Sometimes, risk assessment tools can be compromised by variables derived from a majority population that excludes minority groups. This is precisely the issue that came to the fore in Ewert v Her Majesty the Queen, 2015 FC 1093.
Jeffrey Ewert is an Aboriginal offender in his fifties. He is serving two life sentences for second degree murder and attempted murder. He has spent more than 30 years in federal penitentiaries, oscillating between medium and maximum security facilities. For more than 20 years, Correctional Services Canada (CSC) relied on actuarial risk assessment scores to keep Mr. Ewert in custody, arguing that he was an “undue risk to society”.
Actuarial violence and sexual recidivism risk assessment tools (AVSRRATs) are based on empirical studies carried out on historical forensic and offender populations. Akin to actuarial methods used to make inferences about the behaviour and outcomes of large population groups (e.g., insurance eligibility and rates), some AVSRRATs rely on a set number of statistically derived key variables that assist clinicians in defining personality types. Other AVSRRATs assist in rating the likelihood of violent, criminal, and sexual offence recidivism.
Notably, conclusions provided by the actuarial method are about the flock of geese, as it were, and not about any individual goose. For this reason, when they are used to determine whether an offender is an “undue risk to society” at the parole stage, their accuracy warrants scrutiny, as negative scores risk prolonging the duration of incarceration.
One potential issue with AVSRRATs is the presence of inherent cultural bias. Sometimes, risk assessment tools can be compromised by variables derived from a majority population that excludes minority groups. This is precisely the issue that came to the fore in Ewert v Her Majesty the Queen, 2015 FC 1093.
In Mr. Ewert’s case, CSC primarily relied on the PCL-R (a tool that measures the presence and degree of psychopathy), in addition to several others used to predict sexual recidivism. As a result, Mr. Ewert was denied temporary escort absences, and for several reasons, never had a parole hearing.
In 2007, Mr. Ewert initiated grievances about the cultural appropriateness of the tests, and challenged their validity at the Federal Court. Research by CSC into their applicability to federal Aboriginal offenders commenced in 2003, but remained incomplete. In the 2007 Federal Court (Trial Division) decision referable to this challenge, Justice Beaudry encouraged CSC to get on with its research.
In 2015, Mr. Ewert successfully sued the CSC for using unreliable actuarial risk assessment tools, arguing they had adverse effects on his incarceration and that they were culturally biased. Justice Phelan, of the Federal Court, determined the PCL-R was cross-culturally variant, biased, and insufficiently reliable for Aboriginal offenders.
The PCL-R scores two main “factors”. Factor 1 scores eight items, including interpersonal and affective elements (e.g. one’s superficial charm). Factor 2 is comprised of nine personality features and scores the presence or absence of types of past antisocial behaviour and lifestyle. Factor 1 was deemed “junk” science in relation to Aboriginal offenders. Mr. Ewert scored 100% and 98% on Factor 1, which, taken together with his Factor 2 score, easily reached the cutoff mark to be designated as a psychopath. CSC used his PCL-R score to argue that he was an “undue risk to society”.
use of the PCL-R with this offender group entailed the likelihood of cross-cultural bias, lacked evidence or research proving its validity for Aboriginal offenders
The Court heard evidence from two experts on the reliability of these tools for Aboriginal offenders. The evidence of Mr. Ewert’s expert, Dr. Hart, a forensic psychologist, was ultimately accepted by the Court for its objectivity. Dr. Hart testified that, given the differences between Aboriginal and non-Aboriginal offenders, he would not have applied the actuarial tests. He suggested that “structured clinical assessments”, which provide clinicians with the ability to balance actuarial data with other important risk variables, would better capture the relevant background risk factors of Aboriginal offenders. Dr. Hart also explained that use of the PCL-R with this offender group entailed the likelihood of cross-cultural bias, lacked evidence or research proving its validity for Aboriginal offenders, and that it was developed using studies from a predominantly Caucasian inmate population.
The Court accepted this evidence. It determined that by using unreliable tools of assessment, CSC breached its statutory duty under section 24(1) of the Corrections and Conditional Release Act, S.C. 1992 c. 20, which provides that “[t]he Service shall take all reasonable steps to ensure that any information about an offender that it uses is as accurate, up to date and complete as possible”.
Since the tools had a direct impact on his continued incarceration, Mr. Ewert’s section 7 Charter right was infringed. The continued use of the tools was overbroad, operating beyond the purpose and objective of the legislation, and directly impacted his liberty. CSC was enjoined from using the tools in respect of Mr. Ewert and other Aboriginal inmates, until a study could confirm their reliability.
