B0233
Title: Sensitivity analysis for null results: Implications for studies of racially biased policing
Authors: Jake Bowers - University of Illinois @ Urbana-Champaign (United States) [presenting]
Thomas Leavitt - Harvard University (United States)
Luke Miratrix - Harvard University (United States)
Abstract: A method of formal sensitivity analysis is proposed for causal inference that addresses the problem of null results: a null result in an observational study is no more or less likely to emerge because of hidden confounding than a strong result. The motivation comes from the problem of null results in the study of the causal effects of race of civilians on police use of force and shows how it adds to existing critiques of null results. We show how in a small simulated dataset, a pattern of hidden confounding and a pattern of post-treatment missingness like that seen in datasets used to study race and police can combine to produce a misleading null effect. And we show how our method of sensitivity analysis for null effects reveals that the null result is, in fact, sensitive to this kind of bias. We compare both an approach for tests of the weak null of no average effects and an approach for the strong null of no effects.
