View this issue as a PDF: 
CrimHighlightsV4N1.pdf
The Minister of Correctional Services says that “Ontario’s first young offender boot camp [is] a success.” He is wrong. The Ministry’s own evaluation shows that boot camp “graduates” are not significantly less likely to commit new offences than are youths in standard institutions.
Background. Ontario set up a “strict discipline” boot camp approximately four years ago for young offenders aged 16-17 years old. The “evaluation” of this institution has been released and the Ministry’s press release suggested that the boot camp was responsible for a drop in recidivism from 50% (for a “comparable sample of youth who were not exposed to the program”) to 33%.
Differences between groups are typically examined to see if they are “statistically significant.” If a sample is small, or the difference is not large, what appears to be a “real” difference may be due simply to chance variation. The difference between 50% and 33% may seem large, but it would not be impressive if the 50% figure merely reflected 2 out of 4 youths recidivating and the 33% data simply represented two recidivist youths in a group of six. On the other hand, if the sample size were 300 for each of the two groups, the difference (50% vs. 33%) would be more impressive. A statistician would say that it was a “statistically significant” difference because the likelihood of getting a difference this large purely by chance is very small (e.g., less than one in a thousand, or p <.001).
This study compared boot camp graduates and a “comparison” group on two sets of dimensions: psychological changes between the beginning and the end of their custodial experience and recidivism after they were released. The comparison group was comprised of youths who met the criteria for the boot camp, but did not participate because there was no space at the time. They differed from the boot camp group on several dimensions, though it is difficult to know whether these differences were important.
The recidivism data are easy to summarize. The “comparison” group (n=60) was used as a baseline for comparisons with (a) all boot camp participants (n=59 for the highly advertised comparison), and (b) boot camp completers (n=51). This last group is problematic for the obvious reason: the “non-completers” are clearly a troubled group. They have a high rate of recidivism. Their omission from the boot camp group with no attempt being made to eliminate “failures” from the comparison group is a lethal methodological error for two reasons. First, the two groups are no longer comparable, since the “worst” kids have been excluded from the boot camp group but not from the other sub-sample. Second, we are no longer looking at the impact of the institution itself: we are looking only at the impact of the institution on a subset of youths.
For almost all recidivism comparisons, no standard statistics are presented in the report. However, they can be calculated from the data that are available. The main comparison that is highlighted by the government (p. 47 of the report) shows differences which do not even approach normal statistical significance for the contrast of the comparison group and all boot camp participants. Furthermore, when one looks at the “boot camp completers” vs. the comparison group, the difference does not approach statistical significance when the appropriate statistical test is carried out (a “corrected” chi-square test). Even when the somewhat inappropriate “uncorrected” chi-square test is performed on this inappropriate comparison, the difference is not statistically significant. In other words, using traditional conservative, common-sense statistics, there is no difference between boot-camp participants and a group that the evaluators claim to be comparable. In any normal social science report, then, the conclusion would be a simple one: looking at the comparison that the government chose to highlight, there is no reliable impact of the boot camp experience on recidivism.
The report is very thorough in its investigation of differences between subsets of boot camp graduates and the comparison group. There are over 30 sets of comparisons drawn between subsets of boot camp and non-boot-camp youths (for various periods of time). In none of these comparisons was a proper “statistically significant difference” found between the groups. When the report is quoted as saying that “recidivism rates [for boot camp graduates] were consistently lower than the rates observed for a comparable sample of youth…” (p.1 of 24 March 2001 Government of Ontario press release), what is not quoted are statements, also in the report, that point out that the results are not statistically significant. Even the inappropriate comparison emphasized by the government and described in the report as having “approached conventional levels of statistical significance (p<.10)” (p. 43) can only be stated in this way when conventionally conservative ‘corrections’ (related, in part, to the relatively small sample size) are not included in the calculations.
Psychological changes were also examined. The government claims that boot camp graduates “also showed more positive changes in behaviour, self esteem and respect for the law” (p. 1 of press release). The press release omits to report an increase in their scores on a scale measuring the tendency to lie. The “positive results”, too, have to be examined carefully. Changes in the boot camp participants are examined on approximately 57 dimensions. They did show changes on about 23 dimensions. However, these improvements could, of course, be due to such things as re-testing or simple maturation. That is why God invented comparison groups. The comparison group data, instead of being in the report itself, are in an appendix. Nonetheless, the results are summarized by the evaluators: the data suggest that while both groups “made positive gains on the majority of the measures, greater or more significant changes could not be attributed to either group” (p. 29). So much for the special benefits of the boot camp experience. Moreover, on tests of “academic competence”, “the pattern of results suggested that greater gains were made by comparison offenders than by [boot camp] offenders” (p. 29) on all but two measures.
