Criminological Highlights Vol. 11, No. 6 – April 2011

View this issue as a PDF: PDF iconCrimHighlightsV11N6.pdf

This issue of Criminological Highlights addresses the following questions:

  1. How can crime and imprisonment both be reduced?
  2. If ordinary citizens were sentencing offenders, would sentences be harsher?
  3. What can be done to reduce reoffending by those on parole?
  4. Does sending offenders to prison reduce the likelihood that they will reoffend?
  5. How long does it take until those with a criminal record are no more likely to offend than those without records?
  6. Should governments spend money on sex offender registries and community notification systems?
  7. How do victim impact statements affect sentencing decisions?
  8. Why do girls in some neighbourhoods commit violent acts at a rate that approaches that of boys?

Rather than focusing on severity-based policies that increase already harsh sentences, policy makers should shift their attention to programs that use the police to make the risks and consequences of crime more clear and certain. Such a policy shift holds the promise of reducing both crime and imprisonment.  

Imprisonment rates in many countries, most notably the United States, are a concern in part because of the various costs of imprisonment and the fact that high imprisonment rates appear to have little effect in reducing crime.  There is a substantial amount of research suggesting that increasing the severity of sentences from current levels will not increase the (general) deterrent impact of the criminal justice system (see Criminological Highlights, 6(2)#1) and is not efficient in reducing crime through incapacitation (Criminological Highlights, 3(1)#1, 10(2)#5). 

This paper points out that deterrence always depends on both certainty and severity.  But variation in sentence severity – within levels that are plausible in western societies – does not appear to have much, if any, impact on crime.  Given the various costs of imprisonment (financial as well as social), a very attractive criminal justice approach to crime prevention is one that reduces both crime and imprisonment levels. Incapacitation does not qualify as such a policy since it “necessarily will increase the rate of imprisonment. In contrast, if the policy also prevents crime by deterrence, then it is possible that it will be successful in reducing both imprisonment and crime” (p. 16).  In addition, to the extent that the experience of prison is criminogenic (see Criminological Highlights,11(1)#1, 11(1)#2, 11(4)#2), policies that reduce imprisonment have an additional advantage.  There are, of course, many other ways to reduce crime.  However, given that substantial amounts of public money are spent on the criminal justice system, the question that should be addressed by criminal justice policy makers is a simple one: how can this “criminal justice budget” best be used?

A careful analysis of the data suggests that a fundamental shift should occur – from focusing on sentence severity to focusing on the certainty of apprehension.  A shift of this sort does not mean that by increasing police budgets, crime rates will automatically be lowered.  Instead this analysis suggests that targeted increases in police activity that increase the likelihood that offenders will be apprehended can prevent crime in the first place and thereby avert the need for punishing an apprehended offender.  In other words, averting crime also averts punishment.  For example, regular drug testing of probationers to enforce prohibitions against drug use resulted in more certain but shorter imprisonment periods (1-2 days); this, in turn, was quite effective in deterring probationers from drug use and other probation violations. In this way, the certainty of apprehension averted the need for exacting further punishment. What is crucial, of course, is that potential offenders must believe that their likelihood of apprehension and punishment is high. 

The difficulty is that achieving certainty in delivering punishments is elusive.  Not all police programs – or programs that simply increase the number of police in a neighbourhood – achieve high levels of real or perceived certainty of punishment.  However, “the key empirical conclusions… are that at prevailing levels of certainty and severity, relatively little reliable evidence of variation in the severity of punishment having a substantial deterrent effect is available and that relatively strong evidence indicates that variation in the certainty of punishment has a large deterrent effect, particularly from the vantage point of specific programs that alter the use of police” (p. 37). 

Conclusion:  If policy makers are committed to using criminal justice budgets effectively, shifting funds from imprisonment to policing could be effective in reducing both crime and imprisonment.  Since people are likely to be deterred by programs that increase the (perceived) likelihood of apprehension, those program will prevent crime and those people who are deterred will not end up in prison. There are, obviously, potential costs to such programs as well.  Hence programs that appear to be effective in one location need to be continually evaluated as they are implemented in new locations.  And these evaluations need to examine not just the impact on crime, but also other impacts on communities and residents. 

Reference: Durlauf, Steven N. and Daniel S. Nagin (2011).  Imprisonment and Crime: Can Both Be Reduced?  Criminology and Public Policy, 11(1), 9-54.

