A general approach for conducting power analysis in two- and three-level hierarchical linear models (HLMs) is described. The method can be used to perform power analysis to detect fixed effects at any level of a HLM with dichotomous or continuous covariates. It can easily be extended to perform power analysis for functions of parameters. Important steps in the derivation of this approach are illustrated and numerical examples are provided. Sample code implementing this approach is provided using the free program R.
Pan, Yi and McBee, Matthew T.
"A Flexible Method for Conducting Power Analysis for Two- and Three-Level Hierarchical Linear Models in R,"
Journal of Modern Applied Statistical Methods:
1, Article 30.
Available at: http://digitalcommons.wayne.edu/jmasm/vol13/iss1/30