Risk-Need Profiles of Serious and Chronic Female Juvenile Offenders: Implications for Female Juvenile Correctional Programming

Abstract

Understanding differences and similarities between male and female juvenile offenders is critically important for determining the treatment needs of each group. Less is known, however, about the similarities and differences among female juvenile offenders and the variation in their needs, risks, and psychosocial profiles. Understanding the variation among female juvenile offenders could lead to improvements in gender-responsive interventions and treatment. Latent profile analysis was conducted to construct risk-need profiles in a state-based sample of incarcerated girls (N = 203) based on a range of psychosocial subscales covering family, peer, school and cognitive and emotional processes, psychopathology, and antisocial outcomes. Findings revealed four distinct groups/profiles with varying levels of risk-needs—Aggression Only (51%), Alcohol and Drug Use (19%), Socioemotional and Family Relationship Problems (24%), and Severe Alcohol and Drug Use (6%)—warranting the need for varying levels of treatment intensity and different treatment components across subgroups, ranging from less to more extensive.

Keywords

serious and chronic juvenile female offenders incarcerated girls subtypes latent profile analysis risk-need profiles

Although the base rates of serious and chronic offending among girls are relatively low compared with the rates among boys, the consequences are no less serious. Like their male counterparts, this small subgroup of female juvenile offenders contributes disproportionately to juvenile acts of delinquency and eventually to adult crime. Kempf-Leonard, Tracey, and Howell (2001) found that only 3.5% of the girls in their sample were serious and chronic offenders, but 44% of this group had arrest records as adults. Similarly, Colman, Kim, Mitchell-Herzfeld, and Shady (2009) found that only14% of girls in their sample of incarcerated girls were chronic offenders or low-rising offenders (relatively low arrest rates in late adolescence but relatively high arrest rates in young adulthood). Findings of a study conducted by Lanctôt, Cernkovich, and Giordano (2007) revealed that women with a history of juvenile incarceration were involved in more criminal activities than were their counterparts without a history of juvenile incarceration and faced more problems in adulthood because of substance and alcohol use.

Serious and chronic female juvenile offenders have a variety of multiple overlapping problems, risks, and needs, including persistent and serious substance use, anxiety and posttraumatic stress disorders, academic difficulty and school failure, family conflict, lack of parental/caregiver supervision, aggressive, hostile personality traits, irritability, and histories of trauma, depression, and suicidal ideation (Dixon, Howie, & Starling, 2004; Huizinga, Loeber, Thornberry, & Cothern, 2000; Kataoka et al., 2001; McCabe, Lansing, Garland, & Hough, 2002; Rich, Wilson, & Robertson, 2016; Roe-Sepowitz, 2009). These multiple, interrelated problems must be considered when addressing their treatment needs. Accordingly, a “careful identification of the configuration of problems facing serious [female juvenile] offenders is needed” (Huizinga et al., 2000, p. 6). By identifying and addressing the needs of serious and chronic offenders exhibiting the same constellation of risk-needs, interventions could be customized to subgroups of girls, “rather than proceeding under the assumption that all offenders require similar treatment” (Huizinga et al., 2000, p. 1) and that serious and chronic female juvenile offenders are a homogeneous group (Odgers et al., 2007). Moreover, understanding the variation between and within different subgroups of serious female juvenile offenders could lead to improvements in gender-responsive interventions and treatment.

Using person-centered analytic techniques, which are oriented toward identifying patterns of similarity and dissimilarity among individuals (Bergman, 2001), the aim of the present study was to construct risk-need profiles in a state-based sample population of incarcerated girls with a history of serious and chronic offending. An empirically derived typology of risk-need profiles could provide service providers with diagnostic indicators; could aid them in the development of tailored screening, assessment, and treatment interventions for different subgroups of female juvenile offenders; and could aid in the creation of a classification scheme for determining which combinations of interventions and treatment strategies work best for subgroups of girls with particular constellations of risks, problems, and needs. In short, typologies and profile-based research are critical to the “continued movement toward evidence based, gender-responsive programming that examines how within girl differences impact girls’ responses to various type of interventions” (Matthews & Hubbard, 2008, p. 499).

Conceptual Framework

Although female juvenile offenders have a variety of multiple overlapping risks and needs, the selection of variables for the construction of the risk-need profiles was guided by the biopsychosocial model of the development of adolescent chronic conduct problems (Dodge & Pettit, 2003) and other empirical research on risk factors. The model is based on the premise that multiple factors contribute to antisocial outcomes, including biological, cognitive, and interpersonal factors, and that there are varying paths to conduct problems (Dodge & Pettit, 2003). These paths are characterized as distal, proximal, and intervening. Intervening factors include life experiences with parents, peers, and social institutions. These intervening factors either increase or mediate risks imposed by the distal mechanisms of biological predispositions and sociocultural contexts.

On the basis of this model, Dodge and Pettit (2003) outlined implications for interventions. First, because antisocial behaviors are correlated with other problem behaviors, successful interventions will require addressing multiple risk factors and multiple outcomes at the same time, which in turn will require multicomponent interventions. This aligns well with practice recommendations for addressing the multiple problems, needs, and poor outcomes involved in the development and continuation of serious delinquent behavior among female juvenile offenders (Hipwell & Loeber, 2006; Leve, Chamberlain, & Kim, 2015). Second, because development operates through psychological processes and life experiences with parents, peers, and school, interventions should address these domains. Family and school relationships and association with deviant peers have been found to be robust predictors of female offending (Hubbard & Pratt, 2002; Steketee, Junger, & Junger-Tas, 2013; van der Put et al., 2014), and family factors have been found to be strong predictors of recidivism (van der Put et al., 2014) and adult arrests among female juvenile offenders (Rhoades, Leve, Eddy, & Chamberlain, 2016). And third, interventions must be culturally, racially, and developmentally sensitive.

