« Back to Results
Atlanta Marriott Marquis, International 9
Hosted By:
American Economic Association
To account for this observed heterogeneity we nest the racial bias test into a logit model that controls for officer experience and the spatial dispersion of officer activity (among other things). We find that Syracuse police do not behave monolithically and that their experience and spatial assignment are correlated with their proclivities for bias. In particular, it appears that rookie officers and officers with concentrated patrol assignments are more likely to exhibit racial profiling against black drivers in the city of Syracuse. The implications are 1) there may be scope for "learning by doing" and 2) rookies may not be racially biased but are simply bad at forecasting guilt. We also find that the most productive officers (in terms of total traffic stops) exhibit greater proclivities for bias against blacks.
Race and Policing
Paper Session
Saturday, Jan. 5, 2019 2:30 PM - 4:30 PM
- Chair: Stephen L. Ross, University of Connecticut
Racial Bias in Police Searches: Using Shifts in Police Manpower to Test for Racial Profiling
Abstract
This paper applies a new statistical method for estimating racial profiling among police officers. We extend the model of using “hit rates” from searches of suspects stopped by the police to test for racial profiling with a quasi-experimental design that exploits the sudden shift in the intensity of police searches caused by the declaration of small geographic areas in New York City as high crime “impact zones.” Under this sudden change in focus areas are given extra police officers who are encouraged to stop and question suspects more vigorously to reduce criminal behavior. If the police utility is only focused on reducing crime, then the increase in stops and searches after an area is declared an impact zone should be applied equally to everyone suspected of a crime. We apply the use of doubly robust estimation so that stops and searches before and after an area is declared an impact zone are statistically similar on observables. We then estimate the changes in hit rates for Blacks, Whites, and Hispanics stopped and searched by the police before and after areas during two major expansions of the impact zone program.Whose Help Is on the Way? The Importance of Individual Police Officers in Law Enforcement
Abstract
The public’s perception of police fairness is essential to the willingness of citizens to cooperate with the police and is fundamental to establishing police legitimacy. However, little is known about whether police officers are actually fair and impartial in their application of the law. In this paper, I show that the likelihood of an arrest is not only a function of incident timing, geography, offense type, and other contextual factors but also critically depends on the identity of the police officer who responds to a call for service. The analysis examines detailed data on more than 1,850 police officers responding to over 230,000 offenses reported through calls for service from the Dallas Police Department. I find that police officers are important determinants of arrest outcomes, with individual officer behavior accounting for 10-15% of the explainable variation in arrests. Officers vary widely in their arrest behavior, with a 1 standard deviation increase in an officer’s propensity to arrest resulting in a 32% increase in the likelihood of arrest. Additionally, I apply a test of taste-based racial bias and fail to find conclusive evidence that officer differences are driven by racial bias in this setting.Less-than-lethal Weapons and Police Use of Force: The Case of Tasers and the Chicago Police Department
Abstract
This paper examines the expansion of the use of tasers and its effect on the level of police use of force, suspect injuries, officer injuries, violent crimes, and race distribution of suspects subject to use of force. We examine the effects of a policy that affected the use of tasers in Chicago from 2009 to 2012, by an expansion of the number of available tasers in March 2010. This policy provides an opportunity to study the use of tasers, firearms, and other use of force substitutes. We adopt an event study approach and a differences-in-differences where we used other US jurisdictions as comparison groups. We find that the number of incidents related to firearm discharge and less-than-taser weapons (eg: baton, physical strikes, tackle down) decreased. The total number of use of force incidents increase after the expansion of the number of tasers. The net effect on injuries is ambiguous. On the one hand, officers are less likely to be injured because tasers allow them to not engage in physical confrontation with subjects. On the other hand, subjects’ injuries increase. Finally, we did not find evidence that allowing for tasers changes the racial distribution of subjects involved in incidents where officers discharge their firearms. The number of incidents involving black suspects increased by 7-10%; we did not find any statistically significant effect on white suspects. We provide a model that is consistent with our empirical findings where the officer is utility maximizing when interacting with suspects.Police Officer Experience and Racial Bias in Traffic Stops
Abstract
The taste-based approach to testing for police racial bias from police stop and search data assumes that officers are monolithic in their search and arrest behaviors. That is, economic models of police behavior assume a representative officer. Using a unique panel data set of about 200 police officers from 2006 to 2009, we find that the experience composition of the Syracuse Police Department changed over the period with rookies (officer with less than 5 years of experience) accounting for a larger percentage of police activity in later years. Using spatial dispersion measures of individual officer activity across the city, we also find rookies assigned to more concentrated patrol regions in earlier years, while senior officer activity tended to be more spatially dispersed. These roles reverse in later years.To account for this observed heterogeneity we nest the racial bias test into a logit model that controls for officer experience and the spatial dispersion of officer activity (among other things). We find that Syracuse police do not behave monolithically and that their experience and spatial assignment are correlated with their proclivities for bias. In particular, it appears that rookie officers and officers with concentrated patrol assignments are more likely to exhibit racial profiling against black drivers in the city of Syracuse. The implications are 1) there may be scope for "learning by doing" and 2) rookies may not be racially biased but are simply bad at forecasting guilt. We also find that the most productive officers (in terms of total traffic stops) exhibit greater proclivities for bias against blacks.
Discussant(s)
Jennifer Doleac
,
Texas A&M University
Felipe Goncalves
,
Princeton University
JEL Classifications
- K4 - Legal Procedure, the Legal System, and Illegal Behavior
- J1 - Demographic Economics