Domains of the YLS: Do Some Matter More than Others?

Kirsten L. Witherup^a, PJ Verrecchia^a

ABSTRACT

Data were obtained in Pennsylvania to examine referrals to juvenile courts in the Commonwealth for the 2021-2022 school year that include any felony, misdemeanor or summary offenses that occurred on school property. This study looked at the referrals to see if any of the individual domains of the Youth Level of Service Case Management Inventory 2.0 (YLS/CMI) had more of an effect on how the case was disposed of, which was defined as dismissed or handled informally, or formally adjudicated, in juvenile court. The research found that while the overall YLS risk score influenced juvenile court outcomes, only five of the eight individual domains of the YLS had a significant effect on how a case was processed in juvenile court, which calls into question how closely juvenile probation officers are following the instrument in their practice.

KEYWORDS: Delinquency, treatment, prediction, regression analysis

Received March 2023; Accepted July 2023; Published July 2023             DOI: 10.52935/23.231020.7

DOMAINS OF THE YLS AND JUVENILE COURT DISPOSITION

In the juvenile court in Pennsylvania assessment of juvenile offender’s risk to the community traditionally was conducted using a clinical assessment. A juvenile probation officer would interview an alleged juvenile delinquent and their family members, and then review relevant documents (like school records), before determining the appropriate disposition. If the disposition was out of home placement, a psychological, or less frequently, a psychiatric, evaluation was conducted. However, this was after the clinical assessment by the probation officer deemed that circumstances warranted the youth being removed from the home. Therefore the assessment of any risk the delinquent posed was based on the professional using an unstructured interview process (Gottfredson & Moriarty, 2006). However, studies found that this “unstructured clinical judgment” (Perrault et al., 2012, p. 487) led to poor accuracy rates and equally poor decisions (Borum & Verhaagen, 2006; Calley & Richardson, 2011).

The advent of risk assessment instruments led to more systematic and empirically based judgments of future risk a juvenile delinquent posed, and therefore appropriate intervention strategies (Chu et al., 2014; Grove & Meehl, 1996). The Youth Level of Service Case Management Inventory (YLS/CMI) is a screening tool designed to estimate a juvenile offender’s level of risk and treatment needs relative to various criminogenic factors (Witherup & Verrecchia, 2022), and it is one of the most widely used risk and needs assessment measures not only in America but also Canada, Australia, Japan, the United Kingdom, and Singapore (Chu et al., 2014; Schwalbe et al., 2007). The YLS/CMI uses a face-to-face interview with the juvenile offender and any relevant family members, as well as a review of any pertinent documents to make an assessment (Cuervo & Villaneuva, 2018; Flores et al., 2004). There are 67 counties in Pennsylvania and all of them use the YLS, which was implemented in four phases between 2009 and 2012 across the Commonwealth (Miller et al., 2021). Pennsylvania is a county based juvenile justice system which means that it is administered at the local level and not from the state capitol (Witherup & Verrecchia, 2022). The Juvenile Court Judges’ Commission, the administrative state agency for Pennsylvania’s juvenile justice system, encouraged the counties in Pennsylvania to adopt the YLS as, a “cornerstone of Pennsylvania’s Juvenile Justice System” (Pennsylvania Juvenile Court Judges’ Commission, 2020, p. 1), and they all did.

The YLS/CMI consists of 42 items that relate to the “Central Eight” risk and needs domains, which are prior and current offenses, antisocial attitudes, antisocial friendships, and an antisocial personality (considered “the Big Four”), plus family circumstances, education/employment, substance abuse, and leisure recreation (Witherup & Verrecchia, 2022). The “Central Eight” domains are associated with criminal offending according to the general personality and cognitive social learning (GPCSL) perspective (Andrews & Bonta, 2010). They contain between three (leisure/recreation) and seven (personality and behavior; education/employment) statements that are answered either yes or no. For every yes one point is assigned, and no points are assigned for every no. Under personality and behavior the statements are inflated self-esteem; physically aggressive; tantrums (temper); short attention span; poor frustration tolerance; inadequate guilt feelings; and verbally aggressive, impudent. A score of 5-7 in this domain is considered high risk, a score of 1-4 is moderate risk, and a score of 0 (answer no to all of the statements) is low risk. The higher the total score on the YLS/CMI (out of a possible 42), the higher the risk. For males a score of 29-42 is considered very high risk, and for females very high risk is a score of 32-42. 

