After the Webinar: Crime Prevention with Data Informed Community Engagement. Q&A with Dr. Joel Caplan

Webinar presenter Dr. Joel Caplan answered a number of your questions after his presentation, Crime Prevention with Data-Informed Community Engagement. Here are just a few of his responses.


Audience Question: So, what are some typical agency partners that you’ve seen law enforcement agencies work with? 

Joel Caplan: When it’s spearheaded by a police department, oftentimes they’ll partner with other police departments that they’re readily familiar with. So, other agencies within the jurisdiction. But in addition to that, other city agencies become partners, such as in the example of risk-based policing in Kansas City. For the KCPD, it was easier to establish buy-in with other city agencies, being a city agency itself. Those agencies were much more open to being involved. I can say that one of the ways in which they establish those relationships early on was to invite representatives from those city agencies to their introduction sessions and training sessions that kicked off the initiative. So, by being invited to the commanders’ meetings and to the officer meetings that introduced this new initiative, they got some insight into what was happening and were involved early on. And then, representatives from the police department were invited to those city agency’s meetings. I think in Kansas City, they call it their Regulators Meetings, or something like that. So, we tend to see it initially between city agencies and then slowly begin to include other community partners, both non-profit and business partners.



Audience Question: So then are there any departments or organizations or agencies that may be out of the typical norm that maybe agencies should consider then reaching out to? That maybe they might not immediately think of.

Joel Caplan: Oh, I’ll have to give that a little bit more thought. The best way for me to answer that would actually be to go to the Newark Public Safety Collaborative website. The NPSC is a really great example of DICE in action. They’ve implemented it. They’ve sustained it for the past two years, and they have a lot of case studies and notable successes. One of the web pages includes all of the agency partners, I believe there are over 40. I mean, it might give listeners on this webinar, some ideas of who to invite in their own jurisdiction. It covers everything and ranges from block groups and neighborhood associations, to mercantile associations, all the way up to large corporations headquartered in Newark, plus other small businesses, individual change agents, and city agencies.

Host: Fantastic. Thank you. That’s what I was wondering as if it had to be just another government agency. Or if folks could get creative and reach out into the neighborhoods themselves, the neighborhood associations. It sounds like. That’s what you’re saying is they truly can be as creative as they want to be.

Joel Caplan: They can, and what they’ve learned is that some agencies will have townhall meetings, where people will come and maybe complain. But when community leaders are armed with data and analysis, it allows for the conversation to get past finger-pointing and to kind of a validation of the public’s concerns because when somebody says, “I see the problem with this particular business”, or “this location is problematic”, city agency representatives can then say, “you know what, you’re right. In fact, those businesses increased risk by five times, and that area is an area that we’re focusing on, because it’s one of our highest risk locations.” And so, community members can be involved in the conversation, and add value to it, because they have insight into potential risk narratives and can interpret some of the results in ways that we might not have been able to previously. Even if they’re not involved in the intervention strategy, involving the community, not as a kind of open townhall-style complaint session, but as a community meeting with invitations to people who have insights and expertise to share, can really add some great insights to the analysis and the conversation.



Audience Question: What are some of the associated risk factors to look for when applying RTM to overdose analysis?

Joel Caplan: That is a good question. I don’t have the link on this slide. But, if you go to the Risk Terrain Modeling site there’s a lot more detailed information about that. So, with that said, I’ll answer the question by saying that risk factors can differ. Even within the same jurisdiction, we found that open-air drug markets for different types of drugs can have very different risk terrains. And so, the location where marijuana or heroin or cocaine are sold, can, in fact, be analyzed separately and different risk terrains could emerge. I don’t know any specific risk factors off the top of my head because I haven’t looked at that research recently, or at least haven’t refreshed my memory before this call, so I don’t want to make anything up. But the link that I directed you to has links to other published research, as well as other case studies that provide more details.



Audience Question: What cities have done this analysis process? 

Joel Caplan: The cities that I know of are the ones that have made it publicly known. And so, some of them are included as examples or case studies on some of the websites I already mentioned. Those that have written about their work and that are readily available publicly, for me to share, include places like Newark or Irvington, NJ, where risk terrain modeling began. It’s been used and tested in Atlantic City and Kansas City (MO) and Fayetteville, North Carolina. In fact, one of the commanders over there published an essay in the NIJ Journal and is a member of the American Society of Evidenced-Based Policing, too. We have a great case study from Burlington, North Carolina on the website, from an analyst there who produced the report and let us share it publicly with her permission. These are just some of the agencies that I’m aware of, mainly because I’ve worked either directly with them, or they’ve reached out to me and let me know what they were doing. Then, there’s also a National Institute of Justice study that we conducted in 2015 that tested RTM in 6 different cities. Those cities were Newark, Chicago, Arlington TX), Kansas City (MO), Colorado Springs (CO), and Glendale (AZ). So, those types of examples I remember off the top of my head because of my involvement. See also the handout where Dallas (TX), Jersey City (NJ) and a few others are listed.



