The umpires strike back: Canadian judicial experience with risk-assessment instruments.

• Author: Cole, David P.
• Position: Sentencing and Risk Assessment

A recent national survey of Canadian judges solicited their views about including risk-assessment information in pre-sentence reports (PSRs) (Bonta, Bourgon, Jesseman, and Yessine 2005). (1) The 2005 documented findings indicated that 68.3% of judges thought that risk-assessment information should be included in PSRs. Another 10.6% of those surveyed had no strong opinion. Only a minority of judges (21.2%) were expressly opposed to the inclusion of such information. According to the authors, the qualitative responses disclosed that the objections of many of those judges were based on concerns about the validity and reliability of the instruments being used, rather than on their being opposed to risk-assessment instruments entirely (Bonta et al. 2005). Assuming that this survey is representative of a given benchmark in recent times, it seems clear that most Canadian criminal court judges would like to have access to risk-assessment information that can be factored into their sentencing decision-making processes.

This canvass of judicial attitudes was part of a larger survey, in which the views of other criminal justice professionals--Crown attorneys, defence counsel, and probation officers--were probed about their perceptions of the utility of PSRs in sentencing decisions and correctional placement processes. The entire project was conducted with the encouragement of provincial and federal governmental officials concerned with community corrections. This national initiative is consistent with attempts by many (if not most) provinces to conceptualize PSRs as risk/needs assessment tools, and as the foundation for case-management practices by probation and correctional staff. (3) To date, six Canadian jurisdictions--Northwest Territories, Saskatchewan, Manitoba, Ontario, New Brunswick, and Newfoundland--have, to varying degrees, woven their various risk-assessment tools into their PSR process. It seems likely that most, if not all, provinces and territories will soon move to similar models of PSR preparation (see Cole and Angus 2003).

Because dangerous/long-term offender proceedings under Part XXIV of the Criminal Code so obviously engage risk as the central focus of the hearing, there has, particularly in recent years, been some considerable use of and judicial reflection about the admissibility and adequacy of risk-assessment instruments in Canadian courts (see, e.g., Webster, Harris, Rice, Cormier, and Quinsey 1994; Grant 1998; Jackson 1997; R. v. Neve). I have made a deliberate choice not to review this literature in this article, both because such hearings invariably involve full disclosure of the instruments used and the scoring criteria employed and because they usually (4) involve properly trained professional assessors, knowledgeable about psychometric properties such as reliability, validity, and stratified sampling of populations. In addition, their knowledge usually includes formulation of opinions based on careful integration of quantitative and qualitative information about offender subjects. Furthermore, in such important hearings, these issues are usually tested on the anvil of cross-examination.

My focus in this article is to examine a less well-considered issue--how correctional officials interested in conserving scarce resources are increasingly seeking, in routine cases, to import into bail and sentencing proceedings risk-assessment instruments principally designed for institutional classification, placement, and release decision making. I shall then address a concern, expressed by academics and policy makers, that judges may be so impressed by these instruments that there is a danger that proportionality--the "fundamental principle" of Canadian sentencing law (see Criminal Code s. 718.1; Youth Criminal Justice Act (YCJA) s. 38(2)(c))--may be trumped by considerations of risk, particularly in an era where the Supreme Court of Canada has emphasized that appellate review of sentencing decisions is to be employed rarely.

1 Non-disclosure of the use of risk-assessment instruments

1.1 By correctional officials

Against this backdrop of increased use of risk-assessment tools by correctional officials, it is both surprising and disappointing to learn that, sometimes, judges are deliberately not told when actuarial instruments are being used in the process of risk assessment. In preparation for a 2003 article on the uses and abuses of PSRs (see Cole and Angus 2003), my co-author and I (separately) wrote to all Canadian provincial and territorial jurisdictions and most U.S. state jurisdictions to survey PSR policies and practices and to inquire specifically about current practices and future plans, if any, with respect to the inclusion of formal risk/needs assessment in the pre-sentence investigation, report, or both. We learned that, in some jurisdictions that administer a risk/needs assessment as part of the investigative process, by express policy, those drafting PSRs either avoid referring to the use of the assessment, or pointedly exclude the result from the report to the court, or both. We found evidence of a general reluctance in some American and Canadian jurisdictions to expose risk/needs assessment, even tools based on validated probabilities, to the possibility of cross-examination. For example, a summary of Ontario's policy document forwarded to the second author in 2001 stated,

