Volume 5, Issue 1 • Spring 2016

Table of Contents

Editor's Note

Trauma-Informed Collaborations Among Juvenile Justice and Other Child-Serving Systems: An Update

Looking Forward: A Research and Policy Agenda for Creating Trauma-Informed Juvenile Justice Systems

Psychosocial Interventions for Traumatized Youth in the Juvenile Justice System: Research, Evidence Base, and Clinical/Legal Challenges

Acute and Chronic Effects of Substance Use as Predictors of Criminal Offense Types Among Juvenile Offenders

Examining the Influence of Ethnic/Racial Socialization on Aggressive Behaviors Among Juvenile Offenders

Assessing Probation Officers' Knowledge of Offenders with Intellectual Disabilities: A Pilot Study

Gender and Adolescents’ Risk for Recidivism in Truancy Court

Gender and Adolescents’ Risk for Recidivism in
Truancy Court

Valerie R. Anderson, Department of Pediatrics, Indiana University School of Medicine; Ashlee R. Barnes, and Nordia A. Campbell, Department of Psychology, Michigan State University; Christina A. Campbell, School of Criminal Justice, University of Cincinnati; Eyitayo Onifade, School of Social Work, Florida State University; William S. Davidson, Department of Psychology, Michigan State University.

Correspondence concerning this article should be addressed to: Valerie R. Anderson, 410 W. 10th Street, Suite 1001, Indianapolis, IN, 46202. E-mail: valeande@iu.edu

Keywords: truancy, gender, risk assessment, status offenses, recidivism


The current study investigated the predictive validity of the Youth Level of Service/Case Management Inventory (YLS/CMI) by gender with a sample of youth involved in truancy court (N = 911). The results indicate that the YLS/CMI is a valid predictor of recidivism for truant offenders in general; however, the measure did not predict the recidivism of truant girls. The YLS/CMI is a significant predictor of future delinquency for both boys and girls but is not a significant predictor for future truancy court petitions across gender. The results suggest the need to explore truancy-specific and gender-responsive risk assessment instruments for truancy court–involved youth.


Truancy has long been an issue for the school and juvenile justice systems (Henry, 2007; Maynard, McCrea, Pigott, & Kelly, 2013; Zhang, Katsiyannis, Barrett, & Willson, 2007). Many youth who are truant from school are likely to engage in risky behaviors such as drinking alcohol, using drugs, and having unprotected sex (Dembo & Gulledge, 2009; Dembo et al., 2012, 2014; Henry & Huizinga, 2007a, 2007b; Zhang et al., 2007). Within the context of the justice system, truancy is viewed as a status offense, which is an act that is only considered illegal if committed by a juvenile (Sickmund & Puzzanchera, 2014; Stahl, 2008; Zhang et al., 2010). The other major categories of status offenses include running away, curfew violations, incorrigibility (e.g., being “out of control”), and liquor law violations (e.g., underage drinking) (Sickmund & Puzzanchera, 2014; Stahl, 2008). In this study, we investigate the validity of a widely used criminogenic risk assessment instrument among a sample of youth who were referred to a truancy court intervention. In particular, we examine gender differences in the predictive validity of the assessment on truancy and delinquency (e.g., crimes committed by people younger than age 18) recidivism types.

Literature Review

Truancy is generally defined as chronic absenteeism from school or classes without authority, and truant youth may be handled formally (e.g., processed) or informally (e.g., diverted) by the juvenile justice system (DeSocio et al., 2007). There is a lack of consistency and uniformity in defining truancy among researchers, because most school districts and local juvenile courts have their own definitions of truancy and policies in response to truancy (Gentle-Genitty, Karikari, Chen, Wilka, & Kim, 2015; Reid, 2014), and how cases are handled by the juvenile justice system varies widely (Dembo et al., 2014). Researchers also have different definitions in their studies of truancy’s effects. For example, Barry, Chaney, and Chaney (2011) defined truancy as deliberately skipping school; Epstein and Sheldon (2002) operationalized truancy based on rates of school attendance; Hendricks, Sale, Evans, McKinley, and DeLozier Carter (2010) identified truants by the percentage of hours they spent in school each semester; Lawrence, Lawther, Jennison, and Hightower (2011) defined truancy as students who were absent from school 15 or more times in the school year. This variability had led to issues for researchers who are attempting to synthesize results from studies on truancy (Gentle-Genitty et al., 2015) and for juvenile justice experts in determining how truancy cases are specifically handled across the country (Dembo et al., 2014).

In the United States, truancy cases account for the largest proportion of formally handled status offenses (Sickmund & Puzzanchera, 2014). In 2010, more than 49,000 truancy cases were processed nationally out of an estimated 137,000 status offenses (Sickmund & Puzzanchera, 2014). For formally handled truancy cases, youth ages 14 to 17 represented almost five times more cases per 1,000 than youth ages 10 to 13 (Sickmund & Puzzanchera, 2014). As for gender, boys and girls had similar trends in truancy cases referred to juvenile court between 1995 and 2010 (Sickmund & Puzzanchera, 2014). Except for one racial/ethnic group—American Indian/Alaskan Native youth—truancy is the most common status offense among youth of all race/ethnicities (Sickmund & Puzzanchera, 2014; Stahl, 2008).

Based on survey research with informally processed truant youth—taken from a community sample rather than a juvenile justice sample—Attwood and Croll (2006) reported that students from families with high socioeconomic status (SES), defined by income and education levels, are less likely to engage in truancy than their low-SES counterparts. These differences may be due to high-SES parents being more engaged in their children’s school as well as the promotion of ideas related to academic success from parents and school systems (Zhang et al., 2010). Similar results on the impact of SES on truancy were reported in Henry’s (2007) study using the Denver Youth Survey, which examined the relationship between truancy and several correlates, such as level of parental education and mother’s employment status, which can indicate SES level; academic grades; and peer relationships. After comparing truancy rates of students whose mothers had a college degree to those of students whose mothers had lower levels of education, researchers found that truancy rates were significantly lower for those students whose mothers had a college degree (Henry, 2007).

Researchers looked into other factors that could impact truancy. In her study, Henry (2007) found that the strongest predictors of truancy were delinquent peers and poor school performance. Hunt and Hopko (2009) examined contextual factors predicting truancy among surveyed high school students and identified poor academic performance, depression, and a less-structured home environment as the strongest predictors of truancy. It is important to note that most studies on truancy use community and school-based samples or informally processed youth (e.g., diverted from the juvenile justice system). This is a crucial distinction, because many factors distinguish school-based samples from samples of youth handled formally by the juvenile justice system, and there are few studies that specifically examine truancy with the juvenile justice population.

