Dear Alexander,

In the second-level results tab for ROI to ROI analyses there are two options (in the top right corner), that indicate whether you want to perform an analysis looking only at connections with an individual seed ROI ('GLM analysis results: individual ROIs' option), or if you want to perform an analysis looking at all connections between ROI pairs ('GLM analysis results: all ROIs' option). When selecting the former ('individual ROIs') the FDR multiple comparison correction is applied only across the multiple target-ROIs within each analysis separately (i.e. if you want to repeat those analyses looking at different seed ROIs, that is not contemplated/included in the multiple comparison correction). If you want to be able to correct simultaneously for multiple comparisons across all connections between ROI pairs, select instead the second option ('all ROIs'). That will offer several options to perform that multiple comparison correction, including a simple FDR multiple comparison correction across all connections between ROI pairs (see 'ROI-to-ROI cluster level statistics' section in https://web.conn-toolbox.org/fmri-method... for details).

Hope this helps

Alfonso

Originally posted by Alexander Groessing:

Dear CONN experts,

I am new to CONN and am currently doing an ROI to ROI analysis of a baseline. I correlate with behavioral outcome (SRS 2) and control for age. Between subject contrast [ 0 0 1]

We picked all possible ROIs (164) and chose all ROIs as targets in the 2nd level analysis. 

I'm wondering how the correction works precisely. We chose FDR at 0.05.

Is it corrected per source for 164 targets, which would be 164 tests, or for 164 sources, each having 164 targets? There would be some false positives in the first case, and we would need further correction. 

I should add that I go through the seeds in the list individually, but I am unsure of how CONN handles these corrections.

 

Thank you very much for your assistance.

Best regards

Alex