> The GLT input file has only one row in it (only one linear contrast). Then
> when we compare the F-statistic and t-statistic of the GLT in many
> voxels, the square root of the F-statistic value is not exactly the t-
> statistic value for that voxel.
How much is the difference? It would be more helpful if you provide your 3dDeconvolve script and that line of coding for the contrast.
Gang