hi kids,
I've been thinking about an analysis and i'm not sure how to go about it...
Imagine you have a blocked-design experiment with 3 conditions: A, B, & C. You do multiple regression for each individual subject. C is your baseline, so there are regressors for A & B in the model, but not for C. Thus, there is no beta weight for C. How would you go about doing a group mixed effects analysis to see where in the group A differs from C?
3dttest: you can't compare the means of A & C since there's no beta for C. how inappropriate would it be to compare the mean of A to zero without accounting for the variance of B?
3dAnova: if you wanted to do an ANOVA on beta weights, how is the variance attributable to C accounted for without a beta weight for C? what would a contrast where A = 1 and B = 0 mean?
thanks for your thoughts,
jill