Quote
What is the right tool to use to get the group mean of beta with proper t-score (p-value)?
3dttest seems not right since there is no option to tell it that the measures are repeated (multiple runs).
3dANOVA? gives means for each factor level, but I want the mean of all subjects, all runs.
This depends on what exactly your situation is. 3dttest is definitely not equipped to handle it if you have separate beta for each run.
(1) You can use either 3dMean or 3dcalc to obtain the average beta cross runs for each subject and then run group analysis as usual;
Or
(2) You can use the appropriate ANOVA program and provide all the beta's as input for the ANOVA program. Just be careful with the -type option in that ANOVA program.
Gang