438 CHAPTER 12 Testing Hypotheses

ETHICS IN ACTION

I t has been three years since Mohammed Al-Tamimi opened

his computer repair business, Mo’s Mending Station. Unlike
the well-known Nerd Squad of the big electronics retailer,
Mo’s Mending Station fixes only computers and does not deal
with any other electronics such as TVs, phones, cameras, or
professional, experienced, and friendly repair technicians. In
other words, put the focus on “twice the nerd ” rather than
“half the cost.”
Mohammed thinks that this is a great idea and wonders
if they can also prepare some statistics to strengthen the
appliances. Nor does Mo’s provide any in-home services, such message. Because customer receipts include both when a
as networking or computer setup. Mo’s main objective is clear: computer is brought to Mo’s (date and time) as well as when
to provide standard repair services for computers and laptops, the repair is finished, he asks his office manager to select a
virus and spyware removal, and data recovery, each at a fixed sample so they can estimate the average service time. Based
low price. He charges the competitive rate of $45 per hour for on 36 receipts, she finds a mean service time of 2 hours and
more complicated computer issues. 10 minutes with a standard deviation of 30 minutes. Further
Mo’s slogan is “Get twice the nerd at half the cost!” This statistical analysis yielded a 95% confidence interval for the
strategy has worked well, allowing Mohammed to grow his mean service time of 1.99 to 2.33 hours. Mohammed plans to
business to include six repair technicians and one office advertise that 95% of his customers can expect to wait
manager. However, recent monthly receipts indicate that the between 1.99 and 2.33 hours to get their computers back
demand for Mo’s services may be slowing down. Worried that from repair! He is anxious to include this claim in all of the
the Mending Station might be losing its competitive price Mending Station’s future marketing communication materials.
advantage, M ­ ohammed gathers his staff together for a brain-
• Identify the ethical dilemma in this scenario.
storming session. Ed Ramsey, who has been with Mo’s since
it opened, mentions the possibility that the Mending Station’s • Has Mohammed interpreted the confidence interval correctly?
low prices may give some potential customers the impression • What are the undesirable consequences?
that it offers poor quality service. He suggests hiring a local • Propose an ethical solution that considers the welfare of all
advertising firm to help brand Mo’s Mending Station as stakeholders.
affordable AND high quality by emphasizing its team of

CHAPTER

12 FROM LEARNING TO EARNING

LEARNING Know how to formulate a null and an alternative hypothesis for a question of interest.
OBJECTIVES • The null hypothesis specifies a parameter and a (null) value for that parameter.
• The alternative hypothesis specifies a range of plausible values should we reject the null.

Be able to perform a hypothesis test for a proportion.

• The null hypothesis has the form H0: p = p0.
• We find the standard deviation of the sampling distribution of the sample proportion by
assuming that the null hypothesis is true:

p0q0
SD1pn 2 = .
B n
pn - p0
• We refer the statistic z = to the standard Normal model.
SD1pn 2

M12_SHAR9313_04_GE_C12.indd 438 24/09/2020 20:35

 From Learning to Earning 439

Be able to perform a hypothesis test for a mean.

• To apply the Central Limit Theorem for the mean in practical applications, we must esti-
mate the standard deviation. This standard error is
s
SE1y2 =
2n
• When we use the SE, the sampling distribution that allows for the additional uncertainty is
Student’s t-model on n - 1 degrees of freedom.
y - m0
• We refer the test statistic t = to the Student’s t distribution with n - 1 degrees
of freedom. SE1y2
• Find critical values by technology or from tables.
• Check the assumptions and conditions before using any sampling distribution for inference.

Write clear summaries to interpret a confidence interval or state a hypothesis test’s conclusion.

Understand P-values.
• A P-value is the estimated probability of observing a statistic value at least as far from the
(null) hypothesized value as the one we have actually observed.
• A small P-value indicates that the statistic we have observed would be unlikely were the
null hypothesis true. That leads us to doubt the null.
• A large P-value just tells us that we have insufficient evidence to doubt the null hypothesis.
In particular, it does not prove the null to be true.

Know the reasoning of hypothesis testing.

• State the hypotheses.
• Determine (and check assumptions for) the sampling distribution model.
• Calculate the test statistic—the mechanics.
• State your conclusions and decisions.

Be able to decide on a two-sided or one-sided alternative hypothesis, and justify your decision.

Know that confidence intervals and hypothesis tests go hand in hand in helping us think about models.
• A hypothesis test makes a yes/no decision about the plausibility of the value of a parameter
value.
• A confidence interval shows us the range of plausible values for the parameter.

