I don't think there is a standard way to do this. A vertex-wise analysis is not the same thing as a averaging over a group of vertices. I guess you could constrain your post-hoc analysis to be within the main effect cluster; that would be most consistent. But I don't think you'd have any problems getting it published either way.

On 12/2/2019 3:12 PM, cody samth wrote:

        External Email - Use Caution        

Hi,

I have a statistical question about how to approach reporting results from FreeSurfer analyses containing three groups. I ran a group effect (F-test) and then post-hoc tests looking at pair-wise comparisons between the three groups. My question is why is that we run separate vertex-wise analyses for the post-hocs rather than extracting the values from significant clusters and running post-hocs in a statistical software for just the regions where a significance difference was found (ie the clusters)? As a post-hoc vertex wide analysis can lead to different results.

For example in my group effect contrasts I found a cluster in the parietal lobe. Whereas in my post hoc-vertex wide analyses one of them found two clusters 1) within the parietal 2) within the frontal lobe. If I choose to run these post-hoc analyses via freesurfer (ie vertex-wise) rather than extracting the results to analyze in a statistical program, is it standard to report the second cluster? Even though it didn't come up in the group model? If so is there a paper that people reference that uses this approach?

Thanks,

Cody

_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer