Real power of cluster randomized trials according to the discrepancy between the a priori postulated and a posteriori estimated intraclass correlation coefficients. The effect size to be detected is fixed at 0.25 and power at 80%. g is the number of clusters per arm, m is the average cluster size and N is the total number per intervention arm considering an a priori postulated ICC of 0.005 or 0.02.
Guittet et al. BMC Medical Research Methodology 2005 5:25 doi:10.1186/1471-2288-5-25