View this issue as a PDF: CrimHighlightsV8N4.pdf
This issue of Criminological Highlights addresses the following questions:
- Does ‘broken windows policing’ (the targeting of minor forms of disorder) reduce crime?
- Do legal aid payment structures affect the quality of the preparation of criminal cases?
- Would you buy life insurance after reading this paper?
- For how long is a record of offending predictive of future offending?
- Does the presence of a mandatory minimum sentence affect the manner in which cases are prosecuted?
- Is the Level of Supervision Inventory-Revised (the LSI-R) a useful tool for classifying women?
- Do laws requiring the transfer to adult court of youths charged with serious offences act as a deterrent?
- Do ordinary citizens really want to spend more money on building prisons?
The police strategy of targeting minor disorder on the street – so-called ‘broken windows policing’ – does not reduce crime.
In 1982, in an article in the Atlantic Monthly, James Q. Wilson and George Kelling suggested that if the police targeted minor instances of visible disorder – e.g., panhandling, prostitution – the rates of more serious crimes would drop. Though 25 years later the evidence supporting their theory is at best mixed, there continues to be widespread belief that this strategy works.
This study re-examined an earlier important study that purports to show beneficial effects of ‘broken windows policing’ – a study of crime in New York by Kelling and Sousa. Although Kelling and Sousa were not willing to share their data with the current authors, equivalent data were compiled from original sources which allowed for more stringent assessment of the impact of this policing strategy on crime. The problem in assessing the impact of changes in policing strategies during the 1990s is that “Any study of the influences on American crime patterns during the past 20 years is complicated by the massive period effects that have generated dramatic year-to-year changes in crime across the country… Those cities that experienced the largest increases in crime during the [beginning of] this period [the 1980s] subsequently also experienced the largest drops [in the 1990s]” (p. 291).
This same problem is evident in New York City when one looks at individual neighbourhoods. Crime dropped in New York City during the 1990s. And, crime dropped most in those neighbourhoods in which broken-windows policing was most aggressively implemented. It would appear, at first blush, that broken windows policing was a cause of the drop. However, it is also true that broken windows policing was most aggressively implemented in neighbourhoods that had experienced the largest increases in violent crime during the 1980s. The Kelling-Sousa study essentially related changes in violent crime to levels of misdemeanour arrests, ignoring the fact that the changes were, essentially, reversions to an earlier level of crime. Controlling for the size of the change (generally an increase) in violent crime during the period 1984-1989 eliminated the crime reducing impact of misdemeanour arrests during the period 1989-1998. What goes up comes down, whether or not there is a police officer or city employee nearby fixing broken windows.
The Kelling-Sousa study was not the only published study apparently showing support for the broken windows hypothesis. Another study noted that there was a relationship, for the period 1970-2000 in New York as a whole, between the rate of misdemeanour arrests and violent crime, controlling for known correlates of crime. The problem, once again, is that this ‘effect’ is driven largely by the decrease in crime that occurred in the late 1990s (the period when ‘broken windows policing’ was in vogue in New York). The problem is that attributing a drop in crime that occurred largely in one time period to a single cause is risky. The authors note that one could logically examine the ‘Broken Yankees Hypothesis’ (p. 298) by looking at the cumulative number of New York Yankee (baseball) championship wins as the possible cause, on the theory that New Yorkers are happy when their home team is winning and thus less likely to commit crime. Plugging this variable into the equation, one finds that the ‘Broken Yankee Hypothesis’ fits the data just about as well as the ‘broken windows’ hypothesis.
Conclusion. Though it can be shown that certain police activities – e.g., the targeting of ‘hot spots’ where crime is chronically prevalent – can reduce crime, the suggestion that broken windows policing will reduce violent crime is without empirical support and is most likely an artefact of the practice of focusing police resources (and, in particular, high rates of police charging of minor offenders) in those areas in which crime had been on the rise.
Reference: Harcourt, Bernard E. and Jens Ludwig. (2006) Broken Windows: New Evidence from New York City and a Five-City Social Experiment. The University of Chicago Law Review, 73 (1), 271-320.
Changes in the structure of legal aid payments to lawyers affect the manner in which cases are processed.
Many jurisdictions are concerned about the cost of providing legal services to criminal clients who are dependant on state sponsored legal aid systems. The key question is simple: is there a relationship between the manner in which legal aid payment regimes are structured and the manner in which cases are handled by lawyers? It appears to be well-established that “the [legal] profession will not ever concede publicly that there may have been a decline in the effectiveness of defence work [as a result of the way in which legal aid payments are structured]” because to do so would be “to admit publicly… that financial arrangements play some part in how clients are advised and cases prepared” (p. 741).
