YASI’s Provides High Predictive Validity for Recidivsm in Youth Offenders

Natalie J. Jones, Orbis Partners, Ottawa, Ontario, Canada
Shelley L. Brown, Carleton University
David Robinson, Orbis Partners, Ottawa, Ontario, Canada
Deanna Frey, Alberta Justice and Solicitor General, Edmonton, Alberta, Canada

The primary purpose of this study is to introduce the Youth Assessment and Screening Instrument (YASI; Orbis Partners, 2000), which is a comprehensive assessment protocol gauging a range of risks, needs, and strengths associated with criminal conduct in juvenile populations. Applied to a sample of 464 juvenile offenders bound by community supervision in Alberta, Canada, the Pre-Screen version of the instrument achieved a high level of accuracy in predicting both general and violent offenses over an 18-month follow-up period (Area Under the Curve [AUC] .79). No significant differences in overall predictive validity were found across demographic groups, save for the relatively lower level of accuracy achieved in predicting general reoffending across the subsample of girls (AUC .68). With regard to strengths, a buffering effect was identified whereby high-risk cases with higher levels of strength had superior outcomes compared to their lower strength counterparts. Results suggest that it is advisable to consider the quantitative inclusion of strength-based items in the assessment of juvenile risk.

Youth Assessment and Screening Instrument (YASI), juvenile assessment, strengths, Aboriginal offenders, gender-responsive assessment

“The current article features an introduction of the Youth Assessment and Screening Instrument (YASI), a risk/need/strength assessment tool employed by several jurisdictions across North America to predict recidivism and guide case management efforts in juvenile justice contexts. Consistently shown to exceed the accuracy of unguided clinical judgment, the development and implementation of formal risk assessment protocols like the YASI has burgeoned over the last few decades. Beyond a strict determination of risk to reoffend, information yielded from the assessment process can be integral to decisions regarding security classification in custodial settings, requisite levels of supervision, the prioritization of treatment targets, and institutional release” (p. 182). The tool is available in both Pre-Screen and Full Assessment form. “In each case, the protocol is scored on the basis of a semi-structured interview, with input frequently offered by parents or an alternative legal guardian. Interview-based data are supplemented with a systematic review of collateral sources including police files, probation records, school records, and mental health reports” (p.184). The YASI Pre=Screen “contains 34 items tapping both static and dynamic risk and strength factors across nine domains. Many items are scored on Likert-type scales, comprising up to six response options to maximize the sensitivity of the measure” (p.185). The Pre-Screen total score corresponds to estimated recidivism rates. The YASI Full Assessment “is more comprehensive, gauging additional strength-based factors and dynamic risk factors (or needs). The Full Assessment comprises 90 items (risk, need, and strength factors) across 10 separate domains. The quantitative score yielded from the YASI Pre-Screen is used to assess an offender’s overall risk of recidivism, classifying youth as low, moderate, or high risk. However, the computerized algorithms associated with the full assessment generate separate scores for both the risk/needs and strength components of each domain” (p.185).

The purpose of the current study was to examine the YASI’s internal consistency, predictive utility, and predictive validity. Predictive validity was hypothesized to be moderate to high. It was further hypothesized that the YASI would predict recidivism equally well across gender and ethnic subgroups. Thirdly, it was also hypothesized that the “strength-based content of the tool would enhance predictive accuracy beyond the consideration of risk/need scores. Finally, an exploratory domain-level analysis was conducted to identify those factors most highly predictive of criminal conduct” (p.185).

“Risk assessment intake information reflected in the current dataset was collected between January 2009 and May 2011 and included all youth under community supervision with Alberta Justice and Solicitor General for whom recidivism data were available over an 18-month follow-up period. These initial YASI assessments were completed by probation officers within 45 days of the youth receiving a community sentence” (p.185). “The sample comprised 464 youths on probation (114 female, 350 male), ages ranging from 12.5 to 19.7 years at intake with probation services. Of the total sample, 61.2% were Caucasian, 25.9% were of Aboriginal descent, and 12.9% identified as “Other.” Historically, 58.6% of these youths had engaged in acts of violence” (p.186)..

“The YASI Pre-Screen predicted reoffending with a high level of accuracy…Recidivism rates and risk scores were significantly higher for Aboriginal versus non-Aboriginal offenders as predicted by extant literature. Contrary to our hypotheses, the YASI Pre-Screen achieved only a moderate degree of accuracy in predicting general reoffending for girls, in contrast to the high levels of accuracy achieved for males. However, the assessment protocol forecasted violent reoffending with an equivalent level of accuracy across all demographic subgroups under investigation” (p.188). Although lower than the male participants, the moderate predictive validity for the female subsample is consistent with other risk assessments measures for youth..

