Rethinking approaches to tackling violent crime

In many countries, the approach to tackling violent crime has been dominated by two ideas for generations. On the one hand, use bigger sticks like longer prison sentences to deter bad actors. On the other, use bigger carrots like more economic opportunity to make crime a less attractive choice. Yet both of these pillars of policy, enforcement and opportunity are pressed on a fundamental assumption that people are making deliberate decisions before they act. What if for the vast majority of violent incidents that assumption is wrong? What if the story of violence is less about good versus evil or poverty versus wealth and more about the way the human mind reacts to pressure? My name is Michael Hullsworth, I'm Bitt's Chief Bavill Scientist. And for this episode of The Bitt Podcast, I'm coming to you from the Safe Streets Conference in London. In this episode, we'll explore a different way of thinking about violence. One rooted in behavioral science. It's an approach which suggests that much of the violence seen in communities across the world stems not from pre-metastated plans, but from split-second decisions taken in high-stakes situations. If this is the case, then it opens up a whole new set of solutions, the focus not on carrots or sticks, but on influencing thought processes. This begins by hearing from Yens Ludwig, a professor at the University of Chicago and author of the book, Unforgiving Places. Yens highlights just how stubborn the problem of violence has been. Why that demands we question our oldest assumptions if we're to change the status quo. One of the reasons that I'm so excited to be in London right now is, as you know, the national government has set very ambitious goals for public safety here in the United Kingdom to cut youth-knife violence in half and to cut violence against women and girls in half. That also provides a big opportunity. It's a big, ambitious goal, but it also provides a big opportunity to step back and rethink what we're doing on the violence problem in the UK. One of the reasons that there's so much value in stepping back to rethink this is, in the UK, as we see in the US as well, there just has not been very much progress since about 1900 in making major changes in the violence problem. I think the persistence of the problem for 125 years and the new governments desire to really change that, I think, provides an opportunity to step back and think about what the problem is, whether we've been thinking about it the correct way and I think behavioral science gives us some ways to reimagine the causes of the problem and maybe rethink some of our solutions for the problem as well. So one of the key arguments that I make in the book, I think maybe one of the most important arguments that I make in the book, is that interpersonal violence is not what we think. And I argue that's true for gun violence in the United States and I argue that there are many, many parallels to the problem of night violence and for that matter, violence against women and girls in the UK context. So I think if you look at opinion surveys of the general public and you ask them questions like, why do you think people engage in violence against others? One big segment of people will say the answer to why people engage in violence is they are morally bad people, we're just not afraid of what the criminal justice system is going to do to them. And that perspective on the problem leads you to think that the only thing that you can do is disincentivize violence with the threat of ever bigger sticks, or bigger prison penalties. To the extent to which there are people who don't fall into that camp, most of the remainder of the public fall into a second conventional wisdom camp that says that violence is due to economic desperation, that these are people who will do whatever it takes to feed themselves, feed their families. And the only solution of the violence problem is to disincentivize it by offering people bigger carrots to make the alternatives to crime more lucrative so they choose not to engage in crime. And I think the key insight with the shared assumption with both of those conventional, one of those conventional wisdoms tends to be more common on the right, one of those conventional wisdoms tends to be more common on the left, some people believe both, some people believe another, but most people hold one of those two. And I think what those two conventional wisdoms have in common, often just implicitly, is the view that before anybody engages in violence, that there's a certain amount of deliberate, rational calculation of pros and cons before they act. So if you're on the right, you think the solutions are bigger sticks. If you're on the left, you think the solution are bigger carrots, but you agree that incentives are going to matter because people are really kind of thinking deliberately about what they're doing before they act. And I think that turns out to be true, but for only for a surprisingly modest chair of violent events in the US and in the UK. So in the United States, as far as we can tell, something like 80% of murders in America stem from arguments. And in the UK is best we can tell it's maybe closer to something like 70%. But it still suggests the overwhelming majority of murders are not premeditated, are not driven by deliberate benefit cost calculation. You can see that if you just reflect on the last time that you were in a very heated argument with someone else. This is the last thing in the world that you were thinking about in that moment is what the consequences might be. You are just literally seeing red and acting in a way that is just not deliberate and not at your most deterrible self. And I think that creates the big window for behavioral science to give us a better understanding of what's going on and what to do about it. Chicago crime lab, you've watched on a few different programs. Can you just sort of summarize some of these programs, many of which become quite famous now? Yeah, I think that the key