The appropriate juvenile court intervention should be based on the delinquent’s level of risk. Higher risk youth will benefit from more services and they and society would benefit from more restrictions on their freedom (Andres & Dowden, 2006; Andrews & Bonta, 2010). The YLS/CMI not only predicts risk but identifies areas of treatment as well, which if addressed will result in less risk (Flores et al., 2004).

Review of the Literature

Witherup and Verrecchia (2022) examined the relationship between score on the YLS/CMI and how long juvenile delinquents spent in a residential treatment facility. They hypothesized that a delinquent with a higher YLS/CMI score (deemed high or very high risk) would spend more time in a treatment facility than a delinquent with a lower score. Using a sample of 152 juvenile delinquents from a county in Pennsylvania who were sent to a residential treatment facility between 2017 and 2020, Witherup and Verrecchia found no relationship between score on the YLS/CMI and placement length (2022). 

Like the current study, Flores et al. (2004) not only examined the predictive ability of the YLS but the individual domains as well. Using a sample of juvenile delinquents in Ohio, Flores and colleagues (2004) investigated the predictive ability of the YLS for delinquents in residential settings and under supervision of the probation department. Using follow up data two years after the initial assessment and interviews with correctional staff regarding their perceptions of the YLS, Flores et al. (2004) found that “relatively few of the forty-two items contribute to accuracy in risk classification” (p. 1). Of the eight domains, only three (substance abuse, attitudes/orientations, and prior and current offense) were related to case outcomes. They concluded that for initial risk classification, the YLS “may not be an appropriate instrument” (Flores et al., 2004, p. 2). 

In 2007 Schwalbe conducted a meta-analysis of 11 YLS studies that focused on recidivism of juvenile delinquents, including violations of probation and new offenses. Schwalbe (2007) found that higher scores on the YLS were associated with an increase in recidivism. In another meta-analysis, Olver et al. (2009) found that the YLS had lower predictive validity for recidivism. Examining the psychometric properties of the YLS, Van de Ven (2004) found that the YLS was a reliable predictor of recidivism, although the effect sizes were small.

Perrault et al. (2012) examined whether the implementation of a risk needs assessment (RNA) for juvenile delinquents influenced probation officer recommendations. They addressed whether the self-reported practices of the juvenile probation officers matched their actual case management decisions in an office in a northeastern state which used the YLS/CMI, and two in a southern state which used a different RNA, the Structured Assessment of Violence Risk in Youth (SAVRY) (Perrault et al., 2012). The SAVRY uses a structured clinical assessment and not an actuarial approach to determine a juvenile’s risk (Borum et al., 2006). Almost three quarters (73%) of the probation officers who were interviewed for the research stated that they based their recommendations on risk level, while the remaining officers (27%) reported that they did not (Perrault et al., 2012). 

While the state government strongly encouraged use of the YLS/CMI across Pennsylvania, the implementation was left up to individual county juvenile probation offices. Using a mixed method approach that utilized both interviews (n=86) and surveys (n=117), Miller et al. (2021) examined the implementation process in five counties in Pennsylvania. They found that officer attitudes were “more positive than negative” towards the YLS/CMI specifically, and risk/needs assessment instruments generally (Miller et al., 2021, p. 219). However, there were officers who were skeptical and said that the YLS/CMI is not as effective in determining risk or treatment needs as the experiences and judgement of individual officers. Others shared concerns about the extra work that completing the YLS/CMI entailed (Miller et al, 2021). However, while most officers completed the YLS/CMI, it appears that fewer used it to inform their decisions about client care (Miller et al., 2021), which would suggest a disconnect between completion of the YLC/CMI and its actual use.

It is acknowledged in juvenile justice that boys and girls have different risk factors when it comes to delinquent behavior (see Pusch & Holtfreter, 2018). Kitzmiller et al. (2022) examined 2,384 youths who were adjudicated delinquent in juvenile court in a midsized Midwestern county. The sample over one quarter (26.2%) female and under three quarters male (73.8%). Using Multi-group confirmatory analysis (CFA), Kitzmiller and colleagues found that the YLS/CMI was a solid predictor of risk, and it was adequate for assessing risk differently for boys and girls, since both groups differed in their risk profile (2022). 

