the center and<br />
variability of the sampling distribution of p^ .<br />
How to Calculate μ p^ and σ p^<br />
Suppose that p^ is the proportion of successes in an SRS of size n drawn from a large population<br />
with proportion of successes p. Then:<br />
Teaching Tip<br />
Point out that any description of a<br />
sampling distribution must include<br />
information about shape, center, and<br />
variability. Center and variability will be<br />
examined in more detail in this lesson.<br />
• The mean of the sampling distribution of p^ is m p^ = p.<br />
p(1 − p)<br />
• The standard deviation of the sampling distribution of p^ is s p^ = . Å n<br />
Here are some important facts about the mean and standard deviation of the sampling<br />
distribution of the sample proportion p^ :<br />
• The sample proportion p^ is an unbiased estimator of the population proportion<br />
p. This is because the mean of the sampling distribution m p^ is equal to the<br />
population proportion p.<br />
• The standard deviation of the sampling distribution of p^ describes the typical<br />
distance between p^ and the population proportion p.<br />
FYI<br />
The formulas for m p^ and s p^ are true for<br />
the sampling distribution of p^ no matter<br />
what shape it has.<br />
Teaching Tip<br />
This is consistent with previous<br />
definitions of standard deviation as the<br />
typical distance a value falls from the<br />
mean of a distribution.<br />
18/08/16 5:01 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 425<br />
FYI<br />
The exact formula for the standard deviation<br />
of p^ when sampling without replacement is<br />
p(1 − p)<br />
s p^ = # N − n<br />
. When n is small<br />
Å n Å N − 1<br />
relative to N (less than 10%), the value of<br />
N − n<br />
is approximately 1 and therefore<br />
Å N − 1<br />
p(1 − p)<br />
doesn’t change the value of Å n<br />
N − n<br />
much. The factor is sometimes<br />
Å N − 1<br />
called the finite population correction factor.<br />
18/08/16 5:01 PM<br />
L E S S O N 6.4 • The Sampling Distribution of a Sample Proportion 425<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 425<br />
11/01/17 3:55 PM
426<br />
C H A P T E R 6 • Sampling Distributions<br />
• The sampling distribution of p^ is less variable for larger samples. This is indicated<br />
by the !n in the denominator of the standard deviation formula.<br />
• The formula for the standard deviation of the distribution of p^ requires that the<br />
observations are independent. In practice, we are safe assuming independence<br />
when sampling without replacement as long as the sample size is less than 10%<br />
of the population size.<br />
Alternate Example<br />
Expensive ride?<br />
Mean and SD of the sampling<br />
distribution of p^<br />
PROBLEM: Suppose that 26% of all<br />
high school students at a large school<br />
spend about half or more of their<br />
earnings on a car. A random sample<br />
of 150 students from this school is<br />
surveyed. Let p^ 5 the proportion of<br />
students in the sample who spend about<br />
half or more of their earnings on a car.<br />
(a) Calculate the mean and standard<br />
deviation of the sampling distribution of p^ .<br />
(b) Interpret the standard deviation<br />
from part (a).<br />
SOLUTION:<br />
(a) m p^ = 0.26 and<br />
0.26(1− 0.26)<br />
s p^ =<br />
= 0.036<br />
Å 150<br />
(b) In SRSs of size n 5 150, the sample<br />
proportion of students who spend about<br />
half or more of their earnings on a car<br />
will typically vary by about 0.036 from<br />
the true proportion of p 5 0.26.<br />
a<br />
e XAMPLe<br />
What proportion of students have a smartphone?<br />
Mean and SD of the sampling distribution of p^<br />
PROBLEM: Suppose that 43% of students at a large high school own a smartphone. As part of a<br />
schoolwide technology study, the principal surveys an SRS of n 5 100 students. Let p^ 5 the proportion<br />
of students in the sample who own a smartphone.<br />
(a) Calculate the mean and the standard deviation of the sampling distribution of p^ .<br />
(b) Interpret the standard deviation from part (a).<br />
SOLUTION:<br />
0.43(1 − 0.43)<br />
(a) m p^ = 0.43 and s p^ = = 0.050<br />
Å 100<br />
(b) In SRSs of size n 5 100, the sample proportion of students<br />
who own a smartphone will typically vary by about 0.050 from<br />
the true proportion of p 5 0.43.<br />
d<br />
ThinK ABoUT iT Is the sampling distribution of p^ (the sample proportion of<br />
successes) related to the sampling distribution of X (the sample count of successes)?<br />
Yes!<br />
number of successes in sample<br />
p^ = = X sample size<br />
n<br />
For example, here are dotplots showing the simulated sampling distribution of X 5 the<br />
number of pennies from the 2000s and p^ 5 the proportion of pennies from the 2000s<br />
for samples of size 20 in Mr. Ramirez’s class. The distributions are exactly the same,<br />
other than the scale on the axis.<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
6 8 10 12 14 16 18 20<br />
Number of pennies from 2000s<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d d<br />
d d<br />
d d d<br />
d<br />
In this context, n 5 100 and p 5 0.43. The mean is m p^ 5 p<br />
p(1 − p)<br />
and the standard deviation is s p^ 5 . Å n<br />
FOR PRACTICE TRY EXERCISE 1.<br />
d<br />
d d<br />
d d<br />
d d d<br />
d d d d<br />
d d d d d d<br />
d d d d d d<br />
d d d d d d d<br />
d d d d d d d d<br />
d d d d d d d d d d d<br />
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0<br />
Proportion of pennies from 2000s<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 426<br />
Common Error<br />
Students may refer to the sampling<br />
distribution of the sample proportion of<br />
pennies as a binomial distribution. Tell them<br />
this is not technically correct because a<br />
binomial random variable is a count, not a<br />
proportion. However, these two distributions<br />
are mathematical transformations of one<br />
another.<br />
18/08/16 5:01 PMStarnes_<strong>3e</strong>_CH0<br />
426<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 426<br />
11/01/17 3:56 PM
L E S S O N 6.4 • The Sampling Distribution of a Sample Proportion 427<br />
18/08/16 5:01 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 427<br />
Also, the formulas for the mean and standard deviation of the sampling distribution<br />
of p^ are derived from the formulas we learned in the previous lesson:<br />
Shape<br />
s p^ 5 s X<br />
n<br />
m p^ 5 m X<br />
n = np n = p<br />
"np(1 − p) np(1 − p) p(1 − p)<br />
= = =<br />
n Å n 2 Å n<br />
Both the sample size and the proportion of successes in the population affect the shape<br />
of the sampling distribution of the sample proportion p^ . The following activity helps<br />
you explore the effect of these two factors.<br />
AcT iviT y<br />
Sampling from the candy machine<br />
Imagine a very large candy machine filled with<br />
orange, brown, and yellow candies. When you insert<br />
money, the machine dispenses a sample of candies.<br />
In this activity, you will use an applet to investigate<br />
the sample-to-sample variability in the proportion of<br />
orange candies dispensed by the machine.<br />
1. Launch the Reese’s Pieces® applet at<br />
www.rossmanchance.com/applets. Click the<br />
button for “Proportion of orange” to have the<br />
applet calculate and record the value of p^ 5 the<br />
sample proportion of orange candies.<br />
2. Click on the “Draw Samples” button. An animated<br />
simple random sample of n 5 25 candies should<br />
be dispensed. The screen shot shows the results<br />
of one such sample. Was your sample proportion<br />
of orange candies close to the actual population<br />
proportion, p 5 0.50?<br />
Teaching Tip:<br />
Differentiate<br />
Students who have difficulty with algebra can<br />
ignore the mathematical derivations of m p^<br />
and s p^ at the top of this page. They are not<br />
important for understanding the big ideas in<br />
this and other lessons.<br />
3. Click “Draw Samples” 9 more times, so that you<br />
have a total of 10 sample proportions. Look at the<br />
dotplot of your p^ values. Does the distribution<br />
have a recognizable shape?<br />
4. To take many more samples quickly, enter 990<br />
in the “number of samples” box. Click on the<br />
“ Animate” box to turn off the animation. Then<br />
click “Draw Samples.” You have now taken a<br />
total of 1000 samples of n 5 25 candies from<br />
the machine. Describe the shape of the simulated<br />
sampling distribution of p^ shown in the<br />
dotplot.<br />
5. How does the shape of the sampling distribution<br />
of p^ change if the proportion of orange<br />
candies in the machine is p 5 0.10 instead of<br />
p 5 0.50? Set the probability of orange<br />
candies to p 5 0.10 and draw 1000 samples of<br />
size n 5 25. What if p 5 0.90? Describe how the<br />
value of p affects the shape of the sampling<br />
distribution of p^ .<br />
6. How does the shape of the sampling distribution<br />
of p^ change if the sample size increases?<br />
Set the probability of orange to p 5 0.90 and<br />
the number of candies to n 5 25 and draw<br />
1000 samples. Then, repeat with sample sizes of<br />
n 5 100 and n 5 500. Describe how the value of<br />
n affects the shape of the sampling distribution<br />
of p^ .<br />
Activity Overview<br />
Time: 15–18 minutes<br />
18/08/16 5:01 PM<br />
Materials: An Internet-connected device for<br />
each student or group of students<br />
Teaching Advice: This activity helps students<br />
understand the shape of the sampling<br />
distribution of p^ . Because the sampling<br />
distribution of p^ is closely related to the<br />
sampling distribution of the sample count X,<br />
this activity can be skipped if your students<br />
understood Lesson 6.3 well.<br />
If you don’t have enough devices, students<br />
can work in groups or you can demonstrate<br />
the applet to the entire class. Showing the<br />
applet as a demonstration also saves time,<br />
even if it doesn’t engage students as much.<br />
This applet gives a visual of the<br />
population distribution (the candy<br />
machine), the distribution of one<br />
sample (the candies in the dish), and the<br />
sampling distribution (the dotplot). Point<br />
out these three distributions to your<br />
students.<br />
Beginning at Step 4, emphasize the<br />
difference between “number of candies”<br />
(sample size) and “number of samples.”<br />
Students will often be confused about<br />
the meaning of these terms. Make sure<br />
students are using the correct values.<br />
There is nothing special about 1000<br />
samples. This activity would work just<br />
as well if students generated 10,000<br />
samples. What matters is to repeat the<br />
sampling process often enough to see<br />
the pattern in the sampling distribution.<br />
The shape of the distribution of p^<br />
follows the same rules as the shape of the<br />
distribution of the corresponding binomial<br />
random variable X. This should be evident<br />
from comparing the results of this activity<br />
to the results of the activity in Lesson 6.3.<br />
Answers:<br />
1. Students should launch the applet<br />
and click “Proportion of orange.”<br />
2. Student results will vary. Most students<br />
should get a sample proportion close to<br />
0.50, but some students may not.<br />
3. With only 10 dots on the dotplot,<br />
the distribution shouldn’t have a<br />
recognizable shape.<br />
4. The distribution should look moundshaped<br />
and roughly symmetric.<br />
5. When p 5 0.1, the shape of the<br />
distribution is skewed to the right.<br />
When p 5 0.9, it is skewed to the left.<br />
Values of p that are lower than 0.5<br />
result in right-skewed distributions<br />
and values higher than 0.5 result in<br />
left-skewed distributions.<br />
6. As n increases, the sampling<br />
distribution of p^ gets more<br />
approximately normal. This result<br />
holds true for all values of p.<br />
Teaching Tip<br />
The preceding activity helps students<br />
visualize sampling. This is a good time<br />
to remind them that the word “sample”<br />
does not refer to a single individual,<br />
but to a group of many individuals.<br />
Make sure your students don’t refer to<br />
individuals as “samples.”<br />
Lesson 6.4<br />
L E S S O N 6.4 • The Sampling Distribution of a Sample Proportion 427<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 427<br />
11/01/17 3:56 PM
428<br />
C H A P T E R 6 • Sampling Distributions<br />
As you learned in the activity, the shape of the sampling distribution of p^ will be<br />
closer to normal when the value of p is closer to 0.5 and the sample size is larger.<br />
These relationships are the same as those you discovered in the previous lesson. And<br />
the Large Counts condition is the same as well.<br />
DEFINITION the Large counts condition<br />
Suppose p^ is the proportion of successes in a random sample of size n from a population<br />
with proportion of successes p. The Large Counts condition says that the distribution of<br />
p^ will be approximately normal when<br />
np ≥ 10 and n(1 2 p) ≥ 10<br />
Alternate Example<br />
Cooking the books?<br />
Shape of the sampling distribution of p^<br />
PROBLEM: To audit one department<br />
of a large corporation, an accountant<br />
selects a random sample of n 5 80<br />
transactions. Historically, the true<br />
proportion of transactions that are more<br />
than $1000 is p 5 0.42. Would it be<br />
appropriate to use a normal distribution<br />
to model the sampling distribution of p^<br />
for samples of size n 5 80? Justify your<br />
answer.<br />
SOLUTION:<br />
Yes; because np 5 80(0.42) 5 33.6 ≥ 10<br />
and n(1 2 p) 5 80(1 2 0.42) 5 46.4 ≥ 10,<br />
the sampling distribution of p^ is approximately<br />
normal.<br />
a<br />
a<br />
e XAMPLe<br />
A penny for your thoughts?<br />
Shape of the sampling distribution of p^<br />
PROBLEM: Mr. Ramirez’s class did the Penny for Your<br />
Thoughts Activity from the beginning of this chapter.<br />
In his population of pennies, the proportion of pennies<br />
from the 2000s is p 5 0.627. Would it be appropriate<br />
to use a normal distribution to model the sampling<br />
distribution of p^ for samples of size n 5 16? Justify<br />
your answer.<br />
SOLUTION:<br />
No. Because n (1 2 p ) 5 16(1 2 0.627) 5 5.968 < 10,<br />
the sampling distribution of p^ is not approximately normal.<br />
e XAMPLe<br />
Finding Probabilities Involving p^<br />
When the Large Counts condition is met, we can use a normal distribution to calculate<br />
probabilities involving p^ 5 the proportion of successes in a random sample<br />
of size n.<br />
How far from home do you attend college?<br />
Normal calculations involving p^<br />
<strong>Ch</strong>eck the Large Counts condition to determine if p^ will<br />
have an approximately normal distribution.<br />
PROBLEM: A polling organization asks an SRS of 1500 first-year college students how far away<br />
their home is. Suppose that 35% of all first-year students attend college within 50 miles of home.<br />
Find the probability that the random sample of 1500 students will give a result within 2 percentage<br />
points of this true value.<br />
FOR PRACTICE TRY EXERCISE 5.<br />
Andrew Unangst/Getty Images<br />
Alternate Example<br />
College bound?<br />
Normal calculations involving p^<br />
PROBLEM: In a recent year, a fact sheet<br />
for the state of Pennsylvania stated that<br />
72.6% of all public high school seniors<br />
were planning to attend college the next<br />
year. If an SRS of 400 public high school<br />
seniors in Pennsylvania were surveyed,<br />
what is the probability that the sample<br />
proportion who are going to college next<br />
year will be within 3 percentage points<br />
of the true value?<br />
• Mean: m p^ 5 0.726<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 428<br />
0.726(1 − 0.726)<br />
• SD: s p^ =<br />
= 0.022<br />
Å 400<br />
• Shape: Approximately normal because<br />
np 5 400(0.726) 5 290.6 ≥ 10 and<br />
n(1 2 p) 5 400(1 2 0.726) 5 109.6 ≥ 10<br />
0.696<br />
0.756<br />
0.696 − 0.726<br />
Using Table A: z = = −1.36<br />
0.022<br />
0.756 − 0.726<br />
and z = = 1.36<br />
0.022<br />
P(21.36 ≤ Z ≤ 1.36) 5 0.9131 2 0.0869<br />
5 0.8262<br />
Using technology: Applet/normalcdf<br />
(lower:0.696, upper:0.756, mean:0.726,<br />
SD:0.022) 5 0.8273<br />
18/08/16 5:02 PMStarnes_<strong>3e</strong>_CH0<br />
SOLUTION:<br />
Let p^ 5 the proportion in the sample<br />
who are going to college next year,<br />
where p 5 0.726 and n 5 400.<br />
0.660 0.682 0.704 0.726 0.748 0.770 0.792<br />
Sample proportion who<br />
are going to college<br />
428<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 428<br />
11/01/17 3:56 PM
L E S S O N 6.4 • The Sampling Distribution of a Sample Proportion 429<br />
SOLUTION:<br />
• Mean: m P^ 5 0.35<br />
0.35(1 − 0.35)<br />
• SD: s p^ = = 0.0123<br />
Å 1500<br />
• Shape: Approximately normal because np 5 1500(0.35)<br />
5 525 ≥ 10 and n (12 p) 5 1500(12 0.35) 5 975 ≥ 10<br />
Let p^ 5 the proportion in the sample who attend college<br />
within 50 miles of home, where p 5 0.35 and n 5 1500.<br />
To use a normal distribution to calculate probabilities<br />
involving p^ , we have to know the mean, standard<br />
deviation, and shape of the sampling distribution of p^ .<br />
Recall that the mean is m p^ = p and the standard<br />
p(1 − p)<br />
deviation is s p^ = . Å n<br />
Teaching Tip<br />
In this example, if students wonder what<br />
to do if the shape isn’t approximately<br />
normal, the answer is that the<br />
approximate probability would have<br />
to be found by using a binomial<br />
distribution.<br />
Lesson 6.4<br />
1. Draw a normal distribution.<br />
0.3131 0.3254 0.3377 0.35 0.3623 0.3746 0.3869<br />
0.33 0.37<br />
Sample proportion who live within 50 miles<br />
Using Table A:<br />
