Dear SPMers, I know this question has been asked before, but it doesn't seem that there is a lot of recent consensus. I have done a second-level analysis in SPM2. I was interested in seeing all of the activated voxels in one of my contrast images, so I lowered the threshold to T = 1.1 (1.0 results in a crash, incidentally). When I did that, I noticed a LOT of activity outside the brain. I wondered if that activity outside the brain might result from smoothing, and if so, if masking it out would help boost power. If masking it out would help boost power, would it be reasonable to mask the first-level contrast images, and then include the new, masked contrast images in the second-level analysis? Or would the approach discussed in 2002 by Stephan Kiebal (via Tom Nichols, http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind02&L=SPM&P=R36174&I=-3) be better? (Below) (1) Configure, but don't estimate a model (2) Modify SPMcfg.mat... load SPMcfg.mat xM.TH = -Inf; xM.VM = spm_vol('mask_all.img'); save SPMcfg xM -append (3) Estimate the model Any suggestions, including those not mentioned above, would be appreciated. Thanks in advance, Christy (sorry for the repeat message... I hit "send" before the last version was ready) -- Christy Marshuetz Assistant Professor Department of Psychology Yale University 2 Hillhouse Avenue P.O. Box 208205 New Haven, CT 06520-8205 Phone: (203) 432-4636