Many thanks for the clarification!
I have one final question that is partially related to design but actually is more of a multiple comparisons issue:
We would like to focus our analyses specifically on 3 white matter tracts within TBSS: Uncinate, Cingulum, and Corpus Callosum. We have constructed TBSS-based ROI's for these tracts and will run 3 separate analyses using the design that we discussed before. We will use TFCE for the voxelwise inferences during each ROI analyses, but I suppose we also have to deal with the fact that we are running 3 separate analyses (thus increasing the rate of potential false positives).
I know that the simplest solution is probably the classical bonferroni correction of the corrected-tfce maps (0.05/3=0.016). However, these 3 analyses are not fully independent, at least, I am pretty sure that FA values in these tracts will correlate across participants. So, would a bonferroni correction modified/adjusted for the intercorrelation between FA in these tracts be a more accurate option? And would this be a valid approach?
Also, is it correct that number of ROI's becomes less of an issue when applying FDR correction? In other word, as long as the individual uncorrected-tfce-tstat maps pass FDR (q=0.05) it doesn't matter how manny ROI's are tested/selected? I have come across some posts on this forum and elsewhere that suggest this.
Would be great to hear your thoughts on this.
Cheers
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|