Juvenile offenders and experiences of neglect: The validity of the YLS/CMI with dual-status youth

Highlights

• Examine the predictive validity of YLS/CMI with dual-status youth

• The YLS/CMI exhibits differential predictive validity for dual-status youth.

• Dual-status youth may be a unique set of offenders.

Abstract

The extant literature suggests that youth present in both child welfare and juvenile justice systems (dual-status) represent a unique subset of young offenders. Sparse attention has been given to the predictive validity of risk assessment measures with this uniquely vulnerable subpopulation. The focus of this research was the validity of a commonly used recidivism risk/need assessment, the Youth Level of Service/Case Management Inventory (YLS/CMI), with dual-status adolescents. One hundred twenty-eight dual-status youth from a probation division were administered the YLS/CMI and monitored during a twenty-four month follow-up for recidivism. The YLS/CMI demonstrated poor predictive validity for the dual-status youth. Recommendations for future research and policy are presented.

Introduction

Youth who are under the jurisdiction of both juvenile justice and child welfare systems (referred to as crossover youth or dual-status youth) are at greater risk for undesirable and maladaptive behaviors (Cuevas et al., 2007, Ryan et al., 2013). Dual involvement poses a challenge for justice systems that rely on risk assessments in case management and disposition decisions. The fundamental question underlying this challenge is whether risk instruments used by juvenile justice systems are valid delinquency prediction and case management tools when used on dual-status youth. The utility of risk assessment hinges on the validity of the relationship between a broad spectrum of risk factors and recidivism (Dowden and Andrews, 1999, Onifade et al., 2009). Investigations of the risk–recidivism relationship have suggested that certain combinations of risks, in addition to levels of risk can be used to predict future delinquency outcomes (Andrews & Dowden, 2007). This serves the primary function of allowing system officials to steer youth through the juvenile justice system based on likelihood of re-offense (i.e., offenders receive services based on level of need) rather than severity of offense or crime type (i.e., violent offenses, property offenses, drug offenses, etc.). The practice of risk assessment theoretically reduces financial costs for systems, reduces caseloads for practitioners, decreases overall delinquency in the community, and improves youth outcomes (Andrews & Dowden, 2006).

The risk assessment approach rests on the assumption that the risk–recidivism relationship is consistent across communities and population subgroups. However, investigators have identified subgroups (e.g., girls, young sex offenders, and members of marginalized groups), in which popular risk assessment instruments perform differentially (Onifade et al., 2009, Onifade, Petersen, Bynum and Davidson, 2011, Onifade, Wilkins, Davidson, Campbell and Petersen, 2011). Given that maltreated youth may have a heightened risk for delinquency, the risk–recidivism relationship for offenders who have experienced abuse and/or neglect should not be assumed to be congruent with population norms. The existing literature suggests that dual-status youth represent a unique subset of offenders (Bender, 2010, Herz et al., 2010). This study focused on the predictive utility of a commonly used recidivism risk assessment measure with both exclusively delinquent and dual-status adolescents.

There are few national estimates reflecting the number of youth involved in both child welfare and juvenile justice systems. Despite this limitation, the literature is replete with studies evidencing the link between maltreatment and delinquency. For example, estimates indicate that at least 70% of young offenders have experienced some level of childhood and/or adolescent trauma (Greenwald, 2002). Maltreated youth are 47% more likely to engage in delinquent behavior during adolescence than their counterparts (Morris & Freundlich, 2004). Due in part to early onset of delinquent behavior relative to their peers (Rivera & Widom, 1990), these youth commit nearly twice as many offenses and are more likely to be incarcerated as adults (English, Widom, & Brandford, 2004). Young offenders under the jurisdiction of the child welfare system are more likely to be committed to detention than youth who are solely under the jurisdiction of the juvenile justice system (Bender, 2010). English et al. (2004) found substantial disparities in arrest rates for violent crimes among adolescents and adults who were victims of abuse and/or neglect as children. Given their increased risk for delinquency, research indicated that maltreated youth demonstrated higher recidivism rates than their peers (Chang et al., 2003, Herz et al., 2010). This finding is particularly germane to the issue of risk assessment validity with dual-status offenders as the risk–recidivism relationship is different for maltreated youth, giving rise to the possibility of risk assessments underestimating likelihood of re-offense for this subpopulation of offenders.

To the extent considerable evidence exists for the heightened risk for delinquency that maltreated youth exhibit (Bender, 2010, Cuevas et al., 2007); the mechanism for this link is still widely debated and unconfirmed. When groups experience disproportionate contact with the juvenile justice system, the disparity is a product of both differences between the subgroup and the larger offender population with regard to 1) the levels of delinquency within the subgroup (i.e., proximal level maltreatment pathways to delinquency) and 2) differences in the manner in which the juvenile justice system treats members of those subgroups (i.e., distal level maltreatment pathways to delinquency) (Onifade, Petersen, Bynum and Davidson, 2011, Onifade, Wilkins, Davidson, Campbell and Petersen, 2011).