In recent years, in a number of jurisdictions, including Scotland, standard or block fee arrangements have been seen as a way for governments to control legal aid costs. The Scottish legal aid system involves a fixed fee paid for everything that is done on a case through the first 30 minutes of an actual trial. After that, lawyers are paid a flat amount for each day of trial. A higher tariff is attached to a case that goes to the Sheriff Courts (the higher level court) than cases dealt with in the District Courts (courts generally presided over by lay justices). Although a case could be deemed to be ‘exceptional’ and, as a consequence, eligible for extra funding, almost no cases received this designation.
From interviews and an examination of available records, it would appear that ‘specialist’ criminal law firms were initially hard hit by the change in the payment structure, but recovered within a couple of years by “sharply increasing the caseloads undertaken” (p. 727). In addition, levels of contact with clients decreased sharply. Scottish defence lawyers, instead of receiving written documents disclosing the prosecution’s case, are given the names of prosecution witnesses and can arrange to have them interviewed and have statements taken. These were previously “widely regarded by defence solicitors as vital to the preparation of a case for trial and conducting a trial” (p. 729). Approximately 60% of the 60 solicitors interviewed thought that the use of this process of learning about the case against the accused had decreased as a result of the new fixed payment system. Privately and anonymously, the lawyers would admit to the researchers that “The general level of preparation is less than it was before” (p. 729).
In the Scottish courts, a hearing is required before a trial is held to “review whether the parties are ready for trial… [and to attempt to] reduce the number of trials cancelled during, on, or shortly before the date of trial…” (p.733). This requirement increased the proportion of cases resolved before the day of trial. When the change in the legal aid payment structure was introduced, the proportion of cases completed early in the process decreased slightly, as did the trial rate. The number of cases resolved at this intermediate hearing or on the day of the trial increased somewhat. As one solicitor noted “The most economic use of fixed fees is simply to plead everyone not guilty, apply for legal aid, get legal aid granted, then plead them all guilty in the intermediate [pretrial hearing]” (p. 735). In general, the stage at which cases were concluded moved toward what the solicitor described as the most economically advantageous for the lawyer.
Conclusion. It would appear that the legal aid tariff structures do, indeed, affect the manner in which cases are handled by defence counsel. A number of those interviewed suggested that “the system of fixed payments seems to have led to a reduction in client contact and a decline in overall levels of preparation and case investigation…. Many… suggested that… the overall effectiveness of defence work had diminished” (p. 739).
Reference: Tata, Cyrus and Frank Stephen (2006). “Swings and Roundabouts”: Do Changes to the Structure of legal Aid Remuneration Make a Real Difference to Criminal Case Management and Case Outcomes? Criminal Law Review, 2006, 722-741.
Deceptive sales practices in the life insurance business have become part of ‘normal business’: sales agents are taught by companies how to be deceptive and not get caught.
The easiest way for a large corporation to respond to allegations of fraud or dishonesty is to suggest that there are a few ‘bad apples’ in every organization, but that the company’s policies and practices are honest. Life insurance companies, however, would have a hard time making this argument. A practice, referred to in some locations as ‘misselling’ of insurance policies – “violating industry rules, guidelines and codes of ethics, if not the law” – occurs quite regularly. These practices involve “institutionally endorsed manipulation, deception and sometimes fraud” (p. 994).
It is not surprising that fraud exists. What is interesting in this context is that companies appear routinely to engage in what might be called ‘moral risk’ – the ways in which an insurance relationship ‘fosters behaviour … that immorally increases risk to the others’ (p. 994). That this is institutionalized can be demonstrated by the fact that companies appear not to modify these practices when their own employees bring it to their attention. Part of the problem with life insurance and one reason that customers are vulnerable is that it is difficult to assess exactly what is being purchased in a life insurance policy: “Customers have difficulty knowing what it is that they are buying and often even knowing how much they are actually paying” (p. 998). An industry booklet “reinforced the culture of misconduct” by stressing that “Agents need to assume and work with the belief that everyone is under-insured” (p. 1006). It was clear that life insurance agents who were interviewed as part of this study understood the nature of their deceptive practices with customers. A number of them indicated that they had changed jobs within the company “after discovering the market misconduct that was expected of them” (p. 1004). Indeed under the so-called “career agent” system, it appears that “most agents [leave] the business once they have exhausted their own family, friends and extended social networks” (p. 1005) as potential customers.
