Thomas Lumley • 4/1/2018

svylme

The article details the development of an experimental R package for fitting two-level linear mixed models to data from two-stage complex survey samples. It explains the use of pairwise composite likelihood for estimation, current limitations like standard errors only for fixed effects, and plans for future extensions including more model levels and generalized linear mixed models. The package relies on `lme4` for starting values and optimization techniques.

0 comments

#R #Statistical Modeling #Mixed Models