Gender Differences in Truancy

Though national statistics demonstrate nearly equal trends of new truancy cases in juvenile courts across gender (Sickmund & Puzzanchera, 2014; Stahl, 2008), these trends can vary by geographic region, jurisdiction, and divisions of court. For example, Onifade, Nyandoro, Davidson, and Campbell (2009) found that compared to other court divisions (e.g., standard delinquency or the intake division), there was a disproportionately higher number of girls in the truancy division.

Scholars have studied the role that gender plays in predicting truancy recidivism, and this research has produced mixed findings. In a study that investigated the predictive validity of the Youth Level of Service/Case Management Inventory (YLS/CMI) with formally processed truancy cases, Onifade and colleagues (2009) found that gender, criminogenic risk score, and criminogenic risk level were not significant predictors of future truancy. Zhang and colleagues (2007) compared the risk profiles of formally processed truant youth and delinquent (nontruant) youth. When examining demographics, they found that gender was a significant risk factor for recidivism. More specifically, truants who were male and younger at the time of the first offense were more likely to recidivate. In contrast, Henry and Huizinga (2007b) investigated gender as a covariate in their study on the relationship between truancy and substance abuse (i.e., marijuana, alcohol, and tobacco use) and found differing results. Although there were no gender differences in the onset of alcohol or marijuana use, the relationship between truancy and the onset of tobacco use was stronger for boys than girls (Henry & Huizinga, 2007b). Overall, the literature on the relationship between gender, delinquency, and truancy has demonstrated that boys are at higher risk for engaging in delinquency and truancy reoffenses.

Risk Factors for Truancy

In addition to gender, there are other risk factors associated with initial truancy and truancy reoffenses. The most cited risk factors include individual and demographic variables, family and school, economic influences (e.g., low SES), and educational variables (Dembo et al., 2012; Nolan, Cole, Wroughton, Clayton-Code, & Riffe, 2013; Zhang et al., 2007). Family risk factors include a lack of adequate parental supervision, youth and family substance abuse, and domestic violence. Risk factors associated with the family system can manifest themselves in many ways. For instance, researchers found that students who had zero or limited unsupervised time after school were significantly less likely to engage in truant behaviors (Henry, 2007).

Additional studies have investigated demographic characteristics associated with truancy recidivism and found that truants who were male, racial/ethnic minorities, younger in age at the time of the first truancy offense, and enrolled in special education courses were at higher risk to commit new truancy offenses (Nolan et al., 2013; Zhang et al., 2007). Student-specific variables such as substance abuse and lack of social skills are also risk factors for truancy (Henry & Huizinga, 2007a, 2007b; Hunt & Hopko, 2009; Zhang et al., 2007, 2010).

Macro-level systems (e.g., neighborhood, school policies) can also impact the onset of truancy or truancy reoffenses. Important school and educational variables include school size, flexibility of learning environments, and strict consequences (e.g., at-home suspension) for chronic absenteeism (Zhang et al., 2007). Though it is much easier to blame truancy solely on the truant youth’s individual characteristics, research indicates that students are less likely to consistently attend classes if they perceive the teachers as uncaring or the school as unsafe, or if levels of student disengagement are especially high (Henry & Huizinga, 2007a). These issues may also intersect with the socioeconomic status of the school and the availability of resources (e.g., new textbooks, healthy food options).

It is important to note that most research describing truancy risk factors describe factors that influence the initial onset of truancy and not necessarily repeat truancy or future delinquency. Although there is some research on factors that predict repeat truancy and delinquency recidivism (see Dembo et al., 2012; Dembo et al., 2014; Onifade et al., 2009; Zhang et al., 2007, 2010), most of the truancy research is investigated with community and school-based samples.

Problems Associated With Truancy

Truancy has long-term and short-term negative consequences that can impact youth’s health, education, and social development. In the long-term, truant youth are more likely to be incarcerated, unemployed, and in unstable marriages as adults (Henry, 2007). Short-term negative outcomes include poor academic performance (Zhang et al., 2007, 2010), increased risk of school dropout (Henry & Huizinga, 2007a), engaging in risky behaviors (e.g., unprotected sex, substance abuse, driving under the influence, violence; Bazemore, Stinchcomb, & Leip, 2004; Dembo et al., 2012; Henry & Huizinga, 2007b), and increased likelihood of being formally processed by the juvenile justice system (Zhang et al., 2010).

In a recent study, Dembo and colleagues (2012) identified subgroups of truant offenders using latent class analysis (a statistical method to categorize people based on observed characteristics). Overall, truant youth reported juvenile justice system involvement, mental health problems, and substance abuse issues. Expectedly, the high-risk subgroup (28%), having higher levels of justice involvement, substance abuse, and mental health issues, demonstrated significantly higher levels of these characteristics/negative outcomes than the low-risk subgroup (Dembo et al., 2012). It is important to note, however, that many youth who are arrested for a status offense are not formally involved with the juvenile justice system (e.g., status offenders are commonly diverted from the system). Nonetheless, truancy is considered a developmental pathway to delinquency (Polansky, Villanueva, & Bonfield, 2008; Sickmund & Puzzanchera, 2014).

Zhang and colleagues (2010) examined this developmental pathway when they investigated the differences between juvenile offenders whose first offense was truancy and those whose first offense was a delinquent act (e.g., assault, larceny). Compared to those with delinquency initial offenses, truancy-first offenders more frequently received probation referrals and commitments to secured facilities (Zhang et al., 2010). Given the negative outcomes associated with being truant, researchers must work to identify factors that increase the likelihood that a youth will become a repeat offender. One such strategy is risk assessment.

Risk Assessment

Risk assessment instruments are composed of criminogenic risk factors designed to predict future delinquency (Onifade et al., 2008a). These risk factors include association with delinquent peers, lack of involvement in organized activities, negative attitudes toward authority, substance abuse, low achievement, unstable family structure, and antisocial personality characteristics (Cottle, Lee, & Heilbrun, 2001). Over the past few decades, risk assessment tools have markedly improved the prediction of recidivism. They have progressed from first-generation instruments that relied on the experiential judgment of clinicians, to fourth-generation risk assessment instruments that are composed of several factors, including dynamic (conditions that can change over time) and static (measures of prior delinquency) factors, and risk, need, and responsivity factors (show the person’s readiness for change and ability to respond to particular treatments and programs) (Andrews, Bonta, & Wormith, 2006). Studies have shown that employing assessments that target core criminogenic risk factors can significantly reduce recidivism (Lipsey, Howell, Kelly, Chapman, & Carver, 2010). Risk assessment instruments are important because they are designed to standardize probation and placement decision-making so that assessed juvenile offenders are treated primarily based on their level of risk for future delinquency (Onifade et al., 2008a). These instruments can also be used to develop specific case management plans to reduce recidivism for assessed youth.