idea behind most of these programs, we've worked on studies of a bunch of different programs developed by incredibly clever and committed nonprofit organizations in the city of Chicago, including a program called Becoming a Man that was developed by youth guidance and many others now, that I think all have a shared flavor of essentially trying to hold up a mirror for people and help them see how much of their behavior is really driven by automatic system one responses and helping them become, helping their more deliberate, slow thinking system to selves, recognize that they should be a little bit more skeptical of these automatic system one responses, especially in potentially high-stake situations. I think at a very high level that is what these programs are all about. And we've now studied a bunch of these in different randomized control trials of the sort that provide gold standard evidence and medicine and seen remarkable impacts on violence across the board. And I think one of the big opportunities I think that we have started to see from our work in Chicago is one of the studies that we did was in collaboration with the juvenile temporary detention center in Chicago where the teenagers at highest risk of violence involvement were detained awaiting resolution of their case. What the administrator of the juvenile detention center decided to do was he said, "I don't have any money akin to what the US and the UK government, the budget constraints they're facing." The administrator says, "I don't have any money to introduce any new gold standard programming into the facility." What I do have is a bunch of kids who would benefit from some more help. I've got existing staff, I've got a facility, and I've got a lot of low opportunity cost time. That is low value time in the facility, they're just watching TV. And so what the administrator decided to do is just train the detention guards to deliver a ban like program that helps kids sort of through this behavioral science lens recognize this sort of system one automatic responses that they're prone to, that we're all prone to, but that can lead to particularly severe consequences for them. And we evaluated that with our randomized control trial and fined a 20% reduction recidivism for the super-hybrus kids. Now you might look at that and say 20%, I'm not sure how impressed to be. Let me just point out that this is a program where the marginal cost of delivering this program is basically zero, right? Because you've got the guards, you've got the kids, you've got the building, there's the key cost drivers right there. The marginal cost of implementing this program is like a week of training for the staff and the cost of mimiographing the program booklets. So as close to free as anything in public policy ever gets, I'm not a math professor at the University of Chicago, but I think that if I've got a 20% reduction in recidivism and I assign the value to society of that and I divide it by a cost of zero, I've got a very, very favorable benefit cost ratio. But what does it actually feel like to deliver these programs or participate in them? To find out, I spoke to Dr. Chico Tillman, Executive Director of the Community Violence Intervention Academy at the University of Chicago. He started by telling me about the Ready Program. So what Ready does is two things. Ready, outreaches to those individuals through relentless engagement. Help these individuals go through a program which is centered around two things. One is CBI, which is similar to what you just talked about in this book, which would be system two thinking, cognitive behavior, their real intervention, and secondly, workforce development. So the reason why I wanted to bring up the system to thinking of the cognitive behavior intervention, because a lot of people talk about the sociogenic properties of giving somebody a job. But if you give them a job without preparing them to enter into the workforce, then you're bringing actually introducing violence or violent behavior into the workforce. So what we wanted to do was help individuals be able to self-regulate their behavior before we give them the opportunity so that individuals wouldn't get an opportunity and blow an opportunity of a lifetime. So we prepared them to go into the workforce and teach them not only soft skills, but also teach them alternatives to resolve and conflict outside of aggression. All individuals went through the program and were so blown away the high change. Just like once he finished the program, he got hired on as an outreach worker because he said I got to recruit other people to process them and get them what I got, because not only did it change his life in a way that it gave him an opportunity for employment, but it gave him a way to think about life different and not to make permanent decisions in temporary situations. Because what it does is a great example is in a ready program what we would do is deconstruct something that turned out fatal. For example, say an individual post this something somebody responded on a post, they got into a conflict, it ended up in a homicide. We would deconstruct that scenario, a real life scenario. And we would say he posted that what happens if you block it and usually people would look in amazement and look at you strange and then they would say it's over. I say how many people you know we would go through the whole thing, how many people went to prison then, how many people were arrested, none, how many people died, none. And so a life up comes on you mean I could have just blocked him as opposed. So now not only are we, it's a difference between giving directives and educate, now we gave them a tool, if you in a situation, I could just block you as opposed to allowing it to escalate because now I know where it can go to if everybody in the community who are entrenched and maladaptive behavior getting far off. Dr. Tillman also highlights the fact that the messenger matters almost as much as the content of the program. The person delivering message needs to be credible to the