Abstract
Since the mid-1990s, a number of initiatives intended to address gang, gun and drug-related violence have arisen and demonstrated promise in reducing levels of violent crime. These initiatives have employed some combination of focused deterrence and problem-solving processes. These strategies formed the basis for Project Safe Neighborhoods (PSN), a national program implemented by the Department of Justice and coordinated by US Attorneys’ Offices. This paper is an initial attempt to assess the potential impact of the nationally implemented PSN initiative through an analysis of violent crime trends in all US cities with a population of 100,000 or above. While a number of site specific studies exist examining the potential impact of locally implemented PSN programs, to date no national-level study has examined whether PSN may have had an impact on violent crime trends. Cities included in the current study are distinguished on the basis of whether they were considered a treatment city by the PSN task force and by the level of implementation dosage of the PSN program. This allowed a comparison of 82 treatment cities and 170 non-treatment cities as well as a variable of dosage level. Hierarchical Generalized Linear Models (HGLM) were developed that controlled for other factors that may have affected the level of violent crime across the sample of cities. The results suggested that PSN treatment cities in higher dosage contexts experienced statistically significant, though modest, declines in violent crime whereas non-target cities and low dosage contexts experienced no significant changes in violent crime during the same period. The limitations of this initial analysis are noted but the evidence seems to suggest that the multi-agency, focused deterrence, problem solving approach holds promise for reducing levels of violent crime. At a minimum, these findings call for continued programmatic experimentation with data-driven, highly focused, deterrence-based violence reduction strategies.
Similar content being viewed by others
Notes
Rosenfeld et al. (2005) used growth curve models and controlled for factors that affect city-level violent crime rates and report less robust impact of the Boston gun project, though it should be noted that Berk (2005) raised a number of methodological limitations with the Rosenfeld et al. critique. In addition, Wellford et al. (2005) found the evaluation of and the reduction in youth homicide in the Boston project compelling despite some of their own methodological concerns.
Rosenfeld et al. (2005) relied on the Law Enforcement Management and Administration Statistics (LEMAS) surveys for their police density measure. However, their study had a significant lag between their period of interest and their analyses, and thus LEMAS data were available for the period of time they examined (i.e., the 1990’s). Complete LEMAS data through 2006 were not available at the time of our study and we substituted with the use of UCR employee data.
These offense percentages were obtained from the Uniform Crime Reports (FBI 2000–2007).
Unfortunately national UCR reports do not distinguish aggravated assaults with a gun from total aggravated assaults and thus were not available for the present study.
In future analyses we plan to use Supplemental Homicide Reports (SHR) in regression models to focus more extensively on gun homicides. The limitation of the SHR data in this preliminary assessment is that many of the cities have small numbers of homicides and thus the population of cities becomes further restricted due to data power issues. As noted subsequently we do provide a basic analysis of SHR firearms homicides as a supplemental analysis.
The selection of cities with a population of 100,000 or greater was based on maximizing the sample size while also providing sufficient base rates of violent offenses to support the analyses.
Results are available upon request.
We were concerned with cases where at least one of the measures (homicide, robbery, or assault) had missing data, but the other offenses had complete data. If aggregation occurred under this circumstance there would be a bias in the measure. Thus, we imputed missing data values prior to aggregation.
Missing homicide data were an issue in 29/1,764 cases, 1.6%. In 16 of these cases, we were able to supplement the missing annual homicide count with the Supplementary Homicides Reports (SHR) data, given that both data sources were initially housed by the FBI reporting system and are created from incident information. For 10 of the remaining 16 cases, we used within-city regressions to impute a missing value for missing homicide data. In the remaining 6 cases, we simply left the homicide count as missing due to the ‘multiple-missing’ data issue.
Missing robbery data were an issue in 34/1,764 cases, or 1.9% of the cases. Missing assault data were an issue in 33/1,764 cases, or 1.8%.
None of the ‘chronic missing data’ cities (Westminster, Co; Olathe, KS; Overland Park, KS; Warren, MI; Akron, Oh; Alexandria, VA; and Chesapeake, VA) were designated as PSN treatment sites.