Audience Question: How is RTM different than predictive policing and the model developed by Mohler. 

Joel Caplan: I believe Mohler and his colleagues are connected with the PredPol software. I make it a habit to not comment on proprietary commercial products. So, I don’t really have anything to say specifically about that. But what I do know is that risk terrain modeling is significantly different from hotspot mapping and predictive policing. For one, it’s not predictive policing. It’s a diagnostic tool that helps us understand problems. It doesn’t, it’s not designed to simply deploy resources. It provides insights and information into a problem that we want to analyze, allows the human element to be involved, to talk about the analytic outputs to come up with the risk narratives and strategies for response. And it also allows for transparency in those actions because the places we’re focusing on are clearly explained by the results of the analysis, which is also shared, as part of the output. There are some other ways that we can discuss how it differs. People also have different definitions, themselves, of what predictive policing actually is. So, I guess the quick answer is those few things that I mentioned, but I’d be happy to discuss further details after the webinar.



Audience Question: So, data governance is an issue that always seems to get in the way of innovative data analysis. How do you manage to get around sharing data that is sufficiently detailed in order to use that in RTM between agencies? Do you have any advice?

Joel Caplan: Yeah, you know, it’s an interesting question because 10 years ago, well, I guess RTM started around 10 to 12 years ago; You know, around that time, we might have wondered where all of this data can come from. There was a time where all of this data wasn’t even digitized. Now, these datasets are increasingly readily available. There are dozens of open public data portals for large cities and small towns across the country that are making the information, not just for crime data, but also environmental feature data, 3-1-1 calls-for-service data, and more, readily available for download. So, in terms of data governance, the data that we use is inherently public-record data already — crime incident locations, and business retail infrastructure, and physical features of the environment. Now, some agencies add additional data to that, that might be proprietary or that might be considered more confidential such as, some police data or data sets for the City Ordinance Inspector who can add data about places that have a high volume of ordinance violations. But for the most part, the way that we get people to share the data is to talk to them about how there’s mutual benefit in doing so. There’s a very big difference between data and analysis. And while data can be made available easily or to anyone, the analysis that RTM offers, and the process that DICE allows, kind of democratizes the data and analysis and lets people who wouldn’t otherwise have access to it; that is, access to not only the data but the analysis and the software and the tools that were once only available to police agencies (because they have the budgets to pay for them or the analysts to use them). And so, when people realize the value of the data, they’re more willing to share it, and they’re also more receptive to listening to the conversations that are born from the analysis.



Audience Question: Is this a tool that anyone has access to? 

Joel Caplan: Risk Terrain Modeling is a process that was developed at Rutgers. And there are online resources for using it and thinking about how to apply it. The software was also originally developed at Rutgers and funded, in part, by the National Institute of Justice. There’s a number of ways in which that software is made accessible to individuals, agencies, non-profit organizations, and other groups. Everything from free on a project-specific basis for a limited period of time, to enterprise subscriptions, as well as, access to colleges and universities for in-class instruction. There’s information about all these different options, including student access, at, which is also on the handout.



Audience Question: Did the group choose higher risk areas in order to put something there to make them less risky or did they choose less risky houses on the outset? 

Joel Caplan: As inputs for the analysis, or as priority deployment areas? I think the way that I understand it is the RTM analysis was performed based upon the existing physical and structural environment of the jurisdiction. The analysis was run for the observed crime patterns. And the places, the environmental features, that connected with those crime patterns, the pattern of crime locations, were identified as significant, and then weighted to produce the final risk terrain map. The final map showed the highest risk locations, and then the vacant properties and abandoned buildings were identified from within those high-risk locations.

Host:  This was the group that purchased old houses to make low-income housing that used RTM to choose which empty buildings to purchase.

