1 Does deterrence work?
We begin with the most basic question: Does deterrence work? Strong statements are often made in the negative. âDeterrence is a principal with much immediate appeal,â wrote England's Home Office in 1990,
but much crime is committed on impulse [âŚ] and it is committed by offenders who live from moment to moment [âŚ] It is unrealistic to construct sentencing arrangements on the assumption that most offenders will weigh up the possibilities in advance and base their conduct on rational calculation.1
In fact, the world is soaked in deterrence, something that easily escapes notice because of its utter ordinariness. The class of people who persistently put their hands on hot stoves, cross the street without looking, and steal cars in front of police officers is very small. The class of New York City neighborhoods in which drivers routinely ignore alternate-side parking regulations (and, in parallel, fail to respond adroitly to the occasional lifting of those regulations) is precisely nil, a finding confirmed by the author's extensive participant-observer research.
The expectation of official action clearly does not deter all offending, but it equally clearly deters a lot of offending, and when it is altered or removed, people's behavior changes. There is an abundance of examples of such influence and adaptation, much of it fascinating. Many prosecutors, for instance, have office policies setting nonstatutory minima for drug prosecutions, with arrests for less than a set quantity generally being dismissed or charged as possession rather than trafficking. The experience of police in those jurisdictions is that drug offenders soon learn those minima and scrupulously stay under them, carrying, for example, exactly twenty-seven pieces of crack rather than the informal and arbitrary but well-understood twenty-eight.2 Ethnographic research into illicit gun markets in Chicago showed that gangs restrict their members from carrying and using guns, one reason being that police attention following gun violence interferes with money-making crime such as drug-dealing.4 Motorcycle gang members in the Midwest were subject, as convicted felons, to substantial federal penalties for carrying guns; they took, instead, to carrying nineteenth-century percussion firearms exempted under an obscure âcurios and relicsâ provision in the governing criminal statute.5
Ordinary experience validates the common sense of deterrenceâat least with some people, with respect to some offenses, some of the timeâand the idea that existing criminal-justice activities are in fact creating deterrence. When police departments go on strikes, bad things tend to happen.6 A great deal of research, much of which this book will draw on, ends up pointing to much the same conclusion. âMy assessment,â writes Philip Cook, one of our most thoughtful and thorough students on the subject, âis that the criminal justice system, ineffective though it may seem in many areas, has an overall crime deterrent effect of great magnitude.â7
The equally obvious frequent failure of deterrence is not probative that the general idea is mistaken or that sanction structures aimed at deterrence are having no impact. Part of the problem here is the frequent commission of the prison warden's fallacy: that the prisons are filled with men manifestly not deterred by legal threats is not evidence that other men were not deterred. Strong claims that deterrence is ineffective are often in fact arguments that the limits of deterrence have been reached and that additional enforcement and penalties will not produce additional deterrent impact. That may be correct, but it is not evidence that deterrence as such does not work. It does raise the considerably more interesting question of whether deterrence can be extended beyond those limits, and if so how: a question that will consume most of this book.
Assessments of particular policy interventions also support the operation of deterrence: in, again, given contexts. There are very few controlled experiments in deterrence, but quasi-experiments and more ordinary evaluations frequently show impact. In a set of controlled trials, mandatory arrest for domestic violence reduced recidivism in a number of cities;8 in another controlled trial, increased police presence in high-crime areas reduced crime in those areas.9 A controlled trial that applied largely deterrent tactics reduced crime in drug-trafficking hot spots.10 Intensive police patrol reduced robberies in New York City subways.11 The prospect of greatly enhanced drunk-driving enforcement in the United Kingdom reduced both drunk driving and related accidents.12 Steps to increase the apprehension of airline hijackers in the early 1970s all but ended hijacking, not because of actual apprehensions but because the likelihood of being detected rose to such a prohibitive level that people stopped trying.13 A mandatory one-year sentence for the illegal possession of a firearm reduced gun-carrying and gun crime in Massachusetts.14 Mandatory sentences for gun offenses reduced violent crime in a pool of six cities.15 A study of arrest ratios in 171 cities found that increased enforcement deterred street crime.16 Crackdowns on street dealing dried up drug markets in Lynn, Massachusetts.17 The announcement that a vigorous and established powder cocaine market in Houston would be, on a date certain, the focus of an interagency law enforcement crackdown dried up the market before the crackdown could actually be implemented.18 Intensive directed patrol aimed at identified offenders reduced homicide and gun crime in Indianapolis, Indiana.19 The record thus shows that deterrence is not just operating, it can be deliberately created in at least some settings and against at least some forms of offending.
A long tradition of âperceptual deterrenceâ social-science research has gone at the deterrence question in a different way.20 This research sought to unpackâusing individual-level cross-sectional survey and scenario studies, and panel survey studiesâhow and why people reacted to what they thought, or were told, would happen if they offended; why they did, or did not, care about that; and how it shaped their subsequent behavior. This research, too, is generally supportive of the idea that deterrence is operating. While varied, the cross-sectional and scenario studies tended to find that higher perceived risks of sanction were associated with lack of offending or of likelihood to offend. Waldo and Chiricos, for example, found that college students who thought that they would be arrested and convicted for marijuana use and theft were less likely to say that they would commit those offenses.21 In general, the certainty and, to a lesser extent, the swiftness of sanction mattered more than the severity of sanction, to the extent that many researchers concluded that severity was all but or in fact irrelevant.
