Chapter 9: Introduction to Hypothesis Testing
Chapter 9: Introduction to Hypothesis Testing
Chapter 9: Introduction to Hypothesis Testing
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
376 CHAPTER 9 • INTRODUCTION TO HYPOTHESIS TESTING<br />
W HY Y OU N EED T O K NOW<br />
As a business decision maker, you will encounter many<br />
applications that will require you <strong>to</strong> estimate a population<br />
parameter. <strong>Chapter</strong> 8 introduced statistical estimation and<br />
included many examples. Estimating a population parameter<br />
based on a sample statistic is a component of business<br />
statistics called statistical inference. An important component<br />
of statistical inference is hypothesis testing. In<br />
hypothesis testing, we begin with some belief regarding<br />
what the value of a population parameter is, or should be.<br />
We then use sample data <strong>to</strong> either refute or support the<br />
initial belief.<br />
For example, suppose the Coca-Cola plant in Tampa,<br />
Florida, produces approximately 400,000 cans of Coke<br />
products daily. Each can is supposed <strong>to</strong> contain 12 fluid<br />
ounces. However, like all processes, the au<strong>to</strong>mated filling<br />
machine is subject <strong>to</strong> variation and each can will contain<br />
either slightly more or less than the 12-ounce target. The<br />
important thing is that the mean fill is 12 fluid ounces.<br />
Every two hours, the plant quality manager selects a random<br />
sample of cans and computes the sample mean. His<br />
initial belief is that the filling process is providing an average<br />
fill of 12 ounces. This is his hypothesis. If the sample<br />
mean is “close” <strong>to</strong> 12 ounces, then the sample data would<br />
tend <strong>to</strong> support the hypothesis and the machine would be<br />
allowed <strong>to</strong> continue filling cans. However, if the sample<br />
mean is “significantly” higher or lower than 12 ounces,<br />
the data would refute the hypothesis and the machine<br />
would be s<strong>to</strong>pped and adjusted.<br />
<strong>Hypothesis</strong> testing is performed regularly in many<br />
industries. Companies in the pharmaceutical industry must<br />
perform many hypothesis tests on new drug products<br />
before they are deemed <strong>to</strong> be safe and effective by the<br />
federal Food and Drug Administration (FDA). In these<br />
instances, the drug is hypothesized <strong>to</strong> be both unsafe and<br />
ineffective. Then, if the sample results from the studies<br />
performed provide “significant” evidence <strong>to</strong> the contrary,<br />
the FDA will allow the company <strong>to</strong> market the drug.<br />
<strong>Hypothesis</strong> testing is the basis of the legal system in<br />
which judges and juries hear evidence in court cases. In a<br />
criminal case, the hypothesis in the American legal system<br />
is that the defendant is innocent. Based upon the <strong>to</strong>tality<br />
of the evidence presented in the trial, if the jury concludes<br />
that “beyond a reasonable doubt” the defendant committed<br />
the crime, the hypothesis of innocence will be rejected<br />
and the defendant will be found guilty. If the evidence is<br />
not strong enough, the defendant will be judged not guilty.<br />
<strong>Hypothesis</strong> testing is a major part of business statistics.<br />
<strong>Chapter</strong> 9 introduces the fundamentals involved in<br />
conducting hypothesis tests. Many of the remaining chapters<br />
in this text will introduce additional hypothesis-testing<br />
techniques, so you need <strong>to</strong> gain a solid understanding of<br />
concepts presented in this chapter.<br />
Null <strong>Hypothesis</strong><br />
The statement about the<br />
population parameter that will be<br />
tested. The null hypothesis will<br />
be rejected only if the sample data<br />
provide substantial contradic<strong>to</strong>ry<br />
evidence.<br />
Alternative <strong>Hypothesis</strong><br />
The hypothesis that includes all<br />
population values not included in<br />
the null hypothesis. The alternative<br />
hypothesis is deemed <strong>to</strong> be true if<br />
the null hypothesis is rejected.<br />
9.1 <strong>Hypothesis</strong> Tests for Means<br />
By now you know that information contained in a sample is subject <strong>to</strong> sampling error. The<br />
sample mean will almost certainly not equal the population mean. Therefore, in situations in<br />
which you need <strong>to</strong> test a claim about a population mean by using the sample mean, you can’t<br />
simply compare the sample mean <strong>to</strong> the claim and reject the claim if x and the claim are different.<br />
Instead, you need a testing procedure that incorporates the potential for sampling error.<br />
Statistical hypothesis testing provides managers with a structured analytical method for<br />
making decisions of this type. It lets them make decisions in such a way that the probability<br />
of decision errors can be controlled, or at least measured. Even though statistical hypothesis<br />
testing does not eliminate the uncertainty in the managerial environment, the techniques<br />
involved often allow managers <strong>to</strong> identify and control the level of uncertainty.<br />
The techniques presented in this chapter assume the data are selected using an appropriate<br />
statistical sampling process and that the data are interval or ratio level. In short, we<br />
assume we are working with good data.<br />
Formulating the Hypotheses<br />
Null and Alternative Hypotheses In hypothesis testing, two hypotheses are formulated.<br />
One is the null hypothesis. The null hypothesis is represented by H 0<br />
and contains an<br />
equality sign, such as “,” “,” or “.” The second hypothesis is the alternative hypothesis<br />
(represented by H A<br />
). Based on the sample data, we either reject H 0<br />
or we do not<br />
reject H 0<br />
.<br />
Correctly specifying the null and alternative hypotheses is important. If done incorrectly,<br />
the results obtained from the hypothesis test may be misleading. Unfortunately, how you