The YLS/CMI is a widely used risk-assessment instrument that has been demonstrated to accurately predict recidivism risk (Bechtel, Lowenkamp, & Latessa, 2007; Catchpole & Gretton, 2003; Flores, Travis, & Latessa, 2003; Onifade et al., 2008a, 2008b; Schmidt, Campbell, & Houlding, 2011; Schmidt, Hoge, & Gomes, 2005). Further, the YLS/CMI was designed to be a universal assessment tool for juvenile court systems (e.g., universal use for all offenders regardless of crime type, age, gender, race/ethnicity). YLS/CMI is a multidimensional assessment comprising the static and dynamic factors that best predict criminogenic risk for recidivism (Andrews et al., 2012; Onifade et al., 2009; Schwalbe, 2007). Subscales include prior and current offenses, education, leisure and recreation, family and parenting, substance abuse, personality and behavior, attitudes and orientation, and peer relationships (Andrews et al., 2012). The purpose of the assessment is to uncover areas of need so youth can receive services for those needs. It also uses a low-, moderate-, and high-risk classification system that accurately predicts the potential for recidivism at each level (Onifade et al., 2009).

The YLS/CMI is a valid classification tool for assessing juvenile risk for recidivism (Catchpole & Gretton, 2003; Flores et al., 2003; Onifade et al., 2008a). Onifade and colleagues (2008a) identified significant differences in offense rates and time to recidivism across risk levels that were determined by the YLS/CMI (e.g., youth classified as high risk on the YLS reoffended at a faster rate than those classified as low or moderate risk). Bechtel and colleagues (2007) also found that the YLS/CMI accurately predicted recidivism for juveniles in the community (e.g., probationers) and those in institutions (e.g., detention), demonstrating more accurate predictions for community-based offenders.

Few studies have investigated the ability for any risk assessment instrument to specifically predict future truancy. The only study we found was one by Onifade and colleagues (2009); they investigated the predictive validity of the YLS/CMI with a sample of truant offenders. The researchers aimed to determine whether the YLS/CMI was a valid predictor for truancy, and they looked at which criminogenic risk typologies could be identified among truant youths (Onifade et al., 2009). The researchers found that neither risk level nor risk score significantly predicted truancy recidivism. In addition, five subgroups of offenders emerged with distinct criminogenic risk typologies (minimal risk, antiauthority risk, drug-involved peer risk, court-involved group, and comprehensive-risk group); nearly half of the truant offender sample belonged to the minimal risk group. Interestingly, those with the highest rate of truancy recidivism belonged to the minimal risk group as well. In addition, there were two moderate-risk groups with similar offense rates (but different risk profiles), and two high-risk groups with high criminogenic risk and high rates of delinquent reoffenses but low rates of truancy reoffenses (Onifade et al., 2009). The study concluded that the YLS/CMI was not a good risk assessment for predicting repeat truancy, but it performed adequately in predicting delinquency among first-time truancy offenders.

YLS/CMI and Gender

Although some studies have reported that the YLS/CMI predicts delinquency recidivism equally across gender (Catchpole & Gretton, 2003; Flores et al., 2003; Onifade et al., 2008a, 2008b; Schmidt et al., 2011, 2005), others have reported nonsignificant findings (Bechtel et al., 2007; Onifade et al., 2009). For instance, the comparison of YLS/CMI scores based on gender is equivocal at best, with some findings suggesting that girls exhibit significantly lower risk than boys (Onifade et al., 2008a), and some findings suggesting that girls tend to score higher than boys (Flores et al., 2003). Furthermore, Flores and colleagues (2003) found gender differences across the eight domains of the YLS/CMI.

Overall, studies that investigated truancy, gender, and justice system involvement reported that girls were less likely to be rearrested (Flores et al., 2003); the distribution of girls differed across YLS/CMI criminogenic risk profiles in that girls were overrepresented in the low-risk group (Onifade et al., 2008b); and in general, the YLS/CMI is better at predicting risk of recidivism among boys (Schmidt et al., 2011). In a recent meta-analysis, Schwalbe (2008) examined 19 studies (4 of which utilized the YLS/CMI) that specifically investigated the predictive validity of risk-assessment tools across gender. The effect sizes in gender differences for the YLS/CMI studies were not statistically different between boys (r = .32) and girls (r = .40). Schwalbe (2008) concluded that although risk assessments effectively predicted recidivism for female offenders, there was evidence of gender bias (e.g., practitioners scoring girls systematically higher than boys on criminogenic risk measures) in juvenile justice processing and decision making.

Current Study

The variability in the previous studies examining the predictive validity of the YLS/CMI for delinquent youth calls into question the extent to which the YLS/CMI adequately predicts general recidivism for both boys and girls. Previous researchers have investigated gender differences in risk assessment and delinquency but not in the context of a truancy court intervention. In particular, our study used an innovative approach by examining the gender-based validity of the YLS/CMI to understand overall recidivism as well as delinquency and truancy recidivism. Our study provided an in-depth examination of specific subscales of the YLS/CMI to determine if certain subscales are better predictors of recidivism by gender than others, or if the overall risk score is a better predictor of recidivism by gender. Previous studies have demonstrated that the YLS substance abuse subscale is a stronger predictor of female juvenile recidivism (Andrews et al., 2012) and the family subscale is a stronger predictor of recidivism for female juvenile offenders (Onifade et al., 2009). Thus, further research is needed to investigate the role of gender in the YLS/CMI’s predictive validity, particularly for truant youth. Because girls tend to comprise at least half of truancy petitions (Onifade et al., 2009), our current study focuses primarily on gender differences among truant youth. Given the paucity of research on the topic of gender, risk assessment tools, and truancy, it is clear that further research on gender and risk assessment in the context of truancy is necessary for both research and intervention purposes. Therefore, our study aims to fill this gap in the literature by examining gender and truancy in relation to the YLS/CMI.

Our research site was the family division of a juvenile court in a midsized, midwestern county with three major units: intake, truancy, and delinquency. Youth at intake are generally low-risk, first-time offenders, and youth who are supervised in the delinquency division are formally adjudicated. The truancy court is separate from both the intake (informal) and delinquency (formal) divisions of court in that it processes truancy petitions submitted by the local public school system. Youth younger than age 16 are eligible for truancy court referrals in the county of interest, yet the court typically processes middle school–aged youth to promote prevention. Truancy court exclusively handles all school referrals for chronic absences. In conjunction with the local public school system, the truancy court of interest defines chronic absenteeism as missing 10 or more class periods during the academic year. The overall mission of the truancy court is to eliminate barriers to education as well as provide academic opportunities to local youth who are referred to the court system. The truancy court judges established an “on time, every time” policy in hopes that this court supervision will remove barriers, increase youth and parental commitment to education, and motivate overall changes in school attendance behaviors.