Back to top

Those citizens – jury members – who have intimate knowledge of specific criminal cases are quite content with sentences imposed by judges in those cases.   

Public opinion polls in most western countries suggest that the vast majority of people – typically about 70-80% – say that sentences, in general, are too lenient.  Extensive research carried out in many countries suggests that the answers to such questions reflect a belief based on inadequate knowledge of cases and the sentences actually handed down. Instead, the answers that people give to questions about ‘sentence severity’ appear to be based on people’s beliefs about sentences or the sentencing process rather than being carefully considered conclusions based on evidence of what goes on in court. 

This study – carried out at the suggestion of the Chief Justice of the High Court of Australia –examines how sentences, as handed down by the courts, are perceived by a group of ordinary citizens who have extensive knowledge of a single case: jurors in the Australian state of Tasmania who decided on the guilt of the accused in criminal trials.  Before the judge handed down the sentences in 138 trials in which there was a guilty verdict, jurors were asked to indicate the sentence they thought should be imposed.  Overall 52% chose a sentence that was more lenient than the sentence actually imposed by the judge, 44% chose a more severe sentence, and 4% gave exactly the same sentence as the judge. There was some variation across offence types but in all cases about half or more of the jurors recommended the same or a more lenient sentence than did the judge.  Ninety percent thought that the actual sentence handed down by the judge was very or fairly appropriate. 

Those whose preferred sentence was more lenient than the sentence actually handed down by the judge were significantly more likely to say that the judge’s actual sentence was very appropriate than were those who had selected a more severe sentence than the judge. “In other words, jurors who were more punitive were less tolerant of the judge’s sentence and less malleable in their views than the more lenient jurors” (p. 5). 

The responses of the jurors in this study to questions about sentencing generally were typical of those who answer such questions on public opinion polls.  These jurors were asked their opinion about sentences in general.  The majority thought that, in general,  sentences were too lenient for all offence types, most notably for sex and violence where 80% and 76%, respectively, thought sentences were too lenient.   Though jurors were slightly less likely to say that sentences generally were “much too lenient” after they heard the judge’s sentence in “their” case, the majority of jurors still believed that, in general,  judges’ sentences are too lenient.  Hence it would seem that this one exposure to a ‘complete’ case did not have a dramatic impact on jurors’ overall views of sentencing. Apparently, in general, the 698 jurors who participated in the study saw their case as being exceptional in the sense that the judge handed down an appropriate sentence. 

As in other studies, those jurors who thought that sentences, generally, were too lenient were more likely to think that crime in their state had increased (when, in fact, it had decreased in recent years).  Thinking that sentences were too lenient was also correlated with overestimating the proportion of crime that involves violence and underestimating the likelihood of imprisonment for those convicted of rape. 

Conclusion: The basic findings – that jurors are not more punitive than judges in recommending sentences for actual cases when jurors and judges have the same information – are consistent with other findings on public attitudes to sentencing. These findings underline the importance of responding sensibly to public opinion on sentencing. Most citizens have little if any information about the details of criminal cases. Hence their view that sentences are too lenient is best thought of as a ‘belief’ rather than an attitude based on a careful assessment of information.

Reference: Warner, Kate, Julia Davis, Maggie Walter, Rebecca Bradfield, and Rachel Vermey (2011).  Public Judgement on Sentencing: Final Results from the Tasmanian Jury Sentencing Study. Australian Institute of Criminology: Trends & Issues in Crime and Justice, No. 407. 

Back to top

Providing social services in neighbourhoods in which many former prisoners live is associated with lower rates of recidivism.

When prisoners are released, they typically return to the communities from which they came. It has frequently been pointed out that programs to help inmates reintegrate into their communities can be important in reducing crime (see Criminological Highlights, 3(2)#5, 6(1)#1).  Factors that make it difficult for former inmates to re-establish contact with offending peer groups can reduce future offending (Criminological Highlights, 10(5)#1).  What seems reasonably clear from the existing research is that re-offending is affected, in part, by the circumstances of the offenders in the communities they return to.  This paper examines the impact on subsequent offending of the availability of services to former inmates in the immediate neighbourhood to which an offender returns. 

