Preventing Sexual Assault: Measurement Error and Violence in the Informal Settlements of Nairobi, Kenya

Oregon Convention Center 

There has been a major breakthrough in preventing gender-based violence. In this talk, we will focus on how a small team of statisticians was involved with the early stages of an academic discipline that is transitioning from predominantly advocacy and theory into a discipline grounded in empirical science. We will discuss the intervention our team has studied and sketch out the two cluster-randomized trials we completed in the informal settlements of Nairobi, Kenya. We will also describe the handful of methodological innovations we've had to make in order to do our work: (i) randomization in the presence of interference, (ii) a potential outcome framework for differential measurement error (e.g., addressing bias from self-reports of sensitive outcomes), and (iii) developing a model-selection framework using community-input.