0.33 − 0.35<br />
0.37 − 0.35<br />
z = = −1.63 and z = = 1.63<br />
0.0123<br />
0.0123<br />
P (21.63 ≤ Z ≤ 1.63) 5 0.94842 0.05165 0.8968<br />
Using technology: Applet/normalcdf (lower:0.33, upper:0.37,<br />
mean:0.35, SD: 0.0123) 5 0.8961<br />
2. Perform calculations.<br />
(i) Standardize the boundary value and use Table A to<br />
find the desired probability; or<br />
(ii) Use technology.<br />
FOR PRACTICE TRY EXERCISE 9.<br />
L e SSon APP 6. 4<br />
Lesson App<br />
Answers<br />
18/08/16 5:02 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 429<br />
What’s that spot on my potato chip?<br />
A potato-chip producer and its main supplier agree<br />
that each shipment of potatoes must meet certain<br />
quality standards. If the producer finds convincing<br />
evidence that more than 8% of the potatoes in the<br />
entire shipment have “blemishes,” the truck will be<br />
sent away to get another load of potatoes from the<br />
supplier. Otherwise, the entire truckload will be used<br />
to make potato chips. To make the decision, a supervisor<br />
will inspect a random sample of potatoes from the<br />
shipment. Suppose that the proportion of blemished<br />
potatoes in the entire shipment is p 5 0.08 and that<br />
the supervisor randomly selects n 5 500 potatoes for<br />
inspection.<br />
Teaching Tip<br />
The supervisor has made a mistake about<br />
the proportion of all blemished potatoes<br />
in the truckload, because getting a sample<br />
proportion of 0.11 or higher if the true<br />
proportion in the truckload is 0.08 would<br />
be highly unlikely to happen just by chance.<br />
In <strong>Ch</strong>apter 8, we’ll learn that this is called a<br />
Type I error, or a false positive. Not rejecting<br />
a truckload with a proportion of blemished<br />
potatoes higher than 0.08 would be a Type II<br />
error, or false negative.<br />
1. Calculate the<br />
mean and<br />
standard<br />
deviation of the sampling distribution of p^ .<br />
Interpret the standard deviation.<br />
2. Justify that the sampling distribution of p^ is<br />
approximately normal.<br />
3. Calculate the probability that at least 11% of the<br />
potatoes in the sample are blemished.<br />
4. Based on your answer to Question 3, what should<br />
the supervisor conclude if he selects an SRS of size<br />
n 5 500 and finds p^ 5 0.11? Explain.<br />
Africa Studio/Shutterstock<br />
18/08/16 5:02 PM<br />
p(1− p)<br />
1. m p^ = p = 0.08; s p^ =<br />
Å n<br />
0.08(1− 0.08)<br />
=<br />
= 0.012<br />
Å 500<br />
In SRSs of size n 5 500, the sample<br />
proportion of blemished potatoes will<br />
typically vary by about 0.012 from the<br />
true proportion of p 5 0.08.<br />
2. Because np = (500)(0.08) = 40 ≥ 10<br />
and n(1 − p) = (500)(1 − 0.08) = 460<br />
$ 10, the sampling distribution of p^ is<br />
approximately normal.<br />
0.11 − 0.08<br />
3. z = = 2.5; P( p^ ≥ 0.11)<br />
0.012<br />
= P(Z ≥ 2.5) = 1 − 0.9938 = 0.0062<br />
Using technology: Applet/normalcdf<br />
(lower:0.11, upper:1000, mean:0.08,<br />
SD:0.012) 5 0.0062<br />
4. Send the shipment back. Assuming<br />
the true proportion of blemished<br />
potatoes is 0.08, there is only a 0.62%<br />
chance of getting a sample proportion of<br />
0.11 or higher purely by chance. Because<br />
this result is unlikely (less than 5%), we<br />
have convincing evidence that more<br />
than 8% of the potatoes in this shipment<br />
have blemishes.<br />
L E S S O N 6.4 • The Sampling Distribution of a Sample Proportion 429<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 429<br />
11/01/17 3:56 PM
430<br />
C H A P T E R 6 • Sampling Distributions<br />
TRM chapter 6 Activity:<br />
Sampling Movies<br />
This activity reviews the sampling<br />
distribution of p^ by sampling from a<br />
population of movies. Access this resource<br />
by clicking on the link in the TE-book,<br />
logging into the Teacher’s Resource site, or<br />
accessing this resource on the TRFD.<br />
TRM full Solutions to Lesson 6.4<br />
Exercises<br />
You can find the full solutions for this<br />
lesson by clicking on the link in the TEbook,<br />
logging into the Teacher’s Resource<br />
site, or accessing this resource on the TRFD.<br />
Answers to Lesson 6.4 Exercises<br />
p(1− p)<br />
1. (a) m p^ = p = 0.20; s p^ =<br />
Å n<br />
0.20(1− 0.20)<br />
=<br />
= 0.089<br />
Å 20<br />
(b) In SRSs of size n 5 20, the sample<br />
proportion of orange Skittles® will<br />
typically vary by about 0.089 from the<br />
true proportion of p 5 0.20.<br />
p(1 − p)<br />
2. (a) m p^ = p = 0.55; s p^ =<br />
Å n<br />
0.55(1 − 0.55)<br />
= = 0.022<br />
Å 500<br />
(b) In SRSs of size n 5 500, the sample<br />
proportion of Democrats will typically<br />
vary by about 0.022 from the true<br />
proportion of p 5 0.55.<br />
p(1 − p)<br />
3. (a) m p^ = p = 0.90; s p^ =<br />
Å n<br />
0.90(1 − 0.90)<br />
= = 0.03<br />
Å 100<br />
(b) In SRSs of size n 5 100, the sample<br />
proportion of orders shipped within three<br />
working days will typically vary by about<br />
0.03 from the true proportion of p 5 0.90.<br />
p(1 − p)<br />
4. (a) m p^ = p = 0.59; s p^ =<br />
Å n<br />
0.59(1 − 0.59)<br />
= = 0.07<br />
Å 50<br />
(b) In SRSs of size n 5 50, the sample<br />
proportion of couples in which both<br />
parents work outside the home will<br />
typically vary by about 0.07 from the true<br />
proportion of p 5 0.59.<br />
5. No; because np = (10)a 16<br />
120 b =1.33<br />
< 10, the sampling distribution of p^ is<br />
not approximately normal.<br />
6. No; because np = (7)a 42<br />
100 b = 2.94<br />
< 10, the sampling distribution of p^ is<br />
not approximately normal.<br />
430<br />
Exercises<br />
C H A P T E R 6 • Sampling Distributions<br />
Lesson 6.4<br />
WhAT DiD y o U LeA rn?<br />
LEARNINg TARgET EXAMPLES EXERCISES<br />
Calculate the mean and standard deviation of the sampling<br />
distribution of a sample proportion p^ and interpret the standard<br />
deviation.<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 430<br />
7. Yes; because np = (100)(0.90) = 90 ≥ 10<br />
and n(1 − p) = (100)(1 − 0.90) = 10 ≥ 10,<br />
the sampling distribution of p^ is approximately<br />
normal.<br />
8. Yes; because np = (50)(0.59) = 29.5 ≥ 10<br />
and n(1− p) = (50)(1− 0.59) = 20.5 ≥ 10,<br />
the sampling distribution of p^ is approximately<br />
normal.<br />
p(1 − p)<br />
9. Mean: m p^ = p = 0.70; SD: s p^ =<br />
Å n<br />
0.70(1 − 0.70)<br />
= = 0.028<br />
Å 267<br />
Shape: Approximately normal because<br />
np = 267(0.70) = 186.9 ≥ 10 and<br />
n(1 − p) = 267(1 − 0.70) = 80.1 ≥ 10<br />
0.75 − 0.70<br />
z = = 1.79; P( p^ ≥ 0.75)<br />
0.028<br />
= P(Z ≥ 1.79) = 1 − 0.9633 = 0.0367<br />
p. 426 1–4<br />
Determine if the sampling distribution of p^ is approximately normal. p. 428 5–8<br />
If appropriate, use a normal distribution to calculate probabilities<br />
involving p^ .<br />
Mastering Concepts and Skills<br />
1. Orange Skittles ® The makers of Skittles claim that<br />
20% of Skittles candies are orange. You select a<br />
random sample of 20 Skittles from a large bag. Let<br />
p^ 5 the proportion of orange Skittles in the sample.<br />
(a) Calculate the mean and the standard deviation of<br />
the sampling distribution of p^ .<br />
(b) Interpret the standard deviation from part (a).<br />
2. Registered voters In a congressional district, 55%<br />
of registered voters are Democrats. A polling<br />
organization selects a random sample of 500 registered<br />
voters from this district. Let p^ 5 the proportion<br />
of Democrats in the sample.<br />
(a) Calculate the mean and the standard deviation of<br />
the sampling distribution of p^ .<br />
(b) Interpret the standard deviation from part (a).<br />
3. On-time shipping A large mail-order company<br />
advertises that it ships 90% of its orders within<br />
3 working days. You select an SRS of 100 of the<br />
orders received in the past week for an audit. Let<br />
p^ 5 the proportion of orders in the last week that<br />
were shipped within 3 working days.<br />
(a) Calculate the mean and the standard deviation of<br />
the sampling distribution of p^ .<br />
(b) Interpret the standard deviation from part (a).<br />
4. Married with children According to a recent U.S.<br />
Bureau of Labor Statistics report, the proportion of<br />
married couples with children in which both parents<br />
work outside the home is 59%. 12 You select<br />
an SRS of 50 married couples with children and<br />
let p^ 5 the sample proportion of couples in which<br />
both parents work outside the home.<br />
(a) Calculate the mean and the standard deviation of<br />
the sampling distribution of p^ .<br />
(b) Interpret the standard deviation from part (a).<br />
pg 426<br />
Lesson 6.4<br />
p. 428 9–12<br />
5. Airport safety The Transportation Security Administration<br />
(TSA) is responsible for airport safety. On<br />
pg 428<br />
some flights, TSA officers randomly select passengers<br />
for an extra security check before boarding. One<br />
such flight had 120 passengers—16 in first class and<br />
104 in coach class. TSA officers selected an SRS of<br />
10 passengers for screening. Would it be appropriate<br />
to use a normal distribution to model the sampling<br />
distribution of p^ 5 the proportion of first-class passengers<br />
in the sample? Justify your answer.<br />
6. Only vowels? In the game of Scrabble, each player<br />
begins by drawing 7 tiles from a bag containing<br />
100 tiles. There are 42 vowels, 56 consonants, and<br />
2 blank tiles in the bag. Cait chooses an SRS of<br />
7 tiles. Would it be appropriate to use a normal<br />
distribution to model the sampling distribution of<br />
p^ 5 the proportion of vowels in her sample? Justify<br />
your answer.<br />
7. Model shipping? Refer to Exercise 3. Would it be<br />
appropriate to use a normal distribution to model<br />
the sampling distribution of p^ 5 the proportion of<br />
orders in the last week that were shipped within 3<br />
working days? Justify your answer.<br />
8. A model marriage? Refer to Exercise 4. Would<br />
it be appropriate to use a normal distribution to<br />
model the sampling distribution of p^ 5 the sample<br />
proportion of couples in which both parents work<br />
outside the home? Justify your answer.<br />
9. Women on diets Suppose that 70% of all college<br />
women have been on a diet within the past 12<br />
pg 428<br />
months. A sample survey interviews an SRS of 267<br />
college women. Find the probability that 75% or<br />
more of the women in the sample have been on a diet.<br />
10. Percentage of Harleys Harley-Davidson motorcycles<br />
make up 14% of all the motorcycles registered<br />
in the United States. You plan to interview an SRS<br />
Using technology: Applet/normalcdf(lower:0.75,<br />
upper:1000, mean:0.70, SD:0.028) 5 0.0371<br />
p(1 − p)<br />
10. Mean: m p^ = p = 0.14; SD: s p^ =<br />
Å n<br />
0.14(1 − 0.14)<br />
= = 0.0155<br />
Å 500<br />
Shape: Approximately normal because np =<br />
(500)(0.14) = 70 ≥ 10 and n(1− p) = 500<br />
(1− 0.14) = 430 ≥ 10<br />
0.20 − 0.14<br />
z = = 3.87;<br />
0.0155<br />
P( p^ ≥ 0.20) = P( Z ≥ 3.87) ≈ 0<br />
Using technology: Applet/normalcdf(lower:0.20,<br />
upper:1000, mean:0.14, SD:0.0155) 5 0.0001<br />
18/08/16 5:02 PMStarnes_<strong>3e</strong>_CH0<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 430<br />
11/01/17 3:56 PM
L E S S O N 6.4 • The Sampling Distribution of a Sample Proportion 431<br />
18/08/16 5:02 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 431<br />
of 500 motorcycle owners. Find the probability<br />
that 20% or more of the motorcycle owners in the<br />
sample own Harleys.<br />
11. Success on Kickstarter The fundraising site Kickstarter<br />
regularly tracks the success rate of projects<br />
that seek funding on its site. Recently, the percentage<br />
of projects that were successfully funded was<br />
38.7%. 13 You select a random sample of 50 Kickstarter<br />
projects. What is the probability that less<br />
than 30% of them were successfully funded?<br />
12. Parlez-vous français? Quebec is the only province in<br />
Canada where the one official language is French.<br />
According to a recent census, 79.7% of Quebec<br />
residents identify French as their mother tongue.<br />
You select an SRS of 165 Quebec residents. What is<br />
the probability that less than 80% of them identify<br />
French as their mother tongue?<br />
Applying the Concepts<br />
13. Drinking the cereal milk? A USA Today poll asked<br />
a random sample of 1012 U.S. adults what they do<br />
with the milk in their cereal bowl after they have<br />
eaten. Let p^ be the proportion of people in the sample<br />
who drink the cereal milk. A spokesman for the<br />
dairy industry claims that 70% of all U.S. adults<br />
drink the cereal milk. Suppose this claim is true.<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of p^ . Interpret the standard<br />
deviation.<br />
(b) Justify that the sampling distribution of p^ is<br />
approximately normal.<br />
(c) Calculate the probability that at most 67% of the<br />
people in the sample drink the cereal milk.<br />
14. Who goes to church? A Gallup poll asked a random<br />
sample of 1785 adults if they attended church during<br />
the past week. Let p^ be the proportion of people<br />
in the sample who attended church. A newspaper<br />
report claims that 40% of all U.S. adults went to<br />
church last week. Suppose this claim is true.<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of p^ . Interpret the standard<br />
deviation.<br />
(b) Justify that the sampling distribution of p^ is<br />
approximately normal.<br />
(c) Calculate the probability that at least 44% of the<br />
people in the sample attended church.<br />
15. Who drinks the cereal milk? Refer to Exercise 13. Of<br />
the poll respondents, 67% said that they drink the<br />
cereal milk. Based on your answer to part (c), does<br />
this poll give convincing evidence that less than 70%<br />
of all U.S. adults drink the cereal milk? Explain.<br />
16. Do you go to church? Refer to Exercise 14. Of the<br />
poll respondents, 44% said they attended church<br />
last week. Based on your answer to part (c), does<br />
p(1− p)<br />
11. Mean: m p^ = p = 0.387; SD: s p^ =<br />
Å n<br />
0.387(1 − 0.387)<br />
= = 0.069<br />
Å 50<br />
Shape: Approximately normal because<br />
np = (50)(0.387) = 19.35 ≥ 10 and<br />
n(1 − p) = 50(1 − 0.387) = 30.65 ≥ 10<br />
0.30 − 0.387<br />
z = = −1.26; P( p^ < 0.30)<br />
0.069<br />
= P(Z < −1.26) = 0.1038<br />
Using technology: Applet/normalcdf(lower:<br />
−1000, upper:0.30, mean:0.387, SD:0.069)<br />
5 0.1037<br />
p(1 − p)<br />
12. Mean: m p^ = p = 0.797; SD: s p^ =<br />
Å n<br />
0.797(1 − 0.797)<br />
= = 0.031<br />
Å 165<br />
this poll give convincing evidence that more than<br />
40% of all U.S. adults attended church last week?<br />
Explain.<br />
Extending the Concepts<br />
17. More milk drinkers Refer to Exercise 13. What<br />
sample size would be required to reduce the standard<br />
deviation of the sampling distribution to onehalf<br />
the value you found in part (a)? Justify your<br />
answer.<br />
18. Off to college The example on page 428 used the<br />
sampling distribution of the sample proportion p^<br />
to find the probability that the random sample of<br />
1500 students from a population where p 5 0.35<br />
will give a p^ between 0.33 and 0.37. You can also<br />
find this probability using the sampling distribution<br />
of the sample count X, where X 5 the number<br />
of students in the sample who attend college within<br />
50 miles of their home.<br />
(a) Find the mean and standard deviation of the sampling<br />
distribution of the sample count X.<br />
(b) Justify that X has an approximately normal<br />
distribution.<br />
(c) Find the values of X that would result in sample<br />
proportions of p^ 5 0.33 and p^ 5 0.37.<br />
(d) Calculate the probability that X is between the two<br />
values from part (c).<br />
Recycle and Review<br />
19. Waiting with intent (1.8, 3.9) Do drivers take longer<br />
to leave their parking spaces when another car<br />
is waiting? Researchers hung out in a parking lot<br />
and collected some data. The graphs and numerical<br />
summaries display information about how long it<br />
took drivers to exit their spaces.<br />
Someone waiting?<br />
Yes<br />
No<br />
30 40 50 60 70 80 90<br />
Time (sec)<br />
Descriptive Statistics: Time<br />
*<br />
*<br />
Waiting N Mean StDev Min Q 1 Median Q 3 Max<br />
No 20 44.42 14.10 33.76 35.61 39.56 48.48 84.92<br />
Yes 20 54.11 14.39 41.61 43.41 47.14 66.44 85.97<br />
Shape: Approximately normal because<br />
np = (165)(0.797) = 131.5 ≥ 10 and<br />
n(1 − p) = 165(1 − 0.797) = 33.5 ≥ 10<br />
0.80 − 0.797<br />
z = = 0.10;<br />
0.031<br />
P(p^ < 0.80) = P(Z < 0.10) = 0.5398<br />
Using technology: Applet/normalcdf(lower:<br />
−1000, upper:0.80, mean:0.797, SD:0.031)<br />
5 0.5385<br />
p(1 − p)<br />
13. (a) m p^ = p = 0.70; s p^ =<br />
Å n<br />
0.70(1 − 0.70)<br />
= = 0.014<br />
Å 1012<br />
In SRSs of size n 5 1012, the sample<br />
proportion of people who drink the cereal<br />
milk will typically vary by about 0.014 from<br />
the true proportion of p 5 0.70.<br />
18/08/16 5:02 PM<br />
(b) Because np = (1012)(0.70) = 708.4<br />
≥ 10 and n(1 − p) = (1012)(1 − 0.70)<br />
= 303.6 ≥ 10, the sampling distribution<br />
of p^ is approximately normal.<br />
0.67 − 0.70<br />
(c) z = =−2.14; P( p^ ≤ 0.67)<br />
0.014<br />
= P( Z ≤ −2.14) = 0.0162<br />
Using technology: Applet/normalcdf<br />
(lower:−1000, upper:0.67, mean:0.70, SD:<br />
0.014) 5 0.0161<br />
p(1 − p)<br />
14. (a) m p^ = p = 0.40; s p^ =<br />
Å n<br />
0.40(1 − 0.40)<br />
=<br />
= 0.012<br />
Å 1785<br />
In SRSs of size n 5 1785, the sample proportion<br />
of people who attended church<br />
will typically vary by about 0.012 from<br />
