It is a block design. The first scan had 13 subjects. Of those 13, only 9 completed the second scan.
We ran two deconvolution analyses. The first analysis was run on the first scan only (i.e., before concatenation). After this analysis, a single-sample 3dTtest was run on the deconvolution output. The second analysis was run on the concatenation of both scans. Again, after this analysis was run, we used a single sample 3dTtest on the output. The reason we feel that that the results are off, is that the two analyses produce drastically different results for conditions in the first scan.
We believe this may be due to the different number of subjects used in the two analyses. Is there a better way to compare these two scans?
Attached please find the script for the deconvolution analysis on the concatenated files.