Objectives:

     1) Analysis of Covariance (ANCOVA)

     2) testing for interactions

Objective 1: The Analysis of Covariance (ANCOVA)

 You may want to refer back to lab #8 where I first introduced ANOVAs. This week, we're going to add a covariate and run an ANCOVA. In addition, to make things more interesting, we'll also add another independent variable to the model. As before, you will need to used the Jam/U.S. data set for this objective.
 

• Get means and SDs for variable AGGRESS (i.e., aggression) by CULTURE and SEXCB (separately)
• Run an ANOVA with CULTURE and SEXCB as the independent variables and AGGRESS as the dependent variable. Since this is not new material, I am not providing the SAS code.
• Now, rerun the same analyses, but this time use AGE as a covariate. Substitute the following model statement for the one you had in step #2.

• Look at Type III output. Does partialing out AGE make a difference? How do you know?
• Run a post-hoc test to determine the direction for all significant effects (see lab #8 for details)

Objective 2: testing for interactions So far, we've only looked at main effects in all our analyses. However, as all of you know, sometimes there are interactions between two or more different variables. It's important to test for these interactions. Here's how to do it with PROC GLM. Please note that the procedure is different for PROC REG.
 

• Go back to your ANCOVA in objective 1, step 3.
• Substitute the following model statement for the one you had.

• As you can see, I asked for all possible interaction combinations so that you can see how they are specified. However, there is also a shorter way to get all interactions. I'll show it to you in class.