Well, my design is a little unusual, as I am looking at changes in mental state.
The first scan is 24 minutes long: 8 minute baseline, 8 minutes of state A, then another 8 minute baseline. The questions are: 1) is state A different from baseline, and 2) Are the 2 baseline the same, or has state A had carry-over effects into the second baseline. I treated the second baseline as a different state (eg made a different column for it in the .1D file), then calculated the regression coefficients for it and for state A, as well as a glt for A vs the second baseline.
The second scan is 40 minutes long. There is an 8 minute baseline, 8 minutes of state B, then 24 minutes of state C, which we arbitrarily divide into 3 8-mintue blocks for analysis (we expect some things to change and some to stay the same across the 24 minutes). States B and C are counterbalanced across subjects. The questions are: what does each of the states (B, C1, C2 and C3) look like relative to baseline, and what is different between each of the 4 blocks. I created coefficient maps for each block, as well as glts for B vs C1 or C3; and C3 vs C1.
For both scans I would like to report group effects, including statistical significance and percent change. As for which statistical test to use, I don't know enough stats to know which is the proper one (I thought I could just do a bunch of t-tests, but apparently that is not the case).
I have just re-read the message from Christine, and she also suggested making a conjunction map to find regions of activation in common across the group. Would this be an acceptable alternative to a stat test? I presume not, but it would certainly be a way around the problem of the differences between significance in individuals vs group maps (ie .0001 in individuals but only 0.1 in the group map).
While I have your attention, I might as well ask a few more questions. What is the best (or usually done) order of steps post deconvolution? Specifically, does it matter whether I smooth or Talairach transform first (and when best to add teh -dxyz=1 step)? I have tried different orders and seen slightly different maps, which is disconcerting, to say the least. Any advice?
Thanks much!
Sara