1. BIAS IN POLICE STOPS?
In the late 1990s, popular, legal, and political concerns were raised across the United States about police harassment of minority groups in their everyday encounters with law enforcement.
These concerns focused on the extent to which police were stopping people on the highways for “driving while black” (seeWeitzer 2000; Harris 2002; Lundman and Kaufman 2003).
Additional concerns were raised about racial bias in pedestrian stops of citizens by police predicated on “zero-tolerance” policies to control quality-of-life crimes and policing strategies concentrated in minority communities that targeted illegal gun possession and drug trafficking (see Fagan, Zimring, and Kim 1998; Greene 1999; Skolnick and Caplovitz 2001; Fagan and Davies 2000, 2003; Fagan 2002; Gould and Mastrofski 2004).
These practices prompted angry reactions among minority citizens that widened the breach between different racial/ethnic groups in their trust in the police (Lundman and Kaufman 2003; Tyler and Huo 2003;Weitzer and Tuch 2002), provoking a crisis of legitimacy with legal, moral, and political dimensions (see Wang 2001; Russell 2002; Harris 2002).
In an era of declining crime rates, policy debates on policing strategies often pivot on the evaluation of New York City’s policing strategy during the 1990s, a strategy involving aggressive stops and searches of pedestrians for a wide range of crimes (Eck and Maguire 2000; Skogan and Frydl 2004). The policy was based on the lawful practice of “temporarily detaining, questioning, and, at times, searching civilians on the street” (Spitzer 1999). The U.S. Supreme Court has ruled police stopand- frisk procedures to be constitutional under certain restrictions (Terry v. Ohio 1968). The approach of the New York City Andrew Gelman is Professor, Department of Statistics and Department of Political Science, Columbia University, New York, NY 10027 (E-mail:[email protected]). Jeffrey Fagan is Professor, Law School and School of Public Health, Columbia University, New York, NY 10027 (E-mail: [email protected]). Alex Kiss is Biostatistician, Department of Research Design and Biostatistics, Sunnybrook andWomen’s College Health Sciences Center, Toronto, Ontario, Canada. The authors thank the New York City Police Department, the New York State Division of Criminal Justice Services, and the Office of the New York State Attorney General for providing data for this research. Tamara Dumanovsky and Dong Xu made significant contributions to the analysis. Joe Bafumi, Rajeev Dehejia, Jim Liebman, Dan Rabinowitz, Caroline Rosenthal Gelman, and several reviewers provided helpful comments. Support for this research was provided in part by National Science Foundation grants SES-9987748 and SES-0318115. All opinions are those of the authors. Police Department (NYPD) during the 1990s has been widely credited as a major source of the city’s sharp crime decline (Zimring 2006).
But near the end of the decade there were repeated complaints of harassment of minority communities, especially by the elite Street Crimes Unit (Spitzer 1999). These complaints came in the context of the well-publicized assault by police of Abner Louima and the shootings of Amadou Diallo and Patrick Dorismond. Citizen complaints about aggressive “stop and frisk” tactics ultimately provoked civil litigation that alleged racial bias in the patterns of “stop and frisk,” leading to a settlement that regulated the use of this tactic and established extensive monitoring requirements (Kelvin Daniels et al. v. City of New York 2004).
We address this dispute by estimating the extent of racially disparate impacts of what came to be known as the “New York strategy.” We analyze the rates at which New Yorkers of different ethnic groups were stopped by the police on the city streets, to assess the central claim that race-specific stop rates reflect nothing more than race-specific crime rates. This study is based on work performed with the New York State Attorney General’s Office (Spitzer 1999) and reviewed by the U.S. Commission on Civil Rights (2000). Key statistical issues are the baselines used to compare rates (recognized as a problem by Miller 2000; Walker 2001; Smith and Alpert 2002) and local variation in the intensity of policing, as performed by the Street Crimes Unit and implicitly recommended byWilson and Kelling (1982) and others. We use multilevel modeling (see Raudenbush and Bryk 2002 for an overview and Sampson, Raudenbush, and Earls 1997; Sampson and Raudenbush 1999;Weidner, Frase, and Pardoe 2004 for examples in studies of crime) to adjust for local variation in comparing the rates of police stops of different ethnic groups in New York City.
Were the police disproportionately stopping ethnic minorities? We address this question in several different ways using data on police stops and conclude that members of minority groups were stopped more often than whites, both in comparison to their overall population and to the estimated rates of © 2007 American Statistical Association Journal of the American Statistical Association September 2007, Vol. 102, No. 479, Applications and Case Studies DOI 10.1198/016214506000001040