The coverage of the media campaign is impossible to measure in a fashion that would allow measurement of variation across jurisdictions. It included a national campaign that involved television and radio public service announcements (PSAs) and each district included its own campaign that also included PSAs as well as billboards, posters, and other creative mediums.
Fourteen of 68 districts included in the analyses had multiple large cities that were the focus of PSN intervention.
Zimmermann (2006) notes that additional elements framed by DOJ including media outreach strategies and formal training exhibited extremely low variability across districts and were considered constants and were thus dropped in the aggregation of the overall policy adoption, or dosage variable.
These case studies provided seven “tests” of PSN impact. In all seven cases violent gun crime declined. In two of these sites, the decline was either not statistically significant (Durham, NC) or was observed in comparison sites as well (St. Louis). See McGarrell et al. 2009.
The reality is that for many districts, it was not until 2003 or later that the task force was truly operational and various enforcement, intervention, and prevention components were actually implemented. Thus, the reliance of a common 2002 treatment date results in a conservative test of PSN’s impact as it may discount impact observed in late adopter jurisdictions. The 2002 date is justifiable based on federal prosecution trends. This makes sense in that it is a strategy under the control of the US Attorney’s Office and thus was often the earliest indicator of PSN implementation.
The limitation of this approach is discussed and addressed throughout the results and discussion sections.
We did not include PSN treatment as a level 2 measure and PSN dosage as a level 1 measure within the same model due to the high inter-relationship between these two measures. Specifically, dosage only increased in PSN target cities. However, when examining a model that included both treatment and dosage at different levels, the results were virtually identical to those presented here-in (i.e., the dosage-violent crime relationship remained the same at level 1 with or without the treatment estimate at level 2).
A series of Independent Samples T-Tests comparing measures from the eleven cities excluded from the mixed-regression models were performed. None of the tests were statistically significant (P > .05) where measures existed (including violent crime rates in a given year, level 1 covariates in a given year, and level 2 covariates that were treated as time-invariant), indicating that the non-treatment cities excluded from the analyses were not significantly different than non-treatment cities that were included in the regressions presented here-in.
We felt it necessary to address the concern that PSN implementation could have led to an increase in law enforcement agents at the city level and an increase in incarceration rates at the state level. In this case, increases in police density and incarceration could actually constitute an indirect reflection of PSN implementation. To address this concern, we estimated a growth curve model where annual police density (the outcome at level 1) was a function of a city being designated a PSN site at level 2 (0 = non-treatment site, 1 = treatment site). The estimated effect was actually negative and statistically significant, indicating that PSN sites actually had a larger decline in police per 100,000 residents than did non-PSN sites. The same was true when state incarceration changes were modeled as the outcome variable at level 1. Thus, we find contradictory evidence to the concern that PSN sites actually led to a significant increase in state incarceration rates as well as increases in law enforcement density These relationships were actually negative.
The reduction of 52.1% of the residual variance from the unconditional to the conditional model was computed as a percentage change: (.2066−.4320)/.4320 = −.521 or −52.1%.
Results available upon request.
Refer to Appendix C for the equation for this model.
[exp(−.0252)] = −.024 or −2.4%.
This was a minimal concern in the research done in Chicago by Papachristos et al. (2007) because they included additional relevant time-variant measures such as prosecution changes and sentences associated with federal prosecution in their linear growth models. Thus, they included both static and dynamic PSN treatment measures in one overall model.
Results available upon request.
The above models were also analyzed excluding aggravated assaults from the dependent variable (i.e., based on homicides and robberies). The results were consistent with those presented above and are available upon request.
Twenty-five percent of all US cities with a population of 100,000 or more averaged 3.85 homicides per year, or .32 homicides per month. Fifty percent averaged 11.28 homicides per year, or .94 per month. Seventy-five percent of all large US cities averaged 30 homicides per year, or 2.5 homicides per month.
