This illustration is part of a series of Social Science Bites illustrations by scientific illustrator Alex Cagan. We’ve looked through our archives and chosen some of our favorite episodes from over the years, which Alex has brought to life in these visualizations. We’ll be unveiling new illustrations in this series through June and July 2020 on our Twitter page. Catch each new illustration as it’s released at the hashtag #SSBillustrated and click here to view all the illustrations so far.
Lawrence Sherman is a professor of criminology at Cambridge University and a keen advocate of experimental criminology. In this episode of the Social Science Bites podcast he outlines his approach and gives some examples of its successes. Social Science Bites is made in association with SAGE.
Click HERE to download a PDF transcript of this conversation. The full text also appears below.
To directly download this podcast, right click HERE and “Save Link As.”
Nigel Warburton: There are many theories about crime, its causes and treatment. So how do we decide which ones are effective? Take the case of restorative justice when criminals and their victims meet face to face. Some critics argue that this approach is too soft on perpetrators and doesn’t work. But is this true? Lawrence Sherman of Cambridge University believes that theories about crime can and should be put to the test. He’s a passionate advocate of experimental criminology.
David Edmonds: Lawrence Sherman, welcome to Social Science Bites.
Lawrence Sherman: Thank you.
David Edmonds: The topic we’re going to talk about today is criminology. I guess we’d better start with the definition of what criminology is.
Lawrence Sherman: Criminology is the science of law-making, law-breaking, and law-enforcing. The starting point for my own preference is to have criminology become the science of making better decisions about how to make laws and how to respond to law breaking or to prevent in the first place.
David Edmonds: And you’re a pioneer in something called experimental criminology, a branch or criminology. Tell us what experimental criminology is.
Lawrence Sherman: Experimental criminology is a field that is defined by a method, much like experimental physics or experimental biology. The method of course embraces a wide range of questions, but in the case of criminology, it’s a bit more profound in its implications because for most of its history criminology has been essentially a descriptive or observational science, sort of like astronomy. We don’t think that we can intervene in the way the planets revolve around the sun, the big dispute was whether or not they did. So there’s all these very important descriptive questions in any science, but in medicine, the descriptive questions translate very quickly into prescriptive questions of how you treat patients who are sick, how you prevent people from getting sick in the first place. And by developing a field of experimental criminology what we accept is that the core concerns of a discipline of criminology have to be how societies make decisions and what decisions they should make to deal with their crime problems. So that goes well beyond the descriptive and the observational, the purely theoretical. It requires having very hard empirical evidence, especially randomized control trials, which is the primary method in experimental criminology. Fifteen years ago I founded the Academy of Experimental Criminology, now we have a Journal of Experimental Criminology, we have a division in the American Society of Criminology, we even have an application group called the Society of Evidence-Based Policing, designed to promote the conduct of experiments in policing, the use of the results of those experiments in structuring police practices, improving police methods. You could have a society for evidence-based corrections, evidence-based prosecution. Prosecutors are about the most reluctant group to get involved in experimental research; they more than any other part of the criminal justice system tend to think they have all the answers. Evidence-based sentencing is very big, there’s a lot of interest on the part of judges now who say that it’s unethical for them to be sentencing people without knowing the consequences of their sentencing decisions. And this is all coming together in the 21st century to reframe the environment of criminology, to expect criminology to provide the same kind of interventionist guidance that medicine provides, as opposed to biology, as opposed to chemistry. We are an interventionist science and not just observational.
It’s an option that’s much more damaging potentially psychologically than just sitting in your prison cell and having your lawyer do all your talking for you. Damaging psychologically in the sense that it is painful in the moment, not necessarily damaging in the long run.
David Edmonds: So you say you’re interventionist and not just observational, but are you dragging the rest of criminology with you, or would you say most criminology is still practised in the old descriptive style?
Lawrence Sherman: Yes, most criminologists alive today would have been heavily influenced by the role of social science in the latter 20th century as a source of social criticism, as a source of values that were contrary to conventional values at the time, greater tolerance for diverse lifestyles, greater human rights – lots of good things that social science was associated with. If you consider a book like The American Dilemma by Gunnar Myrdal which helped us to come to grips with the fundamental immorality of the segregationist laws in the United States which were being challenged in the 1960s when police research first became visible, and my PhD supervisor Albert Reese did systematic observation of things like police arresting black people more than white people, using police brutality more against black people, and that’s really what drew me into the field: the fact that his research was observationally and descriptively documenting all this helped us to accept that there were problems and that we had to do something about it. But in my own career development I was very fortunate in having the guidance of an observationalist scientist to help me become an experimentalist. My teacher never did an experiment in his career, but he very much encouraged me and gave me good advice about how to do experiments, in part because I had the chance to do it since I had spent some time as a research analyst in the New York City police department before I got my PhD. That cocktail, that mix of practical experience with social science at a very high level has been the basis for me pursuing and promoting this agenda of interventionist criminology and that means experimental criminology.
