Research Seminars @ Illinois

Abstract

Policymakers, researchers, and other stakeholders often care about the prevalence of complex, sensitive outcomes or behaviors, such as human trafficking and forced labor, gender-based violence, child labor, or corruption. But the measurement of such issues is challenging: Comprehensive administrative data is usually unavailable, survey respondents may define these concepts very differently than existing legislation, and social stigma may lead to substantial underreporting. In this paper, we set up a conceptual framework that identifies four theoretically distinct types of prevalence rates. We then propose a survey design approach that is practically feasible in many contexts and that, with only one additional assumption beyond those commonly made in the literature, can recover all the conceptual information. We apply that approach to estimate the prevalence of labor trafficking in two countries, Malawi and Zambia.