Correctional Service Canada
Symbol of the Government of Canada

Common menu bar links

Research Reports

Warning This Web page has been archived on the Web.

High-Risk Violent Offenders in Canada


James Bonta
Policy Branch
Solicitor General Canada

Laurence L. Motiuk
Research Division
Correctional Service Canada

A paper presented at the 104th Annual Convention
of the American Psychological Association
in Toronto, Canada, 1996.

August, 1996

Executive Summary

Canada has both Dangerous Offender (DO) legislation and provisions which permit detaining high-risk violent inmates until expiry of sentence. These laws are based, in part, on the premise that violence can be reliably predicted. However, neither law requires the use of actuarial risk assessment. The files of designated Dangerous Offenders and detained offenders were reviewed in order to evaluate how well high-risk violent offenders were being identified. Furthermore, a follow-up of detained inmates who recidivated violently permitted the profiling of some very high-risk, violent offenders. The results showed that both designated Dangerous Offenders and detention cases are indeed, high-risk offenders. Although some empirically-based instruments could be useful in assessing the potential for violence, further research is needed to improve the predictive accuracy of such instruments.


Interest in high-risk violent offenders draws the attention of the public, politicians, correctional agencies and researchers. Understanding the nature of violent offending and managing violent offenders is now a major public priority. In order to minimize the harm posed to the public by some offenders, jurisdictions throughout North America have enacted various laws that attempt to incapacitate high-risk, violent offenders. These legislative initiatives include lengthy periods of determinate sentences, indeterminate sentences, and civil commitments. One of the assumptions underlying many of these laws is that individuals who have the potential to seriously harm others can be reliably identified. How well we can target high-risk violent offenders who require extraordinary restrictions of liberty is the focus of the present research.

Canadian Legislation and Violent Offenders

Canada has a number of pieces of legislation intended to deal with offenders who pose a serious risk of re-offending in a violent manner. The most widely known legislation is the Dangerous Offender provisions of the Criminal Code of Canada (1977). The Dangerous Offender provisions developed from the earlier Dangerous Sexual Offender law (1960) which provided indeterminate sentences for certain sexual offenders. The Dangerous Sexual Offender law was criticized for being too restrictive and subsequently, it was broadened to allow for the inclusion of violent, non-sexual offenders.

To meet the criteria for a Dangerous Offender (DO) designation, the offender must have committed a "serious personal injury offence" (such as sexual assault, manslaughter, aggravated assault). Murder is not included since a conviction results in an automatic life sentence. In addition to having committed a serious personal injury offence, evidence must be brought to show any of the following: a) repetitive and persistent behaviour that is likely to lead to injury or death, b) likelihood of injury through a failure to control sexual impulses, or c) the act is so "brutal" that one can only conclude that it is unlikely the offender can inhibit his/her behaviour in the future. All the criteria imply an ability to predict future behaviour. The application for a DO designation occurs after conviction and forms part of the sentencing hearing.

Another important piece of legislation deals with offenders who are already incarcerated and serving determinate sentences. In Canada, federal inmates (i.e., those serving sentences of two years or more) are released prior to the end of their sentence either through full parole (eligibility is usually after one-third of sentence) or through statutory release (at two-thirds of sentence completion). The automatic nature of statutory release raised concerns when certain inmates were seen as posing a significant risk for violence and were to be released.

In response to situations involving the release of high-risk, violent offenders, the Corrections and Conditional Release Act includes detention provisions. An inmate who has committed a violent offence and for whom there are reasonable grounds to believe that the offender would commit another violent offence prior to sentence completion could be detained in prison until sentence expiration. Statutory release would be denied and this decision is made by the National Parole Board.

A common theme in both pieces of preventative detention legislation described is that violent behaviour can be predicted. Research has shown however, that predicting violent behaviour is an extremely difficult task (Monahan, 1981). Violent crime typically shows low base rates making prediction particularly challenging (Quinsey, 1980). Clinical judgements of dangerousness have been particularly notorious for their conservatism in identifying many as dangerous who subsequently do not behave in a violent manner (false positives). Actuarial prediction methods have generated more accurate predictions, but still produce significant error rates.

The Canadian preventative detention laws do not require the use of actuarial assessments and allow clinical judgements of dangerousness. In the DO provisions, the testimony of two psychiatrists or psychologists is required. The law is silent on whether the testimony must include empirically validated methods of predicting violent behaviour. The same is true with the detention provisions. The only empirically-based requirement is the commission of a violent offence. As DOs and detained inmates represent individuals judged as the most high-risk violent offenders in the country an analysis of their characteristics would contribute to a better understanding of violence and our ability to predict violent recidivism.


