The National Flagging System: Identifying and responding to high-risk, violent offenders 2005-04

PDF Version (241 KB)

Acknowledgements

This research project began with the support of Carol Snell who was Chair of the National Flagging System coordinators at the time that the original data was collected. During the data collection phase, six coordinators provided their time and support to give us access to their files. We would like to thank Ron Hurt, Bob Mettlewsky, Dean Sinclair, Gregg Lawlor, Michael Phillips, Wayne Gorman and Loraine Minish-Cooper (deceased). Finally, we would like to thank Mia Dauvergne for conducting many of the initial file reviews and Dean Sinclair and Cliff Yumanski for reading an earlier draft of this report.

The opinions expressed are those of the authors and do not necessarily reflect those of Public Safety and Emergency Preparedness Canada.

James Bonta and Annie K. Yessine
Public Safety and Emergency Preparedness Canada

Correspondence concerning this report should be addressed to:

James Bonta, Corrections Research,
Public Safety Canada,
340 Laurier Avenue West, Ottawa, Ontario,
K1A 0P8; email: jim.bonta@ps-sp.gc.ca.

Executive Summary

The National Flagging System (NFS) was established to track high-risk, violent offenders. In general, offenders who are judged to be suitable candidates for a Dangerous Offender (DO) and/or Long Term Offender (LTO) application should they reoffend are placed on the national CPIC system. Thus, if they do reoffend, authorities are quickly alerted to their status and steps taken to consider a DO/LTO application.

The present research was aimed at empirically investigating the effectiveness of this policy initiative in identifying and responding to potentially dangerous offenders. Specifically, the profile of 256 flagged male offenders was compared with the profile of 97 known high-risk, violent offenders (i.e., Dangerous Offenders and Detention Failures). In addition, the reconviction rates of the flagged offenders were examined with an emphasis on the effect of the NFS in prompting DO and LTO applications.

Results demonstrated that, with a few exceptions, although the flagged offenders showed less serious and persistent criminality characteristics than the known high-risk offenders, both groups were comprised of relatively high-risk cases, compared to general offender populations. Furthermore, the violent/sexual recidivism rates of the flagged offenders were much higher than those reported among typical Canadian male federal offender populations. Even among the violent (including sexual) recidivist offenders, however, the rates of DO and/or LTO applications and successful designations were low.

Taken as a whole, despite the fact that the NFS appeared successful in appropriately identifying offenders who pose a risk to the community, this policy initiative's role in facilitating the use of DO and/or LTO provisions was less than expected. Recommendations for the development of guidelines to assist criminal justice professionals in screening, monitoring and processing dangerous offenders are made.

The National Flagging System: Identifying and Responding to High-Risk, Violent Offenders

Over the past decades, violent and sexual offences have elicited a great deal of community fear and revulsion. It is therefore not surprising to see that governments in North America and elsewhere in the world have proposed a variety of legislative policies and program reforms directed towards individuals convicted of such crimes. Responses aimed at minimizing the danger posed by high-risk, persistent offenders and alleviating public concern have taken numerous forms, including criminal law and sentencing reforms (e.g., civil commitment of violent offenders), treatment services, prohibitions as well as other preventative measures such as employment screening procedures, community notification and public registries (e.g., sex offender registration (SOR) programs, child abuse registries). Throughout the years, although sometimes aimed at violent offenders in general, most jurisdictions have adopted methods specifically targeted towards sex offenders, and even more particularly sex offenders who victimize children.

An examination of dangerousness policies and practices adopted in the United States , the United Kingdom and Australia demonstrates a profound shift towards more rigorous alternatives to identify and respond to dangerous offenders. With the introduction of community notification protocols as well as sex offender and child abuse registries, Canada seemed to have followed the international trend towards tighter responses and more severe penalties for violent and/or sexual offenders.

Historically, the first preventative measure adopted for high-risk, persistent offenders by the Federal government in Canada , which has exclusive jurisdiction to legislate criminal law and procedures, was enacted in the form of the 1947 Habitual Offender and the 1948 Criminal Sexual Psychopath provisions of the Criminal Code of Canada (C.C.C.). In 1960, however, following several criticisms, a number of amendments were effected to the Criminal Sexual Psychopath provisions. Most noteworthy was the replacement of the term “Criminal Sexual Psychopath”, which was judged rather vague and unscientific, for the term “Dangerous Sexual Offender”.

In subsequent years, the continued dissatisfactions with the reformed provisions (e.g., disparities in application, targeting of non-violent sexual and property offenders, failure to include dangerous non-sexual offenders) led to the enactment of Bill C-51, the Criminal Law Amendment Act. This, consequently, prompted the repeal of the Habitual Offender and Dangerous Sexual Offender (DSO) legislations in 1977 and the introduction of the Dangerous Offender (DO) provisions under Part XXIV of the C.C.C.. The statutory criteria for the establishment of a Dangerous Offender application consist of offenders convicted of a Serious Personal Injury Offence (SPIO; Section 752 of the C.C.C., R.S.C. 1985, c. C-46) who exhibited a repetitive and persistent pattern of aggressive behaviours manifested by the failure to restrain those behaviours and/or to control sexual impulses. Despite the fact that the main objective of the new DO provisions was to extend the DSO law to include non-sexual offenders who pose a serious threat to the community, studies found very little change in the type of offenders being targeted by the revised legislation (Bonta, Harris, Zinger & Carrière , 1996; Bonta, Zinger, Harris & Carrière , 1998; Ministry of the Solicitor General, 1993; Trevethan, Crutcher & Moore, 2002).

