An agenda for crime-control research

If we wanted to know how to have less crime and fewer people behind bars, what research would we need to do? I nominate 17 candidate studies.

Like, I suspect, most people who do research, my focus is mostly on the stuff I’m doing, or plan to do.  I try to keep current on new findings and theories, but I don’t have much occasion to think about a broader crime-research agenda.

That changed recently, when I was asked to join the Committee on Law and Justice of the National Research Council. Looking over the Committee’s portfolio of current and proposed projects, I found myself dissatisfied by how unadventurous it was. (Does the world really need one more death-penalty study?) That led me to formulate an alternative agenda, less focused on resolving theoretical and empirical puzzles and more focused on ways of reducing the level of criminal risk faced by individuals and communities while also reducing the prison headcount specifically and the damage done by the criminal justice system: especially, but not exclusively, to poor minority neighborhoods.

The list is below the jump. Comments welcome. (Offers of research funding even more welcome; many of the ideas are far away from my specific competence, it wouldn’t be hard to find smart folks to carry out any of these projects.)


Imprisonment of the actually innocent: prevalence, systemic causes, and cures

Capital-punishment litigation has revealed that the risk of conviction of the actually innocent is not small, especially in cases dependent on eyewitness identification of strangers. That suggests that the much larger population of non-capital cases, which are not subject to similar post-conviction scrutiny, might also include a substantial number of the innocent. But there is no published estimate of the fraction of all prison inmates serving times for crimes they did not commit. A study might usefully put bounds on that number, identify the case characteristics and enforcement practices most associated with false conviction, and propose strategies for mitigation, including the importation of error-finding practices from medicine and aviation.

Individual social capital, collective efficacy, and crime

Social isolation as measured by lack of friends is a risk factor for both offending and victimization. Neighborhoods with many socially isolated people also tend to be neighborhoods with low collective efficacy: limited capacity to secure public-goods contributions of all kinds from residents.  What are the actual patterns of social isolation, low collective efficacy, and criminal victimization? What policies might increase the individual social capital of the friendless and the collective efficacy of low-morale neighborhoods, and what might be the impact on crime rates?

Drug law enforcement and incarceration: costs, benefits, alternative strategies

Half of all federal prisoners, and about a fifth of state prisoners, are serving time for drug law violations. The U.S. has more drug-law prisoners per capita than any other OECD country has total prisoners per capita. Massive incarceration has not prevented price decreases for the major illicit drugs.  Is there a rationale for continuing heavy expenditure on drug investigations, prosecutions, and incarceration?  Are very long sentences for drug-dealing cost-justified? What would be the consequences of 50%, 75%, or 90% reduction in drug enforcement effort? Of changing drug sentencing to shorten base sentences computed from drug and quantity and increase the importance of enhancements based on other conduct, such as violence?

Reducing drug-dealing violence in the U.S. and abroad

Drug-related violence continues to make up a substantial fraction of U.S. homicides, and Mexico and northern Central America face massive problems of dealers killing one another, officials, and ordinary citizens. What law enforcement and sentencing strategies could create effective disincentives for the use of violence among drug dealers, both domestically and in source and transit-country markets?

Illicit tobacco distribution

Heavy tobacco taxation reduces rates of smoking, but creates opportunities for markets in untaxed products. Worldwide, the volume of evaded taxes is estimated at $50 billion annually, comparable to the revenues of the illicit opiate trade. The problem is compounded when high-tax jurisdictions are convenient to low-tax jurisdictions: New York to North Carolina, for example. Untaxed cigarettes constitute an estimated three-quarters of the market in the South Bronx, and a third of the market in the United Kingdom. Chinese factories produce enough illicit cigarettes each year to serve the entire U.S. tobacco market.  Current enforcement efforts combined with current price differentials make tobacco smuggling a high-reward, limited-risk criminal enterprise: much more attractive than dealing in purely illicit drugs. There appears to be a risk of “tipping” from the current equilibrium of generally high compliance to a low-compliance equilibrium. What are the patterns and consequences of illicit tobacco dealing, and what regulatory, tax, and enforcement strategies could do a better job of controlling it?

