I think the resounding answer will be that you should not use the uncorrected statistics, and you should never "play" with the thresholds. Stick with p<0.05 (or something else if you can justify it).
From: FSL - FMRIB's Software Library [[log in to unmask]] on behalf of Amandine Pelletier [[log in to unmask]]
Sent: Monday, April 08, 2013 10:49 AM
To: [log in to unmask]
Subject: [FSL] images corrected and uncorrected for multiple comparisons
Dear FSL experts,
I have a question concerning TBSS results corrected and uncorrected for multiple comparisons:
Using the image corrected for multiple comparisons (called tbss_tfce_corrp_tstat1), we can play on the p-value image with fslview. For example, when I set min=0.99 and max=1, the image corresponds to the results thresholding p<0.01.
As the same manner, is it correct if we play on the p-value image with the image uncorrected for multiple comparisons (called tbss_tfce_tstat1) ?
Thanks for your time and your help,