I admit I'm probably a bit behind the times, but I just recently started using GLTs in 3dDeconvolve and I'm wondering about the difference between doing a GLT and doing a paired ttest. My old approach would be to run the regression and get the beta coefficients for my stimuli of interest (e.g. A, B). Then I would enter the betas for all my subjects into a paired ttest using 3dttest to produce a contrast of A-B across all subjects
Does this differ from doing a GLT of +A -B? Theoretically, couldn't I just run the GLT for each subject individually, and then use the betas for the GLT as the input for the group stats.
If there's little difference, which is the better way to run this contrast?