Longitudinal data analysis based on generalized linear partially varying-coefficient models

ABSTRACT

In this article, we consider a generalized linear partially varying-coefficient model for longitudinal data analysis. A local quasi-likelihood method is proposed to estimate the constant-coefficient and varying-coefficient functions simultaneously based on the local polynomial kernel regression. The corresponding standard error estimates are derived. Large sample properties are investigated. The proposed methodologies are demonstrated by extensive simulation studies and a real example.

KEYWORDS:

• Generalized estimating equations
• local polynomial kernel regression
• varying-coefficient models
• working correlation matrix

MATHEMATICS SUBJECT CLASSIFICATION:

• 62G09

Funding

The work of the second author is partially supported by a grant from the Research Grant Council of the Hong Kong Special Administration Region (Project No. UGC/FDS14/P01/16). The work of the third author was partially supported by the National Natural Science Foundation of China (No. 11271368), the Beijing Planning Office of Philosophy and Social Science (No. 12JGB051), the Fundamental Research Funds for the Central Universities, and the Research Funds of the Renmin University of China (No. 10XNL018, No. 10XNK025), the China Statistical Research Project (No. 2011LZ031), the Government of Chaoyang District Postdoctoral Research Foundation (No. 2014ZZ-13), and the Beijing Research Base of Social Science Fund Project (No. 14JDJGC040).