OUTLINE FOR THE 2^{nd} HALF OF CHAPTER
7 IN THE 6^{th} EDITION
Hypothesis Testing
(NOTE: This outline covers pages
144159 of Ch. 7. A different outline covers pages 131144 of
this chapter.)
 The Final 2 Steps (#5 & #3) of the Basic Hypothesis
Testing Procedure
 Step #5: The Criterion for Evaluating the Sample Evidence
 The primary question: "Are the sample data inconsistent with
what would likely occur if Ho
were true?"
 Three possible outcomes when the numerical summary of the
sample data is compared against Ho's
pinpoint number . . . and what should be decided about Ho
 They're identical . . . and therefore Ho
obviously cannot be rejected
 A difference exists, but it's a "small" difference (i.e.,
small enough to be within the limits of expected sampling
error). . . and therefore it's not logical to reject Ho
 A difference exists, and it's a "big" difference (i.e.,
larger than what's likely to have been produced simply by
sampling error) . . . and thus Ho
should be rejected
 Two procedures for deciding whether an observed difference
(between the numerical summary of the sample data and Ho's
pinpoint number) should be considered "small" or "large"
 Compare the calculated value (obtained in Step #4) against
a tabled "critical value"
 Compare the databased pvalue (obtained in Step #4) against
the "level of significance"
 Step #3: Selecting the Level of Significance
 The level of significance as a "scientific cutoff point"
for deciding whether Ho should
be rejected
 Popular levels of significance: .05 and .01 (and sometimes
.001) . . . with .05 used most often
 Different ways researchers talk about the level of significance:
alpha, a, p = .05, p < .05
 Type I & Type II errors (and how a
defines the former, influences the latter if a
is changed)
 Two points of possible confusion regarding the level of
significance
 Results That Are Highly Significant and Near Misses
 The kind of p
that causes a result to be "highly significant"
 Using different plevels within the same study (even
though a constant a is used)
 The kind of p that causes a result to be a "near miss"
 Four alternative labels for an outcome that is a "near miss"
 The allornothing approach to hypothesis testing
 A Few Cautions
 Two meanings of "alpha"
 The importance of Ho
 The ambiguity of the word "hypothesis"
 When p is reported to be equal to or less than zero
 The meaning of the term "significant"
