Seminar 5 Outline

PRINT THIS
CHAPTER 8 & 11
Sections 8-1 - 8-5
Sections 11-1 - 11.2

This week you will learn how to test a claim about a population parameter. This type of
test is a Hypothesis Test.

In previous chapters we have discussed the idea of a rare event. Hypothesis Tests use the
concept of the Rare Event Rule for Inferential Statistics. This rule is stated on page 386
of your book but I will also include it here:

Rare Event Rule for Inferential Statistics

If, under a given assumption, the probability of a particular observed event is
exceptionally small, we conclude that the assumption is probably not correct.

The steps for performing a Hypothesis Test are:

1.) Identify the Null and Alternative hypothesis.

2.) Calculate the Test Statistic using sample data.
3.) Identify the Critical Value.
4.) Identify the P-value given by the Test Stat.
5.) State the conclusion of the Hypothesis Test.

In chapters eight and eleven we will learn to analyze a sample in an attempt to distinguish
between results that can easily occur from results that are highly unlikely. More
specifically, we will test claims about a population parameter. "The average life span of
75-watt light bulbs is at least 1000 hours," is a claim about the mean lifetime of all 75-
watt light bulbs, but will data from a sample of fifty 75-watt light bulbs support the claim
or fail to support the claim? If we get a sample, which fails to support the claim, could
this easily occur or is this highly unlikely? These are the kinds of questions we will learn
to address in this chapter.

This week we introduce the basic concepts and procedures used for testing claims made
about population parameters. We will develop the technique and vocabulary to test claims
about proportions, about means with σ known (very rare – use z-test) and about means
with σ unknown (this will be used most of the time – use t-test).

NOTES:

Please read carefully the section about forming your own claims on page 390. Knowing
how to define the Null Hypothesis and the Alternative Hypothesis can be confusing. If
you want to use a hypothesis test to support your claim, the claim must be worded so that
it becomes the alternative hypothesis. By rejecting the null hypothesis you can support
your claim. As you work through this section be sure to post as many questions as you
need to until this is clear.

The actual hypothesis tests are not difficult and the formulas will resemble the formulas
used to create confidence intervals. Interpreting the test statistic is what is important in
this chapter. These concepts are important and are used by many professionals including:
medical researchers, manufacturers, psychologists, and educators just to name a few.

Hypothesis Testing
Print this to use as a guide
It can be difficult sometimes to know how to set up a hypothesis test. Use these ten steps
as a guide to setting up your Null and Alternative Hypothesis. Then continue on through
the steps until you’ve reached your final conclusion.

1. STATE THE CLAIM USING ONE OF THE FOLLOWING: <, <, >, >, =, or ?.
The claim is always about a population parameter – mean or proportion.
If the claim contains “less than” use <
If the claim contains “more than” use >
If the claim contains “less than or equal to” or “at most”, use <
If the claim contains “greater than or equal to” or “at least”, use >
If the claim contains “is” or “equals” or “is equal to”, use =

2. STATE THE OPPOSITE IDEA OF THE CLAIM

= and ≠ are opposites
< and > are opposites
> and < are opposites

3. LABEL ONE OF THE ABOVE STATEMENTS H0 AND THE OTHER H1

H0 always contains the “=” or “>” or “<”
H1 always contains “ ≠” or “<” or “>”.

4. STATE THE FORMULA FOR THE TEST STATISTIC AND ITS DISTRIBUTION
If the claim is about:
a. A proportion where n*p and n*q are > 5 use the formula on page 408
b. A mean with σ known then use the formula on page 419
c. A mean with σ unknown then use the formula on page 426
d. Two proportions use the formula on page 457
e. Two means with large independent samples use the formula on page 470
f. Two means of matched pairs of data use the formula on page 485
Each of these formulas starts out either z = … or t = … The ones that start out z = … are
the test statistics that have a standard normal distribution. The ones that start out t = …
are the test statistics that have one of the student t distributions.

5. STATE IF THE TEST IS RIGHT TAILED, LEFT TAILED, OR TWO TAILED

If H1 contains > then the test is right tailed
If H1 contains < then the test is left tailed
If H1 contains ≠ then the test is two tailed

6. STATE THE CRITICAL VALUE OR VALUES (CV) OF THE DISTRIBUTION

ABOVE
Use table A-2 or A-3. (pages 772-774)
If the test is left tailed then the critical value is negative
If the test is right tailed then the critical value is positive
If the test is two tailed then there are two critical values – one positive and one negative

7. DESCRIBE THE CRITICAL REGION (REJECTION REGION)

If the test is left tailed, the critical region is all numbers less than the (negative) critical
value.
If the test is right tailed, the critical region is all numbers greater than the (positive)
critical value.
If the test is two tailed, the critical region comes in two parts – all numbers less than the
negative critical value together with all numbers greater than the positive critical value

8. COMPUTE THE VALUE OF THE TEST STATISTIC (TS) FROM THE SAMPLE
DATA.
Take the formula you selected in step 4 and substitute the values known about the sample
statistics.

9. STATE YOUR DECISION TO EITHER REJECT H0 OR FAIL TO REJECT H0

If the computed value of the test statistic (TS) is in the critical region, then REJECT H0

If the test is left tailed, the TS < CV, then the TS is in the critical region.
If the test is right tailed and the TS > CV, then the TS is in the critical region.
If the test is two-tailed and the TS > the positive CV or TS < the negative CV, then the
test statistic is in the critical region.

If the computed value of the test statistic is not in the critical region, then FAIL TO
REJECT H0.

10. STATE THE FINAL CONCLUSION

Use the flow chart on page 397 to find the correct wording of the final conclusion.
Section 11.2

A multinomial experiment is very similar to a binomial experiment except there are more
than 2 categories. When there are more than 2 categories the methods that we learned in
Chapter 8 cannot be used to perform a hypothesis test. We will learn how to perform a
Goodness of Fit Hypothesis test in this section.

In a multinomial experiment the data is categorical and is separated into different

categories. A goodness of fit test is used to determine if the observed data fit a claimed
distribution. For this test we need to understand a new distribution.

The Chi-Squared Distribution

Critical values of the Chi-Square distribution are found in Table A-4

The degrees of freedom for a Multinomial Experiment are found by k-1, where k is the
number of categories.

The test Statistic is:

(O − E ) 2
χ =∑
2

0 = Observed Frequencies
E = Expected Frequencies.

Goodness-of-fit tests are always right-tailed.