recipient. Think about high school. If somebody from the high school that you grew up with, two people was in to it and you knew both of them, you could talk to both of them. That's what a credible messenger is, it's somebody that's born in. But also especially if it's somebody who has been system impacted, unafraid and courageous because the reason why I say you have to be courageous because you don't have a blunt and in cases you talk to somebody who might be armed. So there's a level of respect when a person think you understand what they're coming from. That's why it's important that a person is credible meeting there from that community. Bayeral Science is clearly a powerful tool which can help those at risk of committing violence. What about those on the streets of our towns and cities who have to make split second decisions to de-escalate situations that could turn violent? For police officers who often operate in high stakes environments where their actions can be the difference between life and death, it's a difficult challenge to resolve. This issue has been studied extensively by Andrilla Dubay, professor of global conflict studies at the University of Chicago's Harris School of Public Policy. She told me about the situational decision making program or CITD that she ran with the Chicago Police Department. I would say a lot of the inspiration for this program actually came from work that we saw being done on youth violence and programs like the becoming a man program which aimed to curb youth violence by reducing automatic thinking. When we saw the results from studies like the becoming a man intervention that showed that by curbing automatic thinking among youth, it could lead to a reduction in violence among youth, we started thinking maybe police officers are actually also engaging in this form of automatic thinking and if that's true, then helping them with interventions that train them out of that automatic thinking might be helpful for reducing violence in their interactions with civilians. I would say policing regularly places officers in situations that are ambiguous where it's difficult to figure out what's going on and that leads to this tendency for officers to resort to making assumptions to make decisions. So consider a situation in which a police officer sees a man in a dark alley and he's shouting and stumbling and suddenly the officer catches a glimpse of something in the man's hand. It might be natural for the officer to go with the default interpretation that that glint is a knife but of course the officer needs to consider the possibility that maybe that glint is actually just a beer bottle and the man is shouting because he's drunk or maybe that's just a cell phone he's using to call for help. And so what does the program do to try and change those automatic reactions? The crux of the program is really training officers to make the most of whatever little time they have in these types of ambiguous high-stress situations to enable them to rapidly consider alternative interpretations of the situation so they don't just go with the default assumption. And the training tries to achieve that in two ways. A big component of it is actually teaching officers about cognitive biases or what we call thinking traps that can emerge in these situations. So here we really draw on behavioral science to lay out these types of behavioral biases like over generalizing in the situation on the basis of past situations or catastrophizing which is assuming the worst most extreme outcome like assuming that glint is actually a knife. Once we teach the officers about these cognitive biases then we use a scenario-based training to actually keep them from falling into these thinking traps and to enable them to be able to rapidly develop these alternative explanations for what could be going on. That's really interesting that you do training because one of the concerns about, you know, debiasing is this notion of the bias blind spot that you can learn about a bias and actually sometimes it can make things worse because you don't think you do it or you don't connect it with your behavior but it seems that you're putting people in the situation, getting them to experience it rather than just theoretically know that there's a bias. Absolutely. The integration of that sort of conceptual learning which does take place in the classroom with the scenario-based training is really integral to the success of the program. And the way that we do the scenario-based training to make it engaging is we actually use this technology called force-option simulators which are like giant x-boxes in which the officers are interacting with characters, projected onto a screen and they carry with them all the weapons that they would usually have on the street. So in this way, firstly they begin to see how the assumptions they're making about a situation map to their interpretation and how that interpretation guides what force-options they use. So after an officer goes to a scenario there's a really extensive debrief where they're asked, "Okay, so what did you think was going on? What else could have been going on?" And in this way they start to see what assumptions they might have made and what automatic thinking traps they may have fallen into in the course of the situation. So that's the way in which the concepts and the behavioral biases that they learn about come to life and I think that's part of what helps them avoid these types of thinking traps. So how did you make this force option simulator immersive? Because a skeptic would say people know it's a game and they are it's not the same as the real world but it seems that well maybe we'll come onto this but seems that you're getting results in the real world. Absolutely and before entering into one of these I might have shared the very skepticism of how this can be immersive but when I observed how this actually plays out I could see that the situation feels very realistic. I went through a simulated scenario myself and I had a tendency to freeze because I