help > Functional connectivity network ROI FDR and not FWE
Showing 1-2 of 2 posts
Jan 9, 2021 08:01 PM | Aser A
Functional connectivity network ROI FDR and not FWE
Hi all,
I would like to ask why is the cluster based inferences on the functional network connectivity is set to FDR corrections and not FWE?
Thank you
Aser
I would like to ask why is the cluster based inferences on the functional network connectivity is set to FDR corrections and not FWE?
Thank you
Aser
Jan 10, 2021 12:01 PM | Alfonso Nieto-Castanon - Boston University
RE: Functional connectivity network ROI FDR and not FWE
Hi Aser,
That is a good question, and both options would be perfectly appropriate. CONN offers FDR in this context mainly because a FDR correction is often a bit more sensitive than a FWE correction, but also because in practice in order to compute FWE-corrected p-values for each network we would need to either make a few more assumptions about the dependence between the different network statistics (e.g. to use a Satterthwaite dof correction) and/or use non-parametric randomization/permutation methods.
Hope this helps
Alfonso
Originally posted by Aser A:
That is a good question, and both options would be perfectly appropriate. CONN offers FDR in this context mainly because a FDR correction is often a bit more sensitive than a FWE correction, but also because in practice in order to compute FWE-corrected p-values for each network we would need to either make a few more assumptions about the dependence between the different network statistics (e.g. to use a Satterthwaite dof correction) and/or use non-parametric randomization/permutation methods.
Hope this helps
Alfonso
Originally posted by Aser A:
Hi all,
I would like to ask why is the cluster based inferences on the functional network connectivity is set to FDR corrections and not FWE?
Thank you
Aser
I would like to ask why is the cluster based inferences on the functional network connectivity is set to FDR corrections and not FWE?
Thank you
Aser