Modified BIC Criterion for Model Selection in Linear Mixed Models

Linear mixed-effects models are widely used in applications to analyze clustered, hierarchical, and longitudinal data. Model selection linear mixed is more challenging than that of as the parameter vector a model includes both fixed effects variance component parameters. When selecting components random effects, must be non-negative parameters may lie on boundary space. Therefore, classical methods cannot directly handle this situation. In article, we propose modified BIC for with can solve case when Through simulation results, found performed better regular most cases models. The was also applied real dataset choose most-appropriate model.