How Data Helps Law Enforcement Turn Criminal 'Playgrounds' Into Safe Spaces

Spatial analysis makes smart urban planning into passive, bullet-free policing.

Pip Wilson/Flickr

Though it’s tempting to diminish criminals to nothing more than motivations or mistakes, it’s critical to remember that wrongdoers exist in physical spaces that influence their behavior. Lawbreakers are nearly as predictable as the rest of us. They don’t try to rob banks located next to police stations. They don’t break into day care centers. They don’t flee down highways on foot. If that all sounds obvious, consider this: Modern law enforcement is largely set up to respond to anything. This is why criminologists are starting to leverage advanced mapping technologies to understand what criminals do in specific environments and how those environments can be altered to keep the peace.

Leslie Kennedy and Joel Caplan, academics within Rutgers School of Criminal Justice, combine the strengths of these disciplines with their work concerning Graphic Information Systems designed for mapping and spatial analysis. With help from criminal justice professor Eric Piza they invented Risk Terrain Modeling, a method of determine the spatial dynamics of crime and figuring out how to alter environments in order to prevent wrongdoing instead of reacting to it. While other crime maps point out the hot spots, Kennedy and Caplan’s model prevents them. They are the blister-preventing Band-Aids of violence.

The Risk Terrain Modeling Diagnostics Utility is free to download and use, which is precisely what crime analysts around the world are doing. And it’s working: When police in Glendale, Arizona started using the model, the city experienced a 42 percent reduction in robberies over three months. That’s wild success with no shots fired — clearly a step forward for American law enforcement.

Inverse spoke with Kennedy and Caplan about their work, assessing risk, and how to data can turn urban planning into a form of soft power.

How did you both come to focus on GIS mapping and spatial analysis as a subfield of criminal justice?

Joel Caplan: I started using GIS because I like understanding the spatial aspects of the qualitative, theoretical, and quantitative side of criminology. Looking at the spacial perspective makes sense because everybody operates somewhere in that space.

For decades, researchers in criminology have been limited to data that is often collected at the aggregate level — census data, racial demographics data, household income data — which is collected for certain purposes and applied to matters of research and evaluation in criminal justice. That’s been the data that exists.

Leslie Kennedy: I’ve been working at this for a long time. My original interest was in urban studies and a lot of research had been done for years in criminology focused on the geography of crime. But it didn’t really talk very much about what happened at the micro-level and how features of the environment could influence individual behavior.

What the GIS allows us to do is actually take locations on the map and put characteristics, or meta-data, to those locations. Then we can start trying to figure out the relationship between, say, the placement of bars, or the placement of schools and the sort of outcome measures like delinquency, or crime, or violence. We can explore testing theoretical concepts and go beyond just noticing the concentration of crime in certain locations. We add all this extra information that sits on that location, try to tap into that, and understand spatial influences on these outcomes.

That sounds incredibly complicated. How do you think of potential correlations and relationships?

LK: It’s important to visualize those relationships on a map because you need to get the people who are influencing outcomes — agencies or police or whatever — to understand that if they go to those locations and deal with the issues that are impacting crime there, that might matter. And then you can create another map and show the impact of their interventions on those risk factors and see if there was a change in crime behavior.

Can you walk me through what a standard inquiry might look like?

LK: You may have some theoretical idea — like that bars lead to violent crime at the location of bars. What does the location of bars have to do with the outcome of crime? There are all these probable factors. We can analyze them because the necessary data sets are readily available.

Is it fair to say that this is a both a pre-emptive tool to let police and security know where to go, and also to keep it from happening in the future?

LK: The idea is that right now they can go and determine whether or not they’ve had success, see if the crime has continued or they’ve gotten a certain number of arrests. So, if they arrest a lot of people, the crime rate goes up, right?


LK: Ultimately, it’s supposed to go down — the arrests are supposed to lead to a reduction, but they’ll keep going back to the same places over and over again. So they’ll say, ‘Yeah, but that’s where the crime occurs.’ And we say, ‘Yes, that’s where the crime occurred, because you haven’t gotten rid of the underlying reason why it’s happening there.’ If you go there, you need to try to address the crime based on the things that are happening in that location. Rather than going to a park to arrest robbers, you go to the subway stations where the robberies are actually taking place. Prediction isn’t necessarily about where the next crime is going to take place. We predict where the highest risk of the crime is and the likely underlying reasons.

The idea is that you can go back afterward and evaluate whether or not whatever you did — like putting up CCTV or putting more police at subway entrances —worked instead of just chasing the hot spot over and over again.

JC: The example we give is the playground example. If I tell you that a bunch of kids play at the same place over and over again, it tells you something about where these kids are playing. But if you take your focus off the kids, and look at the environment, you might observe that there are features of the landscape that define the spot as a playground: swings, slides, see-saws. We would expect kids to be attracted to this location, as opposed to other places that don’t have such entertaining qualities. Because of the diagnostic process of understanding what attracted these kids over and over again, we can anticipate the types of behaviors that we would expect at places with similar conditions. So if we don’t want kids to play anymore, we can take the slide away and make it boring.

That’s a benign example of what we do when trying to understand why certain locations are so suitable for crimes to continue. Diagnosing the underlying conditions that attract crime over and over again, makes accurate forecasts — inherently it’s a diagnostic procedure that informs us not just where to go, but what to focus on when [law enforcement] gets there. If you don’t want people to go to a place that attracts them, do something about the environment that makes it less attractive.

A mock Risk Terrain Modeling map.


The work you do seems to be contingent on the advancement of technology. Do you find that this field is rapidly changing because of that? What do you see as the future of this sort of work?

LK: That’s an interesting question. I think that there is a certainly an awareness among police agencies that they need evidence-based practice and they need to advance through these types of techniques. The federal government has moved really aggressively into providing the means to help these agencies to build up their data collection systems and to improve their analytical skills. The answer is: I think this [is] very important in the professional area. In the academic area, it probably hasn’t taken off quite as fast because there is a fairly steep learning curve among the academic practitioners — to learn how to use the GIS and to integrate it into their analysis. And the field really isn’t that demanding of them to do that yet — it’s certainly accepting of it, but it can also accept other types of research. Which is probably not a bad thing, it keeps different perspectives involved.

But yes, it’s not taking off as fast in the academic field as it is in the professional field. This direction is so central to what the overall theoretical orientation in the field has been for 100 years. It’s basically allowing people to create tools to test old, long-standing issues about crime and place. However, many police agencies in this country still don’t collect data using micro-management systems and [are] not totally automated. There is a long way for them go to — but there seems to be a pretty strong commitment to getting them up to speed. That’s because the only way that you’re going to be able to keep track of what is going on is to have the data. This is not only to track crime, but to track the activity of police in these communities.

JC: I would add to that — I’m not completely convinced that technology is going to reshape policing. I actually think that the future, or the next wave of innovation in policing, is going to come from the mentality of police and policing culture. I think that the technology has forced them to rethink how they utilize evidence, how to utilize the results of evidence-based research and technology in a way that is effective and transparent to the populations that they serve.

A lot of what Les and I have been doing now has been to work with police agencies to develop responses to the problems they’re assessing through spatial analysis. If all you make is a pretty map, then you’re not doing anything of much use to people on the ground. We need to help them figure out what to do with that information. Thats really where I think the next wave of innovation is going to come in — the mentality.

This interview has been edited and condensed for clarity.