Who Will Become Productive Adults? Longitudinal Patterns of Gainful Activities Among Serious Adolescent Offenders

Abstract

The juvenile justice system is charged with the welfare of the children it serves, yet less is known about the prosocial behaviors of adolescent youthful offenders. This study identifies patterns of prosocial behavior for 7 years among serious adolescent offenders, the correlates of each pattern, and associated patterns of secure placement. Using 7 years of monthly data from the Pathways to Desistance Study (N = 1,354), we used group-based trajectory models to identify longitudinal patterns of positive youth behaviors related to school and work among serious adolescent offenders and a joint trajectory model to assess the relationship between trajectories of institutional placement and positive youth behaviors. Four groups were identified that demonstrated a high, low, medium, and dips-then-rises likelihood of gainful activities throughout the study period. Gainful activities were negatively associated with risk for delinquency across multiple domains. Juvenile justice interventions should consider prosocial promise in addition to risk for delinquency.

Keywords

trajectories life-course theory incarceration desistance from crime juvenile delinquency

In 2017, 43,580 youth were in residential placement in the United States (Sickmund et al., 2019). In spite of this relatively high reliance on institutional care in the United States (Hazel, 2008), juvenile justice intervention does not clearly facilitate desistance, and instead, possibly has a harmful effect (Barrick, 2014; Gilman et al., 2015; Lee et al., 2017; Murray et al., 2014). Yet, beyond the impact of juvenile justice intervention on desistance, there remains the developmental question as to how justice practices can best be delivered to foster healthy adolescent development that contributes to adult productivity. We lack systematic knowledge of how juvenile justice secure placement may contribute to the later success for these young people. For example, among serious adolescent offenders, around 80% eventually desist (Piquero et al., 2013), but the impact of juvenile justice interventions on outcomes is highly variable and equivocal (Mulvey et al., 2010). Improving policies to promote positive youth behaviors and development is wise not only from the “do no harm” perspective, but also from a practical standpoint regarding policies that should reduce the burden of unproductive citizens to taxpayers through loss of taxes from future wages and potential costs in terms of public assistance (Apel & Sweeten, 2010; Gilman et al., 2015).

This study contributes to our understanding of which serious adolescent offenders are most likely to become productive adults by identifying longitudinal patterns of employment and school attendance (referred to as gainful activities) during the transition to adulthood. Limited educational attainment and uneven employment histories during the transition to adulthood period are strong predictors of later offending (Aaltonen et al., 2011). This study explores the factors in adolescence that may differ between our identified trajectory groups and how longitudinal patterns of gainful activities compares to contemporaneous longitudinal patterns of secure placement.

Theoretical Rationale

We take a developmental approach in order to identify how to best support healthy development. Developmental criminology, which focuses on explaining intra-individual differences, has garnered less overall attention (Sampson & Laub, 1997). Criminologists have applied the life course perspective to generate developmental theories of deviance (Moffitt, 1993; Sampson & Laub, 1990, 1997; Thornberry, 1987). This perspective acknowledges an individual’s agency while recognizing that individuals are embedded within social relationships (Elder, 1994). In the life course perspective, “trajectories can be charted by linking states across successive years, the states of employment, for example” (Elder, 1985, p. 31). Within trajectories, “turning points [are] a change in course” (Elder, 1985, p. 32). Thus, identifying both trajectories and turning points of gainful activities can help identify factors that distinguish between those who become productive adults, thereby identifying potential targets for intervention.

Moffit’s (1993) classic work has posited a dual developmental taxonomy which differentiates between individuals who engage in life-course persistent and adolescence-limited antisocial behaviors. Moffit (1993) argues that individuals who engage in life-course-persistent antisocial behavior experience cumulative and contemporary consequences that result from an individual’s interactions with their environment, which maintain the antisocial behavior throughout the life-course, although the behaviors change in each developmental stage. In contrast, individuals who engage in adolescence-limited antisocial behavior experience a maturity gap in adolescence, during which they experience biological adulthood but are not socially recognized adults (Moffitt, 1993). Thus, these individuals engage in mimicry of their life-course-persistent peers that appear more adult-like (Moffitt, 1993). A recent study that employed computer simulation to examine Moffit’s dual taxonomy was able to generate the results predicted by Moffit, yet their intervention analysis suggested that individuals can move between the two groups (Leaw et al., 2015).

