 |
 |
Dear Hieu Kim Huynh If the correction deems that the multiple-comparisons corrected T value should be ~0.001, then you will see similar results. FDR simply tells you (and sets) what statistical threshold is appropriate. Uncorrected activations at 0.001 (where FDR or FWE has revealed nothing) are very likely to comprise noise. Hope this helps! S. On 3 January 2012 14:59, Huynh Kim Hieu <[log in to unmask]> wrote: > Dear SPMers, > > > > For some contrasts, I found no thing in the brain with FDR p<0.05 (False > Discovery Rate - http://www-personal.umich.edu/~nichols/FDR/) but I found > some activation/de-activation with Uncorrected p-value adjustment to control > p<0.001. > > However, for some contrast, there is no difference between the FDR and > Uncorrected p-value. > > Can you explain me why and in which situation there are these results ( the > same and different results for FDR and Uncorrected) ? > > Thank you very much and Happy New Year, > > > > Sincerely, > > > > Hieu Kim Huynh -- Vicente Grabovetsky, Alejandro (Sasha) Postdoctoral researcher Donders Centre for Cognitive Neuroimaging http://www.doellerlab.com