Gang,
How should we now handle averaging of beta coefficients from the output iresp functions when using 3dMEMA. Typically we use TENT basis functions to estimate the hemodynamic response, take the average of the beta coefficients in an specific window of time, and use these averages for ANOVA analyses. Should we do the same with respect to the t-statistics indicating reliability of the beta estimates for an individual subject? Would this really represent a reliability estimate of the average activity over the specified time window?
Vinny