Thanks for the feedback, Vinny.
> One question I had is whether its worth redoing analyses that utilized 3dLME or 3dANOVA?
If you're satisfied with your previous results, you probably don't need to re-analyze them with 3dMEMA.
> Hypothetically, let's say when looking at ROI derived estimates of signal
> change for a group of subjects there appears to be one or two subjects who
> are outliers in one of two groups of interest. Would 3dMEMA yield radically
> different results than those obtained through the other two programs?
That depends. Unreliable outliers (small or big beta with small t-statistic) will be nicely handled through downgrading in the final result. However, reliable outliers (small or big beta with big t-statistic) will still dramatically change/bias the group effect. I'm still working on the 2nd situation.
Gang