The enactment of Bill C-55 in 1997 introduced some revisions to the Dangerous Offender law. One of the more important amendments was the establishment of the Long Term Offender (LTO) category. The court was now authorized to impose a term of up to ten years of community supervision to offenders following completion of a prison sentence of two or more years. Long term supervision was designed to provide a structured sentence including community supervision especially suited to manage sex offenders. A court can impose a long term supervision order if it is satisfied that there is a substantial risk that the offender will reoffend, but a reasonable possibility that such risk can be managed in the community with appropriate supervision and intervention. Accordingly, the premises of LTO provisions are tailored towards both safe reintegration and risk management. That is, this new sentencing option is based on the assumption that the successful reintegration of particular classes of offenders can be achieved without increasing the risk posed by such individuals to the community.

The Long Term Offender provisions came into effect following a number of recommendations made by a joint Federal, Provincial and Territorial (FPT) Task Force on High-Risk Offenders. This working group was also responsible for the creation of the National Flagging System (NFS), one of the more recent policy and legislative responses from Provincial and Territorial governments across Canada to better protect children and other vulnerable people from acts of violence and sexual abuse.

The NFS was announced on March 10, 1995 to track high-risk, violent offenders. In essence, the rationale for such a system is to help Crown Attorneys deal more effectively with high-risk, persistent offenders at the time of prosecution. Due to concerns regarding the relative ease with which offenders can move across a large country as well as the large number of offenders prosecuted in each jurisdiction, the NFS was developed to ensure that prosecutors are aware of potential information held elsewhere regarding an offender's likelihood of a high and continuing risk of future violent conduct. Such background information is basically intended to inform prosecutors of the need to review a particular case for a possible Dangerous Offender (DO) and/or Long Term Offender (LTO) application¹. That is, in the event of a new encounter with the criminal justice system, knowing that an offender was previously judged to be high-risk can assist prosecutors faced with making decisions on the laying of appropriate charges and prosecution strategy.

An offender is placed on the National Flagging System if a review of the available information indicates that the offender is a reasonable prospect for a DO and/or LTO designation if he or she reoffends. Each province or territory has an assigned coordinator who accepts referrals to the NFS from local Crowns, police and/or correctional agencies. The coordinator gathers and reviews information on the offender and makes the decision to flag an offender. When the decision to flag an offender is made, the Provincial/Territorial coordinator has the additional responsibility to communicate the decision to, and exchange information with, the police, corrections, Crown prosecutors and other Provincial/Territorial coordinators. NFS coordinators are therefore designated as contact persons on the CPIC system, and so liaise with the local police CPIC representative to place flagged offenders on the national police identification system as a “person of special interest”.

The criteria used by Provincial/Territorial coordinators to include an offender on the NFS are closely related to the Dangerous Offender and Long Term Offender provisions of the Criminal Code. Factors to be taken into consideration for flagging a person on the system can be either offence-based, offender-based or both. Likely candidates include, for example, cases involving a previous unsuccessful DO application, a conviction for an offence that did not meet the DO criterion of a SPIO, but where significant concerns exist as a result of other observations about the circumstances of the crime or the general pattern of behaviours of the offender, or a plea or sentence negotiation for a previous SPIO conviction. When a flagged candidate is placed on the system, the NFS coordinator is responsible for maintaining a file on the offender and update and review the file as required. When available, NFS files should ideally contain the following information: criminal record; psychiatric, probation, pre-sentence, and/or correctional reports; court transcripts; names and addresses of victims; and names of police officers and Crown prosecutors who have relevant experience with the offender.

The Present Study
The present study was aimed at empirically investigating the effectiveness of the National Flagging System in identifying and tracking high-risk, persistent offenders. Stated differently, the goal of the research was to attempt to answer whether the system currently in place was meeting its intended objective, that is to facilitate the early identification and the proper management of those offenders assessed at high-risk to reoffend violently and/or sexually. The evaluation used three different analytic approaches. First, the profile (i.e., demographic, social, actuarial risk) of the flagged offenders was explored and compared with the profile of known high-risk, violent offenders (i.e., Dangerous Offenders (DOs) and Detention Failures (DFs)). Second, the recidivism outcomes, both non-violent and violent (including sexual) of the flagged offenders were examined. Third, an analysis of the effect of the NFS in prompting the successful application of Dangerous Offender and/or Long Term Offender designations was conducted.

Method

Participants
Participants consisted of male offenders under Provincial or Federal jurisdiction. All participants were identified by various Canadian authorities as being high-risk or potentially dangerous offenders. Three different samples were used. The first group was comprised of 256 offenders, screened under the NFS policy initiative as being at a high-risk to recidivate with a violent offence. Included in the second group were 64 Dangerous Offenders (DOs) held under Part XXIV of the C.C.C. Finally, the third group consisted of 33 Detention Failures (DFs). A federally sentenced offender who is judged likely to commit an offence causing death or serious harm prior to expiration of sentence may be detained until warrant expiry under the provisions of the Corrections and Conditional Release Act (CCRA; 1992). Offenders from the Detention Failures group were drawn from a study by Motiuk, Belcourt and Bonta (1995) that followed detention cases upon release from prison². DFs were thus defined as offenders detained until expiration of sentence and who subsequently recidivated with a violent offence following release.

Procedures
Whereas the groups of DOs and DFs were drawn from Ontario and British Columbia (Bonta et. al., 1996), the flagged offenders were obtained from the records held by Provincial and Territorial NFS coordinators across the country. All male offenders placed on the system since implementation of the NFS (1995) until December, 1999 were included in the study³. At the time when the information was collected, eight jurisdictions had operational flagging systems: Alberta , British Columbia, Manitoba , Newfoundland, the Northwest Territories, New Brunswick, Ontario and Saskatchewan. Five jurisdictions (Quebec, New Brunswick, Prince Edward Island, Yukon and Nunavut Territories) whose flagging system was either not fully implemented or still in its developmental stages did not have any offender officially registered and therefore did not contribute to the NFS group. Any court declared DOs prior to having been flagged were excluded from the initial sample selection. Otherwise, a comprehensive review of the files of the flagged offenders in all jurisdictions was conducted, with the exception of British Columbia where, because of the large number of flagged offenders, a 33% randomized sample was drawn.

