 |
 |
Hi Anderson, Are you arguing that randomise should not be used to do group comparisons of tractography? If my tractography data were categorical (i.e that a tract either passes through a given voxel or not for a given group), then I would understand your point. However, I looked at the 4D concatenated file of fdt_paths, and they all overlap. They simply look more or less robust, as I would expect given that my ROIs are relatively close together anatomically. In this case, randomise should have worked because it is a parametric test, so I still don’t understand how it's possible that no subsets of a significant comparison are significant. Thanks, Zak