In the present study, multiple risk-need factors and multiple antisocial behaviors were examined concurrently. Key constructs of the biopsychosocial model were examined, including cognitive processes and life experiences with family, peers, and school.

Method

Procedure and Sample

Cases and data analyzed in the present study were collected as part of a larger study designed to evaluate interventions at a female youth development center, the only youth development facility for girls, in a southeastern state that housed girls aged 10 to 18 years, who committed a criminal offense. For the larger study, data were collected from case records associated with each youth committed to the youth development center over a 4-year period. Specifically, the following data were collected from case records: mental health and psychological assessments; demographic, social, and offense history profiles; mental, physical health, and substance use questionnaires; admission and discharge histories; risk and needs assessments; and standardized assessments (Multidimensional Adolescent Assessment Scale [MAAS], Massachusetts Youth Screening Instrument–Version 2 [MAYSI-2], and Problem Oriented Screening Instrument for Teenagers [POSIT]). Data from the case records were coded, entered, and stored in the Residential Substance Abuse and Treatment (RSAT) database.

Cases in the RSAT database included all females committed to the youth development center over a 4-year period. There were a total of 218 cases during this period. However, 15 of those cases included girls who were recommitted to the youth development center. Because of duplication, those cases were eliminated, resulting in a final sample of 203 cases. Of the 203 females comprising the sample population, 70.4% were African American, 26.6% were Caucasian, and 4.9% were designated as “Other.” The average age of girls in the sample was 15 (M = 14.91, SD = 0.985). Their average age at first court referral was 13 (M = 12.88, SD = 1.409). On average, the girls in the sample had three prior adjudications (M = 2.71, SD = 1.769). Nearly 70% of the girls in the sample met criteria for a Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM-IV-TR; American Psychiatric Association, 2000) substance use disorder, and 97.4% met criteria for a mental health disorder.

Measures

Data from the MAAS and POSIT were used in the present study. The MAAS is a self-report tool designed to evaluate a youth’s personal and social functioning across 16 domains. Items are rated on a 7-point scale (1 = none of the time, 4 = some of the time, 7 = all of the time). The composite scores for the subscales range from 0 to 100, with higher scores indicating greater problems in the respective domain of functioning. A previous study established reliability scores through the coefficient alpha; the coefficient alphas for 15 of the subscales range from .84 to .96, and .74 for the Depression subscale (Mathiesen, Scottye, & Hudson, 2002).

The Problem Oriented Screening Instrument for Teens (POSIT) is a self-report tool, comprised 139 yes/no questions, designed to evaluate a youth’s level of functioning across 10 domains, including family relationships, peer relationships, and educational status. The scoring of the POSIT is based on a three-category risk rating: low, middle, and high. Low scores indicate less risk in a respective area of functioning and high scores indicate high risk in a respective area of functioning. The POSIT has been widely used in clinical practice and juvenile correctional facilities and has been shown to have good reliability and validity with adolescents in juvenile correctional facilities (Danseco & Marques, 2002; Dembo, Schmeidler, Borden, Sue, & Manning, 1998; Latimer, Winters, & Stinchfield, 1997; McLaney, Boca, & Babor, 1994).

Antisocial behaviors

Three subscales of the MAAS were used to assess three antisocial behaviors: problem alcohol use, substance use, and aggression. The MAAS Alcohol Abuse subscale consists of 15 items. The MAAS Drug Use subscale consists of 10 items. The MAAS Aggression subscale consists of 10 items that assess tendencies toward aggressive and violent behavior. The clinical cutoff score for all three subscales is 15.

Psychological processes

Two subscales of the MAAS were used to assess psychological processes. The MAAS Depression subscale consists of 12 items that assess severity of depression. The MAAS Self-Esteem subscale consists of 12 items that assess problems related to self-esteem and the clinical cutoff score for both subscales is 30.

Life experiences: Family, peers, and schools

The MAAS and the POSIT were used to assess the life experience domains of family, peers, and schools. The MAAS Family Relationship Problems subscale consists of 13 items that assess the general family atmosphere and problems in family relationships; the clinical cutoff score is 30. The POSIT Family Relationship subscale consists of 11 items that assesses family management and parenting practices. The MAAS Problems with Friends subscales consists of 13 items that assess peer acceptance/rejection; the clinical cutoff score is 30. The POSIT Peer Relationships subscale consists of 10 items and assesses problems due to negative influences and negative behaviors by the youth’s peers, such as truancy, property damage, and theft. The MAAS Problems with School consists of 10 items and assesses school bonding/connection. The POSIT Educational Status consists of 26 items and screens for learning disabilities or academic underachievement due to problems with cognitive functioning.

External covariates

Several external covariates were used to characterize and validate the identified latent classes. Because “it is important that delinquency researchers illuminate the specific factors, such as demographic characteristics, antisocial behavior, or prior trauma, found in this [juvenile offenders] heterogeneous population” (Montgomery, Vaughn, Thompson, & Howard, 2013, p. 1341), the following variables were selected as covariates: race, age at current commitment, age at first court referral, the number of prior adjudications, traumatic experiences, and history of victimization (physical, sexual, or emotional abuse).

Moreover, because trauma, drug and alcohol abuse, and mental health problems are common among incarcerated girls, psychological assessments based on the DSM-IV-TR which classify mental health and substance use disorders were used to support the self-reports of the girls in the sample. Prior experiences of trauma were assessed using the MAYSI-2 Traumatic Experience subscale.