Other developmental criminologists have provided a unified theory about the “dynamic processes that alter future outcomes” (Sampson & Laub, 1997, p. 2). A developmental theory which focuses on the interactions between an individual and their social environment to explain both continuity and changes in delinquent behavior is Thornberry’s (1987; Thornberry & Krohn, 2005; Thornberry et al., 1991) interactional theory of delinquency. In addition to taking a life course perspective, a key premise of interactional theory is bidirectional causality (Thornberry & Krohn, 2005). This is similar to other theorists’ description of an individual’s interaction with others, such as Moffit’s (1993) description of cumulative and contemporary consequences of an individual’s behaviors and Sampson and Laub’s (1997) description of cumulative and interactional continuity. Yet, in interactional theory, not only do factors from an individual’s social context change the risk for youth delinquency, but the youth’s decision to engage in delinquent behavior has causal implications for their lives by closing (or opening) opportunities (Thornberry & Krohn, 2005). Thus, interactional theory points to the importance of examining not only antisocial trajectories, but also prosocial trajectories since these provide an indication of the opportunities the individual has been able to maintain. In other words, antisocial and prosocial trajectories are intertwined.

Juvenile Justice and Education and Employment Outcomes

The evidence associating juvenile justice interventions with negative educational and employment outcomes is robust. Juvenile justice interventions are associated with a higher likelihood of high school dropout and lower rates of college enrollment (Aizer & Doyle, 2015; Hjalmarsson, 2008; Kirk & Sampson, 2013; Sweeten, 2006; Tanner et al., 1999; Widdowson et al., 2016). It is important to note, however, that these associations between juvenile justice involvement and educational and employment outcomes may be heavily influenced by selection bias. Youth involved with the juvenile justice system may have traits that would predispose them to poor educational or employment outcomes.

Studies have used advanced statistical techniques to address potential selection bias, and have found evidence of a causal effect of secure juvenile placement on reducing the likelihood of high school graduation (Hjalmarsson, 2008). One study used an instrumental variable approach by exploiting the random assignment of judges, using the judge’s tendency for sentencing as an instrumental variable (Aizer & Doyle, 2015). Using 10 years of administrative data following 35,000 youth in Chicago, IL, Aizer and Doyle found that secure juvenile placement reduced the likelihood of high school graduation by 13 percentage points. This study suggests that youth are unlikely to return to school after incarceration, or if they do, they are often placed into an alternative educational program.

Similarly, the evidence supports a negative association between secure placement and employment outcomes (Apel & Sweeten, 2010; Bernburg & Krohn, 2003; Huebner, 2005). One study used propensity score matching and fixed-effects models to address potential selection bias. Apel and Sweeten (2010) used the National Longitudinal Study of Youth (NLSY97) and found that a youth’s first secure juvenile placement was associated with a reduction in the likelihood of formal employment by 11 percentage points. These few studies provide evidence that there is a causal relationship between juvenile secure placement and negative education and employment outcomes. While the association between secure placement and both education and employment outcomes is robust, less is known about education and employment experiences during the transition to adulthood, and how these experiences may be co-produced with experiences of secure placement.

Given the importance of acquiring human capital during adolescence in preparation for becoming independent adults, disruptions may have an especially deleterious effect on later outcomes. Juvenile justice interventions may have a stronger negative effect on youth than criminal justice interventions on adults (Barrick, 2014). For example, juvenile detention has a stronger negative effect on later offending among 15 and 16 year olds (Aizer & Doyle, 2015). Longitudinal studies that follow youth over time can provide additional insight into how disruptions impact youth, moving beyond the mere understanding that there is a negative impact. Rather, the emphasis can be on the nature and type of the impact, and the factors that differentiate youth who are likely to follow distinct trajectories of school and work over time. Collectively these can assist with helping to target appropriate interventions.

Other Correlates of Education and Employment Outcomes

While juvenile justice interventions may have a negative effect on education and employment income, other factors are also related to both juvenile justice involvement and education and employment outcomes. Race, neighborhood, and socioeconomic status are highly correlated with each other and with both juvenile justice involvement, education, and employment outcomes (Sampson et al., 2002). This reflects Sampson et al. (2002) conclusion that there is social inequality between neighborhoods, and that social problems cluster within neighborhoods. Moreover, these findings have been robust with variations in how neighborhoods have been defined and operationalized (Sampson et al., 2002).

There are several hypothesized mechanisms that operate in disadvantaged neighborhoods and contribute to individual outcomes, which include neighborhood disorganization, both social and physical (Sampson et al., 2002; Wodtke et al., 2011). One study found that concentrated disadvantaged was a predictor of secure detention (Rodriguez, 2013). In part, this was because court officials preferred to remove youth from neighborhoods that they perceived to provide few prosocial and many antisocial opportunities (Rodriguez, 2013). Thus, given the geographic isolation of African Americans (Sampson et al., 2002), it is not surprising that race is also highly correlated with juvenile justice involvement. Youth of color are more likely to be placed in secure detention (Rodriguez, 2013), comprise the majority (67%) of youth in residential placement, and were detained longer than white youth (Hockenberry, 2020).