The report notes that “[s]ome evaluators and researchers would be highly speculative of the findings that we have qualified as non-significant trends. Based on lack of statistical significance, they might dismiss any positive findings, arguing that without statistical significance there is no evidence to conclude that [the boot camp] has any impact on post-release recidivism. However, we conclude that the pattern of findings, albeit statistically non-significant, was consistently evident across three examinations of the data (2 interim reports and the current final report)” (p. 74). Statistical tests are used for good reason: they help us evaluate whether what we are examining is “real” or likely to be the result of chance variation. To say that the data are consistent across three reports is meaningless. Essentially what is being said is that as the data were collected over time, these same data (presumably with some new data added in each subsequent report) showed the same non-significant results. Finding the same thing three times on the basis of largely the same evidence does not make it more “true” than finding it only once.
Conclusion. Using traditional social science standards of evaluation, the findings from Ontario’s boot camp are easy to describe: a very thorough examination of the data found no significant differences on recidivism between boot camp participants (or boot camp completers) and a comparison group. It did not matter whether one looked at recidivism at six months, 12 months, or whatever length of time the youth had been in the community. There was also no evidence of any overall beneficial psychological or academic impact of the boot camp experience over a standard correctional institution. The generalized failure of Ontario’s boot camp to show positive effects on youth is consistent with evaluations elsewhere.
Reference: T3 Associates Training and Consulting. Project Turnaround Outcome Evaluation – Final Report. Report obtained from the Ontario Ministry of Correctional Services, April 2001.
Community notification laws appear to be more effective than they really are. Even if a community has a law requiring “community notification” of the presence of sex offenders, few, if any, crimes would be prevented.
Background. It is attractive to think that “if only we knew who the offenders living in our communities were”, we would be safe. The problem is that the creation of a registry is one issue, and the effectiveness of that registry in reducing predatory crimes is quite another. Knowing that there is a person living in the neighbourhood who has committed sex crimes may sound useful, but what one does with this information to avoid victimization is another. Additionally, the implicit underlying theory of these laws is that a centralized database would exist which would help solve sex crimes rapidly. However, there are costs. First, fear and concern may be raised that are not offset by comparable crime reduction effects. Second, there are obvious financial costs of notification projects. Finally, concerns exist that the information may be used by communities in ways that decrease the likelihood of offenders being integrated back into society.
Massachusetts law requires the state to keep a registry of those convicted of any of 11 sexual or kidnapping crimes. The law also obligates those who have served their sentence to register with local police departments. Citizens can inquire as to whether any individual is a “sex offender.” Moreover, depending on the perceived risk of reoffending, the police have an obligation to carry out various levels of community notification.
This study is “an optimistic assessment” (p. 145) of the public safety potential of the Massachusetts law. It assumes that all offenders will comply completely with the law and that the police will carry out all of the required notification. Hence, it explicitly “overestimates the law’s public safety potential” (p. 146). It looked at a sample of 136 “clinically diagnosed… habitual or compulsive” (p. 146) sex offenders. Again, these subjects provided an “overly generous assessment of the preventive potential” (p. 147) of the law since the study examines what notification would have done for this group of the “worst of the worst.”
Only 36 (27%) of these 136 offenders had a prior registry-eligible crime conviction. Said differently, 73% of these “habitual or compulsive sex offenders” could not have been in the registry before their current offence. When one looks at the sex offences that they committed (which lead to their imprisonment), only 12 of these 36 committed “predatory-stranger” crimes while the others committed offences against those known to them (e.g., close family friends, close family incest). A careful examination of these 12 “predatory-stranger crimes” suggests that in only 4 of the 12 would there have been a good chance of notification providing positive effects (and in 2 of the other cases, there was a “poor to moderate” chance). These results are due, in large part, to the fact that the other offences were committed in an area beyond the limits of which the notification would have taken place (e.g., the offender lived in a different jurisdiction from the victim). However, even in these cases, it would have depended on an aggressive and expensive notification effort on the part of the police. Furthermore, it would have depended on the victim acting effectively to avoid the crime.
