Misreporting in sensitive health behaviors and its impact on treatment effects : an application to intimate partner violence

A growing literature seeks to identify policies that could reduce intimate partner violence. However, in the absence of reliable administrative records, this violence is often measured using self-reported data from health surveys. In this paper, an experiment is conducted comparing data from such surveys against a methodology that provides greater privacy to the respondent. Non-classical measurement error in health surveys is identified as college-educated women, but not the less educated, underreport physical and sexual violence. The paper provides a low-cost solution to correct the bias in the estimation of causal effects under non-classical measurement error in the dependent variable.