LAPD Research Paper “Fighting Crime in the Information Age: The Promise of Predictive Policing”

Fighting Crime in the Information Age: The Promise of Predictive Policing

  • William Bratton, Altegrity Security Consultants
  • John Morgan, National Institute of Justice
  • Sean Malinowski, Los Angeles Police Department
  • 15 pages
  • Draft
  • November 18, 2009

Download

WHAT IS PREDICTIVE POLICING?

Innovative strategies like Compstat and community-oriented policing have revolutionized police practice in the last twenty years. Unlike previous generations of police managers, we have greater capability than ever before to use information to drive innovative strategies. We can consider a broader, deeper set of policing or criminal justice interventions because we can understand our environment better, even forecasting what is likely to happen in the future. The term, predictive policing, is therefore not meant to tie a police chief or anyone else to a particular strategy or cause us to abandon proven police techniques. Rather, police leaders need to be thinking about what’s going to happen that may affect law enforcement and crime, what the implications are for their work, and how they may improve the future environment by acting today. The toolkit they may employ may involve any of the myriad strategies or interventions or innovations that are available to us.

What is critical is that police agencies develop and use good information and cutting-edge analysis to inform forward-thinking crime prevention. Predictive policing connects technology, management practices, real-time data analysis, problem-solving and information-led policing to lead to results—crime reduction, efficient police agencies, and modern and innovative policing. Just as Compstat and mapping technologies have enabled management accountability at the precinct level and below, advanced information tools now permit predictive policing and the proactive deployment of police resources to respond to crime trends as they occur or even before. Research suggests that changes in crime rates at the city level are driven by changes that are highly localized, perhaps even at the block or building level. Good police chiefs have known this for many years. We can track crime at the block level and connect it to how police are managed through Compstat, or Comprehensive Computer Statistics, which incorporates intelligence, resource allocation, tactics, and follow-up. Accountability is enforced typically at the precinct commander level, because that individual has tactical situational awareness, high levels of training and experience, and interaction with higher-level executives in a police department. In other words, the precinct commander intuitively knows which blocks or buildings are the problems in the local area and which police tactics will work best in those places. With this knowledge, the local commander can employ problem-oriented policing solutions tailored to the local community. Essentially, Compstat is an early information-age product, since it is dependent on real-time intelligence data and analysis about crime at the precinct level, something made possible by the advent of modern computer systems. A typical Compstat session will overlay geographic distributions of particular types of crime to inform tactical deployments in a precinct. However, what we call Compstat is itself undergoing radical change. We now hear of police departments that identify a particular problem, like vehicle theft or drug activity, and decide that they want to “compstat it.” They mean that they want to track the performance of their department against an outcome that is meaningful to their community and hold themselves accountable for the improvement of that outcome. Furthermore, this analysis should have a predictive element. Analysts can identify, track and correlate repeat or potential offenders, hot spots, victims and witnesses, response times, patrol activity, and other variables. Advanced analysis can be based on intelligence gleaned from the community, incident reports, economic data such as housing foreclosures, and even weather reports. In this way, it is possible to predict where crime may increase next week or next month, thus enabling proactive deployment or other, fundamental crime prevention strategies. This is a major turning point for law enforcement and turns the old, reactive model of policing completely on its head. Now, we intend to be proactive, focusing on lead times and performance instead of response times and last year’s crime rates. Further, we intend to use all of the tools at our disposal to anticipate criminal activity, including all elements from the city planning department and youth services to patrol and investigation, and everything in between.