Conclusion. A careful examination of the criminal histories of a sample of the worst sex offenders concludes that “the public safety potential of the… registry law to prevent stranger-predatory crimes.. is limited” (p. 154). In only 4 of 136 cases could the law have stood a good chance of avoiding the victimization and only then if great effort had been expended.
Reference: Petrosino, Anthony J. and Carolyn Petrosino. The Public Safety Potential of Megan’s Law in Massachusetts: An Assessment from a Sample of Criminal Sexual Psychopaths. Crime and Delinquency, January 1999, 45 (1), 140-158.
Does corporal punishment of children lead to antisocial behaviour? Not necessarily. It depends on the parents’ level of warmth/control (e.g., support, monitoring, explaining what the child did wrong) and on how high the level of corporal punishment actually is.
Background: Some researchers have suggested that there is convincing evidence that “children who are subjected to corporal punishment are at risk of delinquent and criminal behaviour” (p. 48). Whether one likes corporal punishment or believes, as the authors of this paper do, that “parents should strive to use less coercive forms of punishment”, it is noteworthy that surveys suggest that the majority of parents “sometimes” use corporal punishment to discipline their children. It turns out that the different parenting styles may have different impacts in different cultures. For example, there is some evidence that for African-American children living in disadvantaged neighbourhoods, corporal punishment is inversely related to delinquency. The problem is that the studies showing a positive impact of corporal punishment on delinquency have generally failed to take into account other dimensions of parenting. Inept parenting practices, for example, tend to be associated (though not invariably) with higher levels of the use of corporal punishment.
This study looks at corporal punishment in the context of other child rearing characteristics: parental warmth, parental monitoring of their children, and parental use of “inductive reasoning” with their children (e.g., explaining what was wrong with their behaviour). Two-parent families in Iowa and Taiwan were used in this study. The latter group is important because corporal punishment is used more frequently in that culture. Data were collected from both the children (grade 7) and the parents. The majority of the data came from the child’s report (use of corporal punishment, self-report delinquency, warmth and control of the child by the parents) although several measures (oppositional/defiant behaviour on the part of the child) were from the parents.
The results demonstrate the importance of looking at interrelationships among parenting measures. For both samples, the more that parents used corporal punishment, the less warm they were and the less they monitored the behaviour of their children. Looking only at the relationship between corporal punishment and conduct problems, generally speaking the more corporal punishment was used, the more conduct problems there were.
However, this effect of corporal punishment on delinquent behaviour disappeared for the boys in the Iowa sample when “oppositional-defiant” behaviour and warmth and control by the parents were taken into account. Interestingly, the effect of corporal punishment was more complex for girls. For mothers with low warmth and control, increased corporal punishment augmented the amount of delinquency. For high warmth/control mothers, increased corporal punishment tended to lead to less delinquency. The Taiwanese data for boys are somewhat different. First of all, it should be noted that the level of corporal punishment of Taiwanese boys by their fathers is much higher than for Taiwanese girls as well as for either boys or girls in the Iowa sample. It appears that very high levels of corporal punishment lead to delinquency in children, regardless of parenting styles.
Conclusion: Typically, the use of corporal punishment at low to moderate rates is not associated with delinquency once other parenting style variables (e.g., warmth, control) are taken into account. Hence, one may conclude that in terms of its impact on delinquency, the levels of corporal punishment generally used in our culture are not likely to be a cause of delinquent behaviour. However, very high levels of corporal punishment are associated with delinquent behaviour, regardless of parenting practices.
Reference: Simons, Ronald L., Chyi-In Wu, Kuei-Hsiu Lin, Leslie Gordon, and Rand D. Conger. A Cross-Cultural Examination of the Link between Corporal Punishment and Adolescent Antisocial Behaviour. Criminology, 2000, 38(1), 47-79.
If “young offender” aged youths understand right from wrong, comprehend that their behaviour is governed by the same rules as everyone else, and are aware of the risks that they take, why do we deal with them under different criminal laws than those governing adults? One reason is that they are deficient in their ability to make “mature decisions.” This skill increases throughout adolescence until at least age 18, the age at which youths are typically deemed to be adults.
