Hi Rick,
Thank you very much! I have run the 3dANOVA again in the way that you suggested.
3dANOVA -DAFNI_FLOATIZE=YES \
-levels 3 \
-dset 1 participant_1_S1+tlrc \
-dset 1 Participant_2_S1+tlrc \
-dset 2 participant_1_S2+tlrc \
-dset 2 Participant_2_S2+tlrc \
-dset 3 participant_1_S3+tlrc \
-dset 3 Participant_2_S3+tlrc \
-ftr ftest \
-diff 1 2 S1vsS2 \
-contr 1 -0.5 -0.5
-mask group_mask+tlrc
-bucket contrast
My mask has 1005930 voxels (1*1*1 mm).
And whe I look the result of '-diff S1vsS2', p=0.0749, and q=0.9999. I do not know why.
Another question is the difference between the result of the 3dANOVA and the 3dttest. I also used the 3dttest to do 'S1 vs S2', and compared with result from 3dANOVA, when I get the similar activation map, p= 0.005 for 3dttest, and p=0.0749 for '-diff S1vsS2'. Why is there so great difference?
when I look the result of '-contr 1 -0.5 -0.5 ', p=0.0088, q=0.6593. Is it because that my mask is too big?
Best,
Pengmin