View this issue as a PDF: CrimHighlightsV18N5.pdf
This issue of Criminological Highlights addresses the following questions:
- How have Canada’s immigrants affected the country’s crime rate?
- When trying to explain the reasons for a mass shooting, why isn’t it useful to look for evidence of ‘mental illness’ in the shooter?
- How does the crime rate of the neighbourhood in which a business is located affect the likelihood that a Black applicant will be seriously considered for a job at that business?
- How do ‘risk assessments’ affect the sentencing of poor and wealthy defendants?
- Why do school suspensions increase crime?
- Should personality assessment tools like the Psychopathy Check List be used to make crucial decisions about individuals?
- What general lessons can be learned from the failure of Canada’s “conditional sentence of imprisonment” to reduce imprisonment?
- Can those serving prison sentences for murder be released safely into the community?
When the proportion of residents of Canada’s cities who are foreign born increases, crime generally decreases and when the proportion of foreign born goes down, crime goes up.
About 21% of Canada’s total population in 2011 were foreign born. But the origins of those immigrants are changing. Prior to 1971, about three quarters were from Europe, but now more are from Asia than elsewhere in the world. This paper examines whether changes in the proportion of foreign-born people in Canada’s cities are associated with changes in crime rates.
Theoretical arguments can be cited for expecting either that increased immigration would be associated with increases in crime (e.g., immigrants are often disadvantaged or face barriers in getting jobs; the increased heterogeneity associated with increased immigration may weaken informal social control) or decreases in crime (e.g., they may revitalize poor urban areas, or they may bring with them pro-social values that protect against criminal behavior).
This paper looks at crime rates in Canada’s 32 urban areas (populations greater than 100,000) as a function of changes in immigration as measured by the Statistics Canada census at 8 points in time between 1976 and 2011. Immigrant populations are very unevenly distributed across Canadian cities. In Toronto, for example, the proportion of the population that was foreign born increased from 35% in 1976 to 45% in 2011. In contrast, in Saskatoon, between 1976 and 2001, the proportion of foreign born decreased from 14% to 7.5%.
Immigration was assessed by combining two highly related measures: the portion of the population that was born outside of Canada and the proportion with a mother tongue other than one of Canada’s two official languages – English and French. Three different measures of crime were used – total (non-traffic) crime, property crime, and violent crime. In all cases, Statistics Canada crime incident data were estimated for each metropolitan area as the average rate for the three-year period surrounding each of the census years.
Fixed-effects linear regression models were used because the focus of the study was on within-city changes in crime and immigration. This approach controls for within-city differences that are stable over time. But in addition, various controls that do change over time within cities (e.g., percent home ownership, percent low income, labour force participation) were included in the model.
The effect of changes in the main index of immigration (% foreign born and % mother tongue other than English/French) was quite consistent. Analysis of changes within cities demonstrated that after controlling for demographic and socio-economic factors, increases in immigration were associated with declines in total, property, and violent crime. Other analyses demonstrated that the overall pattern held both for cities where immigration increased and for cities in which the proportion of immigrants decreased: increased immigration was associated with decreased crime. But in addition, decreasing immigration was associated with increases in crime. These results are generally consistent with the findings from studies carried out elsewhere (see Criminological Highlights, 5(4)#6, 10(6)#7, 11(1)#4, 13(6)#7, 16(1)#2, 17(1)#3, 17(4)#1).
Conclusion: Canada’s overall crime rate in its cities dropped by an average of 52% between 1981 and 2011. In this same 30-year period, the immigration index used in this study showed an increase of 250%. The results of this study suggest that these changes are related: when immigration increased in a city, crime went down. But similarly, when the proportion of immigrants declined, crime tended to increase. Immigrants, it would seem, should be held responsible for lowering Canada’s crime rates.
Reference: Jung, Maria (2020). Immigration and Crime in Canadian Cities: A 35-Year Study. Canadian Journal of Criminology & Criminal Justice, 62(1), 71-97.
The popular explanation for mass shootings – that the assailant must have been mentally ill – not only isn’t very helpful in explaining or preventing these events, but also may undermine more productive approaches as well as harming those who already suffer the stigma of mental illness.
When mass killings occur, there is often a search for evidence that the killer was mentally ill. The US National Rifle Association, for example, suggests that these killings are caused by “mentally ill people who turn guns on innocents” (p. 86). These explanations often have a circularity to them in that the explanation is typically offered because it is assumed that nobody but a mentally ill person would do such a thing.
