Dear all, I'm performing a two-Sample Paired T-test VBM using a statistical analysis with randomise TFCE. 1) Results obtained by estimating the 2 paired differences A-B and B-A showed to be different from each other (what was expected) but one was not the opposite of the other: why? 2) More generally, how can one optimally choose the number of permutations in randomise TFCE? Thanks in advance Arsene