Previous studies addressing adolescents’ risk of recidivism in truancy court have not deeply explored gender and assessment in the context of truancy court interventions (e.g., Onifade et al., 2009; Zhang et al., 2007, 2010). This study aims to fill this gap in the literature by examining the gender-based performance of the YLS/CMI with a sample of youth involved in truancy court in predicting recidivism.

Research Questions

  1. Are there gender differences in risk of recidivism among youth in truancy court?
  2. Are there gender differences in risk of recidivism based on type of recidivism (e.g., truancy or delinquency)?
  3. Are there gender differences in the predictive validity of the YLS/CMI’s composite score and eight domains among truant youth based on any type of future petition (either delinquency or truancy) to court?
  4. Are there gender differences in the predictive validity of the YLS/CMI’s composite score among truant youth disaggregated by future petition type: (a) future delinquency petitions, and (b) future truancy petitions?



This study examined how well the YLS/CMI predicted recidivism overall for male and female youth who entered the juvenile justice system through a truancy court by type of recidivism (future truancy or delinquency petitions). Data in this study were collected in the truancy court division of a juvenile court in a midwestern county. Juvenile court officers (JCOs) administered the risk assessment from 2004 to 2011 to all youth referred to truancy court. Two-year recidivism was the dependent variable and was measured from the time the JCO administered the initial YLS/CMI assessment to each youth (e.g., the beginning of the truancy court case). Recidivism was coded as a dichotomous variable and defined in two ways. First, recidivism was defined as any new petition to court—delinquency or truancy (e.g., 0 = no petition, 1 = delinquency or truancy petition). Second, recidivism was broken down by type of recidivism to identify gender differences in the predictive validity of the YLS/CMI based on future delinquency petitions and future truancy petitions. Adult records were also checked for recidivism during the same time intervals if youth aged out of the court system. Identical to the juvenile records except for the inclusion of status offenses, the adult records included any future criminal justice petitions (e.g., assaults, property offenses) in the county of interest.

The YLS/CMI is a 42-item multidimensional criminogenic risk measure designed to predict future offending and provide guidance for case management for youth in the juvenile justice system (Hoge, Andrews, & Leschied, 2002). The items for each of the eight subscales are dichotomously scored (no = 0, yes = 1); thus, scores can range from 0 to 42. The eight subscales, commonly referred to as the “big eight” criminogenic risk domains, assess both static and dynamic risk factors for future offending (Hoge et al., 2002); they are as follows: Official offense history has five items (e.g., three or more prior convictions), school performance and behavior has seven items (e.g., low achievement), use of free time has three items (e.g., lack of organized activities), characteristics of acquaintances and friends has four items (e.g., lack of positive acquaintances), drug and alcohol use/abuse has five items (e.g., occasional drug use), family relationships and parental behavior has six items (e.g., inadequate supervision), antisocial tendencies has five items (e.g., not seeking help), and disruptive behavior and personality characteristics has seven items (e.g., short attention span). The items within each of these subscales are computed to create a summated score for each risk domain, with scores ranging from 3 to 7 (Hoge et al., 2002), depending on the number of items in each subscale. Risk level for recidivism is determined by the total score of all items on the scale: Low Risk = 0 to 8; Moderate Risk = 9 to 22; High Risk = 23+.

The truancy court started this assessment project in 2003, and the court selected the YLS/CMI because it had been widely used and validated in many juvenile justice settings (Bechtel et al., 2007; Olver, Stockdale, & Wormith, 2014; Onifade et al., 2008a, 2008b; Schwalbe, 2007). The standard delinquency division implemented use of the YLS/CMI at the same time. The truancy court implemented the YLS/CMI to assess the criminogenic risk of truant youth in the same standardized manner as youth involved in the standard delinquency division of the court.


During 2004–2011, 911 youth were referred to truancy court and received the YLS/CMI. The sample included 49.2% boys (n = 448) and 50.8% girls (n = 463). Table 1 includes descriptive statistics of the sample. The JCO administered the YLS/CMI to the youth predisposition via face-to-face interview format; the JCO then scored it. All new truancy assessments scores were entered into the court data management system. There were no missing data or duplicate cases during the study time period. That is, as every youth referred to truancy court received one initial YLS/CMI, our study’s sample only represented unique cases. In collaboration with court administration and management staff, we provided JCOs with extensive training on administering and scoring the YLS/CMI. These trainings consisted of interrater reliability checks, listening to taped cases, and group discussions about scoring and case planning.

Table 1. YLS/CMI Descriptive Statistics


Girls (n = 463)

M (SD) or n (%)

Boys (n = 448)

M (SD) or n (%)

YLS Total Score

12.51 (5.76)

12.49 (5.93)

Offense History

0.20 (0.59)

0.17 (0.58)


2.16 (1.64)

2.02 (1.59)


3.41 (1.61)

3.64 (1.56)

Peer Relations

1.83 (1.16)

1.71 (1.15)

Substance Abuse

0.52 (1.07)

0.67 (1.15)*

Leisure and Recreation

1.91 (0.86)

1.74 (0.90)*


2.02 (1.66)

1.96 (1.66)

Attitudes and Orientation

0.46 (0.81)

1.96 (1.00)*

Low Risk

124 (26.8%)

128 (28.6%)

Moderate Risk

310 (67.0%)

290 (64.7%)

High Risk

29 (6.3%)

30 (6.7%)


13.77 (1.12)

13.73 (1.14)


155 (33.5%)

151 (33.7%)


61 (13.2%)

59 (13.2%)


160 (34.6%)

144 (32.1%)


80 (17.3%)

78 (17.4%)


7 (1.5%)

14 (3.1%)

Notes. Independent samples t-tests and chi-squares were used to test differences.

*p < .05

Two boys were missing race/ethnicity data.



The independent variables in the study included total score, risk level, and subscale scores for each of the eight domains on the YLS/CMI. A correlation matrix of these variables is presented in Table 2.

Table 2. Correlation Matrix of YLS/CMI Scores











1. Total Score



2. Risk Level




3. Prior History





4. Education






5. Leisure







6. Peers








7. Substance Abuse









8. Family










9. Attitudes










10. Personality










*p < .05

A receiver operating characteristic (ROC) area under the curve (AUC) statistic was calculated to examine the predictive validity for the overall sample and the disaggregated samples of boys and girls. AUCs are robust to low base rates, making this a more ideal analysis than a binary logistic regression (Fawcett, 2006). These statistics range from 0 to 1 with an AUC of .5 or below specifying the acceptance of the null hypothesis. AUCs at approximately .6 indicate adequate predictive validity, and values above .7 are considered strong (Fawcett, 2006; Schmidt et al., 2005). In addition to predicting delinquency and truancy recidivism by gender, our study aggregated recidivism type by gender to test if the YLS/CMI is predictive of future truancy petitions and/or future delinquency petitions by gender.


