help > Inconsistency - connectome ring vs 2nd level results table
Showing 1-2 of 2 posts
Jul 10, 2020 08:07 AM | Paul Summers
Inconsistency - connectome ring vs 2nd level results table
After running a second level analysis Roi to Roi analysis with a
couple of covariates,
moving to the results explorer, I've generated the connectome ring using alternative settings ROI-based inference parametric multivariate stats
As far as F stats and p-values these give the same results, but, whereas the table on the 2nd level results listing cuts-off based on pFDR 0.05, the
ring cuts-off based on both pUncorr <0.01 and pFDR < 0.05, which means it lacks a few areas in some analyses.
Is there a way to override the pUncorr limitation so that the ring is just showing pFDR < 0.05 ?
thanks,
Paul S
moving to the results explorer, I've generated the connectome ring using alternative settings ROI-based inference parametric multivariate stats
As far as F stats and p-values these give the same results, but, whereas the table on the 2nd level results listing cuts-off based on pFDR 0.05, the
ring cuts-off based on both pUncorr <0.01 and pFDR < 0.05, which means it lacks a few areas in some analyses.
Is there a way to override the pUncorr limitation so that the ring is just showing pFDR < 0.05 ?
thanks,
Paul S
Jul 10, 2020 10:07 PM | Paul Summers
RE: Inconsistency - connectome ring vs 2nd level results table
Tracked down to the first entry in the fifth tuple in the
assignment of data.thres_defaults in conn_displayroi.m (line
212 in my copy of 19.c)
Or setting the connection threshold p <� in the advanced family-wise error control settings tab for a less invasive solution.
Lovely (if extremely intricate) code by the way.
Or
From my side, I consider this resolved.
Paul S
Or setting the connection threshold p <� in the advanced family-wise error control settings tab for a less invasive solution.
Lovely (if extremely intricate) code by the way.
Or
From my side, I consider this resolved.
Paul S