Fig. 4

Association of the three most prevalent risk of generalizability biases with pilot and efficacy/effectiveness sample size. Note: The x- and y-axis represent the log of the total sample size per study. The tick marks represent the actual total sample size across the range of sample sizes in the studies.