It should also be noted that as part of the requirements for PSN funding, many districts submitted firearm related offense data to the MSU PSN research team. However, a focused preliminary analysis of these data would have two major flaws: First, only those districts that reported data would be included in the analysis, which would have a tautological relationship with our PSN dosage measure since one of the components of dosage was the use and quality of district crime data. Second, we would not be able to compare target site offense data to non-target sites.
As noted previously, analyses based on SHR reports are being conducted in subsequent stages.
For example, Papachristos et al. (2007) reported a homicide decline of approximately 39% for the specific police districts where PSN was implemented in Chicago. This accounted for much of the city’s decline in homicide. Our measure of the decline in violent crime in Chicago incorporates the target-specific decline but is a more modest decline than that observed in the PSN target areas.
References
Bayley D (1994) Police for the future. Oxford University Press, New York
Berk RA (2005) Knowing when to fold ‘em: an essay on evaluating the impact of Ceasefire, Compstat, and Exile. Criminol Public Policy 4:451–466
Blau JR, Blau PM (1982) The cost of inequality: metropolitan structure and violent crime. Am Sociol Rev 47:114–129
Blumstein A (2006) Crime drop in America: an exploration of some recent crime trends. J Scan Stud Criminol Crim Prev 7:17–35
Blumstein A, Wallman J (2000) Crime drop in America. Cambridge University Press, New York
Braga AA (2008) Pulling levers focused deterrence strategies and the prevention of gun homicide. J Crim Just 36:332–343
Braga AA, Pierce GL (2005) Disrupting illegal firearms markets in Boston: the effects of Operation Ceasefire on the supply of new handguns to criminals. Criminol Public Policy 4:717–748
Braga AA, Kennedy DM, Piehl AM, Waring EJ (2001a) Measuring the impact of Operation Ceasefire. in reducing gun violence: the Boston gun project’s Operation Ceasefire. National Institute of Justice, Washington, DC
Braga AA, Kennedy DM, Waring EJ, Piehl AM (2001b) Problem-oriented policing, deterrence, and youth violence: an evaluation of Boston’s Operation Ceasefire. J Res Crime Delinq 38:195–225
Braga AA, McDevitt J, Pierce GL (2006) Understanding and preventing gang violence: problem analysis and response development in Lowell, Massachusetts. Police Q 9:20–46
Brame R, Bushway S, Paternoster R (1999) Can panel designs tell us anything about continuity and change in offending over time. Criminology 37:588–641
Browning CR, Wallace D, Feinberg SL, Cagney KA (2006) Neighborhood social processes, physical conditions, and disaster-related mortality: the case of the 1995 Chicago heat wave. Am Sociol Rev 71:661–678
Bureau of Justice Statistics (2008) http://www.ojp.usdoj.gov/bjs. Accessed 5 March 2008
Bushway SD, McDowall D (2006) Here we go again-can we learn anything from aggregate-level studies of policy interventions? Criminol Public Policy 3:461–470
Center for Disease Control (2008a) http://webappa.cdc.gov/sasweb/ncipc/leadcaus10.html. Accessed 5 March 2008
Center for Disease Control (2008b) www.cdc.gov/ncipc/dvp/YVP/YVP-conque.html. Accessed 5 March 2008
Chamlin MB, Cochran JK (1997) Social altruism and crime. Criminology 35:203–228
Chen HT (1990) Theory-driven evaluations. Sage, Newbury Park
Cohen J, Ludwig J (2003) Policing crime guns. In: Ludwig J, Cook PJ (eds) Evaluating gun policy. Brookings Institution Press, Washington, DC
Cook PJ, Ludwig J (2000) Gun violence: the real costs. Oxford University Press, New York
Corsaro N, McGarrell EF (2009) Testing a promising homicide reduction strategy: re-assessing the impact of the Indianapolis “pulling levers” intervention. J Exp Criminol 5:63–82
Decker SH, Huebner BM, Watkins A, Green L, Bynum T, McGarrell EF (2007) Project safe neighborhoods case study report: eastern district of Missouri. US Department of Justice, Office of Justice Programs, Washington, DC
Fagan J, Zimring FE, Kim J (1998) Declining homicide in New York City: a tale of two trends. J Crim Law Criminol 88:1277–1323
