Why the TFCE results aren't identical to the ones produced by randomise?

There are two reasons for this:

Why are the degrees of freedom non-integers?

For the v and G statistics, the degrees of freedom can indeed be non-integers. These values, even if non-integers, can be used to compute parametric p-values by referring to the t and F distributions (if the parametric assumptions are met).


PALM/FAQ (last edited 13:13:21 26-08-2016 by AndersonWinkler)