Dear Bryn,
What does the surface_fdr script report for the "FDR threshold"?
If it is zero then that means that nothing passed the multiple
comparison correction.
If you are concerned that there is a problem then please feel
free to send us the relevant files via our secure upload server:
http://www.fmrib.ox.ac.uk/cgi-bin/upload.cgi
and send us the registration number.
All the best,
Mark
On 1 Sep 2011, at 22:44, Bryn Levitan wrote:
> Hello,
>
> I ran the T1 images of two groups (PBD vs HC) through segmentation using FSL- FIRST and then did vertex analysis. There were fairly large group differences for some of the individual structures (green-blue range coloring = higher statistical significance), but after running multiple comparison correction (FDR or Fthresh), all of the structures were solid red, which according to the instructions means that none of the differences were significant. The red color corresponded to a value of 2, however, on the scale when I view it in FSLview (not 0 as the guide shows). If all of my structures are entirely red, does this indeed mean that none of the differences between the group's structures are significant? Has anyone else had this problem where before multiple comparison correction the differences had a fairly large statistical significance that would presumably pass multiple comparison correction, but then after running multiple comparison corrections all of the structures showed no statistical significance? My group sizes are 47 for HC and 48 for PBD.
>
> Thanks for your help!
> Bryn
>
|