Nic,
One possible approach is to run group analysis using their ranks across subjects. In doing so, the results would be insensitive to the extreme values or outliers. How many groups do you have? And are you comparing two or more conditions, or just one condition? For example, with one or two conditions, you can try 3dWilcoxon.
Gang
Edited 1 time(s). Last edit at 08/11/2014 04:46PM by Gang.