the true proportion of p 5 0.40.<br />
(b) Because np = (1785)(0.40) = 714 $<br />
10 and n(1− p) = (1785)(1 − 0.40) =<br />
1071 ≥ 10, the sampling distribution of<br />
p^ is approximately normal.<br />
0.44 − 0.40<br />
(c) z = = 3.33; P( p^ ≥ 0.44)<br />
0.012<br />
= P( Z ≥ 3.33) = 1 − 0.9996 = 0.0004<br />
Using technology: Applet/normalcdf<br />
(lower:0.44, upper:1000, mean:0.40,<br />
SD:0.012) 5 0.0004<br />
15. Yes; assuming the true proportion<br />
of people who drink the cereal milk is<br />
0.70, there is only about a 2% chance<br />
of getting a sample proportion of 0.67<br />
or lower purely by chance. Because this<br />
result is unlikely (less than 5%), we have<br />
convincing evidence that less than 70% of<br />
all U.S. adults drink the cereal milk.<br />
16. Yes; assuming the true proportion<br />
of people who attended church is 0.40,<br />
there is about a 0.0004% chance of getting<br />
a sample proportion of 0.44 or higher<br />
purely by chance. Because this result is<br />
unlikely (less than 5%), we have convincing<br />
evidence that more than 40% of all U.S.<br />
adults attended church last week.<br />
17. Because the standard deviation is<br />
found by dividing by !n, using 4n for the<br />
sample size halves the standard deviation<br />
¢"4n = 2"n≤ ; we would have to<br />
sample 1012(4) = 4048 adults.<br />
18. (a) m X = np = 1500(0.35) = 525;<br />
s X = "np(1−p) ="1500(0.35)(1−0.35)<br />
= 18.47<br />
(b) Because np = 1500(0.35) = 525 ≥ 10<br />
and n(1− p) = 1500(1− 0.35)= 975 $ 10,<br />
the sampling distribution of X is approximately<br />
normal.<br />
Answers 18(c,d)–19 are on page 432<br />
Lesson 6.4<br />
L E S S O N 6.4 • The Sampling Distribution of a Sample Proportion 431<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 431<br />
11/01/17 3:56 PM
Answers continued<br />
X<br />
18. (c) p^ =<br />
n , so X = p^ n; (0.33)(1500) =<br />
495 and (0.37)(1500) = 555<br />
495 − 525<br />
(d) z = = −1.62;<br />
18.47<br />
555 − 525<br />
z = = 1.62<br />
18.47<br />
P(495 # X # 555)= P(−1.62 ≤ Z # 1.62)<br />
= 0.9474 − 0.0526 = 0.8948<br />
Using technology: Applet/normalcdf(lower:<br />
495, upper:555, mean:525, SD:18.47)<br />
5 0.8957. This answer matches the answer<br />
from the example that uses the sampling<br />
distribution of p^ , except for rounding.<br />
19. (a) Shape: Both distributions are<br />
skewed to the right. Center: Drivers<br />
generally take longer to leave when<br />
someone is waiting for the space. Spread:<br />
There is more variability for the drivers<br />
with someone waiting. Outliers: There<br />
were no outliers for those with someone<br />
waiting, but there were two high outliers<br />
for those with no one waiting.<br />
(b) Not necessarily; the researchers<br />
merely observed what was happening,<br />
and they did not randomly assign the<br />
treatments of either having a person<br />
waiting or not to the drivers of the cars<br />
leaving the lot.<br />
20. (a)<br />
Relative frequency<br />
1.05<br />
1<br />
0.95<br />
0.9<br />
0.85<br />
0.8<br />
0.75<br />
0.7<br />
0.65<br />
0.6<br />
0.55<br />
0.5<br />
0.45<br />
0.4<br />
0.35<br />
0.3<br />
0.25<br />
0.2<br />
0.15<br />
0.1<br />
0.05<br />
0<br />
Key<br />
Other<br />
Hazel<br />
Green<br />
Brown<br />
Blue<br />
432<br />
C H A P T E R 6 • Sampling Distributions<br />
(a) Write a few sentences comparing these distributions.<br />
(b) Can we conclude that a waiting car causes drivers to<br />
leave their spaces more slowly? Why or why not?<br />
20. Those baby blues (2.1, 4.4) The two-way table<br />
summarizes information about eye color and gender<br />
in a random sample of 200 high school students.<br />
Gender<br />
Male Female Total<br />
Blue 21 29 50<br />
Brown 35 40 75<br />
Eye color green 14 21 35<br />
Hazel 12 23 35<br />
Other 2 3 5<br />
Total 84 116 200<br />
(a) Is there an association between eye color and gender<br />
in this group of students? Support your answer<br />
with an appropriate graphical summary of the data.<br />
(b) Select a student at random. Are the events “Student<br />
is male” and “Student has blue eyes” independent?<br />
Justify your answer.<br />
Lesson 6.5<br />
The Sampling Distribution<br />
of a Sample Mean<br />
L e A r n i n g T A r g e T S<br />
d<br />
d<br />
Find the mean and standard deviation of the sampling distribution of a sample<br />
mean x and interpret the standard deviation.<br />
Use a normal distribution to calculate probabilities involving x when sampling<br />
from a normal population.<br />
When sample data are categorical, we often use the count or proportion of successes<br />
in the sample to make an inference about a population. When sample data are quantitative,<br />
we often use the sample mean x to estimate the mean m of a population. When<br />
we select random samples from a population, the value of x will vary from sample<br />
to sample. To understand how much x varies from m and what values of x are likely<br />
to happen by chance, we want to understand the sampling distribution of the sample<br />
mean x.<br />
Male<br />
Gender<br />
Female<br />
Yes, because the percentages for a<br />
given eye color are not the same for<br />
each gender. In other words, knowing<br />
a person’s gender helps us predict eye<br />
color. Males are more likely than females<br />
to have brown eyes, while females are<br />
more likely than males to have hazel or<br />
green eyes.<br />
(b) P(blue eyes | male) 5 21/84 5 0.25;<br />
P(blue eyes | female) 5 29/116 5 0.25.<br />
Because the probabilities are equal,<br />
the events “male” and “blue eyes” are<br />
independent. Knowing that a student<br />
is male does not change the probability<br />
that he has blue eyes.<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 432<br />
Learning Target Key<br />
The problems in the test bank are<br />
keyed to the learning targets using<br />
these numbers:<br />
d 6.5.1<br />
d 6.5.2<br />
BELL RINGER<br />
Thinking back to the “A penny for your<br />
thoughts?” activity for samples of<br />
5 pennies, did every sample produce the<br />
same sample mean year? What is the<br />
name for this phenomenon?<br />
Teaching Tip<br />
Be picky with students about using the correct<br />
symbols! The sample mean is denoted x and<br />
the population mean is denoted m.<br />
Common Error<br />
This lesson and the next are about means, not<br />
proportions. Make sure students don’t use the<br />
symbols p and p^ when working with means.<br />
18/08/16 5:02 PMStarnes_<strong>3e</strong>_CH0<br />
432<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 432<br />
11/01/17 3:56 PM
L E S S O N 6.5 • The Sampling Distribution of a Sample Mean 433<br />
DEFINITION Sampling distribution of the sample mean x<br />
The sampling distribution of the sample mean x describes the distribution of values taken<br />
by the sample mean x in all possible samples of the same size from the same population.<br />
When Mr. Ramirez’s class did the Penny for Your Thoughts activity at the beginning<br />
of the chapter, his students produced the “dotplot” in Figure 6.9 showing the simulated<br />
sampling distribution of x 5 the sample mean year of pennies in 50 samples of size n 5 5.<br />
x<br />
xx<br />
xxxxxx<br />
x<br />
x<br />
x<br />
1990 1995 2000 2005 2010 2015<br />
Sample mean year (n = 5)<br />
This distribution is slightly skewed to the left, with a mean of about 2002 and a<br />
standard deviation of about 5 years. By the end of Lesson 6.6, you should be able to<br />
anticipate the shape, center, and variability of distributions like this one without having<br />
to do a simulation.<br />
Center and Variability<br />
When we select random samples of size n from a population with mean m and standard<br />
deviation s, the value of x will vary from sample to sample. As with the sampling<br />
distribution of p^ , there are formulas that describe the center and variability of the<br />
sampling distribution of x.<br />
How to Calculate μ x and σ x<br />
Suppose that x is the mean of an SRS of size n drawn from a large population with mean m<br />
and standard deviation s. Then:<br />
• The mean of the sampling distribution of x is m x = m.<br />
• The standard deviation of the sampling distribution of x is s x = s "n .<br />
The behavior of x in repeated samples is much like that of the sample proportion p^ :<br />
• The sample mean x is an unbiased estimator of the population mean m. This is<br />
because the mean of the sampling distribution m x is equal to the mean of the<br />
population m.<br />
• The standard deviation of the sampling distribution of x describes the typical<br />
distance between the sample mean x and the population mean m.<br />
• The distribution of x is less variable for larger samples. This is indicated by the<br />
!n in the denominator of the standard deviation formula.<br />
• The formula for the standard deviation of the distribution of x requires that the<br />
observations be independent. In practice, we are safe assuming independence<br />
when we are sampling without replacement as long as the sample size is less<br />
than 10% of the population size.<br />
These facts about the mean and standard deviation of x are true no matter what shape<br />
the population distribution has.<br />
FigUre 6.9 Simulated<br />
sampling distribution of<br />
the sample mean year<br />
x in 50 samples of size<br />
n 5 5 from a population<br />
of pennies.<br />
Teaching Tip<br />
In Figure 6.9, ask students what the<br />
“dot” at 1997 represents. It is the sample<br />
mean/average year for one sample of<br />
5 pennies.<br />
Teaching Tip<br />
This figure is a good opportunity to refer<br />
to the dotplots made by your students in<br />
the “A penny for your thoughts?” activity<br />
from Lesson 6.1. Compare the results<br />
from your class with those from<br />
Mr. Ramirez’s class.<br />
FYI<br />
The formulas for m x and s x are true for<br />
the sampling distribution of x no matter<br />
what shape it has.<br />
Teaching Tip:<br />
Differentiate<br />
There are many symbols in this section,<br />
which may be difficult for some students.<br />
These students may find it easier to<br />
read the bullet points by substituting<br />
the words “sample mean” for x and<br />
“population mean” for m.<br />
Teaching Tip<br />
This is consistent with previous<br />
definitions of standard deviation as the<br />
typical distance a value falls from the<br />
mean of a distribution.<br />
Lesson 6.5<br />
18/08/16 5:02 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 433<br />
18/08/16 5:02 PM<br />
L E S S O N 6.5 • The Sampling Distribution of a Sample Mean 433<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 433<br />
11/01/17 3:56 PM
434<br />
C H A P T E R 6 • Sampling Distributions<br />
Alternate Example<br />
Thick hair?<br />
Mean and standard deviation of the<br />
sampling distribution of x<br />
PROBLEM: Suppose that the true mean<br />
number of hair follicles on a human head<br />
is 100,000 with a standard deviation of<br />
40,000 follicles. The mean number of hair<br />
follicles on the heads of 20 randomly<br />
selected humans will be computed.<br />
(a) Calculate the mean and standard<br />
deviation of the sampling distribution of x.<br />
(b) Interpret the standard deviation<br />
from part (a).<br />
SOLUTION:<br />
(a) m x = 100,000 follicles and<br />
s x = 40,000 = 8944 follicles<br />
"20<br />
(b) In SRSs of size n 5 20, the sample<br />
mean number of hair follicles will typically<br />
vary by about 8944 follicles from the<br />
population mean of 100,000 follicles.<br />
Activity Overview<br />
Time: 15–18 minutes<br />
Materials: An Internet-connected device<br />
for each student or group of students<br />
Teaching Advice: This activity helps<br />
students understand the shape of the<br />
sampling distribution of x. Although<br />
the applet doesn’t have high-resolution<br />
graphics, it is an excellent visual display<br />
of key concepts in this lesson.<br />
If you don’t have enough devices,<br />
students can work in groups or you<br />
can demonstrate the applet to the<br />
entire class. Showing the applet as a<br />
demonstration also saves time, although<br />
it doesn’t engage students as much.<br />
Even if students are doing the activity<br />
individually, it is helpful to show them<br />
the layout of the applet and demonstrate<br />
taking a few samples. Point out that<br />
sample size is denoted with a capital N,<br />
instead of the usual lowercase n.<br />
This applet gives a visual of the<br />
population distribution (the top/first<br />
number line), the distribution of one<br />
sample (the second number line), and<br />
the sampling distribution (the third and<br />
fourth number lines). Point out these three<br />
distributions. Note that this activity doesn’t<br />
make use of the fourth number line.<br />
There are two mysterious values<br />
reported by the applet: skew and kurtosis.<br />
Neither is important for this course.<br />
a<br />
e XAMPLe<br />
Seen any good movies lately?<br />
Mean and standard deviation of the sampling distribution of x<br />
PROBLEM: The number of movies viewed in the last year by students at a large high school has<br />
a mean of 19.3 movies with a standard deviation of 15.8 movies. Suppose we take an SRS of 100<br />
students from this school and calculate the mean number of movies viewed by the members of<br />
the sample.<br />
(a) Calculate the mean and standard deviation of the sampling distribution of x.<br />
(b) Interpret the standard deviation from part (a).<br />
SOLUTION:<br />
(a) m x = 19.3 movies and s x = 15.8 = 1.58 movies<br />
"100<br />
(b) In SRSs of size n 5 100, the sample mean number of<br />
movies will typically vary by about 1.58 movies from<br />
the population mean of 19.3 movies.<br />
AcT iviT y<br />
Shape<br />
Sampling from a normal population<br />
Professor David Lane of Rice University has<br />
developed a wonderful applet for investigating<br />
the sampling distribution of x. In this activity, you’ll<br />
use Professor Lane’s applet to explore the shape of<br />
the sampling distribution when the population is<br />
normally distributed.<br />
1. go to http://onlinestatbook.com/stat_sim/<br />
sampling_dist/ or search for “online statbook<br />
sampling distributions applet” and go to the website.<br />
When the BEgIN button appears on the left<br />
side of the screen, click on it. You will then see a<br />
yellow page entitled “Sampling Distributions” like<br />
the one in the screen shot.<br />
2. There are choices for the population distribution:<br />
normal, uniform, skewed, and custom. The<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 434<br />
Skewness is a measure of the skewness of the<br />
distribution; kurtosis measures how light or<br />
heavy the tails of the distribution are relative to a<br />
normal distribution.<br />
Answers:<br />
1. Students should launch the applet.<br />
2. The black boxes represent individuals<br />
being randomly selected from the<br />
population. The blue square represents the<br />
sample mean of the sample on the second<br />
number line.<br />
3.<br />
• The simulated sampling distribution has<br />
an approximately normal shape.<br />
Recall that m x = m and s x = s "n .<br />
FOR PRACTICE TRY EXERCISE 1.<br />
The shape of the sampling distribution of the sample mean x depends on the shape of<br />
the population distribution. In the following activity, you will explore what happens<br />
when sampling from a normal population.<br />
default is normal. Click the “Animated” button.<br />
What happens? Click the button several more<br />
times. What do the black boxes represent? What<br />
is the blue square that drops down onto the<br />
plot below?<br />
• The mean and median of the sampling<br />
distribution are 16, just like the population.<br />
(It is possible that some students will get<br />
values slightly different from 16.)<br />
• The standard deviation of the sampling<br />
distribution is smaller than the standard<br />
deviation of the population.<br />
4. The sampling distribution of x for n 5 20<br />
has the same shape and center, but the<br />
variability is even less than the sampling<br />
distribution for n 5 5.<br />
5. The shape of the sampling distribution is<br />
normal when the population distribution<br />
has a normal shape.<br />
18/08/16 5:03 PMStarnes_<strong>3e</strong>_CH0<br />
434<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 434<br />
11/01/17 3:57 PM
L E S S O N 6.5 • The Sampling Distribution of a Sample Mean 435<br />
3. Click on “Clear lower 3” to start clean. Then click<br />
on the “100,000” button under “Sample:” to simulate<br />
taking 100,000 SRSs of size n 5 5 from the<br />
population. Answer these questions:<br />
• Does the simulated sampling distribution<br />
(blue bars) have a recognizable shape? Click<br />
the box next to “Fit normal.”<br />
• To the left of each distribution is a set of<br />
summary statistics. Compare the mean of<br />
the simulated sampling distribution with the<br />
mean of the population.<br />
Describing a Sampling Distribution of a Sample Mean<br />
When Sampling from a Normal Population<br />
• How is the standard deviation of the simulated<br />
sampling distribution related to the<br />
standard deviation of the population?<br />
4. Click “Clear lower 3.” Use the drop-down menus<br />
to set up the bottom graph to display the mean<br />
for samples of size n 5 20. Then sample 100,000<br />
times. How do the two distributions of x compare:<br />
shape, center, and variability?<br />
5. What have you learned about the shape of the<br />
sampling distribution of x when the population<br />
has a normal shape?<br />
As the activity demonstrates, if the population distribution is normal, so is the<br />
sampling distribution of x. This is true no matter what the sample size is.<br />
