Dear Anderson,
Thank you very much for your helpful answer.
In the meantime another question arose: using a f-contrast in a shape analysis an effect showed up. Checking the time series of the 4d file as described in the UserGuide to determine the direction of change, it pointed towards an atrophy of the patients (first 18 Inputs in the matrix). However, when I wanted to confirm this effect checking the t-contrast nothing showed up (in fact the value was 0).
I used the matrix/contrasts from the attached file shape2.fsf and the following randomise options: -D -c 3 -F 3.
Am I mistaking the signs of the EVs/contrast and/or is a t-contrast less sensitive for cluster-wise analysis?
Any help would be really great!
Best regards and thank you one more time in advance
Falk
|