Tutorial: Mixed-Effects Models in R

presented by Douglas Bates

The lme4 package for R provides replacements lmer and nlmer for the lme and nlme functions in the nlme package. The new package allows useRs to fit linear mixed models or generalized linear mixed models or nonlinear mixed models with multiple levels of random effects in nested or crossed or partially crossed configurations. The underlying computational methods provide the ability to work with very large data sets, such as are encountered in a value-added analysis of standardized test scores in education research, without sacrificing efficiency in smaller models.

The new package has been developed in conjunction with the latest capabilities in the lattice package for exploratory or presentation graphics of grouped data. It also provides sophisticated methods of assessing parameter precision using Markov Chain Monte Carlo samples from the posterior distribution of the parameters and for hypothesis testing using a parametric bootstrap to derive a reference distribution for the test statistic.

The tutorial will introduce useRs to the capabilities of this package using case studies.

Back to Tutorials