Department of Statistics Seminar
North Carolina State University

presents

Daowen Zhang

University of Michigan

"Generalized Additive Mixed Models"

ABSTRACT

A new class of models, generalized additive mixed models (GAMMs), is proposed for analyzing correlated continuous and discrete data. GAMMs use random effects to account for correlation among observations and use additive nonparametric functions to allow for flexible dependence of an outcome variable on covariates. Smoothing splines are used to estimate the nonparametric functions and marginal quasi-likelihood is used to estimate the smoothing parameters and the variance components simultaneously. In view of often intractable numerical integration, double penalized quasi-likelihood is proposed to draw approximate inference for the model components. A bias correction is used to improve the performance of the double penalized quasi-likelihood estimates for sparse data. A linearity test and a variance component test are developed. Extensive simulation studies were conducted to evaluate the performance of the proposed inference procedure. Our simulation results show that the proposed procedure performs well in estimating the nonparametric functions and the variance components. The proposed models are illustrated through application to data sets from biomedical research.

Thursday, January 15, 1998

8:00 - 9:00 am

124 Dabney Hall

Refreshments will be served in 124 Dabney Hall at 7:45.