As stated earlier, the YLS/CMI is used internationally, and Cuervo and Villaneuva (2018) conducted a study to see if this screening instrument could predict recidivism in juvenile offenders in a Spanish province. The YLS/CMI was completed by a “technical team in the juvenile court” who had been trained on the use of the YLS/CMI for a month (Cuervo & Villanueva, 2018, p. 3566). The juveniles were then measured two years after the initial assessment and it was defined as any charge the juvenile received after that evaluation. Using a sample of 382 juvenile offenders (mean age 16.33 years), it was found that boys were more likely to recidivate than girls. In addition, the juvenile recidivists had a higher mean score on the YLS/CMI than the non-recidivists and scores on the YLS/CMI were correlated with the number of charges during the two year follow up period (Cuervo & Villanueva, 2018). In other words, the YLS/CMI was a significant predictor of recidivism. Of the Central Eight factors the three that were the most significantly correlated with recidivism were school and employment problems, antisocial personality, and delinquent peers.

Chu and colleagues (2014) looked at the validity of the YLS/CMI-SV (Youth Level of Service/Case Management Inventory-Screening Version) as a tool for predicting recidivism of delinquents. Using a sample of 3,264 juveniles who were referred for probation services in Singapore from 2004-2008, Chu et al. concluded that the YLS/CMI-SV was a significant predictor of recidivism at one year, three years, and five years post intervention for both male and female delinquents (2014).

Gomis-Pomares et al. (2022) examined the predictive ability of the YLS/CMI as it relates to recidivism for a group of Roma and non-Roma juvenile delinquents in Spain. There were 88 Roma youth and 135 non-Roma youth for a sample size of 223 (n=223). There was no difference in the average age (Roma m=15.86, sd=1.09; non-Roma m=15.88, sd=1.03) and genders (Roma=68.2% male; non-Roma=66.7% male) of the two groups. Gomis et al. conducted this study to determine if there are differences in the predictive ability of the YLS/CMI for minority groups (2022). Recidivism was measured as reoffending or not, and by the number of delinquent charges after the initial YLS/CMI assessment conducted by the justice department in a province in Spain. Results showed that while the Roma youth reoffended more than the non-Roma youth in terms of number of subsequent delinquent acts, there was no difference between the groups on the dichotomous measure of recidivism (Gomis et al., 2022). Furthermore, the YLS/CMI proved to be a reliable predictor of recidivism, correctly predicting recidivism for 75.3 percent of the Roma group and 75.9% of the non-Roma group.

McGrath et al. (2018) examined the predictive ability of a revised version of the YLS/CMI that is specific to delinquent youth in Australia, the YLS/CMI-AA (Australian Adaptation). They examined almost 5,000 (n=4,887) juvenile offenders as well as an in-depth qualitative analysis of 26 case files. McGrath et al. found that while there were differences in reoffending among boys and girls (girls had higher recidivism rates) and Indigenous and Ethnic youth (Indigenous youth had higher recidivism rates), the “predictive validity [of the YLS/CMI-AA] remained robust” (2018, p. 835).

In 2021, Huang et al examined the applicability of the YLS/CMI with a sample of Indigenous (n=205) and non-Indigenous (n=193) youth in Canada to test the “cross-cultural applicability of forensics practices” (p. 503). The sample of 398 juvenile delinquents in Ontario, Canada, were individually matched on gender and age to examine how well the YLS/CMI assessed risk of recidivism, which was operationalized as “being convicted of one or more offenses within a three year follow up period” from their initial assessment (Huang et al., 2021, p. 506). The three-year recidivism rate for the sample was 70%, but the Indigenous youth had a higher rate than the non-Indigenous youth (82% to 58%). There were no differences in the type of re-offenses. The differences in reoffending were mainly in the low and moderate risk categories, while there were no differences in the high-risk category (Huang et al., 2021). Huang and colleagues (2021) found that the YLS/CMI was a better predictor of recidivism for the non-Indigenous youth than the Indigenous youth, raising questions about its cross-cultural applicability.

The purpose of the current study is to examine the relationship between score on the individual domains of the YLS/CMI and its relationship with disposition decisions in juvenile court, specifically, if the referred juvenile’s case was dismissed or handled informally, or handled formally. It would stand to reason that delinquents with a higher YLS score (high risk) would require a formal disposition because (1) they need more services than a delinquent who is moderate or low risk and (2) they pose a greater risk to the community. Most of the research around the YLS/CMI looks at its predictive ability in regards to disposition length and recidivism. The current study looks inside the YLS/CMI to see if one or more of the “Central Eight” domains is a better predictor of how a case is disposed of in the juvenile court than another.

METHODS

To continue reading, download the full open access article.