The proportion of violence in the US that could be attributable to mental illness is estimated to be under 5%. Yet a 2016 study reported that 38% of US newspaper stories about mental illness over a 20-year period focused on violence. The problem is that “mental illness is a highly elastic clinical term that can mean many things but is often used without definition in the mass violence narrative” (p. 87). Although people often agree, after a mass shooting episode, that there is a need for more mental health services, there is no evidence that such services would actually touch that tiny number of people most at risk for committing mass shootings. Similarly, there is no indication that laws that exist in some locations requiring ‘reporting’ of ‘dangerous patients’ (p. 87) have had any useful impact.
The difficulty in the context of mass shooters, starts from a simple problem: Some estimates would suggest that about half of the US population will meet the formal criteria for one or more mental illnesses during their lifetime. Hence, even taking “informal diagnoses of mental illness in the shadow of… shooting sprees” (p. 90) at face value, diagnoses of “mental illness” are not very useful in explaining why this person, and not the millions of others, acted the way they did. Finding out that a person could possibly have been described as mentally ill is not much better than finding out that the person could possibly have been described as male. Simply put, even though there may be a small association between mental illness and ordinary violence, “most people with serious mental illness living in the community are not violent” (p. 91).
The most important challenge, however, is that mass violence is very rare. And “the most robust risk factors for violence are shared by people with and without mental illness” (p. 93). These include being male, young, having a history of victimization and early exposure to violence, and having substance abuse problems. Mass shootings where the shooter is psychotic do occur but are rare, and, again, most of those who suffer from symptoms of psychosis do not act violently.
What if one sets these concerns aside and attributes mass shootings to mental illness? Then, it would appear that two solutions follow -- fixing the mental health system and restricting the liberty of those with identified mental illness. There are many good reasons for addressing the lack of treatment options for the mentally ill, but ‘fixing’ the system is a bit more difficult than fixing a flat tire on one’s automobile. Restricting liberty for those diagnosed with a problem is also, from a practical perspective, impossible given that the number of people who would have to be incapacitated is enormous compared to the number of mass shooters. Even restricting access to guns for those apparently suffering a mental illness is likely to be ineffective. One study suggests that it would reduce, at best, fewer than 5% of these incidents. But restricting firearms more generally may be helpful in various ways: Firearms are likely to turn a non-lethal crime into a homicide and increases in gun availability also are associated with increased gun accidents (Criminological Highlights 17(3)#2, 17(4)#3).
Conclusion: Stopping mass shootings by focusing on the mentally ill is almost certainly not an effective approach. Hence it is likely not only to stigmatize mental illness, but it discourages society from thinking more broadly about the problem. This review of the evidence concludes that “improvements in the delivery of psychiatric services to people with serious mental illness will do little to prevent common violence, gun violence, or mass violence” (p 103).
Reference: Skeem, Jennifer and Edward Mulvey (2020). What Role Does Serious Mental Illness Play in Mass Shootings, and How Should We Address It? Criminology & Public Policy, 19, 85-108.
For white job applicants, having a criminal record reduces the likelihood of being hired. Equivalent black applicants, however, have a lower likelihood of getting the job than white applicants whether they have a criminal record or not. But black applicants – whether or not they have a criminal record – are especially disadvantaged if the work establishment happens to be in a neighbourhood with an above-average rate of reported violence.
There is now a substantial amount of information suggesting that people with criminal records face special challenges in getting employment, housing, and admission to college (e.g., Criminological Highlights 6(3)#2, 18(3)#6, 15(1)#7, 17(2)#6, 18(4)#2). This paper extends this earlier research by looking at the impact of recent violence in the employer’s neighbourhood on the success of black and white job applicants who either volunteered the fact that they had a criminal record or apparently had no record.
Specifically, the paper suggests that “for employers, any kind of recent exposure to proximate violent crime amplifies fears, activates stereotypes about blacks and criminals, and draws attention to information that reinforces stereotypes about blacks and criminality” (p. 71).
The design of the study was straightforward. A total of 368 résumés of hypothetical, but realistic, male job applicants were created and sent to those advertising, on the online service Craigslist, for food, beverage and hospitality industry jobs in Oakland, California, that required only a high school diploma. The race of the applicant was manipulated by using names likely to be associated with white, black or Hispanic males. Half of each group of applicants signaled that they had a criminal record by listing work in a correctional facility as the most recent employment experience.