Social service agencies in the immediate neighbourhood of a returning offender can provide help in addressing the factors that affect re-offending.  It would appear, however, from public health findings that, especially for the poor, it is important that these services be located relatively near their target populations.  Services that are distant aren’t likely to be accessed, in part because of the costs of getting to the service.  Welfare recipients, for example, have been shown to be much more likely to access services if these services are close to where they live. 

This study looks at California parolees who were released in 2005 or 2006.  Parolees are required to live at a specific address and each parolee’s presence at that address is likely to be verified by a parole officer. The locations of social service agencies that provide services to parolees are known because parole officers guide parolees to those services.   Hence the availability of services to parolees can be measured.  In this study, the measure of ‘proximate social service agencies’ was the number of different programs available to parolees within two miles of the parolee’s home.  

Various other factors were controlled statistically including residential stability, racial composition and level of disadvantage of the neighbourhood;  the age, race, and sex of the offender;  years the parolee spent in prison, and the number of violent and property offences committed by the parolee. The outcome measure was whether or not the parolee was returned to prison for any reason (a new offence or a technical violation of the conditions of release).

Those parolees returning to disadvantaged neighbourhoods and to neighbourhoods surrounded by other disadvantaged neighbourhoods were more likely to recidivate than were offenders living elsewhere.  But controlling for neighbourhood and individual characteristics, those parolees returning to neighbourhoods with many services available to them were less likely to be returned to prison.  On the other hand, if there was a large demand for services (measured by the number of parolees in the neighbourhood), there was increased recidivism, presumably because of the difficulty parolees had in accessing these services.

Conclusion:  With thousands of people being released from prison every year, it is inevitable that some will re-offend and be returned to prison.  However, it appears that public policy – in this case providing services in the neighbourhoods in which parolees live – can reduce future re-offending.

Reference: Hipp, John R., Joan Petersilia, and Susan Turner (2010).  Parolee Recidivism in California: the Effect of Neighbourhood Context and Social Service Agency Characteristics.  Criminology, 48(4), 947-979.

Back to top

Being sent to prison does not decrease subsequent offending.

Recent research (see Criminological Highlights 11(1)#1, 11(1)#2, 11(4)#2) suggests that sending offenders to prison is likely, if anything, to increase slightly the likelihood that they will re-offend compared to what would have occurred had they been given some other sentence.  Given that prison sentences are expensive (in Canada, about $322 per prisoner per day for federal prisoners and about $161 for provincial prisoners), if sentences – particularly short sentences – cannot be shown to reduce subsequent offending, it would appear to make sense to search for less expensive alternatives. 

This study, carried out in New South Wales, Australia, examined the criminal careers of two sets of offenders: those convicted of burglary and those convicted of non-aggravated assault.  For each offence type, pairs of convicted offenders were located one of whom had been imprisoned for the offence, the other who had received a non-custodial sentence.  The members of each pair were matched on variables that have been shown to relate to recidivism such as prior record, prior imprisonments, and whether bail had been refused (as an indicator of concern about reoffending). 

The results show that those who were imprisoned for assault were more likely to reoffend even after various factors not used for matching purposes were controlled for statistically.  For those convicted of burglary, the results were similar, but the difference in the likelihood of reoffending for those imprisoned and not imprisoned was not significant. 

A second study, also carried out in New South Wales, using a relatively similar approach, compared those given prison sentences to those given suspended sentences – non-custodial sentences similar to Canada’s conditional sentence of imprisonment.  In this study, scores measuring an offender’s ‘propensity to reoffend’ were calculated using 16 demographic (e.g., age, economic disadvantage of home neighbourhood) and criminal justice measures (e.g., criminal record, offence seriousness).  Pairs with the same ‘propensity scores’ were created with one of each pair going to prison and the other receiving a suspended sentence. The dependent measure was the length of time the offender remained free of offending in the community. 

A total of 2,650 pairs of convicted offenders with no prior prison sentences – one of whom was sentenced to prison, the other who received a suspended sentence –  were followed for about 1100 days.  There was no difference between the two groups in the likelihood of being reconvicted.  When examining the 1661 pairs of offenders with prior prison experience, those sent to prison were likely to reoffend earlier than were those who received a suspended sentence. 

