As part of the response to allegations of racially biased police practices many police agencies began collecting information on the stops made by their officers. Social scientists have attempted to use these administrative data on stop decisions to assess the existence or extent of racially biased policing and, in the process, have developed a number of benchmarks for comparison to police stop data. This chapter describes an array of benchmarking methods that have been used around the country including the use of U.S. Census population estimates, non-at fault driver crash data, crime and arrest data, drivers' license data, red light cameras, observations, instrumental variables, assessments of post-stop outcomes, and officer-to-officer comparison via internal benchmarks. Each method's application, strengths, and weaknesses are discussed in the context of their ability to establish a reasonable estimate of the population at risk for being stopped by the police and to draw a causal inference about the extent to which race is a relevant factor in police decision-making on whom to stop, question, and search.
Posted here with permission from Race, Ethnicity, and Policing: New and Essential Readings, edited by Stephen K. Rice and Michael D. White, March 2010, chapter 7, pages 180-204. Copyright 2010 NYU Press.