In order to conduct an a priori sample size calculation for a repeated-measures t-test, researchers have to hypothesize how much of a difference will exist between the pre/baseline observation and the post-observation as a result of treatment. The absolute difference between observations of means and standard deviations and proportions is the effect effect size. Researchers should seek out evidenced-based measure of effect from a published article that is theoretically, conceptually, or pathophysiologically similar. This adds internal validity to the study and is preferable to just "throwing" numbers around regarding sample size. The methods for conducting and interpreting an a priori sample size calculation for repeated-measures t-test in G*Power. http://www.scalelive.com/sample-size-for-repeated-measures-t-test.html