found the situation so frightening. So I think it's actually quite successful in creating those realities out on the street of experiencing stress of feeling like there's a lot of information to take in and feeling like there's different ways in which the situation could evolve. So in that way especially with the weapons that you actually carry on yourself when you're going through the scenario I think it does quite a good job of sort of emulating the reality out there on the streets. I think the possibilities for the future are absolutely endless because for example VR headsets are coming online and many companies are actually innovating on this front as well and that opens up a whole new world of possibilities because it's very easy to create your own scenarios in the VR headsets. So that means whatever the situations are that a particular department needs to deal with whatever the you know greatest problems in a city might be. If it's drug-related crimes you can have scenarios that speak to that. If it's domestic violence you can create scenarios that match that. You can really tailor the scenarios to essentially have officers practice these ideas for scenarios that are relevant to the needs of their jurisdiction. So how many people have been through the program now? You launched it in Chicago originally? How have things gone? Yeah so we actually just completed the study. So in Chicago there were you know we did a randomized control trial with 2,000 officers or so and so a thousand of them received the training and we have the impact evaluation results from that group. Now we are in the process of expanding CITD to departments around the country. I would say the headline finding are the impacts that we observe out in the field. We see that the training led to a 23% reduction in uses of non-lethal force. We can't identify the effect on lethal force because there's very few shootings in the sample. But everything else from using the taser to kicking, punching, canine, baton, we see a 23% reduction in that outcome. We also observe a 23% reduction in discretionary arrests. These are arrests for low-level charges like disobeying an officer. These are sort of disorderly conduct. These are charges that tend to come up when officers are irritated with what the subject is doing but they may not be the best way of resolving the situation. So we think these two outcomes are very important because they suggest that the training is really enabling officers to solve problems in other ways that avoid uses of force and avoid arrests that are not necessary. I would add to that there's one other finding that's very important which is that the training led to a 49% reduction in days off taken for officer injury. That is to say these improvements in community safety went hand in hand with improvements in officer safety. In other words, by finding other ways of resolving these situations officers were also keeping themselves safer. We actually observed a large closing of the race gap in arrest rates. That is to say we observed a fall in the overall arrests of black subjects without any change in the arrests of subjects of any other races and we observed a much larger fall in the discretionary arrests of black subjects than the subjects of any other races. This really struck us because this was not a training about racial bias. It was a training about these general cognitive biases but the reason I think the training ended up having this effect is because to the extent that officers are more prone to making assumptions or exhibiting these cognitive biases in situations when subjects are black by training officers out of these cognitive biases you're going to see a larger effect on black subjects. A key thread here is evaluation. A program may be having an unexpected or unintended effect on behavior but we won't know if we aren't measuring the impacts. Jennifer Doliak is exactly vice-president at Arnold Ventures. A philanthropic organization aimed at tackling some of the biggest issues in America. Her work is used new methods to uncover the surprising effects of policies intended to reduce crime. My co-author Jill Carr and I did a study of juvenile curfews in Washington DC and what we were interested in was both the effect of a curfew but also the challenge of studying the impact of a program that can also affect the measurement of the outcome. In this case you have people under a certain age are required to be home by a certain time in the evening and then the question is how that affects public safety, how that affects crime out on the street. The traditional way of measuring that would be to say okay we'll let's look at reported crime rates but both the incidence of crime and the likelihood that it's being reported could be affected by the curfew. If everybody is suddenly inside and no one's out on the street anymore then crime could still be happening but no one's reporting it to police right or police are spending time in different places now enforcing the curfew instead of instead of you know looking out for burglars or something then that burglary is no longer reported and so we were using data from shot spotter which is an acoustic gun shot sensor as a more objective measure of a type of crime that matters a lot so gun gunfire and so basically we're saying okay well here's here's a different way to measure what the impact of this curfew is that should not be affected by the policy itself so whether or not the curfew is an effect or not the gun sensors are still going to hear gunfire and so what we found was that after the curfew went into effect gunfire actually increased in in these areas of DC rather than decrease which was the goal of the policy was to make the community safer so what seems to be happening here the intuition that we have in mind is that you've got police that are enforcing the curfew instead of doing whatever they were doing instead which presumably was valuable and you have you know people who would have been eyes on the street in Jane Jacobs terms who could be you know potential