The Exploration of Risk and Protective Score Differences Across Juvenile Offending Career Types and Their Effects on Recidivism

Abstract

Despite an abundance of research on serious and violent juvenile offenders, few studies have linked juvenile offending career categories to juvenile court risk assessments and future offending. This study uses juvenile court referrals and assessment data to replicate earlier categorizations of serious, violent, and chronic offenders; to examine risk and protective score differences across these categories; and to assess whether risk and protective score constructs differentially predict adult criminality across these offender categories. Based on a sample of 9,859 juvenile offenders who aged out of Connecticut’s juvenile justice system between 2005 and 2009, we found that (1) our categorization of juvenile career types mirrored earlier work, (2) comparing risk and protective factors across and within juvenile career types identified distinct patterns, and (3) the juvenile risk and protective assessment subscales were not predictive of adult arrests for chronic offenders but were predictive for nonchronic juvenile career types.

Keywords

juvenile offending career types juvenile risk assessments juvenile recidivism

Delinquency theory and research have commonly categorized juvenile offenders based on the seriousness, violent nature, and chronicity of their offending careers (Baglivio, Jackowski, Greenwald, & Howell, 2014; Howell, 2003a; Howell, Lipsey, Wilson, & Howell, 2014; Johansson & Kempf-Leonard, 2009; Kempf-Leonard, Tracy, & Howell, 2001; Loeber & Ahonen, 2014; Loeber & Farrington, 2000; Loeber, Farrington, & Waschbusch, 1998; Snyder, 1998; Vaughn & Howard, 2005; Wilson & Howell, 1993). The Office of Juvenile Justice and Delinquency Prevention’s (OJJDP) efforts in the 1980s and 1990s led to accepted research definitions of seriousness, violence, and chronicity. Snyder (1998), in particular, empirically validated eight categories of juvenile offenders using juvenile court data: serious, violent, and chronic (SVC); serious and violent; serious and chronic; violent and chronic; serious; violent; chronic; and not serious, not violent, not chronic. His analysis of court referrals supported prior self-reported delinquency research that found a small percentage of juvenile offenders are SVC offenders, but they commit a significant amount of juvenile crime (Hamparian, Schuster, Dinitz, & Conrad, 1978; Moffitt, 1993; Nagin & Farrington, 1992; Nagin, Farrington, & Moffitt, 1995; Wolfgang, Figlio, & Sellin, 1972).

Since the early 1990s, a significant amount of research has focused on SVC juvenile offenders. This research has consistently found that (1) a small percentage of juveniles commit a majority of juvenile crime (Baglivio et al., 2014; Barnes, 2013; DeLisi & Piquero, 2011; Vaughn et al., 2011; Vaughn, Salas-Wright, DeLisi, & Maynard, 2014), (2) early age of onset of antisocial behavior leads to an escalation of offense seriousness and longer juvenile offending careers (Ayers et al., 1999; Baglivio, Wolff, Piquero, & Epps, 2015; DeLisi, 2006; DeLisi, Neppl, Lohman, Vaughn, & Shook, 2013; DeLisi & Piquero, 2011; Howell et al., 2014; Loeber & Farrington, 2000; Loeber, Stouthamer-Loeber, Van Kammen, & Farrington, 1991; McCluskey, McCluskey, & Bynum, 2006; Moffitt, 1993; Nagin et al., 1995; Piquero, Brame, & Lynam, 2004; Zara & Farrington, 2013), and (3) the number of risk factors present or the number of absent protective factors is more predictive of SVC offending than any particular type of risk or protective factors (Borduin & Ronis, 2012; Hawkins et al., 1998; Huizinga & Jakob-Chien, 1998; Johansson & Kempf-Leonard, 2009; Krohn, Lizotte, Bushway, Schmidt, & Phillips, 2014; Loeber & Farrington, 2000; Valois, MacDonald, Bretous, Fischer, & Drane, 2002; Zara & Farrington, 2013).

Despite the abundance of research on serious and violent juvenile offenders, only a few studies have linked juvenile offending career categories to juvenile court risk assessments and future offending. In fact, a common research recommendation has been to use juvenile court risk assessments to identify risk and protective score differences between SVC subgroups and assess the predictive value of these on recidivism (Baglivio et al., 2014; Guerra, 1998; Kempf-Leonard et al., 2001). The current study is based upon this recommendation and contributes to the SVC literature by using juvenile court referrals and assessments to replicate earlier categorizations of SVC offenders, to examine risk and protective score differences across the juvenile offender categories, and to assess whether risk and protective score constructs differentially predict adult criminality across these offender categories.

Literature Review

Development and Application of SVC Categories

The OJJDP established the Violent Juvenile Offender Research and Development Program in 1981 to assess the juvenile justice system’s ability to be innovative when dealing with SVC juvenile offenders (Wilson & Howell, 1993). This initial effort led to the OJJDP Program of Research on the Causes and Correlates of Delinquency in 1986 which consisted of three coordinated research projects (Browning & Huizinga, 1999; Browning & Loeber, 1999; Browning, Thornberry, & Porter, 1999). These longitudinal studies of urban youth included the Denver Youth Survey (Huizinga, Weiher, Menard, Espiritu, & Esbensen, 1998), the Pittsburgh Youth Study (Loeber, Farrington, Stouthamer-Loeber, Moffitt, & Caspi, 1998), and the Rochester Youth Development Study (Thornberry, Krohn, Lizotte, Smith, & Porter, 1998). The initiative culminated in the formation of an OJJDP study group of experts representing different aspects of serious and violent juvenile offender research and practice (Loeber & Farrington, 1998).

The OJJDP initiative added to existing knowledge of the causes of serious juvenile offenders and, perhaps most importantly, synthesized juvenile research across various academic disciplines to formulate a comprehensive strategy to address SVC juvenile offenders (Howell, Krisberg, Hawkins, & Wilson, 1995; Loeber & Farrington, 1998; Wilson & Howell, 1993). While scholars used the terms “serious,” “violent,” and “chronic” to describe juvenile offenders prior to the OJJDP initiative, it was this initiative and its many subsequent publications that added the phrase “SVC juvenile offenders” to the juvenile justice vernacular and provided definitive definitions of them.

Snyder’s (1998) work was one of the first attempts to empirically validate these categories using juvenile court data. He analyzed 151,209 juvenile offender careers from Maricopa County, AZ, who had “graduated” out of juvenile court from 1980 to 1995 (i.e., those juvenile offenders who turned 18 years old and aged out of the juvenile court system). Violent offenses consisted of at least one juvenile court referral for murder, nonnegligent manslaughter, kidnapping, arson, violent sexual assault, robbery, and aggravated assault. Youth were considered serious and nonviolent offenders if they had at least one referral for burglary, serious larceny, motor vehicle theft, weapons offenses, and drug trafficking. Nonserious delinquent offenses were simple assault, possession of controlled substances, disorderly conduct, vandalism, nonviolent sex offenses, minor larceny, liquor law offenses, and all other delinquent offenses (status offenses and traffic violations were excluded). Youth were considered chronic if they had four or more referrals throughout a juvenile’s offending career.

Snyder’s study supported earlier findings that a small number of juvenile offenders were SVC, but they account for a majority of referrals to juvenile court (Hamparian et al., 1978; Wolfgang et al., 1972). Snyder found that 15% of offenders were considered chronic and accounted for 45% of all juvenile referrals and 60% of violent referrals (Wolfgang et al. found that 18% of males were involved in 52% of all delinquent acts). Snyder (1998) also found that 3.3% of his sample could be defined as SVC juvenile offenders, while 17.7% were serious, 7.9% were serious and chronic, 3.2% were violent, 2.5% were chronic, 0.9% chronic and violent, and 0.7% were serious and violent. The majority of juveniles (63.9%) referred to court were nonchronic, nonserious, and nonviolent offenders.