Pain control and drug diversion: improving the terms of the tradeoff

Under-treated chronic pain and the diversion of prescription opioids to drug abuse are both widespread phenomena, as is suboptimal prescribing of pain medication by under-trained physicians. Controls on diversion threaten to worsen the under-treatment problem and compromise physician autonomy and patient confidentiality.  Is it possible to design policies that perform better on all outcome dimensions?

Alcohol taxation, user controls, and crime

Problem drinkers account for a large share of violent crime, along with very heavy health care costs. There is reason to think that higher taxes or minimum per-unit-alcohol pricing (recently adopted in Scotland) could reduce crime. But by how much? What would be its other effects?

An alternative or complementary strategy would be to make access differentially difficult for problem drinkers (as evidence by convictions for alcohol-related crimes). Sobriety 24/7 does this with alcohol testing and sanctions, with apparent success. An as-yet-untried option would be to require that all alcohol purchasers – not just those of youthful appearance – be “carded,” and to change the markings on the drivers’ licenses of convicted alcohol offenders to make them “non-drinkers’ ” licenses.

What are the design issues involved in implementing such policies, and what
range of benefits might be achieved?

Focused deterrence strategies: evidence and policy implications

“Focused deterrence” embraces two related sets of ideas: (1) selecting some mix of offenders, offenses, offending groups, and geographic areas for concentrated attention with the goal of “tipping” situations from high-violation to low-violation equilibria, thus creating potentially lasting impacts with temporary costs; and (2) using direct and explicit warnings to reduce the amount of punishment that needs to be actually inflicted in the process. A variety of interventions work on these principles. What is known about their efficacy, their characteristic risks and failure modes, and the circumstances under which they are beneficial or the reverse?

Potential contributions of social service agencies and health care providers to crime control

Offender populations, patient populations, and recipients of social services overlap. Police and criminal justice agencies have been justly criticized for a one-dimensional view that treats offenders as only offenders without reference to their health and social-service needs, even when serving those needs might reduce offending behavior. But the reverse is also true: social-service, education, and health-care providers and agencies pay little attention to the offending activity of their clients. No one expects the head of a TANF office or the principal of a school to know how many of the children being served by that institution got arrested last year, still less holds them accountable for changing that figure. Nor is a trauma center expected to worry about how many of the gunshot victims it patches up go out and shoot someone else. An earlier NRC report testifies (per Carl Bell) to the value of the Good Behavior Game in improving outcomes for the children whose first-grade classrooms use it. But it continues to spread only slowly, and its potential contribution to crime reduction has yet to be estimated quantitatively. Some social interventions, such as the Nurse-Family Partnership program, have been evaluated for their crime-control benefits, that idea has yet to penetrate most routine operations. What is the range of potential interventions? What is the evidence of their possible efficacy, and how large a contribution might each one make to reducing crime and incarceration? What are the managerial challenges and risks involved in pursuing policy interventions in this direction?

Back to the Panopticon: applications of low-cost information-gathering to prison management

Prisons are, notoriously, run more by their inmates than by the guards, and more by the guards than by the warden. Part of the problem is that those with nominal authority can’t wield it effectively without information, and information is scarce and hard to come by and under the control of prisoners and guards. The results include prisoner-on-prisoner assault and sexual assault, guard-on-prisoner assault, dealing in contraband (and therefore drug abuse), and the flourishing of prison gangs.

But gathering some sorts of information has become radically less expensive with the falling costs of computing power. A webcam costs a few dollars. A model airplane capable of hovering and equipped with high-definition video and audio and transmission capacity costs $199 at Brookstone.

That raises the possibility, first imagined by Bentham, of making real the nominal power of the warden to control the prison. What would be required to make this possibility actual, and what risks and disadvantages would such an attempt encounter?


Crowdsourcing policing: organizing citizens to gather and transmit real-time crime information

The fiscal crisis of state and local governments means that the United States is now, in all probability, at Peak Cop, Peak Jail, and Peak Prison. More and more of the correctional job will have to be done while offenders are in the community rather than behind bars, and the job of policing will have to be accomplished with fewer sworn officers.

In the UK, police have made aggressive use of fixed cameras in public places to observe offending behavior. That approach has found more resistance in the U.S., but some cameras are up and audio “shot-spotter” technology has now been deployed in several cities.

