Hi Gang-
Thanks for the quick response. In this case, I think the t score corresponding to the beta would be converted to a z, and then these values would be used in a 2-factor mixed effects ANOVA. I am inclined to use the approach advocated on your site, which is to do the ANOVA on percent signal change. However, I would like to have a good reason rather than "that's the way AFNI says to do it" for choosing one over the other. As I said, I have seen it done the other way, and I can think of some reasons why the t (or z) approach seems wrong (maybe it does not provide an accurate estimate of error related to individual subject differences?), but I would like to defer to the statistical experts for a second opinion.
Anthony