Snyder’s (1998) analysis has been replicated in other jurisdictions using different types of data (Baglivio et al., 2014; Kempf-Leonard et al., 2001; Loeber, Van Kammen, & Fletcher, 1996 as cited in Loeber et al., 1998). A comparison of these studies shows fairly similar results (Table 1). Of the three studies that provided information about nonserious, nonviolent, and nonchronic, it was found that these youth represented a high percentage of court referrals or police contacts. In addition, all four studies found that close to 30% of serious offenders were chronic offenders. A primary difference across the four studies is observed when looking at chronic offenders who were also violent and for violent offenders who were also chronic. Kempf-Leonard, Tracy, and Howell (2001) and Baglivio, Jackowski, Greenwald, and Howell (2014) had a higher percentage of chronic offenders who were also violent while Baglivio et al. (2014) had a much lower percentage of violent offenders who were also chronic.

Table 1.

Comparison of Studies Using SVC Categories.

Source of records	Snyder (1998)	Loeber, Van Kammen, & Fletcher (1996)	Kempf-Leonard et al. (2001)	Baglivio et al. (2014)
Source of records	Court (%)	Police (%)	Police (%)	Court (%)
Sample size of offenders	151,209	1,517	11,192	363,617
Chronic offenders who were also violent offenders	29	36	58	59
Violent offenders who were also chronic offenders	53	45	51	31
Serious offenders who were also chronic offenders	35	35	38	26
Offenders who were not serious, violent, or chronic offenders	64		49	44

Note. SVC = Serious, violent, and chronic.

There were methodological differences across the four studies that may have produced these differences. For instance, Loeber et al. (1996) only included males in their sample, whereas the other three consisted of both males and females (although one can assume that Loeber et al.’s percentages should have been higher since male offending rates are higher than females). In using police reports, Kempf-Leonard et al. (2001) used police rap sheets and more informal information reports (such as police field contacts and investigative reports), which likely increased the number of overall police contacts for chronic offenders. They also used different criteria for categorizing serious offenders (e.g., violent offenders or juveniles committing felony property offenses compared to Snyder’s categorization using specific offenses). Finally, Baglivio et al.’s (2014) serious definition consisted of any felony committed by youth and their violent definition used personal felonies and a weapons offense, where Snyder (1998) defined a violent offender as one who was referred to court for “murder and nonnegligent manslaughter, kidnapping, violent sexual assault, robbery, and aggravated assault” (p. 429).

Differences Between SVC and Non-SVC Juvenile Offenders

Loeber and Farrington (1998) identified common themes across studies involving SVC juvenile offenders that have been consistently supported by subsequent research. First, they concluded that serious and violent juveniles are distinct from other offender groups and have an array of personal, behavioral, school-related, and mental health problems (Howell, 2003b; Howell et al., 2014; Johansson & Kempf-Leonard, 2009; Kennedy, Burnett, & Edmonds, 2011; Krohn et al., 2014; Zara & Farrington, 2013). Second, their age of onset for antisocial behaviors and delinquency is significantly earlier, and their offending careers last much longer (DeLisi, 2006; DeLisi et al., 2013; Howell et al., 2014; McCluskey et al., 2006; Piquero et al., 2004). Third, the majority of juvenile crime is committed by chronic juvenile offenders with the majority of chronic offenders being serious and violent as well (Baglivio et al., 2014; Kempf-Leonard et al., 2001). Fourth, the highest rates of serious and violent offending are among African American youth (Baglivio et al., 2015; Kempf-Leonard et al., 2001; Trulson, Caudill, Haerle, & DeLisi, 2012; Vaughn et al., 2014).

Research more specific to the defined SVC categories has also found differences across these categories. For instance, Baglivio et al. (2014) found that males and African Americans were much more likely to be SVC offenders than females and other racial/ethnic groups. This study also compared risk and protective factors to SVC groupings and found that SVC youth are clearly more risky across all of their assessment domains (overall risk, social history and criminal history, school, leisure time use, employment, relationships, family and living arrangement, substance use, mental health, attitude/behavior, and aggression). The largest risk factor differences between their SVC and non-SVC groupings were found for school history, current aggression, history of relationships, current attitude/behavior, and mental health history. Additionally, they found that protective scores were higher for the non-SVC group for school history, current school, current aggression, current attitude/behavior, and current leisure time use.

More recent research has explored the relationship of trauma has on SVC youth. Fox, Perez, Cass, Baglivio, and Epps (2015) compared the type and frequency of adverse childhood experiences (ACEs) between SVC youth to “one and done” youth (youth who had only one adjudicated case in juvenile court). They found large significant differences between these two groups with SVC offenders having double the ACEs as one and done youth. These findings have been replicated in other research as well (Baglivio et al., 2015).

Predicting Adult Offending

Most youth who exhibit antisocial behaviors at a young age do not commit crime as adults. However, youth who do display antisocial behavior at a young age are likely to continue these behaviors into adolescence and adulthood (Moffitt, 1993; Piquero, Hawkins, & Kazemian, 2012), but the majority of children and adolescents who are antisocial do not continue this behavior into adulthood (Laub & Sampson, 2001; Monahan, Steinberg, Cauffman, & Mulvey, 2013; Sampson & Laub, 1993). Studies of continuity and desistance generally find that 30–60% of juvenile offenders continue to commit crime into adulthood (see Piquero et al., 2012) with serious and violent juvenile offenders having higher rates of adult criminality than other offender types. The Kempf-Leonard et al.’s (2001) sample of 6,287 delinquents had a 32% adult criminality rate with the rate being much higher for serious juvenile offenders (48%), violent offenders (53%), and chronic juvenile offenders (58.5%). These rates were highest for serious and chronic (62.5%) and violent and chronic (63%) juvenile offenders. Similarly, 35% of the Baglivio et al. (2014) sample were arrested as adults or rereferred to juvenile court. SVC juvenile offenders were twice as likely as non-SVC youth and 2.77 times more likely to be arrested as juvenile offenders who were not serious, violent, or chronic offenders.

Longitudinal research has found that juvenile offenders who continue their criminal behavior into adulthood are more likely to have decreased psychosocial maturity (Monahan et al., 2013); negative changes to their social context and social capital (e.g., school, family, friends; Giordano, Cernkovich, & Holland, 2003; Mulvey et al., 2004), biosocial differences (DeLisi & Piquero, 2011; Nagin et al., 1995), and more psychopathy characteristics (Corrado, McCuish, Hart, & DeLisi, 2015; McCuish, Corrado, Hart, & DeLisi, 2015; Vaughn & Howard, 2005). Despite the abundance of this research, there are few studies that have used juvenile risk/need assessments for SVC youth in predicting adult criminality. One such study, conducted by Baglivio et al. (2014), used Florida juvenile court records and adult arrests for 34,497 juvenile offenders who were assessed with the Positive Achievement Change Tool. They categorized youth as SVC, non-SVC, serious, violent, or chronic offenders. When predicting an adult arrest or new referral to juvenile court within 1 year of youths’ completing juvenile court placements, they found that SVC recidivists were more likely to be males, minorities, have more extensive criminal histories, have higher substance abuse needs, and better social skills. The non-SVC study group recidivists had the same predictors as the SVC group but also had higher risk scores for school and relationships. The recidivism predictors in their study group consisting of juvenile offenders who were either serious or violent or chronic were similar to the non-SVC study group. One difference between these groups was that the serious or violent or chronic groups had lower mental health risk scores than the non-SVC group.

Focus of the Current Study

The current study builds upon prior SVC research in using assessments combined with juvenile career categories. First, we compare risk assessment subscale scores across the risk categories. Based on prior research, we expect to find differences in risk and protective assessment scores across the career categories, especially for SVC youth. These SVC youth are expected to have higher scores for more of the assessment subscales than the other career types. Second, we will test the assessment subscales as predictors of adult criminality within each of the career types. It is hypothesized that SVC youth will have higher rates of adult criminality and that their recidivism will be explained by more of the assessment subscales than non-SVC youth.