Suppose that a population is normally distributed with mean m and standard deviation s. Then<br />
the sampling distribution of x for SRSs of size n has a normal distribution with mean m and<br />
standard deviation s "n.<br />
FYI<br />
The exact formula for the standard<br />
deviation of x when sampling without<br />
replacement is s x = s # N − n<br />
"n Å N − 1 .<br />
When n is small relative to N (less than<br />
N − n<br />
10%), the value of Å N − 1 is<br />
approximately 1 and therefore doesn’t<br />
s<br />
change the value of much. The<br />
"n<br />
N − n<br />
factor is sometimes called the<br />
Å N − 1<br />
finite population correction factor.<br />
Lesson 6.5<br />
18/08/16 5:03 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 435<br />
In the next lesson, you will learn what happens when sampling from a non-normal<br />
population.<br />
Finding Probabilities Involving x<br />
Now we have enough information to calculate probabilities involving x when the<br />
population distribution is normal.<br />
Are those peanuts underweight?<br />
Probabilities involving x<br />
PROBLEM: At the P. Nutty Peanut Company, dry-roasted, shelled peanuts<br />
are placed in jars labeled “16 ounces” by a machine. The distribution of<br />
weights in the jars is approximately normal with a mean of 16.1 ounces and<br />
a standard deviation of 0.15 ounces. Find the probability that the mean<br />
weight of 10 randomly selected jars is less than the advertised weight of<br />
16 ounces.<br />
e XAMPLe<br />
SOLUTION:<br />
Let x 5 the sample mean weight of 10 randomly<br />
• Mean: m x - 5 16.1 ounces<br />
selected jars. To find P(x ≤ 16), we have to know the<br />
• SD: s x - = 0.15<br />
mean, standard deviation, and shape of the sampling<br />
"10 = 0.047 ounces distribution of x. Recall that m x = m and s x = s "n .<br />
<strong>Ch</strong>arles Nesbit/Getty<br />
Images<br />
18/08/16 5:03 PM<br />
Alternate Example<br />
How hungry are the hounds?<br />
Probabilities involving x<br />
PROBLEM: The local SPCA (Society for the<br />
Prevention of Cruelty to Animals) feeds the<br />
animals it shelters, but the amount of food<br />
needed per day varies. The distribution of<br />
the weight of dry dog food used per day<br />
is approximately normal, with a mean of<br />
30 pounds and standard deviation of 5.1<br />
pounds. Find the probability that the mean<br />
weight of dry dog food for 6 randomly<br />
selected days is greater than 33 pounds.<br />
SOLUTION:<br />
Let x 5 the sample mean weight of dog<br />
food on 6 randomly selected days.<br />
• Mean: m x = 30 pounds<br />
• SD: s x = 5.1 = 2.08 pounds<br />
!6<br />
• Shape: Approximately normal<br />
because the population distribution is<br />
approximately normal<br />
33<br />
23.76 25.84 27.92 30.00 32.08 34.16 36.24<br />
Sample mean weight of dry dog food<br />
33 − 30<br />
Using Table A: z = = 1.44 and<br />
2.08<br />
P(Z ≥ 1.44) 5 1 2 0.9251 5 0.0749<br />
Using technology: Applet/normalcdf(lower:<br />
33, upper: 10000, mean: 30, SD: 2.08) 5<br />
0.0746<br />
L E S S O N 6.5 • The Sampling Distribution of a Sample Mean 435<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 435<br />
11/01/17 3:57 PM
436<br />
C H A P T E R 6 • Sampling Distributions<br />
• Shape: Approximately normal because the population<br />
distribution is approximately normal<br />
1. Draw a normal curve.<br />
Teaching Tip<br />
After the “Are those peanuts<br />
underweight?” example, you can have<br />
students calculate the probability that<br />
the weight of a single peanut X is less<br />
than 16 grams. Using technology,<br />
P(X < 16) 5 0.2525. This should make<br />
intuitive sense to students because the<br />
weight of a single peanut is more likely<br />
to be far from the true mean than the<br />
average weight of 10 peanuts (which<br />
is likely to include a mix of light and<br />
heavy peanuts). These probabilities are<br />
illustrated in Figure 6.10.<br />
Lesson App<br />
Answers<br />
1. m x = m = 64.5 inches;<br />
s x = s !n = 2.5 = 0.645 inches<br />
!15<br />
2. In SRSs of size n 5 15, the sample<br />
mean heights of young women will<br />
typically vary by about 0.645 inch from<br />
the true mean of 64.5 inches.<br />
66.5 − 64.5<br />
3. z = = 3.10;<br />
0.645<br />
P( x > 66.5) = P(Z > 3.10) = 1 − 0.9990<br />
= 0.0010<br />
Using technology: Applet/normalcdf<br />
(lower:66.5, upper:1000, mean:64.5,<br />
SD:0.645) 5 0.001<br />
4. Assuming the true mean height of<br />
women at this college is 64.5 inches,<br />
there is about a 0.1% chance of selecting<br />
an SRS of 15 women and getting a<br />
sample mean of 66.5 or higher purely<br />
by chance. Because this result is unlikely<br />
(less than 5%), we have convincing<br />
evidence that the mean height for all<br />
young women at this college is greater<br />
than 64.5 inches.<br />
TRM sAmpling Distribution Summary<br />
<strong>Ch</strong>art<br />
15.959 16.006 16.053 16.1 16.147 16.194 16.241<br />
16<br />
Sample mean weight x<br />
–<br />
(oz)<br />
Using Table A:<br />
16 − 16.1<br />
z = =−2.13<br />
0.047<br />
P (Z ≤ 22.13) 5 0.0166<br />
Using technology: Applet/normalcdf(lower: 21000, upper: 16,<br />
mean: 16.1, SD: 0.047) 5 0.0167<br />
FigUre 6.10 Normal<br />
curves showing the<br />
distribution of weights<br />
for individual jars of peanuts<br />
(purple curve) and<br />
distribution of sample<br />
mean weights for SRSs of<br />
10 jars (blue curve).<br />
L e SSon APP 6. 5<br />
Are college women taller?<br />
A helpful summary chart is available in the<br />
Teacher’s Resource Materials. This chart helps<br />
students organize the sampling distributions<br />
for sample counts (Lesson 6.3), sample<br />
proportions (Lesson 6.4), and sample means<br />
(Lesson 6.5). Access this resource by clicking<br />
on the link in the TE-book, logging into the<br />
Teacher’s Resource site, or accessing it on the<br />
TRFD.<br />
Individual values vary more than averages, so randomly selecting a single jar that<br />
is under the advertised weight is more likely than getting a sample mean for 10 jars<br />
that is less than the advertised weight. This is illustrated in Figure 6.10.<br />
Population distribution<br />
2. Perform calculations.<br />
(i) Standardize the boundary value and use Table A to<br />
find the desired probability; or<br />
(ii) Use technology.<br />
FOR PRACTICE TRY EXERCISE 5.<br />
Sampling distribution of x –<br />
16 16.1<br />
The fact that averages of several observations are less variable than individual observations<br />
is important in many settings. For example, it is common practice in science and<br />
medicine to repeat a measurement several times and report the average of the results.<br />
The heights of young women follow a normal distribution with mean m 5 64.5<br />
inches and standard deviation s 5 2.5 inches.<br />
1. Calculate the mean and standard deviation of the sampling distribution of x<br />
for SRSs of size 15.<br />
2. Interpret the standard deviation from Question 1.<br />
3. Find the probability that the mean height of an SRS of 15 young women<br />
exceeds 66.5 inches.<br />
4. Suppose that the mean height in a sample of n 5 15 young women from a local college is x 5 66.5. Based on your<br />
answer to Question 3, what would you conclude about the mean height for all young women at this college?<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 436<br />
© Dave & Les Jacobs/Blend Images/Corbis<br />
18/08/16 5:03 PMStarnes_<strong>3e</strong>_CH0<br />
436<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 436<br />
11/01/17 3:57 PM
L E S S O N 6.5 • The Sampling Distribution of a Sample Mean 437<br />
Lesson 6.5<br />
18/08/16 5:03 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 437<br />
WhAT DiD y o U LeA rn?<br />
LEARNINg TARgET EXAMPLES EXERCISES<br />
Find the mean and standard deviation of the sampling distribution of<br />
a sample mean x and interpret the standard deviation.<br />
Use a normal distribution to calculate probabilities involving x when<br />
sampling from a normal population.<br />
Exercises<br />
Mastering Concepts and Skills<br />
1. Short songs? David’s iPod has about 10,000 songs.<br />
The distribution of the play times for these songs is<br />
heavily skewed to the right with a mean of 225 seconds<br />
and a standard deviation of 60 seconds. Suppose<br />
we choose an SRS of 10 songs from this population<br />
and calculate the mean play time x of these songs.<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of x.<br />
(b) Interpret the standard deviation from part (a).<br />
2. Grinding auto parts A grinding machine in an autoparts<br />
factory prepares axles with a target diameter<br />
of m 5 40.125 millimeters (mm). The machine has<br />
some variability, so the standard deviation of the<br />
diameters is s 5 0.002 millimeter. The machine<br />
operator inspects a random sample of 4 axles each<br />
hour for quality-control purposes and records the<br />
sample mean diameter x.<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of x.<br />
(b) Interpret the standard deviation from part (a).<br />
pg 434<br />
Lesson 6.5<br />
3. Fresh tomatoes A local garden center says that a<br />
certain variety of tomato plant produces tomatoes<br />
with a mean weight of 250 grams and a standard<br />
deviation of 42 grams. You take a random sample<br />
of 20 tomatoes produced by these plants and calculate<br />
their mean weight x.<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of x.<br />
(b) Interpret the standard deviation from part (a).<br />
4. Fuel efficiency Driving styles differ, so there is variability<br />
in the fuel efficiency of the same model automobile<br />
driven by different people. For a certain car<br />
model, the mean fuel efficiency is 23.6 miles per<br />
gallon with a standard deviation of 2.5 miles per<br />
gallon. 14 Take a simple random sample of 25 owners<br />
of this model and calculate the sample mean<br />
fuel efficiency x.<br />
TRM full Solutions to Lesson 6.5<br />
Exercises<br />
You can find the full solutions for this lesson<br />
by clicking on the link in the TE-book, logging<br />
into the Teacher’s Resource site, or accessing<br />
this resource on the TRFD.<br />
Answers to Lesson 6.5 Exercises<br />
1. (a) m x = m = 225 seconds;<br />
s x = s !n = 60 = 18.97 seconds<br />
!10<br />
(b) In SRSs of size n 5 10, the sample mean play<br />
time of songs will typically vary by about 18.97<br />
seconds from the true mean of 225 seconds.<br />
p. 434 1–4<br />
p. 435 5–8<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of x.<br />
(b) Interpret the standard deviation from part (a).<br />
5. How much cereal? A company’s cereal boxes advertise<br />
pg 435 that 9.65 ounces of cereal are contained in each box.<br />
In fact, the amount of cereal in a randomly selected<br />
box follows a normal distribution with mean m 5 9.70<br />
ounces and standard deviation s 5 0.03 ounce. What<br />
is the probability that the mean amount of cereal x in<br />
5 randomly selected boxes is at most 9.65?<br />
6. Finch beaks One dimension of bird beaks is<br />
“depth”—the height of the beak where it arises<br />
from the bird’s head. During a research study on<br />
one island in the Galapagos archipelago, the beak<br />
depth of all Medium Ground Finches on the island<br />
was found to be normally distributed with mean<br />
m 5 9.5 millimeters and standard deviation s 5 1.0<br />
millimeter. 15 What is the probability that the mean<br />
depth x in 10 randomly selected Medium Ground<br />
Finches is at least 10 millimeters?<br />
7. Estimating cholesterol Suppose that the blood cholesterol<br />
level of all men aged 20 to 34 follows a<br />
normal distribution with mean m 5 188 milligrams<br />
per deciliter (mg/dl) and standard deviation s 5 41<br />
mg/dl. In an SRS of size 100, find the probability<br />
that x estimates m to within ±3 mg/dl.<br />
8. Bottlers at work A company uses a machine to fill<br />
plastic bottles with cola. The contents of the bottles<br />
vary according to a normal distribution with mean<br />
m 5 298 milliliters and standard deviation s 5 3<br />
milliliters. In an SRS of size 16, find the probability<br />
that x estimates m to within ±1 milliliter.<br />
Applying the Concepts<br />
9. Why won’t the car start? An automaker has found<br />
that the lifetime of its batteries varies from car to<br />
car according to a normal distribution with mean<br />
m 5 48 months and standard deviation s 5 8.2<br />
months. The company installs a new battery on an<br />
SRS of 8 cars.<br />
2. (a) m x = m = 40.125 mm;<br />
s x = s !n = 0.002 = 0.001 mm<br />
!4<br />
18/08/16 5:03 PM<br />
(b) In SRSs of size n 5 4, the sample mean<br />
axle diameter will typically vary by about<br />
0.001 mm from the true mean of 40.125 mm.<br />
3. (a) m x = m = 250 grams;<br />
s x = s !n = 42 = 9.39 grams<br />
!20<br />
(b) In SRSs of size n 5 20, the sample mean<br />
weight of tomatoes will typically vary by<br />
about 9.39 grams from the true mean of<br />
250 grams.<br />
4. (a) m x = m = 23.6 miles per gallon;<br />
s x = s !n = 2.5 = 0.5 mile per gallon<br />
!25<br />
(b) In SRSs of size n 5 25, the sample<br />
mean fuel efficiency will typically vary by<br />
about 0.5 mile per gallon from the true<br />
mean of 23.6 miles per gallon.<br />
5. Mean: m x = m = 9.70;<br />
SD: s x = s !n = 0.03<br />
!5 = 0.013<br />
Shape: Normal because the population<br />
distribution is normal<br />
9.65 − 9.70<br />
z = = −3.85;<br />
0.013<br />
P( x ≤ 9.65) = P(Z ≤ −3.85) ≈ 0<br />
Using technology: Applet/normalcdf<br />
(lower:−1000, upper:9.65, mean:9.70,<br />
SD:0.013) 5 0.0001<br />
6. Mean: m x = m = 9.5;<br />
SD: s x = s !n = 1.0<br />
!10 = 0.316<br />
Shape: Normal because the population<br />
distribution is normal<br />
10 − 9.5<br />
z = = 1.58; P(x ≥ 10) =<br />
0.316<br />
P(Z ≥ 1.58) = 1−0.9429 = 0.0571<br />
Using technology: Applet/normalcdf<br />
(lower:10, upper:1000, mean:9.5,<br />
SD:0.316) 5 0.0568<br />
7. Mean: m x = m = 188;<br />
SD: s x = s !n = 41<br />
!100 = 4.1<br />
Shape: Normal because the population<br />
distribution is normal<br />
185 − 188<br />
z = = −0.73;<br />
4.1<br />
191 − 188<br />
z = = 0.73<br />
4.1<br />
P(185 ≤ x ≤ 191) = P(−0.73 ≤<br />
Z ≤ 0.73) = 0.7673 − 0.2327 = 0.5346<br />
Using technology: Applet/normalcdf<br />
(lower:185, upper:191, mean:188, SD:4.1)<br />
5 0.5357<br />
8. Mean: m x = m = 298;<br />
SD: s x = s !n = 3<br />
!16 = 0.75<br />
Shape: Normal because the population<br />
distribution is normal<br />
297 − 298<br />
z = = −1.33;<br />
0.75<br />
299 − 298<br />
z = = 1.33<br />
0.75<br />
P(297 ≤ x ≤ 299)= P(−1.33 ≤ Z ≤<br />
1.33) = 0.9082 − 0.0918 = 0.8164<br />
Using technology: Applet/normalcdf<br />
(lower:297, upper:299, mean:298,<br />
SD:0.75) 5 0.8176<br />
Lesson 6.5<br />
L E S S O N 6.5 • The Sampling Distribution of a Sample Mean 437<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 437<br />
11/01/17 3:57 PM
438<br />
C H A P T E R 6 • Sampling Distributions<br />
9. (a) m x = m = 48 months;<br />
s x = s !n = 8.2 = 2.90 months<br />
!8<br />
(b) In SRSs of size n 5 8, the sample mean<br />
battery life will typically vary by about 2.90<br />
months from the true mean of 48 months.<br />
(c) The sampling distribution is normal<br />
because the population distribution is<br />
normal.<br />
42.2 − 48<br />
z = = −2.00; P( x < 42.2) =<br />
2.90<br />
P( Z < −2.00) = 0.0228<br />
Using technology: Applet/normalcdf<br />
(lower:−1000, upper:42.2, mean:48,<br />
SD:2.90) 5 0.0228<br />
10. (a) m x = m = 3.4 kg;<br />
s x = s !n = 0.5 = 0.129 kg<br />
!15<br />
(b) In SRSs of size n 5 15, the sample mean<br />
birth weight will typically vary by about<br />
0.129 kg from the true mean of 3.4 kg.<br />
(c) The sampling distribution is normal<br />
because the population distribution is<br />
normal.<br />
3.55 − 3.4<br />
z = = 1.16; P( x > 3.55) =<br />
0.129<br />
P( Z > 1.16) = 1− 0.8770 = 0.1230<br />
Using technology: Applet/normalcdf<br />
(lower:3.55, upper:1000, mean:3.4,<br />
SD:0.129) 5 0.1225<br />
11. Yes; assuming the true mean battery<br />
life is 48 months, there is only about a 2%<br />
chance of getting a sample mean of 42.2<br />
or lower purely by chance. Because this<br />
result is unlikely (less than 5%), we have<br />
convincing evidence that the population<br />
mean battery life is less than 48 months.<br />
12. No; assuming the true mean birth<br />
weight is 3.4 kg, there is about a 12%<br />
chance of getting a sample mean of 3.55<br />
or higher purely by chance. Because this<br />
result is plausible (greater than 5%), we<br />
do not have convincing evidence that<br />
the population mean is more than 3.4 kg.<br />
13. (a) Less likely; individual values<br />
vary more than averages, so getting an<br />
individual value that is close to the true<br />
mean is less likely.<br />
185 − 188<br />
(b) z = = −0.07;<br />
41<br />
z = 191−188 = 0.07<br />
41<br />
P(185 ≤ X ≤ 191)= P(−0.07 ≤ Z<br />
≤ 0.07) = 0.5279 − 0.4721 = 0.0558<br />
Using technology: Applet/normalcdf<br />
(lower:185, upper:191, mean:188, SD:41)<br />
5 0.0583. This probability of 0.0583 is<br />
much less than probability calculated in<br />
Exercise 7 (0.5357).<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of x for SRSs of size 8.<br />
(b) Interpret the standard deviation from part (a).<br />
(c) Find the probability that the sample mean life is<br />
less than 42.2 months.<br />
10. Birth weights The birth weights of males born full term<br />
are normally distributed with mean m 5 3.4 kilograms<br />
and standard deviation s 5 0.5 kilogram. 16 A large<br />
city hospital selects a random sample of 15 full-term<br />
males born in the last six months.<br />
(a) Calculate the mean and standard deviation of<br />
the sampling distribution of x for SRSs of size 15.<br />
(b) Interpret the standard deviation from part (a).<br />
(c) Find the probability that the sample mean weight is<br />
