Dear SPMers,
in my second-level analysis, I would like to calculate the interaction of a
factor (three independent groups) and a covariate (psychometric score). As
the factorial designs don't provide me with the interaction, I tried to do it by
basically setting up a GLM (centered covariate, centered factor level (-1 0 1)
and their interaction terms). Firstly, I am not sure how to test the interaction-
terms for significance correctly (for my hypothesis that group membership
impacts the effect of the covariate). Secondly, I am concerned about the
unequal sample sizes in each of my three independent groups. Obviously, the
regressors will correlate.
Is there any way to do this properly with SPM5?
Thanks in advance. Best,
Tim
|