The data in Table 12.21 consist of hypothetical aggressiveness scores for 6 monkeys who have been exposed to three types of drugs (A, B, and C). In order to control for potential order effects, a Latin square design is utilized. In particular, two randomly constituted squares are formed with three monkeys randomly assigned to each square and also randomly assigned to the particular row of the square that assigns the order that they are exposed to treatment

The hypothesis of primary interest is whether the drugs differ in their effectiveness in controlling aggression in the monkeys. Thus, the test of the condition main effect is of primary interest. The current procedure reproduces the results for test of the condition main effect shown on page 615.

1. Click Analyze, then General Linear Model, then Univariate.

2. Place the dependent variable (dv) in the Dependent Variable box,time and cond in the Fixed Factor(s) box, and subject in the Random Factor(s) box.

3. Click Model in the upper right hand corner of the Univariate screen, and then click Custom in order to specify the model of interest. Since the analysis of Latin square designs does not involve interactions, under the Build Terms box in the middle of the screen click on Main Effects. Then place time, cond, and subject in the Model box on the right hand side of the Univariate:Model box. Click Continue and then OK.