Monte Carlo Likelihood Approximation for Generalized Linear Mixed Models

• Activity Number: 650
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Computing
• Author(s): Christina Knudson*+ and Charles Geyer and Galin Jones
• Companies: and University of Minnesota and University of Minnesota
• Keywords: Monte Carlo ; likelihood approximation ; likelihood-based inference ; random effects ; maximum likelihood ; generalized linear mixed model

Abstract

When random effects are incorporated into a model, the likelihood is expressed as an integral. Monte Carlo likelihood approximation (MCLA) can be used to approximate this integral. This enables maximum likelihood and other likelihood-based inference. Until this point, software for performing likelihood-based inference for generalized linear mixed models was limited. Our newly-developed R package approximates the model's likelihood via Monte Carlo, then performs maximum likelihood. Other model-based information is also reported, which can be used for other types of likelihood-based inference.