The Indianapolis harmspot policing experiment
Introduction
Recently, the National Academies of Sciences, Engineering, and Medicine published a comprehensive report, Proactive Policing: Effects on Crime and Communities (National Academies of Sciences, Medicine, et al., 2018), summarizing the collective evidence of various proactive policing strategies on crime and community impact. Among several conclusions and recommendations were that 1) place-based approaches yielded significant crime reduction benefits with minimal displacement and community dissatisfaction; 2) problem-solving strategies aimed at root-causes of issues generated short-term reductions; and 3) community-based programs that leverage existing community resources appear promising, though are limited by evaluations with weak designs. These various approaches to proactive policing are grounded by different logic (see (Weisburd et al., 2019)), each which demonstrates the complex and nuanced mechanisms police must consider when attempting to fashion crime prevention strategies.
Place-based proactive strategies exhibit the strongest evidence that police can reduce crime (National Academies of Sciences et al., 2018), primarily achieved through hot spots policing strategies (Braga, Turchan, Papachristos, & Hureau, 2019). What remains less certain are the methods through which police and scholars identify hot spots for effective intervention. Low levels of spatial overlap among hotspots of various crime types has been observed (Haberman, 2017), suggesting police departments may have to pick and choose offense types for optimal outcomes, or triage the fewest places with highest overall crime. Telep and Hibdon extend this operational challenge and contend hot spots policing should move beyond crime counts as “…analyses in these hot spots should be guided by additional, non-police data sources…[and]…that a reliance on crime data (incidents or calls) alone may oversimplify or distort the distribution of crime or other harmful activity” (Telep & Hibdon, 2017, p. 662).
There exists a growing recognition that placed-based policing, and policing more generally, should move beyond crime to focus on “social harm”. Ratcliffe articulates this broadened approach to policing and urges the inclusion of multiple crime types, various drug and forgery offenses, and traffic crashes to be integrated as “The inclusion of these supplementary metrics is more reflective of the multidimensional responsibilities of the police in the community” (Ratcliffe, 2015, p. 176). Moreover, recent work has shed light on the co-occurrence of substance abuse, mental health, and physical health problems pervasive within crime hot spots (Goldberg, White, & Weisburd, 2019; White & Goldberg, 2018; White & Weisburd, 2018; Wood, 2020). Furthermore, recent research also suggests that micro-places for police intervention can measure “harm” as opposed to event counts. Harm is an event-weighted calculation of severity which more accurately captures the seriousness of issues facing micro-places as well as enables police to identify different harms plaguing different geographies (Ratcliffe, 2015; L. Sherman, Neyroud, & Neyroud, 2016; Weinborn, Ariel, Sherman, & O'Dwyer, 2017).
In sum, place-based police interventions are effective in reducing crime and there is currently a movement to enhance methods used to identify intervention micro-places. This movement suggests that events should be weighted with respect to severity, or impact on society, and that such event types should extend beyond traditional crime counts to include incidents that reflect harms experienced by different communities. To date, there has yet to be an empirical test of a dynamic proactive place-based policing intervention that leverages a broader set of social harm incidents, each weighted by severity. These harms include criminal events, self-harm, and potential harm from racially biased policing. The current study reports the findings of a federally-funded experiment in Indianapolis which sought to engage these specific questions. Lastly, the present study reports community perceptions of data-driven analytics that guide police activity.