David Edmonds: So give me a couple of examples of experiments you’ve carried out.
Lawrence Sherman: In 1981, my former supervisor in the New York City Police Department had become the chief of police in Minneapolis and at my request he obtained unanimous approval from the Minneapolis City Council to randomly assign arrest. It was the first clinical trial in the world, randomized control trial in the use of arrest for any offense. It was in the context of police not having made arrests for misdemeanor domestic violence, common assault, and a new law that gave them the power to do it even if they hadn’t witnessed the offense, in Minnesota. We got the city councils approval to enlist 40 police officers to act as doctors would in a randomized control trial, randomly assigning their patients to different treatments. Now there’s some criticism in criminology of calling an arrest a treatment: it’s a sanction, it’s a step in the process of prosecution. But from the standpoint of the individual who gets arrested, it can be an intervention that changes their life for the better or for the worse.
David Edmonds: So in this trial, some people were arrested after allegations of domestic violence, and others were merely warned at the scene of the crime, and there was a random approach to who got what treatment, and then you looked at the effect of that?
Lawrence Sherman: We did, and the initial effect was quite impressive. There was actually a third option which was that the police would ask the offender to leave the home for the night. And the lowest repeat offending rate over the next six months was the group that had been arrested, and this made headlines all over the world. It actually provoked changes in the laws of 28 states in the US, became policy in the UK and Australia and other places were convinced that this good news that in a scientific experiment, punishment actually worked to repeat offending, showed the wisdom of a retributivist policy that was both morally satisfying and empirically effective. And the bad news is that, as we said at the time, you have to replicate a finding like this to be certain of its generalizability, and this one turned out not to replicate.
David Edmonds: So it worked in one specific part of the United States but didn’t work elsewhere in America or elsewhere in the world?
Lawrence Sherman: Yes. In the early 1980s Minneapolis was a booming economy, very low unemployment rate and when we went to Milwaukee with high structural unemployment and segregation, vast areas of Milwaukee is a black underclass with very low employment rates,, in that context what we found out was that when you mix arrest for domestic violence with unemployment, either for individuals or even neighborhoods of high unemployment, arrest backfires: it doubles the risk of repeat offending, in contrast to the effect where you have, even in Milwaukee where people were employed or the neighborhoods had high employment, arrest was an effective deterrent. So we begin to see this connection between the social context of individual offending and the effects of an intervention. Just as there’s some evidence in medicine that some kinds of medicine work well for some kind of people or some contexts but work very badly in others. And it’s this sort of specification that experimental science is capable of doing, whereas theoretical science can’t do it on its own. It’s got to follow the experiments.
David Edmonds: But presumably, that’s a problem for the entire discipline because there are so many causal factors involved that you’ll never be sure that your experiment can be replicated in the next town, let alone the next country.
Lawrence Sherman: Well this is actually true in other kinds of science. Darwin famously took 20 years to publish – a lot of people, including some distinguished sociologists thought that was just because he was uncertain and they cite him as a great scientist who was purely observational, but they don’t know Darwin’s work. What he was doing for 20 years was experiments, and he needed, in his own view, to have those experiments confirm his proposed laws of natural selection.
David Edmonds: That first experiment you undertook, you found that there were powerful effects within a six month period. How do you know that the results that you found are going to survive longer than that? Presumably, you can’t repeatedly go back to the same people?
Lawrence Sherman: Well, the possibility of doing that is actually quite great, and while we didn’t do it in Minneapolis, a much better experiment done in Milwaukee a few years later, 1987 to 1988, is one that we have just followed up for 24 years. Milwaukee was the experiment in which we first found that the effect of arrest depended on whether the suspect was employed so that among the unemployed suspects, arrest doubled the rate of repeat domestic violence and cut it in half among employed suspects. So the question was how long would that last? Many other things happened in their lives: they can get arrested for other crimes, there can be economic conditions changing. Most theorists I think would say that the impact of a randomly assigned arrest in 1987 or ’88 is unlikely to persist into 2012, and they would be wrong. In fact, the effects got bigger around 12 and 15 years out, and the negative effect of arrest on unemployed people is the most powerful persisting effect. There’s no positive benefit from arresting employed people that lasted 24 years. There was a slight difference but it wasn’t what we call statistically significant. It could’ve been due to chance, but what was clearly not due to chance was the 24-year impact of causing more domestic violence among people who were arrested and unemployed.
David Edmonds: I’m fascinated by the arrest experiment that you’ve been talking about. Can you give us another example of an experiment that you’ve been working on?