Life insurance sales earnings are based almost exclusively on commissions, “creating a strong incentive to market aggressively and to sell the highest commission product rather than the right product for the consumer’s needs” (p. 997). Large commissions to the seller lead, among other things, to a practice known as ‘churning’ – persuading customers to change their coverage even if the costs of doing so makes the change economically unwise for them. Replacement policies sometimes result in higher commissions than new client policies. Indeed, one insurance executive interviewed for the study freely admitted that instead of using agents on commission, the company would assign salaried employees to talk to certain customers in circumstances in which it was important to the insurance company to be completely trustworthy to a favoured client. On the other hand, “because promotion in the company ranks seemed to be based solely on productivity, agents who engaged in rule-breaking were likely to be promoted” (p. 1006).
Conclusion. It is clear that deceptive practices in the life insurance business are the result of organizational strategies, not individual deviance. Agents are, in effect, taught how to be deceptive with customers. Nevertheless, when faced with a scandal resulting from systematic market misconduct, companies attempt to view the agents as being responsible. Ironically, an employment practices insurer told the researchers that life insurance companies are seen as being among the worst risks regarding employment practices, because ‘life insurance companies faced potential liability suits from agents on the argument that the agents were structurally induced to participate in market misconduct” (p. 1009). It appears that on occasion within the insurance community, the “marketing practices of life insurers were themselves too flawed by moral risks to be insurable” (p. 1009).
Reference: Ericson, Richard V. and Aaron Doyle (2006). The Institutionalization of Deceptive Sales in Life Insurance. British Journal of Criminology, 46, 993-1010.
Offenders who have gone six or seven years without committing a new offence are only slightly more likely to offend than are people who have no criminal record at all.
It is common for employers to ask job applicants whether they have a criminal record. Though in some places, such as Canada, most offenders can apply for a pardon if they have lived ‘crime free’ for a certain period of time (3-5 years after the end of the sentence in Canada, depending on the offence), being pardoned does not mean that one does not, for certain purposes, have a criminal record. For example, although in Canada a criminal record cannot be used to deny a pardoned individual a job that is regulated by federal law, other employers can still refuse to hire someone who has a criminal record but has been pardoned. These decisions would appear to be informed by the notion that “once a criminal, always a criminal.” But are ex-offenders who live crime-free for extended periods of time, in fact, any more likely to offend than people who have no criminal record?
Two very similar studies (by the same three authors and published nearly simultaneously in different journals) use different data sets to examine the re-offending records of groups of male offenders. In the first study, the records of reoffending (to age 26) for all 13,160 males who were born in Philadelphia, Pennsylvania in 1958 were examined. In the second study the records of males born in Racine, Wisconsin in 1942 were examined up to their 32nd birthday. In each of these studies, the likelihood of subsequent offending could be compared for two different groups: those who, by a certain age, had offended and those who apparently had not offended by the time they reached this same age. In one study an arrest for a criminal charge was used as a proxy for re-offending; in the other the measure was a ‘contact’ with the police in relation to a criminal matter.
The findings from the two studies are quite consistent. Early in their adult lives (e.g., when they were age 20), the likelihood of offending for those who had offended at least once by age 18 was about four times as high as the likelihood for those who had been crime free at age 18 (8% vs. <2%). However, those people who had offended before they turned 18, but had been crime free up until age 25, had only a 2% likelihood of offending at age 25. This was, however, slightly higher than the likelihood of offending for those who had been crime free at age 18 (<0.5%). In the second study, the data show that those who had been juvenile offenders were more likely to offend as young adults, but “in any given year after the mid-20s, there appears to be little difference in offending likelihoods between juvenile offenders who have avoided offending during early adulthood and those with no record at all” (p. 72). A similar analysis was carried out comparing those who, by age 20, had not offended to those who had at least one police contact between the ages of 18 and 20. In their early 20s, these two groups differed dramatically: those who had offended as young adults (age 18-20) were more likely to offend in their early 20s. However, if a youth made it to about age 25 without re-offending, the likelihood of reoffending was no different than for those who had never offended.
Conclusion. This analysis suggests that those who are using criminal history information “should place [this information] into a context that pays close attention to the recency of the criminal record as well as the [mere] existence of a criminal record. That is, if a person with a criminal record remains crime free for a period of about 7 years, his or her risk of a new offence is similar to that of a person without any criminal record” (p. 80). It would appear that there is empirical justification for legal procedures like the pardon that recognize that former offenders who have not reoffended after a period of time do not, in fact, present special risks to society.