Data and Method

Connecticut has a unified juvenile court system with a centralized administration and database for juvenile court matters and probation services. Information for all Connecticut youth who are referred to juvenile court is housed in the Connecticut Judicial Branch’s Management Information System (CMIS). The CMIS contains juvenile court data relevant to specific cases and individual charges (e.g., charges and dispositions), pre- and post-adjudicatory assessments (the widely used Juvenile Assessment Generic [JAG] is discussed below), and juvenile probation officers’ reports and case notes. Adult recidivism data were obtained from the Connecticut Department of Public Safety and Emergency Services’ Criminal History database. These electronic data facilitated the application of Snyder’s (1998) typology to Connecticut juvenile court data. While court records tend to underreport the actual amount of crime an offender commits when compared to self-reported offenses (Farrington, Piquero, & Jennings, 2013), they still provide valuable information in the study of delinquency careers. Piquero, Schubert, and Brame’s (2014) work lends support to prior research by finding a moderate to strong agreement between self-reported and actual arrests with no subgroup differences across race/ethnicity and gender.

Using the CMIS, we first identified all youth with a delinquency referral to the juvenile court who aged out of the court’s jurisdiction between 2005 and 2009 at 16 years of age (n = 30,236). (Connecticut Statutes defined 16- and 17-year-old offenders as adults until 2010.) Once a master list was created, we collected demographic, official juvenile court records, and risk assessment data from CMIS. These records were then matched to youths’ subsequent adult arrests and court dispositions. The juvenile court data consisted of demographics (age, gender, race/ethnicity), current offense (number of charges, most serious offense), age of first juvenile court referral, and risk assessment scores. Criminal history data were comprised of adult arrests after youths’ 16th birthday.

Creating Juvenile Career Types

Having established the study population, we next determined the most serious statute for every juvenile offender. All dockets for which the most serious statute was a status offense, violation of probation or court order, infraction, or local ordinance violation were excluded. The remaining dockets were then categorized based on the most serious charge as violent, serious, and nonserious. Consistent with Snyder (1998), violent and serious offenses were identified based on felony status, while the nonserious (violent and nonviolent) offenses were of misdemeanor status.

Violent offenses included murder, nonnegligent manslaughter, kidnapping, arson, violent sexual assault, robbery, and aggravated assault.

Serious nonviolent offenses: burglary, serious larceny, motor vehicle theft, weapons offenses, and drug trafficking.

Nonserious delinquent offenses: simple assault, possession of controlled substances, disorderly conduct, vandalism, nonviolent sex offenses, minor larceny, liquor law offenses, and all other delinquent offenses.

Aggregating total dockets (violent, serious, and nonserious) by unique juvenile provided the chronic measure from Snyder’s (1998) SVC typology. Each aged-out juvenile’s court history was sorted into a career type based on the following rules: violent, one or more violent offense; serious, one or more serious offense; and chronic, four or more court referrals yielding eight distinct careers types (serious/violent/chronic, serious/chronic, violent/chronic, serious/violent, serious, violent, chronic, and not serious/violent/chronic).

The Connecticut population of juvenile offenders closely mirrors Snyder’s findings when his career-type definitions are applied (Table 2). A very small percentage of all juvenile offenders (1.9–3.3%) were defined as SVC youth while the majority of juveniles were not serious, violent, or chronic offenders (67.0–63.9%). Serious offenders comprised the largest group of any of the SVC categories (15.2% and 17.7%, respectively). The two offender populations were also very similar when comparing chronic offenders who were also violent (21–29%), violent offenders who were also chronic (48–53%), and serious offenders who were also chronic (34– 35%).

Table 2.

Comparison Between Connecticut Juvenile Offender Career Types and Snyder (1998).

Career Type	Connecticut	Snyder (1998)
Career Type	(n = 30,236), %	(n = 151,209), %
Serious, violent, and chronic	1.9	3.3
Serious and violent	0.4	0.7
Violent and chronic	1.1	0.9
Serious and chronic	6.0	7.9
Violent	2.9	3.2
Serious	15.2	17.7
Chronic	5.4	2.5
Not serious, violent, or chronic	67.0	63.9
Chronic offenders who were also violent	21	29
Violent offenders who were also chronic	48	53
Serious offenders who were also chronic	34	35

JAG

The next step was to match the juveniles to their risk assessments. The Connecticut Judicial Branch Juvenile Probation Services uses the JAG as the main risk/need assessment for court involved juveniles (Bogue, Vanderbilt, & Ehret, 2005). Juvenile court referred youth are administered the JAG prior to the court’s disposition by a juvenile probation officer. For youth placed on probation, the JAG is readministered 6 months into their supervision and at discharge from probation.

A third-generation actuarial tool, the JAG, comprises 47 items and measures risk and protective factors in areas of criminal history, substance abuse/risk taking, distress/family, peers/stake in conformity, and personal values. The JAG generates five risk factor scale scores sensitive to static and dynamic qualities influencing criminal recidivism and rankings that highlight target treatment areas. The static items score 0 or 1 while the dynamic items score 0–3. Each item is scored for a particular frame (i.e., ever, last 6 months, last year). The dynamic items score both risk and protective with scores of 0 or 1 indicating risk and 2 or 3 no risk. Scores of 1, 2, or 3 indicate varying levels of protective influence. So, a dynamic item score of 1 contributes to both the risk and the protective subscale scores. A 0 contributes 1 risk point but nothing to the protective scale, while scores of 2 or 3 add 0 to the risk subscale and either 2 or 3 to the protective subscale.

Criminal history

Intended to capture the nature, attachment to and versatility of a youth’s delinquent history, this subscale is made up of 5 static items. Criminal history examines the nature of the current offense (did it occur during existing supervision and did it involve theft) and past criminal history such as the number of prior probation violations, prior admissions/adjudications, and the number of prior incarcerations (both preadjudicatory detention and post-adjudicatory residential placement).

Substance abuse/risk taking

This subscale measures use of substances and the degree of willingness to engage in risky behaviors related to substances and is comprised of 5 static and 4 dynamic items. The static items include (1) use of tobacco, (2) use of mood altering substances other than alcohol, (3) substance use interfering with daily functioning, (4) crimes due to poor judgment under the influence, and (5) crimes to support/obtain drugs/alcohol. The dynamic items gauge the regularity and/or degree of use (1) regular use of alcohol, (2) alcohol abuse, (3) regular use of marijuana, and (4) regular use of other substances.

Distress/family

The nature of the family and family-related distress is measured in this subscale by 3 dynamic items and 7 static items. The dynamic items center on parental relationship and the style of family functioning: (1) relations with (step)mother, (2) relations with (step)father, and (3) chaotic family. The static items include (1) has been in out-of-home placement, (2) psychiatric history (father, mother, or sibling), (3) has been the victim of physical abuse by person in a position of trust, (4) has been the victim of sexual abuse, (5) lives away from parent/guardians, (6) youth has a history of psychological intervention, and (7) psychological intervention recommended.

Peers/stake in conformity

This scale measures peer influences generally and in relation to youths’ attachment to and degree of conformity or nonconformity in various contexts. There are 9 dynamic and 4 static items. Dynamic items about school/work explore (1) school achievement (grades, effort, extracurricular activities), (2) classroom behavior in the prior year, (3) relations with student peers/coworkers, (4) relations teachers/supervisors, concerning use of time, (5) could make better use of time (degree of structured vs. unstructured), (6) degree of parental supervision and structure, (7) absence of reasonable future plans and toward delinquency, (8) allegiance to criminal peers, and (9) supportive of delinquency. Static items focus on antisocial influences and attitudes: (1) no prosocial interests, (2) has very few prosocial acquaintances, (3) has very few prosocial friends, and (4) poor attitude toward sentence/disposition.

Personal values

Traits and characteristics associated with criminality are gauged in this scale through 3 dynamic and 7 static items. Dynamic items capture the degree to which a youth demonstrates (1) a dominating attitude, (2) no empathy, and (3) manipulative attitudes or behavior. The static items include (1) history of sexual offending, (2) sexual offender treatment, (3) blaming others (for failures or poor behavior), (4) narcissism, (5) need for structure and control, (6) anger management, and (7) psychological/emotional impairments.

