[Freesurfer] mri_glmfit-sim question

Dear FS experts,

I am trying to do a paired analysis between 2 time points obtained from the longitudinal pipeline and had some trouble interpreting the results. I have 2 different treatment groups I and II. I ran the paired analysis at p = 0.05, p = 0.01 and p = 0.001 vertex-wise thresholds. For p = 0.05, group I shows greater differences between the 2 time points whereas at p = 0.01 and p = 0.001, group II shows greater differences between the 2 time points. Is there any reason that might explain this discrepancy? If not, which result do I trust? The sample size in both groups is comparable (approximately 120 subjects in each group).

Thanks Koushik