Abstract
The purpose of this study was to examine the effect of race/ethnicity on recidivism outcomes with a sample of juveniles involved with a truancy court. Three regression models were conducted to examine the influence of race/ethnicity on receiving any new court petition (N = 1,206), including petitions for delinquency offenses or any new status offense petition within 2 years of their initial contact with the court. Results suggest that racial/ethnic disparities exist for juveniles involved in truancy court, especially with regard to new delinquency petitions. These findings are important to take into consideration to understand how truancy courts may facilitate the school-to-prison pipeline for non-White youth.
According to the Office of Juvenile Justice and Delinquency Prevention, racial/ethnic disparity refers to the overrepresentation of certain racial/ethnic groups in the justice system, especially people of African American and Hispanic descent (Bales & Piquero, 2012; Leiber, Peck, & Beaudry-Cyr, 2016). Even with the decline in juvenile delinquency over that last decade, a large racial/ethnic disparity remains in the U.S. juvenile justice system (Bishop & Frazier, 1988; Lehmann, Chiricos, & Bales, 2017; Leiber et al., 2016; Rodriguez, 2010). There are two hypotheses commonly used to explain why racial/ethnic minority groups represent a larger proportion of offender populations. One hypothesis suggests that there is differential offending, which suggests that specific groups have a higher propensity of involvement in specific types of behavior. Conversely, the second hypothesis is differential treatment. Differential treatment suggests that there are structural inequalities (e.g., community surveillance leading to arrest or strict disciplinary policies within urban schools), which exacerbate the risk of involvement in the justice system. Given issues of disproportionality remain, there are growing social concerns on how to address issues of overrepresentation of specific minority groups. Truancy courts are one component of the juvenile justice system in which disparate treatment may lead to racial/ethnic disparities. The current study aims to examine the nature and extent of disparate outcomes in truancy court.
An important aspect that should be considered when understanding disproportionality is the role formal social controls play in racial disparities (Bishop & Frazier, 1988). Research suggests that community policing, schools, and courts jointly have the potential to perpetuate and/or reduce the overrepresentations of racial/ethnic minorities at all stages of the juvenile court process (Crutchfield, Skinner, Haggerty, McGlynn, & Catalano, 2012; Rodriguez, 2007; Stevens & Morash, 2015). For instance, strict disciplinary practices such as “zero tolerance” policies within schools lead to increase in court referrals, especially in urban and racial/ethnic minority communities (American Psychological Association Zero Tolerance Task Force, 2008; Bishop & Leiber, 2011). The school-to-prison pipeline has been well documented as an important pathway to consider when creating strategies to reduce disproportionate contact among racial/ethnic minorities. Examining the role of formal social controls such as schools is imperative, given that overinvolvement with courts may exacerbate criminogenic risk of recidivism and risk of future involvement at all stages of the juvenile system (Mears, Cochran, & Lindsey, 2016). The goal of the present study is to examine potential biases within truancy court, an aspect of the juvenile justice system that heavily relies on school referrals.
Research on racial/ethnic disparity is imperative in all divisions of the juvenile justice system to observe the problem, assess the problem, and then find solutions to remedy the problem (Mears et al., 2016). The majority of research has focused on race/ethnicity in formal juvenile probation; however, there is limited research on juvenile truancy courts as it relates to racial/ethnic disparities and recidivism. This study aims to fill that gap and examine the impact of race/ethnicity in truancy court.
Juvenile Justice System and Truancy
Some scholars believe that the primary goals of the juvenile justice system are to rehabilitate youth and identify youth who show early signs of intervention needs (Steinberg & Scott, 2003). However, the extent to which these goals remain a core focus is debatable. Steinberg and Scott (2003) argue that youth are still developing and may have gaps that fail to allow them to weigh the pros and cons of an action (Steinberg & Scott, 2003). Youth offenders are seen as needing guidance compared with adult offenders; because of this, the juvenile court is divided into different components (Steinberg and Scott, 2003). For instance, many juvenile courts around the country have developed truancy courts to address chronic absenteeism among youth with the goal of intervening before youth commit more serious offenses. This was, in part, created because some researchers believe that status offenses such as truancy act as a risk factor for general delinquency (Baker, Sigmon, & Nugent, 2001; Huizinga & Jakob-Chien, 1998). As a result, truancy court was established to ensure that the juvenile court could intervene early in youth misconduct to prevent future truant and/or delinquent behavior.
Juvenile courts often have multiple divisions which include formal or informal probation in addition to specialty courts. Within juvenile justice, specialty courts commonly involve addressing status offenses such as underage drinking and tobacco use, running away, breaking curfew, and truancy. Among status offenses, truancy is one of the most prevalent infractions; because of this, many states have a separate truancy court division (Stahl, 2008). Truancy refers to unexcused absences, skipping school, and extensive tardiness. Youth identified as truant are often referred to court by school officials or attendance officers hired by the school districts (Baker et al., 2001). Youth may be truant for a number of reasons: (a) family factors (e.g., lack of parental supervision), (b) school factors (e.g., poor education environment or failure to engage students from different cultures), (c) economic factors (e.g., high mobility or lack of transportation), and (d) student factors (e.g., lack of friendly peers, bullying, substance abuse, lack of social and academic support; Baker et al., 2001).