Arrest Data

Adult arrest data were collected for the study group. These data were obtained from the Connecticut Criminal History database through a matching process that employs a series of algorithms using name, gender, date of birth, social security number, and a Connecticut State Police identification number. Adult arrest data were collected for 5 years after youth had aged out of the juvenile justice system.

Assessment Sample

The last available assessment for each aged-out juvenile was used for this study. The match of assessments to juveniles was constrained by availability of JAG scores in the study population. The JAG is only administered to youth who are being placed on probation. In many instances, referrals to juvenile court are processed out through case dismissals or through the discretion of the prosecutor (e.g., suspended prosecution, nolo contendre) and never reach the point of requiring a risk assessment. Additional youth are screened out of the assessment process through the use of a short-form assessment. The short-form risk assessments function as a low-risk screen. If the short-form assessment determines that a client is low risk for reoffending, the assessment process ends; however, if the short form indicates the client is something other than low risk, then a JAG is administered.

Our assessment sample consists of 9,859 juveniles with a delinquency referral to the juvenile court who aged out of juvenile jurisdiction between 2005 and 2009 at 16 years of age (Table 3). The majority of juveniles were males (70.4%). White offenders comprised the highest percentage within race/ethnicity (39.8%) followed by African American youth (34.1%). For age of first juvenile court referral, the highest percentages were 13 years old (25.1%), 14 years old (24.5%), and 12 years old (17.6%).

Table 3.

Assessment Group Demographics.

Demographic		Frequency	Percentage
Gender	Males	6,941	70.4
	Females	2,918	29.6
Race/ethnicity	White	3,868	39.8
	African American	3,309	34.1
	Hispanic	2,351	24.2
	Other	181	1.9
	Missing	150
Age of first referral (years)	Under 11	851	8.6
	11	991	10.1
	12	1,735	17.6
	13	2,485	25.2
	14	2,412	24.5
	15	1,385	14.0

Analysis

Two sets of analyses were performed. First, we looked at differences in risk and protective score subscales across the eight nominal juvenile offender career types using multivariate analysis of variance (MANOVA) tests. Second, several negative binary logistic regression models were constructed for each juvenile offender career type to assess the risk and protective subscale effects on juveniles’ rearrest as adults up to 2 years after they aged out of the juvenile justice system. The logistic regression models also included gender, race/ethnicity, and age of first juvenile referral.

Results

Risk and Protective Score Differences Across Juvenile Career Types

The subscale scores were standardized prior to conducting the MANOVA since they were scored on different metrics. Tables 4 and 5 present the mean z-scores for each risk subscale across the juvenile offender career types. The MANOVA analyses for risk and protective scores produced distinct results across the eight career types.

Table 4.

MANOVA Result of Career Types by JAG Risk Subscales.

Career Type	n	Criminal History	Substance Abuse	Family	Peers	Personal Values
CSV	492	0.996	0.229	0.282	0.745	0.496
Chronic/serious	1,536	0.730	0.245^a	0.122^a	0.375	0.112
Chronic/violent	280	0.750^b	−0.018^a	0.361^a,b	0.414^b	0.346^a,b
Chronic	1,224	0.361	−0.030^c	0.316^a,c	0.176	0.226^b,d
Serious/violent	95	0.112^d	−0.192^c,d	−0.109^a,b	−0.027^d	0.125^a,b,d
Serious	1,993	−0.290	−0.031^c,d,e	−0.325^e	−0.303^f	−0.296
Violent	419	−0.286^f	−0.284^c,e	−0.241^ef	−0.298^f	−0.071^b,e
None	3,820	−0.413^g	−0.065^c,d,e,f	−0.014^e	−0.142^f,g	0.045^e,g
F		472.39*	25.34*	68.36*	136.66*	63.48*
Partial η²		.25	.02	.05	.09	.04

Note. MANOVA = multivariate analysis of variance; JAG = Juvenile Assessment Generic; CSV = chronic, serious, and violent.

^aNot statistically significantly different from CSV at p < .05. ^bNot statistically significantly different from chronic/serious at p < .05. ^cNot statistically significantly different from chronic/violent at p < .05. ^dNot statistically significantly different from chronic at p < .05. ^eNot statistically significantly different from serious/violent at p < .05. ^fNot statistically significantly different from serious at p < .05. ^gNot statistically significantly different from violent at p < .05.

*Pillai’s trace F = 93.643, p < .0001, df = 35, partial η² = .062.

Table 5.

MANOVA Result of Career Types by JAG Protective Subscales.

Career Type	n	Substance Abuse	Family	Peers	Personal Values
CSV	492	−0.166	−0.361	−0.736	−0.445
Chronic/serious	1,536	−0.183^a	−0.159	−0.347	−0.145
Chronic/violent	280	−0.025^a,b	−0.325^a,b	−0.356^b	−0.236^a,b
Chronic	1,224	0.029^c	−0.279^a,c	−0.201^c	−0.187^b,c
Serious/violent	95	0.161^c,d	0.225	0.094^d	0.041^b,c,d
Serious	1,993	0.050^c,d,e	0.352^e	0.303^e	0.228^e
Violent	419	0.315^e	0.319^e,f	0.312^e,f	0.194^e,f
None	3,820	0.031^c,d,e,f	0.000^e	0.130^e	0.051^e,g
F		14.65*	78.86*	128.47*	46.89*
Partial η²		.01	.05	.08	.03

Note. MANOVA = multivariate analysis of variance; JAG = Juvenile Assessment Generic; CSV = chronic, serious, and violent.

^aNot statistically significantly different from CSV at p < .05. ^bNot statistically significantly different from chronic/serious at p < .05. ^cNot statistically significantly different from chronic/violent at p < .05. ^dNot statistically significantly different from chronic at p < .05. ^eNot statistically significantly different from serious/violent at p< .05. ^fNot statistically significantly different from serious at p < .05. ^gNot statistically significantly different from violent at p < .05.

*Pillai’s trace F = 40.017, p < .0001, df = 28, partial η² = .028.

The chronic, serious, and violent (CSV) career type had much higher risk scores and the lowest protective scores for almost every subscale (with the exception of substance abuse and family risk). This career type had especially high subscale scores for peers and personal values, meaning that CSV youth had an extremely low stake in conformity, a very high attachment to delinquent peers, and exhibited a high amount of antisocial personality characteristics. These youth also had high levels of family dysfunction and substance abuse. Chronic and serious youthful offenders had similar risk and protective scores as the CSV group for substance abuse and family risk but had lower risk levels for peers and personal values and had higher protective scores for family, peers, and personal values than CSV youth. Chronic and violent youth did not have the same amount of substance abuse risk as chronic and serious offenders but had similar high-risk and protective scores for family, peers, and personal values as CSV and serious/chronic offenders. Chronic offenders had less criminal history than the other three chronic offender groups as well as lower substance abuse and peer risk scores. One interesting finding here was that chronic youth had risk and protective scores for family dysfunction that mirror CSV youth.

There is a clear distinction in risk and protective scores for chronic and nonchronic offender groups. Most of the risk scores were significantly lower for the nonchronic offender groups than any of the chronic career types while the opposite was true for protective scores. There were also fewer significant differences within the nonchronic career types than there were within the chronic career types. For instance, there were no significant differences between the serious career type, the violent career type, and no career type for peer risk scores. There were more mean differences within the nonchronic career groups for the protective scores but not as many as the chronic career groups. There were significant differences between the violent career and the serious career types for substance abuse risk while the other subscales were not different. Serious offenders had higher substance risk scores (and lower protective scores) and lower personal values risk scores than violent offenders.