Farrington DP, Langan PA, Tonry M (2004) Cross-national studies in crime and justice. US Department of Justice, Bureau of Justice Statistics, Washington, DC
Harcourt M (2001) Illusion of order: the false promise of broken windows policing. Oxford University Press, Cambridge
Hipple NK, O’Shea T, McGarrell EF (2008) Project Safe Neighborhoods case study report: Southern District of Alabama. Case study submitted to US Department of Justice, Office of Justice Programs, Washington, DC
Horney JD, Osgood DW, Marshall IH (1995) Criminal careers in the short-term: intra-individual variability in crime and its relation to local life circumstances. Am Sociol Rev 60:655–673
Kane RJ (2006) On the limits of social control: structural deterrence and the policing of “suppressible” crimes. Justice Q 23:186–213
Kelling G, Sousa WH Jr. (2001) Do police matter? An analysis of the impact of New York City’s police reforms. Manhattan Institute Civic Report (December)
Kennedy DM, Braga AA, Piehl AM (2001) Developing and implementing Operation Ceasefire. In Reducing Gun Violence: The Boston gun project’s Operation Ceasefire. National Institute of Justice, Washington, DC
Klofas J, Hipple NK, McDevitt J, Bynum T, McGarrell EF, Decker SH (2006) Crime incident reviews: A Project Safe Neighborhoods strategic intervention. US Department of Justice, Office of Justice Programs, Washington, DC
Koper CS (2005) Purchase of multiple firearms as a risk factor for criminal gun use: Implications for gun policy and enforcement. Criminol Public Policy 4:749–778
Krivo LJ, Peterson RD (1996) Extremely disadvantaged neighborhoods and urban crime. Soc Forces 75:619–650
Kuziemko I, Levitt SD (2004) An empirical analysis of imprisoning drug offenders. J Public Econ 88:2043–2067
LaFree G (1999) Declining violent crime rates in the 1990 s. Annu Rev Sociol 25:148–168
Land KC, McCall PL, Cohen LE (1990) Structural covariates of homicide rates: are there any invariances across time and social space? Am J Sociol 95:922–963
Langan PA, Farrington DP (1998) Crime and justice in the United States, and in England and Wales, 1981–1996. US Department of Justice, Bureau of Justice Statistics, Washington, DC
Levitt SD (2002) Deterrence. In: Wilson JQ, Petersilia J (eds) Crime: public policies for crime control. ICS Press, Oakland
Liska AE, Bellair PE (1995) Violent crime rates and racial composition: convergence over time. Am J Sociol 101:578–610
Ludwig J (2005) Better enforcement, less crime. Criminol Public Policy 4:677–716
Marvell TB, Moody CE Jr (1997) The impact of prison growth on homicide. Hom Stud 1:205–233
McDevitt J, Braga AA, Cronin S, McGarrell EF, Bynum T (2007) Project Safe Neighborhoods case study report: Lowell, District of Massachusetts. US Department of Justice, Office of Justice Programs, Washington, DC
McGarrell EF, Chermak S, Weiss A, Wilson J (2001) Reducing firearms violence through directed police patrol. Criminol Public Policy 1:119–148
McGarrell EF, Chermak S, Weiss A (2002) Reducing gun violence: evaluation of the Indianapolis Police Department’s directed patrol project. National Institute of Justice, Washington, DC
McGarrell EF, Zimmerman C, Hipple NK, Corsaro N, Perez H (2005) The roles of the police in the offender reentry process. Inter J Comp Applied Crim Just 29:53–78
McGarrell EF, Chermak S, Wilson JM, Corsaro N (2006) Reducing homicide through a ‘lever-pulling’ strategy. Justice Q 23:214–231
McGarrell EF, Hipple NK, Corsaro N, Pappanastos E, Stevens E (2007) Project Safe Neighborhoods case study report: Middle District of Alabama. US Department of Justice, Office of Justice Programs, Washington, DC
McGarrell EF, Hipple NK, Corsaro N, Bynum TS, Perez H, Zimmermann CA, Garmo M (2009) Project Safe Neighborhoods - A National Program to Reduce Gun Crime: A Final Project Report. Submitted to National Institute of Justice
Messner SF, Golden RM (1991) Racial inequality and racially disaggregated homicide rates: an assessment of alternative theoretical explanations. Criminology 30:421–447
Messner S, Rosenfeld R (1998) Social structure and homicide: theory and research. In: Smith MD, Zahn MA (eds) Homicide: a sourcebook of social research. Sage, Thousand Oaks