Attending school is a fundamental aspect of academic success (Sutphen, Ford, & Flaherty, 2010). Hendricks et al. (2010) examined four middle school truancy interventions and found that the program did improve school attendance, but only during the semester of the intervention. Hendricks et al. (2010) studied 185 juveniles in truancy court from 2004 to 2008 and separated them into different severity groups (e.g., mild, moderate, and severe). Mild truancy involved youth present in school 88% to 90% of the time, moderate truancy involved youth in school 79% to 87%, and severe truancy was considered any youth who spent less than 79% of time in school (Hendricks et al., 2010). The study’s findings revealed that the truancy intervention worked best for severe truancy juveniles, had limited effects on moderate juveniles, and had no effect on mild juveniles (Hendricks et al., 2010). Another study in Florida asked students why they did not attend school and students replied that they (a) were bored, (b) did not have an interest in school, (c) thought the coursework was not relevant, (d) thought school was too easy or too difficult, (e) had a negative relationship with teachers or other students, or (f) did not feel safe at school (Clement, Gwynne, & Younkin, 2001).
Predictors of Recidivism
Prior research has focused on race/ethnicity, gender, risk assessment, and age in reference to recidivism or receiving future petitions in the juvenile court system (see, for example, Campbell, Papp, Barnes, Onifade, & Anderson, 2018; Leiber & Johnson, 2008; Leiber et al., 2016). There have also been studies examining the role of demographic characteristics of juvenile offenders who have committed status offenses (Anderson et al., 2016; Freiburger & Burke, 2011; Peck, Leiber, & Brubaker, 2014). This literature review will discuss how demographic characteristics such as race/ethnicity, gender, and age influence future juvenile court petitions and the role of juvenile risk assessment instruments.
Race/Ethnicity
Prior literature on race/ethnicity and the juvenile court continue to document racial/ethnic disparities (Bishop & Frazier, 1988; Lehmann et al., 2017; Leiber et al., 2016; Peck et al., 2014; Rodriguez, 2010). According to Leiber and colleagues (2016), racial stereotypes play a role in how the court system makes decisions about juveniles. Minority youth are often subject to disadvantaged outcomes in the juvenile court compared with Whites after controlling for legal and extralegal characteristics (e.g., prior offenses, crime type, age, gender; Leiber & Peck, 2014). One study found that Black/African American, Hispanic/Latinx American, and American Indian youth were treated harsher than White youth at front end court processing, diversion and detention, and back end processing (i.e., out of home placement; Rodriguez, 2010). Another study found that race/ethnicity affected decision-making at the intake level, and that Black/African American youth were less likely to be diverted from the system compared with White youth (Leiber & Johnson, 2008). Each stage of the juvenile justice system has different system actors when a juvenile is processed (Leiber & Peck, 2014). During processing, youth must go through various stages in which different decision makers can affect the outcome of their experience (Leiber & Peck, 2014). Stages that have court officers with the most discretion (i.e., intake) also have the highest potential for racial/ethnic bias, because police officers, social workers, intake officers, and judges all make decisions that affect the juvenile’s court outcome (Leiber & Peck, 2014). However, some researchers believe stages such as adjudication and formal charging are less likely to result in biased decisions based on race/ethnicity or gender, because these stages emphasize prior offenses and crime type (Leiber & Peck, 2014). The reduction in bias is mostly accounted for as the result of implementing risk assessment tools; however, static factors such as criminal history can overestimate risk for certain groups.
In regard to status offenses, race/ethnicity was found to be a significant predictor of adjudication (Freiburger & Burke, 2011). Freiburger and Burke (2011) found that Hispanic youths had the highest odds of adjudication. Meanwhile, Peck et al. (2014) found that African American youth were twice more likely to be adjudicated than Whites. To the contrary, Bishop and Frazier (1996) found that racial/ethnic disparities rarely existed for status offenses, and when they did occur, White youths were treated more harshly than minority youth. Anderson et al. (2016) investigated gender differences in truancy court, examining 2 years’ recidivism. Anderson et al. (2016) noted that further research is needed on racial/ethnic impacts for truant-involved youth. There is very limited research on truancy court and race/ethnicity, especially with regard to recidivism. Zhang et al. (2010) examined race/ethnicity and truancy offenses; results indicated that minority youth (i.e., Black/African American, Asian Americans, and Hispanic/Latinx Americans) were more likely to receive another truancy petition than White youth.
