Semiparametric Methods for Clustered Binary Data

• Activity Number: 182
• Type: Invited
• Date/Time: Monday, July 30, 2007 : 2:00 PM to 3:50 PM
• Sponsor: SSC
• Author(s): Grace Y. Yi*+ and Wenqing He and Hua Liang
• Companies: University of Waterloo and University of Western Ontario and University of Rochester
• Keywords: Binary data ; Clustered data ; Estimating Equation ; Missing Data ; Semiparametric regression

Abstract

Clustered binary data arise commonly in practice, and generalized estimating equations methods are frequently used to analyze such data. The generalized linear model is often used to modulate the mean response in which covariates pertaining to the responses are present in a linear form through a link function. In practice, however, the relationship between the mean response and covariates may be very complex and linearity may not be adequate to capture that relationship. Under this circumstance, semiparametric regression with both linear and nonlinear terms included may be more flexible to facilitate the relationship between the response and covariates. In this talk, I will discuss semiparametric regression methods for analyzing clustered binary data with the interest centering on estimating both the mean and association parameters. Numerical studies will be reported.