机构地区: 中山大学公共卫生学院
出 处: 《生物数学学报》 2008年第3期496-500,共5页
摘 要: 本文给出可交换条件下多维协变量的带有测量误差的多维结构回归模型,利用该模型研究总体平均处理效应的估计,给出当暴露组和对照组的协变量测量误差同分布时总体平均处理效应的拟极大似然估计及其性质. In this paper,the multi-dimensional structural regression models with multidimensional covariates and the measurement errors under exchangeability condition are given.The models are studied to estimate the population-averaged treatment effects.When the measurement errors of the covariates in the treatment population and the control population have the same distributions,the quasi-maximum likelihood estimator of the population-averaged treatment effect and its characters are given.