greater than 3.55 kilograms.<br />
11. Could it be the battery? Refer to Exercise 9. Suppose<br />
that the average life of the batteries on these<br />
8 cars turns out to be x 5 42.2 months. Based on<br />
your answer to Exercise 9, is there convincing evidence<br />
that the population mean m is really less than<br />
48 months? Explain.<br />
12. More birth weights Refer to Exercise 10. Suppose<br />
that the average birth weight of the 15 babies turns<br />
out to be 3.55 kilograms. Based on your answer to<br />
Exercise 10, is there convincing evidence that the<br />
population mean m is really more than 3.4 kilograms?<br />
Explain.<br />
13. One man’s cholesterol In Exercise 7, you calculated<br />
the probability that x would estimate the<br />
true mean cholesterol level within 63 mg/dl of m in<br />
samples of size 100.<br />
(a) If you randomly selected one 20- to 34-year-old<br />
male instead of 100, would he be more likely, less<br />
likely, or equally likely to have a cholesterol level<br />
within 63 mg/dl of m? Explain this without doing<br />
any calculations.<br />
(b) Calculate the probability of the event described in<br />
part (a) to confirm your answer.<br />
14. One bottle of cola In Exercise 8, you calculated the<br />
probability that x would estimate the true mean<br />
amount of cola within 61 milliliter of m in samples<br />
of size 16.<br />
(a) If you randomly selected one bottle instead of 16,<br />
would it be more likely, less likely, or equally likely<br />
to contain an amount of cola within 61 milliliter of<br />
m? Explain this without doing any calculations.<br />
(b) Calculate the probability of the event described in<br />
part (a) to confirm your answer.<br />
15. Sampling music Refer to Exercise 1. How many<br />
songs would you have to sample if you wanted the<br />
standard deviation of the sampling distribution of<br />
x to be 30 seconds?<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 438<br />
14. (a) Less likely; individual values vary more<br />
than averages, so getting an individual value<br />
that is close to the true mean is less likely.<br />
297 − 298<br />
(b) z = = −0.33;<br />
3<br />
299 − 298<br />
z = = 0.33<br />
3<br />
P(297 ≤ X ≤ 299)= P(−0.33 ≤ Z ≤ 0.33)<br />
= 0.6293 − 0.3707= 0.2586<br />
Using technology: Applet/normalcdf(lower:297,<br />
upper:299, mean:298, SD:3) 5 0.2611. This<br />
probability of 0.2611 is much less than the<br />
probability calculated in Exercise 8 (0.8176).<br />
15. s x = s 60<br />
→ 30 =<br />
!n "n → n 5 4;<br />
we would have to sample 4 songs.<br />
16. Sampling auto parts Refer to Exercise 2. How many<br />
axles would you have to sample if you wanted the<br />
standard deviation of the sampling distribution of<br />
x to be 0.0005 millimeter?<br />
Extending the Concepts<br />
17. Orange overage Mandarin oranges from a certain<br />
grove have weights that follow a normal distribution<br />
with mean m 5 3 ounces and standard deviation<br />
s 5 0.5 ounce. Bags are filled with an SRS of<br />
20 mandarin oranges. What is the probability that<br />
the total weight of oranges in a bag is greater than<br />
65 ounces? Hint: Re-express the total weight of 20<br />
oranges in terms of the average weight x.<br />
Recycle and Review<br />
18. Let’s text (1.2) We used Census at School’s “Random<br />
Data Selector” to choose a sample of 50 Canadian<br />
students who completed the survey in a recent year.<br />
The bar graph displays data on students’ responses<br />
to the question “Which of these methods do you<br />
most often use to communicate with your friends?”<br />
25<br />
Frequency<br />
20<br />
15<br />
10<br />
5<br />
0<br />
Text<br />
In<br />
person<br />
Social<br />
media<br />
Phone<br />
Other<br />
Method of communication<br />
(a) Would it be appropriate to make a pie chart for<br />
these data? Why or why not?<br />
(b) Jerry says that he would describe this bar graph as<br />
skewed to the right. Explain why Jerry is wrong.<br />
19. Shut it down and go to sleep! (5.1, 5.2) A National<br />
Sleep Foundation survey of 1103 parents asked,<br />
among other questions, how many electronic<br />
devices (TVs, video games, smartphones, computers,<br />
MP3 players, and so on) children had in their<br />
bedrooms. 17 Let X 5 the number of devices in a<br />
randomly chosen child’s bedroom. Here is the<br />
probability distribution of X.<br />
number of<br />
0 1 2 3 4 5<br />
devices<br />
Probability 0.28 0.27 0.18 0.16 0.07 0.04<br />
(a) Show that this is a legitimate probability distribution.<br />
(b) What is the probability that a randomly chosen child<br />
has at least 1 electronic device in her bedroom?<br />
(c) Calculate the expected value and standard deviation<br />
of X.<br />
16. s x = s 0.002<br />
→ 0.0005 =<br />
!n !n → n 5 16;<br />
we would have to sample 16 axles.<br />
17. A bag of 20 oranges that weighs 65<br />
ounces would give an average orange weight<br />
of x = 3.25 ounces. So we want to find<br />
P( x > 3.25).<br />
Mean: m x = m = 3; SD: s x = s !n = 0.5<br />
!20 = 0.11<br />
Shape: Normal because the population<br />
distribution is normal z = 3.25 − 3 = 2.27;<br />
0.11<br />
P(total weight > 65) = P( x > 3.25) =<br />
P( Z > 2.27) = 1− 0.9884 = 0.0116<br />
Using technology: Applet/normalcdf(lower:<br />
3.25, upper:1000, mean:3, SD:0.11) 5 0.0115<br />
Answers 18–19 are on page 439<br />
18/08/16 5:03 PMStarnes_<strong>3e</strong>_CH0<br />
438<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 438<br />
11/01/17 3:57 PM
18/08/16 5:03 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 439<br />
Lesson 6.6<br />
The central Limit Theorem<br />
L e A r n i n g T A r g e T S<br />
d Determine if the sampling distribution of x is approximately normal when<br />
sampling from a non-normal population.<br />
d If appropriate, use a normal distribution to calculate probabilities involving x.<br />
In Lesson 6.5, you learned about the sampling distribution of the sample mean x<br />
when sampling from a normally distributed population. The following activity will<br />
help you explore what happens when you sample from non-normal populations.<br />
AcT iviT y<br />
Sampling from a non-normal population<br />
In this activity, we will use an applet to investigate the<br />
sampling distribution of the sample mean x when<br />
sampling from a non-normal population.<br />
1. go to http://onlinestatbook.com/stat_sim/sampling_dist/<br />
or search for “online statbook sampling<br />
distributions applet” and go to the website. Launch<br />
the applet and select the “Skewed” population. Set<br />
the bottom two graphs to display the mean—one<br />
Answers continued<br />
18. (a) Yes, a pie chart is appropriate<br />
here because the categories (method of<br />
communication) form parts of a whole.<br />
(b) The graph should not be described as<br />
skewed to the right because this is a distribution<br />
of categorical data not quantitative data.<br />
The categories could be graphed in any order.<br />
19. (a) The probabilities are all between 0<br />
and 1. Also, the sum of the probabilities is<br />
0.28 1 0.27 1 0.18 1 0.16 1 0.07 1 0.04 5 1<br />
(b) Using the complement rule,<br />
P(X ≥ 1) = 1− P(X = 0) = 1− 0.28 = 0.72<br />
(c) E(X) = m X = 0(0.28) + 1(0.27) + 2(0.18) +<br />
+ 3(0.16) + 4(0.07) + 5(0.04) = 1.59 devices<br />
s X = 1.422 devices<br />
for samples of size 2 and the other for samples of<br />
size 5. Click the “Animated” button a few times to be<br />
sure you see what’s happening. Then “Clear lower 3”<br />
and take 100,000 SRSs. Describe what you see.<br />
2. <strong>Ch</strong>ange the sample sizes to n 5 10 and n 516 and<br />
repeat Step 1. What do you notice?<br />
3. Now change the sample sizes to n 5 20 and<br />
n 5 25 and take 100,000 samples. Did this confirm<br />
what you saw in Step 2?<br />
4. Clear the page, and select “Custom” distribution<br />
from the drop-down menu at the top of the page.<br />
Click on a point on the horizontal axis, and drag<br />
up to create a bar. Make a distribution that looks<br />
as strange as you can. (Note: You can shorten a bar<br />
or get rid of it completely by clicking on the top<br />
of the bar and dragging down to the axis.) Then<br />
repeat Steps 1 to 3 for your custom distribution.<br />
Cool, huh?<br />
5. Summarize what you learned about the shape of<br />
the sampling distribution of x.<br />
439<br />
Learning Target Key<br />
The problems in the test bank are<br />
keyed to the learning targets using<br />
these numbers:<br />
d 6.6.1<br />
d 6.6.2<br />
18/08/16 5:03 PM<br />
BELL RINGER<br />
Thinking back to the “A penny for your<br />
thoughts?” activity, does the sampling<br />
distribution of the sample mean x always<br />
have the same shape? Does it have the<br />
same shape as the population distribution?<br />
Activity Overview<br />
Time: 15–18 minutes<br />
Materials: An Internet-connected device<br />
for each student or group of students<br />
Teaching Advice: This activity helps<br />
students understand the shape of the<br />
sampling distribution of x from nonnormal<br />
populations. Contrast this with<br />
the activity from the previous lesson<br />
where the population was normal. As<br />
noted in previous activities, it is best to<br />
have students work individually or in<br />
pairs, but the applet work can be done<br />
in larger groups or as an entire class. If<br />
your class didn’t do the activity in Lesson<br />
6.5, show the layout of the applet and<br />
demonstrate taking a few samples. In<br />
particular, show the animations for the<br />
second and third number line.<br />
This applet gives a visual of the<br />
population distribution (the top/first<br />
number line), the distribution of one<br />
sample (the second number line), and<br />
the sampling distribution (the third and<br />
fourth number lines). Point out these<br />
three distributions to your students.<br />
Make sure students click “Animated”<br />
in Step 1! Don’t let them miss the<br />
visual reminder of the process of<br />
random sampling. Also in Step 1, make<br />
sure students select “Mean” from the<br />
dropdown box next to the fourth<br />
number line in the applet.<br />
There are two mysterious statistics<br />
reported by the applet: skew and kurtosis.<br />
Neither is important for this course. The<br />
skewness statistic is a measure of the<br />
skewness of the distribution; the kurtosis<br />
statistic measures how light or heavy the<br />
tails of the distribution are relative to a<br />
normal distribution.<br />
Answers:<br />
1. The sampling distribution of the sample<br />
mean x for n 5 2 and n 5 5 have a<br />
mean near 8, which is the mean of the<br />
population. The standard deviation of<br />
the sampling distribution for n 5 5 is<br />
less than the variability of the sampling<br />
distribution for n 5 2, which is less than<br />
the variability of the population. The<br />
shape of both sampling distributions<br />
is skewed right, but the sampling<br />
distribution for n 5 5 seems to be a<br />
little less skewed.<br />
2. The sampling distributions have the<br />
same mean as the population. The<br />
variability in the sampling distribution<br />
decreases as n increases. The shape of<br />
the sampling distribution is less skewed<br />
as n increases.<br />
Activity answers continue on page 440<br />
Lesson 6.6<br />
L E S S O N 6.6 • The Central Limit Theorem 439<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 439<br />
11/01/17 3:57 PM
Activity answers continued<br />
3. Yes, this confirms our observations in<br />
Step 2.<br />
4. Students should paint their own<br />
population distribution. Yes, this is<br />
cool. The results from Steps 1 to 3<br />
hold true for this new population (no<br />
matter what the population looks like)!<br />
5. The sampling distribution of x has<br />
the same mean as the population<br />
distribution. As n increases, the<br />
variability in the sampling distribution<br />
decreases. The shape of the<br />
sampling distribution becomes more<br />
approximately normal as n increases.<br />
440<br />
C H A P T E R 6 • Sampling Distributions<br />
The Central Limit Theorem<br />
It is a remarkable fact that as the sample size increases, the sampling distribution<br />
of x changes shape: It looks less like that of the population and more like a normal<br />
distribution. This is true no matter what shape the population distribution has. This<br />
famous fact of probability theory is called the central limit theorem (sometimes abbreviated<br />
as CLT).<br />
DEFINITION central limit theorem (cLt)<br />
Draw an SRS of size n from any population with mean m and finite standard deviation s.<br />
The central limit theorem (CLT) says that when n is large, the sampling distribution of<br />
the sample mean x is approximately normal.<br />
How large a sample size n is needed for the sampling distribution of x to be close<br />
to normal depends on the population distribution. A larger sample size is required<br />
if the shape of the population distribution is far from normal. In that case, the sampling<br />
distribution of x will also be far from normal if the sample size is small. To use<br />
a normal distribution to calculate probabilities involving x, check the Normal/Large<br />
Sample condition.<br />
Teaching Tip<br />
It is hard to overstate the importance<br />
of the central limit theorem. Students<br />
should know this theorem by name.<br />
Make sure students understand that the<br />
CLT applies to sample means and refers<br />
to the shape (and only the shape) of the<br />
sampling distribution of x.<br />
Teaching Tip<br />
There is nothing magical about 30 as<br />
the boundary value for a “large” sample<br />
size. Some statisticians and authors<br />
recommend n 5 40 as the boundary.<br />
In truth, some populations need<br />
sample sizes much larger than 40 to<br />
have sampling distributions that are<br />
approximately normal. For our purposes,<br />
30 is a reasonable value that works<br />
relatively well for most populations.<br />
Make sure that students understand that<br />
n ≥ 30 is just a guideline (like the Large<br />
Counts condition for proportions).<br />
a<br />
e XAMPLe<br />
A few more pennies for your thoughts?<br />
Sampling from a non-normal population<br />
DEFINITION normal/Large Sample condition<br />
The Normal/Large Sample condition says that the distribution of x will be approximately<br />
normal when either of the following is true:<br />
• The population distribution is approximately normal. This is true no matter what the<br />
sample size n is.<br />
• The sample size is large. If the population distribution is not normal, the sampling<br />
distribution of x will be approximately normal in most cases if n ≥ 30.<br />
PROBLEM: Mr. Ramirez’s class did the Penny for Your<br />
Thoughts Activity from the beginning of this chapter.<br />
The histogram shows the distribution of ages for the<br />
2341 pennies in their collection.<br />
(a) Describe the shape of the sampling distribution<br />
of x for SRSs of size n 5 2 from the population of<br />
pennies. Justify your answer.<br />
(b) Describe the shape of the sampling distribution<br />
of x for SRSs of size n 5 50 from the population of<br />
pennies. Justify your answer.<br />
Frequency<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
1950 1960 1970 1980 1990 2000 2010 2020<br />
Year<br />
Alternate Example<br />
Will you show me the money?<br />
Sampling from a non-normal population<br />
PROBLEM: Early in 2016, the<br />
opening weekend earnings from<br />
the top-earning movies of all time<br />
were reported by the website Box<br />
Office Mojo. The histogram shows the<br />
distribution of earnings, rounded to the<br />
nearest million dollars, for the top 1799<br />
movies based on opening weekend<br />
gross income.<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 440<br />
Frequency<br />
900<br />
800<br />
700<br />
600<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
50 100 150 200 250 300<br />
Opening weekend gross<br />
earnings ($ millions)<br />
(a) Describe the shape of the sampling<br />
distribution of x for SRSs of size n 5 60 from<br />
this population of movies. Justify your answer.<br />
(b) Describe the shape of the sampling<br />
distribution of x for SRSs of size n 5 10 from<br />
this population of movies. Justify your answer.<br />
SOLUTION:<br />
(a) Because n 5 60 ≥ 30, the sampling<br />
distribution of x will be approximately normal<br />
by the central limit theorem.<br />
(b) Because n 5 10 < 30, the sampling<br />
distribution of x will also be skewed to the right,<br />
but not quite as strongly as the population.<br />
18/08/16 5:03 PMStarnes_<strong>3e</strong>_CH0<br />
440<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 440<br />
11/01/17 3:58 PM
L E S S O N 6.6 • The Central Limit Theorem 441<br />
SOLUTION:<br />
(a) Because n 5 2 < 30, the sampling distribution of x will be skewed to the left, but not quite as strongly as<br />
the population.<br />
(b) Because n 5 50 ≥ 30, the sampling distribution of x will be approximately normal by the central limit<br />
theorem.<br />
FOR PRACTICE TRY EXERCISE 1.<br />
Lesson 6.6<br />
The dotplots in Figure 6.11 show the simulated sampling distributions of the sample<br />
mean for (a) 500 SRSs of size n 5 2 and (b) 500 SRSs of size n 5 50.<br />
1970<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
1980 1990 2000 2010<br />
Sample mean (n = 2)<br />
2020<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d d<br />
d ddddddddddddddddddddddddddddddddddddddddddddd d ddddddddddddddddddddddddddddddddddddddddddddddddd d ddddddddddddddddddddddddddddddddddddddddddddddddddddddd d<br />
1998 2000 2002 2004 2006<br />
Sample mean (n = 50)<br />
As expected, the simulated sampling distribution of x for SRSs of size n 5 2 is<br />