Data Analysis

Latent profile analysis was conducted to construct risk-need profiles in the sample. Latent profile analysis is a variant of latent class analysis in which a set of continuous indicator variables are used to construct latent classes (i.e., classes that are not observed directly). In latent profile analysis, a separate set of means, variances, and covariances for each latent class is estimated. Thus, one of the major objectives of this type of analysis is finding latent classes that differ in terms of their means (Magdison & Vermunt, 2002). For the current study, the latent profile analyses–derived classes reflect differences in levels or severity of risk-needs.

Models for the latent profile analysis were estimated with the software MPlus. Two to four classes were estimated, and each class was evaluated using multiple criteria: the Lo–Mendell–Rubin (LMR) test, the Bayesian information criteria (BIC), and entropy values. With the LMR test, a model with K classes can be compared with a model with (K + 1) classes. The LMR test generates a p value that can be used to determine whether there is a statistically significant improvement in fit for the inclusion of a model with one more class (Nylund, Asparouhov, & Muthén, 2007). The BIC is a global measure of parsimony that weighs the fit and parsimony of the model; the lower the BIC, the better the model. And entropy values measure how well the latent classes can be distinguished (Nylund et al., 2007). Values range from 0 to 1, and high entropy values (i.e., closer to 1.00) are indicative of a good-fitting model.

Results

Latent profile analysis was conducted to construct risk-need profiles in a sample-based population of incarcerated girls. On the basis of the LMR test, the BIC, and the entropy value, the model with four classes was selected as the best-fitting model. The LMR test was used to compare a model with K classes to a model with K – 1 classes. Although in the current study, two through four classes were fitted to the model, it has been suggested that the first time the p value of the LMR is nonsignificant, the researcher should stop increasing the number of classes (Nylund et al., 2007). In the current study, the first time the p value of the LMR was nonsignificant was with a three-class solution, suggesting a model with a three-class solution was sufficient relative to a two-class model. In terms of the fit indices, the BIC improved progressively from a two-class solution (17,387.540) to a four-class solution (17,031. 884). The BIC value indicated that the four-class solution was the best fit to the data; the lower the BIC, the better the model. With regard to the entropy value, the entropy value increased from a three-class solution (0.88) to a four-class solution (0.90). The results of the latent class models, comparison of models, and associated statistics are presented in Table 1.

Table 1.

Model Fit for Tests of Two- to Four-Class Solutions.

Number of classes	LMR test	BIC	Entropy value
2	0.0086	17,387.540	0.93
3	0.1056	17,948.581	0.88
4	0.2185	17,031.884	0.90

Note. LMR = Lo–Mendell–Rubin; BIC = Bayesian information criteria.

Class Membership

The analyses revealed that half of the sample (51%) was in Class I, Aggression Only, with an average probability of .95; 19% of the population was in Class II, Aggression and Drug Use, with an average probability of .98; 24% of the population was in Class III, Socioemotional and Family Relationships, with an average probability of .91; and 6% were in Class IV, Severe Alcohol and Drug Use, with an average probability of .99. The mean values for each subscale across each class are illustrated in Figure 1.

Figure 1.

Risk-need profiles.

Class I: Aggression Only

Girls in Class I, the largest class (n = 104), were primarily African American (76.9%) and largely incarcerated for the commission of a crime against property (50%). Girls in this class displayed low to moderate risk-needs. The only clinically elevated or high-risk score in this class was on the subscale of aggression. The girls in this class had middle-risk scores on the subscales of family relationships (i.e., parenting practices), peer relationships, and educational status.

Class II: Aggression and Drug Use

Girls in Class II (n = 38), were primarily Caucasian (52.6%) and largely incarcerated for the commission of a crime against person (47.4%). The girls in this class had clinically elevated scores or high-risk scores on seven of the 11 domains measured—depression, problems with school, aggression, family relationship problems, alcohol abuse, drug use, and peer relationships (i.e., association with delinquent peers)—and middle-risk scores on the subscales of family relationships (i.e., parenting practices) and educational status.

Nearly, half of the girls in this class had a history of victimization. Nearly all girls, 97.4%, in this class met criteria for a DSM IV-TR substance use disorder.

Class III: Socioemotional and Family Relationship Problems

Girls in Class III (n = 48) were primarily African American (87.5%) and largely incarcerated for the commission of a crime against person (56.3%). The girls in this class had clinically elevated scores or high-risk scores on eight of the 11 domains measured. Similar to girls in the Aggression Only class, girls in the Socioemotional and Family Relationship Problems class reported relatively low problems related to alcohol use and substance use and problems with friends and high problems related to aggression. Girls in this class, however, had family relationship problems (i.e., parenting practices), deviant peer associations, and problems related to self-esteem, depression, and school bonding/connection.

Class IV: Severe Alcohol and Drug Use

Girls in Class IV, the smallest class (n = 13), were primarily Caucasian (53.8%) and mainly incarcerated for the commission of a crime against property (61.5%). Girls in this class reported clinically elevated problems related to alcohol use and drug use. Their level of problem severity related to alcohol use and drug use, however, was much higher than girls in the Aggression and Drug Use class. Their level of problem severity related to aggression was also extremely high. Girls in this class had clinically elevated or high-risk scores on nine of the 11 measured domains. All girls in this class met criteria for a DSM IV-TR substance use disorder.