Upon investigating the risk for recidivism based on the total risk score and across each domain, we found there were no significant differences between boys and girls on the total score and five of the eight domains. As illustrated in Table 1, the substance abuse and attitudes/orientation subscales were significantly higher for boys, and the leisure/recreation subscale scores were significantly higher for girls.

Table 3 presents the 2-year recidivism for truancy and delinquency rates by gender. Boys recidivated at a significantly higher rate (40.2%) than girls (31.3%) for any new petition (e.g., truancy or delinquency) to court 2 years following their YLS/CMI assessment. In addition, 2 years following their initial YLS/CMI assessment, the proportion of boys with delinquency petitions (28.4%) was significantly higher than girls with future delinquency petitions (19.9%). In terms of truancy recidivism rates, 14.3% of girls had a future truancy petition, and 13.1% of boys had a future truancy petition. There was a small proportion of boys and girls (n = 24) that had both future delinquency and truancy petitions (not shown in the table); therefore, there was some overlap when broken down by type of recidivism.

Table 3. Two-Year Recidivism Rates by Gender


Girls (n = 463)

n (%)

Boys (n = 448)

n (%)

All Recidivism


145 (31.3%)

180 (40.2%)*


318 (69.7%)

268 (59.8%)*

Delinquency Recidivism


91 (19.9%)

126 (28.4%)*


365 (80.1%)

317 (71.6%)*

Truancy Recidivism


65 (14.3%)

58 (13.1%)


391 (85.7%)

385 (86.9%)

Notes. Independent samples t-tests and chi-squares were used to test differences.

*p < .05

Seven girls and five boys were missing re-offense types.


As seen in Table 4, the YLS/CMI total score was a significant predictor of any recidivism for all youth in truancy court (AUC = .567, p < .01). However, the observed effects are not very strong (all AUCs range from .509 to .590). We also examined the predictive validity of the YLS/CMI subscales. Of the eight subscales, family/parenting and personality/behavior subscales were significant predictors, and the education subscale was the strongest predictor (AUC = .574, p < .01) of any recidivism for the total sample. In terms of the gender-based predictive validity of the assessment, there were several differential findings. We found that none of the subscales or the total score significantly predicted recidivism for girls when examining any type of recidivism. Conversely, the total score, education, family, and personality subscales significantly predicted any type of recidivism for boys (see Table 4).

Table 4. Predictive Validity of the YLS/CMI by Gender for Truant Youth


Overall (N = 911)

AUC [95% CI]

Girls (n = 463)

AUC [95% CI]

Boys (n = 448)

AUC [95% CI]

Total Score

.567 [.528-.606]**

.545 [.488-.601]

.590 [.536-.644]**

Risk Level

.539 [.500-.578]

.528 [.471-.585]

.551 [.496-.605]

Prior History

.516 [.477-.555]

.526 [.468-.583]

.509 [.454-.564]


.574 [.535-.613]***

.556 [.499-.583]

.584 [.530-.639]**


.513 [.473-.552]

.513 [.456-.570]

.522 [.468-.577]


.531 [.492-.571]

.520 [.462-.578]

.548 [.493-.602]


.546 [.508-.585]*

.532 [.476-.588]

.565 [.511-.619]*

Substance Abuse

.540 [.500-.579]*

.527 [.469-.584]

.546 [.491-.601]


.534 [.495-.574]

.530 [.473-.587]

.535 [.480-.590]


.542 [.503-.582]*

.520 [.462-.578]

.565 [.510-.619]*

Notes. CI = confidence interval; AUC = area under the curve
*p < .05. **p < .01 ***p < .001


We conducted a set of posthoc analyses (analogous to a t-test) using MedCalc to test for differences between boys’ and girls’ AUC values. The results revealed that there were no significant differences in AUCs between groups that indicate true differences, but boys had statistically significant AUCs and girls did not. The MedCalc significance tests indicated that the YLS/CMI does not predict differently by gender for truancy youth, which is similar to previous studies with the YLS/CMI that do not predict recidivism for delinquent youth differently by gender (see meta-analysis in Schwalbe, 2008). However, not demonstrating predictive validity at all for a certain subgroup of offenders (e.g., females) is a distinct issue that warrants additional attention for assessments, such as the YLS/CMI, to provide equivalent and accurate risk estimates for all youth (Barnes et al., 2016). Moreover, although the AUCs are statistically significant for the total score and certain subscales for boys, the AUCs are still small in magnitude and should be interpreted with caution. Overall, the YLS/CMI does not appear to be a strong predictor of recidivism for truancy-involved youth or by gender. Therefore, we conducted additional tests to determine the effect of recidivism type.

Significant differences emerged when investigating the predictive validity of the total YLS/CMI score separated by type of recidivism (see Table 5). For the full sample, the YLS/CMI total score did not predict truancy recidivism (AUC = .390). When broken down by gender, the YLS/CMI total score did not predict future truancy for either boys (AUC = .367) or girls (AUC = .411). However, for the entire sample, the YLS/CMI total score was a predictor of future delinquency petitions (AUC = .645), significantly predicting future delinquency for both boys (AUC = .664) and girls (AUC = .626).

Table 5. Gender Differences in the Predictive Validity of the YLS/CMI by Recidivism Type


Overall (N = 911)

AUC [95% CI]

Girls (n = 463)

AUC [95% CI]

Boys (n = 448)

AUC [95% CI]

Truancy Recidivism

.390 [.340-.441]

.411 [.339-.483]

.367 [.298-.436]

Delinquency Recidivism

.645 [.603-.688]***

.626 [.562-.689]***

.664 [.606-.721]***

Notes. CI = confidence interval; AUC = area under the curve
***p < .001



The aim of this study was to identify whether the YLS/CMI was a valid predictor of truancy and delinquency recidivism for youth in the truancy division of a juvenile court. Given that girls are disproportionately more likely to be involved with the juvenile courts for status offenses (e.g., truancy) than delinquency offenses compared to boys (Sickmund & Puzzanchera, 2014), we also investigated the differential predictive validity of the YLS/CMI by gender (Onifade et al., 2009; Zhang et al., 2007). The YLS/CMI is a well-validated criminogenic risk tool for delinquent youth (Bechtel et al., 2007; Catchpole & Gretton, 2003; Flores et al., 2003; Onifade et al., 2008a, 2008b; Schmidt et al., 2011, 2005), but it has not shown promising results to aid in predicting recidivism for youth referred to truancy courts (Onifade et al., 2009). Moreover, practitioners often rely on the criminogenic risk level rather than the total score to guide decision making and aid in case planning. From a practical standpoint, our findings in this study suggest that the YLS/CMI risk scores do not possess strong predictive validity for both boys and girls involved in truancy court.

