Methods to Detect Treatment Effect in Cluster Randomized Trials

• In Event: Multilevel Modeling: Estimation, Model Selection, and Sample Size

Abstract

Cluster randomized trials typically represent a substantial investment of time and money. As the use of cluster randomized trials has become more common, it has become particularly important to investigate the impact of analytical methods on the results of cluster randomized trials. We plan to investigate Type I error rate and power for the test of the treatment effect in a 2-level cluster randomized trial as well the bias of the standardized treatment effect for four analytical methods. The four methods are the result of two factors (a) whether a level-two covariate was an observed mean or latent mean of a level-1 covariate and (b) whether a level-1 covariate was included in the model.

Authors

• Burak Aydin, Recep Tayyip Erdoğan University

• James Algina, University of Florida

• Walter L. Leite, University of Florida