TERMS
Alternative hypothesis The alternative hypothesis proposes what we should conclude if we reject the null hypothesis.
Effect size The difference between the null hypothesis value and the true value of a model parameter.
Hypothesis A model or proposition that we adopt in order to test.
Null hypothesis  he claim being assessed in a hypothesis test that states “no change from the traditional value,”
T
“no effect,” “no difference,” or “no relationship.” For a claim to be a testable null hypothesis, it must
specify a value for some population parameter that can form the basis for assuming a sampling
distribution for a test statistic.
One-proportion z-test A test of the null hypothesis that the proportion of a single sample equals a specified value
pn - p0
1H0: p = p0 2 by referring the statistic z = to a Standard Normal model.
SD1pn 2
One-sample t -test The one-sample t-test for the mean tests the hypothesis H0: m = m0 using the statistic
for the mean
y - m0
tn - 1 = .
SE1y2
The standard error of y is
s
SE1y2 = .
2n

M12_SHAR9313_04_GE_C12.indd 439 24/09/2020 20:35

 440 CHAPTER 12 Testing Hypotheses

One-sided alternative An alternative hypothesis is one-sided 1e.g., HA: p 7 p0 or HA: p 6 p0 2 when we are interested in
(One-tailed alternative) deviations in only one direction away from the hypothesized parameter value.
P-value  he probability of observing a value for a test statistic at least as far from the hypothesized value as the
T
statistic value actually observed if the null hypothesis is true. A small P-value indicates either that the
observation is improbable or that the probability calculation was based on incorrect assumptions.
The assumed truth of the null hypothesis is the assumption under suspicion.
Two-sided alternative An alternative hypothesis is two-sided 1HA: p ≠ p0 2 when we are interested in deviations in either
(Two-tailed alternative) direction away from the hypothesized parameter value.

TECH SUPPORT Hypothesis Tests

Hypothesis tests for proportions are so easy and natural that Null hypothesis Alternative hypothesis
many statistics packages don’t offer special commands for
them. Most statistics programs want to know the “success”
and “failure” status for each case. Usually these are given as Te s t H o : ( s l e e p ) = 7 v s H a : (sleep) < 7
1 or 0, but they might be category names like “yes” and “no.” Sample Mean = 6.6400
Often you just know the proportion of successes, pn , and the t = -1.6737 w/24 df
total count, n. Computer packages don’t usually deal naturally P- val ue = 0 . 0 5 3 5 9
with summary data like these, but the statistics routines found
on many graphing calculators do. These calculators allow
you to test hypotheses from summaries of the data—usually, The t-statistic
all you need to enter are the number of successes and the (and its degrees
of freedom)
sample size.
The P-value is usually
In some programs you can reconstruct the original values. given last
But even when you have reconstructed (or can reconstruct) the
raw data values, often you won’t get exactly the same test sta- The package computes the sample mean and sample
tistic from a computer package as you would find working by standard deviation of the variable and finds the P-value from
hand. The reason is that when the packages treat the propor- the t-distribution based on the appropriate number of degrees
tion as a mean, they make some approximations. The result is of freedom. All modern statistics packages report P-values. The
very close, but not exactly the same. package may also provide additional information such as the
For quantitative data, statistics packages offer convenient sample mean, sample standard deviation, t-statistic value, and
ways to make histograms of the data. Even better for assessing degrees of freedom. These are useful for interpreting the result-
near-Normality is a Normal probability plot. When you work ing P-value and telling the difference between a meaningful
on a computer, there is simply no excuse for skipping the step result and one that is merely statistically significant. Statistics
of plotting the data to check that it is nearly Normal. Beware: packages that report the estimated standard deviation of the
Statistics packages don’t agree on whether to place the Normal sampling distribution usually label it “standard error” or “SE.”
scores on the x-axis (as we have done) or the y-axis. Read the Inference results are also sometimes reported in a table.
axis labels. You may have to read carefully to find the values you need.
Any standard statistics package can compute a hypothesis Often, test results and the corresponding confidence interval
test for a mean. Here’s what the package output might look bounds are given together. And often you must read carefully
like in general (although no package we know gives the results to find the alternative hypotheses. Here’s an example of that
in exactly this form):11 kind of output:

11
Many statistics packages keep as many as 16 digits for all intermediate
calculations. However, for reporting, the rule-of-thumb that says to report
one more decimal place for a statistic than are in the original data is a
good one.

M12_SHAR9313_04_GE_C12.indd 440 24/09/2020 20:35