As a follow up to the previous questions I posted I want to ask an additional related questions. I have previously been calculating ANOVA's on the group data using 3dANOVA2. However, performing group analysis on functinal data in this way has resulted in unecpected results. I think this is the reason why but I want to check with the experts to be sure.
If a statistical comparison is performed on the intensity subbrick of an fim, variability information is lost. Similarily, if group level statistics are performed on the threshold subbrick, amplitude informaiton is lost. Performing the statistics within each subject using regression takes both these varaibles into account.
Does this make sense to anyone?