Characterization and Validation of the Four Latent Classes

To further characterize and validate the classes, chi-square analysis and analysis of variance (ANOVA) were conducted to test for differences across the four classes on the sociodemographic and offense history characteristics listed in Table 2. Chi-square tests revealed differences across classes by race—χ²(6, N = 203) = 33.72, p < .001. Class III, Socioemotional and Family Relationship Problems, had the highest percentages of African American girls (87.5%) and Class IV, Severe Alcohol and Drug Use, had the lowest percentages of African American girls (30.8%). Conversely, Class IV, Severe Alcohol and Drug Use, had the highest percentages of Caucasian girls (53.8%) and Class III, Socioemotional and Family Relationship Problems, had the lowest percentages of Caucasian girls (10.4%).

Table 2.

Comparison of Demographic, Antisocial Behavior, Prior Trauma, Substance Use, and Mental Health Characteristics Across Classes.

	Class I	Class II	Class III	Class IV	χ²/F
	Aggression Onlyn = 104	Aggression and Drug Usen = 38	Socioemotional and Family Problemsn = 48	Severe alcohol and Drug Usen = 13
	% or M (SD)	% or M (SD)	% or M (SD)	% or M (SD)
Race					χ² = 33.732*
African American	76.9	44.7	87.5	30.8
Caucasian	17.3	52.6	10.4	53.8
Other	5.8	2.7	2.1	15.4
Age	14.91 (1.05)	14.84 (0.916)	14.88 (0.890)	15.15 (1.07)	F = 0.446
Age-first court referral	12.90 (1.52)	13.09 (1.28)	12.75 (1.39)	2.67 (0.985)	F = 0.287
Prior adjudications	2.52 (1.52)	2.79 (2.45)	2.93 (1.69)	3.17 (1.47)	F = 0.427
Traumatic experiences	1.62 (1.48)	2.86 (1.58)	2.33 (1.87)	1.91 (1.97)	F = 5.92*
History of victimization	23.8	47.4	34.0	23.1	χ² = 7.622
Diagnosable MH disorder	99.0	97.2	95.7	83.3	χ² = 5.919
Type of MH disorder
Adjustment	3.0	5.6	0.0	10.0	χ² = 4.245
Anxiety	5.0	11.1	8.7	8.3	χ² = 1.687
Attention deficit	18.0	19.4	30.4	25.0	χ² = 2.929
Conduct	90.0	77.8	87.0	58.3	χ² = 8.584*
Learning	8.0	2.8	15.2	0.0	χ² = 6.364
Mood	29.0	50.0	50.0	33.3	χ² = 8.546*
Oppositional defiant	7.0	2.8	2.2	8.3	χ² = 2.395
Diagnosable SU disorder	58.7	97.4	60.4	100	χ² = 36.731
Type of SU disorder
Alcohol	27.9	45.9	41.4	53.8	χ² = 5.222
Amphetamine	0.0	2.7	3.4	0.0	χ² = 3.071
Cannabis	90.2	59.5	96.6	46.2	χ² = 27.017*
Cocaine	13.1	16.2	13.8	7.7	χ² = .666
Hallucinogen	1.6	0.0	3.4	0.0	χ² = 2.060
Polysubstance	3.3	29.7	3.4	38.5	χ² = 22.529*
Sedative	0.0	2.7	3.4	0.0	χ² = 6.382

Note. MH = mental health; SU = substance use.

p < .01.

One-way ANOVA tests revealed that girls in the Aggression and Drug Use class had the highest score on the MAYSI-2 Traumatic Experience subscale. Subsequent analyses indicated that girls in the Aggression and Drug Use class had significantly higher mean scores on the Traumatic Experience subscale than girls in the Aggression Only class. Although not significantly different from girls in other classes, girls in the Aggression Only class had the highest percentage of having a history of victimization.

Similarly, chi-square analyses and ANOVA were conducted to test for differences across the four classes on the substance use and mental health disorders listed in Table 2. Chi-square analyses revealed differences across classes on meeting criteria for a substance use disorder—χ²(3, N = 203) = 36.731, p < .01. Most notably, girls in the Severe Alcohol and Drug Use (100%) and Aggression and Drug Use (97.4%) classes had higher rates of meeting criteria for a substance use disorder than girls in the Aggression Only (58.7%) and Socioemotional and Family Relationship Problems (60.4%) classes. Follow-up chi-square tests were conducted to test for differences across the four classes on meeting criteria for specific types of substance use disorders. Chi-square analyses revealed differences across classes on meeting criteria for a cannabis disorder—χ²(3, n = 140) = 27.017, p < .01—and for a polysubstance use disorder—χ²(3, n = 140) = 22.529, p < .01. Class III, Socioemotional and Family Relationship Problems, had the highest percentage of girls with a diagnosable substance use disorder (96.6%) who met criteria for a cannabis disorder and Class IV, Severe Alcohol and Drug Use, had the lowest percentage (46.2%). Class IV, Severe Alcohol and Drug Use, had the highest percentage of girls with a diagnosable substance use disorder (38.5%) who met criteria for a polysubstance disorder and Class I, Aggression Only, had the lowest (3.3%).

Chi-square analysis revealed that there were no significant differences across classes in meeting criteria for a mental health disorder. However, rates of meeting criteria for specific types of mental health disorders differed. Rates of meeting criteria for a conduct disorder—χ²(3, N = 203) = 8.584, p < .01—and a mood disorder—χ²(3, N = 203) = 8.546, p < .01—differed significantly across classes. Class I, Aggression Only, had the highest percentage of girls (90%) who met criteria for a conduct disorder and Class IV, Severe Alcohol and Drug Use, had the lowest percentage (58.3%). Classes II and III had the highest percentage of girls (50% respectively) who met criteria for a mood disorder and Class I, Aggression Only, had the lowest percentage (29%).