Lawrence Sherman: Yes, in the mid-1990s, the Australian National University asked me to help design a test of a very old method of dealing with crime that they call restorative justice but which has been really the traditional basis for justice in the Middle East, in the aboriginal Canadian tribes, and in many other parts of the world in which the primary purpose of justice is not to do what Immanuel Kant described in the 18th century of inflicting a just measure of pain, no more no less than what each individual deserves for the seriousness of their transgression. The traditional purpose of justice was to repair the damage to relationships that allowed marginally existing communities to go on existing. So there’s a long-standing human, almost evolutionary process of trying to work out a conflict which has been created by a crime that disrupted a relationship. That’s the context, what’s the experiment? The experiment was in Canberra when police identified people who they thought might be appropriate for a meeting between victim and offender with the victim’s family and the offender’s family and instead of prosecuting them in court they would be diverted to this meeting, led by a police officer and at the end of the meeting there would be an agreement that the offender would do something to try to repair the harm to the victim. But prior to getting to that point, there would be a very robust discussion in which the offender would have to begin by saying in front of this group what they did. And the success of the police efforts to have the offenders not only describe the breaking of the law that they did, the harm that they caused, but in most cases, they would voluntarily apologize. And then this discussion about how they could either do community service or in some cases direct personal service, but the victims didn’t really want the money or the compensation. They mostly wanted the apology.
David Edmonds: The success was defined how?
Lawrence Sherman: Well, in the short run the success was holding a conference which didn’t happen in 100 percent of the cases, but it did happen most of the time, and then having everybody walk out of the conference saying ‘yes, this was a good thing to do’. And according to the victims, they felt much better having gone to the conference and they certainly felt much less angry than victims who didn’t have a chance to have this kind of conference and apology. The offenders actually felt terribly ashamed and there’s some evidence they were traumatized by it. They were actually reliving the conference in future days, months, years ahead, having nightmares about it, racing thoughts about how angry some of the people were in the room.
David Edmonds: Is that a good outcome?
Lawrence Sherman: Well, it’s a means to an end, and the end appears to be less repeat offending. We do have pretty good results across ten randomized control trials that would be on par with rehabilitation programs for offenders, at much greater expense, after prosecution, sometimes prison, and the best you can get by way of reducing repeat offending is something we achieved without ever taking these people to court and much cheaper and much quicker with far higher levels of victim satisfaction. So after four of these experiments in Australia, the British government invited us to test the same method but at a different stage of the criminal justice process. So from 2011 to 2005 we ran eight experiments in Britain which were supplements to prosecution, not substitutes for it, and the results were quite comparable. Overall we’re reducing repeat offending measured by convictions compared to the control group by close to 30 percent. I don’t think there is a rehabilitation program in the UK that actually works that well and certainly not for the very low cost of engineering this kind of meeting.
I think the long-term follow up is certainly to be informed by what we have now learned about how the offenders have reacted to having this kind of conference, and because there has been a kind of knee jerk opposition by conventional criminal justice policy-makers to the use of restorative justice as appearing too soft and as something the tabloid papers would criticize, and no politician in a democracy would ever want to offend the tabloid newspapers, but what I think we can say is that it’s not a soft option. It’s an option that’s much more damaging potentially psychologically than just sitting in your prison cell and having your lawyer do all your talking for you. Damaging psychologically in the sense that it is painful in the moment, not necessarily damaging in the long run. We have had people who’ve led miserable lives, one of whom has written a book about his experience in restorative justice. This experience for him got him out of a career of 5,000 burglaries. As far as he’s concerned, even though he still remembers the trauma of that conference, it’s the best thing that ever happened to him in his life.
David Edmonds: What kind of skills do you need to be a really good experimental criminologist? Because you have to come up with a hypothesis about what will work, so presumably you need to draw on, what, economics, psychology, all sorts of other disciplines?
Lawrence Sherman: Well, criminology itself is a multi-disciplinary field. It’s one that has competition from economics, from psychology, sociology, these are all fields in which journal articles are published with crime as one of the measures in the studies they’re doing. But criminology is like a sponge. We welcome the basic science disciplines to do what they are interested in doing. Very often it’s in proving a kind of theoretical premise. The economists are very fond of showing that punishment works because it fits a rational choice model. Daniel Kahneman has blown that model completely out of the water in terms of how people really do make decisions, and his work on how people experience pain has a lot to say about the relationship between punishment and conduct. Just as this notion that you can work with inertia in how people make decisions in a sort of ‘nudge’ context that by changing whether they have to tick the box to go one way or the other or when you send them a text message, they’re more likely to pay a criminal fine. That’s called behavioral economics now, and it absolutely belongs in experimental criminology because it gives us information about how to undertake interventions to get people to obey the law, to comply with legal punishments at the lowest cost possible to the tax payer to make a safer society and a more just society but also to use as much soft power and not to be using harder power than is really necessary in the circumstances. We wouldn’t be thinking that way if it wasn’t for behavioral economics and psychology and other fields that we then incorporate into experimental criminology.
