Hi Anderson,
thank you - that is very helpful!
For that covariate with data in both patients and controls (IQ or task performance), I don't think of a confounder, but rather of the first two of you options...
- Is there a relationship between activation and IQ irrespective of the groups?
--> for this, I guess, I need to put the covariate in a single EV - but I think that is not possible, because I have assigned patients and controls to different groups in the group column?
- Is there a different relationship between activation and IQ between the two groups?
--> here I will split the covariate, and then I can analyse the interaction (with a contrast like [0 0 0 0 1 -1]) - is that right?
Best wishes!
Viola
|