The fact that most people on the street have cell phones with cameras and the ability to send images wirelessly, added to the growth of habits and institutions of on-line information-sharing, might transform policing as radically as has transformed publishing. The same model airplanes that could help a warden manage a prison could also empower “community watch” groups to be able to actually watch their communities in useful ways.

How can citizens be organized and motivated to gather real-time crime information and transmit it in usable form to the police, and how would police departments have to re-organize themselves to make use of that data? What are the risks to civil liberty, and how could abuses by volunteers (Peeping Tom behavior, harassment of personal enemies via surveillance and/or false reports) be controlled?

Prison gangs and their street extensions: problem assessment and policy options

Prisons are, supposedly, places where formal social control reaches its maximum. But the emergence of prison gangs, in the U.S. and elsewhere, challenges that idea; one recent study describes it as “a hole at the center of the state.” The problem is most profound when prison-based gangs, and prison-based extremist groups such as the Aryan Brotherhood, extend their control back into the community. In neighborhoods where a large proportion of young men expect to be incarcerated, the prospect of future incarceration, and the consequent need to ensure against in-prison victimization by the gang, can generate strong incentives to collaborate with gang activities among those still in the community.

How extensive is the prison-gang problem? What is its geographic distribution? What practices of prison management encourage, or could discourage, the growth of prison gangs? What can police, prosecutors, and community-corrections authorities do to limit the power of prison-based criminal organizations?

Enforcing the conditions of community corrections: current and alternative practices

Very high offender-to-staff ratios, inadequate monitoring, and slow, clumsy, and sporadic sanctioning characterize much of the community corrections system (probation, parole, and juvenile community supervision). Little appears to have changed since the report of the Reinventing Probation Council, except for the deployment of elaborate systems of risk and needs assessment to sort offenders into levels of supervision.  The result continues to be high rates of re-offending and re-incarceration.

Alternative practices using more intensive monitoring and swift, certain, mild sanctioning have appeared in various spots around the country, with apparently superior results. What is the evidence for these new practices? What resource and organizational commitments do they require for success? What, quantitatively, are their potential benefits? What barriers stand in the way of their more widespread adoption?

Pretrial release: decision-making and monitoring

Bail reform and the substitution of release on recognizance has gone some distance to remedy the problem of punishment before trial and of the undue pressure on jailed defendants to plead guilty (since doing so can actually lead to less time behind bars than going to trial and being acquitted). It has also, where instituted, eliminated the cost of the bail-bond business without any apparent increase in pre-trial crime or failure-to-appear, or in the capture rate for absconders. But a number of states, including California, continue to employ the bail-bond system.

Position monitoring of offenders

GPS monitoring can be used both to enforce conditions of community corrections – curfews,  requirements to be at particular locations (treatment programs, workplaces) at particular times, or, conversely, not to be at particular locations (drug-dealing streetcorners, residences of prior victims) – and as part of the sanctions process for other conditions (enforcing curfews, community-service requirements, and periods of home confinement assigned for violations). The monitoring itself can also act as a sanction for violations, or be used to make probation more onerous (especially at first) and therefore a closer substitute for incarceration. In addition, the location of those being monitored can be compared with crime locations (either derived from 9-1-1 reports or determined by “shot-spotter” technologies) to deter re-offending and provide evidence about those offenses that are not deterred.

What are the benefits, costs, managerial issues, and risks associated with the use of position monitoring? For what populations of offenders would it be most appropriate? It it possible to estimate quantitatively its potential contribution to improving community corrections outcomes and reducing re-offense rates? Are its net effects on incarceration or re-incarceration positive or negative?

Managing drug-involved offenders: diversion, drug courts, and testing-and-sanctions

Despite the “buzz” about HOPE, treatment diversion programs (TASC, California Proposition 36, Arizona Proposition 200) and treatment drug courts remain far more prevalent. What is the evidence about the costs, risks, and consequences of these alternatives? Is it possible to say which offenders ough to be assigned to which programs? What would constitute an appropriate integrated system?