Based upon a review of the relevant literature on high-risk offenders, a coding manual was developed encompassing 105 variables. The information, which pertained to offender identification and demographics as well as criminal history, emotional functioning and personal circumstances, was primarily collected from files maintained by each jurisdiction. Information from the Offender Management System (OMS) of the Correctional Service of Canada (CSC) supplemented information that was not readily available from the files held by the NFS coordinators. Follow-up records were obtained from the RCMP's Criminal Records Branch. The criminal history records from the RCMP include the arrest, conviction and sentence for offenders. In addition, Provincial/Territorial coordinators from each jurisdiction were contacted and asked to provide recidivism and follow-up information from their own record systems.

The file coders consisted of two researchers. The coding manual reflected the revisions formulated during two pilot testing sessions designed to evaluate the completeness, relevance and inter-rater reliability of the original document (i.e., two cases each from Ontario and British Columbia ). Data from the hard copies file of the coding forms were transferred into an electronic database. In conjunction with the follow-up phase of the study conducted between November 2003 and April 2005, additional variables were entered directly into the electronic database. Subsequent to the adding of new variables and before the analysis, further revisions to the database were effected upon a re-evaluation of OMS records.

Measures
Actuarial Risk Assessment Instruments
Risk assessment processes are fundamental to the appropriate disposition of cases by criminal justice officials, especially when dealing with serious and persistent offenders. Within the context of this study, one of the hazards of the DO legislation lies in the potential to selectively focus on a few sexual and violent offenders who may not demonstrably be more dangerous than most of the offenders from the larger pool of sexual and violent offenders from which they are drawn (Petrunik, 1994). Consequently, a major issue in any discussion of high-risk offenders concerns the ability to predict violent and sexual behaviours (FPT Task Force on High-Risk Offenders, 1995). New policy initiatives such as the National Flagging System can benefit from procedures and/or techniques that permit better identification of high-risk offenders.

In this study, attempts were made to collect sufficient information to allow for the scoring of the Static-99⁴ (Hanson & Thornton, 2000), the Violence Risk Appraisal Guide⁵ (VRAG; Quinsey, Harris, Rice & Cormier, 1998) and the screening version of the Level of Service Inventory (LSI‑SV; Andrews & Bonta, 1998). In addition, offenders' scores on the Statistical Information on Recidivism – R1 Scale⁶ (SIR-R1; Nafekh & Motiuk, 2002) and the Psychopathy Checklist - Revised (PCL-R; Hare, 1991) were obtained, when available, from either the record forms on file or subsequent OMS queries.

Measurement of Recidivism
Recidivism data of the flagged offenders was examined and analyzed with specific attention to a successful application for a DO and/or a LTO designation. Recidivism in this study was defined as a conviction for an offence (i.e., either non-violent, violent and/or sexual) following the original date of flagging and the conviction for the index offence that initiated the flag. Violent recidivism included all sexual offences (e.g., rape, sexual assault) and crimes against the person (e.g., armed robbery, assault) as well as (conspiracy to commit) arson, resisting arrest, and assault of a police officer. In contrast, non-violent recidivism included all crimes against property (e.g., break and enter, theft), crimes against public morals and decency (e.g., indecent phone call, prostitution), narcotics offences (e.g., possession of restricted drug, trafficking), liquor and traffic offences (e.g., driving while disqualified, failure to remain at a scene of an accident) and other

offences such as willful damage to property, causing a disturbance, failure to appear, mischief and probation or parole violation.

A number of offenders were placed on the NFS while in prison and it was possible that some of them could commit a serious offence while in custody. Given that there is no legislative specification restricting the use of DO and/or LTO provisions to a community offence, there was a need to take into account the possibility of recidivism in prison that could trigger a DO or LTO application. Consequently, all convictions subsequent to the index offence, regardless of whether the offence arose from an incident that occurred within the institution or in the community, were recorded. As the conviction process for any offence committed in prison is complicated by additional concerns (e.g., witness testimony), the likelihood of observing a DO/LTO application or successful designation following an institutional offence was expected to be relatively infrequent. Only the most serious incident in each type of recidivism category was recorded. In order to clarify the general pattern of recidivism outcomes, an index representing the presence or absence of a new non-violent or violent (including sexual) offence for each offender was further computed (i.e., any recidivism).

Time-at-Risk
For each type of recidivism, time-at-risk for reoffending was recorded. For non-recidivist offenders, time-at-risk was calculated using the number of days that elapsed between the follow-up end date (i.e., December 31, 2004) and the date of release into the community or, if incarcerated (i.e., never released into the community), the date of the flagging index offence. For offenders who recidivated while in prison, time-at-risk was calculated from the date of the flagging index offence to the date of new conviction. For offenders who recidivated following release in the community, time-at-risk was calculated from the date of release into the community to the date of conviction for the recidivism offence. When the actual date of the offence that led to the new conviction was known, this date was used instead of the conviction date.

Results

This study comprised 256 male offenders identified under the National Flagging System (NFS) as being at a high-risk to recidivate with a violent and/or sexual offence. Table 1 presents the Provincial/Territorial distribution of the flagged offender (FO) sample included in the research. As can be seen, cases from Ontario represented the largest percentage (25.4%) of the FOs in the sample. Due to the randomized sample selection of files from British Columbia, a smaller number of the FOs in this study were cases from this jurisdiction, although they comprised about 20% of the sample.

Table 1. Provincial/Territorial Distribution of Flagged Offenders

Part I: A Comparison of the Profiles of FOs and HROs
One approach to assessing the risk for violent offending of the flagged offenders is to compare their profiles with Dangerous Offenders and Detention Failures. As the 64 DOs and the 33 DFs were very similar in personal-social demographics and criminal history ( Bonta et. al., 1996), both groups were merged into a High-Risk Offenders (HROs) group. Table 2 provides a general summary of personal-social demographic information on the FOs and the HROs. Similarly, mental health (i.e., psychological) characteristics and indicators of criminal history for the two groups of offenders are presented in Tables 3 and 4, respectively. Tables 2 through 4 describe the offenders' information at the time of the index offence that initiated the flag⁷.