Section snippets
Place-based policing interventions and the movement to social harm
Place-based policing strategies leverage the empirical reality that large portions of crime are concentrated within small geographic bandwidths across urban landscapes (Weisburd, 2015). As approximately 50% of crime occurs within 5% of a city geography, police can then identify focal areas for proactive activities, service delivery, and enforcement (Weisburd, 2018). Hot spots policing is the primary vehicle through which crime concentration at place is translated to a policing strategy. In
Study objectives
Our harmspot policing experiment has six primary objectives. We operationalize dynamic hot spot policing with a broadened set of events consistent with recent movements for police to consider social harm as opposed to just crime. Strong evidence suggests hotspots policing generates significant short-term crime reductions, however we seek to observe if these crime prevention benefits can be achieved when identifying intervention micro-places using social harm events and weighting as well as
Experimental methods
Indianapolis, Indiana served as the study location, which is the largest city in the state, the state capital, and a consolidated city-county municipality. Although Marion County and Indianapolis share city-county boundaries, the cities of Beech Grove, Lawrence, Southport, and Speedway are independent municipalities with their own police departments also located within Marion County and thus fall outside of the Indianapolis Metropolitan Police Department's (IMPD) jurisdiction. As of July 2019,
Community survey methodology
To capture citizen perceptions of data-driven policing in Indianapolis, pre- and post-experiment survey waves were administered. The pre-experiment survey was conducted in the month preceding, and the post-survey in the week immediately following, the field trial. Surveys were conducted via phone using live callers and computer assisted telephone interviewing software. Phone number sample was sourced randomly from Dynata and comprised of Indianapolis city residents only. Both cellular and
Results
In Table 4 we show the estimated coefficients for the regression corresponding to Eq. (4). Here we find that the effect of proactive activities in treatment harmspots is significant at the p = .0295 level, whereas the effect of an activity in a control hotspot is not statistically significant. To quantify the effect of a single activity in a harmspot, we add the coefficients β + δ = − 38.6. Given the average proactive activity lasted 10.4 min, this result can be interpreted as a lowering of
Community survey results
In Fig. 3 we display results from the first survey question asking respondents to consider the statement “the police should use data analytics to predict where crime is most likely to happen” and then respond with strongly disagree, somewhat disagree, somewhat agree, or strongly agree. Here we find that the majority of respondents either somewhat or strongly agree that police should use data analytics to predict crime, an attitude that is consistent across racial/ethnic subgroups of the
Discussion
Over the course of a 100 day field trial during the summer of 2019, the IMPD engaged in dynamic harmspot and hotspot policing. Officers utilized a web application to guide proactive activities that included vehicle and foot patrols, high-visibility positioning of marked cars for traffic enforcement, community education of data-driven policing at micro-places, and enhanced awareness of local substance abuse services. Our experiment focused on six primary objectives, and these guide our
Acknowledgements
This research was supported by NSF grants SCC-1737585 and ATD-1737996. We sincerely thank the Indianapolis Metropolitan Police Department for their commitment to this project, and their service to the community. Specifically at IMPD, we thank Chief Randal Taylor, Deputy Chief Kendale Adams, former Chief Bryan Roach, and the entire Command Staff for their support of this project. We also thank Indianapolis Emergency Medical Services, and specifically Chief of IT and Informatics Tom Arkins. We
References (81)
- et al.
The (in) appropriateness of aggregating across crime types
Applied Geography
(2012) - et al.
Mapping crime harm and priority locations in New Zealand: A comparison of spatial analysis methods
Applied Geography
(2017) - et al.
Poor health and violent crime hot spots: Mitigating the undesirable co-occurrence through focused place-based interventions
American Journal of Preventive Medicine
(2020) Area-wide urban traffic calming schemes: A meta-analysis of safety effects
Accident Analysis & Prevention
(2001)- et al.
Mapping suicide mortality in Ohio: A spatial epidemiological analysis of suicide clusters and area level correlates
Preventive Medicine
(2018) Exploring the geography of suicide threats and suicide attempts: An application of risk terrain modeling
Social Science & Medicine
(2020)- et al.
Spatial patterning, correlates, and inequality in suicide across 432 neighborhoods in Taipei City, Taiwan
Social Science & Medicine
(2019) - et al.
The cost of crime to society: New crime-specific estimates for policy and program evaluation
Drug and Alcohol Dependence
(2010) - et al.
Improving social harm indices with a modulated hawkes process
International Journal of Forecasting
(2018) - et al.
Hotspots vs. harmspots: Shifting the focus from counts to harm in the criminology of place
Applied Geography
(2017)