![Sample our Law journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5942/TOC-Subject-Sample-2021-Law.jpg"})
Abstract
The most fitting way to celebrate Joan Petersilia’s numerous advancements within the field of criminal justice is to deliberately practice the two most important lessons that perhaps best define her legacy. First, embedded research should drive collaborative partnerships to foster real-world change based on science. Second, such real-world change should reduce the consequences of crime as well as the social costs of heavy-handed criminal justice sanctions within distressed communities to transform criminal justice policy and practice. In this paper, we outline future directions for policing practice and research as a result of the embedded, collaborative partnerships that led to the Tulsa Community-Based Crime Reduction (CBCR) initiative. The Tulsa CBCR initiative and subsequent evaluation shows that police-driven, collaborative stakeholder partnerships (including business owners, resident steering committees, local government, police, and researchers) significantly reduced crime within targeted areas, allayed many citizens’ perceptions of risk, and improved residents’ attitudes about the Tulsa Police Department.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 Initially the TPD had assigned an experienced patrol officer with a strong community policing orientation and committed to build relationships with residents in the Savanna Landing complex. This officer, however was reassigned to other duties when he was promoted to Sergeant. The TPD committed to the hiring a dedicated Community Resource Officer (using grant funds to initially fund the position), and later retained the officer using city funding at the close of the grant period. While the initial plan was delayed due to these changes, the long-term relationships across partners was enhanced, and enabled the continuation of the work beyond the current project.
2 Project personnel from the technical assistance team - Local Initiatives Support Corporation (LISC) - were also heavily integrated into the project and were important contributors to and managers of various project deliverables.
3 The use of 24-hour private security provided both benefits and challenges. In terms of positive impact, security provided a constant approach to impact disorder, they were on-site to control drug crime problems, and they offered an immediate response to trespassing among non-residents. However, several challenges were also noted during focus group interviews, including residents’ perceptions that security officers were sometimes rude or acted curtly to residents and apartment staff. The partnership between TPD and the Tulsa Community Resource Officer was helpful, as TPD officials offered guidance and suggestions for security personnel to improve demeanor. Resident interviews in 2018 showed that the initial brusque reactions by security personnel were reduced almost immediately after TPD intervention.
4 The no fighting tolerance policy was almost universally for outdoor fights/assaults – and to our knowledge had little to no impact on domestic violence in the complex. The ‘street fights’ were largely among young, male residents. An unintended consequence of this policy was the apparent displacement (though interviews suggest at a much lower frequency) of fights to nearby non-residential alleyways and hidden areas – which went largely unreported to police because they took place outside of public view. Apartment managers were content with this unintended consequence because it minimized the viewing/acceptance of fighting as a suitable approach to conflict resolution in the presence of residents, staff, and children.
5 The convenience samples were obtained conducting ‘walkthroughs’ at Savanna Landing by the CBCR project manager and the Tulsa community resource officer. CBCR team members would stop and speak with residents (and fill out the surveys by hand with the respondent) regarding questions about crime, safety, and perceptions of local initiatives. Often, respondents would ask resident-peers to also talk with the CBCR team and complete the surveys to provide more thorough feedback. A similar survey approach was used in the Austin (TX) Rundberg CBCR (formerly Byrne Criminal Justice Innovation) grant – see Springer, Lauderdale, Landuyt, and Cole (Citation2012).
6 The 2017 survey was both web-based (48%) and paper-based (52%) to all Savanna Landing units, with a final N = 114 respondents (out of a possible 336 unit complex). However, the occupancy rate at that time was roughly 90%, which would equate to roughly a 37.7% response rate. The demographic profile from the sample was comparable to the geographic census tract block data, obtained from the 2010 SF1 Census table (Block 6801, Block group 1014) – where 39.5% of the population were male (47% were male in the survey), 40.3% were Black (43% were Black in the survey), 41.3% were White (34.0% were White in the survey), and 100% of the respondents were renters (100% were renters in the survey). The later convenience samples (2018 N = 84; 2019 N = 46) demographics were similar.
7 We examined the goodness-of-fit statistics for each full regression and chose, where appropriate, negative binomial regression models in place of Poisson models when the Chi-Square p-value statistics were statistically significant (p < .05), which indicates evidence of overdisperion (Long & Freese, Citation2003, p. 270).
8 Comparing variance and mean instabilities from the remainder of Tulsa would not serve as an appropriate counterfactual framework due to parallel trend assumption violations (see Card & Krueger, 1994). The use of the Riverside Division was particularly salient for our counterfactual control analysis because the remainder of the City of Tulsa did not have commensurate crime rate levels relative to the Division area where the Tulsa CBCR initiatives took place. Riverside accounted for over 37% of all the total crimes in the City of Tulsa from 1/2010 to 12/2020.