Gender
According to Leiber and Peck (2015), girls are more likely to be treated informally in the beginning stages of the juvenile system compared with males, unless they are near the ending stages of the court system (e.g., adjudication), then they are more likely to be treated harsher than boys. Also, girls receive longer detention sentences than boys if they commit a status offense or violate probation (Beger & Hoffman, 1998; Tracy, Kempf-Leonard, & Abramoske-James, 2009). This may be due to the idea that girls violated their gender role in society and the juvenile court sentences them to harsher sanctions to further protect girls (Freiburger & Burke, 2011; Peck et al., 2014). Girls who commit status offenses, such as truancy, are overrepresented in the juvenile justice system (Chesney-Lind & Shelden, 2004). In fact, prior research has found evidence that female status offenders are being treated more harshly than male status offenders and male delinquent offenders (MacDonald & Chesney-Lind, 2001; Peck et al., 2014; Tracy et al., 2009). In the early 1990s, there were mixed findings about whether there was a gender bias for status offenses in the juvenile system (Freiburger & Burke, 2011). However, in the late 1990s, researchers examined the gender affect by examining the type of status offense (e.g., underage drinking, running away, and skipping school; Freiburger & Burke, 2011). Stahl (2008) found that a gender disparity occurred with certain types of status offenses commonly committed by girls, including running away or skipping school. Of all the status offenses, truancy is the most prevalent among males and females (Stahl, 2008). Prior research has mixed findings on the influence gender played on truancy recidivism specifically (Anderson et al., 2016). Onifade, Davidson, and Campbell (2009) did not find gender to be a significant predictor of truancy recidivism. However, Zhang, Katsiyannis, Barrett, and Willson (2007) found that gender was a significant predictor of truancy recidivism. Flores, Travis, and Latessa (2003) found that girls with a truancy offense had a decreased likelihood of being rearrested. Also, another study had similar findings that boys were more likely to receive another truancy petition than girls (Zhang et al., 2010). Consequently, gender is an important factor to examine among status offenses; however, previous studies suggest that the interaction of gender and race/ethnicity is also important to examine (Freiburger & Burke, 2011; Peck et al., 2014).
Risk Assessment
Risk assessments are widely used and predict juvenile recidivism. Instruments for assessing juvenile risk were developed to predict future delinquency, intended to do so regardless of the influence of demographic characteristics (Onifade, Nyandoro, Davidson, & Campbell, 2010). Scholars have examined the effects of race, gender, and other sociological-related demographics (Campbell et al., 2018; Schwalbe, Fraser, Day, & Arnold, 2004). Schwalbe et al. (2004) found that risk assessments provide the juvenile court with valuable information to assist in court decision-making; however, the relationship between risk scores and recidivism varies by gender and race/ethnicity (Schwalbe et al., 2004). Campbell et al. (2018) examined gender, race, and risk score and found that Black youth significantly differed from White youth on recidivism risk, especially among males. This study suggested that juvenile justice responses (e.g., surveillance, treatment practices) affect the validity of risk assessments rather than the measures themselves (Campbell et al., 2018). In other words, court practices may confer risk based on gender, race/ethnicity, even when considering risk assessment scores (Baglivio & Jackowski, 2013).
The research on risk assessments in specialty courts (e.g., truancy in the juvenile justice system) is limited, and even more so on the influence of demographic characteristics as it relates to risk and future petitions. Previous research on this topic measures recidivism as a single binary variable (e.g., rearrests; Schwalbe, 2008), rather than specifying differences between status offenders and delinquency. The current study aims to fill these gaps by extending the literature on race/ethnicity and exploring juvenile justice system involvement among truancy offenders.
The Current Study
The current study examines whether youths receive a new petition (i.e., recidivism) within 2 years of being evaluated with a risk assessment in truancy court. The research questions are as follows:
The current study aims to extend previous research by examining 2-year recidivism for both status offense petitions and delinquency petitions. First, the influence of race/ethnicity on receiving any new recidivism petition within 2 years of the juvenile’s initial YLS/CMI (Youth Level of Service/Case Management Inventory) assessment date will be examined. Second, the effect of race/ethnicity on delinquency recidivism petitions will be examined. Finally, the influence of race/ethnicity on status offense petitions will be examined. The current study aims to add to prior research by examining the influence race/ethnicity plays in receiving a new petition after juveniles undergo an initial referral to truancy court. The main hypothesis follows that racially/ethnically non-White juveniles (e.g., Black/African American juveniles, Hispanic/Latinx juveniles, and Multiracial juveniles) will be more likely to receive a new delinquent and status offense recidivism petition than White juveniles, regardless of criminogenic risk scores, age, and gender.
Method
Design
The research employs a cross-sectional design that utilizes secondary data from a Midwestern family court’s truancy division. The sample included 1,206 juveniles, inclusive of all youth referred to truancy court between 2004 and 2015 with complete risk assessment and demographic information. This study examined recidivism (e.g., any delinquency or status petition, any status offense petition, and any delinquency petitions) 2 years after juveniles received an initial referral to the truancy division. At the same time, as the initial referral, a risk assessment was given to each juvenile. Recidivism rates were gathered at two different time points, 1 year and 2 years after the initial risk assessment.
Sample
The target population for this study was juveniles with a petition in truancy court in one county in the Midwestern United States. This study used secondary data from one court in the Midwestern region. The sample included every juvenile referred to a Midwestern family court, truancy division, during the years of 2004 to 2015. Table 1 provides the univariate descriptive statistics for each variable used in the three models.