Background. In many jurisdictions, youth justice systems are under pressure to lower age boundaries – both at the lower end (making very young children “criminally responsible”) and at the upper end (legislatively turning adolescents into adults). In Canada, for example, we will soon be “presumptively” sentencing youths as adults as long as they have committed certain serious offences and are 14 years or older. The research in this area does not support the notion that older adolescents are unaware of the risks that they take or lack logical abilities.
This paper examines the developmental changes through adolescence (into adulthood) on antisocial judgements. Using a sample of 810 adolescents in grades 8, 10, and 12 from the same school district, respondents were given tests that measured the following:
- Responsibility (e.g., respondent’s agreement with questions such as “luck decides most things that happen to me”)
- Perspective (e.g., “I often do things that don’t pay off right away but will help in the long run” or “Before I do something, I think about how it will affect the people around me”)
- Temperance (e.g., “I do things without giving them enough thought” or “I lose my temper and ‘let people have it’ when I’m angry”)
- “Psychosocial maturity” (a combination of the above scales)
- Antisocial decision-making (hypothetical situations were described in which participants had to indicate how they would behave – an anti-social choice and a socially acceptable choice were given).
The results are easy to describe. As psychosocial maturity increases, decision-making becomes less antisocial. Age is related to psychosocial maturity – older youths are more mature than younger ones. (From other data, it appears likely that psychosocial maturity does not increase significantly after approximately age 19). However, age does not add to the predictability of the degree of anti-social attitudes above and beyond psychosocial maturity. In fact, within each age group, the more mature the youth, the less likely that he or she reported antisocial decisions.
Conclusion. It would appear that throughout adolescence, youths are becoming more mature on such dimensions as the ability to have perspective in what they are doing, control their impulses, and see that they, themselves, are responsible for what happens in their lives. These are general characteristics and are not tied to specific types of anti-social behaviours. Hence, policies which treat certain young people as adults (e.g., those charged with certain offences) as well as policies that treat youths as adult for some purposes (e.g., crime) and not others (e.g., drinking) appear to make no sense from a developmental perspective.
Reference: Cauffman, Elizabeth and Laurence Steinberg. (Im)maturity of Judgement in Adolescence: Why Adolescents may be less Culpable than Adults. Behavioral Sciences and the Law, 2000, 18, 731-760.
Ask Canadians sensible questions about sentencing and they give sensible, measured answers. Canadians do not really expect sentencing judges to keep them safe. They do, however, want their political leaders and judges to use resources sensibly. Sensible sentencing appears to be more important to Canadians than “harsh” or “lenient” sentencing.
Background. Every public opinion poll carried out in the past 35 years that has asked Canadians whether sentences are sufficiently severe has found “discontent” with sentencing: a majority of Canadians always say that sentences in criminal courts are not harsh enough. The irony of this answer is, of course, that almost nobody knows what the sentences in Canada actually are. Only recently have we had any systematic information about sentencing patterns in criminal courts. Part of the problem is that most Canadians think that sentencing can accomplish a great deal: deterrence is seen by most Canadians as being an important purpose of sentencing, notwithstanding the evidence which shows that variation in sentencing practices does not have a significant impact on crime levels.
This study looks at an Ontario public attitudes survey in which respondents were asked questions that focused largely on adult and youth crime issues. For both adults and youth, non-punitive approaches (increasing the availability of social programs, addressing unemployment, increased use of non-prison sanctions) were seen as being better strategies for controlling crime than making sentences harsher. In fact, in addressing both youth and adult crime, most Canadians would prefer to invest in prevention or non-prison sanctions rather than pay the cost of a harsher sentencing structure (more prisons).
Moreover, when Canadians appear harsh, one of the reasons may be that they have not thought about the consequences of their harshness. This same survey found that by reminding Canadians that an offender would, if imprisoned, be released after a few months, prison became a less attractive sentence. Similarly, when Canadians are told the cost of imprisonment, the preferred sanction shifts somewhat away from imprisonment.
Harsh sentences (typically involving prison) appear to people, at first blush, to be attractive for a number of reasons. First, they appear to promise something – incapacitation and punishment, at a minimum. In contrast, community sanctions (e.g., community service orders) are viewed by many Canadians with much skepticism. Over 60% of Canadians think that half or fewer community service orders for adults or youth are actually carried out.
