Hello,
I have a question about when to do contrast during analysis.
Within one functional scan we have two conditions/stimuli, A and B.
It seems we have two options:
1) We could run a simple glt contrast (1 -1) at the individual subject level with 3dDeconvolve. We could then pull out the intensity difference between the betas at each voxel, warp them to tlrc space and run a one-sample t-test across subjects with those differenced intensity values
or
2). Pull out the betas for each condition after running 3dDeconvolve at the individual subject level, warp them to tlrc space and run a paired t-test across subjects between conditions A an B.
It seems that it should produce the same result (or not?). However, i tried both methods. The Activation maps look quite similar i most areas in the brain, but they are not exactly the same. And in some areas of the brain there are noticeable cluster size differences when comparing the activation maps resulting from carrying out the two methods described above.
My question is, which method is the better approach. Is there any evidence in the literature to suggest which method is better? Your help would greatly be appreciated.
Thanks,
Michael