afni.nimh.nih.gov

> how can we run t-tests only on clusters that are significant in the whole-brain
> anova, rather than using whole-brain t-tests? This is what the reviewers for
> one paper were asking.

Alessandra, if you have the mask already (e.g., from your whole-brain ANOVA), just run t-test on the whole brain, and then pretend you don't see those voxels you're not interested by masking them out:

3dcalc -a t-testResult -b mask -expr 'a*b' -prefix ttestMasked

Gang