The assessment process for court-ordered report (PSR/PDR) closely parallels the LSI-OR assessment process. It is entirely appropriate and may be very helpful if the author uses the LSI-OR as a tool to assist with the preparation of a PSR/PDR assessment. However, the use of the instrument and the score must not be referenced in the report. (5) One would hope that such intellectually dishonest--and ultimately self-defeating--practices would have disappeared, as Canadian correctional officials have begun to embed risk-assessment tools into documents intended for or likely to find their way into public court processes. (6) Unfortunately, a December 2005 decision of the Supreme Court of Canada strongly suggests that these practices are continuing: May v. Ferndale Institution was the lead among a series of cases challenging decisions, made between November 2000 and February 2001, to return prisoners serving life sentences from minimum to medium security federal institutions in British Columbia, notwithstanding an entire absence of any allegation of fault or misconduct on the prisoners' parts. The involuntary transfer decisions were all based on a computerized reclassification system, the Security Reclassification Scale (SRS), which had been developed to help institutional staff determine the most appropriate level of security at key points throughout an offender's sentence. (7) It provided a security rating, based on data entered with respect to various factors assumed to be related to the assessment of risk. The prisoners involved in the litigation had, on several occasions, requested a copy of the scoring matrix, which provides information on the numerical values to be assigned to each factor and explains the manner in which the final score is generated by a computerized tool. They claimed that, absent this information, they could not formulate a meaningful response to the reclassification decisions. In other words, while the prisoners knew what the factors were, they were unable to discern how the values were assigned to them or how those values factored into the final score (or even whether the overall score generated was accurate).

In the lower courts, the correctional authorities first told the judges hearing the cases that the scoring matrix was unavailable. In addition, they argued that the SRS was simply a preliminary assessment tool that was by no means determinative of the transfer decision, because correctional officials could override it. In other words, "[I]t was thought to be a duplication of information already disclosed" to the prisoners, who had received an explanation for the involuntary transfer decisions in writing. However, by the time the case reached the Supreme Court of Canada, the prisoners had come into possession of the scoring matrix now being used, the cover page of which indicated that this was a June 2001 updated version of the SRS. It was thus patently obvious that the previous version was in existence and could have been produced.

While the Supreme Court split 6-3 on the issue of whether the release of the scoring matrix would ultimately have made a difference to the transfer decisions, both the majority and the minority agreed that the prisoners should have received the materials. In particularly scathing language, the judges in the majority found that correctional officials had "falsely [mis]led" the lower court(s), behaviour which the Court found "highly objectionable." It was "disingenuous" on the part of counsel for the correctional authorities to maintain that the information was believed merely to duplicate what was released to the prisoners (para. 111). For the majority, "crucial information" had been "concealed" from the prisoners and from the courts (para. 120). The Supreme Court ordered that the prisoners be returned to minimum security--but only after five years of litigation, during which time they were housed in more restrictive penal environments.

The tenacity with which correctional officials resisted disclosure of the materials sought seems instructive about how some provincial authorities are, unfortunately, intensely protective of their control over data in their possession. For example, the most recent version of the Ontario Probation, Parole and Conditional Sentence Policy and Procedures Manual continues to maintain the policy previously referred to (Ontario, Ministry of Community Safety and Correctional Services 2007). In the 2006 case of R. v. B.H.D. (discussed below), one of the bases on which Saskatchewan Crown prosecutors resisted disclosure of the...