Dear SPMers,
I've conducted a multi-group : conditions and covariates analysis with MRI
data, with two conditions (baseline and follow-up) and two groups of
patients. I'm interested by the group x condition interaction (ie the
between group difference between baseline and follow-up = contrast 1 -1 -1
1), but also by the effect of condition in each group (ie contrasts 1 -1,
and 0 0 1 -1).
My problem is the following: I need to correct for the effect of age. With
this kind of model, such nuisance variable is implicitely modelled : when
assessing group x condition interactions, potential effects of the
nuisance variable cancell. However, it seems that when assessing the
effect of condition in each group, differences between baseline and follow-
up reflect both the effect of the pathology, and the effet of age if age
has not been included in the model... Is there a possibility to
distinguish between these two processes ?
Thanks a lot,
Gaël.
|