Descriptive Statistics.
Note. YLS/CMI subscale scores show when a youth is high risk (i.e., higher scores) or when a youth is low risk (i.e., lower scores) on a subscale. YLS/CMI = Youth Level of Service/Case Management Inventory.
This variable was not included in the regression model due to multicollinearity.
Data Collection
Data were retrieved through the truancy division of a midsized juvenile and family court from the court’s data management system. The family court from which the data were taken had three components: (a) intake, (b) delinquency, and (c) truancy. This study focuses solely on the truancy division, which is separate from the other two divisions and is unique, in that it receives petitions from the local school district. The local school district keeps a record of the absentees and defines chronic absence when a student has 10 or more missed class periods. When a student meets criteria for chronic absence, a referral is sent to the truancy division. The court aims to prevent truancy, other status offenses, and future delinquency by referring students 16 years of age or younger with a focus on middle school students. The truancy court division implemented risk assessment in 2003 utilizing the YLS/CMI. The YLS/CMI was scored by the assigned juvenile court officer (JCO) through a face-to-face interview, and each new YLS/CMI assessment was scored and entered into the truancy court’s data system following each referral. This method of data collection includes all juvenile truancy referrals, risk assessment scores, and demographic information (e.g., race/ethnicity, gender, age) for every youth with a truant offense between 2004 and 2015.
Data and Measures
The data set complied information from a truancy court from 2004 until 2015 on all juvenile referrals. The information gathered on juveniles with an initial referral from the court was as follows: new petitions, new petition dates, race/ethnicity, date of birth, gender, YLS/CMI assessment risk level, date of YLS/CMI assessment, YLS/CMI’s eight subscales scores, and the scores for each item. The YLS/CMI is one of the most widely used risk assessments (Schwalbe, 2007). The YLS/CMI comprises eight subscales and a total of 42 items (Hoge & Andrews, 1996). The eight subscales are as follows: (a) prior offenses (e.g., current offense, prior adjudications), (b) family and parenting (i.e., family relationships), (c) education (e.g., classroom disturbances, relationship with teacher), (d) peer relationships (e.g., negative acquaintances or friends), (e) substance abuse (e.g., illegal drug use, problem with alcohol), (f) leisure and recreation (e.g., lack of organizational activities), (g) personality and behavior (e.g., aggressive), and (h) attitudes and orientation (e.g., antisocial attitudes; Andrews & Bonta, 2010). Flores et al. (2003) noted the importance of the YLS/CMI as it provides “correctional agencies with an indication of the youth’s overall risk of re-offending” and provides “an in-depth explanation as to what factors are driving a youth’s risk level” (p. 43). Schwalbe (2007) conducted a meta-analysis to examine how well the YLS/CMI predicts reoffense and found that the YLS/CMI is a strong predictor of future offenses. Onifade et al. (2010) had similar findings, that the YLS/CMI predicted delinquency recidivism; however, it did not predict truancy recidivism. In fact, juveniles who had a low risk level on the YLS/CMI were more likely to have a truancy reoffense (Onifade et al., 2010). Anderson et al. (2016) examined gender and the YLS/CMI and found that boys were at a higher risk of recidivism for both delinquency and truancy, compared with girls.
Dependent variable
Recidivism was the dependent variable and was operationalized in three ways: (a) any recidivism petitions (e.g., delinquency and status offense) within 2 years of initial truancy court involvement, (b) any delinquency recidivism petitions (e.g., assault, larceny), and (c) status offense recidivism petitions (e.g., truancy). If a youth aged out of the juvenile system during this period of time, adult records were checked as well for any criminal petitions. Delinquency recidivism was a dichotomous variable that measured whether a juvenile received any new delinquency petitions within 2 years of their initial YLS/CMI assessment date. For this variable, juveniles who had a status offense petition but not a delinquency petition were given a 0 for this measure, as they did not receive any delinquency petitions. Status offense recidivism was defined as any new status offense petition within 2 years of the initial YLS/CMI date. Dates of new petitions were coded into a binary recidivism variable after the initial YLS/CMI assessment. Recidivism was measured by obtaining the offense codes tied to the petition number for every juvenile after the youth was processed into the juvenile system and given an assessment from 2004 to 2015.
Recidivism was calculated in two steps. First, all recidivism petitions were examined for each juvenile. For the purpose of this study, the first five recidivism petitions were utilized; however, the court management system for which the data were obtained includes every recidivism petition for each juvenile. Because of this, the number of recidivism petitions range from zero to five. Each petition was coded as one of the following: (a) Part I offenses, (b) Part II offenses, and (c) status offenses. Second, the most serious offense was coded for all juveniles with a valid recidivism petition using the Uniform Crime Report’s (UCR) offense classifications. The UCR identifies two major offense categories. First, the UCR includes Part I offenses, which include homicide, forcible rape, robbery, aggravated assault, burglary, and arson (Federal Bureau of Investigation, 2004). Second, the UCR contains Part II offenses, which include crimes such as simple assaults, forgery, fraud, stolen property, vandalism, weapons, prostitution, and drug abuse (Federal Bureau of Investigation, 2004). Part I and Part II offense categories were used to classify all juvenile offenses, except for status offenses (i.e., any offense in which minors can be charged, but adults, 18 years and above, cannot). All offense-type codes that were not identified were counted as missing. The offense codes for other petitions for that juvenile were used to determine the most serious crime committed by the individual.