References: Kurlychek, Megan C., Robert Brame, and Shawn D. Bushway. Scarlet Letters and Recidivism: Does an Old Criminal Record Predict Future Offending? Criminology and Public Policy, 2006 (August), 5(3), 483-503. Enduring Risk? Old Criminal Records and Predictions of Future Criminal Involvement. Crime and Delinquency, 2007 (January), 53(1), 64-83.
Oregon’s “get tough on crime” law, passed by voters in 1994, had an impact on the courts: trial rates increased for the first two years after the law came into effect. More people went to prison, and they went to prison for longer periods of time.
On 1 April 1995, a sentencing referendum (Measure 11) brought in by the voters in Oregon resulted in long mandatory minimum sentences for 16 violent and sex related crimes. In addition it prohibited ‘early’ release from prison, and it provided automatic transfer of youths to adult court for these same offences. Five more offences were added to the list shortly thereafter. The theory behind the bill was simple: residents of Oregon were to be safer because offenders would be incapacitated or deterred. However, research on these topics [see, e.g., Criminological Highlights 3(1)#1, 6(2)#1, 7(3)#6, 8(1)#2, 8(3)#6] demonstrates that crime is essentially unaffected by legal changes such as those contained in the referendum.
Previous research has found that there is often “hydraulic displacement of discretion” – meaning that a change in the processing of cases has consequential effects in another part of the system – in circumstances such as that which followed the change in the Oregon law. In the case of mandatory minimum sentences, the typical finding is that “prosecutorial authority to determine which offenders are prosecuted [under the new provisions] is enhanced, whereas judges lose much of their authority over the sentencing process” (p. 11). In addition, it is often found that sentence lengths for ‘non-targeted’ offences increase along side of the ‘targeted’ offences.
In the case of Oregon, it was found that there was a decrease in the prosecution of Measure-11-eligible cases and an increase in the prosecution of ‘alternate’ cases (typically lesser degrees of the same offences which did not attract the mandatory penalty). Trial rates for Measure-11-eligible offences also increased in the first two years after implementation, and then reverted to their former levels. But the nature of pleas changed: there was an increase in the number of cases in which the accused decided to plead to lesser included offences, and a decrease in pleas involving the original charge. The rate of prison sentences, however, increased both for Measure-11 eligible cases and for Measure-11 alternate cases. The group contributing most to the increased use of prison sentences for Measure-11 cases were cases in which the offender had no history of offending. The average prison sentence increased from 77 to 105 months. However, this “success” has to be understood in the context of another effect: sentence lengths for some of the Measure 11-alternate cases decreased. Overall, though, imprisonment rates in Oregon increased during this period.
What seemed to be happening was that after the new law came into effect, rather than being charged with a Measure-11-eligible offence, an offender may be charged with a lesser offence, yet receive approximately the same sentence that the Measure-11-eligible offence would have drawn before Measure 11 came into effect. In other words, “fewer offenders have been sentenced for the [Measure-11] offences, whereas a greater proportion of offenders have been sentenced for Measure-11-alternate offences. [The] analysis suggests that this shift resulted from the use of prosecutorial discretion and the downgrading of cases, that, although technically Measure-11-eligible, were not deemed appropriate for the associated mandatory minimum penalty” (p. 31). Said differently, prosecutors were sometimes willing to downgrade the offence when the mandatory minimum punishment did not fit the crime.
Conclusion: It is clear that “prosecutorial discretion is the force that drives the implementation and… the impact of mandatory minimum sentencing policy” (p.33). There was, in addition and not surprisingly, evidence of variation in the manner in which the new policies were implemented across locations. Clearly, the implementation of mandatory minimum sentences is not as straightforward as it would appear to be in the legislation.
Reference: Merritt, Nancy, Terry Fain, and Susan Turner (2006). Oregon’s Get Tough Sentencing Reform: A Lesson in Justice System Adaptation. Criminology and Public Policy, 5 (1), 5-36.
The Level of Supervision Inventory-Revised (LSI-R), which is widely used to identify offenders who are likely to reoffend, “misclassifies a significant portion of socially and economically marginalized women” (p. 384) whose pathways into crime do not follow typical male patterns.
