If the behavioral measure is different between the two time points, it would be impossible to handle the situation in the framework of the traditional general linear model such as ANCOVA, and linear mixed-effects modeling is the way to go. However, LME is definitely impractical under 3dGroupInCorr with the current computational power.
Gang
Edited 1 time(s). Last edit at 09/24/2014 04:41PM by Gang.