For individual and incremental predictive validity of risk and strength aggregate scores on the YASI Pre-Screen, “univariate analyses suggest that both risk and strength scores individually contribute to predicting new offenses committed over 18-months. The multivariate model assessing the unique contribution of risk and strength components indicates that while the risk scale accounts for the majority of the model’s predictive ability, the incremental contribution of the strength scale is trending toward statistical significance” (p.187). “High-strength scores are particularly effective in attenuating recidivism rates. Specifically, the reoffense rate of high- risk/low-strength cases is approximately twice that of high-risk/ high-strength cases” (p. 188).

“Results of the present study, combined with those of other recent investigations, suggest that the quantitative inclusion of strengths in risk assessment is a worthwhile endeavor that is apt to enhance both predictive and case management functions…The YASI’s quantification of strengths is unique among available assessment tools. Specifically, rather than have qualitative indicators of strengths scored on the basis of check-boxes (e.g., see the YLS/CMI), the YASI assigns a quantitative score to the degree of expression of each strength-based factor using a Likert scale. Item scores are then tabulated to produce a total strength score. One could potentially have two high-risk clients that differ widely in their expression of strengths. Given the buffering effect of strengths, a high-risk/high-strength case would effectively be lower risk overall and should arguably be processed differently than a high-risk/low-strength case,” (p.189).

“From a practical standpoint, if assessments provide both risk and strength metrics, case managers and frontline staff have more information for varying their casework strategies to address the unique issues presented by high-risk cases. Rather than treat all high-risk cases equally, the buffering effect indicates that some high-risk clients who lack strengths may need more attention than their high-risk counterparts with documented strengths. Apart from information relevant to classification decisions, employing a strength-based assessment approach can also help mobilize youth in the development of case plans. By presenting strength-based feedback to youth within the case planning context, case managers are in a better position to motivate youth to engage in a positive discussion about the need and desire to make changes, and to identify and leverage positive supports and resources in a youth’s life,” (p.189).

“It is not surprising to observe variations in reliability and validity across samples, potentially due to differences in implementation practice and measurement of recidivism. As a case in point, a survey of 117 juvenile probation officers across several American states outlined perceived obstacles in implementing risk/needs assessment protocols (Guy, Fusco, Cook, & Vincent, 2010). Of those probation officers surveyed, 63% expressed difficulty in rating items, either due to insufficient information or because they felt that item scoring criteria were not sufficiently clear. In addition to improving data gathering practices and potentially introducing clarifications into scoring manuals based on the feedback of frontline workers, the latter observation speaks to the importance of ensuring sufficient training. In order to facilitate item interpretation and scoring, instruction is relevant both at pre-implementation stages and in the form of booster training. Indeed, beyond the content of a risk assessment protocol, there is evidence to suggest that the integrity of a tool’s implementation achieved through appropriate training is essential to producing valid assessments” (p. 188).

“The Mental Health domain of the YASI did not predict criminal outcome, a finding that is also consistent with the mainstream correctional literature that conceptualizes personal distress as a responsivity factor. The reader should note that the Mental Health domain is fairly comprehensive and includes psychosis, bipolar, other mood disorders, suicidal ideation/attempts, thought/personality disorders, and a global category termed “other.” The impact of mental health indicators on juvenile justice outcomes is complex. For example, Hoeve and colleagues found that while the presence of a mental disorder predicted the severity of reoffending in a juvenile justice population, the relationship only held up when mental health disorders co-occurred with substance abuse disorders. Accordingly, the weak association between recidivism and mental health indicators was not surprising. That stated, future research is warranted before discounting mental health as noncriminogenic. There is some evidence to suggest that specific psychopathological symptoms in the form of extreme depression, self-harm, and suicidal tendencies are salient predictors of criminality for females” (p. 190).

YASI-Girls (YASI-G) is in development to better capture a gender-informed approach for this risk assessment tool. “Much of the tool parallels the YASI in its gender-neutral form but further encompasses items deemed foundational to the criminal behavior of young females according to the feminist and gender-responsive literatures. Specifically, the measure includes additional items pertinent to the nature of one’s relationships, levels of emotional expression, self-efficacy, sexual vulnerability (e.g., prostitution), early parenthood, and features a broader array of potential mental health issues…It is arguable that the additional consideration of female-responsive needs and strengths such as those featured on the YASI-G would serve to further enhance levels of predictive accuracy in the risk assessment of young females,” (p.190-191).

As always, please join the discussion below if you have thoughts or comments to add!