The high correlations between race and neighborhood with both education and employment outcomes contribute to the difficulty in disentangling the unique effect of each factor. There has been a racial gap in educational attainment where fewer black young adults complete college than white young adults (Aud et al., 2010; McDaniel et al., 2011). In turn, educational attainment is associated with employment outcomes: among men with bachelor’s degrees, black men had lower employment rates than white men (McDaniel et al., 2011). Furthermore, neighborhood disadvantage is also associated with worse education and employment outcomes. Growing up in a disadvantaged neighborhood was associated with a reduced likelihood of high school graduation, ranging from 96 to 76% reduced probability of graduation (Wodtke et al., 2011). Similarly, in the Moving to Opportunity study, if children moved to a low-poverty neighborhood prior to turning 13, they reported higher college attendance and increased earnings (31% higher) (Chetty et al., 2015).

The concentration of neighborhood disadvantage results in the concentration of families of lower socioeconomic status (SES). Also related to both neighborhood disadvantage and parental SES are an individual’s intelligence. SES is correlated with brain functioning, specifically language and executive functions (Hackman & Farah, 2009). These deficits may contribute to the association between SES and academic achievemen—SES has been found to have a medium effect on academic achievement (Sirin, 2005). Similarly, poverty has been found to be related to lower IQ and poor academic achievement (McLoyd, 1998), and intelligence is a promotive factor against offending and violent behavior (Ttofi et al., 2016). Thus, race, socioeconomic status, neighborhood context, and individual IQ are related to juvenile justice outcomes as well as gainful activities.

The Current Study

In the current study, we used group-based trajectory modeling to identify longitudinal patterns of gainful activities among a sample of serious adolescent offenders. These are the youth most likely to receive the most stringent sanctions (e.g., secure placement); differentiating effects among these youths is more useful than comparing them to a general population of youth who commit minor offenses that might be considered “normative” adolescent behavior (such as experimenting with alcohol). Next, we identified correlational differences in factors related to risk for offending, education, and employment outcomes, including risk scores based on a standardized risk assessment. Finally, we estimated joint trajectories to identify the association between these longitudinal patterns of gainful activities and concurrent trajectories of institutional placement identified in a prior study (Lee et al., 2018). Thus, we also sought to develop a picture of how these prosocial and antisocial phenomena co-develop during the transition to adulthood.

Method

Data

The Pathways to Desistance study consists of 1,354 youth (654 in Maricopa County, Arizona and 700 in Philadelphia County, Pennsylvania) who were between 14 and 18 years of age who had committed a serious offense and were found guilty (Mulvey et al., 2004; Schubert et al., 2004). These youth were first interviewed in 2000 to 2003 and followed for 7 years, when participants were 21 to 25 years old. An average of 90% of the sample was interviewed at each follow-up interview.

The Pathways to Desistance study includes monthly calendar data about gainful activities and time spent in secure placement, which enabled us to estimate longitudinal patterns at a finely grained focus. The data used for the current analyses were downloaded from the Inter-university Consortium for Political and Social Research (Mulvey, 2013, 2017). The majority of the study participants are male (86.4%), and the sample is diverse (20.2% white, 41.4% black, 33.5% Hispanic).

Measures

The key variable was the monthly measure of gainful activities: regular school attendance or at least part-time employment. This is a prosocial measure of the youth’s acquisition of human capital. Regular school attendance was defined as not missing more than 5 days for the month, and was constructed from a youth’s responses to questions about enrollment, suspensions and expulsions, and whether they missed five or more days of school that month for any reason. A youth was classified as having at least part-time work if they worked two or more weeks of a month for at least 20 or more hours per week. This was constructed from a youth’s responses to questions about whether they had a job, and if so, how many weeks and how many hours per week. If a youth met criteria for regular school attendance or at least part-time employment, they were coded as having engaged in gainful activities for the month (=1). If the youth was in an institution for more than 7 days of the month, they were coded as missing for that month.

The second key variable was the proportion of time the youth spent in a secure residential facility, defined as a facility which does not permit the youth to interact with the community (see Mulvey et al., 2007 for a more detailed description and examples). This variable had values that ranged from 0 (no time spent in detention secure facility) to 1 (the whole time spent in a secure facility) or any range of values between.