The data presented in this study were drawn from the federal penitentiary system. In Canada, jurisdiction over offenders depends upon the length of sentence. Offenders with sentences of two years or more are a federal responsibility. As a result, the most serious offenders are typically found in federal penitentiaries.


Two categories of federal offenders were selected for analysis. The first category consisted of 262 detained inmates for whom there was a minimum follow-up period of one year after their release (Motiuk, Belcourt & Bonta, 1995). This group of inmates also provided a subsample of 34 inmates who recidivated violently after release which was sued as a comparison group for our second category of subjects. The second category consisted of 64 designated Dangerous Offenders (DOs) from the provinces of Ontario and British Columbia (Bonta, Harris, Zinger & Carriere, 1996). At the time when the data were collected there were 146 DOs in Canada.


Computerized files and penitentiary records of DOs and detention cases were reviewed. The detention cases formed two groups: a) 148 inmates where the decision to detain was later rescinded and the offenders were released prior to sentence expiry ("lifted"), and b) 114 inmates detained until the end of their sentence. A comparison of inmates who were detained until sentence expiration and the offenders judged no longer a threat and for whom the detention order was lifted permitted an evaluation of the assessment of dangerousness by correctional staff. Violent recidivism was based upon criminal convictions reported in the RCMP's criminal history records.

Penitentiary and court records of 64 DOs from British Columbia and Ontario were reviewed. This represented 85% of all DOs in British Columbia and approximately half of the DOs in Ontario. The files provided information on a variety of measures ranging from criminal history, the incidence of psychopathy (as measured by the PCL-R; Hare, 1991) and an actuarial measure of risk used by the Correctional Service of Canada. This risk scale, the Statistical Information on Recidivism (SIR), has been empirically validated (Bonta, Hann, Harman, & Cormier, 1996; Nuffield, 1982).

Finally, from the follow-up of detention inmates, we formed a group of 34 offenders who recidivated violently. These offenders were demonstrably high-risk violent offenders. They were assessed in prison as likely to commit a further violent offence and they did. The DOs were then compared to the detained offenders who recidivated violently (referred to as detention failures or DFs). If the courts are accurate in identifying high-risk violent offenders, then the DOs should share similar characteristics with the DFs.


The Detained Inmates

Nearly half (47.3%) of the detained inmates had a sexual offense as their admitting offense. Despite the fact that all the offenders had committed a violent offense, their sentence lengths were relatively short compared to the general penitentiary population. Approximately two-thirds of the detained offenders in this sample had sentences less than four years whereas 39% of the general penitentiary population had sentences under four years. Case characteristics of detained inmates who had their detention order lifted (n = 148) and those held until completion of their sentence (n = 114) are shown in Table 1. Inmates held until sentence expiration were older, more likely to be Caucasian and to have committed a sexual offence.

Table 1. Case Characteristics by Detention Groupings Detention Group

Variable Lifted Fully Detained t/c2 All
Age at admission (yrs) 30.2 33.4* -2.5 31.6
Race (%) Caucasian 62.2 74.6* 7.7 69.9
>Race (%) Aboriginal 27.7 25.4   26.7
Race (%) Other/td> 6.1 0.0   3.4
Sexual Offence (%) 41.2 55.3* 5.1 47.3
Prior federal term (%) 48.0 48.3 ns 48.1

Note: * p < .05; ns = nonsignificant

Table 2 presents the distribution of detained offenders across risk categories as measured by the SIR. The average SIR score was -2.9. Detained offenders who had their order lifted scored higher risk (-4.2, SD = 8.0) than fully detained inmates (-1.3, SD = 7.5; t = -3.8, p < .01) and twice as many offenders from the lifted group were assigned to the Very Poor risk category (chi-square = 16.9, p < .01).

Table 2. Distribution of SIR Risk Categories by Detention Group

Risk Category Lifted Fully Detained
Very Good 13.6% 14.2%
Good 10.9% 26.6%
Fair 23.8% 25.7%
Poor 20.4% 19.5%
Very Poor 31.3% 14.2%

All the detention cases had a minimum follow-up of one year with the average follow-up being 4.1 years. Two cases, one from each group, could not be traced during the follow-up. Within the time frame of the follow-up period, 40.4% had committed a new violent offence. The violent recidivism rate appeared higher for the "lifted" group than it was for the fully detained offenders, but the differences only approach statistical significance (44.9% vs 34.5%; chi-square = 2.86, p < .09). A partial correlation analysis, controlling for the amount of time detained, was unrelated to recidivism (r = -.07, ns).

