A Simple Monte Carlo Method for Estimating Power in Multilevel Designs

• In Event: Statistical Power Considerations in Multilevel Designs

Abstract

This paper describes a Monte Carlo approach to estimating power for multilevel models. Our method allows any number of within-cluster, between-cluster, and covariate effects at either level, cross-level interactions, and random coefficients. Moreover, we do not assume orthogonal predictors, and regressors can correlate at either level. Additionally, our approach accommodates multiple cross-level interaction effects, deriving exact expressions for the variances and covariances of product random variables. Finally, our strategy allows researchers to specify multilevel population parameters using R^2 effect size expressions, and we use these equations to derive solutions for a model’s fixed effect coefficients and variance components. We provide an R function that computes these solutions and automates the process of generating artificial data sets and summarizing the simulation results.

Authors

• Craig K. Enders, University of California - Los Angeles
  Presenting Author

• Michael Woller, University of California - Los Angeles
  Presenting Author