Statistical power analysis is an important topic in research designs. Power analysis methods for a variety of models, such as multiple regression and ANOVA, have been well developed and implemented in popular statistical software (e.g., SAS PROC POWER). The implementation of power analysis for multilevel models, however, is less common. As part of the Webpower project, we developed online software that allows researchers to conduct power analysis for popular multilevel models.
In this week’s QSG, I am going to (1) introduce widely used multilevel designs, including cluster randomized trials and multisite randomized trials; (2) introduce the procedure of power analysis for these multilevel models; and (3) illustrate the application of Webpower.