Discussion

Results of the latent class analysis revealed four subgroups of female juvenile offenders with distinct risk-need profiles: Aggression Only, Aggression and Drug Use, Socioemotional and Family Relationship Problems, and Severe Alcohol and Drug Use. Half of the sample in the present study was in the Aggression Only class (the largest class). Girls in this class displayed low to moderate risk-needs. The only clinically elevated or high-risk score in this class was on the subscale of aggression. Compared with the other classes, girls in this class had the lowest level of problem severity or risk-needs. However, they had one of the highest rates meeting criteria for a conduct disorder. The characteristics of girls in the Aggression Only class appear to be consistent with traits that are conducive to an adolescence-limited developmental pathway of antisocial behavior (i.e., antisocial or delinquent behavior that appears for the first time in adolescence and that does not persist into adulthood) identified by Moffitt (1993).

Conversely, girls in the Severe Alcohol and Drug Use group had clinically elevated scores on nine of the 11 subscales. Compared with the other classes, girls in this class had the highest level of problem severity or risk-needs. The profile of girls in this class appears to coincide with the characteristics of the delayed-onset pathway of antisocial behavior proposed by Silverthorn and Frick (1999). In this pathway, “girls are hypothesized to share many of the vulnerabilities of the early onset boys [i.e., Moffitt (1993) description of childhood-onset/life-course persistent offenders] but do not manifest severe antisocial behavior until adolescence when there are significant changes in girl’s biological and social milieu” (Silverthorn & Frick, 1999, p. 122). These characteristics include cognitive/neurological dysfunction, negative family histories, problematic family relationships, dysfunctional parenting practices, substance abuse and marked aggression (Silverthorn & Frick, 1999). Given their risk-need profile, girls in this class may be more likely to persist with antisocial behavior into adulthood.

External Covariates

The examination of the covariate of race revealed that there were racial differences across classes. Compared with African American girls, Caucasian girls were overrepresented in the two classes with the highest level of risk-needs: Severe Alcohol and Drug Use and Aggression and Drug Use. All girls in Severe Alcohol and Drug Use class and nearly all girls in the Aggression and Drug Use had diagnosable substance use disorders. These findings are consistent with studies examining substance use among female juvenile offenders that Caucasian girls exhibit much higher levels of problem substance use than African American girls (Holsinger & Holsinger, 2005; Mulvey, Schubert, & Chaissin, 2010; Welch-Brewer, Stoddard-Dare, & Mallett, 2011). Moreover, girls in the Aggression and Drug Use class had the highest mean number of traumatic experiences and the highest percentage of victimization (although not significantly different from other classes).

Similarities Across Classes

There were similarities across all classes. Across all classes, girls had clinically elevated scores on the subscale of aggression, indicating that aggression is a problem among all girls in the sample. In addition, across all classes, the mean scores on the subscale of problems with friends (e.g., peer acceptance/rejection) were in the subclinical range, indicating that girls in the sample reported minimal problems related to peer rejection. Conversely, scores on the domain of peer relationships (i.e., association with deviant peers) were in the moderate- and high-risk categories. These findings are consistent with the conceptualization and evidence that the relationships of antisocial adolescents are not always characterized by low bonding or low attachment but are usually characterized by high bonds and attachments to antisocial or deviant peers (Ayers et al., 1999; Catalano & Kosterman, 1996).

There were similarities across classes in the rates of meeting criteria for a mental health disorder and for specific types of mental health disorders, including anxiety disorder, attention deficit hyperactivity disorder, oppositional defiant disorder, and learning disorder. In general, the rates of meeting criteria for these disorders appeared to be evenly distributed across classes. Moreover, the rates of having a history of victimization appeared to be evenly distributed across classes.

Implications for Female Juvenile Correctional Programming and Practice

The variation in risk-need profiles across classes indicates heterogeneity within the sample and suggests to service providers that the needs of female juvenile offenders may not be met optimally using fixed intervention strategies based on a single uniform composition and dosage (Collins, Murphy, & Bierman, 2004). There is a need to adopt adaptive treatment strategies that tailor the treatment components, treatment composition, and dosage to the needs of each group (Murphy & McKay, 2003) or client matching strategies that match an individual to a particular treatment program, treatment modality, or set of interventions based on the characteristics and treatment needs of the individual. Meta-analytic studies have shown that interventions or programs based on client-treatment matching strategies were more effective in reducing recidivism among juvenile and adult offenders than those programs that did not use these strategies (Dowden & Andrews, 1999a, 1999b). By adopting adaptive treatment or client matching strategies, correctional administrators and practitioners can make better use of available resources by distributing interventions among juvenile offenders on the basis of the needs and characteristics of the offenders (Murphy, & McKay, 2003).

The current findings indicate that the level, dosage, and type of interventions provided to subgroups of female juvenile offenders should be tailored to their risk-need profiles. For example, girls in the Aggression and Drug Use and Severe Alcohol and Drug Use classes may benefit from highly structured, intensive substance abuse treatment, such as that provided within an in-custody therapeutic community. Girls in both groups would also benefit from intensive cognitive behavioral therapy that focuses on stress and anger management, problem solving, and relapse prevention therapy based on cognitive behavior principles such as self-monitoring to recognize drug cravings and to identify high-risk situations for use and strategies for refusing drugs, coping with and avoiding high-risk situations and the desire to use, resisting peer pressure to use drugs (Carroll, Rounsaville, & Keller, 1991; Roberts & Welch, 2008). They would also benefit from cognitive behavioral approaches that address problems related to depression. Approaches that have been found to be effective in reducing depressive symptoms and improving social functioning among a sample of juvenile justice involved youths include those that focus on monitoring moods, improving social skills, increasing pleasant activities, decreasing anxiety, reducing depressogenic cognitions, and improving communication and conflict resolution (Rohde, Clarke, Mace, Jorgensen, & Seeley, 2004). Girls in the Aggression and Drug Use class, however, may benefit from cognitive behavioral therapy with a relational approach to better respond to their history of trauma and victimization, mood disorders, aggression, and alcohol and drug use problems.

