Hi Gang,
I am using the GroupAna.m script to run my fMRI group analysis, and I noticed that the 2nd order contrast results is a little "suspicious". Then later I used AFNI 3dANOVA3 as a comparison, and I noticed the t-stat subbrik of all my 2nd order contrasts are very different from the ones generated using GroupAna.m. However, the 1st order contrasts are fine.
You may also want to verify other higher order contrasts as well.
My design is a 3-factor "balanced" design with BxC(A).
Factor A = Group with 2 levels
Factor B = task with 2 levels
Factor C = task with 3 subjects for each group. (one coworker of mine ran his data with 16 subjects per group as well and he discovered the same issue as I mentioned above, so I guess this problem is not due to the number of levels in Factor C)
For the 3dANOVA3, I am using AFNI version:
Version AFNI_2005_12_30_0934
[[Precompiled binary linux_gcc32: Jan 31 2006]]
For the GroupAna.m script, I downloaded it from the link below earlier this year:
[
afni.nimh.nih.gov]
Thanks very much.
Cheers,
Ricky