Detecting Bias in Traffic Searches: Examining false searches of innocent drivers

Purpose: We test current models of racial bias in policing, identify limitations,and propose a test of racial bias, the Overlapping ConditionTest, that accounts for unknown population contraband rate. Methods: We conceptualize police officer search decisions as a 2 (search/nosearch) by 2 (contraband present/absent) table, with missing data (ifthe police did not search, the presence of contraband is unknown). Weconstrain the feasible problem space using properties of a 2 x 2 contingencytable. Then we examine all possible feasible 2 x 2 tables toidentify instances of racial differences in police officer hit and false alarmrates. To do this, we develop a new test of racial bias, the OverlappingCondition Test. We analyze state and county data across 26 differentpolice departments in the United States. Results: We observe thesedepartments have observable racial bias in every feasible 2 x 2 table,in the county of Durham, NC, 10 out of 14 state police departmentsand 9 out of 10 local departments across the United States. That is, forevery feasible real world scenario police officers have lower false alarmrates for White drivers than Black drivers. Conclusion: We interpretthis difference in false alarm rate as a threshold bias. That is, officersuse different criteria for searching Black drivers than White drivers andthis conclusion is not qualified by the unknown contraband rate. Futuredirections should explore how police officers make the decision to searchdrivers and develop interventions to address the racial bias in search rate.