Miethe TD, Hughes M, McDowall D (1991) Social change and crime rates: an evaluation of alternative theoretical approaches. Soc Forces 70:165–185
Miller TR, Cohen MA (1995) Costs of penetrating injury. In: Ivatury R, Clayton CG (eds) Textbook of penetrating trauma. Lee and Civiga, Philadelphia
Moore MH, Braga AA (2003) Measuring and improving police performance: the lessons of Compstat and its progeny. Policing 26:439–453
National Research Council (2004) Fairness and effectiveness in policing: the evidence. National Academies Press, Washington, DC
Office of Justice Programs (2009). www.ojp.usdoj.gov/nij/topics/crime/gun-violence/prevention/project-safe-neighborhoods.htm. Accessed 6 November 2009
Papachristos AV, Meares TL, Fagan JJ (2007) Attention felons: evaluating Project Safe Neighborhoods in Chicago. J Empir Leg Stud 4:223–272
Piehl AM, Cooper SJ, Braga AA, Kennedy DM (2003) Testing for structural breaks in the evaluation of programs. Rev Econ Stat 85:550–558
Piquero AR (2005) Reliable information and rational policy decisions: does gun research fit the bill?”. Criminol Public Policy 4:779–797
Project Safe Neighborhoods (2008) www.psn.gov. Accessed 6 November 2009
Raphael S, Ludwig J (2003) Prison sentence enhancements: the case of project exile. In: Ludwig J, Cook PJ (eds) Evaluating gun policy: effects on crime and violence. Brookings Institute, Washington, DC
Raudenbush SW, Bryk AS (2002) Hierarchical linear models: applications and data analysis methods, 2nd edn. Sage, Thousand Oaks
Raudenbush SW, Bryk AS, Cheong YF, Congdon RT, du Toit M (2004) HLM for Windows version 6.02a. Social Scientific Software publishing, Lincolnwood
Reisig MD, Parks RB (2004) Can community policing help the truly disadvantaged. Crime Delinq 50:139–167
Rennison C (2002) Criminal victimization 2001: changes 2000–2001 with trends 1993–2001. Bureau of Justice Statistics, Washington, DC
Roehl J, Rosenbaum D, Costello S, Coldren J, Schuck A, Kunard L, Forde D (2008) Paving the way for Project Safe Neighborhoods: SACSI in 10 US Cities. National Institute of Justice. Research in Brief, Washington, DC
Rosenfeld R, Fornango RR, Baumer E (2005) Did Ceasefire, Compstat, and Exile reduce homicide? Criminol Public Policy 4:419–450
Sampson RJ, Raudenbush SW (1999) Systematic social observation of public spaces: a new look at disorder in urban neighborhoods. Am J Sociol 105:603–651
Sherman LW, Rogan DP (1995) The effects of gun seizures on gun violence: ‘Hot spots’ patrol in Kansas City. Justice Q 12:673–693
Sherman LW, Shaw JW, Rogan DP (1995) The Kansas City gun experiment. National Institute of Justice, Research in Brief, Washington, DC
Spelman W (2000) The limited importance of prison expansion. In: Blumstein A, Wallman J (eds) The crime drop in America. Cambridge University Press, New York
Symposium (1998) Symposium: why is crime decreasing? J Crim Law Criminol 88:4
Tita G, Riley KJ, Ridgeway G, Grammich C, Abrahamse A, Greenwood P (2003) Reducing gun violence: results from an Intervention in East Los Angeles. RAND Corporation, Santa Monica
Weisburd D, Braga A (2006) Police innovation: contrasting perspectives. Cambridge University Press, New York
Wellford CF, Pepper JV, Petrie CV (2005) Firearms and violence: a critical review. The National Academy Press, Washington, DC
Wilson JW, MacDonald JM, Tita GE (2010) Localized homicide patterns and prevention strategies: a comparison of five Project Safe Neighborhoods sites. Victims Offenders 5:1–19
Xie M, McDowall D (2008) The effects of residential turnover on household victimization. Criminology 46:539–575
Zimmermann CR (2006) Federal incentives to address gun violence: a model of success and failure. Unpublished Doctoral dissertation. School of Criminal Justice, Michigan State University, East Lansing
Zimring FE, Hawkins G (1999) Crime is not the problem: lethal violence in America. Oxford University Press, New York
Acknowledgment
This project was supported by Grant No. 2002-GP-CX-1003 awarded by the National Institute of Justice, Office of Justice Programs, US Department of Justice. Points of view in this article are those of the authors and do not necessarily represent the official position or policies of the US Department of Justice.
