Hi,

I have 'n' subjects and each has 'm' sessions (Multi-Session & Multi-Subject) and I want to do group average using FEAT. According to FSL instructions, I should do three level analysis modelling the "within-subject repeated measure" in the 2nd level and the "between-subjects variance" in the 3rd level. 

My question is regarding the 2nd level analysis since the GUI is time-consuming especially for large group sample. I am wondering what is the formula of the resulted COPE and VARCOPE [that result from averaging the 'm' sessions for each subject (Fixed Effects)], so that I can make a script to automatize the calculation instead of the 2nd level GLM feat (using GUI) and then directly pass the output to the third level. 

Thanks,

--Meena