Hi Gary,
What are the two factors when you run 3dANOVA2? One is condition conctrast?
I believe that the independence assumption in two- and three-way ANOVA is about the factor variables, not about the factor levels. If you set condition contrast as one factor, I think you can still get group mean of those three contrasts without any problem.
If you don't feel comfortable with running 3dANOVA2 this way, you can set the condition (not condition contrast) as one factor, which has three levels. Then use the beta weights of the three conditions as input, and get the group mean of condition contrast with the -adiff or -bdiff option in 3dANOVA2.
I wonder if you can also run one-way ANOVA (3dANOVA or 3dttest) to get the group mean of those condition contrasts. Of course the results (t values) would be a little less legitimate since the part of sum of squares due to the other factor and any potential interaction is not accounted for in the case of one-way ANOVA.
Gang