Hi Christine,
> What is needed is a 3dMEMA version of 3dRegAna. :)
I thought about that a while ago, but haven't taken any action yet.
> I think I am then going to return to the idea of coding a single vector with condition
> number (i.e., 1, 2, or 3), convolving it with a HRF, and then using 3dDeconvolve to
> get a beta and t-score for this vector. Then I would put those betas and ts into a
> one-sample, t-test using 3dMEMA.
Unfortunately it seems to me that would not work. The reason is this. A set of arbitrarily equally-spaced numbers (e.g., 1, 2, and 3) works in 3dRegAna because there is an intercept present in the model that takes care of any baseline value. For 3dMEMA (or 3dttest++) to work, you would need some 'magic' numbers (e.g., coefficients from orthogonal polynomials); in the case of linear trend with three conditions, one set of such magic numbers are -1, 0, and 1. However, the problem is that the middle condition is directly taken into consideration when you extract the trend information.
I forgot to ask: Aside from the reliability issue, are you generally happy with the 3dRegAna results?
Gang
