报告题目：A negative control method for removing confounding
报 告 人：苗旺博士 北京大学
报告时间：2019年10月21日下午14：00-15：00
报告地点：数学楼第二报告厅
报告摘要：
Confounding often arises in observational studies, and it may jeopardize causal inference and lead to biased results. In this talk, I will present a mitigation method we recently developed, which incorporated auxiliary variables called negative controls to eliminate the confounding bias. We show that under certain completeness conditions, the actual causal effect is identified with both a negative control exposure and a negative control outcome. I will also illustrate with several examples including a time-series air pollution study.
报告人简介：
苗旺，现任北京大学光华管理学院助理教授，在北京大学数学学院获统计学博士学位，曾在哈佛大学生物统计系做博士后研究。研究方向包括因果推断，缺失数据分析，及其在生物统计、流行病学和计量经济中的应用。
