Quiz (Chapter 12)
Post Hoc and Planned Comparisons
Introduction
 (T/F) With 3 comparison groups, Ho, if false, might be
false because m1 = m2
while m3 is equal to something
else.
 (T/F) Researchers use planned comparisons more often than they do
post hoc comparisons.
Post Hoc Comparisons: Definition and Purpose
 (T/F) There is confusion among researchers as to what is or is not
a post hoc test.
 What's the statistical focus of a post hoc test that's conducted
in connection with a oneway ANOVA?
 (T/F) Post hoc tests are used when the Ho of a oneway
analysis of variance is not rejected.
Post Hoc Comparisons: Terminology
 A post hoc test is sometimes referred to as a ______ test or a ______
test or an ______ test.
 Within the post hoc tests, what's a oneword synonym for "comparison"?
 (T/F) The F that comes from a oneway ANOVA can be referred to as
the "omnibus Ftest."
 If all possible "pairwise" comparisons had been conducted
in Excerpt 11.9 (Chapter 11, p. 268), how many comparisons would have
been made?
 Suppose we have 3 groups: undergrads, grad students in an M.S. program,
& grad students in a Ph.D. program. If the 2 groups of grad
students are combined and compared against the undergrads, this is a
_____ comparison.
Post Hoc Comparisons: Test Procedures Frequently Used
 Who invented the multiple comparison test that goes by the name "HSD"?
 Who invented the multiple comparison test that's called the "New
Multiple Range Test"?
 Which of the 5 tests used most often to do post hoc comparisons is
most "liberal"? Most "conservative"?
 Which of these testsDuncan, Tukey, NewmanKeuls protects most
against Type I errors? Type II errors?
 What is the name of the multiple comparison test that compares experimental
groups versus the control group (but doesn't compare experimental groups
against each other)?
Post Hoc Comparisons: The Null Hypothesis
 (T/F) A separate Ho exists for every comparison that's
set up in a post hoc investigation.
 In Excerpt 12.18, there were __ post hoc null hypotheses.
Post Hoc Comparisons: Presentation of Results
 (T/F) Pairwise comparisons can be accomplished via Tukey's HSD procedure
. . . but cannot be accomplished by the Tukey a or Tukey B procedures.
 In Excerpt 12.14, how many Tukey tests are being summarized?
 Suppose a Tukey test is used to make all pairwise comparisons among
4 means. Suppose further that these tests indicate that (a) the largest
mean is signifficantly larger than the secondlargest mean, (b) each
of those means is significantly different from the other 2 means, and
(c) the 2 smallest means are not significantly different from each
other. If the "reporting scheme" used in Table 4 of Excerpt
12.18 is used to summarize these results, how many different letters
would appear next to the sample means?
 If we let "1" represent the young group, "2" represent the middle
group, and "3"
represent the older group, which of the following statements accurately
summarizes the results of the post hoc comparisons made in the last
sentence of
Excerpt
12.14?
(a) 3<1<2, (b) 1<2<3, (c) 3<2=1, (d) 2<1<3, (e) 1=3<2
Post Hoc Comparisons: The Bonferroni Procedure as a Post Hoc Technique
 What a would be needed if the Bonferroni
technique were to be used by a ".05" researcher who wants to conduct
a post hoc analysis wherein all possible pairwise comparisons are made
among 5 groups?
Planned Comparisons
 (T/F) Planned comparisons are turned to by the researcher if the
ANOVA F turns out p > .05.
 (T/F) Since test procedures such as Tukey's were invented for use
as post hoc investigations, they can't be used in situations where the
researcher wants to conduct planned comparisons.
 A planned contrast is sometimes referred to as a ________ contrast.
 Return to question #2 and answer it again with these two changes:
(a) insert the words "ought to" after the word "Researchers" and (b)
delete the final 3 words of the sentence.
 (T/F) The planned comparisons set up by most researchers are nonpairwise
(rather than pairwise) in nature.
 (T/F) In any given study having three or more comparison groups,
the researcher can set up some planned comparisons that are nonpairwise
in nature
along with other planned comparisons that are pairwise in nature?
Comments
 If a planned comparison is set up to contrast one group's mean against
the average of two other groups (that have been lumped together), what
will the Ftest's first df number be?
 (T/F) The term "orthogonal" means "independent."
 (T/F) The equal variance assumption is important when a researcher
conducts an omnibus Ftest but it is not important if he/she
performs a planned or post hoc investigation.
 If three samples have means of 12.15, 18.50, and 21.75, the index d would turn
out to be smallest for a comparison of which 2 means?
 (T/F) All of the test procedures developed for doing post hoc and
planned comparisons have a builtin "control mechanism" that prevents
the results from being statistically significant if there's no practical
significance.
 Look again at Excerpt 12.26. The values of d can be thought
of as "large" because
they are larger than ___ .
Click here for answers.