Although the two groups were similar in terms of employment and marital status, the FOs were older and better educated. With regard to the latter difference, whereas no HRO pursued post-secondary education, eleven FOs received some university, college or trade school training and an additional eight FOs completed post-secondary education (i.e., B.A. or college/trade diploma). The FOs were also more likely to be Aboriginal than the HROs. However, the racial differences may be an artifact of the absence of sampling from the Prairie region in the HRO group (Bonta et. al., 1996).

The results concerning the mental health of the FOs and the HROs suggested a higher incidence of a personality disorder among the latter group (Table 3). Even though there were no significant differences between the two groups in the extent to which a psychiatric diagnosis of schizophrenia and/or psychopathy was met, the FOs were less likely than the HROs to have a general personality disorder and antisocial personality disorder (APD) in particular. Relative to the general federal offender population, while the base rates of psychopathy reported among both samples in the present study appeared slightly higher (Hare, 1991), the prevalence of APD was considerably lower (Motiuk & Porporino, 1992), especially for the FO group.

Table 2. Personal-Social Demographic Characteristics of FOs and HROs

1 Includes only elementary/secondary educational levels.
* p < .05; ** p < .01.

Table 3. Mental Health/Psychological Characteristics of FOs and HROs

1 Measured by the Psychopathy Checklist – Revised (PCL-R; Hare, 1991).
** p < .01.

Table 4. Criminal History Information on FOs and HROs

** p < .01.

As can be seen in Table 4, the criminal histories of the FOs and the HROs showed some differences. HROs began their criminal career at an earlier age than the FOs and a greater proportion of the HROs had a juvenile criminal history. HROs were also more likely to have a previous failure while on conditional release, but this was anticipated given that the Detention Failures comprised approximately one-third of the HRO group. With respect to the index offence, both the FOs and the HROs showed a serious criminal profile with the majority of the offenders in the two groups exhibiting violent behaviours. All the HROs (N = 98) and 96.1% of the FOs (n = 246) were charged with and/or convicted of a violent index offence. Furthermore, with only 14.6% (n = 12) of the HROs and 37.9% (n = 97) of the FOs representing non-sexual violent cases, both groups of offenders were mostly comprised of violent sexual offenders. HROs were nonetheless significantly more likely to have been charged with and/or convicted of a sexual offence at index.

Actuarial Risk Assessment Instruments
Results obtained from performing ROC analyses between the risk assessment instruments and recidivism (i.e., non-violent, violent, sexual and any) for the sample of the flagged offenders are presented in Table 5⁸. The Area Under the Curve (AUC) presented in the table of results is used to evaluate and compare the predictive accuracy of the different risk instruments. The larger the value for the AUC, the better the overall predictive accuracy of the risk instrument. Specifically, an AUC of 1.0 represents perfect prediction, whereas an AUC of .50 is chance. An easy way of interpreting for example, an AUC of .81 (such as obtained for the SIR-R1 in the prediction of non-violent recidivism; Table 5) is to say that there is an 81% chance that a randomly selected non-violent recidivist would have a higher score on the SIR-R1 than a randomly selected non-recidivist. An AUC value is statistically significant when its 95% confidence interval does not contain .50.

Both the SIR-R1 and the VRAG Proxy predicted all four measures of recidivism in the sample of FOs, with the SIR-R1 demonstrating substantially greater predictive accuracy. With one exception (sexual recidivism), the LSI-SV also predicted recidivism, both violent and non-violent as well as any recidivism, at statistically significant levels. Results obtained on Static-99 and the VRAG, however, were not outstanding. The VRAG demonstrated predictive validity for the FOs solely with regards to violent recidivism, while Static-99 only significantly predicted non-violent recidivism for offenders who were convicted of, or charge with, a sexual offence at index.

In addition to comparing the personal-social demographic, mental health and criminal history characteristics of the FOs with the HROs, the actuarial risk assessment scores for the two groups were also compared. The findings for the mean total scores are reported in Table 6. For the readers interested, Appendix A presents the distribution of scores for both the FOs and HROs across the various risk level categories. The actual percent recidivism within each risk level category on all instruments for the FO group can be found in Appendix B.

Table 5. ROC Analyses between Actuarial Risk Assessment and Recidivism

Table 6. Actuarial Risk Assessment for FOs and HROs

1 Lower scores on the SIR-R1 are indicative of an increased risk.
* p < .05; ** p < .01.

Mean differences in risk levels between the two groups were statistically significant for all assessment instruments but for the SIR-R1. With one exception, the direction of the relationship further suggested that the risk levels of the HROs were higher than for the FOs. The exception related to the LSI-SV where the FOs were assessed as higher risk. This discrepancy can be partly explained by the FOs' unusually high scores on two of the eight items comprising the LSI‑SV, namely antisocial attitudes and associates where the FO group revealed a significantly more problematic situation than the HROs in terms of attitudes (93.2% vs. 64.4% of the HROs; χ 2  = 42.32 (1, N = 322), p < .01) and patterns of association (79.1% vs. 41.4%; χ 2 = 42.30 (1, N = 322), p < .01). Also notable were the significantly higher scores of the FOs on the item measuring the extent to which patterns of current alcohol/drug use caused problems in the domains of school and/or work. Specifically, 63.0% of the FOs, but only 36.8% of the HROs ( χ 2  = 17.68 (1, N = 322), p < .01) demonstrated problematic circumstances in this area.

