您所在的位置:首页 新闻动态 学术讲座

学术讲座

中国人民大学李伟教授应邀讲座

讲座信息

题目:Identification and Multiply Robust Estimation of Optimal Treatment Regimes with an Invalid Instrument

主讲人:李伟 (中国人民大学 教授)

时 间:2026.11.04 19:30

联系人:罗姗姗(北京工商大学)

报告摘要

The estimation and evaluation of individualized treatment regimes have gained increasing attention in observational studies. Existing approaches, however, typically rely either on the absence of unmeasured confounding or on the availability of valid instrumental variables (IVs). In this paper, we study the nonparametric identification of the optimal treatment regime under unmeasured confounding using a potentially invalid IV. We relax the standard exclusion restriction to a mean-difference exclusion restriction that allows the IV to have a direct effect on the outcome, and introduce additional restrictions on unmeasured effect modification and confounding to identify the conditional average treatment effect (CATE) and, consequently, the optimal regime. Under these conditions, we derive two equivalent identification formulas for the optimal regime. We further establish a necessary and sufficient condition under which the proposed formulas continue to recover the optimal regime even when the CATE itself is not identifiable. Building on these identification results, we develop multiply robust estimators of the optimal regime and evaluate their finite-sample performance through extensive simulations and a real-data application.

主讲人简介

李伟,中国人民大学统计学院教授、博士生导师,先后入选中国人民大学吴玉章青年学者、教育部青年长江学者、北京市通州区“运河英才计划”领军人才。2013年毕业于南开大学数学科学学院,获理学学士学位;2018年毕业于北京大学数学科学学院,获理学博士学位;2018—2020年在北京大学从事博士后研究。2020年加入中国人民大学统计学院,先后任讲师、副教授,2026年破格晋升教授。主要研究方向为因果推断,已在Biometrika, JASA, JRSSB等统计学及相关领域知名期刊发表论文30余篇,出版学术专著1部。