I realize all of the AFNI issued MATLAB scripts are to be used at one's own risk, but I wanted to raise a question about what seems to be an erroneous calculation in the DF for within-subject contrasts for two independent groups.
I used GroupAna to compare to unbalanced groups A (n=18) and B (n=13) in an experiment with 2 conditions. When you check the header info for the GroupAna output file it lists the DF for condition X versus Y in group B as 29, when it should be 12. If you run the specific contrast of X and Y for group B using 3dANOVA2 and compare the t statistics generated via GroupAna and 3dANOVA2 (see attached image) you can see that there is a discrepancy in the t-statistics, while the mean difference between the two conditions generated through each program is identical.
Vinny
