I used 3dClustSim to calculate familywise corrected p-values for clusters, and according to the help text the function by default uses one sided uncorrected voxel-wise p-values to define the clusters. So far so good. However, when I open the AFNI GUI and set the threshold slider to p = 0.01 (which I also used as a voxel-wise threshold for 3dClustSim), the corresponding test value is about 2.70. As I'm looking at results from 3dttest++, I assume that this is a t-vale. When I double checked the threshold in Matlab (using icdf), it seems like 2.70 is the threshold for a TWO-sided test of p = 0.01 (for about 40 degrees of freedom). A one sided test would give a t-value of about 2.4 for p = 0.01.
This made me a bit confused, should I use 2.7 or 2.4 as the voxel-wise threshold?
Is it possible to get one sided p-values in the GUI? I tried checking the box "pos" (for positive) but it did not have any effect on the p-value.
