Hi Gang,

Thanks for your reply.

I am converting my statistic to z scores because I read the following posting by Doug Ward:

"for both the t-test and ANOVA, it is assumed that, under the null hypothesis, the input data is normally distributed. Now, this is true for the estimated parameters ("amplitudes"). However, this is definitely not true for the "statistics". So, if you are going to use a "statistic" as input to 3dttest, 3dANOVA, etc., it will be necessary to first
normalize the "statistic" (i.e., convert to z-score). This can be done using program 3dcalc. "

As for the AlphaSim, I plan to perform the clusters on the tlrc data for the group (threshold determination based on Orig voxel), and then convert the z-values back to p values. Please let me know if this is wrong.


Jasmin