skewed to the left and the simulated sampling distribution of x for SRSs of size n 5 50<br />
is approximately normal—thanks to the central limit theorem.<br />
Probabilities Involving x<br />
Using the central limit theorem, we can do probability calculations involving x even<br />
when the population is non-normal.<br />
Mean texts?<br />
Probabilities involving x<br />
PROBLEM: Suppose that the number of texts sent during a typical day by the population of students<br />
at a large high school follows a right-skewed distribution with a mean of 45 and a standard<br />
deviation of 35. How likely is it that a random sample of 100 students will average at least 50 texts<br />
per day?<br />
SOLUTION:<br />
• Mean: m- x 5 45 texts<br />
• SD: s x - = 35 5 3.5 texts<br />
"100<br />
• Shape: Approximately normal by the CLT because n 5 100 ≥ 30<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
FigUre 6.11 Simulated<br />
sampling distributions of<br />
the sample mean age for<br />
(a) 500 SRSs of size n 5 2<br />
and (b) 500 SRSs of size<br />
n 5 50 from a population<br />
of pennies.<br />
e XAMPLe<br />
Let x = the sample mean number of texts. To find<br />
P(x ≥ 50), we have to know the mean, standard<br />
deviation, and shape of the sampling distribution of x.<br />
Recall that m x -= m and s– x = s "n .<br />
Common Error<br />
Remind your students that the CLT only<br />
addresses the shape of the sampling<br />
distribution of x. It doesn’t tell us about<br />
the center or variability of the sampling<br />
distribution.<br />
Alternate Example<br />
Will you show me more money?<br />
Probabilities involving x<br />
PROBLEM: The opening weekend<br />
earnings for the top 1799 movies of all time<br />
have a distribution that is strongly rightskewed<br />
with a mean of $26.2 million and<br />
standard deviation of $22.9 million. What is<br />
the probability that a random sample of 50<br />
of these movies will have average opening<br />
weekend earnings of less than $19 million?<br />
SOLUTION:<br />
• Mean: m x = $26.2 million<br />
• SD: s x = $22.9 = $3.24 million<br />
"50<br />
• Shape: Approximately normal by the<br />
CLT because n 5 50 ≥ 30<br />
18/08/16 5:03 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 441<br />
18/08/16 5:04 PM<br />
19<br />
16.48 19.72<br />
22.96 26.20<br />
29.44 32.68 35.92<br />
Sample mean opening weekend gross<br />
earnings ($ millions)<br />
19 − 26.2<br />
Using Table A: z = = −2.22<br />
3.24<br />
and P(Z < 22.22) 5 0.0132<br />
Using technology: Applet/normalcdf (lower:<br />
2100000, upper: 19, mean: 26.2, SD: 3.24)<br />
5 0.0131<br />
L E S S O N 6.6 • The Central Limit Theorem 441<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 441<br />
11/01/17 3:58 PM
442<br />
C H A P T E R 6 • Sampling Distributions<br />
Teaching Tip:<br />
Differentiate<br />
By this point in the chapter, advanced<br />
math students may notice similarities<br />
between the sampling distribution of p^<br />
and the sampling distribution of x. This<br />
is not a coincidence. Thinking back to<br />
the “A penny for your thoughts?” activity<br />
in Lesson 6.1, we can let the random<br />
variable X be 0 if the penny is not from<br />
the 2000s, and let it be 1 if it is from the<br />
2000s. That is, X is 0 for a “failure” and 1<br />
for a “success.” Then in a sample of n 5 5<br />
pennies, the sample proportion is just<br />
the sum of the values of X, divided by 5.<br />
In other words, the sample proportion<br />
can be thought of as a special case of the<br />
sample mean.<br />
34.5 38 41.5 45 48.5 52 55.5<br />
50<br />
Sample mean number of texts<br />
Using Table A:<br />
50 − 45<br />
z = = 1.43<br />
3.5<br />
P (Z ≥ 1.43) 5 1 2 0.9236 5 0.0764<br />
Using technology:<br />
Applet/normalcdf (lower: 50, upper: 100000, mean: 45,<br />
SD: 3.5) 5 0.0766<br />
L e SSon APP 6. 6<br />
Keeping things cool with statistics?<br />
1. Draw a normal curve.<br />
2. Perform calculations.<br />
(i) Standardize the boundary value and use Table A to<br />
find the desired probability; or<br />
(ii) Use technology.<br />
FOR PRACTICE TRY EXERCISE 5.<br />
Lesson App<br />
Answers<br />
1. Because n = 70 ≥ 30, the sampling<br />
distribution of x is approximately normal<br />
by the central limit theorem.<br />
2. Mean: m x = m = 1 hour;<br />
SD: s x = s !n = 1.5 = 0.179 hour<br />
!70<br />
1.1 − 1.0<br />
z = = 0.56; P( x > 1.1)<br />
0.179<br />
= P(Z > 0.56) = 1 − 0.7123 = 0.2877<br />
Using technology: Applet/normalcdf<br />
(lower:1.1, upper:1000, mean:1, SD:0.179)<br />
5 0.2882<br />
3. No; there is a 29% chance that the<br />
time allotted will not be enough.<br />
Your company has a contract to perform preventive maintenance<br />
on thousands of air-conditioning units in a large city. Based on<br />
service records from the past year, the time (in hours) that a technician<br />
requires to complete the work follows a strongly right-skewed<br />
distribution with m 5 1 hour and s 5 1.5 hours. As a promotion,<br />
your company will provide service to a random sample of 70 airconditioning<br />
units free of charge. You plan to budget an average of<br />
1.1 hours per unit for a technician to complete the work. Will this be<br />
enough time?<br />
1. What is the shape of the sampling distribution of x for samples of size<br />
n 5 70 from this population? Justify.<br />
2. Calculate the probability that the average maintenance time x for 70<br />
units exceeds 1.1 hours.<br />
3. Based on your answer to the previous problem, did the company<br />
budget enough time? Explain.<br />
simazoran/iStock/Getty Images<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 442<br />
TRM Quiz 6B: Lessons 6.4–6.6<br />
You can find a prepared quiz for Lessons<br />
6.4–6.6 by clicking on the link in the TE-book,<br />
logging into the Teacher’s Resource site, or<br />
accessing this resource on the TRFD.<br />
TRM chapter 6 Activity: Sampling<br />
Movies (The Sequel)<br />
This activity reviews the sampling distribution<br />
of x by sampling from a population of movies.<br />
Access this resource by clicking on the link<br />
in the TE-book, logging into the Teacher’s<br />
Resource site, or accessing this resource on<br />
the TRFD.<br />
18/08/16 5:04 PMStarnes_<strong>3e</strong>_CH0<br />
442<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 442<br />
11/01/17 3:58 PM
L E S S O N 6.6 • The Central Limit Theorem 443<br />
Lesson 6.6<br />
18/08/16 5:04 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 443<br />
WhAT DiD y o U LeA rn?<br />
LEARNINg TARgET EXAMPLES EXERCISES<br />
Determine if the sampling distribution of x is approximately normal<br />
when sampling from a non-normal population.<br />
If appropriate, use a normal distribution to calculate probabilities<br />
involving x.<br />
Exercises<br />
Mastering Concepts and Skills<br />
1. Songs on an iPod David’s iPod has about 10,000<br />
songs. The distribution of the play times for these<br />
songs is heavily skewed to the right with a mean<br />
of 225 seconds and a standard deviation of 60<br />
seconds.<br />
(a) Describe the shape of the sampling distribution of<br />
x for SRSs of size n 5 5 from the population of<br />
songs on David’s iPod. Justify your answer.<br />
(b) Describe the shape of the sampling distribution of<br />
x for SRSs of size n 5 100 from the population of<br />
songs on David’s iPod. Justify your answer.<br />
2. Insurance claims An insurance company claims<br />
that in the entire population of homeowners, the<br />
mean annual loss from fire is m 5 $250 with a standard<br />
deviation of s 5 $5000. The distribution of<br />
losses is strongly right-skewed: Many policies have<br />
$0 loss, but a few have large losses.<br />
(a) Describe the shape of the sampling distribution of<br />
x for SRSs of size n 5 15 from the population of<br />
homeowners. Justify your answer.<br />
(b) Describe the shape of the sampling distribution of<br />
x for SRSs of size n 5 1000 from the population of<br />
homeowners. Justify your answer.<br />
3. How many in a car? A study of rush-hour traffic in<br />
San Francisco counts the number of people in each<br />
car entering a freeway at a suburban interchange.<br />
Suppose that the number of people per car in the<br />
population of all cars that enter at this interchange<br />
during rush hours has a mean of m 5 1.5 and a<br />
standard deviation of s 5 0.75.<br />
(a) Could the distribution of the number of people<br />
per car be normal for the population of all cars<br />
entering the interchange during rush hours?<br />
Explain.<br />
(b) Describe the shape of the sampling distribution of<br />
x for SRSs of size n 5 100 from the population<br />
of all cars that enter this interchange during rush<br />
hours. Justify your answer.<br />
pg 440<br />
Lesson 6.6<br />
TRM full Solutions to Lesson 6.6<br />
Exercises<br />
You can find the full solutions for this lesson<br />
by clicking on the link in the TE-book, logging<br />
into the Teacher’s Resource site,or accessing<br />
this resource on the TRFD.<br />
Answers to Lesson 6.6 Exercises<br />
1. (a) Because n = 5 < 30, the sampling<br />
distribution of x will also be skewed to the right<br />
but not quite as strongly as the population.<br />
(b) Because n = 100 ≥ 30, the sampling<br />
distribution of x is approximately normal by<br />
the central limit theorem.<br />
p. 440 1–4<br />
p. 441 5–8<br />
4. Flawed carpets A supervisor at a carpet factory<br />
randomly selects 1-square-yard pieces of carpet<br />
and counts the number of flaws in each piece. The<br />
number of flaws per square yard in the population<br />
of 1-square-yard pieces varies with mean m 5 1.6<br />
and standard deviation s 5 1.2.<br />
(a) Could the distribution of the number of flaws be<br />
normal for the population of all 1-square-yard<br />
pieces of carpet? Explain.<br />
(b) Describe the shape of the sampling distribution of<br />
x for SRSs of size n 5 60 from the population of all<br />
1-square-yard pieces of carpet. Justify your answer.<br />
5. More songs on an iPod Refer to Exercise 1. What<br />
pg 441 is the probability that the mean length in a random<br />
sample of 100 songs is less than 4 minutes (240<br />
seconds)?<br />
6. More insurance claims Refer to Exercise 2. Suppose<br />
that the insurance company charges $300<br />
for each policy. What is the probability that the<br />
insurance company will make money on a random<br />
sample of 1000 homeowners? That is, what is the<br />
probability that the mean loss for a random sample<br />
of homeowners is less than $300?<br />
7. More people in a car Refer to Exercise 3. What<br />
is the probability that the mean number of people<br />
in a random sample of 100 cars that enter at this<br />
interchange during rush hours is at least 1.7?<br />
8. More flawed carpets Refer to Exercise 4. What is<br />
the probability that the mean number of flaws in<br />
a random sample of sixty 1-square-yard pieces of<br />
carpet is at least 1.7?<br />
Applying the Concepts<br />
9. Where does lightning strike? The number of lightning<br />
strikes on a square kilometer of open ground<br />
in a year has a mean of 6 and standard deviation<br />
of 2.4. The National Lightning Detection Network<br />
(NLDN) uses automatic sensors to watch for lightning<br />
in a random sample of fifty 1-square-kilometer<br />
18/08/16 5:04 PM<br />
2. (a) Because n = 15 < 30, the sampling<br />
distribution of x will also be skewed to the right<br />
but not quite as strongly as the population.<br />
(b) Because n = 1000 ≥ 30, the sampling<br />
distribution of x is approximately normal by<br />
the central limit theorem.<br />
3. (a) No; a count only takes on wholenumber<br />
values, so it cannot be normally<br />
distributed.<br />
(b) Because n = 100 ≥ 30, the sampling<br />
distribution of x is approximately normal by<br />
the central limit theorem.<br />
4. (a) No; a count only takes on wholenumber<br />
values, so it cannot be normally<br />
distributed.<br />
(b) Because n = 60 ≥ 30, the sampling<br />
distribution of x is approximately normal<br />
by the central limit theorem.<br />
5. Mean: m x = m = 225 seconds;<br />
SD: s x = s !n = 60<br />
!100 = 6 seconds<br />
Shape: Because n = 100 ≥ 30, the sampling<br />
distribution of x is approximately<br />
normal by the central limit theorem.<br />
240 − 225<br />
z = = 2.5;<br />
6<br />
P( x < 240) = P(Z < 2.5) = 0.9938<br />
Using technology: Applet/normalcdf<br />
(lower:21000, upper:240, mean:225,<br />
SD:6) 5 0.9938<br />
6. Mean: m x = m = $250;<br />
SD: s x = s !n = 5000<br />
!1000 = $158.11<br />
Shape: Because n = 1000 ≥ 30, the<br />
sampling distribution of x is approximately<br />
normal by the central limit theorem.<br />
300 − 250<br />
z =<br />
158.11 = 0.32;<br />
P( x < 300) = P( Z < 0.32) = 0.6255<br />
Using technology: Applet/normalcdf<br />
(lower:21000, upper:300, mean:250,<br />
SD:158.11) 5 0.6241<br />
7. Mean: m x = m = 1.5 people;<br />
SD: s x = s !n = 0.75 = 0.075 people<br />
!100<br />
Shape: Because n = 100 ≥ 30, the sampling<br />
distribution of x is approximately<br />
normal by the central limit theorem.<br />
1.7 − 1.5<br />
z = = 2.67; P( x > 1.7) =<br />
0.075<br />
P( Z > 2.67) = 1− 0.9962 = 0.0038<br />
Using technology: Applet/normalcdf<br />
(lower:1.7, upper:1000, mean:1.5,<br />
SD:0.075) 5 0.0038<br />
8. Mean: m x = m = 1.6 flaws;<br />
SD: s x = s !n = 1.2 = 0.155 flaws<br />
!60<br />
Shape: Because n = 60 ≥ 30, the sampling<br />
distribution of x is approximately<br />
normal by the central limit theorem.<br />
1.7 − 1.6<br />
z = = 0.65; P( x > 1.7) =<br />
0.155<br />
P( Z > 0.65) = 1− 0.7422 = 0.2578<br />
Using technology: Applet/normalcdf<br />
(lower:1.7, upper:1000, mean:1.6,<br />
SD:0.155) 5 0.2594<br />
Lesson 6.6<br />
L E S S O N 6.6 • The Central Limit Theorem 443<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 443<br />
11/01/17 3:58 PM
9. (a) Because n = 50 ≥ 30, the<br />
sampling distribution of x is approximately<br />
normal by the central limit theorem.<br />
(b) Mean: m x = m = 6 lightning strikes;<br />
SD: s x = s !n = 2.4 = 0.339 lightning<br />
!50<br />
strikes; z = 5 − 6<br />
0.339 = −2.95;<br />
P( x < 5) = P( Z < −2.95) = 0.0016<br />
Using technology: Applet/normalcdf<br />
(lower:21000, upper:5, mean:6,<br />
SD:0.339) 5 0.0016<br />
10. (a) Because n = 45 ≥ 30, the<br />
sampling distribution of x is approximately<br />
normal by the central limit theorem.<br />
(b) Mean: m x = m = 12.8 minutes;<br />
SD: s x = s !n = 7.2 = 1.073 minutes<br />
!45<br />
15 − 12.8<br />
z = = 2.05; P( x > 15) =<br />
1.073<br />
P( Z > 2.05) = 1 − 0.9798 = 0.0202<br />
Using technology: Applet/normalcdf<br />
(lower:15, upper:1000, mean:12.8,<br />
SD:1.073) 5 0.0202<br />
11. (a) Because n = 10 < 30, we can’t<br />
be sure that the sampling distribution of<br />
x will be approximately normal.<br />
(b) Greater than; the variability of the<br />
sampling distribution of x will be greater<br />
with the smaller sample size of 10, making<br />
it more likely to get a sample mean<br />
that is far away from the true mean of 6<br />
lightning strikes (such as x 5 5 or lower).<br />
12. (a) Because n = 5 < 30, we can’t<br />
be sure that the sampling distribution of<br />
x will be approximately normal.<br />
(b) Greater than; the variability of the<br />
sampling distribution of x will be greater<br />
with the smaller sample size of 5, making<br />
it more likely to get a sample mean that<br />
is far away from the true mean of 12.8<br />
minutes (such as x 5 15 or higher).<br />
13. From Exercise 7, we know that when<br />
the true mean number of people in the<br />
car is 1.5, there is almost a 0% chance<br />
that the mean number of people in the<br />
car will be at least 1.7 in a random sample<br />
of 100 cars. Because the observed result<br />
is unlikely to happen purely by chance<br />
(less than 5%), the researcher has good<br />
evidence to conclude that people are<br />
more likely to drive with other people in<br />
the car on Sundays.<br />
14. From Exercise 8, we know that when<br />
the true mean number of flaws is 1.6,<br />
there is about a 26% chance that the<br />
mean number of flaws will be at least<br />
1.7 in a random sample of 60 pieces of<br />
carpet. Because this is a large probability,<br />
it is plausible that the supervisor’s result<br />
occurred purely by chance and that the<br />
machine is still working properly.<br />
444<br />
C H A P T E R 6 • Sampling Distributions<br />
plots of land. Let x be the average number of lightning<br />
strikes in the sample.<br />
(a) What is the shape of the sampling distribution of x for<br />
samples of size n 5 50 from this population? Justify.<br />
(b) Calculate the probability that the average number of<br />
lightning strikes per square kilometer x is less than 5.<br />
10. Please hold The customer care manager at a cell<br />
phone company keeps track of how long each<br />
help-line caller spends on hold before speaking to<br />
a customer service representative. He finds that the<br />
distribution of wait times for all callers has a mean<br />
of 12.8 minutes with a standard deviation of 7.2<br />
minutes. The distribution is moderately skewed to the<br />
right. Suppose the manager takes a random sample of<br />
45 callers and calculates their mean wait time x.<br />
(a) What is the shape of the sampling distribution of<br />
x for samples of size n 5 45 from this population?<br />
Justify.<br />
(b) Calculate the probability that the mean wait time x<br />
is more than 15 minutes.<br />
11. Lightning strikes twice? Refer to Exercise 9.<br />
(a) Explain why you cannot calculate the probability that<br />
the average number of lightning strikes per square<br />
kilometer x is less than 5 for samples of size n 5 10.<br />
(b) Will the probability referred to in part (a) be less<br />
than, greater than, or about the same as the probability<br />
in Exercise 9(b)? Explain.<br />
