SPSS Practice Midterm Exam
Your answers to the questions below should be
hand-written on the examination page.
"Evidence" and "Support" refer to
statistical information that you would select from your SPSS analyses to report in a paper such as means, F ratios,
p values. As you
would do in a paper, present only
relevant information but be sure that your
information completely answers
the question. [e.g., "see SPSS printout" is
insufficient] Your
ability determine what is relevant is part of the test!!!
IN ADDITION, please label your SPSS output file surname.spo (e.g., mine would be ruscher.spo), then email it to ruscher@tulane.edu I will use the print-outs to ascertain whether you performed the analysis correctly and, when appropriate, to award partial credit.
Use your homework data set, and give yourself 30 minutes for the exam
Calculators, notes, homework assignments, formula sheets, and the Howell text are legitimate resources for this portion of the exam. SPSS manuals or facsimiles of manual pages are prohibited.
1. The researcher predicts that the effect of time on evaluative ratings of targets is moderated by gender.
a. Using eval1, eval2, and eval3
as the ratings across time and gender, test the researcher's hypothesis and report the
relevant results. [I won't say what would be relevant on the actual exam, but
you should realize this means the F, df, and p level for the prediction, as
well as the corresponding means] 6points
b.
Assume that the researcher further wished to examine the simple effect of
gender at each level of time. Show these tests. 4 points
2. Another researcher, using the same data set, is interested only the effects of time on evaluation (i.e., she will conduct a 1-way repeated measures ANOVA using eval1, eval2, and eval3). Unfortunately, she is using an old version of SPSS that does not compute epsilon. Using SPSS show the relevant information that she would need to compute it. 4 points
3. A final researcher decides to collapse eval1, eval2, and eval3 into a single index.
a.
What is the internal consistency (alpha) for his new measure? 2points
b. If he examines the effect of gender on this index, what is the observed power of the
test? 2 points
c.
What sample size would give 80 - 90% power? 4 points
d. What is w2 for this
effect? 4 points