Linear mixed models for complex survey data: implementing and evaluating pairwise likelihood

TL;DR

This paper presents an implementation of two-level linear mixed models for complex survey data using pairwise composite likelihood in R, evaluating its efficiency and comparing it to existing methods through simulations and real survey data.

Contribution

It introduces a novel implementation of pairwise composite likelihood for linear mixed models tailored to complex survey data, with performance assessment.

Findings

Pairwise composite likelihood is computationally efficient.

Estimator performs well compared to stagewise pseudolikelihood.

Effective for analyzing complex survey data.

Abstract

As complex-survey data becomes more widely used in health and social-science research, there is increasing interest in fitting a wider range of regression models. We describe an implementation of two-level linear mixed models in R using the pairwise composite likelihood approach of Rao and co-workers. We discuss the computational efficiency of pairwise composite likelihood and compare the estimator to the existing stagewise pseudolikelihood estimator in simulations and in data from the PISA educational survey.