bystanders or potential witnesses but who generally have a deterrent effect on crime because you're increasing the probability that that crime is seen and reported and you have all those people who are now inside law binding so the sense of gun inside right and so that winds have actually increasing gunfire instead of reducing it and the difference here was it looked like there was there were fewer gunshots but that was because people were not reporting them because of the curfew that's what we argue in the paper right so if you used the data from from nine-one calls a reported crime it would look like the curfew was effective it would look like reported crime went down but we looked at the data from the gunshot sensors it actually went up and so so we argue that that's the effect of of the reporting effect that that is the difference there is and so using this more objective data from something like a sensor can give us a better a better sense of what's happening to add the actual incidence of crime if we have enough good quality studies and we can start to step back and put together a menu of the most reliable choices to reduce crime that's something that Jennifer Dolly has been working on herself she recently published research about why crime matters and what to do about it where she reviewed many different studies I asked which intervention should be at the top of the list if for example a mayor wants to reduce youth violence in the next 12 months one big takeaway from a really broad literature is that increasing the probability of getting caught matters a lot so you know we think about when we think about deterring crime for a long time especially in the [BLANK_AUDIO] the United States, we've thought about making prison sentences longer and longer. If you have a really harsh punishment, that should deter people from committing crime now, right? The problem is that most people who are on the margin of committing crime just aren't that forward looking. And so adding years to an already long sentence doesn't affect the decision that they're making today. But it turns out that changing the immediate consequences, the swiftness and certainty of punishment. So the probability of getting caught is the thing that matters a lot. And we do have really nice evidence that that deter's crime quite a bit. So what that means in practice is more and better policing, improving police investigations, increasing the likelihood that crimes are solved through things like DNA databases or cameras. So that is one bucket of interventions that we know is effective. Another bucket is summer employment programs, summer jobs for teens. So a number of cities have run these programs for many years in a way that they're over subscribes. They're extremely popular, which means they're a whole lot more kids that want summer jobs than can have them. And so these cities typically use fair lotteries to just to determine who gets a job and who doesn't. And that's fantastic for research because that basically means they've run a randomized trial. And we've seen across several cities now that summer jobs getting a job offer and participating in the program pretty dramatically reduces violent arrests, even mortality due to things like homicide. And so based on that evidence, my team at Arnold Ventures is trying to scale those programs across the US. And then another set of programs that is really effective is cognitive behavioral therapy. So there have been a number of different programs that include CBT that, again, pretty dramatically reduced, violent arrests or violent crime. I think the whole game then is how to get people to engage in this kind of program when they are not already incarcerated or sitting in a classroom. And so there have been some innovative programs recently that have paired CBT with a job or paired CBT with a mentoring program. And so the job or the mentoring program kind of acts as a carrot to get youth to engage in the program. And then while they're there, they get the CBT that is super effective. And so there, the question is, okay, do these programs replicate in other places? And then if so, how do we scale them? That question, do these programs replicate in other places? Is one that's been taken up by bits ending youth violence lab? As Jennifer just explained, one of the most well-evident programs in the US is summer jobs for young people. Tom McBride was part of the team from the lab who worked to bring this program across from the US, carefully adapting it for a UK context. So in the US, what they call some of youth employment programs are really common. Nearly every major city has a version of it, which gives paid employment to young people during the school summer holidays. A lot of them started with the view of giving young people economic opportunity, helping them to do something productive over the summer months rather than as a public health measure or a youth violence measure. But in recent years, they've been evaluated in Chicago, Philadelphia, Boston and New York, and shown consistently that these approaches can reduce involvement in crime and violence, both well-sewn people are participating in the program and for up to a year beyond. So we wanted to really build on that and see if that could work here. So last year, the ending youth violence lab partnered with UK Youth, Youth Work Charity and through funding from Department of Culture Media and Sports in the Youth Endowment Fund to set up the first of these in the UK. That delivered to 450 young people in London, Manchester and the West Midlands. We successfully in terms of take up, in terms of number of young people completing that program, the feedback from young people themselves, employers and youth workers who participated in that program was fantastic. And on the back of that, that's now progressed into a two-year full-round-and-mores control trial and any day now around 700 young people will be going into paid employment across England and Wales with intense mentoring support from qualified youth workers to really help and build