Statistical Methods for Psychology

Exercises 657

This chapter was based heavily on a study by Pugh (1983) on the blaming-the-victim phe-

nomenon in rape cases. The complete data from Pugh are given below, adding the Gender

of the judge as the final variable.

Stigma (S) Verdict (V)

Gender (G) (Moral Character) Fault (F) Guilty Not Guilty

Male High Low 17 4

High 11 7

Male Neutral Low 36 4

High 23 18

Male Low Low 10 6

High 4 18

Female High Low 25 0

High 12 4

Female Neutral Low 43 8

High 42 23

Female Low Low 22 2

High 13 6

17.16Run the complete analysis on these data. What effect does adding Gender to the analysis

produce? You can compare your conclusions against the results given in Pugh’s paper.

17.17 Agresti (1990) presents data on the relationship of the assignment of the death penalty, the

defendant’s race, and the victim’s race. The data follow.

RaceDefendant RaceVictim DeathPenalty Frequency

11119

1 1 2 132

1210

1229

21111

21252

2216

22297

Note: For Race, 1 5 White; 2 5 Black. For DeathPenalty, 1 5 Yes; 2 5 No.

a. Fit an appropriate model.

b. Summarize the conclusions you would draw.

Discussion Questions

17.18 If you search on PsychINFO under “log-linear” you will find many studies that used such

models. Find an interesting study that contains the necessary cell frequencies and write up a

short example that can be used to illustrate the material covered in this chapter.

17.19 Apply the same data to two or more software packages and note the similarities and differ-

ences in the output. How, if at all, can you resolve the discrepancies?