David Edmonds: You’ve been working with governments around the world, you’ve been working with police federations. Are you worried about getting your hands dirty, about taking money and somehow losing your objectivity?
Lawrence Sherman: Fortunately as long as your salary comes from a university, I don’t think you have to worry about losing objectivity. I have to say that there’s a lot of discussion in criminology, there’s a lot of hand-wringing about chasing government grants and whether your objectivity is compromised. But in criminology, there’s something interesting going on right now which is we’ve been doing experiments without central government grants, especially in the UK and what happens when the local police department wants to do an experiment and they call up the university and ask for help is that if you don’t charge them any money or not very much money they put up an amazing amount of resources, sometimes even including the data analysts who are going to gather the data and work on the study. So it’s a different model. It’s a model that gets away from any party political interference with a crime policy. Domestic violence is a good example: in the United Kingdom, the party political view of it has been mandatory arrest is the only thing you can do. The governments haven’t allowed the kind of experiments we’ve done in the US. They haven’t been allowed in the UK at a central government level. Well now, you’ve got a new system of local control of police departments in the UK and all sorts of experiments can be possible, and if one police and crime commissioner doesn’t want to do the experiment, you can go to another.
David Edmonds: Crime remains stubbornly high despite these experiments. Is that a side of the failure of experimental criminology?
Lawrence Sherman: I think the evidence is against your premise. Serious crime in the United States and the United Kingdom has been falling substantially in recent years and the most common reason that police chiefs give for that in the United States is that they’ve been concentrating police patrols in hot spots of crime, at hot times. And I’m very pleased to say that our research was something that appeared to have launched all that. First of all in Minneapolis we discovered that 3 percent of the street addresses produced over half of all the crime. So that was a descriptive observational statement. Then it was a predictive statement because we said the places that were hot last year were going to be hot next year. Policing historically was trying to give all of the community, all of the landmass, equal attention. But that’s like treating patients who aren’t sick. And so we developed this idea of hot spots policing which would concentrate police resources in a small number of places where most of the crime was occurring and especially at the times when crime was occurring. The first randomized control trial in doing that was in Minneapolis, just like the first control trial and arrest, and Chief Anthony Bouza persuaded the city council to say we’ll take police cars out of low crime neighborhoods and we’ll put them into the high crime hot spots and on average, over the course of the year we double the level of patrol from about 7 percent of the time there would be a police car in these hot spots, it went up to 15 percent of the time, and the difference in the crime rate was about two thirds. Fifty percent reduction in robbery in the hot spots, for example. Now there have been over 20 experiments replicating this and they pretty consistently showing that you push crime down in the hot spots, somewhat less consistency but still overall positive result on not displacing the crime to the areas nearby that hot spot.
More complexity about the question of whether you displace offenders to different kinds of crimes or far away from the hot spot, and in a way, it’s almost metaphysical, because if they go to New Zealand from Minneapolis, we’re not going to know that. But what we can take heart in is the great accumulation of evidence now, the replication on the medical model of repeated experiments producing pretty much the same good news that is now part of what we’re doing in Trinidad where the homicide rate is roughly 50 per 100,000, and where there has been greater use of patrol in the daytime than in the evening but most of the homicides occur between 6 p.m. and 2 a.m. So with a new commissioner there, there’s been a very strong push for evidence-based policing, drawing on experimental criminology, and we’re about to launch a randomized control trial for Trinidad to see if there can be a big reduction in violence and serious crime, using this preventive strategy, and I’m optimistic that the basic approach of altering environments and measuring very carefully whether the changes in the environments can reduce crime over a long time. That’s kind of what happened with public health with clean water and clean air and other environmental strategies that in a way are trying to put doctors out of business. Well, they’ll never be out of business completely but certainly, fewer people getting sick, people living longer, people living longer because they’re not getting murdered, you can see there’s a very close connection between criminology and public health.
David Edmonds: And is that what motivates you? Is it the impact on policy and people’s lives, or is it the intellectual puzzle, working out what works and what doesn’t work?
Lawrence Sherman: All of the above. You can’t take this line of work without having a profound curiosity about how it all fits together, and then for me, what you can do about it. Why would we want to understand more? Certainly because we’re curious, but also because we want to make a difference. A lot of people who are curious are happy to stop with explaining it. Other people like to come up with a strategy, but they don’t want to test it. But they don’t want to test it. Winston Churchill once said that it’s very important that any strategy, no matter how beautiful it is, actually be examined to see if it’s working. And I like to take the whole journey from understanding and explanation to predicting that a certain intervention will work and then testing whether that prediction is correct, reformulating the theory if necessary, and then testing the intervention, we can build up a body of evidence that in the long run will help us have the safest societies in human history.
David Edmonds: Lawrence Sherman, thank you very much.
Lawrence Sherman: Thank you.