Furthermore, of the literature that has examined the efficacy of the YLS/CMI for truant youth, there have not been any comprehensive studies examining gender differences in the predictive ability of the assessment for truancy offenders. The results of our study indicated that the YLS/CMI is a statistically significant but generally poor predictor of recidivism for truant offenders, but it predicted slightly better for male offenders by total score and across specific subscales. For example, in the overall sample, we found that education, family, substance abuse, and personality domains significantly predicted recidivism. However, when disaggregated by gender, this relationship only held up for the subsample of boys, in which the education, family, and personality subscales significantly predicted recidivism.

Our study also added to the literature by incorporating types of recidivism broken down by gender and the predictive validity of the assessment based on type of recidivism. Results indicated that there was not a significant difference in the proportion of males and females that were truancy recidivists. However, when examining delinquency recidivism, there were a significantly greater proportion of male delinquency recidivists than female delinquency recidivists in truancy court. In addition, the YLS/CMI significantly predicted recidivism for boys and girls in truancy court who received a delinquency petition during the 2-year follow-up. These findings are congruent with previous research that has noted that truancy court may act as a pathway into the formal juvenile justice system (Polansky et al., 2008; Zhang et al., 2010).

Finally, the findings yielded are also consistent with Onifade and colleagues’ (2009) results that the YLS/CMI is not a significant predictor of truancy recidivism for both boys and girls. These weak effects may be because the sample of truant youth in this study are generally classified as low-risk offenders. If juvenile courts continue to implement criminogenic risk measures such as the YLS/CMI for truant youth, there is a need to norm the tool (e.g., develop new cut scores by gender for risk levels to improve predictive accuracy) for this subpopulation, given the time and monetary investments associated with its implementation.

In addition to examining the predictive validity of the overall risk score, it is also important to examine the psychometric properties of the YLS/CMI assessment’s subscales. If subscales vary by gender, researchers and practitioners can use this information to address gender differences in needs and subsequently respond to the appropriate—and perhaps differential—needs of boys and girls. To that end, developing more gender-responsive risk assessment instruments for youth involved in the system—perhaps by examining the content validity of items (e.g., are the underlying meanings of the items different by gender?)—would bring about much more accurate risk assessment instruments of truant youth and their specific needs. Gender-responsive assessments, successful in predicting female recidivism, are already common in the adult offender literature (e.g., Salisbury, Van Voorhis, & Spiropoulos, 2009), demonstrating that in assessments for youth offenders, development of more specificity in the items predict more accurately for offending girls. As girls comprise about half of the truancy court population, it is important to accurately assess risk and predict recidivism for girls involved with the juvenile justice system.


As with any study, this research is limited by several factors. One limitation is that at the time of this study, the juvenile court was only administering the YLS/CMI and no other criminogenic risk or mental health assessments. This is an important limitation, because there are a multitude of other validated risk-assessment instruments used in the juvenile justice system (e.g., Youth Assessment Screening Inventory, Positive Achievement Change Tool, and Ohio Youth Assessment System), and the use of other risk measures or assessment types would have allowed for a comparative analysis of measures by gender and recidivism type to determine the best measures for predicting delinquency and truancy reoffending for boys and girls. Furthermore, the YLS/CMI only focuses on risk factors and does not include protective factors (e.g., prosocial attitudes, consistent supervision, commitment to school). Protective factors may play an important role in predicting recidivism and understanding gender differences among youth involved in truancy court and, more broadly, in the juvenile justice system (Stevens, Morash, & Park, 2011). As well, school-related factors may have some influence on youths’ trajectories and exacerbate the school-to-prison pipeline (Nolan et al., 2013; Zhang et al., 2010). Therefore, the use of general criminogenic risk measures in specialized truancy courts may not be the most appropriate tool for this particular context because, as seen in our study, they do not predict future truancy.

The dependent variable of interest in this study was recidivism. Although recidivism is the most common outcome measure for delinquency, there are likely to be many other outcomes of interest for youth specifically involved in truancy court. More proximal outcomes, such as academic achievement, may provide insights into areas where researchers and practitioners can provide assistance to truant youth to facilitate more positive outcomes. Investigating these other variables might have also made it possible to delineate potential pathways from these proximal outcomes to more distal variables such as recidivism.

Another important limitation of this study is the reliance on ROC AUC values. These values only assessed the bivariate relationships between the independent variables and the dependent variable disaggregated by gender and did not control for other potentially relevant factors (e.g., age, race/ethnicity) that could impact the ability of the YLS/CMI to predict recidivism for truancy court–involved youth. Nonetheless, our study is still valuable for having investigated truancy recidivism with an eye toward gender differences.

Implications and Directions for Future Research

Truancy is an issue that spans across the educational, juvenile justice, and social service systems; thus, a comprehensive, coordinated systems response is important for addressing the needs of truant youth (Nolan et al., 2013). Connecting youth to appropriate social service agencies (e.g., child welfare services) may increase school attendance and decrease risky behaviors (Dembo et al., 2014, 2015; Larson, Zuel, & Swanson, 2011). For example, Larson and colleagues (2011) contended that truancy and educational neglect is a child welfare issue rather than a juvenile justice issue and saw improvement in school attendance for truant youth by incorporating a family-centered approach (e.g., interventions that focus primarily on the family unit rather than the individual youth) through the child welfare system.

As shown in our study, the widely used YLS/CMI assessment tool in the juvenile justice system works differently for youth entering the justice system for truancy than for those entering the system for delinquency. Given that on the YLS/CMI, the best predictor for truancy for all youth was the education subscale, even above the total score or risk level, researchers and practitioners may consider developing an instrument that is specific for truant populations (see Dembo et al., 2012) and can predict recidivism by offense type (e.g., whether or not the youth will have chronic issues with truancy and/or penetrate deeper into the juvenile justice system with delinquency petitions). Thus, the development and implementation of more appropriate risk screener tools for truant youth is needed. For example, Kim and Barthelemy (2011) developed a truancy risk screener for use in schools, as there are currently no validated instruments or tools to directly measure truancy risk. This particular assessment was developed and validated in a school context, but it could be adapted for the juvenile justice system and tested for feasibility of use in truancy courts.