We looked at numerous potential predictors of group membership in the school and employment gainful activities trajectories. These variables were measured at baseline to gain a better understanding of what differentiated these groups prior to their current episode with the juvenile justice system. Descriptive statistics for these variables are presented in the first column of Table 2.

Individual demographic characteristics included gender, age, and race/ethnicity. Gender and race/ethnicity are correlated with education and employment outcomes in early adulthood (Danziger & Ratner, 2010). Gender was coded as male (=1) or female (=0). Age was coded based on how the system was likely to respond to the youth, whether ages 14 to 15 (=1) or age 16 (=1) compared to older adolescents ages 17 to 19 (= 0). Race/ethnicity was coded with several dichotomous indicators where white (=0) was the reference group, and the indictors were black (=1), Hispanic (=1), or other race (=1).

Information on their involvement with the juvenile justice system was examined. The grade of the most serious charge indicated whether the youth was charged with a felony or misdemeanor, where the most serious offenses had the lowest values. The number of petitions, including the current petition, was based on court records and ranged from 1 to 15. These provide measures of delinquency and potential disruptions that may impact the youth’s opportunities, and thus contribute to a lower likelihood of prosocial behaviors.

Although standardized risk assessments were not administered to the participants in the study, risk scores created for a previous study were examined (Mulvey et al., 2016). The risk scores were derived from multiple sources of information collected in the Pathways interview, and approximated the Youth Level of Service/Case Management Inventory (YLS:CMI). In addition to a total risk score, the subscales for eight domains were also examined: offense history, family characteristics/parenting, education/employment, peer relations, substance abuse, leisure/recreation, personality/behavior, and attitudes/orientation. Risk scores predict offending behavior and often inform decisions at several points in the juvenile justice system. Thus, understanding how these risk scores may be related to patterns of gainful activities has policy and program implications.

A measure of intelligence was based on the vocabulary and matrix reasoning subtests of the Wechsler Abbreviated Scale of Intelligence (WASI; Wechsler, 1999) with values ranging from 55 to 128. IQ is related to delinquent behaviors and success in schools, and thus likely would be related to gainful activities.

We included measures of mental and behavioral health, since youth involved in the juvenile justice system report high levels of unmet service needs (Maschi et al., 2008; Mulvey et al., 2007), and mental health disorders have been found to be related to educational attainment (Schubert et al., 2018). Untreated mental and behavioral health symptoms provide a measure of individual functioning across multiple domains, and thus is likely to be related to gainful activities. Two measures of mental health were taken from the Composite International Diagnostic Interview (CIDI; (World Health Organization, 1990): a count of posttraumatic stress disorder (PTSD) symptoms ever present (values ranged from 0 to 17) and a measure of major depressive disorder (MDD) symptoms ever present (values ranged from 0 to 9). We used symptom counts to provide a measure of gradations of functioning rather than whether youth crossed a diagnostic threshold. Diagnosis of PTSD and MDD were low (6.4% and 7.2%, respectively), thus symptom counts provided more variability. Finally, we included a measure of the consequences of substance use, which was a modified version of the Substance Use/Abuse Inventory (Chassin et al., 1991). The lifetime total consequences scale had values that ranged from 0 to 16.

Two measures were related to the adolescent’s school risk factors, which would be correlated with gainful activities during the transition to adulthood. Two dichotomous variables measured the youth’s involvement in school: whether the youth was enrolled (=1) or had ever been expelled (=1) from school. Two variables measured the youth’s attachment along two dimensions: teachers (alpha = .65) and school orientation (alpha = .83). These two bonding variables were based on 13 items rated on a scale from 1 to 5 (Cernkovich & Giordano, 1992).

There were three measures of the youth’s family risk factors. A measure of maternal warmth (alpha = .92) indicated the youth’s attachment to this primary family figure, and was measured with the Quality of Parental Relationships Inventory (Conger et al., 1994). Two measures provided an indication of the family’s resources and structure, as families provide important resources that can facilitate a successful transition to adulthood (Waithaka, 2014). The parental index of social position was computed using the Hollingshead (1975) formula, based on both parent education and occupation. Higher values indicated lower socioeconomic status, and values ranged from 11 to 77. A dichotomous variable also indicated whether the youth was from a single parent household (=1).

There were two measures from the Neighborhood Conditions Measure (Sampson & Raudenbush, 1999), ratings of both social (alpha = .87) and physical (alpha = .91) conditions in the neighborhood. Neighborhoods provide employment opportunities for youth, and thus youth from more disorganized neighborhoods are less likely to be engaged in gainful activities. The scale of physical disorder consists of 12 questions, such as graffiti or cigarettes on the street. The scale of social disorder consists of nine questions about neighborhood activities, such as whether adults fight or there are people using needles or syringes. Higher values indicated greater disorder.