Conclusion:  The results of the two papers are fairly consistent. “It would be unwise to imprison offenders when the only reason for doing so is a belief in the specific deterrent effect of prison” (Study 1: page 10).  The results “provide no evidence to support the contention that offenders given imprisonment are less likely to re-offend than those given a suspended sentence” (Study 2, page 10).   Clearly the findings that certain groups are more likely to reoffend when sent to prison are not completely consistent across studies.  However, what is consistent across studies and with other research is the finding that sending offenders to prison does not reduce subsequent reoffending. 

Reference:  Weatherburn, Don (2010).  The Effect of Prison on Adult Re-Offending.  Crime and Justice Bulletin (New South Wales, Bureau of Crime, Statistics, and Research) Number 143.  Lulham, Rohan, Don Weatherburn, and Lorana Bartels (2009).  The Recidivism of Offenders Given Suspended Sentences: A Comparison with Full-Time Imprisonment.  Crime and Justice Bulletin, Number 136.

Back to top

Those employers who use criminal records checks for job applicants should know that for most former offenders who have lived crime-free in the community for about 10 years the criminal record no longer predicts offending. 

There is substantial evidence that job applicants who have a criminal record are severely disadvantaged when they look for jobs (see Criminal Highlights 6(3)#2). Given that a substantial portion of people in many countries have criminal records, it is important to know how predictive these records are of future offending.  Recent research (Criminological Highligths 10(5)#6, 8(4)#4) suggests that, in general, after a few years of living in the community without additional offending, those with criminal records are no more likely to offend than those without records.

This paper looks at a large and varied group of offenders: a representative sample of 3,243 male Dutch offenders whose cases were adjudicated in 1977.  Their pre-1977 criminal records were made available to the researchers as were records of their offending thereafter.  In addition, a representative sample of same age male ‘non-offenders’ – those with no record of offending before 1978 -- was examined.  The question, then, is a simple one: how many years of non-offending does it take until offenders have the same probability of offending (defined as a conviction for a criminal offence) as those who have not previously offended?

Looking at all of the 1977 offenders, it is clear that if offenders reoffend, it is likely to occur very soon after their conviction (or release from prison).  But all offenders do not have the same likelihood of reoffending: older people and those without extensive criminal histories are, generally speaking, less likely to reoffend.  From the perspective of an employer, the question is a straightforward one: when does an offender’s likelihood of reoffending become indistinguishable from those who had never offended in 1977.  This might be called the point at which their offending likelihoods converge. It turns out that this is a function of two quite separate factors.  Younger offenders were more likely to reoffend and those with no criminal record prior to 1977 were less likely to reoffend. 

The youngest groups of offenders with no convictions before their 1977 conviction become indistinguishable from non-offenders after about 10-13 years (depending on the criterion used for being ‘indistinguishable’).  For older offenders, however, the point at which they become indistinguishable from non-offenders occurs earlier – 6 to 10 years for 27 year olds, and 2 years for men 42-46 years old who offended in 1977. 

For those with extensive criminal records in 1977, however, the time it takes for a person who is crime-free in the community to become indistinguishable from non-offenders is considerably longer and, depending on the criterion for being ‘indistinguishable’, they may always have a slightly higher likelihood of reoffending than those who had not offended prior to 1977. 

Conclusion: These data suggest that knowledge that a person once committed a criminal offence gives very little information about the likelihood that he will reoffend.  However, knowing the age of the person at his last offence and his prior criminal record increases the accuracy of prediction.  For those who had no criminal record before being convicted in 1977, ten years of crime-free living brings the probability of reoffending down to the level of non-offenders as it does for those 27 or older who, prior to offending in 1977, had no more than 3 previous convictions. It should also be remembered that, in general, reoffending, if it is to take place, is much more likely to occur shortly after the most recent conviction.

Reference: Bushway, Shawn D., Paul Nieubeerta, and Arjan Blokland (2011).  The Predictive Value of Criminal Background Checks: Do Age and Criminal History Affect Time to Redemption?  Criminology, 49(1), 27-60.

Back to top

Two common policies for dealing with sex offenders do not reduce the incidence of sex crime recidivism: (1) the requirement that sex offenders register their whereabouts with the police and (2) the requirement that police notify people who live in the same neighbourhood as convicted sex offenders of the sex offender’s whereabouts.  

