Dear all,
 
I am performing a ROI-to-ROI analysis on resting-state data from children and adolescents in Conn (40 in total).
I am comparing connectivity in these two groups in 16 ROI's using the contrast [-1 1] (so ado>child).
When I select all ROIs for this contrast and go to the results explorer, it depicts all significant pairs.
I am not sure whether I understand the content of the depicted table in the results interface.
You see, I would expect that if the connection from seed A to seed B differs significantly between the two groups (say T(39)=3.23, p-FDR=0.0381), that the same holds for the connection from seed B to seed A.
This is not always the case in my table, however. Sometimes it says that connection C-D is significant, while connection D-C does not get depicted in the table, from which I conclude it was not found significant.
 
How to interpret/deal with this?
 
Thank you in advance for the advice!
Kind regards,
Mathilde

I was wondering if someone can help me with this, as I still haven't figured it out myself.
I hope I posed my question clearly!

Hello,

Conn does a seed level FDR correction. For a given seed, the correction would depend on all the other seed-to-target correlations (actually Fisher transformed correlation coefficients) for that particular seed (from which you would get your T and p values). Therefore, A-B correlation may be significant when looking at seed A but not vice-versa when looking at seed B as the correction applied is different. Hope that clears it up!


Best
Pravesh
Originally posted by Mathilde Kennis:

Here is an old post where Dr. Alfonso has given a detailed explanation: https://www.nitrc.org/forum/message.php?msg_id=5706


Best
Pravesh

Originally posted by Pravesh Parekh:

Thanks a lot, Pravesh!
Also for digging up that post for me, I was unable to find it myself for some reason.

Cheers,
Mathilde
Originally posted by Pravesh Parekh: