Hi,

This issue has been bothering me for a while. I'm not sure if statistics allows a solution, but here's the problem: We have a dataset where ~30 subjects were scanned a week apart on two stimulus orders of the same task. I want to determine the activation common to the two sessions - we don't care about the differences across Scan 1 and Scan 2.

If subjects were only scanned once, 3dttest would be the answer. However, each subject is repeated, so if my understanding is correct, using 3dttest on the Scan1 + Scan2 dataset is not valid because of this repetition - the result will come out better than it would if we'd have scanned, say, 60 people once.

So, is it possible to perform a test in AFNI to find the common activation (greater than 0) while controlling for the subject fixed effects / repeated subjects with shared variance? All the 3dANOVA programs do not appear to yield anything, because the best one can get is the activation in condition A (Scan 1) and condition B (Scan 2) separately.

To get around this, we have been running 3dttest on Scan 1 and Scan 2 separately and conjoining the results (w/ 3dcalc), but it would be GREAT if it is possible to get a more rigorous test of significance and a more realistic idea of the activation in this task. Or just an answer that says this isn't possible! Let me know if I can clarify anything.

(I realize that one solution would be to concatenate the datasets across sessions and run 3dttest on that data. We have tried some methods as suggested on other board postings, but the results are not at all satisfactory (i.e. the datasets from each subject are always imperfectly joined))

Thanks in advance for any help!

elliott