9 The total number of offenses (N = 1,515) at Savanna Landing were all criminal offenses that occurred within the site; in terms of specifics the crimes were primarily assaults and burglaries (35%) as well as a combination of minor property offenses (larcenies), and public disorder offenses (trespassing, disorderly conduct, narcotic charges), which comprised an additional 33% of all offenses. For crime incidents at the Kwick Stop (N = 180), 33% of the total number of offenses were narcotic offenses, 22% were assaults, while the remainder of offenses were comprised of disorderly offenses, trespassing, and larceny among other less frequent events.
10 The same intervention dates for each set of site-specific models were applied to the remainder of the Riverside Division in order to properly control for potential breaks, trends, or shifts that took place nearby.
11 A series of sensitivity tests were conducted on each of the models – though not all of the results were presented in the tables presented here-in for parsimony. Given that count regression models rely on the use of Maximum Likelihood (ML) estimation and we include the same covariates to control for linear and curvilinear trends and seasonality, this is an appropriate statistical control to account for the first-order autocorrelation process (Harvey, Citation1990). All regression analyses included the exploration of the possibility of broad potential trend influences by adding a simple linear trend variable (to account for linear trends) and a trend-squared variable (trend^2 to account for curvilinear trends) in each model and table presented below. At no point did the included trend measures alter the results in any meaningful or substantive manner, and thus were excluded from presentation. The count regression time-series model(s) can be written as follows: Monthly count outcomes = Intercept + Intervention + Trend + Trend2 + Monthly Dummy Variables + (Lagged Offenses (Yt-1) when Offenses was the DV +) Error Term
12 We present Clogg-Z coefficient difference tests when the treatment areas and control areas both had statistically significant direction changes in crime outcomes at the same point in time (for a relative rate change difference score). In circumstances when the treatment area experienced a significant decline but the control area did not, or when neither area experienced a significant shift, we did not include Clogg-Z scores due to space constraints.
13 The Clogg Z coefficient score relied on the Paternoster, Brame, Mazerolle, and Piquero (Citation1998) standard error summation correction: difference estimate = -.476, st. error = .232.
14 The models did not yield any estimated slope change for the Kwick Stop offenses across any crime type, but given the small sample size at the site the null findings for the site was anticipated.
15 The difference-in-difference estimate was simply modeled as Y = α + β1T + β2X + β3TX + e where T = treatment assignment = 1, control = 0; X = pre-post where 0 = pre-time series and 1 = post-time series; and TX is the difference-in-difference estimator. This model was used to examine model fit regarding the level-change in the time series. The slope estimate was modeled as Y = α + β1T + β2X + β3TX + β4Z+ β5ZT+ β6ZX+ β7ZXT + e where T = time (1-110 for each time series), X = study phase pre-post, TX = number of periods after the shift in the time series, Z = treatment/control assignment; ZT = time for treatment (0 for control); ZX = study phase for treatment (0 for control), and ZTX = time for interruption for treatment (0 for control). The ZTX estimate is presented in the sensitivity analysis section, which equated to a 1% marginally significant (p <.10) shift in the slope of the time series.
16 As an additional supplemental check to ensure the reductions in crime were not calibrated with a rise in aggressive patrol such order maintenance policing we conducted bivariate count trend analysis of arrest patterns for the target neighborhood. Two important findings were observed: 1) all arrests decline post-2017, and 2) standard order maintenance proactive patrol arrests (i.e., OWI, disorderly conduct, and resisting arrests) – see Sampson (Citation1986) – were seasonally-stable each year (averaging roughly 6.0 to 13.3 per month) from 2016 to 2020. These patterns combined with narrative accounts from the community residents indicate that there is little reason to suspect the change in crime outcomes was due to a broad and aggressive increase in general patrol enforcement.
Additional information
Notes on contributors
Nicholas Corsaro
Nicholas Corsaro, Ph.D., is an Associate Professor of Criminal Justice at the University of Cincinnati (UC) and is the Research Director of the International Association of Chiefs of Police/UC Center for Police Research and Policy. He has published over 25 peer reviewed articles and has served as a Principal or Co-Principal Investigator in over 20 grants and contracts. He has extensive experience working with law enforcement agencies to develop, implement, and evaluate promising practices related to crime prevention – and many of his peer reviewed papers appear in Crime and Delinquency, Criminology and Public Policy, Evaluation Review, Journal of Experimental Criminology, Journal of Quantitative Criminology, Journal of Urban Health, and Justice Quarterly.
Robin S. Engel
Robin S. Engel, Ph.D. is a Professor of Criminal Justice at the University of Cincinnati (UC) and Director of the International Association of Chiefs of Police/UC Center for Police Research and Policy. She recently served as UC’s Vice President for Safety and Reform, where her administrative duties included oversight of the daily operations and reform efforts of the University of Cincinnati Police Division. Dr. Engel engages in police research and evaluation, with expertise in empirical assessments of police behavior, police-community relations, and crime reduction strategies. She promotes best practices in policing by establishing academic-practitioner partnerships, has served as Principal Investigator for over 80 research grants and contracts, and published extensively in the fields’ most prestigious peer-reviewed journals.