Independent variable
Race/ethnicity was the independent variable and was recoded into four categorical, binary variables: (a) Hispanic/Latinx American, (b) Black/African American, (c) Multiracial Americans, and (d) White Americans. Prior research found race/ethnicity to be a predictor of receiving a new petition (Zhang et al., 2010). Table 1 shows the descriptive statistics (e.g., M and SD) for these variables. White Americans acted as the reference category. For the first two regression models (i.e., any recidivism and delinquency recidivism), 19.2% of truant juveniles were of a Multiracial descent, 32.9% were Black/African American, 13.9% were Hispanic/Latinx descent, and 33.9% were White (see Table 1). For the status offense model, 18.5% of juveniles were Multiracial, 31% were Black or African American, 14.6% were Hispanic or Latinx, and 35.9% were White (see Table 1). This study examines two major racial groups and one ethnic group defined by the National Research Council (2004). The Multiracial group comprises individuals who are mixed with two or more racial/ethnic ancestries. According to 2010 U.S. Census data, the county the Midwestern juvenile truancy court resided in had a large population of White Americans (69.9%), followed by the second largest racial group, Black/African Americans (12.2%). We could reason that the majority of youth in the Multiracial group are both White and Black as these serve to be the two largest racial groups in that area.
Control variables
The control variables were age, gender, and the eight YLS/CMI subscales. The date of birth was used to calculate each juvenile’s age in years using SPSS. In the full sample, 52.0% (627) of truant juveniles were female and 48.0% (579) were male. This could serve as an example of a gender effect within truancy court. Because truancy courts were created in different states as a response to various issues, these courts often differ in focus and scope. The truancy court in this study was originally created to oversee girls, particularly those involved in sex trafficking or running away. This original goal could potentially lead to increased surveillance and widening the net of girls involved in truancy.
The court did not collect information on the income status of juveniles and their families, nor other factors associated with socioeconomic disadvantage; because of this, the median age, education, and substance abuse subscores of subjects by race were examined. For the status offense recidivism model, White, Hispanic/Latinx, Multiracial, and Black/African American youth had a median age of 14 years, a median education score of 3, and a median substance abuse score of 0. Similarly, the delinquency and any recidivism models had the same median scores for all race/ethnicity groups, except for Black/African American youth. This group had a median age of 14 and substance abuse score of 0 as well, but their median education score was 4, which put youth in a high-risk category (Table 2).
Frequencies of Recidivism Predictors.
Eight YLS/CMI subscales
The YLS/CMI consists of an additive scale of 42 items across eight risk and need subscales (Hoge & Andrews, 1996); however, this study used 41 items because this court system does not use the employment risk variable due to lack of variation. The YLS/CMI is a well-validated and widely used assessment tool, which serves as the most studied juvenile assessment tool to date that reveals a moderate to strong prediction result (Onifade et al., 2009; Schwalbe, 2008).
The YLS/CMI total scores were also examined and added to Table 1 for reference with a mean of 12.69 and a standard deviation of 5.64 for any recidivism and delinquency recidivism models, which indicates a moderate risk level as the mean in this sample of juveniles. The status offense recidivism model held a slightly lower total YLS/CMI score or 11.96 (SD = 5.36), which also indicates the average juvenile in this sample has a moderate risk of reoffending. The total score was created through an additive scale from the eight YLS/CMI subscales (i.e., prior offenses, family and parenting, education, peer relationships, substance abuse, leisure and recreation, personality and behavior, and attitudes and orientation). Low-risk individuals score 0 to 8 overall, moderate-risk individuals score 9 to 22 overall, and high-risk individuals score 23 or more overall (Hoge & Andrews, 1996). The total YLS/CMI score was not included in the analysis due to multicollinearity with the eight YLS/CMI subscales. Removing this variable was not problematic, because prior research found that overall risk scores and risk levels do not predict recidivism of truant offenders (e.g., Onifade et al., 2009).
Statistical Analysis
Three binary logistic regression models were conducted to estimate the influence of race/ethnicity on recidivism outcomes in truancy court. Logistic regression assumptions were generally met with the data: (a) the independent variable (i.e., race/ethnicity) and the dependent variable (i.e., recidivism) were not linearly related; however, some of the continuous predictor variables were linearly related to the log of the odds that recidivism occurs; (b) the dependent variable was dichotomous; (c) the independent variable was not normally distributed, linearly related, nor had equal variance within each group; (d) each categorical variable was mutually exclusive and exhaustive (Weisburd & Britt, 2014).