White and Hispanic job applicants without criminal records were most likely to be called back for job interviews or further investigation (38% and 39% respectively received callbacks). For these groups, having a criminal record reduced substantially the likelihood of a callback (18% and 20%, respectively). For black job applicants, however, the criminal record made no difference in part, perhaps, because the call-back rate was already low (16% for Black applicants without a criminal record and essentially the same, 18%, for those with a criminal record). It would seem that being black or having a criminal record had dramatic effects in lowering the likelihood of getting a job.
Violent crimes reported to the Oakland police were geocoded so that level of violence within 450 metres of the employment establishment in the 70 days prior to the job application being filed could be assessed. For black applicants, with or without a criminal record, applying for a job with an employer whose neighbourhood was exposed to above average level of violence reduced the likelihood of a callback considerably. White applicants (with or without criminal records) were slightly, but not statistically significantly, more likely to get a callback if they were applying in a high crime neighbourhood. For Hispanic applicants, there were no consistent or significant effects of the level of neighbourhood crime.
Another way of looking at these findings is that “race matters more for those employers who are exposed to higher, rather than lower, levels of violence” (p. 91). The people least likely to get callbacks were black applicants who were applying for jobs in high violence areas.
Conclusion: Race and criminal record clearly have effects on the likelihood of obtaining employment. But the effect of neighbourhood violent crime investigated in this paper – demonstrating that blacks are particularly disadvantaged when seeking employment in neighbourhoods with high violent crime rates – illustrates that the effects of local violence extend far beyond victims of crime. “This study provides new evidence of the extent to which perceptions of black men are intertwined with criminality” (p. 98).
Reference: Mobasseri, Sanaz (2019). Race, Place, and Crime: How Violent Crime Events Affect Employment Discrimination. American Journal of Sociology, 125(1), 63-104.
Providing judges with risk assessments at sentencing increases the likelihood that poor defendants will be incarcerated but reduces the likelihood of incarceration for relatively affluent defendants.
There are many empirical and policy reasons to question the appropriateness of ‘predictive’ models of sentencing (see Criminological Highlights 18(4) #3). Risk has a tendency to worm its way into the sentencing process even in proportional sentencing systems. In Canada, for example, a sentence is supposed to be proportional to the gravity of the offence and the responsibility of the offender for it, but its severity can be increased if the offender was, at the time of the offence, serving a sentence for a previous offence or separation from society is seen by the judge as ‘necessary’ (Criminal Code s.718.2(a)(vi); s. 718(c)).
One worry about risk assessments at sentencing is that they will increase socioeconomic disparities in sentencing. The research question asked in this study is straightforward: Does the presence of a risk assessment at sentencing have the same impact on the sentence imposed on wealthy and poor defendants?
Judges in three regions of the US were asked, either at annual judicial conferences or online, to participate in a sentencing study involving them giving a “sentence” to a hypothetical person described in a printed vignette. Having been assured anonymity, 85%-91% of those invited participated. In two of the regions, risk assessments are routinely presented to judges at sentencing. The vignettes varied slightly so as to include relevant local details and risk assessment information that was similar to that which the judges normally saw.
The defendant in these sentencing scenarios, having been convicted of being a participant in selling heroin, was described either as working class (e.g., a casual labourer in construction who had not graduated from high school) or as middle class (e.g., having a BA in computer science and working in the local Apple store). For half of each of these groups, the judge received no risk assessment information. The other half got detailed information about a risk assessment that had been carried out. This assessment included the factors that went into the overall risk assessment such as criminal history, family problems and antisocial associates. It also included the ‘risk’ scores that the defendant received on each dimension. The risk assessment information was identical for the relatively poor and relatively affluent defendant. The overall risk score that each hypothetical defendant received was in the middle of the risk scale. Judges were then asked to indicate the sentence that they thought was appropriate.
For middle class defendants, providing a risk assessment decreased the rate at which judges indicated they would imprison the defendant from 60% to 44%. In contrast, for the relatively poor defendants, providing the risk assessment increased the imprisonment rate from 46% to 61%. “Providing formal risk assessment information to judges reverses the relationship between a defendant’s socioeconomic status and sentencing severity…. This reversal held after controlling for judges’ jurisdictional and personal characteristics” (p. 56).