Beyond COI: measuring the benefits of crime reduction

Estimates of the “economic cost” of one or another form of crime, or of crime generally, abound, and are frequently referred to in benefit-cost analyses of crime-control programs. But most such estimates focus on completed crime and ignore the costs of crime avoidance behavior and of residual anxiety about victimization risk. Moreover, attempts to compute total cost involve, implicitly, comparison with a counterfactual in which there is no crime; such a condition could hardly exist among human beings. Practical choices about crime control police all involve relatively modest reductions in victimization or in the criminal riskiness of the environment; but an estimate of the form “The total cost of crime is $200 billion per year” does not support an estimate that a 1% reduction in crime would be worth $2 billion.

A conceptual exploration of the logic underlying cost-of-crime estimates and their application might save considerable expense of research funds in estimating imaginary quantities while generating a framework within which practically useful work could be done.

Self-command, locus of control, and desistance

The National Institute on Drug Abuse has adopted as official policy Terrie Moffitt’s conclusion that increasing the capacities of actual and potential offenders to exercise self-control could reduce their propensity for both substance abuse and crime. What is the current evidence about techniques for bringing about that result? How does self-control interact with “locus of control”: a person’s perception that his or her outcomes depend strongly on his or her own choices rather than on chance and the whims of others? Can enforcement and community-corrections practices designed to more tightly couple outcomes to offender behavior, combined with direct  communication of deterrent threats, change locus of control, and, if so, how much does that contribute to offenders’ capacity to control their impulses and act on their considered judgments about what serves their interests?

Author: Mark Kleiman

Professor of Public Policy at the NYU Marron Institute for Urban Management and editor of the Journal of Drug Policy Analysis. Teaches about the methods of policy analysis about drug abuse control and crime control policy, working out the implications of two principles: that swift and certain sanctions don't have to be severe to be effective, and that well-designed threats usually don't have to be carried out. Books: Drugs and Drug Policy: What Everyone Needs to Know (with Jonathan Caulkins and Angela Hawken) When Brute Force Fails: How to Have Less Crime and Less Punishment (Princeton, 2009; named one of the "books of the year" by The Economist Against Excess: Drug Policy for Results (Basic, 1993) Marijuana: Costs of Abuse, Costs of Control (Greenwood, 1989) UCLA Homepage Curriculum Vitae Contact:

11 thoughts on “An agenda for crime-control research”

  1. “Does the world really need one more death penalty study?”

    As long as we’re executing innocent people and some of the pols on one side of the aisle are joining with all the pols on the other side in chanting, “Kill, baby, kill!”, I’m gonna say, “Yeah, we probably do need another one or twelve. It’s a waste of money, but until those knuckleheads accept the facts, I guess we have to do it.”

    But that shouldn’t stop your program from being implemented.

  2. While you’re at it, see if you can get a study of a really useful photo lineup procedure.

    Instead of just a few photos (still much better than the in-person lineup system), the person being asked to make an identification gets a computer screen or iPad. Ths photo of ths suspect is entered into the database, and the computer puts it into a multidimensional location based on fitting it into the right niche along the spectra of numerous objectively recognizable qualities in the photos.

    The id procedure begins with the program showing the person a randomly selected image. The subject takes a mouse and moves it around the first photo, which reveals the adjacent photos, ehich vary slightly from the first in an ordered way (lighter to darker, thinner to fatter, clearly male/masculine to “Pat” to clearly female/feminine, etc. The person “drives” through the database by steering the mouse to find pictures that look increasingly like the image that they have in their head …. When they get the sex and color right, they start driving to get the weight right, and then hairstyle (it’s a HUGE database, with potentially many tens of thousands of photo headshots, so each one can be sequenced by all the apparent dimensions), etc.

    The amount of time needed for the person to find the match is measured, along with the “distance” in the multidimensional space between the photo they settle on and the true photo. Anyone who actually lands on the suspects photo and says “that’s the one” is probably making a pretty good ID that should stand up and avoids all the risk of suggestive lineups, especially the notoriously bad cross racial ones. Why don’t we have this already?

    1. How do you know all these images won’t just confuse the witness? I know eyewitness ids are problematic, but this procedure sounds kind of nightmarish.

      But I admire your creative thinking.