Simple solutions to serious problems are often politically attractive.  Sex offenders, in particular, appear to be a magnet for ineffective approaches at reducing crime. Previous research has suggested most of these special ‘sex offender’ policies don’t work.  Residence restrictions are ineffective (see Criminological Highlights, 11(4)#7). Registration and public notification of the whereabouts of sex offenders have negative effects (see Criminological Highlights 7(4)#4, 8(6)#5, 9(2)#7) or are ineffective (see Criminological Highlights 4(1)#2, 5(6)#1, 10(3)#7). Policies such as these are based on the false assumption that a sex offender has an atypically high likelihood of reoffending (See Criminological Highlights 3(3)#3, 5(1)#4, 6(3)#3, 6(6)#8) 8(3)#8, 9(2)#5).  This paper examines the impact of South Carolina’s sex offender registration and notification policy on recidivism. 

The study examines recidivism rates of 6,064 males 16 years old or older who were convicted of sex offences for the first time between 1990 and 2004.  About half were registered at some point during the follow-up period.  Some had prior convictions, but not for sex offences. South Carolina’s registration and notification law came into effect in 1995, applied retroactively, and lasts for life.  “Survival” (no recidivism) was measured from the conviction (or end of incarceration period).  The study controlled for age, race, prior (non-sex offence) record, and whether the original crime involved an underage victim.   The analysis examined the relative risk of recidivism.

Recidivism was defined, in separate analyses, as either a new charge or a conviction for sex crimes, other person offences, or non-person offences.  Across the whole sample, there was an 8% sex crime charge recidivism rate, a rate that is comparable to a U.S. national study which showed a 3-year recidivism rate of 5.3%. The most important finding is simple: for all six measures (charge/conviction by 3 types of offences) there was no impact on recidivism of being registered when other factors were controlled. 

Conclusion: Once again, it has been shown that special restrictions and attempts to track sex offenders in the community are ineffective.  This is not surprising in part because recidivism rates for sex offenders are typically very low.  But in addition, most sex offenders are known to their victims before the offence; hence registration and notification logically add nothing to the ability to identify who is a risk to the community.  Sex offender registration and notification systems use “substantial resources for rigorous monitoring of all sex offenders rather than targeted and intensive supervision of those most likely to reoffend” (p. 455).  Finally, focusing on ineffective solutions to serious problems distracts policy makers from searching for more effective and more cost effective ways to reduce victimization. 

Reference: Letourneau, Elizabeth J, Jill S. Levenson, Dipankar Bandyopadhyay, Debajyoti Sinha, and Kevin S. Armstrong (2010).  Effects of South Carolina’s Sex Offender Registration and Notification on Adult Recidivism (2010).  Criminal Justice Policy Review, 21(4), 415-458.

Back to top

Victim impact statements can increase the likelihood that jurors in death penalty cases will impose the death penalty.  Victim impact evidence encourages jurors to decide on whether to impose the death penalty on the basis of their feelings for the victim and the victim’s family.

Victim impact evidence at sentencing is seen in many jurisdictions as a ‘natural’ way of ‘giving a voice’ to the victim in criminal procedures.  However, if the evidence has any impact, it logically follows that sentencers (judges, or juries in some U.S. capital cases) will be harsher with offenders whose victims (or families of victims) are seen in a more favourable light. 

A 1987 U.S. Supreme Court case forbid the use of victim impact evidence (VIE) in capital cases on the assumption that, among other things, it would divert attention from the culpability of the accused to the character and reputation of the victim. Four years later, with new justices on the court, the same court reversed itself deciding that VIE was relevant because it provided evidence of the harm that the offender had done.  The concern of the dissenters in this decision was that VIE would arouse strong emotions, anger and sympathy, in part because it would focus attention on an ‘identifiable’ victim. The concern was that an identifiable (single) victim would trigger emotional responses about the collateral harm done to the victim’s family and friends.  Said differently, the judges who dissented expressed concern that this evidence would “encourage jurors to decide in favor of death rather than life on the basis of their emotions rather than their reason” (p. 133).  

This study presents the results of an experiment – using materials from an actual capital case – in which people who were qualified to sit on a jury watched a 3.5 hour video of the penalty phase of the trial in a capital case.  For roughly half of the ‘jurors’ the video included the VIE while for the others, the VIE was edited out.  The victim (a police officer) was, of course, described in very favourable terms by his sister. In addition, the evidence provided by the victim’s sister described the impact on the victim’s daughter in vivid language that was almost certain to elicit an emotional response. The sister’s VIE did not include a specific recommendation, but the implication was clear: she asked the jury to impose “a just punishment for an unjustifiable death” (p. 144). 