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix A
Appendix B
The HGLM model presented in Table 2 can be written as follows (Note: All level 1 measures were group-centered, all level 2 measures were grand-mean centered, and no random variance components were estimated for the annual dummy variables because all cities had the same fixed value for this measure):
where η ti = the expected violent crime rate, π0i = β00 + β01(Disadvantage) + β02(PopDensity) + r oi , π1i = β10 + r 1i , π2i = β20 + r 2i , π3i = β30 + r 3i , e π4i + ··· + π9i = β40+ ··· + β90. Thus, the reduced two-level equation can be written as:
The outcome in the HGLM is the count of violent crimes at level 1 and includes time-varying covariates within the cities (i.e., level 1 measures) and time invariant measures at level 2. As seen in the above equations, PSN dosage is included as a time-varying measure at level 1. Here, the HGLM model uses an overdispersed Poisson sampling model at level 1 and a log link function to equate the transformed count into a linear structural model. The log link function in the HGLM is used to equate the transformed count into a linear structural model, consistent with the regression-based analytic approach.
In the analysis, the HGLM Poisson model assumes an expected violent crime count
where λ it is expressed as the violent crime rate of a city i at time t and m it is the exposure measure, which is expressed as the city population in 100,000 s. The expected violent crime rate for a city is transformed through a natural logarithmic function, where η ti = ln(λ it ). The logged event rate, η ti , becomes the dependent variable in the level 1 model.
Appendix C
The HGLM model presented in Table 3, which was the alternative growth curve model where the PSN treatment city designation is a static measure at level 2, is written as follows (Note: all level 1 measures were group-mean centered, and level 2 measures were grand-mean centered):
where π0i = β00 + β01(PSN Treatment City) + β02(Disadvantage) + β03(PopDensity) + r oi , π1i = β10 + r 1i , π2i = β20 + r 2i , π3i = β30 + β31(PSN Treatment City).
Thus, the reduced two-level equation can be written as:
The outcome in the HGLM is the count of violent crimes at level 1 and includes time-varying covariates within the cities (i.e., level 1 measures) and time invariant measures at level 2. As seen in the above equations, PSN treatment is a level 2 static measure. Here, the HGLM model uses an overdispersed Poisson sampling model at level 1 and a log link function to equate the transformed count into a linear structural model. The log link function in the HGLM is used to equate the transformed count into a linear structural model, consistent with the regression-based analytic approach.
In the analysis, the HGLM Poisson model assumes an expected violent crime count
where λ it is expressed as the violent crime rate of a city i at time t and m it is the exposure measure, which is expressed as the city population in 100,000 s. The expected violent crime rate for a city is transformed through a natural logarithmic function, where η ti = ln(λ it ). The logged event rate, η ti , becomes the dependent variable in the level 1 model.
Rights and permissions
About this article
Cite this article
McGarrell, E.F., Corsaro, N., Hipple, N.K. et al. Project Safe Neighborhoods and Violent Crime Trends in US Cities: Assessing Violent Crime Impact. J Quant Criminol 26, 165–190 (2010). https://doi.org/10.1007/s10940-010-9091-9
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10940-010-9091-9