The Level of Supervision Inventory-Revised (LSI-R) is described by its advocates as being equally effective in classifying women as it is for men. The LSI-R consists of 54 items said to assess two kinds of risks and needs that relate to continued criminal activity: “static” risks (life experiences, such as prior convictions, that cannot change) and “dynamic” risks (factors such as peer interactions that change over time). The theory behind the LSI-R appears to be that “offending stems from an assortment of incentives and disincentives regarding criminal and conventional behaviour that arise from various sources, especially family members and peers” (p. 386). These underlying causal mechanisms “are said to explain offending behaviour for all individuals regardless of their gender, race and ethnicity or their pathway to crime” (p. 386). In other words, it is assumed, among other things, that the “causal mechanisms” for men and women are the same. It is important to note, however, that LSI-R scores predict only a small portion of the variance in recidivism (about 12.5%, an estimate based largely on studies with males).
There are empirical reasons to suggest that there may be important limitations in the usefulness of the LSI-R in predicting women’s recidivism. In particular, criminologist Kathleen Daly has suggested that women’s ‘pathways’ to crime may be different from those of men. In particular, she has shown that there are “conditions and circumstances [for women] that spawn violence and illegal forms of economic gain” (p. 390). These differ from the pathways associated with male criminality. Pathways to crime that are much more likely to involve women include committing criminal acts (e.g., prostitution, drug dealing) to survive on the street, being involved in drug offences along with an intimate partner, suffering sexual abuse and neglect as children, and being in abusive relationships with intimate partners. Finally, one group could be considered to have male-like pathways to crime: women whose crimes relate largely to economic issues (e.g., as a result of being economically marginalized or because of simple acquisitive motivations).
As part of this study, women convicted of felonies were interviewed in Minnesota and Oregon prior to beginning community supervision and again approximately one year later. LSI-R measures were obtained during the first interview. Information was also obtained about how the woman ended up being involved in the crime that resulted in the community supervision. Two researchers reviewed the biographical information about each woman and classified her as either following a ‘gendered pathway’ to crime or as following an ‘economic’ pathway to crime. A small number of women could not be classified (e.g., because they had committed very minor crimes or their crimes were seen as being isolated incidents in their lives). A woman was classified as a recidivist if there was official information that she had violated a supervision condition, or had been rearrested, reconvicted, or had her supervision revoked.
Overall, there were no significant differences in the recidivism rates of ‘high’, ‘moderate’, and ‘low’ risk women as measured by the LSI-R. Most interesting was the finding that for the women who got involved in crime for economic reasons (the typical male pathway), the LSI-R did have a small relationship with recidivism: it predicted approximately 9% of the variance in recidivism for this subset of women. But for the women whose pathways to crime were ‘gendered’ (street women, drug-connected, battered, harmed and harming), there was no relationship whatsoever between the LSI-R and reoffending. Said differently, for a large group of women, the LSI-R was useless in predicting their likelihood of recidivism.
Conclusion. The LSI-R predicts only a small portion of the variance in recidivism for men and is even less useful for women. Since the reasons women end up committing crimes are often different from those of men, it makes sense that this instrument is, for many identifiable groups of women, essentially useless as a classification tool. It turns out that an LSI-R score is a predictor of future offending only for those women who are identified as having been involved in crime for reasons that resemble those of men.
Reference: Reisig, Michael D., Kristy Holtfreter and Merry Morash (2006). Assessing Recidivism Risk Across Female Pathways to Crime. Justice Quarterly, 23(3), 384-405.
Laws that require that youths charged with serious offences be tried in adult court do not reduce violent juvenile crime rates.
By 2003, 22 U.S. states had provisions whereby youths charged with serious offences were automatically tried in adult court. Most of the research on transfer provisions in youth court has focused on the impact of being transferred and has generally found that youths, if anything, are more likely to re-offend if they are treated as if they were adults than if they were allowed to remain in youth court (see, e.g., Criminological Highlights, 3(5)#5, 5(5)#3, 2(4)#3, 1(3)#2). However, these transfer provisions are usually justified on the basis of general deterrence: the theory is that youths will decide not to offend if they assume that they will receive harsher penalties from an adult court. This paper examines this hypothesis.
By analyzing violent crime rates in those 21 states for which there existed 5 years of data prior to and 5 years of data after the effective date of the new law, the possible impact on crime in each state of the statutory exclusion law could be assessed. For 17 of the states, there was no significant change in the juvenile violent crime rate (as measured by juvenile arrest rates). For two there were increases and for another two there were decreases. For only one of these states (Maine) was the change as predicted: an abrupt permanent decrease after the change in the law. For these four states, control states were identified that did not have legislative waivers at around the same time as the intervention states and that resembled the intervention state on other dimensions (e.g., size, location, juvenile arrest rate). In no cases did these control states show effects similar to the intervention state suggesting that the changes in crime rates were real. However, the inconsistent direction of these changes in juvenile crime rates suggest that the law change may have been completely irrelevant to the change in crime rates even in these four states. When arrest rates for homicide were examined, there were no significant effects coinciding with the changes in law.