Joel Caplan: It’s a good question. They have limited resources. They’re a non-profit organization. They happen to be in Newark. They have to make decisions about which properties to purchase. So, what they normally do is they’ll partner with the city and they’ll buy properties for $1. It’s not the property that’s expensive. These are city-owned vacant buildings that they purchase. It’s the rehab and revitalization that costs money, so their resources need to be optimized. So, if they’re going to buy these places anyway, they’ll do their own site suitability analysis and maybe have a handful of properties that they’d be interested in given their larger strategic plan. And then they would use the RTM results to help select one of them or to rule out some of them. To allow them not only to purchase the property that they might have done otherwise but to know that they’re also going to have a positive impact on the city’s larger crime problem as a result of doing it. So, it’s kind of, you know, “killing two birds with one stone”, so to speak.



Audience Question: Can this program be integrated with a current system that a department may be using at this current time? 

Joel Caplan: Yes, there are a few ways to answer the question. I’ll answer each of them quickly. Data-informed community engagement is a process, and it’s a process that anybody can replicate. In fact, the Operations Safe Surroundings, or OpSS e-book, is a PDF download. This gives kind of a how-to playbook for the steps that I described on the second to last slide, and that link is on the handout. Risk terrain modeling is performed easily through the RTMDx software. The software is a service accessible through any web browser. It also has API functionality, so you can connect it to your existing dashboards, or you can connect it to your existing CAD or RMS systems to pull data in or push insights out. It can get as complicated or as simple as you’d like to get-up-and-running. The other piece to this is the kind of intervention strategies and planning and data sharing. And that could be done with simple off-the-shelf software products like Google Earth to share maps, or PDFs, or any other more sophisticated systems that exist within your agency. The last piece of this, is you don’t need to get rid of other analytical tools to use RTM. RTM can complement existing practices. Hotspot mapping is valuable, it tells us where crime clusters are.  RTM diagnoses provide new insights about what to do when you arrive and what to focus on and how to coordinate strategies. We’ve seen examples where RTM complements hotspot mapping quite well. Near repeat analysis, and other types of techniques, can all be used in concert with RTM. So, it’s not only possible to use as a complement to your technology, but it also can complement your existing analytical practices. From a technology perspective, if you go to, there’s an eBook that talks about integrating risk terrain modeling with existing technologies. And that might give you some additional insights.


Audience Question: How can RTM be implemented in places where all three of the inputs that you talked about can’t necessarily be gathered, for example, proper shapefiles, crime incidents database, etc. Are there any other alternatives or solutions? 

Joel Caplan: I guess it depends why they wouldn’t be accessible. So, I’m not exactly sure how to answer the question. If they are still on, maybe they could follow up a little bit. But, obviously, the one thing that we need are the incidents that we’re interested in analyzing. The problem, we need at least five different incidents to begin analyzing those patterns. So, whether it’s motor vehicle theft, or the beginning of a burglary spree or series that we want to analyze early in the investigation — to help with that investigation through kind of analyzing the geographic preferences of those offending spots. Or, maybe there’s a long persistent crime problem; that data needs to exist. The environmental features data is often the piece of information that’s harder to acquire. But once it’s obtained, maintaining it is much easier. You know, schools aren’t built or demolished daily. And so, while the terrain and the environmental features are very static, that is, they exist physically in space, their influence on offending is quite dynamic. And different times of the day could be influenced based upon how these facilities are used. Like, just imagine nightclubs at certain times of the night compared to the middle of the day. One of the things that we have done is partnered with private companies that collect this information for anywhere in the country. And it allows us to get up-and-running quickly to get a list of environmental features for a particular area, and then try to supplement that with additional pieces of information. In some cases, you might just want to test one particular thing. You want to know whether or not, you know, convenience stores or liquor stores connect with your crime pattern, and that may be the only data set you analyzed for that particular purpose. It’s not a comprehensive risk terrain modeling analysis, but it would test the spatial correlation between just those environmental features and the crime patterns. So, there’s a scale at which you could begin to analyze the data you currently have, and then get more complicated with your analysis as the data allows.



Click Here to Watch a Recording of Crime Prevention with Data-Informed Community Engagement.  



Additional Resources
4 years ago
Thoughts on Evidence Based Policing from Rachel Tolber
Such a great reminder  about the relevance and importance of data in relation to EBP that Rachel To […]
5 years ago
Beyond Body-Worn: Boulder County’s Digital Evidence Strategy
Body-worn cameras is the future of policing as more and more law enforcement agencies are proactivel […]
5 years ago
Evidence-Based Policing: A Practical Approach
Evidence-Based Policing (EBP) has been around for two decades but only has been getting a lot of tra […]
6 years ago
Evidence Based Best Practices for Public Safety Agencies
Public Safety Agencies are faced with a myriad of issues daily. As part of their duty to serve and p […]