Comparing Systemic and Individual Sources of Racially Disparate Traffic Stop Outcomes
利用数百万次交通拦截数据，构建两种指标评估种族差异是普遍存在（系统性来源）还是局限于少数警察（个体来源），发现搜查中的种族差异广泛存在，且禁止同意搜查比剔除异常警察更能改变执法模式。
Abstract The deaths of George Floyd and Breonna Taylor brought discussions of race and policing to the forefront in the summer of 2020 in the United States, spurring protests and calls for policing reform. However, enacting successful reforms curtailing racially biased policing requires understanding whether bias is widespread, likely tied to systemic sources, or the work of a few racist officers. This study elaborates on these perspectives by drawing on theories of systemic and individual bias that may arise when bureaucrats have ample discretion. Using millions of traffic stops, I construct two measures to evaluate whether disparities are widespread—as indicated by the first perspective—or confined to a few officers—as suggested by the second. This study finds widespread racial disparities in who is searched following a traffic stop and that banning consent searches would alter policing patterns more than eliminating outlier officers in a set of paired t-tests.