power analysis for within-subject, high-repetition experiment