None of the continuous variables were linearly related to the logit of status offense recidivism petitions. These subscales and age were recoded into smaller categories and another regression was conducted for all three dependent recidivism variables. Age was recoded into three categories: (a) juveniles aged 10 to 12 years, (b) juveniles aged 13 to 14 years, and (c) juveniles aged 15 to 16 years. Each YLS/CMI subscale that was not linearly related was recoded into three categories: (a) low risk, (b) moderate risk, and (c) high risk. After recoding, the test was reconducted for each dependent recidivism variable separately. All continuous predictor variables met the logistic assumption of linearity. Finally, the receiver operating characteristic (ROC) area under the curve (AUC) was also examined in SPSS for each variable to determine predictive accuracy (Rice & Harris, 1995; Schmidt, Hoge, & Gomes, 2005).
Results
Any Recidivism
Of 1,206 youth, 422 (35.0%) juveniles had a delinquency or status offense petition within 2 years of their initial YLS/CMI assessment. Of the juveniles who received a new petition within 2 years, 32.0% (135) were status offense petitions, 28.2% (119) were crimes classified as Part II crimes in the UCR, and 39.6% (167) were crimes classified as Part I crimes in the UCR. One recidivism petition was not classified due to an unknown crime-type code.
A binary logistic regression was conducted to examine the effect of race/ethnicity on any new petition within 2 years of the juveniles’ YLS/CMI assessment controlling for gender, age, and the eight YLS/CMI subscales (χ2 = 85.079, p < .001). The independent variables (i.e., race/ethnicity categories) were not significant in this model when controlling for age, gender, and risk score, with the exception of the Multiracial variable that appears to predict 2-year recidivism petitions (Wald = 4.779, p = .029). The control variable, gender, was negatively related to recidivism and significantly contributed to the overall model (Wald = 11.871, p = .001). The odds ratio was 0.645 with a confidence interval = [0.503, 0.828] for the gender variable (see Table 3). Age was a negative significant predictor of 2-year recidivism petitions (Wald = 37.675, p < .001). Of the YLS/CMI subscales, substance abuse (Wald = 12.299, p < .001) and education (Wald = 6.255, p = .014) significantly predicted a new recidivism petition within 2 years of the juveniles YLS/CMI date. The AUC calculated for this model was statistically significant (AUC = 0.627, p < .001).
Logistic Regression Models—Three Types of Recidivism.
Note. Model strength was established by Nagelkerke R2 for any recidivism petitions (0.094), delinquency recidivism petitions (0.150), and status offense recidivism petitions (0.150). CI = confidence interval.
p < .05. **p < .001.
Delinquency Recidivism
Of 1,206 juveniles, 294 (24.4%) juveniles received a delinquency petition and 912 (75.6%) did not receive a delinquency petition within 2 years of their initial YLS/CMI assessment in truancy court. This model classifies status offenses as nonrecidivism. Of the juveniles who received a new delinquency petition, 42.2% (124) were petitioned for an offense consistent with Part II crimes classified by the UCR and 57.8% (170) were petitioned for an offense consistent with Part I UCR crimes.
A binary logistic regression was conducted to examine the effect of different racial/ethnic categories on new delinquency petitions within 2 years of the juveniles’ YLS/CMI assessment controlling for gender, age, YLS/CMI year, and the eight YLS/CMI subscales. The model chi-square was significant (χ2 = 128.322, p < .001), which indicates that the estimated model significantly improved after the addition of predictor variables. The Hosmer and Lemeshow test was also conducted (χ2 = 6.173, p = .628), which suggests a good model fit (Fox, 2016). In the constant logistic classification table, 75.6% were correctly classified; however, when the predictors were added to the model, 76.7% were correctly classified. This increase shows that the model with predictors added improved the overall model (Table 4).
Recidivism Rates by Model.
The independent variables Black/African American (Wald = 4.710, p = .030) and Multiracial (Wald = 4.722, p = .030) were both significant in predicting whether the juvenile received a new delinquency petition within 2 years of their initial YLS/CMI date (see Table 3). This finding suggests that Black/African American or Multiracial youth are more likely to receive a recidivism petition than White juveniles. This finding is important to note as other models (any recidivism and status offense recidivism) did not show the disparities of Black/African Americans. The control variable, gender, was negatively related to recidivism and significantly contributed to the overall model (Wald = 22.840, p < .001).
Race/ethnicity (i.e., Hispanic/Latinx American, Black/African American, Multiracial American, White Americans), gender, age, and the eight YLS/CMI subscales acted as categorical predictor variables; therefore, the strength they contribute to the model can be compared using exponent B. Interestingly, boys were 0.498 times more likely than girls to receive a new delinquency petition (see Table 3). Age was found to have a negative significant relationship with receiving delinquency petitions (Wald = 8.804, p = .003). Also, three of the YLS/CMI subscales, prior offenses (Wald = 5.050, p = .025), education (Wald = 22.277, p < .001), and substance abuse (Wald = 10.078, p = .002) had significant positive relationships with receiving a new delinquency petition. For this model, the AUC was statistically significant (AUC = 0.711, p < .001).