Furthermore, recidivism for truancy-involved youth and gender differences should be examined in more nuanced ways in future research. For example, researchers may disaggregate crime type (e.g., violent or nonviolent offenses) among recidivists to understand potentially gender-specific truancy pathways and violent behavior. Future research should also consider examining gender-specific risk factors for truancy and the needs of truancy court populations (add/remove variables or develop new assessments that are sensitive to gender-based needs; e.g., Emeka & Sorensen, 2009) to improve the predictive validity of risk-assessment instruments for all youth entering truancy court programs. Truancy courts pose unique intervention points in the juvenile justice system for the potential development and provision of more gender-responsive assessment and services. In addition to gender, it is critical to investigate the effects of race/ethnicity on truancy court involvement, as well as potential differences in the predictive ability of risk-assessment instruments for youth by race/ethnicity (e.g., Shepherd, Luebbers, & Dolan, 2013) compared with larger samples of juvenile justice–involved youth. The impact of how changes in risk scores over time may influence recidivism, a growing area of inquiry in the general delinquency literature (e.g., Barnes et al., 2016), is another important consideration for future research.


When addressing truancy, it’s important to consider policies and practices that influence school attendance and engagement and factors that influence truancy and the trajectory of youth into the justice system. One study found no differences in future rates of attendance and academic achievement among youth petitioned to court for truancy compared with those youth who were truant but did not receive a court petition (Thomas, 2011). Therefore, truant youth may fare better when the juvenile justice system handles these cases informally or through diversion programming (e.g., youth mentoring services), as lower-risk youth benefit more from diversion than from further juvenile justice system contact (Onifade et al., 2009). In sum, there is a critical need for more rigorous evaluation of truancy court interventions and the development of risk assessment tools that are both gender-sensitive and valid for juveniles involved in the justice system for truancy.

About the Authors

Valerie R. Anderson, PhD, is a postdoctoral fellow in Adolescent Medicine at Indiana University School of Medicine. Her research focuses on the intersection of gender and the juvenile justice system, with a particular emphasis on girls’ trajectories in the system, as well as the system’s response to female delinquency.

Ashlee R. Barnes, MA, is a doctoral candidate in the Department of Psychology at Michigan State University. Her research interests include understanding risk and protective factors for juvenile offending and evidence-based interventions for juvenile justice-involved youth.

Nordia A. Campbell, MA, is a doctoral candidate in the Department of Psychology at Michigan State University. Her research interests are in the experiences of historically marginalized youth and students, as well as in programming and protective factors that aim to support them. Christina A. Campbell, PhD, is an assistant professor at the University of Cincinnati in the School of Criminal Justice. Her research focuses on the assessment of risk and strengths among juvenile offenders, neighborhood ecology, and juvenile justice and child welfare policy. 

Eyitayo Onifade, PhD, is an assistant professor at Florida State University in the School of Social Work. His research focuses on delinquency prevention, child welfare programming, and policy-level interventions.

William S. Davidson, PhD, is a University Distinguished Professor of Psychology and senior fellow, university outreach and engagement, at Michigan State University. He is also editor emeritus of the American Journal of Community Psychology. His research has included at-risk populations, including juvenile offenders, adult offenders, substance abusers, victims of violence, and adults with dual diagnosis.


Andrews, D.A., Bonta, J., & Wormith, S.J. (2006). The recent past and near future of risk and/or need assessment. Crime and Delinquency, 52, 7–27.

Andrews, D. A., Guzzo, L., Raynor, P., Rowe, R. C., Rettinger, L. J., Brews, A., & Wormith, J. S. (2012). Are the major risk/need factors predictive of both female and male reoffending? A test with the eight domains of the Level of Service/Case Management Inventory. International Journal of Offender Therapy and Comparative Criminology, 56(1), 113–133.

Attwood, G., & Croll, P. (2006). Truancy in secondary school pupils: Prevalence, trajectories and pupil perspectives. Research Papers in Education, 21, 467–484.

Barnes, A. R., Campbell, N. A., Anderson, V. R., Campbell, C., Onifade, E., & Davidson, W. S. (2016). Validity of initial, exit, and dynamic juvenile risk assessment: An examination across gender and race/ethnicity. Journal of Offender Rehabilitation, 55(1), 2138, DOI: 10.1080/10509674.2015.1107004.

Barry, A. E., Chaney, B., & Chaney, J. (2011). The impact of truant and alcohol-related behavior on educational aspirations: A study of US high school seniors. Journal of School Health, 81(8), 485–492.

Bazemore, G., Stinchcomb, J. B., & Leip, L. (2004). Scared smart or bored straight? Testing deterrence logic in an evaluation of police-led truancy intervention. Justice Quarterly, 21(2), 269–299.

Bechtel, K., Lowenkamp, C., & Latessa, E. (2007). Assessing the risk of re-offending for juvenile offenders using the Youth Level of Service/Case Management Inventory. Journal of Offender Rehabilitation, 45 (3/4), 85–108.

Catchpole, R., & Gretton, H. (2003). The predictive validity of risk assessment with violent young offenders: A 1-year examination of criminal outcome. Criminal Justice and Behavior, 30, 668–708.

Cottle, C. C., Lee, R. J., & Heilbrun, K. (2001). The prediction of criminal recidivism in juveniles. Criminal Justice and Behavior, 28, 367–394.

Dembo, R., & Gulledge, L. M. (2009). Truancy intervention programs: challenges and innovations to implementation. Criminal Justice Policy Review, 20(4), 437–456.

Dembo, R., Briones-Robinson, R., Wareham, J., Winters, K. C., Ungaro, R., & Karas, L. (2015). A longitudinal study of truant youths’ involvement in sexual risk behavior. Journal of Child & Adolescent Substance Abuse, 1–16.

Dembo, R., Briones-Robinson, R., Ungaro, R. A., Gulledge, L. M., Karas, L. M., Winters, K. C., Belenko, S., & Greenbaum, P. E. (2012). Emotional psychological and related problems among truant youths: an exploratory latent class analysis. Journal of Emotional and Behavioral Disorders, 20(3), 157–168.

Dembo, R., Briones-Robinson, R., Wareham, J., Schmeidler, J., Winters, K.C., Barrett, K., Ungaro, R., Karas, L.M., & Belenko, S. (2014). Impact of brief intervention services on drug-using, truant youth: arrest changes over time. Journal of Child & Adolescent Substance Abuse, 23(6), 375–388.

DeSocio, J., VanCura, M., Nelson, L., Hewitt, G., Kitzman, H., & Cole, R. (2007). Engaging truant adolescents: Results from a multifaceted intervention pilot. Preventing School Failure, 51(3), 3–11.

Emeka, T. Q., & Sorensen, J. R. (2009). Female juvenile risk: is there a need for gendered assessment instruments? Youth Violence and Juvenile Justice, 7(4), 313–330.

Epstein, J. L., & Sheldon, S. B. (2002). Present and accounted for: improving student attendance through family and community involvement. The Journal of Educational Research, 95(5), 308–318.

