Robust inference for generalized linear models

• 成果类型: Article
• 署名作者: Cantoni, E; Ronchetti, E
• 署名单位: University of Geneva
• 刊物名称: JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
• ISSN/ISSBN: 0162-1459
• DOI: 10.1198/016214501753209004
• 发表日期: 2001
• 页码: 1022-1030
• 关键词: montane ash forests; bounded-influence; arboreal marsupials; central highlands; CONSERVATION; regression; australia; victoria; tests; fits

摘要

By starting from a natural class of robust estimators for generalized linear models based on the notion of qua-si-likelihood, we define robust deviances that can be used for stepwise model selection as in the classical framework. Wc derive the asymptotic distribution of tests based on robust deviances, and we investigate the stability of their asymptotic level under contamination. The binomial and Poisson models are treated in detail. Two applications to real data and a sensitivity analysis show that the inference obtained by means of the new techniques is more reliable than that obtained by classical estimation and testing procedures.