Constructing Race-Specific Driving Patterns to Address Racial Profiling

虽然有大量证据表明，美国各地的警察在对待有色人种时存在种族和民族偏见，但几乎没有经验证据表明，警察在与有色人种接触时对有色人种进行了侧写。众所周知，要确定警察是否以及在多大程度上对交通拦截中的有色人种进行侧写是非常困难的。经验证据的缺乏破坏了正确诊断和干预种族偏见警察行为的努力。此外，除了少数群体与警察的负面经历外，当人们感到受到警察的不公正待遇时，他们就会对执法部门公平代表和执法的权威失去信心。当这种情况发生时，个人就不太可能遵守法律。因此，当警察在交通拦截中针对有色人种时，它助长了对双方的不信任，潜在的犯罪行为，以及几十年来一直困扰美国社会关系，治安和犯罪率的刻板印象的循环。简而言之，准确衡量警察的种族定性是绝对必要的。本研究的目的是综合和分析一套完全不同的公共交通数据，包括人口普查交通规划产品（CTPP），事故数据，以及来自全国家庭旅行调查（NHTS）及其超样本的数据，以开发一个方法框架，用于构建特定种族的驾驶模式，以解决警察对司机的种族貌相问题。该项目的第一个目标侧重于设计创新方法，将地理和交通数据与不同的空间和时间分辨率相结合，以解决分母问题，并创建一个可推广的框架，用于估计种族动机交通停车。第二个目标侧重于衡量和建立用于确定种族定性的方法的有效性，方法是使用开源警务数据来评估个人的种族和民族分布，如果警察没有种族偏见，这些人将被拦截，同时考虑到警察在一个地点的活动。第三个目标是为开发的方法和可重新分发的数据设计一个开源分发平台，并通过文档完备的Python和R前端提供，以促进本项目使用的方法和数据的使用、重用和扩展。将这些测量、方法和统计模型结合起来，将为政策制定者和执法机构提供有关有偏见的警察做法的信息，并为改善结果和减少警察与有色人种社区之间不信任的循环提供蓝图。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

While there is substantial evidence suggesting the presence of racial and ethnic bias in police treatment of people of color throughout the United States, there is little to no empirical evidence indicating that police profile people of color for police contact. Identifying whether and the extent to which the police are profiling people of color for traffic stops is notoriously difficult. This lack of empirical evidence undermines efforts to correctly diagnose and intervene in racially biased policing behavior. Further, in addition to the negative experience(s) that minority groups have with police, when people feel treated unjustly by the police, they lose their faith in law enforcement’s authority to fairly represent and administer the law. When this occurs, individuals are less likely to obey the law. Therefore, when the police target people of color for traffic stops, it feeds a cycle of distrust for both parties, potential criminality, and stereotypes that have plagued societal relations, policing and crime rates in the United States, for decades. In short, it is absolutely essential to accurately measure racial profiling by the police.The purpose of this research is to synthesize and analyze a disparate suite of publicly available transportation data that includes the Census Transportation Planning Products (CTPP), accident data, as well as data from the National Household Travel Survey (NHTS) and its oversamples to develop a methodological framework for constructing race-specific driving patterns to address racial profiling of drivers by the police. The first objective of this project focuses on the design of innovative methods to integrate both geographic and transportation data with varying spatial and temporal resolutions for resolving the denominator problem and creating a generalizable framework for estimating racially motivated traffic stops. The second objective focuses on measuring and establishing the validity of the methods used to determine racial profiling by using open source policing data to evaluate the racial and ethnic distribution of individuals who would be stopped if the police were not racially biased while simultaneously accounting for police activity in a location. The third objective focuses on the design of an open-source distribution platform for the developed methods and redistributable data, and made available through well-documented Python and R front ends to facilitate the use, re-use and expansion of the methods and data used for this project. When combined, these measurements, methods, and statistical models will inform policymakers and law enforcement agencies about biased police practices and provide a blueprint for improving outcomes and reducing the cycle of distrust between the police and communities of color.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.