An examination of the standard deviations and frequency distributions of scores on the instruments indicated a greater variability in risk scores among the FO sample. Compared to the HROs, the risk levels of the FOs were more evenly distributed across the risk categories. For example, 25.8% of the FOs scored within both the Very Low and Very High risk categories, compared to only 15.7% and 31.5% of the HROs, respectively. This general pattern of results was evidenced on all actuarial risk measures. In spite of the aforementioned observed differences between the FOs and HROs, findings suggested that both groups of offenders were comprised of noticeably high-risk cases.

Part II: The Recidivism of FOs
Table 7 presents information relating to the rates of recidivism for the FOs. Among the overall sample, the follow-up for any type of recidivism ranged from 2 days to 12.7 years, with a mean follow-up time of 4.1 years (SD = 2.9 years). Time-at-risk for any recidivism among the sub-sample of offenders who were actually released in the community following the index offence also ranged from 2 days to 12.7 years, but had a mean follow-up of 3.4 years (SD = 2.6 years).

Table 7. Percent of FOs Reconvicted

Note. All offenders include those incarcerated but not released at time of follow-up.

Taken as a whole (any recidivism), 50.0% of the 256 flagged offenders were convicted of at least one new offence following the date of the index offence. Base rates for non-violent, violent and sexual recidivism among those offenders were 33.6%, 36.7% and 11.7%, respectively. The recidivism rates are relatively higher when only the FOs who were released in the community following the index offence were examined (58.3% for any; 41.1% for non-violent; 43.6% for violent; and 14.9% for sexual). This was anticipated as almost one quarter (21.1%) of the offenders in the former group were confined in an environment (i.e., prison) where criminal activity would be restricted.

Findings also demonstrated that most offences for all types of recidivism occurred following release in the community. Among the group of FOs who recidivated following the index offence, 14.1% (n = 18) of the most serious offence (MSO) occurred while incarcerated. Furthermore, approximately three-quarters of the MSO among the 128 recidivist FOs were for a violent offence (72.7% (n = 93) for all offenders and 73.3% (n = 87) for the released offenders only). Within those violent recidivism offences, approximately one-third were of a sexual nature (32.3% (n = 30) for all offenders and 32.2% (n = 28) for the released offenders only). Two of the 30 violent, sexual recidivism offences (6.7%) occurred in prison.

Part III: The Application for DO and LTO Designations
Petrunik (1994) noted that, notwithstanding the presumably large number of offenders that meet the legislative criteria, a relatively small number of individuals are declared Dangerous Offenders. This may, in part, be due to the high resource demands placed on Crown Attorneys in preparing a DO application. The collating of information by National Flagging System coordinators is thus seen by many as a useful tool in facilitating the use of the DO and/or LTO provisions. Consequently, we examined the prevalence of DO/LTO applications resulting from the violent and/or sexual recidivism offence. The results are presented in Table 8. Table 8 displays the outcomes of DO and LTO applications only for the FOs who recidivated violently and/or sexually. That is, all of these FOs committed violent offences, some of which would have made them possible candidates for a DO or a LTO application.

Following a violent recidivism offence, a DO application was initiated for only 13 of the 92 flagged offenders not already court declared DOs (n = 2). This translated into DO applications for approximately 14% of the cases intended by the NFS. Setting aside the two pending DO decisions, less than half (45.5%; 5/11) of the DO applications actually turned out as successful DO designations. A closer examination indicated that two-thirds (66.7%; 4/6) of the unsuccessful DO applications became LTO designations. Along with those 4 “failed DO/successful LTO” cases, 7 additional violent recidivists FOs were designated LTOs. Despite the fact that all LTO applications were successful, the prevalence of successful LTO applications among the flagged offender sample was very low. After accounting for the successful and/or pending DO and LTO applications prompted by the flagged index offence and the successful DO applications following the violent recidivism offence, only 13.1% (11/84) of the violent (including sexual) recidivist offenders received a long term supervision order.

Taken as a whole, a DO and/or LTO application was initiated for 22.0% (20/91) of the offenders who were not already designated a DO (n = 2) or a LTO (n = 1) prior to committing a new violent and/or sexual offence following the flagging date. Ten percent (2/20) of those DO/LTO applications were triggered by an offence that occurred within the institution. Results also indicated that 16 of the 20 cases where a DO and/or LTO application was initiated resulted in a successful DO or LTO designation (2 offenders had a DO decision still pending at the time of writing). Excluding the two cases still pending, only 18.0% (n = 16) of the 89 offenders identified by the National Flagging System as high-risk, persistent offenders who recidivated with a violent offence following the flagging date (i.e., appropriately identified as high-risk, violent offenders) were designated either Dangerous Offenders (n = 5) or Long Term Offenders (n = 11) as a result of the recidivism offence.

Table 8. DO and LTO Applications for the Recidivism Offence

1 The violent recidivist offenders who were already court declared DOs (n = 2) prior to the violent recidivism offence were excluded from the analyses (N = 92).
2 The violent recidivist offenders who were already court declared DOs (n = 2) or LTO (n = 1) prior to the violent recidivism offence, who became DOs (n = 5) following the violent recidivism offence, and who had a DO decision pending (n = 2) were excluded from the analyses (N = 84).
3 The sexual recidivist offenders who became court declared DOs (n = 5) following the violent recidivism offence and who had a DO decision pending (n = 2) were excluded from the analyses (N = 23).

The reasons for the relatively low rates of DO/LTO applications among the flagged offenders who reoffended violently and/or sexually remain, however, unclear. For instance, it is possible that those 20 cases that prompted DO/LTO applications differed in some respect from the rest of the offenders who also recidivated violently and/or sexually following the flagging date. In an attempt to shed some light on this issue, a series of comparison analyses involving a number of personal-social demographic, mental health, criminal history, actuarial risk and recidivism offence variables were performed. That is, the violent and/or sexual recidivist offenders for whom a DO and/or LTO application was initiated (n = 20) were compared with the violent and/or sexual recidivist offenders for whom no submission for a DO/LTO application was made (n = 71). The results are presented in Table 9. Any difference between the two groups on some important factors could help explain why Crown Attorneys failed to initiate a DO or a LTO application for most of the offenders who committed a new violent (including sexual) offence following the flagging date.