Girls in the Severe Alcohol and Drug Use class may require more extensive treatment (i.e., beyond the period of 6 months) than girls in the Aggression and Drug Use class because of their extremely high clinically elevated problems related to alcohol and drug use and their high rates and levels of problem severity related to other risk and need factors. Because of the possible deficits in cognitive functioning observed in girls in this class, they may need to be assessed periodically to determine their readiness or ability to learn the cognitive skills necessary to benefit from cognitive behavioral interventions and approaches. They would also benefit from remedial educational classes, which are an essential element of therapeutic communities (Roberts & Welch, 2008; Winters, Latimer, & Stinchfield, 2001).

Conversely, because girls in the Aggression Only and Socioemotional and Family Relationship Problems classes reported subclinical problems related to alcohol and substance use, girls with a substance use disorder in these respective classes would probably benefit from substance abuse treatment that is less intensive than that provided within a therapeutic community. Substance abuse treatment in the form of individual or group counseling may be more optimal for girls in the Aggression Only class, whereas family-based substance abuse treatment may be more optimal for girls in the Socioemotional and Family Relationship Problems class. Moreover, because many incarcerated girls have a high likelihood of developing substance use problems in adulthood (Lanctôt, Cernkovich, & Giordano, 2007; Lewis et al., 1991), girls without a substance use disorder in the Aggression Only and Socioemotional and Family Relationship Problems classes may benefit from psychoeducational group interventions modeled after prevention education strategies. These strategies include providing information about drugs and alcohol, exploring expectancies and consequences of alcohol and substance use, and providing skills-based training related to drug-refusal skills.

Similar to girls in the Aggression and Drug Use and Severe Alcohol and Drug Use classes, girls in the Socioemotional and Family Relationship Problems class may benefit from interventions targeting their depression. And given their family problems, girls in the Socioemotional and Family Relationship Problems class, as well as girls in the Severe Alcohol and Drug Use class, would benefit from intensive family-based interventions, such as multisystemic therapy, while girls in the Aggression and Drug Use class would benefit from brief family-based interventions, such as family functional therapy.

Despite the heterogeneity across classes, there were some common risk-needs among the girls in the sample. There were common risk-needs related to aggression and association with deviant peers (with the latter ranging from moderate to high risk), indicating that all girls in the sample would benefit from a core set of cognitive behavioral interventions that incorporate social skills training, behavioral management, and self-regulation skills to increase their social and cognitive skills and decrease their aggressive and delinquent behavior and their association with deviant peers (Hipwell & Loeber, 2006).

Conclusion

The findings of the present study add to the small, but growing, body of evidence that serious female juvenile offenders are a heterogeneous group in terms of their risk-need profiles (Espelage et al., 2003; Odgers et al., 2007; Stefurak & Calhoun, 2007). The findings show that the configuration of risk-needs and the level of severity differ across subgroups of female juvenile offenders, warranting the need for varying levels of treatment intensity and different treatment components across subgroups, ranging from less to more extensive. Accordingly, the developers of interventions and programs cannot presume that similar approaches to addressing and treating delinquency and other problem behaviors in girls are warranted. Interventions and treatment strategies should be developed to address the heterogeneous needs of serious female juvenile offenders.

Limitations of Present Study and Future Research

Several limitations of the present study should be noted. First, the findings of the present study were drawn from one site. As such, the findings are state-specific and may not generalize to samples or populations of girls incarcerated in other states. Thus, this study needs to be replicated using samples of incarcerated juvenile females in different states. Second, the size of the sample was relatively small, and many of the measures used in the present study were based on self-report data. Future research would benefit from the use of larger sample sizes and the use of multi-informant measures. Third, the present study was cross-sectional. Given the cross-sectional nature of the current study, only pretreatment characteristics of the sample were examined. Future research is needed that examines the treatment outcomes and posttreatment trajectories of girls with different risk-need profiles. Moreover, as suggested by Vaughn, Freedenthal, Jenson, and Howard (2007), the statistical technique of latent transition analysis could be used to assess whether the base-line risk-need profiles change over time as a function of programming within juvenile correctional facilities and to assess risk-need profiles at the time of discharge to guide community reentry/after care services plan. Also, there is a need for future studies to examine and identify which types of profiles are associated with reoffending and to determine if patterns of differences and similarities are observed across race/ethnicity. Addressing these factors may enhance the treatment outcomes of serious female juvenile offenders involved in the juvenile justice 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.

References

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text rev.). Washington, DC: Author.

Ayers

C. D.

Williams

J. H.

Hawkins

D. J.

Peterson

P. L.

Catalano

R. F.

Abbott

R. D.

(1999). Assessing correlates of onset, escalation, de-escalation, and desistance of delinquent behavior. Journal of Quantitative Criminology, 153, 277-306. doi:10.1023/A:1007576431270

Bergman

(2001). A person approach in research on adolescence: Some methodological challenges. Journal of Adolescent Research, 16, 28-53. doi:10.1177/0743558401161004

Carroll

Rounsaville

Keller

(1991). Relapse prevention strategies for the treatment of cocaine abuse. American Journal of Drug and Alcohol Abuse, 17 (3), 249-265.

