A statistical method used to determine the sample size needed to detect a significant effect with a given level of power.