Table 9. Comparison Analyses between Violent/Sexual Recidivist Offenders for whom a DO/LTO Application Was Submitted or Not Submitted

Note: Numbers vary due to insufficient information. ** p < .01; * p < .05.
1 Includes only elementary/secondary educational levels (3 non-DO/LTO, but no DO/LTO, offenders received some university, college or trade school training).
2 Computed only for offenders whose index offence was of a sexual nature.

Generally speaking, the violent/sexual recidivist offenders for whom either a DO or a LTO application was initiated were not very different from the offenders who also committed a new violent/sexual offence following the flagging date, but for whom Crown Attorneys did not initiate a DO/LTO application. An examination of Table 9 demonstrated that both groups of offenders were similar in terms of age, race, marital status, educational level, employment status and criminal history. Furthermore, whether or not a DO/LTO application was initiated did not appear to be related to actuarial measures of risk, and for the most part, mental health characteristics. As a matter of fact, the general trend in the pattern of results with respect to some of these variables pointed in the opposite direction from what was to be expected. Specifically, non-DO/LTO nominees were significantly more likely to have scored 30 or more on the PCL-R, which qualified them for a diagnosis of psychopathy. The only other statistically significant difference between the groups related to the type of recidivism offence, where sexual offences were more likely to prompt DO/LTO applications than violent, non-sexual offences. Surprisingly, the circumstances of the offence (i.e., whether it occurred in prison or following release in the community) did not seem to have an impact on Crown Attorneys' decision to initiate a DO and/or a LTO application.

Discussion

The principal goal of the National Flagging System is to identify high-risk , violent offenders who fit the profile of potential Dangerous Offenders and/or Long Term Offenders. Ultimately, the process is aimed at alerting Crown Attorneys about the danger posed by some offenders and ensuring that appropriate measures are taken (i.e., initiate a DO and/or a LTO application) to deal with that risk. The present study provided some evidence for the effectiveness of the NFS in appropriately screening those offenders deemed at high risk to reoffend violently and/or sexually. The findings were, however, less convincing with respect to stimulating the use of the Dangerous Offender and Long Term Offender provisions of the C.C.C.

Identifying High-Risk Offenders for Flagging
A comparison of the profile (i.e., personal-social demographics, mental health and criminal history) of the FOs with the profile of a group comprised of court declared Dangerous Offenders (DOs) and a group of Detention Failures (DFs) initially demonstrated that, with a few exceptions, the offenders flagged under the NFS revealed significantly less serious and persistent criminality characteristics than the known high-risk, violent offender (HRO) sample. Specifically, although the two groups of offenders were similar in terms of employment and escape history, the FOs were older and better educated, exhibited fewer antisocial personality characteristics, began their criminal activity later in life and were less likely to have a juvenile criminal history and/or a previous failure while on conditional release.

These preliminary findings were not overly encouraging in light of the existing state of knowledge regarding the theoretical and empirical importance of age, educational level, antisocial personality patterns and criminal history in explaining individual variability in criminal and other rule-violating behaviours, and in predicting instances of future general, violent and sexual recidivism (e.g., Andrews & Bonta, 2003; Gendreau, Little & Goggin, 1996; Pratt & Cullen, 2000). That is to say, compelling evidence for the effectiveness of the NFS in targeting the appropriate group of offenders would have been provided if the flagged offenders demonstrated many similarities (i.e., on several important risk factors and/or criminogenic needs) to a group of offenders known to be dangerous. Supplementary analyses were thus necessary to examine whether the NFS was fulfilling its role of identifying high-risk, violent and persistent offenders.

Indeed, an examination of various actuarial risk assessment scores indicated that both groups of offenders were high-risk samples. The distributions of risk assessment scores for the two groups were generally higher than what is expected based on normative samples of general (Andrews et. al., 1998; Quinsey et. al., 1998) and/or sexual, offenders (Harris, Phenix, Hanson & Thornton, 2003) . Only on the SIR-R1 scale was the distribution of scores comparable to the one reported among the Canadian federal offender population (Nafekh et. al., 2002). For example, as assessed by the VRAG and/or its proxy measure, less than 10% of the FOs and the HROs scored in the lowest three risk categories, whereas about 40-60% of the FOs and the HROs scored in the highest three risk categories. The exact opposite distribution of scores would be expected from the VRAG norms based on the 618 mentally disordered offenders where slightly over one-half scored in the bottom three risk categories and less than 10% in the top three risk categories. Even a conservative comparison finds the FOs and the HROs approximately four times more likely to have been assessed in the highest three VRAG/VRAG Proxy risk categories.

This last series of observations is particularly noteworthy given that, because they combine information of individual risk factors and/or criminogenic needs, actuarial risk scales provide a better overall assessment of offender risk. The previous differences observed on several personal-social demographic, mental health and criminal history variables between the FOs and the HROs therefore appeared partly attributable to the lack of variability, or concentration of high-risk individuals within both groups of offenders under study, relative to general offender populations.

Additional evidence for the effectiveness of the NFS in appropriately identifying high-risk, violent and potentially dangerous offenders was provided from an analysis of the recidivism data of the flagged offenders. Using a comparable follow-up period, the overall rates of recidivism of FOs appeared similar to those reported among the general federal male offender population (Bonta, Rugge & Dauvergne, 2003)⁹. However, the new offences committed by the FOs were more serious. Precisely, the base rates of both violent and sexual reconvictions were much higher among the present sample (43.6% and 14.9% for released offenders), compared to those reported for federal male offenders (16% and less than 2%; Bonta et. al., 2003) . More specifically with respect to groups of sexual offenders, the reconviction rate for sexual crimes among the FOs was also considerably higher than the sexual offence recidivism rate usually reported (10%; Harris & Hanson, 2004). This last finding was anticipated as the majority of the offenders in this study were flagged following a sexual index offence.