Catalano

R. F.

Kosterman

(1996). Modeling the etiology of adolescent substance use: A test of the social development model. Journal of Drug Issues, 26, 429-456. doi:10.1177/002204269602600207

Collins

L. M.

Murphy

S. A.

Bierman

K. L.

(2004). A conceptual framework for adaptive preventive interventions. Prevention Science, 5, 185-196. doi:10.1023/B:PREV.0000037641.26017.00

Colman

R. A.

Kim

D. H.

Mitchell-Herzfeld

Shady

T. A.

(2009). Delinquent girls grown up: Young adult offending patterns and their relation to early legal, individual, and family risk. Journal of Youth and Adolescence, 38, 355-366. doi:10.1007/s10964-008-9341-4

Danseco

E. R.

Marques

P. R.

(2002). Development and validation of a POSIT-Short Form: Screening for problem behaviors among adolescents at risk for substance use. Journal of Child & Adolescent Substance Abuse, 11, 17-36. doi:10.1300/J029v11n0302

Dembo

Schmeidler

Borden

Sue

C. C.

Manning

(1998). Use of the POSIT among arrested youths entering a juvenile assessment center: A replication and update. Journal of Child & Adolescent Substance Abuse, 6, 19-42. doi:10.1300/J029v06n03_02

10.

Dixon

Howie

Starling

(2004). Psychopathology in female juvenile offenders. Journal of Child Psychology and Psychiatry, 45, 1150-1158. doi:10.1111/j.1469-7610.2004.00307.x

11.

Dodge

K. A.

Pettit

G. S.

(2003). A biopsychosocial model of the development of chronic conduct problems in adolescence. Developmental Psychology, 39, 349-371. doi:10.1037/0012-1649.39.2.349

12.

Dowden

Andrews

D. A.

(1999a). What works for female offenders: A meta-analytic review. Crime & Delinquency, 45, 438-452. doi:10.1177/0011128799045004002

13.

Dowden

Andrews

D. A.

(1999b). What works in young offender treatment: A meta-analysis. Forum on Corrections Research, 112, 21-24.

14.

Espelage

D. L.

Cauffman

Broidy

Piquero

A. R.

Mazerolle

Steiner

(2003). A cluster-analytic investigation of MMPI profiles of serious male and female juvenile offenders. Journal of American Academy of Child Adolescent Psychiatry, 42(7), 770-777.

15.

Hipwell

A. E.

Loeber

(2006). Do we know which interventions are effective for disruptive and delinquent girls? Clinical Child and Family Psychology Review, 9, 221-255.

16.

Holsinger

A. M.

(2005). Differential pathways to violence and self-injurious behavior: African American and White girls in the juvenile justice system. Journal of Research in Crime & Delinquency, 42, 211-242. doi:10.1177/0022427804271938

17.

Hubbard

D. J.

Pratt

T. C.

(2002). A meta-analysis of the predictors of delinquency among girls. Journal of Offender Rehabilitation, 34(3), 1-13. doi:10.1300/j076v34n03_01

18.

Huizinga

Loeber

Thornberry

Cothern

(2000). Co-occurrence of delinquency and other problem behaviors. Washington, DC: Office of Juvenile Justice and Delinquency Prevention, U.S. Department of Justice.

19.

Kataoka

S. H.

Zima

B. T.

Dupre

D. A.

Moreno

K. A.

Yang

McCracken

J. T.

(2001). Mental health problems and service use among female juvenile offenders: Their relationship to criminal history. Journal of the American Academy of Child & Adolescent Psychiatry, 40, 549-555. doi:10.1097/00004583-200105000-00014

20.

Kempf-Leonard

Tracey

P. E.

Howell

J. C.

(2001). Serious, violent, and chronic juvenile offenders: The relationship of delinquency career types to adult criminality. Justice Quarterly, 18, 449-478. doi:10.1080/07418820100094981

21.

Lanctôt

Cernkovich

S. A.

Giordano

(2007). Delinquent behavior, official delinquency, and gender: Consequences for adulthood functioning and well-being. Criminology, 45, 131-157. doi:10.1111/j.1745-9125.2007.00074.x

22.

Latimer

W. W.

Winters

K. C.

Stinchfield

R. D.

(1997). Screening for drug abuse among adolescents in clinical and correctional settings using the Problem-Oriented Screening Instrument for Teenagers. The American Journal of Drug and Alcohol Abuse, 23, 79-98. doi:10.3109/00952999709001689

23.

Leve

L. D.

Chamberlain

Kim

H. K.

(2015). Risks, outcomes, and evidence-based interventions for girls in the U.S. juvenile justice system. Clinical Child and Family Psychology Review, 18, 252-279. doi:10.1007/s10567-015-0186-6

24.

Lewis

D. O.

Yeager

C. A.

Cobham-Portorreal

C. S.

Klein

Showalter

Anthony

A. A.

(1991). A follow-up of female delinquents: Maternal contributions to the perpetuation of deviance. Journal of the American Academy of Child & Adolescent Psychiatry, 30, 197-201. doi:10.1097/00004583-199103000-00006

25.

Magdison

J. V.

Vermunt

J. K.

(2002). Latent class models for clustering a comparison with k-means. Canadian Journal of Marketing Research, 20, 37-44.

26.

Mathiesen

S. G.

Scottye

J. C.

Hudson

(2002). The Multidimensional Adolescent Assessment Scale: A validation study. Research on Social Work Practice, 12, 9-28. doi:10.1177/104973150201200103

27.