The partial η² is a measure of effect size commonly used in MANOVA tests showing how much of the total variance is produced by individual mean variances (Cohen, 1988). Based on Cohen’s (1988) effect size interpretations, the .25 partial η² for the criminal history risk score demonstrated a large effect (.10 and higher); the .09 partial η² for peers demonstrated a medium effect (.06–.09); and small effects (partial η²s below .5) for family (.05), personal values (.04), and substance abuse (.02). Criminal history and peers had the most variation across career types, while substance abuse and personal values had the least. CSV youth had the most extensive criminal history and the most serious risk scores for peers. Chronic/serious and chronic/violent youth had similar criminal histories and similar peer scores. On the other hand, the fewest mean differences were seen in substance abuse scores. SVC and chronic/serious youth had the highest risk scores, while there was little significant difference among the remaining groups. The same pattern was present for the protective scores; the most significant differences across groups was observed for peers and the least number for substance abuse scores.

Predicting Adult Arrests

The next analyses used negative binomial regression models to assess which risk and protective subscales were predictive of an adult arrest for the eight individual offending career types. The dependent variable, new adult arrest, was defined as a youth’s first criminal arrest within 2 years after turning 16 years old. Arrests for violations and traffic offenses were excluded from the analysis. The CSV career type had the highest 2-year arrest rate (74%) followed by chronic/serious (73%), chronic/violent (71%), chronic (65%), serious/violent (58%), serious (47%), none (45%), and violent (44%).

A logistic regression model was created for each juvenile career type, first using the risk score subscales as predictors (Table 6) followed by the protective score subscales (Table 7). Each table presents the odds ratios and 95% confidence intervals for the odds ratios for the independent variables. For the risk subscales, the regression model for the entire sample shows that four of the five risk subscales positively predicted adult arrests with criminal history and peers exhibiting the strongest relationship among the risk scores. Gender (males) and race (non-White) were the best overall predictors in this model. In fact, males were 1.74 times more likely to be arrested than females and non-White youth were 1.47 times more likely to be arrested than White youth. For CSV youth, the only significant predictor of arrest was race, with non-White CSV youth being 1.85 times more likely to be arrested than White youth. Interestingly, the risk subscales were weak predictors of arrests for all of the chronic career types while being significant predictors of arrest across the nonchronic career types. This pattern was also observed for the protective score subscales. None of the risk or protective subscales were statistically significant predictors of arrest for the CSV youth while only criminal history (chronic/violent and chronic) and peers (chronic/serious) were predictive for the other chronic career types.

Table 6.

Logistic Regression Risk Scores Odds Ratios and 95% Confidence Intervals Predicting Adult Arrest by Juvenile Offender Career Type.

Independent Variable	Entire Sample	CSV	Chronic/Serious	Chronic/Violent	Chronic	Serious/Violent	Serious	Violent	None
Males	1.74** [1.58, 1.91]	1.23 [0.59, 2.60]	1.50* [1.08, 2.10]	2.00* [1.11, 3.61]	1.88* [1.46, 2.41]	1.31 [0.24, 7.34]	1.93** [1.48, 2.51]	2.48** [1.40, 4.39]	1.64** [1.43, 1.89]
Race: non-White	1.47** [1.34, 1.61]	1.85* [1.05, 3.25]	1.36* [1.05, 1.75]	1.13 [0.57, 2.27]	1.29 [0.99, 1.69]	1.16 [0.41, 3.25]	1.61** [1.32, 1.96]	1.95** [1.22, 3.10]	1.30** [1.13, 1.49]
Age at first referral	1.05** [1.02, 1.08]	1.10 [0.98, 1.25]	1.01 [0.94, 1.09]	1.01 [0.84, 1.21]	1.05 [0.97, 1.14]	1.01 [0.78, 1.30]	1.04 [0.97, 1.11]	1.01 [0.89, 1.17]	1.00 [0.96, 1.05]
JAG: criminal	1.19** [1.16, 1.23]	1.01 [0.87, 1.18]	1.07 [0.98, 1.16]	1.26* [1.03, 1.54]	1.13* [1.03, 1.24]	1.08 [0.82, 1.43]	1.13** [1.05, 1.23]	1.20* [1.00, 1.43]	1.12** [1.06, 1.19]
JAG: substance abuse	1.07** [1.04, 1.10]	1.00 [0.89, 1.13]	1.02 [0.95, 1.09]	1.03 [0.85, 1.25]	1.01 [0.93, 1.10]	1.21 [0.89, 1.67]	1.14* [1.07, 1.21]	1.07 [0.91, 1.27]	1.11** [1.06, 1.16]
JAG: family	1.02* [1.01, 1.05]	1.09 [0.98, 1.22]	1.06 [0.99, 1.12]	1.07 [0.94, 1.22]	1.00 [0.95, 1.07]	1.10 [0.84, 1.44]	1.01 [0.96, 1.07]	1.04 [0.93, 1.16]	1.00 [0.97, 1.04]
JAG: peers	1.07** [1.05, 1.08]	1.04 [0.97, 1.11]	1.05* [1.01, 1.09]	1.07 [0.97, 1.17]	1.00 [0.97, 1.05]	1.01 [0.88, 1.15]	1.13** [1.09, 1.18]	1.13** [1.04, 1.22]	1.05** [1.02, 1.07]
JAG: personal	1.01 [0.98, 1.03]	1.01 [0.89, 1.14]	1.01 [0.94, 1.08]	1.01 [0.85, 1.19]	1.05 [0.98, 1.13]	1.10 [0.87, 1.40]	1.07* [1.01, 1.13]	1.09 [0.97, 1.23]	1.05* [1.01, 1.10]
Constant	2.98	1.92	1.67	5.84	5.10	1.08	5.21	14.08*	5.20
Cox and Snell pseudo-R ²	.09	.04	.02	.06	.03	.03	.09	.11	.05
Model χ²	859.30**	17.58*	37.02**	16.88*	40.99**	2.60	190.14**	47.47**	175.91**
n	9,709	490	1,533	279	1,224	95	1,965	417	3,713

Note. Italics indicate the variable was inversely related to the outcome. CSV = chronic, serious, and violent; JAG = Juvenile Assessment Generic.

*p < .05. **p < .01.

Table 7.

Logistic Regression Protective Scores Odds Ratios and 95% Confidence Intervals Predicting Adult Arrest by Juvenile Offender Career Type.

Independent Variable	Entire Sample	CSV	Chronic/Serious	Chronic/Violent	Chronic	Serious/Violent	Serious	Violent	None
Males	1.79** [1.63, 1.96]	1.18 [0.56, 2.49]	1.47* [1.05, 2.4]	1.93* [1.08, 3.48]	1.84** [1.44, 2.35]	1.68 [0.30, 9.36]	1.87** [1.45, 2.43]	2.53** [1.43, 4.48]	1.64** [1.43, 1.88]
Race: non-White	1.52** [1.39, 1.66]	1.80* [1.04, 3.12]	1.33* [1.04, 1.70]	1.13 [0.57, 2.24]	1.23 [0.95, 1.60]	1.13 [0.41, 3.13]	1.67** [1.38, 2.02]	2.17** [1.39, 3.39]	1.27** [1.11, 1.45]
Age at first referral	1.08** [1.05, 1.11]	1.11 [0.99, 1.25]	1.02 [0.95, 1.10]	1.04 [0.86, 1.24]	1.05 [0.97, 1.14]	1.00 [0.78, 1.29]	1.03 [0.97, 1.10]	1.03 [0.90, 1.19]	1.01 [0.96, 1.06]
JAG: substance abuse	1.06** [1.04, 1.08]	1.01 [0.91, 1.11]	1.0 [0.97, 1.08]	1.01 [0.88, 1.16]	1.03 [0.97, 1.10]	1.01 [0.78, 1.31]	1.09** [1.04, 1.14]	1.14* [1.02, 1.28]	1.08** [1.04, 1.11]
JAG: family	1.05** [1.02, 1.08]	1.09 [0.96, 1.25]	1.04 [0.97, 1.12]	1.18 [1.00, 1.40]	1.01 [0.94, 1.09]	1.05 [0.79, 1.39]	1.01 [0.95, 1.07]	1.10 [0.97, 1.25]	1.05* [1.01, 1.09]
JAG: peers	1.06** [1.05, 1.07]	1.02 [0.97, 1.08]	1.05* [1.02, 1.09]	1.08 [1.00, 1.17]	1.01 [0.98, 1.05]	1.02 [0.90, 1.16]	1.09** [1.06, 1.12]	1.08* [1.01, 1.15]	1.04** [1.02, 1.06]
JAG: personal	1.01 [0.98, 1.04]	1.02 [0.90, 1.17]	1.01 [0.93, 1.10]	1.07 [0.88, 1.30]	1.06 [0.98, 1.15]	1.11 [0.78, 1.58]	1.04 [0.97, 1.11]	1.05 [0.91, 1.23]	1.02 [0.97, 1.07]
Constant	5.75**	8.67	4.37*	1.89	2.41	1.036	3.00	2.09	1.91
Cox and Snell pseudo-R ²	.07	.04	.03	.06	.03	.01	.08	.12	.04
Model χ²	705.61**	17.53*	38.13**	17.07*	30.45**	0.95	161.33**	51.07**	144.35**
N	9,709	490	1,533	279	1,217	95	1,965	417	3,713

Note: Italics indicate the variable was inversely related to the outcome. CSV = chronic, serious, and violent; JAG = Juvenile Assessment Generic.