the skills that we think will help keep them safe from serious violence. We've really had to think quite hard about what their training needs are up front, what pastoral support they might need and practical support with transport, with clothing and so on. You have to think about real local challenges like, are young people safe to enter certain post-codes and so on within their city. And then we had to think about how do you keep a young person in employment and sticking with a job and making sure that they are. I feel they have the support to deal with difficult and challenging situations that arise in the world of work. So we worked really closely with UK youth to think about the pre-employment training that these young people had and the ongoing support they had during their five weeks of employment and that's been really, really successful. Over 95% of the young people who enrolled in that program completed it, which for those of you who know the world of evaluation is a surprisingly high number and one that we're really proud of and I think it's a testament to the level of support we put in place. As we've heard, this behavioural science approach is helping to turn the tide on violence. What are the next frontiers for this work? Fuchiko Tillman, the future is about fundamentally changing the public safety system itself, making community violence intervention or CVI an equal partner to law enforcement. My north star is to make CVI a permanent part of the community safety or public safety ecosystem, where when you think about the work of public safety or community safety, you don't just think about law enforcement but you think about CVI because CVI is a component that does things that's complimentary to law enforcement. Law enforcement does things that CVI doesn't do and CVI does things that law enforcement can't and won't do. For example, one of the things we know is law enforcement is responsible in nature. Just say if you feel it, you wake up today, right? Say if you get in the argument right now, right? You know I got a background and you feel an account of shaky and everybody knows that I'm going to do something to you. I haven't said anything and you call the police and you call it and you say, they say, what's going on? You say, what me and Chiko got in the argument and I just feel nervous. I need the police to come out. What do you think they're going to say? So you and everybody in the community might know like you disrespect the Chiko, it's going to be problems. It's just a matter of time. Law enforcement won't get involved because their response of the nature, they aren't able to really do nothing until a crime has been committed or a threat. It has to be something done that allowed them to utilize or leverage their power. Where CVI is different, you could call some CVI guys and say, hey, here's the situation. Me and Chiko had an argument and they go, no, all Chiko is, oh my God. He has a preclivenly to commit violent acts. I might have said, you know, it got heated. Did he threaten you now? But okay, well, let's get in touch with him and sit y'all above there and resolve the issue before it escalates to somebody getting shot. That's the difference. We work with things on the front end to resolve it so that it doesn't result in a shooting. Whereas with law enforcement, you can call them until after you got shot. Consulate Week also thinks there's an opportunity to move into areas where interventions can become more proactive and preventative. The other major touch point that we're starting to think about, whether we could use that as a platform for delivering these sorts of interventions, is the public school system. And you know, in the US context, and I think what we've heard yesterday at this conference that we're having here at the University of Chicago, Woodson or in London, is in the UK as well, that you also have health class requirements for students that aren't always teaching people the most important things. And you know, that's another place to deliver these sorts of programs. And I think one of the reasons that that's so important is especially for, you know, for street violence, for knife violence, what you see is most of the people have some prior interaction with the criminal justice system. For violence against women and girls. There's so little reporting for totally understandable reasons, but so much of violence against women and girls goes unreported that we often don't know who the offenders are until a woman winds up in an emergency room setting. And so I think that makes it particularly important to do what public health people call primary prevention. You might call it like universal prevention. And the public schools provide a particularly important opportunity for doing that. As with so many of our episodes, these behavioral science approaches have sat alongside a clear right commitment to evaluation. And perhaps the growing role of what works in criminal justice policy has the biggest implication for the future. As Jennifer Dullerke explains, "I know that a lot of people who work in the criminal justice space and maybe in policy more broad. are really pessimistic. They see a really big complicated system that is just full of problems. I am extremely hopeful and optimistic every day. I mean, I am excited about all of the new research that's coming out constantly giving us new solutions that we can run with. Even when a study comes out and says something doesn't work, that's useful information, right? We can stop going down that road and try something else. And so I really love that this policy area has become one where we're all more open to trying stuff, but aiming to fail fast rather than assuming we're not going to fail at all. Because that's the way we're going to be able to just iterate and try something else and eventually find solutions. If you'd like to know more about the work we're doing here at the BehaviveLinsense team, then make sure you visit its website. We've added the link to the description of this podcast. Thanks for listening.