Hello Gang-
We have run the 3dcalc program for fixed effects analysis you suggested above. For comparison, we also re-ran the individual subject deconvolutions in python combining the regressors (correct and incorrect) at that stage for comparison. Viewing the results, we see some differences (please see data we will upload to AFNI site, comparing "weight" files with their counterpart (the python generated combinations). Before we proceed with the remainder of our analysis, we'd like to ask if these results (and the differences) are reasonable to you. If so, do you know if there is any published study we could site in our methods section to support use of this weighting approach?
Many thanks,
Elizabeth