Conclusion. Canadians appear to want a “response” to wrongdoing by adults and youth. It need not involve imprisonment. In fact, a focus on the fact that the offender will soon be in the community makes prison less attractive. However, the sanction must be seen as being carried out. Therefore, it is not surprising that – at least for minor offences – family group conferences are seen as more sensible responses to offending: such “accountability” sessions have the elements that are important to the public. Perhaps what is needed, then, are policies that respond to the public, rather than pander to it.
Reference: Doob, Anthony N. Transforming the Punishment Environment: Understanding Public Views of What Should be Accomplished at Sentencing. Canadian Journal of Criminology, 2000, 42, 323-340.
Cases that result in a sentence of death in the U.S. are riddled with errors. A study of 4578 cases from 1973 to 1995 shows that in 68% of them, courts found serious reversible errors.
Background. Traditionally, concern about the correctness of verdicts in capital cases has focused on single cases. This study looks more generally at the process by which a person is sentenced to death and helps to understand, among other things, why it appears to take so long for cases to go from trial to execution. Two sets of political activities are now happening. On the one hand, moratoriums on executions are being considered or have been declared. On the other hand, attempts have been made to speed up the process of killing offenders.
This study looked at the existing court records of all cases resulting in sentences of death in the U.S. between 1973 and 1995. Courts at various levels can find errors. After the trial court imposes sentences of death, a state high court and/or state post-conviction courts can review the case. Other reviews (federal habeas corpus petitions) are also possible. This study defined a serious error as something that “substantially undermines the reliability of the guilt finding or death sentence imposed at trial” (p. 5).
The results of this study describe a system that is very prone to error:
Of the 5760 death sentences imposed in 1973-1997, only 5.4% resulted in execution during this period.
The overall error rate (proportion of cases overturned at some stage of the appeals process) was 68%.
The most common cause of errors was “egregiously incompetent defence lawyering” (p. 5). The next most common cause was prosecutorial suppression of evidence.
Some of these cases went back for retrial during the period of the study. Of the 301 that were fully retried, 82% (247 cases) resulted in a sentence less than death, and in 7% of the cases (22 cases), the defendant was found not guilty of the capital offence.
The states with high error rates in capital cases are highly dispersed. Illinois, which declared a moratorium on executions, has an error rate slightly lower than the national average.
The errors are found at each stage of the (typically) three stage appeals process, suggesting that additional inspections of these cases might lead to additional errors being found.
The average time between the imposition and the execution of the sentence (for those cases resulting in actual execution) was nine years. The reason that the number of executions has been increasing in recent years is simply because more people have finally made it through the review process. It is still the case that tiny proportions of people are being executed, in part because of this high error rate.
Since the study period ended in 1995, little seems to have changed. Few executions have taken place and errors still appear to be occurring.
Conclusion. These results provide strong evidence that the court system that imposes death sentences in the U.S. also finds, when given a chance, a substantial number of errors. These errors are widespread geographically and appear to be due to serious problems that can affect any case (incompetent defence lawyers and inappropriate prosecutorial behaviour).
Reference: Liebman, James, Jeffrey Fagan, and Valerie West. A Broken System: Error Rates in Capital Cases, 1973-1995. Columbia University School of Law, 2000.
Men’s use of alcohol and wife assault are related, but alcohol may not be the cause. It may be that attitudes about the correctness of male degradation and control of women which are associated with male drinking is the real cause.
Background. It is well established that alcohol use is associated with wife assault. Offenders appear to be more likely to be drinking at the time of the assault than one might expect. Heavy and binge drinkers are more likely to assault their wives than non-drinkers. Therefore, the question is not whether there is a link, but what the link means. Various theories have been suggested, including the disinhibition effect of alcohol. Alternatively, heavy alcohol use may reflect more basic pathologies which are directly important as causes of wife abuse. Finally, there are a set of explanations that suggest that in certain groups, maleness is associated with heavy drinking and social domination of women.
This study examines the link between alcohol and violence against wives using the 1993 Statistics Canada Violence Against Women survey. Violence was defined by the responses to a series of questions asking women whether their spouses had physically or sexually assaulted them. Alcohol abuse and attitudes and beliefs toward male control and devaluation of female partners were obtained from the respondents.