Matthews

Hubbard

D. J.

(2008). Moving ahead: Five essential elements for working effectively with girls. Journal of Criminal Justice, 36, 494-502. doi:10.1007/s10567-006-0012-2

28.

McCabe

Lansing

Garland

Hough

(2002). Gender differences in psychopathology, functional impairment, and familial risk factors among adjudicated delinquents. Journal of American Academy of Child & Adolescent Psychiatry, 41, 860-867. doi:10.1097/00004583-200207000-00020

29.

McLaney

M. A.

Boca

F. D.

Babor

(1994). A validation study of the Problem-Oriented Screening Instrument for Teenagers (POSIT). Journal of Mental Health, 3, 363-376. doi:10.3109/09638239408997947

30.

Moffitt

T. E.

(1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychology Review, 100, 674-701. doi:10.1037/0033-295X.100.4.674

31.

Montgomery

K. L.

Vaughn

M. G.

Thompson

S. J.

Howard

M. O.

(2013). Heterogeneity in drug abuse among juvenile offenders: Is mixture regression more informative than standard regression? International Journal of Offender Therapy and Comparative Criminology, 57, 1326-1346. doi:10.1177/0306624X12459185

32.

Mulvey

E. P.

Schubert

C. A.

Chaissin

(2010). Substance use and delinquent behavior among serious adolescent offenders. Washington, DC: Office of Juvenile Justice and Delinquency Prevention, Office of Justice Programs, U.S. Department of Justice.

33.

Murphy

S. A.

McKay

J. R.

(2003). Adaptive treatment strategies: An emerging approach for improving treatment effectiveness. Clinical Science: Newsletter of the Society for a Science of Clinical Psychology, Section III of Division 12 of the APA, pp. 7–13.

34.

Nylund

K. L.

Asparouhov

Muthén

B. O.

(2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14, 535-569. doi:10.1080/10705511.2014.947375

35.

Odgers

C. L.

Moretti

M. M.

Burnette

M. L.

Chauhan

Waite

Reppucci

N. D.

(2007). A latent variable modeling approach to identifying subtypes of serious and violent female juvenile offenders. Aggressive Behavior, 33, 339-352. doi:10.1002/ab.20190

36.

Rhoades

K. A.

Leve

L. D.

Eddy

J. M.

Chamberlain

(2016). Predicting the transition from juvenile delinquency to adult criminality: Gender specific influences in two high-risk samples. Criminal Behavior and Mental Health, 26, 336-351. doi:10.1002/cbm.1957

37.

Rich

S. L.

Wilson

J. K.

Robertson

A. A.

(2016). The impact of abuse trauma on alcohol and drug use: A study of high-risk incarcerated girls. Journal of Child & Adolescent Substance Abuse, 25, 194-205. doi:10.1080/1067828X.2015.1007197

38.

Roberts

A. C.

Welch

(2008). Incarcerated female teens and substance abuse: The holistic enrichment for at-risk teens (heart) program. Handbook of Prevention and Intervention Programs for Adolescent Girls, 270-303. doi:10.1002/9781118269848.ch9

39.

Roe-Sepowitz

D. E.

(2009). Comparing male and female juveniles charged with homicide: Child maltreatment, substance abuse, and crime details. Journal of Interpersonal Violence, 24, 601-617. doi:10.1177/0306624X11411152

40.

Rohde

Clarke

G. N.

Mace

D. E.

Jorgensen

J. S.

Seeley

J. R.

(2004). An efficacy/effectiveness study of cognitive-behavioral treatment for adolescents with co-morbid major depression and conduct disorder. Journal of the American Academy of Child & Adolescent Psychiatry, 43, 660-668. doi:10.1097/01.chi.0000121067.29744.41

41.

Silverthorn

Frick

(1999). Developmental pathways to antisocial behavior: The delayed onset pathways in girls. Development & Psychopathology, 11, 101-126. doi:10.1017/S0954579499001972

42.

Stefurak

Calhoun

G. B.

(2007). Subtypes of female juvenile offenders: A cluster analysis of the Millon Adolescent Clinical Inventory. International Journal of Law and Adolescence, 30, 95-111. doi:10.1016/j.ijlp.2006.04.003

43.

Steketee

Junger

Junger-Tas

(2013). Sex differences in the predictors of juvenile delinquency: Females are more susceptible to poor environments; males are influenced more by low self-control. Journal of Contemporary Criminal Justice, 29, 88-105.

44.

van der Put

C. E.

Deković

Hoeve

Stams

G. J. J.

van der Laan

P. H.

Langewouters

F. E

. (2014). Risk assessment of girls: Are there any sex differences in risk factors for re-offending and in risk profiles? Crime & Delinquency, 60, 1033-1056.

45.

Vaughn

M. G.

Freedenthal

Jenson

J. M.

Howard

M. O.

(2007). Psychiatric symptoms and substance use among juvenile offenders: A latent profile investigation. Criminal Justice and Behavior, 34, 1296-1312. doi:10.1177/0093854807304624

46.

Welch-Brewer

C. L.

Stoddard-Dare

Mallett

C. A.

(2011). Race, substance abuse, and mental health disorders as predictors of Juvenile court outcomes: Do they vary by gender? Child and Adolescent Social Work Journal, 28, 229-241. doi:10.1007/s10560-011-0229-x

47.

Winters

K. C.

Latimer

W. W.

Stinchfield

(2001). Assessing adolescent substance use. In Wagner

E. F.

Waldron

H. B.

(Eds.), Innovations in adolescent substance abuse interventions (pp. 1-29). New York: Pergamon.