      1. This process works pretty well for making color and tone corrections to photos — the trick is that there aren’t a lot of alternatives at each point in the process. But you wold have to figure out the navigation, and you would probably need a database (real or synthetic) big enough so that all the changes appeared incremental.

  3. Child and partner abuse as a long-term contributor to violence, and ways of reducing it through early childhood intervention and access to child care. I seem to think you have written about nurse visits?

    Restorative justice (the matrix of pros and cons, evaluation of international initiatives) and the problem of giving victims a genuine say in punishment.

    Education and training in prison, and rewards for learning. (In preference to the creepy Panopticon stuff.)

    Environmental factors. Lead has been identified and (independently) more or less fixed. What else is out there?

    Traffic crime and the potential of automated in-vehicle and in-road systems. (Bentham comes back as a freeway Panopticon!)

  4. Mark, the only nod to economic or white-collar crime in your agenda is the study of cigarette smuggling. Granted, the relationship between economic crime and politics makes it difficult to get (federal) funding for this area, it shouldn’t be neglected.

  5. An area of particular interest to me is a subset of the ‘Individual social capital, collective efficacy, and crime’ category.

    In non-Establishment-culture neighborhoods (immigrant, minority, local minority, cultural minority), collective efficacy is often defined as crime (for a well-known example, the opening scene of “The Godfather”.) I see this in my neighborhood–historically very poor white (basically, a ghetto–but all white). It has always had a reputation as a place where only certain kinds of crime are allowed–and the non-allowance of breaking-and-entering, for example, wasn’t historically enforced by the police. (The police attitude has tended to be–“It’s __–what do you expect?”)

    I’m not sure what the well-formed question is–maybe “how can the government interact with collective efficacy so as to limit spill-over effects?”

    1. I thought that item was interesting too. I am a little skeptical that people in bad neighborhoods are socially isolated. Aren’t they just poor and politically non-powerful? Do relatives count as friends? I would be interested to know more.

      Also the social services agency item. Do schools count?

      And how much would another war on poverty itself help? What about more psychotherapy for those who’ve survived traumas, which I believe are quite common when you’re poor?

  6. Some additional topics:

    According to the DOJ, “Of the 272,111 persons released from prisons in 15 states in 1994, an estimated 67.5% were rearrested for a felony or serious misdemeanor within 3 years, 46.9% were reconvicted, and 25.4% resentenced to prison for a new crime.” I would like to see research into whether prison conditions affect recidivism rates. James Wimberley mentioned availability of prison education and training, these are the most obvious, and are probably already well studied. But what about crime victimization rates for prisoners, especially victimization rates for sexual violence. Does a greater victimization rate lead to a greater recidivism rate? What about prison size and crowding — any effect there? How about opportunities for animal care, access to the outside world (a window with a view), prison food quality and nutrition, even prison architecture, maintenance and upkeep, and decoration? Prison looks to be a soul-destroying, PTSD-inducing experience; is there any crime reduction benefit to making it less traumatic?

    I’d also like to see research into differential crime and imprisonment rates between the states. In an American Conservative article skewering the myth that Hispanic immigration increases crime rates, Ron Unz makes a point that there are substantial difference in imprisonment rates between similar demographics in different states, but that the causes for this are not well understood. For instance, white males age 18-29 are more than 4x more likely to be incarcerated in Florida than in Massachusetts. Is it possible to formulate any hypotheses about these differences between similar demographics in different regions? Is there a difference in policing, or in crime? If in policing, does the additional imprisonment actually improve anything? If the difference is in actual criminality, what might be the cause of that? This sort of comparative study would seem to be more likely than most to actually produce actionable policy recommendations.

    Perhaps such studies exist already — if so, I’d be glad to know about them.

  7. And the first item is hugely important. I think there are some programs that are working too, like video’g interrogations, and having a devil’s advocate office in DA offices. But I don’t know that they are very widespread.

  8. The obvious problems with crowdsourcing policing is that a) a significant fraction of images that people will be motivated to send will involve police misbehavior and b) even without the part about mobile phones identifying their users, any sustained amount of audio or video will contain identifying information. It would be really interesting to study ways that information could be preserved against selective use while simultaneously being at least mostly scrubbed of ID.

Comments are closed.