It can happen with TFCE that significant t-tests don't show up in the F-tests, and vice-versa. One workaround to this is not do the F-tests at all, and instead correct across the t-tests. This is available in PALM with the option -corrcon.

All the best,

Anderson

On Tuesday, 1 August 2017, Lara Foland-Ross <[log in to unmask]> wrote:

Hello FSLers,

We are running an F test in randomize to identify structural differences between 3 groups in our VBM analysis.

randomise -i GM_mod_merg_s2 -o GM_mod_merg_s2_output -d design.mat -t design.con -f design.fts -m GM_mask -n 10000 -T

We’re not finding any areas of significance from the F test (using this command). However we are seeing differences between the groups in pairwise comparisons, which makes us second guess whether the above syntax for the F test is correct.

Any advice/input would be greatly appreciated!

Thanks in advance,
Lara