Risk Assessment Protocol Crucial to Success of Juvenile Probationers

Gina M. Vincent, University of Massachusetts Medical School
Laura S. Guy, Simon Fraser University
Rachael T. Perrault, University of Massachusetts Medical School
Bernice Gershenson, University of Massachusetts Medical School

There is a strong movement toward juvenile justice agencies’ use of risk assessment and risk-need responsivity approaches to improve case management decisions for young offenders. However, little is known about whether adoption of risk assessment actually effectuates any changes in the way young offenders are handled. This was a multisite study of the impact on case processing of implementation of the Structured Assessment of Violence Risk in Youth (SAVRY) or Youth Level of Service/Case Management Inventory in 6 juvenile probation offices using a prepost design and 1,694 propensity score-matched young offenders. Consistent with the risk principle, there were significant changes to at least some areas of case processing in all but 1 site, most notably with respect to decreases in the amount of supervision youth received and in rates of out-of-home placement. The nature and extent of the impact varied as a function of sites’ characteristics and implementation quality, not as a function of the risk assessment used. No increases in recidivism were observed in any site, and there was a significant reduction in recidivism in 1 site. The key benefits of implementation of valid risk assessment and case management procedures were improved resource allocation and fewer instances of inappropriate interference in youths’ lives without an apparent increased risk to public safety.

SAVRY, YLS/CMI, implementation study, juvenile, RNR

“Considerable attention has been directed toward reforming juvenile justice over the past decade by integrating research evidence and principles of adolescent development into practice (National Research Council [NAS], 2013). One prominent recommendation for reform has been to base individual programming decisions on risk and criminogenic needs. The NAS (2013) strongly recommended structured risk and need assessment tools (RNAs) be used to identify low-risk youth who could be handled less formally, to match youth to appropriate treatment, and to target high-risk youth for more intensive interventions. Similarly, the Council of State Governments (Seigle, Walsh, & Weber, 2014) listed use of valid RNAs for supervision, service, and resource-allocation as one of the four core principles for reducing recidivism and improving outcomes for youth. Consequently, most states today have instituted a RNA in juvenile probation for use in case planning (Wachter, 2015). These recommendations stem from evidence that individualized case management models, such as risk-need-responsivity (RNR; Andrews & Bonta, 2003, 2010; Andrews, Bonta, & Hoge, 1990), are effective means for reducing recidivism whereas more global approaches toward punishment are not (e.g., Gatti, Tremblay, & Vitaro, 2009; Lipsey, 2009; Lipsey & Cullen, 2007; Loughran et al., 2009; MacKenzie, Wilson, & Kider, 2001; Petrosino, TurpinPetrosino, & Guckenburg, 2010)” (p. 683).

“Studies of probation officers (POs) have found that the actual use of RNAs in decision-making in the justice field is limited…Very few studies have examined actual changes that result in the handling of probation cases after implementation of a RNA” (p. 684).

“The current study focused on the impact of implementing the risk principle in these decisions, anticipating a decrease in use of restrictive monitoring and incarceration once a RNA and RNR practices were implemented. Specifically, we hypothesized decreases in rates of (a) more restrictive dispositions, (b) out-of-home placements, and (c) more intensive supervision levels. We also hypothesized that, in accordance with the risk principle, (d) restrictive dispositions, placements, and levels of supervision would be positively related to risk, and (e) high-risk youth would receive more services than lower risk youth. Lastly, we hypothesized that (f) reoffending would not increase after implementation of the RNA and RNR” (p. 685).

A pre-post, quasi-experimental design was utilized to collect data from 6 county probation offices and 1,694 propensity score-matched offenders. The sample included cases that met the criteria for a RNA to be implemented. Juvenile Probation Officers (JPOs) were trained in risk/need assessment and were told about the implementation procedures. Then, policies for risk/need assessments and case planning were created. Administrators tailored the case plans to be consistent with criminogenic needs identified on the RNA. Finally, JPOs were trained on the fundamentals of the RNR approach to treatment, and completed workshops on the YLS/CMI or SAVRY. Each JPO used information from the juvenile’s file and interviews with the juvenile, parent, and juvenile and parent together. Semi-structured interview scripts were given to JPOs who also administered the SAVRY or YLS/CMI. The study examined four dependent variables: most restrictive disposition, out of home placements, community supervision, number of service referrals, and recidivism.

Results

“There was a significant shift toward less restrictive dispositions in four of the five sites. Risk level was associated with at least some, if not all, disposition decisions in each of these sites, indicating the risk principle was followed” (p. 693).

“As hypothesized, the risk principle influenced placement decisions in all five sites. On a positive note, most sites placed only 50 to 75% of their high-risk youth, suggesting the RNA and RNR training was effective in communicating that many of these youth could be managed safely in the community. A crucial takeaway is that the direction of the impact of the RNA differed as a function of each site’s placement rates before the RNA was used” (p. 693).

“The most consistent area of impact of implementation of a RNA was on level of probation supervision. In every site at which supervision level could be tested (with the exception of YLS/CMI Site 3) there were significant decreases in the use of maximum and moderate levels of supervision, and increases in the use of minimum levels of supervision…Most sites followed the risk principle in their service allocation such that high-risk youth on average received one to three more services than low-risk youth” (p. 693).

“Consistent with the hypotheses, in most sites rates of new petitions [rates of recidivism] did not change [except for site 1]… The stability in recidivism rates after implementation of a valid RNA and RNR approach may be a surprise and a disappointment. After all, the primary benefit of the RNR approach has been touted as recidivism reduction. However, most studies that have reported RNR leads to reductions in recidivism have demonstrated this at the macro level for services that address criminogenic needs (Dowden & Andrews, 1999; Romani, Morgan, Gross, & McDonald, 2012), for young offenders with strong service-to-need matching (Luong & Wormith, 2011; Peterson-Badali et al., 2015; Vieira et al., 2009), or in comparisons between POs with and without intensive training in RNR-related case management (Bonta et al., 2013, 2011). The present study differed in that it examined whether implementation of a valid RNA with RNR-related policies led to reductions in reoffending within a jurisdiction as a whole” (p.694).

Although it is important to implement the RNR approach in juvenile justice agencies, “we stress that focusing on recidivism as the most important or even sole outcome variable of interest when studying RNA implementation could be detrimental. Future studies should examine other outcomes such as reduction in agencies’ human and financial costs and improvement in youths’ educational attainment or employment” (p. 695).

“A major implication of this study is that quality implementation of risk assessment and RNR will conserve resources and serve more justice-involved youth in the community without an increased risk to public safety…Another implication of this work is that agencies must implement the RNA well to reap its benefits and avoid losses of time and other resources” (p. 694).

“A final implication is that the actual RNA instrument used should not matter as long as the tool is being completed reliably by staff and has been demonstrated to be a valid predictor of delinquent behavior for the type of population and setting where it is being used” (p. 694).

The present study did not examine the quality of implementation of the need principle at any of the sites when measuring recidivism. Although the JPOs were trained, there was no standard assessment or evaluation for program delivery or management. Additionally, the authors speculated that sufficient time was not given to see any significant reductions in reoffending. “Good implementation of a new intervention can require 3 years, particularly before any benefits can be realized (Fixsen et al., 2005; Flores et al., 2006)” (p. 695). Finally, developing evaluations for all three stages of the RNR model during the time of implementation would allow future researchers to isolate both the efficient and inefficient aspects of the treatment.

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