I have what I hope is a super simple question but I want to make sure I'm thinking about it right. I have a design matrix with columns:
g1, g2, beh@g1, beh@g2, cov1, cov2, cov3
where beh@g1 and beh@g2 are individual columns for a behavioral measure per group.
I'm using TFCE to look at group and behavior in VBM.
I know that the contrast [0, 0 , 1, -1, 0, 0, 0] will give me the difference in slopes (g1 > g2). If my interaction term is not significant and I look at the contrast [1, -1, 0, 0, 0, 0, 0], will this give me the group effect (i.e, difference in intercepts, g1 > g2), accounting for my covariates, interaction term, and main effect of behavior?
Thanks in advance.
|