Dear experts,
I have some questions regarding the multiple-comparison correction in dual-regression. I did the dual-regression in order to analyze the resting-state of two groups. I am interested in the question in which regions of which RSNs these two groups might differ, when adjusting for the covariates age, gender, handedness and education. First, I chose to extract 30 components using concat ica in melodic. 10 out of these might be of potential interest.
Primarily, I am analyzing whether there is any difference in activity in any network between these two groups. So, (using fslview) I am looking at the output from dual-regression for each component one by one. Do I have to set the significance threshold to 0.05 (so the min./max. in fslview to 0.975/1) OR do I have to correct? When correcting, I would get 0.05/(10*2) = 0.0025 (min./max. = 0.9975/1)?
Kind regards,
Stefan Lenzen
|