*p < .05. **p < .01.

The risk and protective subscales were predictive of adult arrests for the nonchronic career types and showed differences across the career types. Similar to the chronic career types, gender and race were the most predictive independent variables. Of the risk subscales, peer influences had the most effect on adult arrests for serious and for violent youth but not for youth with no defined career type (none). As seen in the MANOVA, substance abuse risk is present for serious youth but not for violent youth. All of the risk subscales except for family had small effects on adult arrests for the none career type. The protective subscales mostly mirrored these findings: Serious youth who were arrested had low substance abuse and peer protective scores while arrested violent youth had lower family protective scores.

Discussion

The purpose of this study was to replicate Snyder’s SVC career types and to explore risk and protective score differences within these career types. Analyzing juvenile court records and assessment data with a sample of 9,859 youth, our study yielded three distinct findings. First, our categorization of juvenile career types mirrored Snyder’s (1998) work. Second, the MANOVA analysis comparing risk and protective factors across and within juvenile career types identified distinct patterns, especially between chronic and nonchronic offenders. Third, when predicting adult arrests 2 years after youths’ juvenile commitment ended, the juvenile risk and protective assessment subscales were not predictive of adult arrests for chronic offenders. They were, however, predictive for nonchronic juvenile offenders. These findings are discussed in more detail below.

Snyder’s Career Types

One of the most consistent findings in juvenile justice research is that a very small percentage of youth commit the most serious and violent offenses (Barnes, 2013; Baglivio et al., 2014; DeLisi & Piquero, 2011; Hamparian et al., 1978; Moffitt, 1993; Nagin & Farrington, 1992; Nagin et al., 1995; Snyder, 1998; Vaughn et al., 2011, 2014; Wolfgang et al., 1972). Our findings add to this long history of research and reaffirm Snyder’s (1998) categorization of juvenile offender career types using official delinquency records. Although retrospective, Snyder’s career types are valuable by providing a snapshot of the types of offenders in the juvenile justice system. We find it remarkable that his categorization of juvenile offending career types remains extremely accurate nearly 20 years after his initial research (his data were collected in 1990) and in a region of the United States that is very different (ideologically, social economically, and demographically) from Maricopa County, AZ. Additionally, despite significant structural changes and philosophy in the juvenile justice system over the past 20 years in adjudication and sentencing, alternatives to detention and youth prisons, confidentiality and expungement, and disproportionate minority contract (Brown, 2015; National Juvenile Justice Network, 2012), the types and proportion of offenders in the juvenile justice system have remained the same.

Risk and Protective Factors Across Career Types

Risk assessments commonly assign a level of recidivism and/or compliance risk based on a total score that is used to guide court supervision and treatment referrals. Overlaying the career types on the risk scores demonstrated the limits of this approach. For instance, CSV youth had consistently high-risk and low protective subscales scores suggesting they have a myriad of problems that all contribute to being CSV offenders. Chronic/serious youth had the highest substance abuse risk scores and the lowest protective scores but did not score as high for family, peers, and personal values as chronic/serious/violent and chronic/violent youth. These differences suggest the frequency and types of offenses committed by chronic/serious youth are driven primarily by substance abuse and are very different from chronic/serious/violent and chronic/violent youth.

One clear finding was the differences between chronic and nonchronic offenders. All types of chronic offenders had higher risk and lower protective scores than nonchronic offenders, suggesting that juveniles with repeated system involvement should be treated differently than first or second time offenders, regardless of the instant offense. For nonchronic offenders, the most prevalent area of risk was family dysfunction. They had less criminal history, lower substance abuse risk, had some risk with peers, and some poor personal values.

These findings support discussions by Skeem, Scott, and Mulvey (2014) and Howell, Lipsey, Wilson, and Howell (2014) on how to significantly reduce the continued court involvement of high-risk juveniles with evidence-based juvenile justice systems. One of Howell et al.’s (2014) core principles of a comprehensive juvenile justice strategy is to identify a small group of SVC offenders. Skeem et al. (2014) conceptualized high risk as a “nondistinct group of youth” that have a multitude and combination of risk factors that greatly increases their likelihood of continuous and serious offending. Based on our findings and these discussions, an efficient and practical approach to dealing with juveniles most likely to recidivate would be to first identify CSV youth using Snyder’s typology and then conduct more in-depth risk assessments to best create individualized case plans and treatment.

Adult Arrests

The logistic regression analysis of adult arrests produced similar findings to the MANOVA analysis, in that, chronic offenders were clearly different than nonchronic, regardless of career type. When regressing risk and protective assessment scores on adult arrest by each juvenile career category, we found that neither the risk nor protective score subscales differentiated arrests for chronic career types. One explanation for this finding is the limited variability in the percentage of youth who were rearrested. The second explanation is that the very high-risk scores limited the ability of the subscales to predict arrest. For instance, the SVC group had little variation in their risk scores since youth were very high risk across multiple subscales.

The lack of criminogenic predictors of recidivism among serious juvenile offenders has become a somewhat consistent finding. Baglivio et al. (2014) also had limited success in identifying predictors of reoffending for SVC youth. They did not address why few factors predicted reoffending for SVC youth but recommended that future research explores combinations of risk and protective factors across SVC subgroups. Other research involving serious juvenile offenders, but not specifically SVC youth, have similarly found little relationship between criminogenic factors and rearrest (Corrado et al., 2015; McCuish et al., 2015; Trulson et al., 2012). These studies did identify other factors related to recidivism such as gang involvement (Trulson et al., 2012) and psychopathy (Corrado et al., 2015; McCuish et al., 2015).

We found that high-risk youth have multiple risk factors that preclude the influence of any one specific factor. Consistent with prior research, the preponderance of risk factors increases youths’ propensity for continued criminal behavior (Borduin & Ronis, 2012; Hawkins et al., 1998; Huizinga & Jakob-Chien, 1998; Johansson & Kempf-Leonard, 2009; Krohn et al., 2014; Loeber & Farrington, 2000; Valois et al., 2002; Zara & Farrington, 2013). Because of this, Kempf-Leonard et al. (2001) recommended that risk assessments be developed that can distinguish between delinquent career types.

The subscales did contribute to the explanation of adult arrests for nonchronic offenders. For nonchronic/serious juveniles, those who were arrested within 2 years following their juvenile commitment had high peer and substance abuse risk scores and low personal value risk scores. These juveniles were not invested in delinquent or criminal lifestyles nor were they antisocial. Violent juvenile offenders had high peer risk scores, suggesting their limited violent behavior was attributed to their relationships with a high-risk peer group. Youth who did not fit into any of the juvenile career types (i.e., none) but were rearrested within 2 years had higher risk scores for substance abuse, peers, and personal values than similar youth who were not arrested. These youth appeared to be typical adolescents who continued their risky behaviors after they turned 16 years old. This group had minor drug use, slightly negative peer groups, and negative personal values.

