Another question on 3dMEMA for mixed-effects analysis: I'd like to try 3dMEMA on a two-way interaction test, but all the posts I've read on the forum indicate that you need to get the coefficients and t-stats for the within-subject contrast using a GLM in 3dDeconvolve/3dREMLfit. My problem is that my within-subject effect is still a between-scan effect: I have two groups (between), each of whom is scanned before and after an intervention (within). So for a given subject I have two 3dREMLfit outputs, one from the pre-stimulation scan and one from the post-.
Is there some way I can use the 3dREMLfit output from the single runs to get the appropriate inputs for the interaction test, or do I need to concatenate the data and do the GLT? As I write this post, the latter option seems less and less impossible, but still grungy given the way I've set up my analysis so far. (I assume that for a given regressor I'd code its pre- and post-intervention timing files separately, then do the GLT between pre and post... but other suggestions are welcome.)
Thanks in advance!