Though there are no clear explanations for why there might have been a significant drop in crime that coincided with the implementation of legislative transfers of violent juvenile offenders to adult court in two states (Maine and Wisconsin), these findings need to be considered along side the increases in crime that took place in two other states (Indiana and Missouri). Once again, as in the case of the deterrent impact of harsh adult sentencing laws (see Criminological Highlights 7(3)#6) isolated instances in selected jurisdictions of ‘success’ in lowering crime through harsh practices need to be evaluated in a larger context. In these instances, there are as many significant negative impacts of the transfer provisions as there are positive impacts.
Conclusion. “Legislative waiver of juveniles to adult criminal court does not have a deterrent effect on violent juvenile crime. Indeed this has been the case with other get-tough legislation…. [T]his may not sit well with policy makers who suggest toughening the juvenile justice system’s response to serious and violent juvenile crime as the panacea to reducing its occurrence….” (p. 50). These data need to be interpreted in conjunction with other data suggesting that there are no positive effects for those youths who are transferred to adult court.
Reference: Steiner, Benjamin, Craig Hemmens, and Valerie Bell (2006). Legislative Waiver Reconsidered: General Deterrent Effects of Statutory Exclusion Laws Enacted Post-1979. Justice Quarterly, 23 (1), 34-59.
A nationally representative sample of U.S. residents report overwhelming support for increased spending on preventing youth crime, for drug treatment for non-violent offenders, and for the police, but they show little support for spending money on building more prisons.
A serious problem with many public opinion polls concerning public policy is that members of the public are typically not forced to make tradeoffs among programs that they favour. For example, a question like “Should more money be spent on the police to reduce crime?” doesn’t offer the respondent any choices of other strategies that they might prefer. It is easy to be in favour of something if nothing has to be given up. If one wants to know what the public would do if faced with real fiscal choices, one needs to ask how they would allocate a fixed budget to various priorities. In an earlier study it was found, for example, that Canadians would generally prefer to invest in the prevention of crime or in non-prison sanctions rather than pay for more prisons (Criminological Highlights, 4(1)#5).
In this study, 1300 interviews were carried out during the summer of 2000 with a representative sample of U.S. adults. Respondents were asked to put themselves in the shoes of their local mayor and imagine that the Federal Government had just given their municipality a sum of money which could be allocated to crime control or crime prevention, or it could be given back to local residents in the form of a tax rebate. Five strategies were listed for each respondent and respondents were asked to allocate the money across these strategies. Across the sample, 37% of the money was allocated to prevention programs to keep youths out of trouble, 22% to drug treatment for offenders convicted of non-violent crimes, 21% for more police on the street, and 8% was allocated for more prisons. Residents allocated 12% for a cash rebate to local residents.
Black Americans were more likely than white and Latino Americans to want to allocate funds for programs to keep youths out of trouble, and were less likely than members of these groups to want to allocate funds for prisons. Those who indicated that they worried a lot about crime indicated that they would spend more of the money on prisons and on drug treatment for non-violent offenders and on the police, and less money on prevention programs to keep youths out of trouble. “It appears that those who currently worry about crime are more concerned about immediate responses to crime at the expense of long-term youth crime prevention” (p. 327). On the other hand, those who had reported having been victims of crime “tended to give less money to prisons and police and more to prevention (though these [effects] are significant only for certain groups of victims)” (p. 330). Income had very little impact on the allocation of funds: “the lowest income levels… had remarkably similar responses to these questions as those with the highest income” (p. 330).
Conclusion. These findings are consistent with other studies carried out with less nationally representative samples which showed that “despite the overall punitiveness of the public toward criminals, there is also significant support for both rehabilitation of offenders and early intervention programs designed to prevent high risk youth from later engaging in criminal activity” (p. 333). Though the public would spend considerably more of any allocation of funds on the police than they would on the building of more prisons, even the police would not receive as high a proportion of any special ‘crime prevention’ funds as would prevention programs.
Reference: Cohen, Mark A., Roland T. Rust, and Sara Steen (2006). Prevention, Crime Control or Cash? Public Preferences Towards Criminal Justice Spending Priorities. Justice Quarterly, 23 (3), 317-335.