First of all, nice job with the AFNI homepage.
I'd like to verify, if I'm thinking correctly here:
I'm doing an experiment, where I count the number of active voxels above a certain correlation threshold and try to calculate the theoretical p-value (without taking the lack of independece between tests into account).
I have 24 volumes in a series (sampling rate of 10sec). I'm omitting the first image, so I'm left with 23. This would then get me df=22 for the correlation and a critical correlation of 0.4 at p<0.05. Clustering my voxels into groups of at least 3 neighboring voxels improves my p-value to 0.000125, applying unadjusted Bonferroni corrections for ROIs of about 80 voxels gets me a p-values of 0.01 for the test. Any objections?
Thanks,
W.