FDR Settings

Submitted by Clguy on Sun, 07/17/2016 - 12:54

 

I have been using DPABI to conduct some statistical analyses of fMRI data.  I have used the viewer to see the results when FDR is invoked to address the p-value correction for multiple comparisons. With GFR in DPABI, one can enter a p-value for each voxel and then a p-value for clusters.  However, for FDR there is only a value for q entry.  When running FDR, is there a preset p-value for the voxels, and the q-value is for FDR of clusters?  Has the p-value for  the voxels when running FDR been set at z>3.1 as is suggested for GFR in your 2016 paper?
 
 
By chance, is there a manual that is available for DPABI users?  I have gone carefully through your paper in Neuroinform and the R-fMRI Course V2.0 Part1: DPABI 140826, but I am unclear when using FDR if that is for voxels or for clusters or both

Forums:

YAN Chao-Gan

Wed, 07/20/2016 - 06:49

Hi,

Here the the Benjamini–Hochberg procedure for FDR correction (https://en.wikipedia.org/wiki/False_discovery_rate) is used. Thus, there is no voxel-level preset p values.

We are trying to build a wiki for DPABI/DPARSF and eventually for R-fMRI. Here we encourage all the users edit the pages at http://wiki.rfmri.org.

Thanks,

Chao-Gan

 

Clguy

Wed, 07/20/2016 - 11:00





Thank you for replying to my inquiry.
Charlie



Sent from my iPad

Clguy

Wed, 07/20/2016 - 11:02

Thank you for the clarification.