East Carolina University
Department of Psychology

Four Variable Logit Analysis Homework: The 1992 Sexual Harassment Study

In the file "HARASS92.SAV", on Karl's SPSS Data Page, are data from a mock jury study. Consider the following categorical variables: VERDICT (1 = guilty, 0 = not guilty), GENDER of mock juror (1 = male, 0 = female) PLAINT, (1 = the plaintiff was described as being socially undesirable, 2 = the plaintiff was described as being socially desirable), and DEFEND (1 = defendant was described as being socially undesirable, 2 = defendant was described as being socially desirable). The female plaintiff in this civil case has accused the male defendant of sexually harassing her. We wish to determine whether our dependent variable, VERDICT, is affected by PLAINT, DEFEND, SEX, and/or any combination of these three categorical independent variables. Use HILOGLINEAR to conduct an initial screening analysis. Obtain tests of partial effects and conduct a backwards selection analysis. Use LOGLINEAR to conduct a LOGIT analysis (including only effects that involve the dependent variable) with a saturated model (including all effects that involve the dependent variable). Determine which effects have significant (.05) parameters and use that information (along with the results of the screening run in HILOGLINEAR) to construct a reduced model. Report the goodness-of-fit test statistics for reduced models you test. Use CROSSTABS to conduct unconditional Chi-square analyses of the two-way effects that you retained in your reduced model and to obtain conditional probabilities that will help you interpret such effects.

Write up your results in an APA-style results section and attach it to email to Karl. Be prepared to discuss your results in class and to show your statistical output to any interested parties.

snake on a stick

spider in web
Contact Information for the Webmaster,
Dr. Karl L. Wuensch

This page most recently revised on 18. February 2008.