12. Please continue to hold Refer to Exercise 10.<br />
(a) Explain why you cannot calculate the probability<br />
that the average wait time for customer service x is<br />
more than 15 minutes for samples of size n 5 5.<br />
(b) Will the probability referred to in part (a) be less<br />
than, greater than, or about the same as the probability<br />
in Exercise 10(b)? Explain.<br />
13. Even more people in a car Refer to Exercise 7. A<br />
researcher selects a random sample of 100 cars<br />
that enter this interchange on a Sunday and finds<br />
x 5 1.7 people per car. Because the sample mean<br />
is greater than 1.5, the researcher concludes that<br />
people are more likely to drive with other people<br />
in the car on Sundays. Based on your answer to<br />
Exercise 7, what would you say to the researcher?<br />
14. Even more flaws in carpets Refer to Exercise 8.<br />
A supervisor selects a random sample of sixty<br />
1-square-yard pieces of carpet and finds that<br />
x 5 1.7 flaws. Because the sample mean is more<br />
than the expected mean of 1.6 flaws, the supervisor<br />
is thinking about shutting down the machine<br />
for inspection. Based on your answer to Exercise 8,<br />
what would you say to the supervisor?<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 444<br />
15. No; the graph of the sample will resemble<br />
the shape of the population distribution,<br />
regardless of the sample size. The student<br />
should say that the graph of the sampling<br />
distribution of the sample mean (x) looks more<br />
and more normal as you take larger and larger<br />
samples from a population.<br />
16. No; the central limit theorem is only<br />
about the shape of the sampling distribution<br />
of the sample mean. The statement is<br />
otherwise correct.<br />
17. A total of $45,000 for 50 students yields<br />
an average cost per student of $900. We want<br />
to find P(x > 900).<br />
Mean: m x = m = $836;<br />
SD: s x = s !n = 388<br />
!50 = $54.87<br />
15. What does the CLT say? Asked what the central<br />
limit theorem says, a student replies, “As you take<br />
larger and larger samples from a population, the<br />
graph of the sample values looks more and more<br />
normal.” Is the student right? Explain your answer.<br />
16. Is this what the CLT says? Asked what the central<br />
limit theorem says, a student replies, “As you take<br />
larger and larger samples from a population, the<br />
variability of the sampling distribution of the sample<br />
mean decreases.” Is the student right? Explain your<br />
answer.<br />
Extending the Concepts<br />
17. Cost of textbooks The cost of textbooks for students<br />
at a particular college has a mean of m 5 $836<br />
per year with a standard deviation of s 5 $388.<br />
What is the probability that a random sample of<br />
50 students spends a total of more than $45,000<br />
on books this year? Hint: Re-express the total cost<br />
in terms of the average cost per student x.<br />
Recycle and Review<br />
18. Are rich people mean? (3.1, 3.6) Psychologist<br />
Paul Piff from the University of California,<br />
Berkeley, studies the relationship between<br />
wealth and lawful behavior. In one such study,<br />
he had assistants cross a road at a crosswalk and<br />
recorded if drivers obeyed the law and stopped<br />
to let the person cross or kept driving and cut<br />
off the pedestrian. He compared the response of<br />
people driving expensive cars and inexpensive<br />
cars. Here are his results. 18 type of car<br />
Driver<br />
behavior<br />
Yielded to<br />
pedestrian<br />
Cut off<br />
pedestrian<br />
Expensive car<br />
Inexpensive car<br />
32 67<br />
26 27<br />
(a) The report on this study stated that the researcher<br />
who determined if the cars could be classified as<br />
expensive or inexpensive was “blind to the hypothesis<br />
of the study.” Explain what this means.<br />
(b) Is this an observational study or an experiment?<br />
Justify your answer.<br />
19. How do rich people drive? (4.3, 4.4) Suppose we<br />
choose a driver at random from the results of the study<br />
in Exercise 18. Show that the events “Yielded to pedestrians”<br />
and “Expensive car” are not independent.<br />
Shape: Because n = 50 ≥ 30, the sampling<br />
distribution of x is approximately normal by<br />
the central limit theorem.<br />
900 − 836<br />
z = = 1.17; P(total cost > 45,000)<br />
54.87<br />
= P( x > 900) = P(Z > 1.17) = 1 − 0.8790<br />
= 0.1210<br />
Using technology: Applet/normalcdf(lower:900,<br />
upper:10000, mean:836, SD:54.87)5 0.1217<br />
18. (a) This means that the researcher<br />
who determined whether the cars could be<br />
classified as expensive or inexpensive didn’t<br />
know the other variable being measured<br />
(driver behavior). If the researcher knew the<br />
hypothesis of the study, it could have affected<br />
the classification of the cars.<br />
Answers 18(b)–19 are on page 445<br />
18/08/16 5:04 PMStarnes_<strong>3e</strong>_CH0<br />
444<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 444<br />
11/01/17 3:58 PM
<strong>Ch</strong>apter 6 Main Points<br />
445<br />
18/08/16 5:04 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 445<br />
© Kirsty Pargeter/Alamy Stock Photo<br />
Main Points<br />
The Idea of a Sampling Distribution<br />
j<br />
j<br />
<strong>Ch</strong>apter 6<br />
Answers continued<br />
18. (b) Observational study. There were no<br />
treatments imposed on the drivers. In other<br />
words, drivers weren’t assigned to drive an<br />
expensive car or an inexpensive car.<br />
19. P(yielded to pedestrians | expensive car)<br />
32<br />
5<br />
32 + 26 = 32<br />
58 = 0.55<br />
P(yielded to pedestrians | inexpensive car)<br />
67<br />
5<br />
67 + 27 = 67<br />
94 = 0.71<br />
Because the probabilities are not equal, the<br />
events “Yielded to pedestrians” and<br />
“Expensive car” are not independent.<br />
Knowing that a randomly selected car is<br />
expensive decreases the probability that the<br />
car yielded to pedestrians.<br />
STATS applied!<br />
How can we build “greener” batteries?<br />
Refer to the STATS applied! on page 399. When the manufacturing<br />
process is working properly, the distribution of battery lifetimes<br />
has mean m 5 17 hours with standard deviation s 5 0.8 hour, and<br />
73% last at least 16.5 hours.<br />
1. Assume that the manufacturing process is working properly, and let p^ 5 the sample<br />
proportion of batteries that last at least 16.5 hours. Calculate the mean and standard<br />
deviation of the sampling distribution of p^ for random samples of 50 batteries.<br />
2. Describe the shape of the sampling distribution of p^ for random samples of 50<br />
batteries. Justify your answer.<br />
3. In the sample of 50 batteries, only 68% lasted at least 16.5 hours. Find the probability<br />
of obtaining a random sample of 50 batteries where p^ is 0.68 or less if the<br />
manufacturing process is working properly.<br />
4. Assume that the process is working properly, and let x 5 the sample mean lifetime<br />
(in hours). Calculate the mean and standard deviation of the sampling distribution<br />
of x for random samples of 50 batteries.<br />
5. Describe the shape of the sampling distribution of x for random samples of 50<br />
batteries. Justify your answer.<br />
6. In the sample of 50 batteries, the mean lifetime was only 16.718 hours. Find the<br />
probability of obtaining a random sample of 50 batteries with a mean lifetime of<br />
16.718 hours or less if the manufacturing process is working properly.<br />
7. Based on your answers to Questions 3 and 6, should the company be worried that<br />
the manufacturing process isn’t working properly? Explain.<br />
<strong>Ch</strong>apter 6<br />
A parameter is a number that describes some characteristic<br />
of the population. A statistic is a number<br />
that describes some characteristic of a sample. We<br />
use statistics to estimate parameters.<br />
The sampling distribution of a statistic is the distribution<br />
of values taken by the statistic in all possible<br />
samples of the same size from the same population.<br />
j<br />
To determine a sampling distribution, list all possible<br />
samples of a particular size, calculate the value<br />
of the statistic for each sample, and graph the<br />
distribution of the statistic. If there are many possible<br />
samples, use simulation to approximate the<br />
sampling distribution: Repeatedly select random<br />
samples of a particular size, calculate the value of<br />
the statistic for each sample, and graph the distribution<br />
of the statistic.<br />
18/08/16 5:04 PM<br />
TRM chapter 6 Learning Targets Grid<br />
You can find a grid with all of the learning<br />
targets for this chapter by clicking on the link<br />
in the TE-book, logging into the Teacher’s<br />
Resource site, or accessing this resource on<br />
the TRFD. An extra column has been added for<br />
students to track their progress.<br />
Teaching Tip:<br />
STATS applied!<br />
This STATS applied! reviews the skills and<br />
ideas from this chapter. Many students<br />
will have difficulty switching between<br />
sample proportions and sample means.<br />
They may also struggle with using the<br />
correct symbols. Monitor their progress<br />
and provide help and practice as needed.<br />
Answers to STATS applied!<br />
p(1− p)<br />
1. m p^ = p = 0.73; s p^ = Å n<br />
0.73(1 − 0.73)<br />
= = 0.063<br />
Å 50<br />
2. Because np = (50)(0.73) = 36.5 $ 10<br />
and n(1− p) = (50)(1− 0.73)=13.5 $ 10,<br />
the sampling distribution of p^ is<br />
approximately normal.<br />
0.68 − 0.73<br />
3. z = = −0.79;<br />
0.063<br />
P( p^ ≤ 0.68) = P( Z ≤ −0.79) = 0.2148<br />
Using technology: Applet/normalcdf<br />
(lower:21000, upper:0.68, mean:0.73,<br />
SD:0.063) 5 0.2137<br />
4. m x = m = 17 hours;<br />
s x = s !n = 0.8 = 0.113 hour<br />
!50<br />
5. Because n = 50 ≥ 30, the sampling<br />
distribution of x is approximately normal<br />
by the central limit theorem.<br />
6. z = 16.718−17 = −2.50;<br />
0.113<br />
P( x ≤ 16.718) = P( Z ≤ −2.5) = 0.0062<br />
Using technology: Applet/normalcdf<br />
(lower:21000, upper:16.718, mean:17,<br />
SD:0.113) 5 0.0063<br />
7. From Question 3, we know that if<br />
the manufacturing process is working<br />
properly (p = 0.73), there is about a<br />
21% chance that the sample proportion<br />
of batteries that last less than 16.5<br />
hours would be less than 0.68. Because<br />
this outcome is plausible, the sample<br />
proportion of 0.68 does not provide<br />
strong evidence that the true proportion<br />
is less than 0.73, and the company<br />
should not be worried.<br />
However, from Question 6, we know<br />
that if the manufacturing process is<br />
working properly (m = 17 hours), there<br />
is only about a 1% chance that the mean<br />
battery life will be 16.718 hours or less.<br />
Because this outcome is unlikely (less<br />
than 5%), the sample mean of 16.718<br />
provides strong evidence that the true<br />
mean battery life is less than 17 hours,<br />
and the company should be worried.<br />
Main Points<br />
C H A P T E R 6 • Main Points 445<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 445<br />
11/01/17 3:58 PM
446<br />
C H A P T E R 6 • Sampling Distributions<br />
Answers to <strong>Ch</strong>apter 6 Review<br />
Exercises<br />
1. Population: All eggs shipped in one<br />
day. Sample: The 200 eggs examined.<br />
Parameter: The proportion p of eggs<br />
shipped that day that had salmonella.<br />
Statistic: The proportion of eggs in<br />
the sample that had salmonella,<br />
p^ = 9<br />
200 = 0.045.<br />
2. (a)<br />
Sample #1: 64, 66, 71 Median 5 66<br />
Sample #2: 64, 66, 73 Median 5 66<br />
Sample #3: 64, 66, 76 Median 5 66<br />
Sample #4: 64, 71, 73 Median 5 71<br />
Sample #5: 64, 71, 76 Median 5 71<br />
Sample #6: 64, 73, 76 Median 5 73<br />
Sample #7: 66, 71, 73 Median 5 71<br />
Sample #8: 66, 71, 76 Median 5 71<br />
Sample #9: 66, 73, 76 Median 5 73<br />
Sample #10: 71, 73, 76 Median 5 73<br />
d<br />
d<br />
d<br />
d<br />
d<br />
d<br />
66 67 68 69 70 71 72 73<br />
Sample median book length<br />
66 + 66 + 66 + 71 + 71 +<br />
71 + 71 + 73 + 73 + 73<br />
(b) m median =<br />
10<br />
= 701<br />
10 = 70.1<br />
The sample median is a biased estimator<br />
of the population median. The mean<br />
of the sampling distribution is equal to<br />
70.1, which is less than the value of the<br />
population median of 71.<br />
(c) The sampling distribution of the<br />
sample median will be less variable<br />
because the sample size is larger. The<br />
estimated median book length will<br />
typically be closer to the true median<br />
book length. In other words, the estimate<br />
will be more precise.<br />
3. (a) m X = np = 500(0.24) = 120<br />
people; s X = "np(1 − p)<br />
= "500(0.24)(1 − 0.24) = 9.55 people<br />
(b) If many samples of size 500 were<br />
taken, the number of people who are<br />
under 18 years old would typically vary<br />
by about 9.55 from the mean of 120.<br />
(c) The sampling distribution of X<br />
is approximately normal because<br />
np = 500(0.24) = 120 ≥ 10 and<br />
n(1 − p) = 500(1 − 0.24) = 380 ≥ 10<br />
j<br />
j<br />
j<br />
We can use sampling distributions to determine<br />
what values of a statistic are likely to happen by<br />
chance alone and how much a statistic typically<br />
varies from the parameter it is trying to estimate.<br />
A statistic used to estimate a parameter is an<br />
unbiased estimator if the mean of its sampling<br />
distribution is equal to the value of the parameter<br />
being estimated. That is, the statistic doesn’t<br />
consistently overestimate or consistently underestimate<br />
the value of the parameter when many<br />
random samples are selected.<br />
The sampling distribution of any statistic will<br />
have less variability when the sample size is larger.<br />
That is, the statistic will be a more precise estimator<br />
of the parameter with larger sample sizes.<br />
Sample Counts and Sample Proportions<br />
j<br />
j<br />
Let X 5 the number of successes in a random sample<br />
of size n from a large population with proportion<br />
of successes p. The sampling distribution of a<br />
sample count X describes the distribution of values<br />
taken by the sample count X in all possible samples<br />
of the same size from the same population.<br />
j The mean of the sampling distribution of X<br />
is m X = np. The mean describes the average<br />
value of X in repeated random samples.<br />
j The standard deviation of the sampling distribution<br />
of X is s X = !np(1 − p). The standard<br />
deviation describes how far the values of X typically<br />
vary from m X in repeated random samples.<br />
j The shape of the sampling distribution of X<br />
will be approximately normal when the Large<br />
Counts condition is met: np ≥ 10 and n(1 2 p)<br />
≥10.<br />
Let p^ 5 the proportion of successes in a random<br />
sample of size n from a large population with proportion<br />
of successes p. The sampling distribution of a<br />
sample proportion p^ describes the distribution of values<br />
taken by the sample proportion p^ in all possible<br />
samples of the same size from the same population.<br />
j The mean of the sampling distribution of p^ is<br />
m p^ = p. The mean describes the average value<br />
of p^ in repeated random samples.<br />
j The standard deviation of the sampling<br />
p(1 − p)<br />
distribution of p^ is s p^ 5 . The<br />
Å n<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 446<br />
100 − 120<br />
(d) z = = −2.09;<br />
9.55<br />
110 − 120<br />
z = = −1.05<br />
9.55<br />
P(100 ≤ X ≤ 110) ≈ P(−2.09 ≤<br />
Z ≤ − 1.05) = 0.1469 − 0.0183 = 0.1286<br />
Using technology: Applet/normalcdf<br />
(lower:100, upper:110, mean:120, SD:9.55)<br />
5 0.1294<br />
j<br />
standard deviation describes how far the values<br />
of p^ typically vary from p in repeated random<br />
samples.<br />
The shape of the sampling distribution of<br />
p^ will be approximately normal when the<br />
Large Counts condition is met: np ≥ 10 and<br />
n(1 2 p) ≥ 10.<br />
Sample Means<br />
j<br />
j<br />
Let x 5 the mean of a random sample of size n<br />
from a large population with mean m and standard<br />
deviation s. The sampling distribution of<br />
a sample mean x describes the distribution of<br />
values taken by the sample mean x in all possible<br />
samples of the same size from the same<br />
population.<br />
j The mean of the sampling distribution of x is<br />
m x = m. The mean describes the average value<br />
of x in repeated random samples.<br />
j The standard deviation of the sampling distribution<br />
of x is s− x = s . The standard deviation<br />
describes how far the values of x typically<br />
"n<br />
vary from m in repeated random samples.<br />
The shape of the sampling distribution of<br />
x will be approximately normal when the<br />
Normal/Large Sample condition is met: The<br />
population is normal or the sample size is large<br />
(n ≥ 30). The fact that the sampling distribution<br />
of x becomes approximately normal—<br />
even when the population is non-normal—as<br />
the sample size increases is called the central<br />
limit theorem.<br />
Probability Calculations<br />
j<br />
j<br />
When the sampling distribution of a statistic is<br />
approximately normal, you can use z-scores and<br />
Table A or technology to do probability calculations<br />
involving the statistic.<br />
To determine which sampling distribution to use,<br />
consider whether the variable of interest is categorical<br />
or quantitative. If it is categorical, use the<br />
sampling distribution of a sample count X or the<br />
sampling distribution of a sample proportion p^ . If<br />
it is quantitative, use the sampling distribution of<br />
a sample mean x.<br />
TRM chapter 6 Review Exercise Videos<br />
