Testing Causative Hypotheses in Nonparametric Models with a Mismeasured or Coarsened Confounder

• Activity Number: 135 - Novel Non/Semiparametric Developments for Risk Perception with Censored and/or Missing Data
• Type: Invited
• Date/Time: Monday, July 31, 2017 : 10:30 AM to 12:20 PM
• Sponsor: Section on Risk Analysis
• Author(s): Wang Miao* and Lan Liu
• Companies: Peking University and University of Minnesota at Twin Cities
• Keywords: Air pollution ; Causative hypothesis ; Coarsening ; Confounder ; Measurement error

Abstract

Suppose we are interested in a causal effect that is confounded by a mismeasured or coarsened confounder. We propose a general framework that can account for both measurement error and coarsening such as censoring of the confounder, to test the hypothesis of a null causal effect. The proposed test method does not necessarily depend on any parametric assumptions on the causal model, except for certain regularity conditions. Moreover, it has good power against familiar alternatives. We apply such a test to accessing the immediate effect of PM2.5 on mortality in an air pollution study.