Gang,
I have been struggling with the same issue. Thank you for your thoughts. I actually complete agree with you and think that your approach allows for a more hypothesis-specific analysis strategy. I have received a hard time from more than one reviewer for this approach, but I still like it.
My follow-up question relates to the interaction coding for a t-test. You state"
"The F-test for the interaction has the same story: It basically tests against the same hypothesis as the t-test for (A1B1-A1B2)-(A2B1-A2B2) or (A1B1-A2B1)-(A1B2-A2B2), which can be obtained with 3dMEMA or 3dtest"
How do I code (A1B1-A1B2)-(A2B1-A2B2)? It seems I would need to use 3dcalc to make the (A1B1-A1B2) and (A2B1-A2B2) files before running the t-test. Is it valid to just subtract to data files like that (assuming each was created with the same 3dDeconvolve)? Is there another way to do it?
Thanks.
sean