Hi Gwenaelle,
My measure is VBM and I am interested only in differences among our three
groups. I've some nuisance regressors for example age, e.c.
I've three regressors of interest, one for each group. Do I have to use
ANOVA or T-test is enough as well?
If I used T-test I'd do the following:
The regressors of interest are binary. An example for 2subjects/group:
EV1(interest1) EV2(interest2) EV(interest3) mean e.c.
1 0 0 1
0 1 0 1
1 0 0 1
0 1 0 1
0 0 1 1
0 0 1 1
My question is: Do I need to put "the mean" regressor in the matrix if I run
randomise without -D? I think yes, but the Glm say at least one EV is linear
combination of the others. It's true, because mean=EV1+EV2+EV3, but I think
I need to demean the data. Is it right?
Contrasts:
EV1 EV2 EV3 mean
1 -1 0 0
-1 1 0 0
0 1 -1 0
e.c.
Is it right? And if not, could somebody help me how to make an accurate
design matrix for it?
Thanks a lot,
Gabor
|