Hi Marco
> In my experiment I have 4 conditions, let's say R (rest), A, B and C. In the
> first level I made 3 contrasts: con_01 (A-R), con_02 (B-R) and con_03 (C-R).
> I entered these contrasts in a second level model.
> Now I would like to assess the activations specific to condition A for
> example, I'm wondering if it makes sense to specify a t-contrast like [1 0
> 0].
I think It would be better to do a separate analysis as a one-sample
t-test on just the (A-R) con* images. The way you describe I think
that you will end up inflating your degrees of freedom (since your
model includes the B-R and C-R images, which aren't being tested).
Hope this helps,
Jonathan
|