Differential behavior patterns and increased risk of delinquency among maltreated youth may be attributed to proximal-level pathways. To that end, maltreated youth may possess a unique combination of risk factors that are not common to the general young offender population. Ford (2002) suggested that maltreated youth likely resort to negative coping as a matter of survival, justifying violent behaviors in order to protect them from future abuse. For example, maltreated youth are more likely than their peers to become homeless either due to running away or being thrown out of their homes by abusive guardians (Ford, 2002). In addition, maltreated youth experience greater instability in residential placement and consequently endure repeated disruption of their education as they navigate Child Protective Services (Ford, 2002). As a result of trauma/abuse, youth that struggle with internalizing disorders like depression or anxiety experience greater difficulty socializing with their peers and resort to establishing deviant peer-group networks (Schwartz & Protor, 2000), increasing their likelihood of engaging in delinquency. Further, as outlined below, increased contact with the child welfare system may expose youth to increased standards for behavior and increased system monitoring.

While proximal pathways to delinquency (e.g. substance abuse, mental health, and deviant peers) are established risk factors for youth in general, they appear to be doubly so for maltreated youth, resulting in delinquency spanning longer stretches of development than is the case for non-maltreated youth who typically desist from delinquency by the conclusion of adolescence (Halemba et al., 2004, Morris and Freundlich, 2004, Onifade, Petersen, Bynum and Davidson, 2011, Onifade, Wilkins, Davidson, Campbell and Petersen, 2011, Ryan et al., 2007).

There is evidence that increased delinquency risk is also a result of factors pertaining to maltreated youth's concurrent involvement in juvenile justice and child welfare systems. Delinquency risk is further exacerbated by the incapacity of child welfare and juvenile justice systems to effectively cooperate in the processing of dual-status offenders. This results in fewer advocacies on behalf of maltreated youth, and a greater likelihood of these youth being subjected to punitive sanctions as they traverse the juvenile justice system. Ryan et al. (2007) found substantial bias in juvenile justice processing of maltreated youth. Children in foster care were significantly less likely to receive probation than non-foster care youth, and were more likely to be placed in detention facilities, even when controlling for offense type and seriousness (Conger and Ross, 2001, Ryan et al., 2007). Halemba et al. (2004) found similar results, reporting that dual-status youth were nine times as likely as delinquency-only youth to be placed in a detention facility. There is considerable evidence that harsh dispositions that fail to account for the individual needs of the offender actually increase the likelihood of future delinquency (Andrews & Bonta, 2006). The differential treatment of maltreated youth in this way is thus its own risk factor for delinquency (Bender, 2010).

In the span of two decades considerable work has been done to both evaluate and enhance the predictive validity of risk assessment measures (Catchpole and Gretton, 2003, Jung and Rawana, 1999, Onifade et al., 2008, Schmidt et al., 2011, Schmidt et al., 2005, Vieira et al., 2009). To date, systematic reviews of the literature related to risk assessment validity suggest that these measures correctly classify youth in determining likelihood of re-offense (Olver et al., 2009, Schwalbe, 2007, Schwalbe, 2008, Shepherd et al., 2013). Measures that classify youth as having low, moderate, or high risk of recidivism have been demonstrated to reveal substantial and significant differences between the levels of risk in delinquency outcomes (Onifade, Petersen, Bynum and Davidson, 2011, Onifade, Wilkins, Davidson, Campbell and Petersen, 2011, Onifade et al., 2009). Moreover, these risk categories have also differed in time to re-offense outcomes and offense levels (Onifade et al., 2008).

Unfortunately, mounting evidence has suggested that these measures are subject to differential predictive validity with some subgroups of offenders (e.g. females, non-White offenders) (Baglivio and Jackowski, 2012, Onifade et al., 2010, Shepherd et al., 2013, Vincent et al., 2011, Worling and Langstrom, 2003). Although risk assessments include a cross-section of risk factors, groups vary in the degree to which these factors correctly predict their outcomes. In general, validation studies that investigated the relationship between risk score and recidivism have found race as a moderator (e.g., Onifade et al., 2009, Schwalbe, 2007, Vincent et al., 2011), while others found opposing evidence (Jung and Rawana, 1999, Meyers and Schmidt, 2008, Schwalbe, 2008). Findings specific to the predictive validity across race/ethnicity and gender for the Youth Level of Service/Case Management Inventory (YLS/CMI) are also mixed (Betchel et al., 2007; Onifade et al., 2010; Takahaski, Mori, Takemi, Kroner, 2013; Thompson & McGrath, 2012). Onifade et al. (2009) evaluated the predictive utility of the YLS/CMI using White and African American juvenile offenders. The authors found the YLS/CMI total risk score a valid predictor for the full sample and race/ethnic groups, however subscale scores did not perform equally across subgroups (Onifade et al., 2009). The researchers specifically found that the YLS/CMI over-predicted recidivism for White female youth. Family circumstances, education, and personality subscales exhibited significant validity coefficients for African Americans than for White youth.

