Policing predicted crime areas: An operationally-realistic randomized, controlled field experiment

This proposal presents a place-based, randomized experiment to study the impact of different police strategies on violent and property crime in predicted criminal activity areas. Predictive policing is an emerging tactic relying in part on software predicting the likely locations of criminal events. At present the field lacks robust evidence to suggest the appropriate policing tactic in predicted areas. That is the subject of this timely study. We will answer the question of whether different varieties of theoretically informed but also operationally realistic police responses to crime predictions estimated by a predictive policing software program can reduce crime.
The research team will randomly assign 20 Philadelphia Police Department (PPD) districts into one of four experimental conditions. Five districts will act as controls, with a business-as-usual patrol strategy. In five districts, officers will be made aware of the predicted high crime activity areas at roll call and asked to concentrate there when able; this is an awareness model. Five districts will receive the awareness model treatment as well as an additional patrol car solely dedicated to the predicted crime areas. Finally, five districts will receive an intelligence-led, investigative response with an unmarked unit dedicated to the predicted areas. This project is a collaboration between Temple University academics and the PPD who have worked together for many years.
In terms of theory, the design permits comparing deterrence theory, with two levels of uniform patrol response, and incapacitation theory. Block randomization will ensure a spread of high to low crime areas in each experimental condition. This pragmatic research design reflects realistic responses of police to predictive estimates of future crime locations. Once the experiment is carefully set up, PPD will spend three months focusing on property crime (burglary and vehicle theft), followed by three months focusing on violent crime. Statistical analyses will generate estimates of the impacts and differential impacts in predicted crime areas of the three theory-linked policing approaches. Different analyses using different assumptions will help gauge robustness of findings. Qualitative research will verify the experimental fidelity and generate insight into the statistical patterns that will emerge.
Results from this research will be disseminated in a variety of electronic formats and made available to leading policing organizations. The research team also will disseminate findings online in short, accessible formats to help spread the research to the police professional community. A dedicated website and numerous academic journal articles will provide an ongoing resource. ca/ncf