we run a regression analysis on TBSS data with Randomise and our results are quite amazing.. even too much! We obtained a significant threshold of 0.99978 in the tbss_tfce_p_tstat and almost all the mean FA skeleton is significant.
Are this results real? Or do we observe a result caused by a problem in the analysis? We use 5000 permutations on a 45-subjects group with one interest and two confound regressors, perhaps do we need more permutations?
Thanks for any suggestion,