A core theme in deterrence research has to do with what ultimately came to be called the âexperiential effectâ: the observation that, as time passes, many people come to lower their estimates of the risks of offending. The debate began in interpreting survey research showing that low risk estimates were associated with higher levels of offending. The initial conclusion was that low estimates of risk led to offending, with associated ideas of criminogenic personality types resistant to deterrence.22 Subsequent thinking shifted to the idea that, instead, offenders were simply more informed and realistic: that âmost people (young and old) have over pessimistic estimates of their ability to get away with breaking the law [âŚ] but that when they participate in illegal acts they soon discover that they can commit them with relative impunity.â23 According to this idea, as offenders commit crimes and escape sanction, or see others do so, they adjust their risk estimates downward. This is not a problem with the idea of deterrence, or with some sort or sorts of undeterrable personality, but with the objective characteristics of the sanction regime supposed to generate deterrence. Experience breaches the âshell of illusionâ that one can't get away with offending.24
This set of issues led to perceptual deterrence research using panel studies that allowed assessment of how attitudes toward risk evolved over time on the basis of concrete experience with offending and sanction, and to other research designed to get at these issues. This research generally supported both the idea that deterrence was operating and that its effects could be vitiated when people thought that their objective risks were low.25 Much of the research supporting this conclusion was conducted outside populations of serious offenders and assessed attitudes by subjects such as college students to low-level offenses, but the basic propositionâpeople learn from experience that they face rather low risks and act accordinglyâmay plausibly be generalized. The rare studies of these dynamics with more or less serious offenders bore them out in that population. Horney and Marshall surveyed incarcerated felons and found that those with more experience of offending had lower risk estimates, and that the actual rate of prior arrests moved those perceptions: For most crimes in the survey, those offenders who had been arrested relatively often, compared to the number of crimes they had committed, felt more at risk.26 Lochner, analyzing data in large national surveys, found much the same dynamic amongst youth offenders.27 This turned out to be one of the main reasons that some perceptual research turned up an absence of deterrence effects: Subjects knew perfectly well that risks were low and maintained that knowledge even after being sanctioned. Though they also often updated their risk assessments, they remained low enough to justify further offending. As we shall see, this can be true even for extremely serious crimes.
So, deterrence effects are there to be found. The strong caseââdeterrence does not workââis clearly false. Deterrence effects are also, often, found to be absent. The weaker caseâthat deterrence does not work very well or does not work where we would particularly like it to, or is not worth the various costs of producing itâis still a live option. Even those who believe in deterrence can be realistically dubious about its actual impact and the prospect for enhancing that impact. âThus the problem faced by those responsible for the everyday enforcement of law in democratic societies is that deterrent effects are apparently modest [and] certainty of punishment is low under many circumstances,â wrote the Committee to Review Research on Police Policy and Practices of the National Academy of Sciences. âAs a consequence, deterrence strategies alone are unlikely to be a sufficient basis for an effective system of social regulation.â28
As on the positive side, the most telling evidence is ordinary and everyday: In the face of a great deal of punishment, offending continues. The death penalty manifestly fails to deter a good many murderers. Recidivism is very high amongst offenders who have had prior arrests and convictions and who thus have good reason to fear sanctions: A 2002 Bureau of Justice Statistics study showed that 67.5 percent of those released from US prisons in 1994 were rearrested for a new crime within three years.29 Striking increases in sanctions, as for drug crimes in the mid-1980s, failed to curb widespread drug epidemics. Explicit official promises of particular sanctions for particular crimes, as with stiff federal mandatory sentences for gun carrying, fail to reduce gun homicide.30 Research with actual criminals frequently reveals that the mechanisms designed to produce deterrence are not operating remotely as intended. Doob and Webster report on a study based on interviews with sixty experienced, and presently incarcerated, burglars and armed robbers. âThe respondents were blunt in reporting that neither they nor other thieves whom they knew considered legal consequences when planning crimes. Thoughts about getting caught were put out of their minds.â31 This is devastating to the deterrence project: If offenders not only do not fear consequences but willfully refuse to even consider them, we simply cannot reach them.
All this cannot be taken to mean that deterrence is not operating, in principle or in fact. Offending might be even higher than it is in the absence of these sanctions; higher sanctions might yet work. But it is at least evidence that existing deterrence regimes, some involving policies on which we rely heavily, including some involving very stiff sanctions, are not working as intended. This, again, accords with the common sense. We are reluctant to think that consequences do not matter, but it is plain that for many people whose behavior concerns us deeply, the consequences they in fact face do not matter much, or enough, or at all.
Recent research from Cincinnati, Ohio provides a case in point. A project aimed at understanding and addressing homicide and violent crime identified some sixty-nine violent groupsâdrug crews and the likeâcomprising something less than 1,000 individuals. Nearly 750 of those individuals were identified by name and found in police records. Those individuals, most of them fairly young, averaged more than fourteen prior misdemeanor charges, more than ten misdemeanor convictions, more than seven felony charges, three felony convictions, nearly thirteen charges as juveniles, and nearly nine convictions as juveniles: which is to say, all told, they averaged nearly thirty-five charges and nearly twenty-two convictions apiece. Nearly 57 percent had ten or more misdemeanor charges; nearly a third had ten or more felony charges. This is the very model of failed deterrence. And here, as often, when deterrence fails it matters a great deal: These groups of offenders were associated with nearly three-quarters of the homicides in Cincinnati.32
Here too, evaluations of real-world policies support the failure of deterrence. The seminal Kansas City Preventive Patrol Experiment varied vehicular police patrol in Kansas City and found no difference in crime, fear, or public perception between a standard level of patrol, twice that level, and no patrol at all.33 This is often taken to imply that patrol is not effective, and even that policing is not effective; it means neither, but it can be taken to imply that patrol more or less as it is in fact usually practiced is not effective. An evaluation of police foot patrol in Newark found reductions in residentsâ fear but no reduction in serious crime.34 Studies of mandatory sentences for gun crime...