Initiating DO and LTO Applications
An analysis of the usefulness of the NFS in prompting DO and/or LTO applications was, however, not as encouraging. Consistent both with previous research (Petrunik, 1994) and expectations (given the long term and serious consequences DO verdicts entail as well as administrative matters such as plea negotiations, etc.), this study suggested that Crown Attorneys were reluctant to initiate DO/LTO applications. Although the majority of DO and/or LTO applications were successful (with most unsuccessful DO applications reduced to LTO designations), the actual number of DO/LTO submissions was somewhat disconcerting.

Low rates of DO/LTO applications and/or successful designations were observed among the flagged offenders in general, and the violent (including sexual) recidivist offenders in particular. Out of the 91 flagged offenders who were not already court declared DOs or LTOs and who were found guilty of a violent/sexual offence following the flagging index offence, Crown Attorneys initiated either a DO or LTO application in only 20 (22.0%) cases. Furthermore, comparison analyses between the violent/sexual recidivist offenders for whom a DO/LTO was initiated and the violent/sexual recidivist offenders for whom a DO/LTO was not initiated failed to clarify the reasons why Crown Attorneys did not proceed with more DO or LTO applications.

Also worthy of discussion is the finding that sexual offences were more likely than non-sexual violent offences to prompt DO and/or LTO applications and further result in successful DO/LTO designations. One possibility is that some of the violent, non-sexual recidivism offences were not Serious Personal Injury Offences or offences that could support a DO or a LTO application. Notwithstanding, sexual offenders appear to remain the target population in the eyes of both Crown Attorneys and the courts. This is so even though the NFS coordinators were attentive to former calls for a broader use of the revised Dangerous Offender legislation (Bonta et. al., 1996; Ministry of the Solicitor General, 1993b; Trevethan et. al., 2002). Specifically, slightly more than one-third of the flagged offenders represented non-sexual violent cases. Taken as a whole, the findings nevertheless suggest that the violent/sexual recidivist offenders for whom either a DO and/or a LTO application was made shared more similarities than differences with the offenders for whom such process was not initiated.

Limitations of Findings, Implications, and Suggestions for Future Research
Even though the National Flagging System appeared fairly successful in terms of identifying those offenders who were likely to reoffend violently and/or sexually, this study found that taking the next steps intended by the NFS, that is, initiating a DO or LTO application was infrequent. To demonstrate exactly where and how the progression actually goes amiss would have required a different methodology that perhaps would have questioned the motives of Crowns for proceeding or not with a DO or LTO application at the violent recidivism event. It is nevertheless possible to speculate on some potential problematic areas.

One primary concern is the fact that the small number of court declared DOs and LTOs did not allow separate analyses to be conducted on the two groups. Although a certain percentage of LTO designations resulted from failed DO applications, DO and LTO provisions stipulate different criteria for eligibility. If a larger sample were available, analyses could have been undertaken to try and understand why Crown Attorneys did not initiate a DO or a LTO application for the majority of violent/sexual recidivist offenders. It is likely that in some cases, the assigned prosecutors entered into sentence negotiations because of the relative weakness of their cases. Equally important is the need to understand the factors that result in a successful DO or LTO application. Few would argue for the possibility that the successful applications differ in some way from the failed applications. As previously noted, this limitation largely impacts on the extent to which definitive conclusions could be reached regarding the effectiveness of the National Flagging System in facilitating the use of DO and LTO provisions.

Finally, it is worth mentioning that this study was conducted over the first four years of implementation of the NFS, that is, when the system was getting off the ground and operational steps were being developed. Had the study been conducted in the most recent years, different results would perhaps had been obtained. In light of the aforementioned, it is recommended that a second phase of this study, including a larger sample of flagged offenders and/or a longer follow-up period, be undertaken.

Notwithstanding these limitations and taking into consideration that so many factors come into play when making a DO/LTO application (e.g., lack of evidence, plea bargain) and that, over the past seven years, the courts have imposed an average of approximately 25-30 DO designations and 40 LTO orders annually (Public Safety and Emergency Preparedness Canada, 2004), perhaps the 20-25% rates of DO/LTO designations observed among the FOs in this study is all that could be expected.

Based on the results of this study, it nonetheless appears that the practical value of the National Flagging System could be augmented if a more effective protocol for responding to those offenders who pose a threat to the community is implemented. First, improvements may be achieved if additional criteria for flagging are added that reflect what is known about DOs. This study found that FOs differed from HROs in terms of juvenile criminal history, prior failure on conditional release and Antisocial Personality Disorder (Tables 3 and 4). Ensuring that flagged offenders have these characteristics may lead to better identification of candidates suitable for DO applications. The criterion, Antisocial Personality Disorder, may be especially important to persuade the courts to the unlikelihood of the offender changing his behaviour.

Second, we do not know how well Crown Attorneys respond to criminal records that note a ''person of special interest''. That is, we are uncertain of the extent to which the assigned Crown prosecutors understand how well they can benefit from the information the NFS coordinators have at their disposal. It is possible that a lack of communication exists between the various intended users of the system and that, in some cases, police and prosecutors simply did not “pick up” on the flag. Finally, we need to know the underlying reasons why Crown Attorneys have failed to follow-up on those offenders identified by the NFS in order to improve the way the current system operates.

Conclusion

On the one hand, findings from this study provided sound evidence for the effectiveness of the National Flagging System in identifying high-risk, violent offenders. As illustrated by the relatively low rates of Dangerous Offender and Long Term Offender applications among the violent/sexual recidivist flagged offenders, the value of this policy initiative in tracking and responding to those persistent offenders who pose a risk to the community was, however, less convincing. Amongst other implications, the present findings, along with previous research on the identification and prediction of violent and/or sexual reoffending (Bonta et. al., 1996; Hanson & Morton-Bourgon, 2004), therefore encourage the development of guidelines to assist corrections and criminal justice professionals in being more effective in screening, monitoring and processing high-risk, violent and potentially dangerous offenders.