Dangerous Offenders

The 64 DOs were compared to 34 detained inmates incarcerated in Ontario penitentiaries who recidivated violently. The same file coding rules were applied to both groups. Table 3 summarizes some of the findings.

Table 3. Comparison of Dangerous Offenders (DOs) and Detention Failures (DFs)

Characteristic DO(n) DF(n) t/c2 p
IQ 94.9 (56) 88.2 (25) 1.89 ns
age 34.4 (56 26.7 (32) 4.01 .001
grade 8.5 (62) 8.0 (33) 1.04 ns
% single 48.4 (62) 64.7 (34) 2.36 ns
% unemployed 63.3 (49) 72.4 (29) 0.69 ns
Index Offence:
% sexual offence 92.2 (64) 35.3 (34) 36.01 .001
% female victim1 86.2 (58) 91.7 (12) 0.21 ns
# of victims1 3.2 (58) 1.5 (12) 3.07 .01
% victim under age 161 58.8 (51) 41.7 (12 1.16 ns
% any brutality 69.6 (56) 48.4 (31) 3.82 .05
% weapon used 50.0 (60 64.7 (34) 1.90 ns
% under influence 46.3 (41) 76.7 (30) 6.59 .01
Criminal History:
% juvenile record 75.0 (56) 70.0 (30) 0.25 ns
% prior incarceration 88.5 (61) 96.4 (28) 1.47 ns
% probation/parole failure 73.0 (63) 97.1 (34) 8.45 .01
% prior assault 45.9 (61) 53.6 (28) 0.45 ns
Actuarial Assessment:
SIR -1.6 (62) -5.3 (28) 1.85 ns
PCL-R 27.7 (48) 27.0 (34) 0.50 ns

Note: numbers vary due to insufficient information

1 Sex offenders only

Scores on the SIR ranged from -18 to 24 (lower scores indicate higher risk). The two groups showed no statistical difference in mean SIR scores. A surprisingly high percentage of DOs (43.6%) scored within the Good and Very Good risk categories. Only 25% of the DFs scored in this range.

An examination of the sexual histories of the DOs and DFs who were convicted of a sexual crime (n=12) revealed some similarities and differences. Both groups showed early sexual offending behaviours. All of the DFs and 96.6% of the DOs had forced sexual activity on a victim prior to the age of 16. In terms of differences, the DOs reported many more sexual offenses for which they were not charged compared to the DFs (27.2 vs 0.82, t = 2.61, p < .01). In addition, the DOs had more adult female victims (2.8 vs 1.2, t = 2.77, p < .05) and more female child victims (2.6 vs 0.9, t = 2.80, p < .01). This may however, be an artifact of age as the DOs were older than DFs.

Evidence of an antisocial personality disorder was measured in three ways: 1) a clinical diagnosis based upon professional judgement, 2) a diagnosis based upon DSM-IV criteria (American Psychiatric Association, 1994) and 3) an assessment of psychopathy using the Psychopathy Checklist (PCL-R; Hare, 1991). Regardless of how antisocial personality was assessed, no differences were found between the DOs and the DFs. A psychiatric diagnosis of antisocial personality was reported in 72.9% of DO files and 73.1% of DF files. DSM-IV criteria found 54% of DOs and 64.7% of DFs meeting the diagnostic criteria (c2 = 1.04, df = 1, ns). There were no differences in the average score on the PCL-R between the DOs and the DFs. Using a cut-off score of 30 as recommended by Hare (1991), 39.6% of the DOs were classified as psychopaths compared to 32.4% of the DFs (chi-square = .45, df =1, ns).


The results from the analysis of detention cases and court designated DOs leads to a number of conclusions. First, the courts and correctional agencies tend to equate high-risk violent offenders with sex offenders. Those who are detained in prison until the completion of their sentence tend to be sex offenders (55.3%). Those who are designated as Dangerous Offenders are overwhelmingly sex offenders (92.2%). Non- sexual, but potentially violent offenders are less likely than sexually violent offenders to receive the restrictions on their liberty that are available through detention and Dangerous Offender provisions. Even though the laws provide special confinement conditions for non-sexual violent offenders, they appear not to be used to the extent possible.