Taken together, our findings support Kempf-Leonard et al.’s (2001) recommendation. General risk assessments do discriminate for lower risk offenders and, by themselves, can be useful in identifying areas of risk/need in developing supervision and treatment strategies. However, high-risk youth require more advanced screening instruments/assessments to more successfully treat them (Loeber & Ahonen, 2014).

Limitations

A major limitation to this study was the reliance on official court records and assessments. The assessment used by the Connecticut Judicial Branch only contained traditional measures of risk and protection (criminal history, substance abuse, family, peers, and personal values). While the literature supports the use of these measures when predicting the recidivism of less serious offenders, they have been found to be weak recidivism predictors for serious and violent juvenile offenders. Future research of serious and violent juvenile offenders should attempt to incorporate other risk factors that have recently been found to be more predictive of recidivism for this offender population. These factors are gang membership/involvement (Johansson & Kempf-Leonard, 2009; Thornberry, Krohn, Lizotte, & Chard-Wierschem, 1993; Trulson et al., 2012), psychopathy (Corrado et al., 2015; McCuish et al., 2015; Vaughn et al., 2011; Vaughn & Howard, 2005), and childhood trauma (Baglivio et al., 2015; Fox et al., 2015).

Conclusion

Snyder’s (1998) offending career typology has only been recently reapplied to juvenile justice populations. Our findings, along with Baglivio et al. (2014), show that it is still a useful tool in assessing and understanding juvenile offending. The usefulness of this typology is perhaps more so in the present than in 1990 due to the increased use of automated juvenile justice data systems. In 1990, most of the information for the typologies would have been collected using hard copy files, while many agencies today have this information readily available. One limitation of applying the typology to practice is that it is retrospective and does not aid in the early identification of chronic/serious/violent youth. The findings presented here, however, suggest that juvenile justice agencies should use the career types in conjunction with risk assessments. First, agencies and researchers should attempt to establish risk assessment score tipping points that may identify the first emergence of chronic offending. This step would better allow for interventions to be more risk based and targeted when youth first enter the juvenile justice system. Second, our research has shown a clear delineation in risk profiles of chronic and nonchronic juvenile offenders. As youth become more chronic, juvenile justice agencies should rely less on risk assessments and use more specific youth-related clinical assessments that will result in more targeted interventions.

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

Ayers

C. D.

Williams

J. H.

Hawkins

J. D.

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, 15, 277–306.

Baglivio

M. T.

Jackowski

Greenwald

M. A.

Howell

J. C.

(2014). Serious, violent, and chronic juvenile offenders: A statewide analysis of prevalence and prediction of subsequent recidivism using risk and protective factors. Criminology & Public Policy, 13, 83–116.

Baglivio

M. T.

Wolff

K. T.

Piquero

A. R.

Epps

(2015). The relationship between adverse childhood experiences (ACE) and juvenile offender trajectories in a juvenile offender sample. Journal of Criminal Justice, 43, 229–241.

Barnes

J. C.

(2013). Analyzing the origins of life-course-persistent offending: A consideration of environmental and genetic influences. Criminal Justice and Behavior, 40, 519–540.

Bogue

Vanderbilt

Ehret

(2005). Report on the validation study for the Juvenile Assessment Generic (JAG) and the Substance Use Survey (SUS) for CSSD Juvenile Probation. Wethersfield: Connecticut Judicial Branch, Court Support Services Division.

Borduin

C. M.

Ronis

S. T.

(2012). Individual, family, peer, and academic characteristics of female serious juvenile offenders. Youth Violence and Juvenile Justice, 10, 386–400.

Brown

S. A.

(2015). Trends in juvenile justice state legislation 2011–2015. Washington, DC: National Conference of State Legislatures.

Browning

Huizinga

(1999, 4). Highlights of findings from the Denver Youth Survey (OJJDP Fact Sheet No. 106). Washington, DC: Office of Justice Programs, U.S. Department of Justice.

Browning

Loeber

(1999, 2). Highlights of findings from the Pittsburgh Youth Study (OJJDP Fact Sheet No. 95). Washington, DC: Office of Justice Programs, U.S. Department of Justice.

10.

Browning

Thornberry

T. P.

Porter

P. K.

(1999, 4). Highlights of findings from the Rochester Youth Development Study (OJJDP Fact Sheet No. 103). Washington, DC: Office of Justice Programs, U.S. Department of Justice.

11.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates.

12.

Corrado

R. R.

McCuish

E. C.

Hart

S. D.

DeLisi

(2015). The role of psychopathic traits and developmental risk factors on offending trajectories from early adolescence to adulthood: A prospective study of incarcerated youth. Journal of Criminal Justice, 43, 357–368.

13.

DeLisi

(2006). Zeroing in on early arrest onset: Results from a population of extreme career criminals. Journal of Criminal Justice, 34, 17–26.

14.

DeLisi

Neppl

T. K.

Lohman

B. J.

Vaughn

M. G.

Shook

J. J.

(2013). Early starters: Which type of criminal onset matters most for delinquent careers? Journal of Criminal Justice, 41, 12–17.

15.

DeLisi

Piquero

A. R.

(2011). New frontiers in criminal careers research, 2000–2011: A state-of-the-art review. Journal of Criminal Justice, 39, 289–301.

16.

Farrington

Piquero

A. R.

Jennings

W. G.

(2013). Offending from childhood to late middle age: Recent results from the Cambridge study in delinquent development. New York, NY: Springer.

17.

Fox

B. H.

Perez

Cass

Baglivio

M. T.

Epps

(2015). Trauma changes everything: Examining the relationship between adverse childhood experiences and serious, violent, and chronic juvenile offenders. Child Abuse & Neglect, 46, 163–173.

18.

Giordano

P. C.

Cernkovich

S. A.

Holland

D. D.

(2003). Changes in friendship relations over the life course: Implications for desistance from crime. Criminology, 41, 293–327.

19.

Guerra

N. G.

(1998). Serious and violent juvenile offenders: Gaps in knowledge and research priorities. In Loeber

Farrington

D. P.

(Eds.), Serious & violent juvenile offenders: Risk factors and successful interventions (pp. 399–404). Thousand Oaks, CA: Sage.

20.

Hamparian

D. M.

Schuster

Dinitz

Conrad

J. P.

(1978). The violent few. Lexington, MA: Lexington Books.

21.

Hawkins

J. D.

Herrenkohl

Farrington

D. P.

Brewer

Catalano

R. F.

Harachi

T. W.

(1998). A review of predictors of youth violence. In Loeber

Farrington

D. P.

(Eds.), Serious & violent juvenile offenders: Risk factors and successful interventions (pp. 106–146). Thousand Oaks, CA: Sage.

22.

Howell

J. C.

(2003a). Diffusing research into practice using the comprehensive strategy for serious, violent, and chronic juvenile offenders. Youth Violence and Juvenile Justice, 1, 219–245.

23.

Howell

J. C.

(2003b). Preventing and reducing juvenile delinquency: A comprehensive framework. Thousand Oaks, CA: Sage.

24.

Howell

J. C.

Krisberg

Hawkins

J. D.,

Wilson

J. J.

(Eds.). (1995). Sourcebook on serious, violent, and chronic juvenile offenders. Newbury Park, CA: Sage.

25.

Howell

J. C.

Lipsey

M. W.

Wilson

J. J.

Howell

M. Q.

(2014). A practical approach to evidence-based juvenile justice systems. Journal of Applied Juvenile Justice Services, 1–21. Retrieved from http://npjs.org/jajjs/wp-content/uploads/2014/12/JAJJS-Article-Howell-et-al-edited-kd.pdf

26.

Huizinga

Jakob-Chien

(1998). The contemporaneous co-occurrence of serious and violent juvenile offending and other problem behaviors. In Loeber

Farrington

D. P.

(Eds.), Serious & violent juvenile offenders: Risk factors and successful interventions (pp. 47–67). Thousand Oaks, CA: Sage.

