Hi,
no need to weight the contrasts: 1 -1 is just fine. Feat takes care of it.
Cheers,
Andreas
________________________________________
Von: FSL - FMRIB's Software Library [[log in to unmask]] im Auftrag von John Dewey [[log in to unmask]]
Gesendet: Dienstag, 21. September 2010 23:45
An: [log in to unmask]
Betreff: [FSL] unequal cell sizes
Hello,
Does anybody know how fsl handles contrasts between events with unequal cell sizes?
Example:
I want to find areas showing greater activation to Event B than to Event A. Every participant is in both conditions (paired samples), but by design Event B occurs less frequently than Event A, because it's supposed to be a surprising event. My question is, if I set up a simple 1,-1 contrast (method of subtraction), does FSL do some automatic differential weighting of the means or otherwise account for the difference in cell-sizes in an appropriate way? What is the usual way of handling this issue in fMRI analyses?
-John
|