Estimate (Generalized) Linear Mixed Models with Factor Structures.
PLmixed
The purpose of PLmixed
is to extend the capabilities of lme4
to allow factor structures (i.e., factor loadings and discrimination parameters) to be freely estimated. Thus, factor analysis and item response theory models with multiple hierarchical levels and/or crossed random effects can be estimated using code that requires little more input than that required by lme4
. All of the strengths of lme4
, including the ability to add (possibly random) covariates and an arbitrary number of crossed random effects, are encompassed within PLmixed
. In fact, PLmixed
uses lme4
and optim
to estimate the model using nested maximizations. Details of this approach can be found in Jeon and Rabe-Hesketh (2012). A manuscript documenting the use of PLmixed
is currently in preparation.
Installation
PLmixed
can be installed from CRAN with:
install.packages("PLmixed")