27.

Huizinga

Weiher

A. W.

Menard

Espiritu

Esbensen

(1998, 11). Some not so boring findings from the Denver Youth Survey. Paper presented at the meeting of the American Society of Criminology, Washington, DC.

28.

Johansson

Kempf-Leonard

(2009). A gender-specific pathway to serious, violent, and chronic offending? Crime & Delinquency, 55, 216–240.

29.

Kempf-Leonard

Tracy

Howell

J. C.

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

30.

Kennedy

T. D.

Burnett

K. F.

Edmonds

W. A.

(2011). Intellectual, behavioral, and personality correlates of violent vs. non-violent juvenile offenders. Aggressive Behavior, 37, 315–325.

31.

Krohn

M. D.

Lizotte

A. J.

Bushway

S. D.

Schmidt

N. M.

Phillips

M. D.

(2014). Shelter during the storm: A search for factors that protect at-risk adolescents from violence. Crime & Delinquency, 60, 379–401.

32.

Laub

J. H.

Sampson

R. J.

(2001). Understanding desistance from crime. Crime and Justice: A Review of Research, 28, 1–69.

33.

Loeber

Ahonen

(2014). What are the policy implications of our knowledge on serious, violent, and chronic offenders? Criminology & Public Policy, 13, 117–125.

34.

Loeber

R.,

Farrington

D. P.

(Eds.). (1998). Serious and violent juvenile offenders: Risk factors and successful interventions. Thousand Oaks, CA: Sage.

35.

Loeber

Farrington

D. P.

(2000). Young children who commit crime: Epidemiology, developmental origins, risk factors, early interventions, and policy implications. Development and Psychopathology, 12, 737–762.

36.

Loeber

Farrington

D. P.

Stouthamer-Loeber

Moffitt

T. E.

Caspi

(1998). The development of male offending: Key findings from the first decade of the Pittsburgh Youth Study. Studies in Crime and Crime Prevention, 7, 141–172.

37.

Loeber

Farrington

D. P.

Waschbusch

D. A.

(1998). Serious and violent juvenile offenders. In Loeber

Farrington

D. P.

(Eds.), Serious and violent juvenile offenders: Risk factors and successful interventions (pp. 13–29). Thousand Oaks, CA: Sage.

38.

Loeber

Stouthamer-Loeber

Van Kammen

Farrington

D. P.

(1991). Initiation, escalation and desistance in juvenile offending and their correlates. Journal of Criminal Law & Criminology, 82, 36–82.

39.

Loeber

Van Kammen

W. B.

Fletcher

. (1996). Serious, violent, and chronic offenders in teh Pittsburgh Youth Study: Unpublished data. Pittsburgh, PA: Western Psychiatric Institute and Clinic.

40.

McCluskey

C. P.

McCluskey

J. D.

Bynum

T. S.

(2006). Early onset offending and later violent and gun outcomes in a contemporary youth cohort. Journal of Criminal Justice, 34, 531–541.

41.

McCuish

E. C.

Corrado

R. R.

Hart

S. D.

DeLisi

(2015). The role of symptoms of psychopathy in persistent violence over the criminal career into full adulthood. Journal of Criminal Justice, 43, 345–396.

42.

Moffitt

T. E.

(1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674–701.

43.

Monahan

K. C.

Steinberg

Cauffman

Mulvey

E. P.

(2013). Psychosocial (im)maturity from adolescence to early adulthood: Distinguishing between adolescence-limited and persisting antisocial behavior. Development and Psychopathology, 25, 1093–1105.

44.

Mulvey

Steinberg

Fagan

Cauffman

Piquero

A. R.

Chassin

L.,

… Losoya

S. H.

(2004). Theory and research on desistance from antisocial activity among serious juvenile offenders. Youth Violence and Juvenile Justice, 2, 213–236.

45.

Nagin

D. S.

Farrington

D. P.

(1992). The onset and persistence of offending. Criminology, 30, 501–523.

46.

Nagin

D. S.

Farrington

D. P.

Moffitt

T. E.

(1995). Life-course trajectories of different types of offenders*. Criminology, 33, 111–139.

47.

National Juvenile Justice Network. (2012, July). Advances in juvenile justice reform: 2009–2011. Washington, DC: National Juvenile Justice Network.

48.

Piquero

A. R.

Brame

Lynam

(2004). Studying criminal career length through early adulthood among serious offenders. Crime & Delinquency, 50, 412–435.

49.

Piquero

A. R.

Hawkins

J. D.

Kazemian

(2012). Criminal career patterns. In Loeber

Farrington

D. P.

(Eds.), From juvenile delinquency to adult crime: Criminal careers, justice policy, and prevention (pp. 14–46). New York, NY: Oxford University Press.

50.

Piquero

A. R.

Schubert

C. A.

Brame

(2014). Comparing official and self-report records of offending across gender and race/ethnicity in a longitudinal study of serious youthful offenders. Journal of Research in Crime and Delinquency, 51, 526–556.

51.

Sampson

R. J.

Laub

J. H.

(1993). Turning points in the life course: Why change matters in the study of crime. Criminology, 31, 301–326.

52.

Skeem

J. L.

Scott

Mulvey

E. P.

(2014). Justice policy reform for high-risk juveniles: Using science to achieve large-scale crime reduction. Annual Review of Clinical Psychology, 10, 709–739.

53.

Snyder

H. N.

(1998). Serious, violent, and chronic juvenile offenders: An assessment of the extent of and trends in officially recognized serious criminal behavior in a delinquent population. In Loeber

Farrington

D. P.

(Eds.), Serious & violent juvenile offenders: Risk factors and successful interventions (pp. 428–444). Thousand Oaks, CA: Sage.

54.

Thornberry

T. P.

Krohn

M. D.

Lizotte

A. J.

Chard-Wierschem

(1993). The role of juvenile gangs in facilitating delinquent behavior. Journal of Research in Crime and Delinquency, 30, 55–87.

55.

Thornberry

T. P.

Krohn

M. D.

Lizotte

A. J.

Smith

C. A.

Porter

P. K.

(1998, 11). Taking stock: An overview of findings from the Rochester Youth Development Study. Paper presented at the meeting of the American Society of Criminology, Washington, DC.

56.

Trulson

C. R.

Caudill

J. W.

Haerle

D. R.

DeLisi

(2012). Cliqued up: The postincarceration recidivism of young gang-related homicide offenders. Criminal Justice Review, 37, 174–190.

57.

Valois

R. F.

MacDonald

J. M.

Bretous

Fischer

M. A.

Drane

J. W.

(2002). Risk factors and behaviors associated with adolescent violence and aggression. American Journal of Health Behavior, 26, 454–464.

58.

Vaughn

M. G.

DeLisi

Gunter

Beaver

K. M.

Perron

B. E.

Howard

M. O.

(2011). The severe 5%: A latent class analysis of the externalizing behavior spectrum in the United States. Journal of Criminal Justice, 39, 75–80.

59.

Vaughn

M. G.

Howard

M. O.

(2005). The construct of psychopathy and its potential contribution to the study of serious, violent, and chronic youth offending. Youth Violence and Juvenile Justice, 3, 235–252.

60.

Vaughn

M. G.

Salas-Wright

C. P.

DeLisi

Maynard

B. R.

(2014). Violence and externalizing behavior among youth in the United States: Is there a severe 5%? Youth Violence and Juvenile Justice, 12, 3–21.

61.

Wilson

J. J.

Howell

J. C.

(1993). Comprehensive strategy for serious, violent, and chronic juvenile offenders. Washington, DC: U.S. Department of Justice, Office of Juvenile Justice and Delinquency Prevention.

62.

Wolfgang

M. E.

Figlio

R. M.

Sellin

(1972). Delinquency in a birth cohort. Chicago, IL: University of Chicago Press.

63.

Zara

Farrington

D. P.

(2013). Assessment of risk for juvenile compared with adult criminal onset implications for policy, prevention, and intervention. Psychology, Public Policy, and Law, 19, 235–249.