Predictive Policing Research

On this page find:

Predictive Policing Demonstration and Evaluation

For the first phase of a two-phase effort, NIJ issued a competitive solicitation for Fiscal Year 2009 seeking proposals from law enforcement agencies to plan predictive policing models in their jurisdictions. Seven agencies won awards to develop plans to implement a predictive policing demonstration and evaluation program. For the second phase, NIJ issued a solicitation for Fiscal Year 2011 that invited Phase 1 awardees to submit proposals to implement the plans they developed. Two agencies — the Chicago Police Department (CPD) and the Shreveport (La.) Police Department — won grants to continue their work by putting their plans in place.

RAND Corp. won an award to provide technical assistance to the agencies in Phase 1 and to serve as an independent evaluator in Phase 2. In its role as evaluator, RAND will evaluate the efficacy of the concept of predictive policing as seen in Chicago and Shreveport’s predictive policing models.

The Chicago Program
CPD received an award to implement a pilot that in part will evaluate the ability of modified pattern-matching software currently used for medical diagnostic purposes to predict crime patterns. It will also evaluate the efficacy of a software tool that quantifies and maps gang activity to predict emerging areas of gang conflict. RAND will evaluate the pilot in terms of accuracy of prediction, process and impact, using randomized, retrospective and quasi-experimental studies. CPD’s research partner is the Illinois Institute of Technology (IIT).

The Shreveport Program
The Shreveport Police Department received an award to implement a pilot that will use a randomized experimental design using experimental and control groups involving six of the highest-crime policing districts in Shreveport. The pilot will evaluate the “broken windows” theory of policing in an operational setting and employ a predictive model using leading indicators related to that theory such as juvenile complaints, loud music, disorderly persons, suspicious activity, loitering, disputes and prowlers. RAND will measure the efficacy of the pilot in terms of its ability to reduce tactical crimes such as shootings, robbery, burglary, auto break-ins, outside residential thefts, outside business thefts and auto thefts.

Geospatial Police Strategies

In Fiscal Year 2012, NIJ awarded grants to the Police Foundation and Rutgers University to study geospatial strategies in policing.

Risk Terrain Modeling Experiment: A Multi-Jurisdictional Place-Based Test of an Environmental Risk-Based Patrol Deployment Strategy, Rutgers University, NIJ grant 2012-IJ-CX-0038 (See grant details)
Does sending police patrols to areas at high-risk for crime affect the number and location of new crimes? Rutgers University is conducting an experiment to measure the extent of the effect. Researchers are using Risk Terrain Modeling (RTM) to define high-risk areas.[1] The project has two primary goals: 1) to replicate and validate RTM in multiple jurisdictions and across many crime types; and 2) to evaluate theoretically- and empirically-grounded risk-based interventions targeted at high-risk micro-level environments.[2] High-risk areas are matched with equivalent control areas through a propensity score matching technique. Participating police agencies are Arlington, Texas; Chicago, Ill.; Colorado Springs, Col.; Glendale, Ariz.; Kansas City, Mo.; and Newark, N.J.

Translating “Near Repeat” Theory into a Geospatial Police Strategy: A Randomized Experiment Testing a Theoretically-Informed Strategy for Preventing Residential Burglary, The Police Foundation, NIJ grant 2012-IJ-CX-0039 (See grant details)
Research has shown that once a burglary occurs on a street, the homes on that street and on nearby streets are at a much higher risk of burglary over the next one to two weeks. This research finding has not yet been translated into actionable crime prevention strategies for police agencies. To address this gap, the Police Foundation will use a randomized controlled trial to test whether quickly notifying community residents that they are at an increased risk for a burglary and providing them with burglary prevention tips reduces incidents of further burglary in the high-risk time period. Participating police agencies are Baltimore, Md., and Redlands, Calif.


[1] RTM is an approach to spatial risk analysis that uses a geographic information system to attribute qualities of the real world to places on a digitized map. It operationalizes the spatial influence of risk factors to common geographic units, then combines separate layers to produce “risk terrain” maps showing the presence, absence or intensity of all risk factors at every location throughout the landscape. Theoretically- and empirically-grounded risk terrain maps show where conditions are conducive for crimes or other hazardous events to occur in the future. RTM offers a statistically valid way to articulate and communicate criminogenic and vulnerable areas at the micro level.

[2] The term micro-level environments has no set definition and can refer to a variety of units smaller than a city, such as a neighborhood, block group, block, street face or even an individual address.

Date Created: January 13, 2014