Conclusion: “Providing judges with risk assessment information transformed low socioeconomic status from a circumstance that reduced the likelihood of incarceration (perhaps by mitigating perceived blameworthiness) to a factor that increased the likelihood of incarceration (perhaps by increasing perceived risk)” (p. 56). In other words, giving judges information that encouraged them to think about risk increased the likelihood that a poor person would be incarcerated. But providing the same risk information about a middle-class person decreased the likelihood that the offender would be incarcerated. Focusing the judge’s attention on risk, then, disadvantaged the already disadvantaged defendant at sentencing.
Reference: Skeem, Jennifer, Nicholas Scurich, and John Monahan (2020). Impact of Risk Assessment on Judges’ Fairness in Sentencing Relatively Poor Defendants. Law and Human Behavior, 44(1), 51-59.
Suspending young people from school increases the likelihood that their friendship networks will change and that they will associate with antisocial peers.
Previous research (see Criminological Highlights 16(6)#1) has found that being suspended from school increases the likelihood of being involved in crime. This paper examines a possible mechanism for this finding -- that a suspended youths’ friendship networks are disrupted, and the suspended youth becomes more involved with youths who are committing offences.
School suspensions appear to be concentrated among disadvantaged and racialized youths. It is also likely that being suspended means being separated from same-grade friends. But in addition, youths who are suspended may be stigmatized (particularly in smaller communities or in local neighbourhoods) and may, as a result, be more likely to develop friendships with other youths who have engaged in some form of misbehavior.
Changes in friendships in those suspended might be a cause of the suspension (youths get involved with other youths with behaviour problems and as a result are suspended). Alternatively, changes in friendship patterns may be a consequence of the suspension from school as suspended youths find new friends who, like themselves, are likely to be troublesome.
In order to separate out these two explanations (suspensions cause changes in friendships vs. changes in friendships cause suspensions), this paper used data that followed two cohorts of youths from the beginning of Grade 6 to the end of Grade 9, with data collected 5 times (beginning and end of Grade 6, and annually to the end of Grade 9). That way it was possible to see if the change in friendship patterns preceded or followed the school suspension. In addition, it provided a natural comparison group -- equivalent youths who were not suspended.
The schools in which the research took place serviced relatively large rural areas in Iowa and Pennsylvania. At each point in time, students were asked to name their two closest friends and up to 5 other close friends in their same grade and school. Suspension data were collected from self-reports from the students and their parents for the 12 months preceding each data collection. The main outcome variable was the within-individual change in nominations of friends made by each youth from one data collection to the next. In addition, “rejections” as friends were measured as the number of nominations received in one wave that were not also received in the next wave of data collection. Substance use (frequency of smoking, drinking alcohol, getting drunk, using marijuana) and delinquency measures (frequency of involvement in such behaviour as thefts, fighting and vandalism) were obtained for all youths for the previous year. Approximately 30 control variables (many measured at each wave) were also collected.
Suspended students had an increased likelihood of discontinuing a friendship they reported having in the previous data collection period. This was particularly true of those suspended more than once. In addition, being suspended made it more likely that a youth would lose a nomination as a friend compared to the previous wave. In other words, being suspended increased the likelihood of a less stable friendship network: A suspended student was less likely than a non-suspended student to have the same friends as the previous year and less likely to be nominated as a friend in comparison to those who were not suspended. Compared to non-suspended youths, suspended youths were more likely to nominate as friends those who were involved in substance use and offending.
Conclusion: Being suspended from school appears to disrupt friendship networks. But in addition, suspended youths are more likely to choose as friends those who are involved in substance use and delinquency. Suspended youths were also more likely than those not suspended to have friends from other grades and other schools. Combined with data on other negative impacts of school suspensions (e.g., Criminological Highlights 18(3)#8), these findings suggest that the long term consequences of school suspensions should be considered by those endorsing it as a solution to troublesome behaviour.
Reference: Jacobsen, Wade C. (2020). School Punishment and Interpersonal Exclusion: Rejection, Withdrawal, and Separation from Friends. Criminology, 58, 35-69.
The Psychopathy Check List (revised) and perhaps other “validated” personality assessment tools should not be used in circumstances where precise and accurate predictions of future behaviour are required concerning individual cases.
The Psychopathy Check List – Revised (PCL-R) -- is sometimes described as the “gold standard” for assessing whether a person is likely, in the future, to engage in rule-breaking and troublemaking across settings.
Some would suggest that the PCL-R is ‘accurate’ in its assessments of almost any population of people. This paper, looking at its use for prisoners – most notably those facing the possibility of execution – suggests that it is inappropriate to use the PCL-R “to draw conclusions about an individual’s risk for committing serious violence in high-security custodial facilities” (p. 134).
