I am testing small regions for activation using masks, but SPM often shows some voxels and clusters significantly negative, while other voxels or clusters in the region with significantly positive activation. How do I combine all the t-tests in the small volume to test whether the region overall has significantly higher activation or significantly lower? I have searched the statistical papers of Friston but have found no such statistic, and I have searched the SPM discussion list and have found nothing. Thanks for any help! Mike Michael Hagerty, Professor Emeritus University of California, Davis http://faculty.gsm.ucdavis.edu/~mrhagert/