Status Offense Recidivism
Of 906 juveniles included in this model, 15.0% (136) of juveniles received a status offense petition within 2 years of their initial YLS/CMI assessment in truancy court. A binary logistic regression was conducted to examine the effect of race/ethnicity on any new status offense petition within 2 years of the juveniles’ YLS/CMI assessment controlling for gender, age, and the eight YLS/CMI subscales. The model chi-square was significant (χ2 = 8.857, p < .001), which indicates that the estimated model significantly improved after the addition of predictor variables. None of the race/ethnicity categories significantly predicted status offense recidivism petitions, which differs from the delinquency model. The control variable, age, was negatively related to recidivism and significantly contributed to the overall model (Wald = 52.418, p < .001).
Of the eight YLS/CMI subscales, two of them significantly predicted recidivism, which means that controlling for these variables had significant impact on the difference in probabilities for recidivism. These subscales include (a) leisure and recreation and (b) peer relationships. The YLS/CMI subscale leisure and recreation significantly predicted recidivism (Wald = 3.991, p = .046), which suggests that the higher risk the youth scored on the leisure subscale, the more likely the youth would receive a recidivism petition. Interestingly, the peer relationship subscale negatively predicted status offense recidivism (Wald = 4.064, p = .044), which suggests that the lower the youth scored on this item, the more likely they would receive a new recidivism petition within 2 years of their initial YLS/CMI assessment. Leisure was found to be a stronger contributor to the model, exp(B) = 1.379. The AUC calculated for this model was statistically significant (AUC = 0.723, p < .001).
Discussion
This study adds to the literature by examining racial/ethnic disparities in truancy court outcomes. These findings are important to take into consideration for juvenile justice policy and practice. In particular, this study aims to understand more about racial/ethnic disparities in truancy courts, how truancy initiatives may further facilitate the school-to-prison pipeline, and, especially, the iatrogenic effects of these practices for non-White youth. The current study examined the influence of race/ethnicity on receiving new petitions for youth who received a truancy referral as their first contact with the juvenile system using three models. These three models controlled for age, gender, and YLS/CMI subscale scores. First, the effect of race/ethnicity on any recidivism within 2 years of the juvenile’s initial YLS/CMI date was examined. This model identified that race did significantly predict receiving a new delinquency or status offense petition for Multiracial youth compared with White youth; however, this model did not find any other significant effects by race/ethnicity. Second, the influence of race/ethnicity on delinquency recidivism within 2 years of their initial YLS/CMI date was conducted. The delinquency model found Black/African American youth and Multiracial youth were more likely to receive a new delinquency petition within 2 years of their initial YLS/CMI assessment compared with White youth. Finally, the influence of race/ethnicity on status offense petitions was examined. This model did not find any racial/ethnic effects on receiving a new status offense petition within 2 years of juveniles’ initial risk assessment.
This study extends prior research and provides insight into an unstudied area in two unique ways. First, risk assessments serve as a widely used tool in the juvenile courts to predict recidivism and aid case planning, but little research has examined the use and impact of risk tools in specialty courts such as the truancy courts in the juvenile system. The current study finds that the YLS/CMI predicts recidivism in some cases, typically with delinquency recidivism; however, the tool does not predict recidivism for status reoffenses. This finding was consistent with prior research that the YLS/CMI predicts delinquency recidivism but fails to predict future truancy offenses (Onifade et al., 2010). Our findings suggest that half of the YLS/CMI items predicted recidivism for delinquency petitions, which is not completely consistent as prior research has found that the YLS/CMI predicts delinquency reoffenses (Schwalbe, 2007). This study found that only one YLS/CMI item predicted status offense petitions, which is consistent with prior research that the YLS/CMI typically does not accurately predict status reoffenses (Onifade et al., 2010). Second, this study examined the influence of different racial/ethnic groups on receiving a new petition within 2 years of the juveniles’ YLS/CMI assessment date. With any new petition, Multiracial youth were more likely to receive a new petition. In contrast, when we separated the petitions into delinquency and status reoffenses, the results changed in terms of racial/ethnic disparities. There were no racial/ethnic differences found between groups for new status offense petitions. This finding was not consistent with previous research, which has indicated that race/ethnicity influences the relationship between status offending and court adjudication (Peck, Leiber, Beaudry-Cyr, & Toman, 2015). Black/African American status offenders receive increased levels of social control than White status offenders (Peck et al., 2014). To the contrary, Freiburger and Burke (2011) found that Black status offenders were no more likely than White offenders to receive adjudication, but Hispanic status offenders were more likely to be adjudicated compared with White juveniles. Our findings also do not align with Zhang et al. (2010), as they found that racial/ethnic minority youth were more likely to receive a truancy petition compared with White youth; however, it is important to note the paucity of research in this area.