The scale itself has 20 items. Standard administration involves a semi-structured interview by an expert and a review of relevant records to determine, on a three point scale, how much each characteristic (e.g., “glibness/superficial charm,” “lack of remorse of guilt,” “early behavioural problems,” p. 144) applies to an individual. Hence although it is quantitative, subjective decisions have to be made (e.g. whether, on each dimension, the item does not apply, applies to a certain extent, or applies).
The PCL-R is often described as having high levels of agreement between raters (inter-rater reliability) for offenders and forensic psychiatric patients. However, recent research carried out by those not connected to the scale’s development suggests that examiners in adversarial settings tend to disagree with each other much more than would be expected on the basis of information in the test manual and that the scores tend to ‘drift’ substantially toward the position favoured by the party that hired the expert.
More generally, when making predictions about a specific individual, the problem of variability across those experts administering the test is crucial. Equally serious is the fact that the validity of the test – i.e, whether the test scores actually predict future behaviour – appears to vary across outcomes, settings, and samples. When trying to assess whether a particular forensic patient/prisoner is likely to commit offences in the future, the scale does not appear to provide predictions that are accurate enough to warrant depending on it when crucial decisions about an individual are being made.
Conclusion: “The interrater reliability of PCL-R scores [the similarity of scores from two separate experts] in field settings, and in particular in adversarial contexts is problematically low. Second, the overall association between PCL-R scores and violence at the group level is only moderate in terms of effect size both in absolute terms and relative to the effect size of other established risk factors for violence; the association between PCL-R scores and violence in institutional settings is small … and the association between PCL-R scores and serious institutional violence is negligible” (p. 137). These problems obviously support the conclusion that these scores should not be used in cases involving important decisions about an individual. More generally, however, the analysis of the findings with respect to this particular test raise very similar questions about the reliability and validity of any personality or behaviour prediction test used in a forensic setting, not just for assessing risk for institutional violence.
Reference: DeMatteo, David, Stephen D. Hart, Kirk Heilbrun and 10 others (2020). Statement of Concerned Experts on the Use of the Hare Psychopathy Checklist-Revised in Capital Sentencing to Assess Risk for Institutional Violence. Psychology, Public Policy, and Law, 26(2), 133-144.
Giving judges another sentencing option to reduce the use of imprisonment won’t do anything unless the new option is well-designed and implemented sensibly. Canada’s “Conditional Sentence of Imprisonment” failed to reduce imprisonment because it didn’t follow these obvious rules.
In the 1980s and 1990s, Canada’s imprisonment rate was creeping upwards. In an explicit attempt to control this growth, the government introduced a new sanction: the conditional sentence of imprisonment that was supposed to be used, in some cases, as a substitute for ‘real’ imprisonment. Conditional sentences bear a striking resemblance to suspended sentences and probation. Judges were encouraged to impose this sanction in cases where imprisonment (of under 2 years) was appropriate but where having the sentence “served” in the community would not be a risk to public safety.
There were many problems with the initial legislative drafting. The law was written in such a way that it looked like the judge would first impose a sentence of imprisonment and then decide that it could be served in the community. But the law also said that imprisonment shouldn’t be used if a sensible alternative that was proportionate to the offence seriousness could be crafted. Furthermore, the legislation was written as if the non-custodial “conditional sentence” was to be of the same length as a prison sentence. Judges were to impose conditions that made it equally punitive, though it was hard to see how that was possible. The conditional sentence was often referred to as if it required “home arrest” (even by some academics) though house arrest has never been required. And there were other technical problems in the initial law. For example, if a person on a conditional sentence simply disappeared, they automatically became a free person if they were not apprehended before the end of the sentence. Not surprisingly, the sanction was seen as a ‘get out of jail free’ card. And, of course, its name didn’t help. It is called a ‘conditional sentence of imprisonment’ where people weren’t imprisoned (nor even required, in legislation, to imprison themselves in their own homes).
Though some have suggested that, notwithstanding these problems, the conditional sentence reduced the use of imprisonment, this paper suggests that various artifacts are more plausible explanations for the apparent reduction in the use of ordinary prison sentences that followed the introduction of the conditional sentence. For example, Canada experienced a large growth in the pretrial detention population at around the time when the conditional sentence became available. Sentenced imprisonment rates decreased because of an increasing number of people who spent time in pretrial detention. They were given credit for time that they had already spent in prison awaiting trial. Hence sentenced imprisonment rates decreased. More generally, many of those who received conditional sentences almost certainly would have received a suspended sentence and probation if the conditional sentence choice was not available. Said differently, the conditional sentence was more likely to be an “alternative to an alternative to imprisonment” than an alternative to imprisonment.