63% of those who saw the VIE preferred a death sentence, compared to only 18% of those who did not see this evidence.  Those who saw the VIE were more likely to report feeling upset and hostile. In addition the VIE elicited more feelings of sympathy and empathy for the victim and victim’s family.  Even though the VIE did not include any evidence about the offender, the offender was described in more negative terms by those who were exposed to the VIE.  Not surprisingly, the VIE created more positive impressions of the victim and his family.  A multivariate analysis demonstrated that the decision to impose the death penalty was mediated in part by feelings of anger and vengefulness toward the offender, sympathy and empathy toward the victim, and favourable views of the victim and the victim’s family. 

Conclusion: In ordinary criminal cases as well as capital cases in the U.S., victim impact evidence has been deemed to be relevant to sentencing outcomes.  This study demonstrates clearly that this evidence has an effect on the outcome of jury decisions (in capital cases) and probably other cases in which juries make recommendations (e.g., second degree murder cases in Canada). More generally, however, this evidence is likely to affect the manner in which sentences are handed down by shifting the focus from the crime and the offender to the character and impact of the crime on the victim.  Though the effect of victim impact evidence on judges (or parole board members) has not been examined by this study, it would be hard to argue that judges or parole board members are not affected by the same human processes that are responsible for these effects.

Reference:  Paternoster, Ray and Jerome Deise (2011).  A Heavy Thumb on the Scale: The Effect of Victim Impact Evidence on Capital Decision Making. Criminology, 4(1), 129-161.

Back to top

The differences in violent offending rates between adolescent girls and boys decreases as neighbourhood disadvantage increases.   Violent peers in disadvantaged neighbourhoods have a larger impact on offending by girls than they do on boys’ offending.

It is well established that males commit violent offences at a higher rate than females. However, the “structural correlates of female offending closely resemble those of male offending” (p. 959).  For example, cities with high rates of economic disadvantage tend to have higher rates of serious crimes for both males and females, but “females account for a greater share of arrests in economically distressed cities” (p. 960) than they do in less economically distressed locations.  

Exposure to violent peers –  which is more likely to occur in disadvantaged neighbourhoods –  is one of the strongest correlates of violent behaviour. This paper investigates the possibility that the difference in the relative rates of male and female offending may be a result, in part, of differences in exposure to violent peers.  The differences between males’ and females’ exposure to violent peers may be less in disadvantaged neighbourhoods in which girls cannot be monitored. 

A total of 1502 girls and boys (as well as their primary caregivers) were interviewed three times when they were, on average, roughly 13, 15, and 18 years old.  These youths lived in 78 different Chicago neighbourhoods that varied dramatically in economic disadvantage (assessed in terms of the percent of neighbourhood residents below the poverty line, receiving public assistance, unemployed, etc.).   Youths were asked about the violent behaviour of their friends and, in addition, filled out self-report measures of their own violent offending.   Various other factors (e.g., victimization history, family structure) were controlled for statistically. 

Overall, males and those living in disadvantaged neighbourhoods were more likely to engage in violent behaviour.  However, the difference between male and female violent offending rates decrease as concentrated disadvantage increased.   The strength of these findings was reduced only slightly when individual characteristics (e.g., whether the respondent had been victimized, or a measure of self-control) and family structure were controlled for.  

Males and those in disadvantaged neighbourhoods were more likely to have violent friends. There was no indication that girls were especially likely to be exposed to violent peers in disadvantaged neighbourhoods.  However, “the effect of peer violence on self-reported violent offending is significantly stronger for females than for males” (p. 969).

Conclusion: It would appear that “the nature of peer influence on [violent] behaviours is not universal; rather it varies by gender” (p. 974).  Specifically, males and females are more likely to be exposed to violent peers if they live in disadvantaged neighbourhoods.  But the influence on girls of having violent friends appears to be greater than it is for boys.  These findings suggest that if communities are concerned about negative influences of violent peers on adolescents, it would be wise to focus, especially, on the impact on girls.

Reference: Zimmerman, Gregory M. and Steven F. Messner. (2010) Neighbourhood Context and the Gender Gap in Adolescent Violent Crime. American Sociological Review, 75(6), 958-980.

Back to top

Publication Type

Volume Number


Issue Number