Conversely, our findings indicate that Black/African American youth and Multiracial American youth receive more delinquency petitions after truancy court involvement compared with White youth when controlling for risk factors. This finding suggests that there are racial disparities within truancy court for delinquency petitions. One explanation is that Black/African American and Multiracial youth have been exposed to disproportionate minority contact by police. According to UCR data, Black/African American juveniles are more likely to be arrested for violent crime, property crime, and drug offenses than White youth (U.S. Department of Justice, 2008). However, self-report surveys indicate that Black/African American youth are less likely than White youth to commit a property crime or use an illegal substance (Felson & Kreager, 2015). Prior research found that police and probation officers viewed Black/African American youth as more blameworthy and viewed criminal behavior as an internal characteristic compared with White youth (Fagan, 2010). Because police and probation officers are in contact with youth in the juvenile system, viewing Black/African Americans as more blameworthy may lead them to arrest more Black/African American youth than other racial groups. In addition, there is no reason to believe that Black/African Americans and Multiracial youth are committing more offenses, especially because this sample is from truancy court, which is a lower risk group of youth with no to minimal official history of general delinquency. Instead, it is likely that these data suggest that truancy may have a net-widening effect for specific racial/ethnic groups. Furthermore, the county in question has a policy; wherefore, juveniles who are referred to truancy court are required to attend school for 90 days without any unexcused absences and tardiness. These youth are under increased surveillance and monitored by the court. It is possible that such stringent policies lead to disproportionate contact and undermine the diverse challenges and social contexts (e.g., lack of support from teachers, parents, or peers) that interfere with school attendance. In addition, for this county, youth who fail to meet the expectations of the court receive additional sanctions, which may result in programming that involves youth from other court divisions (i.e., formal probation). Given the potential for truancy court to lead to negative consequences that may further disrupt a youth’s educational experiences, more evidence-based research is needed that further identifies strategies and responses that better address absenteeism.
Limitations
The current study identified confounding variables as an internal validity threat. To address this threat, the present study controlled for gender, age, and YLS/CMI subscale scores. It is of great importance to control for every variable that may be a predictor of receiving a new petition; however, the present study was limited by the data available. Furthermore, this study obtained data from a secondary data set, which included information from 2004 to 2015 on one Midwestern family court’s truancy division. Thus, this study may not reflect the general target population of truancy offenders throughout the country. There was no exploration of other truancy courts of other regions in the United States, which indicates that this study may not be generalizable throughout the country.
This study had several limitations. First, this study relied on a secondary data set and used a secondary data analysis. The current analysis was limited by the Midwestern family court truancy division’s reported data. The Multiracial category serves as a limitation as we were not able to parse out different racial categories, because the court did not collect this information; however, we were able to examine the racial/ethnic demographics of the county the court was located in from U.S. Census data and the reason that the Multiracial category most likely included youth who were both Black and White. This truancy court managed all school-based referrals for students with chronic absences, which was defined as missing 10 or more course periods throughout the academic year. There were no school data examined for the purposes of this study. The truancy court examined only used the YLS/CMI assessment to screen youth for risk of future delinquency or, in this case, delinquency and truancy. As noted by Anderson et al. (2016), this serves as a limitation because other risk assessments, which include risk scores and protective factors, were not utilized by the court. Also, this study did not control for socioeconomic status due to the limited data available. However, this study did examine the median age, education, and substance abuse subscores for youth by race/ethnicity across each model. This serves as a limitation as socioeconomic status could act as a confounding variable, influencing the relationship between race/ethnicity and receiving a new status offense or delinquent petition. Prior research found that a juvenile from a family with a lower socioeconomic status (i.e., low income and low education levels) was more likely to receive a truancy petition than juveniles from a family with a high socioeconomic status (i.e., high income and high education levels; Attwood & Croll, 2006).
Future Research
Further research is needed on the impact of racial/ethnic disparities in truancy courts across the country. There is also a need to further deepen our understanding of what is occurring within these courts and why certain juveniles receive new delinquency or status offense petitions compared with others of different demographics. Using random sampling in future studies is also needed to make bolder statements about the target population (i.e., juveniles with a truancy petition across the country). The interaction of race/ethnicity and socioeconomic status (i.e., family education levels and family yearly earnings) and truancy should also be examined to address the confounding threat. Qualitative and mixed-method studies would help contextualize these findings. Future research on the interaction of race/ethnicity and gender in the context of truancy court is also needed to further understand who is being referred to these courts and the extent to which there are disparate outcomes.
Conclusion
This is one of the first studies to examine the racial/ethnic disparities by separating offense types (delinquency and status offense petitions) for juveniles involved in truancy court. Our findings indicate that Black/African American and Multiracial juveniles are more likely to receive a delinquency petition within 2 years of their initial risk assessment date. No differences were found with status offenses. This could suggest that Black/African American and Multiracial youth have more deleterious outcomes than White juveniles in truancy court. Furthermore, these youth may experience disproportionate minority contact by police, probation officers, or other court actors which could explain the disparity in delinquency petitions. Further research is needed on exploring why these racial disparities are occurring and ways to change the juvenile justice response to be more equitable for all youth in the system.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