PREDICTIVE POLICING AND OTHER POLICING MODELS

Policing in America has proven to be an extraordinary laboratory for experiments with innovative practices. There are those who have a tendency to apply labels to particular policing strategies, and that has led to a proliferation of so-called policing paradigms. The list includes problem-oriented policing, community-oriented policing, performance-based policing, evidence-based policing and professional model policing, as well as many others. Some models are related to predictive policing. In the 1980’s, Jean-Paul Brodeur coined the term, “high policing” to refer to policing that involves domestic intelligence gathering or national-security-related activities. In the United States, high policing is the responsibility of the Federal Bureau of Investigation, Secret Service, and many regional drug intelligence activities. Because these efforts are tied now to state and local policing through fusion centers and other activities, so we may expect that predictive analysis will play a big role in how police work together to respond to terrorism, organized crime, and similar threats. Intelligence-led policing has become a popular term since the terrorist attacks of September 11, 2001. Advocates of intelligence-led policing have established terrorism-focused fusion centers and information sharing standards. More broadly, information-led policing encompasses the use of many forms of information by police, including mobile data terminals in police cars and sensor systems (e.g., gunshot detection systems). Information-led policing and intelligence-led policing strategies each anticipate close connections between information analysts and law enforcement operations, which is a necessary part of predictive policing. Evidence-based policing encourages partnerships between police and researchers, but many evidence-based policing programs fail because departments don’t have the information to inform interventions or track results. Thus, a predictive policing environment will provide fertile ground for evidence-based approaches to achieve broader success. Problem-oriented approaches are directly tied to predictive methods, because predictive methods permit the early identification of crime problems and development of critical solutions.

Often, one is left with the unfortunate impression that a police chief must choose a particular model to the exclusion of all others. So, the failure of a department may then be attributed to the poor decision of a police chief about the model of policing employed or inadequate execution of the model. In fact, no single model should be employed in any department in the complex environment of real policing. Each department faces unique challenges that must be overcome for any improvement to occur. Also, no department could completely employ any individual strategy or intervention in its purest form, nor should it.

ELEMENTS OF PREDICTIVE POLICING

We suggest that predictive policing, while just a label like any other, is an important way to think about policing innovation and management. Predictive policing enables other strategies but is not a substitute for them. Nonetheless, there are specific elements of a successful predictive policing environment that can be recognized, regardless of the other methods that are employed.

  • Integrated information and operations: Large police departments maintain dozens of databases. It is unusual to see these computer systems linked together to enable effective analysis. It is even more unlikely that other information sources, such as gunshot detection systems or dispatch systems, are linked into police analytical or fusion centers. Finally, police departments do not link their operations and information systems to other parts of the justice system or social services system. Thus, poor information sharing prevents good analysis and investigation. Even more troubling, poor information sharing can undermine efforts to intervene with individuals or neighborhoods to stop the cycle of violence. The best way to see the future and act appropriately is to have a complete picture of the current situation. Police must integrate their information and activities to enable situational awareness.
  • Seeing the Big Picture: Few crimes are isolated incidents. Most crime is part of a larger pattern of criminal activity and social issues. This means that police need to be able to see these patterns in communities. Police need to use predictive intelligence to get out of the trap of day-to-day crisis control. Day-to-day activities dominate the time and energy of the police, despite reforms of the last thirty years. Response times to 911 calls and case clearance rates are used to measure performance, but these measures can be traps that prevent the police from acting effectively. Most of the emerging strategies of the last twenty years have been attempts to get law enforcement to broaden police work to see the big picture. For these interventions to work, it is necessary to build police organizations to use information to see the big picture patterns of what is going on around them.
  • Cutting-edge analysis and technology: Police departments are information-rich but anlaysis-poor. Tomorrow’s forward-thinking department must rely on good information that has been fully analyzed and disseminated. Good information comes from police reports, intelligence, sensors (e.g., license plate readers), forensic technology, and integrated information from multiple sources. To be useful, the information must be analyzed by police analysts with the proper tools and turned into usable products for the front-line officer. Predictive analysis may include tools that link people or activities, statistical techniques, geospatial tools, or visualization of complex interrelationships. With the advent of fusion centers, we now take for granted that we should use police intelligence to deal with terrorism. We should also use these kinds of resources to deal with domestic violence or identity theft, re-entry of parolees into communities, and other crime problems.

Share this:

Facebooktwitterredditlinkedinmail