Assessing the validity of weighted generalized estimating equations

• 成果类型: Article
• 署名作者: Qu, A.; Yi, G. Y.; Song, P. X. -K.; Wang, P.
• 署名单位: University of Illinois System; University of Illinois Urbana-Champaign; University of Waterloo; University of Michigan System; University of Michigan
• 刊物名称: BIOMETRIKA
• ISSN/ISSBN: 0006-3444
• DOI: 10.1093/biomet/asq078
• 发表日期: 2011
• 页码: 215224
• 关键词: Longitudinal Data; inference; models

摘要

The inverse probability weighted generalized estimating equations approach (Robins et al. 1994; Robins et al. 1995), effectively removes bias and provides valid statistical inference for regression parameter estimation in marginal models when longitudinal data contain missing values. The validity of the weighted generalized estimating equations regarding consistent estimation depends on whether the underlying missing data process is properly modelled. However, there is little work available to examine whether or not this condition holds. In this paper we propose a test constructed from two sets of estimating equations: one set is known to be unbiased, but the other set is not known. We utilize the quadratic inference function (Qu et al. 2000) method to assess their compatibility, which is equivalent to testing for the validity of the weighted generalized estimating equations approach. We conduct simulation studies to assess the performance of the proposed method. The test procedure is illustrated through a real data example.