Consistent with previous research, the results showed that the more the male spouse drank, the higher the likelihood his wife had been assaulted in the previous year. Moreover, men who were usually drinking at the time of the assault were more likely to use very serious acts of violence against their wives. Demographic variables also had an independent impact. The following characteristics of the men were associated with increased violence against their wives: being less than 45 years old, having less education, being unemployed (long or short term), and living common law or being together for less than 3 years.
Male dominance attitudes were measured by such things as the woman’s reporting that he used names and put-downs, that he was jealous and kept her away from other men, and that he limited contacts with others. These are positively associated with wife assault. The most interesting finding is that the male’s heavy drinking has no impact on the likelihood of wife assault above and beyond these attitudes.
Conclusion: It appears that attitudes that support control and dominance of women are associated with heavy drinking and that it is these attitudes which may be more important than alcohol use in the understanding of wife assault. If this is correct, “interventions that focus on alcohol treatment [of male batterers] in the belief that addressing alcohol abuse will at the same time reduce violence may be misguided” (p. 70). These programs may be useful, but “the results of this study suggest that unless interventions also aim to address the negative effects of unemployment and [to] alter male attitudes and beliefs in the rightness of male dominance and control over women, they are unlikely to be successful” (pp.70-71).
Reference: Johnson, Holly. Contrasting Views of the Role of Alcohol in Cases of Wife Assault. Journal of Interpersonal Violence, January 2001, 16 (1), 54-72.
When is a group of youths a gang? It turns out that different definitions include somewhat different types of youths. When very restrictive definitions are used, one is most likely to include those youths with the most antisocial attitudes and behaviours.
Background. There is no consensus on what constitutes a youth “gang” or how one defines whether a given youth is a member of a gang. “Experts” do not appear to agree as to what is the best single definition. Hence, it is easy to get vastly different estimates of the prevalence of gangs (or changes in the prevalence) simply because there is no consensus regarding what constitutes membership in a gang. Some “gang” researchers suggest that a gang is a group of youths who are seen as a distinct group, recognize themselves as a group (usually with a name), and have been involved in a great enough number of anti-social acts that people see them in a negative way. The problem is that the first two criteria would fit the Boy and Girl Scouts or university fraternities. The third might also fit university fraternities. Aside from anything else, that which constitutes “gang” behaviour is also associated with other “memberships” (e.g., class, ethnicity, neighbourhood).
This paper looks at data from questionnaires given to 5935 grade 8 students in 42 schools (315 classrooms) in 11 American cities. It looks at the relationships of 5 different definitions of gang membership to attitudes and behaviours. Different definitions obviously are important:
- 17% report ever being in a gang;
- 9% report currently being in a gang;
- 8% report being in a gang that does delinquent things;
- 5% report being in “organized” gangs that have leaders, symbols, initiation rights;
- 2% define themselves as being “core” members of organized gangs.
Gang members, using any of these definitions, tended to be more likely than non-gang members to be male, racial or ethnic minorities, from single parent homes, and have parents without a high school education. However, the more restrictive definitions did not result in higher proportions of people who had these characteristics (e.g., the portion of youths in gangs (defined in any of the five ways listed above) who came from single parent homes was roughly the same (40-41%) across all five definitions).
However, demographic variables become relatively less important in predicting gang membership as the definition of gang becomes more restrictive. Having delinquent peers is important no matter which definition one uses. In addition, for all definitions, those who indicate that they would not feel guilty if they engaged in certain delinquent activities and those who claim that fighting is an appropriate response in specific situations are also common factors. Those youths who no longer claim gang status (but once did) are “substantially more pro-social in both attitudes and behaviour than are those persisting in their membership” (p. 124).
Conclusion. Those youths claiming gang membership are likely to show anti-social attitudes and behaviour but that which constitutes “gang membership” varies according to the definition that one uses. As one moves to more restrictive definitions, “the theoretical predictors from social learning theory (especially association with delinquent peers, perceptions of guilt, and neutralizations for fighting) supercede the importance of demographic characteristics” (p. 124).
Reference: Esbensen, Finn-Aage, L. T. Winfree, Jr., Ni He, T. J. Taylor. Youth Gangs and Definitional Issues: When is a Gang a Gang and Why Does it Matter? Crime & Delinquency, 2001, 47 (1), 105-130.
Publication Type
- Regular Issue