Focused information criterion and model averaging for varying-coefficient partially linear models with longitudinal data

Abstract

We study model selection and model averaging in varying-coefficient partially linear model with longitudinal data. The model is estimated by combining spline approximations and the generalized estimating equations. The corresponding linear coefficient estimators are shown to be asymptotically normal. Then we propose a focused information criterion and a frequentist model average estimator, and establish the asymptotic properties of the proposed estimators. Simulation studies show the proposed methods are superior to the existing ones even if the covariance structure is misspecified. The procedures are further applied to a real data case.

Keywords:

• Focused information criterion
• Generalized estimating equation
• Longitudinal data
• Model averaging
• Varying-coefficient partially linear model

2000 MATHEMATICAL SUBJECT CLASSIFICATION:

• Primary 62G08
• Secondary 62G20
• 62G99

Acknowledgments

The authors would like to thank the reviewers and editors for their careful reading and constructive comments.