Review Session Questions

A few questions from the review session were raised in class today, and I just wanted to clarify these items.

1. Multicollinearity is just as you learned in regression. It refers to predictors in a regression being too highly correlated with each other. In ANCOVA, the regression that is computed is the regression with the dependent variable being predicted from the covariates. Thus, multicollinearity can only occur in ANCOVA when the covariates are too highly related to each other. The IV and the DV are both irrelevant for the examination of multicollinearity.

2. Having said that, it is also true that the covariate should be unrelated to the IV. This, however, is not multicollinearity, but independence of covariate and IV.

3. You should report a measure of central tendency (i.e., the mean), a measure of variance around that mean (i.e., SD or SE), and an n for the cells you actually analyzed. This means that in the second homework problem, you should have reported the means, SD, and n for the cells defined as PREP at Time 1, PREP at Time 2, Control at Time 1, and Control at Time 2. You should not have reported the marginal means for PREP (averaging over time) or Control (averaging over time) or for the total sample (averaging over both group and time) or for time 1 (averaging over group) or for time 2 (averaging over group): none of those are the model you actually analyzed.