Gym Attendance and Body Weight in Female and Male College Students

Gym Attendance and Body Weight in Female and Male College Students

Jenn Selensky gathered data from students in an introduction to psychology course. The data with which you will work for this assignment are weights, sex/gender, and whether or not the student worked-out in the gym. You will conduct a factorial ANOVA, evaluating the effects of sex/gender, gym, and their interaction. The cell sizes are not equal and not proportional, so be sure to use Type III sums of squares. There was some confusion about the coding of sex/gender. I coded the sex/gender with the distinctly larger weights as male.

First, conduct a contingency table analysis relating sex to gym. You will note that the obtained value of 2 is greater than zero, indicating that the two factors are not independent of each other.

Second, conduct the Factorial ANOVA, relating sex, gym, and their interaction to weight. Obtain means, sample sizes, and standard deviations for each cell and for the marginals. For each of the three effects in the omnibus analysis, find 2 and obtain a 90% confidence interval (semipartial2, not partial 2). Test the simple main effects of gym attendance separately for men and for women using the pooled error term. In SAS one uses the LSMEANS command with SLICE to do this. I’m not sure how to do this with SPSS, but I suspect it can be done. It might require the create use of contrasts. Obtain the LSMEANS for both sex and for gym. These will be estimates of the marginal means were sex and gym independent of each other (which they are not). SPSS calls these “adjusted means.” Compare the LSMEANS with the unadjusted means.

Third, test the simple main effects of gym at each level of sex using individual error terms. In SAS, this involves using Proc Sort and By statements. In SPSS it involves splitting the data by sex. For each of the two simple main effects, compute 2 and obtain a 90% confidence interval (semipartial2, not partial 2). Since there are only two levels of gym, an alternative here would be to use Cohen’s d (with CI) as the magnitude of effect estimator for the simple main effects.

The data are in a SPSS sav file on my SPSS data page. The file is named SexGymWeight. SPSS save files are easily imported into SAS.