References

Andrews, D. A., & Bonta, J. (1998). The Level of Service Inventory – Revised: Screening Version (LSI-SV). Toronto: Multi-Health Systems Inc.

Andrews, D. A., & Bonta, J. (2003). The Psychology of Criminal Conduct , 3 rd ed. Cincinnati: Anderson Publisher Ltd.

Bonta, J., Harris, A., Zinger, I., & Carrière, D. (1996). The Crown Files Research Project: A Study of Dangerous Offenders. Research Report No. 1996-01. Ottawa: Ministry of the Solicitor General of Canada.

Bonta, J., Rugge, T., & Dauvergne, M. (2003). The Reconviction Rate of Federal Offenders . Research Report No. 2003-02. Ottawa: Ministry of the Solicitor General of Canada .

Bonta, J., Zinger, I., Harris, A., & Carrière, D. (1998). The dangerous offender provisions: Are they targeting the right offenders? Canadian Journal of Criminology, 40, 377-400 .

Federal/Provincial/Territorial Task Force on High-Risk Violent Offenders (1995, January). Strategies for Managing High-Risk Offenders. Report presented to FPT Ministers responsible for Justice. Ottawa: Canada.

Gendreau, P., Little, T., & Goggin, C. (1996). A meta-analysis of the predictors of adult recidivism: What works! Criminology, 34, 401-433.

Hanson, R. K., & Morton-Bourgon, K. (2004). Predictors of Sexual Recidivism: An Updated Meta-Analysis. Research Report No. 2004-02. Ottawa: Public Safety and Emergency Preparedness Canada.

Hanson, R. K., & Thornton, D. (2000). Improving risk assessments for sex offenders: A comparison of three actuarial scales. Law and Human Behavior, 24, 119-136.

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

Harris, A., & Hanson, R. K. (2004). Sex Offender Recidivism: A Simple Question . Research Report No. 2004-03. Ottawa: Public Safety and Emergency Preparedness Canada.

Harris, A., Phenix, A., Hanson, R. K., & Thornton, D. (2003). Static-99 Coding Rules – Revised 2003 . Ottawa: Ministry of the Solicitor General of Canada .

Jakimiec, J., Porporino, F., Addario, S., & Webster, C. D. (1986). A Descriptive Study of Incarcerated Dangerous Offenders. Research Report No. 1986-44. Ottawa: Ministry of the Solicitor General of Canada .

Ministry of the Solicitor General (1993). Proposals to Amend the Corrections and Conditional Release Act, the Criminal Code, the Prisons and Reformatories Act and the Transfer of Offenders Act. Ottawa.

Motiuk, L. L., Belcourt, R., & Bonta, J. (1995). Managing High-risk Offenders: A Post-detention Follow-up. Research Report R-39. Ottawa: Correctional Service of Canada.

Motiuk, L. L., & Porporino, F. J. (1992). The Prevalence, Nature and Severity of Mental Health Problems among Federal Male Inmates in Canadian Penitentiaries . Research Report R-24. Ottawa: Correctional Service of Canada.

Nafekh, M., & Motiuk, L.L. (2002). The Statistical Information on Recidivism - Revised 1 (SIR-R1) Scale: A Psychometric Evaluation. Research Report No. R-126. Ottawa: Correctional Service of Canada.

Petrunik, M. (1994). Models of Dangerousness: A Cross Jurisdictional Review of Dangerousness Legislation and Practice. Research Report No. 1994-02. Ottawa: Ministry of the Solicitor General of Canada .

Pratt, T. C., & Cullen, F. T. (2000). The empirical status of Gottfredson and Hirschi's general theory of crime: A meta-analysis. Criminology, 38, 931-964.

Public Safety and Emergency Preparedness Canada (2004). Corrections and Conditional Release Statistical Overview. Ottawa.

Quinsey, V. L., Harris, G. T., Rice, M. E., & Cormier, C. A. (1998). Violent Offenders: Appraising and Managing Risk. Washington, DC: American Psychological Association.

Trevethan, S., Crutcher, N., & Moore, J.-P. (2002). A Profile of Federal Offenders Designated as Dangerous Offenders or Serving Long-Term Supervision Orders. Research Report No. R-125. Ottawa: Correctional Service of Canada .

Appendix A - Risk Level Distribution for FOs and HROs

Appendix B - Percent Recidivism within Risk Level Categories for FOs

End notes

1. Although the NFS was initially intended to identify suitable candidates for Dangerous Offender designations, Provincial/Territorial coordinators agreed in 1998 that the decision the flag an offender should also reflect his or her potential suitability of becoming a Long Term Offender.

2. One offender was excluded from the DF group for being comprised in the flagged offenders sample.

3. The study was restricted to male offenders as there was only one female offender who was flagged.

4. In contrast to the other instruments, Static-99 was computed only for offenders who were convicted of, or charged with, a sexual offence at index.

5. Due to issues surrounding the availability of data, two adjustments were made to compute VRAG scores. First, Item 4 (i.e., marital status) was scored using a twelve-month rule of cohabitation instead of the usual six-month criteria. Second, and more importantly, for many offenders Item 12, or the Psychopathy Checklist - Revised (PCL-R; Hare, 1991) score was unavailable. For these offenders a VRAG Proxy total score that did not include Item 12 was computed.

6. As CSC policy does not mandate the application of the SIR-R1 to Aboriginal offenders (Standard Operating Practice 700-04), scores on the SIR-R1 were, for the most part, only available for non-Aboriginal offenders.

7. Note that numbers in the tables of results vary due to missing information.

8. Note that numbers in the table of results vary due to insufficient information.

9. A re-analysis of the first release cohort from the Bonta et. al. (2003) study was conducted using a 3.4-year follow-up.

Date modified:

2018-01-31