Hi Conn Community,
I've been working on a gPPI analysis in conn contrasting two different conditions for some time now, and have some potentially interesting results. I've come across something that I'm curious about and was hoping for some input.
When I look at the ROI to ROI connections (I am looking at Schaefer parcellations in 3 networks specifically), I have found some sig. and interesting clusters using TFCE correction in the Results Explorer. The problem is that while all of the clusters are very significant, if I expand the statistics, most of the connections themselves are not (they are FDR corrected around p=0.05 and a bit higher in the results explorer). Are the sig. clusters themselves enough to draw conclusions or the connections also need to be sig. ? I thought TFCE in the results explorer would also correct the edges and show only sig. connections as well but if the table is correct then it does not.
Please advise if possible and thanks again.
Best,
Kylie I.
I've been working on a gPPI analysis in conn contrasting two different conditions for some time now, and have some potentially interesting results. I've come across something that I'm curious about and was hoping for some input.
When I look at the ROI to ROI connections (I am looking at Schaefer parcellations in 3 networks specifically), I have found some sig. and interesting clusters using TFCE correction in the Results Explorer. The problem is that while all of the clusters are very significant, if I expand the statistics, most of the connections themselves are not (they are FDR corrected around p=0.05 and a bit higher in the results explorer). Are the sig. clusters themselves enough to draw conclusions or the connections also need to be sig. ? I thought TFCE in the results explorer would also correct the edges and show only sig. connections as well but if the table is correct then it does not.
Please advise if possible and thanks again.
Best,
Kylie I.