A0239
Title: A generalized linear model in the presence of population heterogeneity
Authors: Yichuan Zhao - Georgia State University (United States) [presenting]
Abstract: Generalized linear models (GLMs) are highly effective for modeling mean responses under nonstandard conditions, accommodating both discrete and continuous data distributions. In the analysis of individual-level data, numerous parameter estimation methods for GLMs have developed, with the quasi-likelihood method being particularly noteworthy. Recently, leveraging auxiliary information from external sources to enhance the estimation efficiency of model parameters has become a prominent research topic. The aim is to construct estimating equations based on auxiliary information from external sources and combine these with the quasi-likelihood estimating equations derived from individual-level data to form a unified estimating equation. To address the population heterogeneity among different sources of information, bias parameters are defined, and those are included in the unified estimating equation. A generalized method of moment-based is utilized on the unified estimating equation to estimate the coefficients in GLMs. Simultaneously, an adaptive LASSO penalty term is introduced to characterize the potential population heterogeneity among different sources of information. Simulation studies are conducted in various settings to examine the finite sample properties of the proposed method and compare its performance with that of the quasi-likelihood method. Finally, the proposed method is applied to analyze a real-world dataset.