Video solutions to the <strong>Ch</strong>apter 6 Review<br />
Exercises are available to teachers and<br />
students. Access them by clicking on the link<br />
in the TE-book, logging into the Teacher’s<br />
Resource site, or accessing this resource<br />
on the TRFD.<br />
18/08/16 5:04 PMStarnes_<strong>3e</strong>_CH0<br />
446<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 446<br />
11/01/17 3:58 PM
<strong>Ch</strong>apter 6 Review Exercises 447<br />
<strong>Ch</strong>apter 6 Review Exercises<br />
1. Bad eggs (6.1) People who eat eggs that are contaminated<br />
with salmonella can get food poisoning. A large<br />
egg producer takes an SRS of 200 eggs from all the eggs<br />
shipped in one day. The laboratory reports that 9 of<br />
these eggs had salmonella contamination. Identify the<br />
population, the parameter, the sample, and the statistic.<br />
2. Five books (6.1, 6.2) An author has written 5 children’s<br />
books. The number of pages in these books are<br />
64, 66, 71, 73, and 76.<br />
(a) List all 10 possible SRSs of size n 5 3, calculate the<br />
median number of pages for each sample, and display<br />
the sampling distribution of the sample median on a<br />
dotplot.<br />
(b) Show that the sample median is a biased estimator of<br />
the population median for this population.<br />
(c) Describe how the variability of the sampling distribution<br />
of the sample median would change if the sample<br />
size was increased to n 5 4.<br />
3. Kids these days (6.3) According to the 2010 U.S.<br />
Census, 24% of U.S. residents are under 18 years old.<br />
Suppose we take a random sample of 500 U.S. residents.<br />
Let X 5 the number of people in the sample<br />
who are under 18 years old.<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of X.<br />
(b) Interpret the standard deviation from part (a).<br />
(c) Justify that it is appropriate to use a normal distribution<br />
to model the sampling distribution of X.<br />
(d) Using a normal distribution, calculate the probability<br />
that the number of people under 18 years old in a random<br />
sample of size 500 is between 100 and 110.<br />
4. Five-second rule (6.1, 6.4) A report claimed that 20%<br />
of respondents subscribe to the “5-second rule.” That<br />
is, they would eat a piece of food that fell onto the<br />
kitchen floor if it was picked up within 5 seconds. Assume<br />
this figure is accurate for the population of U.S.<br />
adults. Let p^ = the proportion of people who subscribe<br />
to the 5-second rule in an SRS of size 80 from<br />
this population.<br />
(a) Calculate the mean and the standard deviation of the<br />
sampling distribution of p^ .<br />
(b) Interpret the standard deviation from part (a).<br />
(c) Justify that it is appropriate to use a normal distribution<br />
to model the sampling distribution of p^ .<br />
(d) In an SRS of size 80, only 10% subscribed to the<br />
5-second rule. Does this result provide convincing evidence<br />
that the proportion of all U.S. adults who subscribe<br />
to the 5-second rule is less than 0.20? Calculate<br />
P(p^ ≤ 0.10) and use this result to support your answer.<br />
5. Normal IQ scores? (6.5, 6.6) The Wechsler Adult<br />
Intelligence Scale (WAIS) is a common “IQ test” for<br />
adults. The distribution of WAIS scores for persons<br />
over 16 years of age is approximately normal with<br />
mean 100 and standard deviation 15. Let x 5 the<br />
mean WAIS score in a random sample of 10 people<br />
over 16 years of age.<br />
(a) Calculate the mean and standard deviation of the<br />
sampling distribution of x.<br />
(b) Interpret the standard deviation from part (a).<br />
(c) What is the probability that the average WAIS score is<br />
105 or greater for a random sample of 10 people over<br />
16 years of age? Show your work.<br />
(d) Would your answers to any of parts (a), (b), or (c) be<br />
affected if the distribution of WAIS scores in the adult<br />
population were distinctly non-normal? Explain.<br />
6. Watching for gypsy moths (6.1, 6.6) The gypsy moth<br />
is a serious threat to oak and aspen trees. A state agriculture<br />
department places traps throughout the state<br />
to detect the moths. Each month, an SRS of 50 traps<br />
is inspected, the number of moths in each trap is recorded,<br />
and the mean number of moths is calculated.<br />
Based on years of data, the distribution of moth<br />
counts is strongly skewed with mean 0.5 and standard<br />
deviation 0.7.<br />
(a) Explain why it is reasonable to use a normal distribution<br />
to approximate the sampling distribution of x for<br />
SRSs of size 50.<br />
(b) Calculate the probability that the mean number of<br />
moths in a sample of size 50 is at least 0.6 moths.<br />
(c) In a recent month, the mean number of moths in an SRS<br />
of size 50 was 0.6. Based on this result, should the state<br />
agricultural department be worried that the moth population<br />
is getting larger in their state? Explain.<br />
(c) The sampling distribution of x is<br />
approximately normal because the<br />
population distribution is approximately<br />
normal.<br />
105 − 100<br />
z = = 1.05; P ( x ≥ 105)<br />
4.74<br />
= P(Z ≥ 1.05) = 1− 0.8531 = 0.1469<br />
Using technology: Applet/normalcdf<br />
(lower:105, upper:1000, mean:100,<br />
SD:4.74) 5 0.1457<br />
(d) The answer to parts (a) and (b)<br />
would be the same because the<br />
mean and standard deviation do not<br />
depend on the shape of the population<br />
distribution. We could not answer part<br />
(c) because we could not be sure that the<br />
sampling distribution is approximately<br />
normal.<br />
6. (a) Because n = 50 ≥ 30, the<br />
sampling distribution of x is approximately<br />
normal by the central limit theorem.<br />
(b) Mean: m x = m = 0.5 moth;<br />
SD: s x = s !n = 0.7 = 0.099 moth<br />
z =<br />
!50<br />
0.6 − 0.5<br />
0.099 = 1.01;<br />
P( x ≥ 0.6) = P(Z ≥ 1.01) = 1 − 0.8438<br />
= 0.1562<br />
Using technology: Applet/normalcdf<br />
(lower:0.6, upper:1000, mean:0.5,<br />
SD:0.099) 5 0.1562<br />
(c) Assuming the true mean number of<br />
moths is 0.5, there is about a 16% chance<br />
that the mean number of moths will<br />
be at least 0.6 in a sample of 50 traps.<br />
Because this result is plausible (more<br />
than 5%), we do not have convincing<br />
evidence that the moth population is<br />
getting larger.<br />
Review<br />
18/08/16 5:04 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 447<br />
Answers 1–3 are on page 446<br />
4. (a) m p^ = p = 0.20;<br />
p(1− p)<br />
s p^ = Å n<br />
= 0.045<br />
= Å<br />
0.20(1− 0.20)<br />
80<br />
(b) In SRSs of size n 5 80, the sample<br />
proportion of people who subscribe to the<br />
5-second rule will typically vary by about<br />
0.045 from the true proportion of p 5 0.20.<br />
(c) Because np = (80)(0.2) = 16 $ 10 and<br />
n(1− p) = (80)(1− 0.20) = 64 ≥ 10, the<br />
sampling distribution of p^ is approximately<br />
normal.<br />
0.10 − 0.20<br />
(d) z = = −2.22;<br />
0.045<br />
P( p^ ≤ 0.10) = P( Z ≤ −2.22) = 0.0132<br />
18/08/16 5:04 PM<br />
Using technology: Applet/normalcdf<br />
(lower:21000, upper:0.10, mean:0.20, SD:0.045)<br />
5 0.0131.<br />
Assuming the true proportion of people who<br />
subscribe to the 5-second rule is 0.20, there is<br />
only about a 1% chance of getting a sample<br />
proportion of 0.10 or less purely by chance.<br />
Because this result is unlikely (less than 5%),<br />
we have convincing evidence that the<br />
proportion of all U.S. adults who subscribe to<br />
the 5-second rule is less than 0.20.<br />
5. (a) m x = m = 100;<br />
s x = s !n = 15<br />
!10 = 4.74<br />
(b) In SRSs of size n 5 10, the sample mean<br />
WAIS score will typically vary by about 4.74<br />
from the true mean of 100.<br />
TRM full Solutions to <strong>Ch</strong>apter<br />
Review Exercises and Test<br />
You can find the full solutions by clicking<br />
on the link in the TE-book, logging into<br />
the Teacher’s Resource Site, or accessing<br />
this resource on the TRFD.<br />
C H A P T E R 6 • Review Exercises 447<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 447<br />
11/01/17 3:58 PM
448<br />
C H A P T E R 6 • Sampling Distributions<br />
Answers to <strong>Ch</strong>apter 6 Practice<br />
Test<br />
<strong>Ch</strong>apter 6<br />
Practice Test<br />
1. b<br />
2. c<br />
3. c<br />
4. b<br />
5. a<br />
6. a<br />
7. b<br />
Section I: Multiple choice Select the best answer for each question.<br />
1. A study of voting chose 663 registered voters at random<br />
shortly after an election. Of these, 72% said they<br />
had voted in the election. Election records show that<br />
only 56% of registered voters voted in the election.<br />
Which of the following statements is true about these<br />
percentages?<br />
(a) 72% and 56% are both statistics.<br />
(b) 72% is a statistic and 56% is a parameter.<br />
(c) 72% is a parameter and 56% is a statistic.<br />
(d) 72% and 56% are both parameters.<br />
2. Vermont is particularly beautiful in early October<br />
when the leaves begin to change color. At that time of<br />
year, a large proportion of cars on Interstate 91 near<br />
Brattleboro have out-of-state license plates. Suppose a<br />
Vermont state trooper randomly selects 50 cars driving<br />
past Exit 2 on I-91, records the state identified<br />
on the license plate, and calculates the proportion of<br />
cars with out-of-state plates. Which of the following<br />
describes the sampling distribution of the sample proportion<br />
in this context?<br />
(a) The distribution of state for all cars in the trooper’s<br />
sample of cars passing this exit<br />
(b) The distribution of state for all cars passing this exit<br />
(c) The distribution of the proportion of cars with<br />
out-of-state plates in all possible samples of 50 cars<br />
passing this exit<br />
(d) The distribution of the proportion of cars with<br />
out-of-state plates in the trooper’s sample of 50 cars<br />
passing this exit<br />
3. A polling organization wants to estimate the proportion<br />
of voters who favor a new law banning smoking<br />
in public buildings. The organization decides to<br />
increase the size of its random sample of voters from<br />
about 1500 people to about 4000 people right before<br />
an election. The effect of this increase is to<br />
(a) reduce the bias of the estimate.<br />
(b) increase the bias of the estimate.<br />
(c) reduce the variability of the estimate.<br />
(d) increase the variability of the estimate.<br />
4. A machine is designed to fill 16-ounce bottles of<br />
shampoo. When the machine is working properly, the<br />
amount poured into the bottles follows a normal distribution<br />
with mean 16.05 ounces and standard deviation<br />
0.1 ounce. Assume that the machine is working<br />
properly. If 4 bottles are randomly selected and the<br />
number of ounces in each bottle is measured, then<br />
there is about a 95% chance that the sample mean<br />
will fall in which of the following intervals?<br />
(a) 16.00 to 16.10 ounces<br />
(b) 15.95 to 16.15 ounces<br />
(c) 15.90 to 16.20 ounces<br />
(d) 15.85 to 16.25 ounces<br />
5. The central limit theorem is important in statistics<br />
because it allows us to use the normal distribution to<br />
find probabilities involving the sample mean if the<br />
(a) sample size is reasonably large for any population<br />
shape.<br />
(b) sample size is reasonably large and the population is<br />
normally distributed.<br />
(c) population size is reasonably large for any population<br />
shape.<br />
(d) population size is reasonably large and the population<br />
is normally distributed.<br />
6. At a high school, 85% of students are right-handed.<br />
Let X 5 the number of students who are right-handed<br />
in a random sample of 10 students from the school.<br />
Which one of the following statements about the<br />
mean and standard deviation of the sampling distribution<br />
of X is true?<br />
(a) m x 5 8.5; s x ≈ 1.129<br />
(b) m x 5 8.5; s x ≈ 0.113<br />
(c) m x 5 8.5; s x ≈ cannot be determined from the information<br />
given.<br />
(d) Neither the mean nor the standard deviation can be<br />
determined from the information given.<br />
7. The student newspaper at a large university asks an<br />
SRS of 250 undergraduates, “Do you favor eliminating<br />
the carnival from the end-of-term celebration?”<br />
In the sample, 150 of the 250 undergraduates are<br />
in favor. Suppose that 55% of all undergraduates<br />
favor eliminating the carnival. If you took a very<br />
large number of SRSs of size n 5 250 from this<br />
population, the sampling distribution of the sample<br />
proportion p^ would have which of the following<br />
characteri stics?<br />
(a) Mean 0.55, standard deviation 0.03, shape unknown<br />
(b) Mean 0.55, standard deviation 0.03, approximately<br />
normal<br />
(c) Mean 0.60, standard deviation 0.03, shape unknown<br />
(d) Mean 0.60, standard deviation 0.03, approximately<br />
normal<br />
Starnes_<strong>3e</strong>_CH06_398-449_Final.indd 448<br />
18/08/16 5:04 PMStarnes_<strong>3e</strong>_CH0<br />
448<br />
C H A P T E R 6 • Sampling Distributions<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 448<br />
11/01/17 3:58 PM
<strong>Ch</strong>apter 6 Practice Test<br />
449<br />
8. Scores on the mathematics part of the SAT exam in a<br />
recent year followed a normal distribution with mean<br />
515 and standard deviation 114. You choose an SRS<br />
of 100 students and calculate x5 mean SAT Math<br />
score. Which of the following are the mean and standard<br />
deviation of the sampling distribution of x?<br />
(a) Mean 5 515, SD 5 114<br />
114<br />
(b) Mean 5 515, SD 5<br />
"100<br />
(c) Mean 5 515 114<br />
, SD 5<br />
100 100<br />
(d) Mean 5 515<br />
100 , SD 5 114<br />
"100<br />
9. In a congressional district, 55% of the registered voters<br />
are Democrats. Which of the following is closest to<br />
the probability of getting less than 50% Democrats in<br />
a random sample of size 100?<br />
(a) 0.157 (b) 0.496 (c) 0.504 (d) 0.843<br />
10. A statistic is an unbiased estimator of a parameter<br />
when<br />
(a) the statistic is calculated from a random sample.<br />
(b) in all possible samples of a specific size, the distribution<br />
of the statistic has a shape that is approximately<br />
normal.<br />
(c) in all possible samples of a specific size, the values of<br />
the statistic are very close to the value of the parameter.<br />
(d) in all possible samples of a specific size, the values of<br />
the statistic are centered at the value of the parameter.<br />
Section II: Free Response<br />
11. Here are histograms of the values taken by three<br />
sample statistics in several hundred samples from the<br />
same population. The true value of the population<br />
parameter is marked with an arrow on each histogram.<br />
Which statistic would provide the best estimate<br />
of the parameter? Explain.<br />
A B C<br />
12. The amount that households pay service providers for<br />
access to the Internet varies quite a bit, but the mean<br />
monthly fee is $48 and the standard deviation is $20.<br />
The distribution is not normal: Many households pay<br />
a base rate for low-speed access, but some pay much<br />
more for faster connections. A sample survey asks an<br />
SRS of 500 households with Internet access how much<br />
they pay per month. Let x be the mean amount paid<br />
by the members of the sample.<br />
(a) Calculate the mean and standard deviation of the sampling<br />
distribution of x. Interpret the standard deviation.<br />
(b) What is the shape of the sampling distribution of x?<br />
Justify.<br />
(c) Find the probability that the average amount paid by<br />
the sample of households exceeds $50.<br />
13. According to government data, 22% of American children<br />
under the age of 6 live in households with incomes<br />
less than the official poverty level. A study of learning in<br />
early childhood chooses an SRS of 300 children.<br />
(a) Let X 5 the count of children in this sample who live in<br />
households with incomes less than the official poverty<br />
level. What is the shape of the sampling distribution of<br />
X? Justify your answer.<br />
(b) Find the probability that more than 20% of the sample<br />
are from poverty-level households.<br />
8. b<br />
9. a<br />
10. d<br />
11. Statistic A. Both statistics A and B<br />
appear to be unbiased, with the center of<br />
their sampling distributions equal to the<br />
value of the parameter, but statistic A has<br />
less variability than statistic B.<br />
12. (a) m x = m = $48;<br />
s x = s !n = 20<br />
!500 = $0.89<br />
In SRSs of size n 5 500, the sample mean<br />
amount paid for Internet will typically<br />
vary by about $0.89 from the true mean<br />
of $48.<br />
(b) Because n = 500 ≥ 30, the sampling<br />
distribution of x is approximately normal<br />
by the central limit theorem.<br />
50 − 48<br />
(c) z = = 2.25; P( x > 50)<br />
0.89<br />
= P(Z > 2.25) = 1 − 0.9878 = 0.0122<br />
Using technology: Applet/normalcdf<br />
(lower:50, upper:1000, mean:48, SD:0.89)<br />
5 0.0123<br />
13. (a) The sampling distribution of<br />
X is approximately normal because<br />
np = 300(0.22) = 66 ≥ 10 and<br />
n(1 − p) = 300(1 − 0.22) = 234 ≥ 10.<br />
(b) 20% of 300 is 60, so we want to find<br />
P( X > 60).<br />
m X = np = 300(0.22) = 66;<br />
s X = "np(1 − p)<br />
= "300(0.22)(1 − 0.22) = 7.17<br />
60 − 66<br />
z = = −0.84;<br />
7.17<br />
P(X > 60) = P(Z > −0.84) = 0.7995<br />
Using technology: Applet/normalcdf<br />
(lower:60, upper:1000, mean:66, SD:7.17)<br />
5 0.7987<br />
Practice Test<br />
18/08/16 5:04 PMStarnes_<strong>3e</strong>_CH06_398-449_Final.indd 449<br />
18/08/16 5:04 PM<br />
C H A P T E R 6 • Practice Test 449<br />
Starnes_<strong>3e</strong>_ATE_CH06_398-449_v3.indd 449<br />
11/01/17 3:58 PM