Your Name:__________________________________________
Instructions
This is an open-book exam: you may refer to books or notes, and use calculators during it, but you may not discuss the exam except with the instructor.
You will have a maximum of 50 minutes available to complete this exam during the scheduled class. If you finish before the time has expired, you may leave after giving me the exam. There are 100 points possible on this exam. The first two problems are required, and count 30 points each; choose your favorite four from among items 3 through 8 (the latter are worth 10 points each; you can do all of them but a maximum of four can count). You can write on the back of these sheets if you prefer. Please write your name on any additional sheets, and attach them to the exam when you turn it in. Be sure to indicate any relevant formulae and clearly answer all parts of each question.
1. Professor B. suspects that Pallid bats are more fearful (indexed by mg of bat guano produced during a 1-hour observation period--large numbers indicate high fearfulness) on average than are Townsend's Big-Eared bats. He collects the data below (based on 5 Pallid bats, and 6 Townsend's bats). Should he conclude that the mean for Pallid bats is significantly (p = .05) larger? Indicate the basis for your conclusion.
Pallid Townsend's
7 6
6 4
7 5
6 5
7 6
6
2. Lisa, recently promoted to Executive Associate Bartender at CBT, wonders whether students with different majors tip differently on average. She keeps track of the total amount of tips (in dollars) left by randomly selected students from each of three majors (Psychology; HM [Hotel Management]; and MIS [Management Information Systems]), with 4 students in each major, 12 in all.
Psych HM MIS 3 3 5 3 1 3 2 1 3 1 0 4
Maximum of four from the following can count, worth 10 points each:
3. Circle "True" or "False": In general, comparing transformed data with the original "raw" scores, you can assume that the resulting F from an ANOVA would be unchanged.
True False
4. Circle the letter of the best answer: Which of the following approaches would represent the best compromise between controlling the Family-Wise Type I error rate, and maintaining adequate statistical power in an experiment involving 5 groups, where you want to contrast all possible pairs of groups?
A. Bonferroni-corrected multiple t-tests.
B. Fisher's LSD procedure.
C. Tukey's Honestly Significant Difference procedure.
D. All possible t-tests done separately.
5. Circle letters of all that apply: Which of the following would not be a test of an hypothesis about means of groups?
A. Analysis of Variance
B. Independent groups t-test (pooled variance version)
C. Levene's Test
D. Pearson's r
6. For Problem 2 above, (a) compute eta-squared and (b) very briefly indicate what it represents conceptually.
(a)
(b)
7. In boot-strapped estimation approaches to statistical inference, technically what type of sampling is involved? (Correct answer is not "bootstrap"!)
8. Circle letters of all that apply: Which of the following refers to a category or type of reliability in psychological test development?
A. Predictive
B. Test-retest
C. Internal consistency
D. Concurrent