More research has been dedicated to examining differences in predictive utility across gender and these findings are also contradictory (Betchel et al., 2007; Jung and Rawana, 1999, Olver et al., 2012, Olver et al., 2009, Onifade et al., 2008, Onifade et al., 2010, Schmidt et al., 2005, Schmidt et al., 2011, Schwalbe, 2008). Olver et al. (2012) investigated the predictive accuracy of the YLS/CMI by examining different subgroups within a Canadian sample of serious offenders and found the risk assessment predicted recidivism equally as well for males and females. Conversely, Betchel et al. (2007) indicated that the YLS/CMI performed poorly for the community-based female sample compared to males with an admittedly small proportion of female offenders (13%).

Given that risk assessments are commonly additive in nature (i.e., the more risk factors present the greater likelihood of re-offense), they are relatively simple to administer. Consequently, for certain instruments, each risk factor is assumed to have the same weight, which directly contradicts what is known about the etiological differences in delinquency for certain subgroups of the young offender population (Onifade et al., 2010). More often than not, systems turn a blind eye to differential predictive validity or fail to properly validate these instruments with subpopulations (Baglivio, 2009, Onifade et al., 2008, Schwalbe et al., 2004, Thompson and McGrath, 2012, Vitopolous et al., 2012). For a detailed review see Shepherd et al., 2013.

The aforementioned proximal and distal pathways to delinquency for maltreated youth give rise to concern about the risk–recidivism relationship upon which a valid risk assessment measure would rely in assessing this subpopulation. The Child Welfare League of America (2002) reported that fewer than ten percent of juvenile justice systems utilize valid assessments of abuse and neglect in the processing of offenders. To that end, few jurisdictions have developed co-management protocols for establishing custody and case management of maltreated youth. This speaks to the necessity of developing a means by which to share information between the two systems (Chuang and Wells, 2010, Huang et al., 2012). This means has yet to emerge universally as jurisdictions have taken drastically different approaches to addressing the needs of dual-status offenders (Child Welfare League of America, 2002).

Again, the literature is replete with studies documenting increased risk of delinquency maltreated youth display (Bender, 2010, Cuevas et al., 2007, Mersky et al., 2012). Consequently, juvenile justice systems may underestimate or overestimate the risk–recidivism relationship of maltreated youth if this subpopulation is assessed with a general risk assessment normed on non-maltreated youth. Although the literature does offer a number of assessments that are ideal for identifying risk of abuse/neglect and even risk of negative outcomes useful in developing treatment interventions for internalizing and externalizing behaviors (Bender, 2010), there are no identifiable studies validating general risk assessments that are often used with maltreated youth. However, there are studies that have used general risk assessments to investigate the risk–maltreatment–recidivism relationship with dually involved youth (Herz et al., 2010, Ryan et al., 2013). Herz et al. (2010) and Ryan et al. (2013) used general risk assessments to help identify the effect of being dual status offenders on juvenile justice decision-making and likelihood of recidivism, respectively. Both studies used risk assessment scores to predict likelihood of recidivism, however neither study indicated the predictive utility of these measures with dually involved or crossover youth (Herz et al., 2010, Ryan et al., 2013).

This study attempts to fill a gap in the literature by assessing the validity of a widely used risk assessment, the Youth Level of Service/Case Management Inventory (YLS/CMI), with dual-status youth. The YLS/CMI represents an ideal instrument to validate given its inclusion of risk factors underlying the maltreatment–delinquency relationship (i.e., drug involvement, peer association, personality and attitude, family circumstances, etc.). Moreover, the jurisdiction in which this study took place possesses a means to identify youth with dual-involvement with the juvenile justice and child welfare systems. The family court system in which this study took place includes both a juvenile crime division and an Intensive Neglect Services (INS) division. The INS division targets families who enter the court with juvenile crime charges and also have pending neglect charges or who enter with delinquency charges but court staff identify the youth's needs as originating from family or neglect problems. In this unit, both the youth and the family receive services as the youth are typically at high risk of being removed from the home. Consequently, the goals of the unit are to maintain family unification as well as prevent juvenile delinquency and/or reduce recidivism.

The current study aims to measure the predictive validity of the YLS/CMI when there are comorbid delinquency and neglect risk factors. This study sought to answer the following questions:

• Do dual-status youth have a higher risk score for recidivism than delinquency-only youth?

• What is the predictive validity of the YLS/CMI for dual-status youth?