The contrast with Canada’s successful reduction in the rate of imprisonment of youths (Criminological Highlights 18(3)#2) is clear. Unlike Canada’s youth legislation, the conditional sentence’s legislation was ‘aspirational’ rather than prescriptive. Judges were given little legislative guidance on when it should be used. They were only told what the limits of its use were. The youth justice legislation was implemented in a manner that made its legislative intent and its operation clear. The conditional sentence of imprisonment was simply dropped into Canada’s Criminal Code. Furthermore, unlike the substantial effort expended to educate judges and others about the new youth law, there was no systematic education of judges and others in the criminal justice system on what use to make of the conditional sentence.
Conclusion: The more general lesson is that to be effective in reducing imprisonment, “there must be a demonstrable need for a new sanction, and even then, it is fundamental to identify what the specific need is and subsequently target the sanction to address it. Simply adding alternatives to incarceration is insufficient, particularly if they are crafted without sufficient guidance and training, as well as sensitivity to natural public concerns with leniency” (p. 196).
Reference: Webster, Cheryl Marie and Anthony N. Doob (2019). Missed Opportunities: A Postmortem on Canada’s Experience with the Conditional Sentence. Law and Contemporary Problems, 82(1), 163-197.
In times when many prison systems are looking for people who can be safely released on parole in order to reduce prison populations because of COVID-19, those who have already been in prison for a reasonable length of time for murder are a good bet.
The release from prison of people who have done very serious crimes is often controversial, even if they have spent a long time in prison. Hence some jurisdictions do not allow murderers to be paroled, or, they create very long parole ineligibility periods.
Ironically, those who have been convicted of murder appear to be some of the safest people to parole (see also Criminological Highlights 15(1)#2). “Available data on recidivism… demonstrates that even though there may be a small number of convicted murderers who are forever dangerous to the community, the risk of violent and homicidal recidivism is minimal for the overwhelming majority” (p. 257). The problem is a lot of publicity is directed at the very small number of cases where a person who has been convicted of a homicide offence and is then released commits another serious violent offence. Successful cases, where a person serves time in prison for murder is released and never commits another offence, receive very little publicity. This latter group, however, represents the vast majority of people convicted of murder and then released into the community.
It is important to remember that, of those convicted of homicide who are eventually released on parole, most ‘failures’ on parole involve breaches of conditions. Repeat murders are rare. In one Australian study of 1088 homicide offenders released on parole, it was found that there was only one case where the person repeated the offence while on release (and this was a person who had a long history of violence from the time he was very young up until his imprisonment for murder at age 21). These results are similar to a New Zealand study that found no repeat murderers, but that 5% were, at some point after release, reconvicted for some violent offence. A long-term follow-up of Canadian prisoners had similar results, with fewer than 1% repeat murders and a low rate (9.2%) of any type of subsequent indictable offence. A study of those prisoners released from mandatory life sentences in England & Wales found only 2.2% reoffended in any way, compared to a repeat offending rate of 46.7% of those in the general prison population.
Retrospective studies that have been carried out on those who have been convicted of a homicide offence have similar results. In Finland, for example, a study of 1,649 homicides that occurred between 1981 and 1993 found that only 36 (2.2%) were committed by those with a previous homicide offence. Notably, however, 24 of the offenders were alcoholics and 23 had a personality disorder.
US studies show the same pattern: relatively low rates of repeat violence and very small numbers of repeat murders. When capital punishment was (temporarily) found unconstitutional in the US, a number of those who were imprisoned awaiting execution were released into the community. In a follow-up of at least 5 years after release, only one of 239 people who were followed committed a subsequent homicide.
Conclusion: It is clear from studies carried out in many western countries that people in the community who once committed murder are not a large source of violent offending in any country. More generally, “the risk of any form of violent and homicidal recidivism in this distinctive group of offenders is very low” (p. 265). The argument either for “life without parole” or impossibly long parole ineligibility periods (e.g., the consecutive parole ineligibility periods now available in Canada for those convicted of two or more murders) cannot be made on the basis of sensible use of public funds for public safety.
Reference: Anderson, John L. (2019). Recidivism of paroled murderers as a factor in the utility of life imprisonment. Current Issues in Criminal Justice, 31(2), 255-268.