Both categories of offenders studied in this project, the detention cases and DOs, were high-risk violent offenders. The violent recidivism rate for the detention cases was 40.4%. Bonta and Hanson (1995) found a violent recidivism rate of 48.6% for a sample of over 3,000 federal offenders. However, the follow-up period in the Bonta and Hanson (1995) study was 11 years. Thus, the detention cases almost match the level of violence of general penitentiary inmates but in a much shorter time frame.

For the DOs it is impossible to measure their violent recidivism rate since they are not released. However, by comparing them to the DFs, an estimate of their "dangerousness" could be made. We found significant similarities between the two groups in their index offence, criminal histories, objective measure of risk and incidence of an antisocial personality. These similarities lead us to conclude that the offenders designated by the courts are indeed high-risk violent offenders.

Both the detention process and the procedures in a DO application involve subjective assessments of risk. Actuarial assessments may form part of the clinical judgement but they are not obligatory. The mean risk scores, as measured by the SIR were -2.6 for the detained inmates and -1.6 for DOs. For the general inmate population, the average is +2.6. Empirically, both groups of offenders are high-risk as measured by the SIR. Yet, there was considerable variability within the SIR risk categories. Significant numbers of DOs and detention cases fell into the Good and Very Good categories. Obviously, other factors besides the SIR scale, appear to influence release decisions. Recall also from the detention cases that those offenders who had their detention order lifted scored higher on the SIR than those who were detained until the end of their sentences.

Although empirically based offender risk scales add to risk assessment, solely relying on any one scale such as the SIR, is insufficient for release decisions. Rather a combination of factors and empirically-based assessments appear reasonable in making such decisions. Recently, there has been renewed interest in the role of antisocial personality and psychopathy in the assessment of dangerousness. In particular, scores on the Psychopathy Checklist (PCL-R; Hare, 1991) have been shown to be associated with violent recidivism (e.g., Harris, Rice & Cormier, 1991). Using the PCL-R to assess psychopathy, approximately 40% of DOs and 32% of DF could be classified as psychopaths. This finding also shows, however, that the majority of high- risk violent would not be identified by the PCL-R.

As many high-risk violent offenders are sex offenders, more attention to the predictors of sexual recidivism is needed. Some predictors are specific to sexual recidivism and these should be included in the assessment of potentially dangerous offenders. For example, the research literature shows that early sex offending is an important predictor of sexual recidivism (Hanson & Bussière, 1996). This was evident in our own findings where we found that early forcible sexual behaviour was evident in the DF and DO groups. Encompassing a variety of factors, from different domains, appears to be a fruitful strategy in the assessment of violence.


American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed), DSM-IV. Washington: American Psychiatric Association.

Bonta, J., & Hanson, R. K. (1995). Violent recidivism of men released from prison. Paper presented at the 103rd Annual Convention of the American Psychological Association, New York.

Bonta, J., Hann, R. G., Harman, W. G., & Cormier, R. B. (1996). The prediction of recidivism among federally sentenced offenders: A re-validation of the SIR scale. Canadian Journal of Criminology, 38, 61-79.

Bonta, J., Harris, A., Zinger, I., & Carriere, D. (1996). The Crown Files Research Project: A study of Dangerous Offenders. Ottawa: Solicitor General Canada

Hanson, R. K., & Bussière, M. (1996). Predictors of sexual recidivism: A meta-analysis. User Report #1996-04. Ottawa: Solicitor General Canada.

Hare, R. D. (1991). The Hare Psychopathy Checklist - Revised. Toronto: Multi-Health Systems.

Harris, G. T., Rice, M. E., & Cormier, C. A. (1991). Psychopathy and violent recidivism. Law and Human Behaviour, 15, 625-637.

Monahan, J. (1981). Predicting violent behavior: An assessment of clinical techniques. Beverly Hills, CA: Sage.

Motiuk, L. L., Belcourt, R., & Bonta, J. (1995). Managing high-risk offenders: A post-detention follow-up. Research Report R-39. Ottawa: Correctional Service Canada.

Nuffield, J. (1982). Parole decision-making in Canada. Ottawa: Communication Division, Solicitor General Canada.

Quinsey, V. L. (1980). The base rate problem and the prediction of dangerousness: A re-appraisal. Journal of Psychiatry and Law,8, 329-340.