协变量含误差和输出变量误分类时处理组平均处理效应的估计  

Estimation of the Average Treatment Effect on the Treated with Error-prone Covariates and Misclassified Outcomes

作　　者：魏少杰 谢田法[1] 张忠占[1] WEI Shaojie;XIE Tianfa;ZHANG Zhongzhan(Faculty of Science,Beijing University of Technology,Beijing 100124)

机构地区：[1]北京工业大学理学部,北京100124

出　　处：《系统科学与数学》2022年第10期2834-2846,共13页Journal of Systems Science and Mathematical Sciences

基　　金：国家社科基金(21BTJ041);北京市自然科学基金(1202001)资助课题。

摘　　要：经典的逆概率加权估计方法依赖于变量的精确测量,然而实际上这一条件有时得不到满足.文章研究了当协变量带有测量误差和输出变量误分类时,处理组平均处理效应的估计问题.基于纠偏的思想和条件得分方法,讨论了加权估计方法中加权函数的构造和估计方法,并进一步给出了处理组平均处理效应的相合估计.模拟研究和实例分析表明了所提出方法的优越性.The classic inverse probability weighting method relies on the precise measurement of variables.However,in practice,this condition is often violated.This paper focuses on the estimation of the average treatment effect on the treated with error-prone covariates and misclassified outcomes.Based on correction and conditional score methods,the weighting function that can guarantee a consistent inverse probability weighting estimator is discussed,and the consistent estimator of the average treatment effect on the treated is further given.Simulation studies and data analysis demonstrate the superiority of the proposed method.

关 键 词：Rubin因果模型 测量误差 误分类 处理组平均处理效应 逆概率加权 

分 类 号：O212.1[理学—概率论与数理统计]