Sorry, Gang, I just thought of something else that is closely related to our conversation above, but specifically about TENT models. In slide 6 of the "Deconvolution Signal Models" talk from bootcamp, it says:
"For input to group analysis, usually want to compute average signal change... In previous slide, with 6 Beta weights, average signal change is
1/2 B0 + B2 + B2 + B3 + B4 + 1/2 B5
... [beta weights] can be combined into a single number using 3dcalc."
My main question is this: To combine the beta weights into a single number for the purposes of group analysis, should they simply be added (e.g., 1/2 B0 + B1 + B2 + B3 + B4 + 1/2 B5), or should they be averaged (e.g., (1/2 B0 + B1 + B2 + B3 + B4 + 1/2 B5)/6)? The slide seems to imply that they should simply be added. But, prior to noticing this, I've been averaging them.
Is there ultimately any difference between adding and averaging the beta weights, when it comes to a standard group analysis (using the 3dANOVA* programs) that would recommend doing it one way over the other? If not, what is the standard way of doing things, adding or averaging?
(I know that you maintain that you should look at the shape of the HDR curve as revealed by TENTS rather than combining the beta weights. I'll do that for a future analysis. But I would like to start by combining them, either by adding or averaging, whatever you recommend.)