The Impact of Police Stops on Precinct Robbery and Burglary Rates in New York City, 2003-2010

Abstract

New York City experienced a dramatic reduction in crime during the 1990s and continuing through the first decade of the current century. Researchers and commentators have debated the role of policing in New York’s crime drop, including the crime impact of New York’s policy of “stop, question, and frisk” (SQF)—yet, prior research on the crime-reduction effects of SQF is limited in important respects. We seek to overcome many of these limitations in a study of SQF effects on yearly precinct-level robbery and burglary rates in New York between 2003 and 2010. Contrary to prior research, the study reveals few effects of SQF on robbery and burglary. We caution against drawing definitive conclusions from both the current and prior research and recommend that future research address both the impact of SQF on crime and possible collateral effects on the rights and liberties of citizens in the communities most affected by the policy.

Keywords:

• police stops
• crime rates

Notes

1. Excerpted from New York state criminal procedure law § 140.50, quoted in Jones-Brown et al. (2010, pp. 2, 3).

2. A related study by the same researchers evaluated the crime-reduction effects of the New York Police Department’s “Operation Impact,” which involved heightened police patrols of high-crime areas in New York City (see Smith & Purtell, 2007).

3. See http://ccrjustice.org/ourcases/past-cases/daniels,-et-al.-v.-city-new-york.

4. Officers are to fill out and submit a “UF-250” form for each stop meeting these criteria. When completed, the form indicates the date, time, and location of the stop; identification of the officer and suspect; suspect sex, race, ethnicity, age, height, weight, and other physical characteristics; whether an arrest was made or a summons was issued; a description of any weapons or contraband found; and the circumstances leading to the stop (e.g. suspect “casing” a crime location, carrying objects used to commit a crime, fits description of known offender, or engaging in “furtive movements” (Jones-Brown et al. 2010, pp. 8, 9).

5. The SQF data are from the NYPD website (http://www.nyc.gov/html/nypd/html/analysis_and_planning/stop_question_and_frisk_report.shtml).

6. The average precinct population during the period under investigation was just over 110,000.

7. The precinct crime data were provided by the NYPD.

8. Roughly the same percentage of stops resulted in the issuance of a summons (not shown; see Jones-Brown et al., 2010, p. 10).

9. The results for the Operation Impact analyses are not summarized here.

10. We thank a reviewer for this observation.

11. We dropped the precinct covering Central Park from the analysis, because it has no residential population. The crime data were provided by the NYPD.

12. See Langan and Durose (2004) and Rosenfeld and Lauritsen (2010) for assessments of the measurement reliability of New York City robbery and burglary trends.

13. We regressed the 2005-2009 ACS data on the 2000 census data and a linear counter to obtain annualized measures of the covariates. The tract-level data were aggregated to the precinct level for analysis. Population characteristics of tracts spanning multiple precincts were calculated according to the proportionate geographic area in each precinct.

14. No hard-and-fast rule exists for determining the optimal size of the instrument matrix, but a common rule of thumb is to not allow the number of instruments to exceed the number of panels (Roodman 2006). Where necessary, we met this standard by restricting the maximum number of lags of the crime rates and endogenous measures used as instruments in the equations.

15. All results not shown are available from the authors by request.

16. To insure that the number of instruments did not exceed the number of panels in the final models of Tables23, we limited the maximum lags of the dependent and endogenous variables to be used as instruments to five.

17. The Sargan test results for model 5 in Table 2 are = 29.32, p >.05. For model 5 in Table 3, the results with five lags of the outcome and endogenous variables are = 74.48, p < .01. When four lags are used as instruments, the Sargan results for model 5 in Table 3 are = 53.05, p > .05. Failure to reject the null hypothesis in the Sargan test indicates valid over-identifying restrictions.

18. We retain 600 observations in the misdemeanor arrest models, because we have complete data for the 2000-2010 time series for both arrest and crime rates.

19. We restricted the lags of the crime rate and endogenous variables in these models to a maximum of three to insure that the number of instruments would not exceed the number of panels. The Sargan test results for model 5 in Tables45 are = 60.87, p > .05 and = 59.04, p > .05, respectively.

20. Smith and Purtell (2007, 2008) did investigate the effect of a hot-spots strategy, Operation Impact, on crime rates, but their analysis simply partitions police precincts according to whether they contained an “impact zone” and is not based on within-precinct microspatial data.

Additional information

Notes on contributors

Richard Rosenfeld

Richard Rosenfeld is curators professor of Criminology and Criminal Justice at the University of Missouri—St Louis. He has published widely on the institutional sources of crime, crime trends, and crime statistics. He is a Fellow and past president of the American Society of Criminology and currently serves on the Science Advisory Board of the Office of Justice Programs, US Department of Justice.

Robert Fornango

Robert Fornango completed his PhD in criminology and criminal justice at the University of Missouri—St Louis in 2007. His research explores the dynamic processes shaping crime trends in the US at the national, regional, city, and neighborhood level. His interests include the influences of social structural change, criminal justice policies, and spatial dynamics in explaining social organization and violent crime. Additionally, his work examines the correlates of recidivism among released prisoners. Recent co-authored work has appeared in the journals Criminology and Criminology & Public Policy.