Fawcett, T. (2006). An introduction to ROC analysis. Pattern Recognition Letters, 27(8), 861–874.

Flores, A. W., Travis, L. F., & Latessa, E. J. (2003). Case classification for juvenile corrections: an assessment of the Youth Level of Service/Case Management Inventory. Cincinnati, OH: Division of Criminal Justice.

Gentle-Genitty, C., Karikari, I., Chen, H., Wilka, E., & Kim, J. (2015). Truancy: a look at definitions in the USA and other territories. Educational Studies, 41(1–2), 62–90.

Hendricks, M. A., Sale, E. W., Evans, C. J., McKinley, L., & DeLozier Carter, S. (2010). Evaluation of a truancy court intervention in four middle schools. Psychology in the Schools, 47(2), 173–183.

Henry, K. (2007). Who’s skipping school: Characteristics of truants in 8th and 10th grade. The Journal of School Health, 77(1), 29–35.

Henry, K. & Huizinga, D. (2007a). School-related risk and protective factors associated with truancy among urban youth placed at risk. Journal of Primary Prevention, 28(3), 505–519.

Henry, K. & Huizinga, D. (2007b). Truancy’s effect on the onset of drug use among urban adolescents placed at risk. Journal of Adolescent Health, 40(3), 9–17.

Hoge, R. D., Andrews, D. A., & Leschied, A. (2002). Youth Level of Service/Case Management Inventory™ (YLS/CMI™): User’s Manual. Toronto, Canada: MultiHealthServices.

Hunt, M., & Hopko, D. (2009). Predicting high school truancy among students in the Appalachian South. Journal of Primary Prevention, 30(3), 549–567.

Kim, H., & Barthelemy, J. J. (2011). A tool for assessing truancy risk among school children: Predictive and construct validity of the risk indicator survey. Journal of Social Service Research, 37(1), 50–60.

Larson, A., Zuel, T., & Swanson, M. (2011). Are attendance gains sustained? A follow-up on the educational and child welfare outcomes of students with child welfare involvement for educational neglect. Children & Schools, 33(1), 25–35.

Lawrence, S. A., Lawther, W., Jennison, V., & Hightower, P. (2011). An evaluation of the early truancy intervention (ETI) program. School Social Work Journal, 35(2), 57–71.

Lipsey, M., Howell, J., Kelly, M., Chapman, G., & Carver, D. (2010). Improving the effectiveness of juvenile justice programs: A new perspective on evidence-based practice. Georgetown University. Washington DC: Center for Juvenile Justice Reform.

Maynard, B. R., McCrea, K. T., Pigott, T. D., & Kelly, M. S. (2013). Indicated truancy interventions for chronic truant students: A Campbell systematic review. Research on Social Work Practice, 23(1), 5–21.

Nolan, J. R., Cole, T., Wroughton, J., Clayton-Code, K. P., & Riffe, H. A. (2013). Assessment of risk factors for truancy of children in grades K-12 using survival analysis. Journal of At-Risk Issues, 17(2), 23–30.

Olver, M. E., Stockdale, K. C., & Wormith, J. S. (2014). Thirty years of research on the Level of Service scales: A meta-analytic examination of predictive accuracy and sources of variability. Psychological Assessment, 26, 156–176.

Onifade, E., Davidson, W., Campbell, C., Turke, G., Malinowski, J., & Turner, K. (2008a). Predicting recidivism in probationers with the Youth Level of Service Case Management Inventory (YLS/CMI). Criminal Justice and Behavior, 35(3), 474–483.

Onifade, E., Davidson, W., Livsey, S., Turke, G., Horton, C., Malinowski, J., et al. (2008b). Risk assessment: Identifying patterns of risk in young offenders with the Youth Level of Service/Case Management Inventory. Journal of Criminal Justice, 36, 165–173.

Onifade, E., Nyandoro, A.S., Davidson, W., & Campbell, C. (2009). Truancy patterns of criminogenic risk in a young offender population. Youth Violence and Juvenile Justice, 1–16.

Polansky, M., Villanueva, A., & Bonfield, J. (2008). Responses to violence related questionaires by delinquent, truant, and state-dependent boys receiving treatment in an extended day program. Journal of Offender Rehabilitition, 47(4), 191–203.

Reid, K. (2014). Managing school attendance: Successful intervention strategies for reducing truancy. Abingdon, Oxon: Routledge.

Salisbury, E. J., Van Voorhis, P., & Spiropoulos, G. V. (2009). The predictive validity of a gender-responsive needs assessment: An exploratory study. Crime & Delinquency, 55(4), 550–585.

Schmidt, F., Campbell, M., & Houlding, C. (2011). Comparative analyses of the YLS/CMI, SAVRY, and PCL: YV in adolescent offenders: A 10-year follow-up into adulthood. Youth Violence and Juvenile Justice, 9(1), 23–42.

Schmidt, F., Hoge, R. D., & Gomes, L. (2005). Reliability and validity analyses of the Youth Level of Service/Case Management Inventory. Criminal Justice and Behavior, 32(3), 329–344.

Schwalbe, C. S. (2007). Risk assessment for juvenile justice: a meta-analysis. Law and Human Behavior, 31(5), 449.

Schwalbe, C. S. (2008). A meta-analysis of juvenile justice risk assessment instruments predictive validity by gender. Criminal Justice and Behavior, 35(11), 1367–1381.

Shepherd, S. M., Luebbers, S., & Dolan, M. (2013). Gender and ethnicity in juvenile risk assessment. Criminal Justice and Behavior, 40(4), 388–408.

Sickmund, M., & Puzzanchera, C. (2014). Juvenile Court Statistics 2010. Pittsburgh, PA: National Center for Juvenile Justice.

Stahl, A. (2008). Petitioned status offenses cases in juvenile courts, 2004. Pittsburgh, PA: Office of Justice.

Stevens, T., Morash, M., & Park, S. (2011). Late-adolescent delinquency: Risks and resilience for girls differing in risk at start of adolescence. Youth & Society, 43(4), 1433–1458.

Thomas, G. (2011). Truancy in Washington State: Trends, student characteristics, and the impact of receiving truancy petition. Olympia, WA: Washington State Center for Court Research.

Zhang, D., Katsiyannis, A., Barrett, D., & Willson, V. (2007). Engaging truant adolescents: Results from a multifaceted intervention pilot. Remedial & Special Education, 28(4), 244–256.

Zhang, D., Willson, V., Katsiyannis, A., Barrett, D., Song, J., & Wu, J. (2010). Truancy offenders in the juvenile justice system: A multicohort study. Behavioral Disorders, 35(3), 229–242.

OJJDP Home | About OJJDP | E-News | Topics | Funding
Programs | State Contacts | Publications | Statistics | Events