Ziad and Fred,
Thanks for the suggestions, but I don't think that either of these programs will get me what I need. 3dTcorrelate only deals with 2 datasets at a time, and I want to correlate across all 9 subjects. 3ddot also only deals with 2 datasets at a time, and in addition, only gives you one number, but I want a correlation at each voxel.
Let me try to be clearer about the purpose of my request. I would like to see how the reliability of the task varies across voxels. That is, in my group of 9 healthy individuals, how correlated are fit coefficients at Time 1 and Time 2? If I had a particular region of interest, I would use 3dROIStats to get an average value in that ROI for each participant at each time point, and then use a statistics program to correlate Time 1 and Time 2 values (with n=9). I want to do the same thing at the level of the individual voxel, but would like to avoid having to dump all 20,000 voxel values into SPSS. As I mentioned before, I suspect that there is some relationship between the correlation that I want and the voxel-wise results of the paired 3dttest, but I'm not sure exactly what the relationship is...
We are working on "hand-calculating" the correlation using a complicated 3dcalc statement, but I wanted to see if there was an easier way before we invest too much more time on it.
On a related issue, we've also been thinking of trying to calculate the intraclass correlation between the datasets. This requires knowing the Mean Square Within and Mean Square Between from the repeated measures ANOVA. Are these calculated as part of the 3dANOVA programs, and if so, is it possible to have an option to save them in a BRIK file?