The Crown appealed. The Federal Court of Appeal overturned the decision (2016 FCA 203). The Court of Appeal concluded that Mr. Ewert should have established, on a balance of probabilities, that the tools generated false results and conclusions for Aboriginal persons. The mere fact that there is an absence of research on cross-cultural bias did not suffice. While Dr. Hart testified to the subtle effect of cultural bias in the PCL-R, he did not testify to the magnitude of that bias. Thus, there was no Charter breach. Mr. Ewert sought leave to appeal to the Supreme Court of Canada. Leave was granted on March 9, 2017.
the holding is a call for Courts to rise to their gatekeeper function in upholding the primacy of the scientific method when it comes to the reliability of actuarial tools.
The decision of the Federal Court is critical for at least three reasons. Firstly, the holding acknowledges the preeminent cultural values articulated by Gladue/Ipeelee (and progeny), which requires courts to take judicial notice of factors including but not limited to systemic racial discrimination, and higher levels of incarceration among Aboriginal offenders. These principles are critical at every waypoint in the criminal justice system, including the parole stage. While this is not a case about judicial notice, that CSC was barred from applying these tools reaffirms, and rightfully so, that Aboriginal offenders are unique and must be treated as such. Dr. Hart’s testimony further solidifies this point.
Secondly, the holding is a call for Courts to rise to their gatekeeper function in upholding the primacy of the scientific method when it comes to the reliability of actuarial tools. It will also mean that mental health experts on either side of the fence must ensure that their research is scientifically sound and assessments are objective. Counsel should ensure that these types of tools are not applied pro forma. This should especially be the case in dangerous/long-term offender hearings, where Aboriginal offenders are regrettably overrepresented, and actuarial tools are relied upon to determine whether an offender should be sentenced indefinitely. Counsel should also be critical of these tools when used with offenders whose culture is markedly different than the ones used to establish the tool’s validity (e.g. extremist offenders charged with terrorism offences). Overall, Courts should be wary of these issues when performing their gatekeeping function pursuant to the Mohan/Abbey criteria.
Thirdly, the decision highlights the relationship between the weight of scientific evidence as it affects an offender’s liberty, and Charter interests. Moreover, it reflects the broad remedial scope of section 24(1) of the Charter. Counsel would be wise to resort to section 24(1) to fashion creative remedies moving forward, particularly in the parole context.
Jeffrey Ewert's fate at the Supreme Court will be critical to the application of violence risk assessment tools in the future, especially to Aboriginal offenders, and offenders with different cultural backgrounds. The value of empirically derived tools in the assessment of risk should not be taken to mean that they are infallible. Subject to what the Supreme Court says about the issue, and if nothing else, Ewert is a call to decision makers, lawyers and clinicians to assiduously work towards avoiding bias where science could end up dictating the deprivation of liberty. For when science meets liberty, the stakes are much higher.
About the authors
Reem Zaia, (Hons.) B.Soc., J.D., LL.M. is an associate lawyer at Henein Hutchison LLP. She recently attended the University of Toronto for her Master of Laws, where she was the Nathan Strauss Q.C. Fellowship in Constitutional Law. Her scholarship primarily focuses on privacy law and the Charter, as well as and terrorism prosecutions. She is the author of several legal articles on criminal and constitutional law. Dr. Bloom and Ms. Zaia work collaboratively on research in the area of mental health and extremism/radicalization.
Hy Bloom LL.B. M.D., F.R.C.P(C.) is a forensic psychiatrist and lawyer in Toronto. He has conducted numerous assessments of offenders using actuarial tools. He is also an assistant professor in the Department of Medicine at the University of Toronto and an alternative chairperson of the Ontario Review Board. He is the renowned author of several books and scholarly articles on mental health and the law and forensic psychiatry. He has assessed hundreds of accused persons in cases including but not limited to the Toronto-18, dangerous/long-term offenders and homicides.
 VRAG (Violence Risk Appraisal Guide) is used to predict violent recidivism; the SORAG (Sex Offender Risk Appraisal Guide) violent sexual recidivism; the Static 99 the probability of sexual and violent recidivism among adult mates convicted of at least one sexual offence, and the VRS-SO, sexual recidivism to measure and link treatment changes and inform their delivery.
 See also White Burgess Langille Inman v Abbott and Haliburton Co, 2015 SCC 23;  2 SCR 182.