Site was coded as a dichotomous variable: Philadelphia County = 1 and Maricopa County = 0. Indicators for each of the four interview years (2000–2003) were also included.

Analytic Approach

Using the monthly gainful activities dichotomous variable over the course of 87 months, we used GBTM to estimate trajectories of gainful activities. Group-based trajectory modeling (GBTM) assumes clusters of developmental trajectories that may reflect distinct etiologies, in contrast to latent growth analysis, which assumes a common process of growth with individual variability (Nagin, 2005). We estimated models based on study time, where month 1 represented the first month the youth was in the study. We were interested in whether there were common trajectories of prosocial behaviors after becoming involved in the juvenile justice system for a serious offense. The Bayesian Information Criterion and Akaike Information Criterion statistics were used to evaluate model fit (Nagin, 2005). We used both model parsimony and substantive considerations in order to select the best solution (Nagin, 2005). Once we selected the best solution, we examined other fit indicators to evaluate the suitability of the selected solution. These other indicators included comparing the group probability and proportion assigned, as well as examining the average posterior probability, where an average posterior probability over 0.70 is considered good (Nagin, 2005). We then tested bivariate associations (chi-square and ANOVA) between the trajectory group membership and individual, family, and neighborhood characteristics, as well as risk scores. We also conducted poc hoc tests (e.g., Bonferroni and Scheffes for robustness) for significant bivariate associations to identify the groups that differed significantly (Castaneda et al., 1993; Wilcox, 1987).

Finally, we estimated a joint trajectory model, where we simultaneously estimated patterns of gainful activities and patterns of secure institutional placement. A joint trajectory model allowed us to model two distinct but related developmental processes that evolve simultaneously (Nagin, 2005). We used a previously estimated four-group solution of institutional placements (Lee et al., 2018). The four trajectories of secure institutional placement included youth who spent: the whole study period in the community (34.2%); the first few years in secure institutional placement followed by declining time in placement (24.4%); varying time in placement throughout the study period (22.5%); and consistent time in institutional placement throughout the study period (18.8%). The joint trajectory model allowed us to observe associations between prosocial behaviors and the most extreme juvenile justice sanction, thus enabling us to explore potential bidirectional causality. We report on the conditional probabilities estimated in the joint trajectory model.

Results

The final model we selected included four group-based trajectories. The model fit indicators (AIC and BIC) are presented in Table 1; both indicators steadily decreased in absolute value. Thus, as recommended by Nagin (2005), we selected the four-group model based on uniqueness of trajectories and ease of interpretation. The three-group solution included groups with consistently (1) high, (2) low, and (3) moderate levels of gainful activities, none of which reflected any turning points and was thus of limited usefulness. The four-group solution indicated divergence in the trajectory that started with a moderate level of engagement in gainful activities, and thus provided a balance between concerns for parsimony and useful new information. The additional fit indicators, presented in the bottom of Table 1, show that the four-group solution was appropriate: the estimated probability of group membership was comparable to the proportion assigned, and the average posterior probability for all four groups was above 0.85, well above the recommended 0.7 (Nagin, 2005).

Table 1.

Model Fit Statistics.

Number of groups	BIC (N = 82,876)	BIC (N = 1,093)	AIC	LL
1 Group	−40,507	−40,499	−40,489	−40,485
2 Groups	−34,099	−34,081	−34,058	−34,049
3 Groups	−32,791	−32,763	−32,728	−32,714
4 Groups	−32,094	−32,056	−32,009	−31,990
5 Groups	−31,394	−31,346	−31,286	−31,262
6 Groups	−30,923	−30,865	−30,792	−30,763
7 Groups	−30,553	−30,485	−30,400	−30,366
8 Groups	−30,231	−30,153	−30,056	−30,017
Group number	Group name	N	π Estimated probability of group membership (%)	P_i proportion assigned to group (%)	Average posterior probability
Group 1	Low gainful activities	293	26.8	26.40	0.954
Group 2	Medium gainful activities	388	33.0	34.95	0.878
Group 3	Dips-then-rises gainful activities	175	17.3	15.77	0.930
Group 4	High gainful activities	254	22.8	22.88	0.940

The Four Trajectory Groups

The four groups are presented in Figure 1, and the confidence intervals for these trajectories are indicated by the dotted lines. The low group (26.8%) reflected the lowest levels of gainful activities, with declining levels throughout the study period. The high group (22.8%) indicated high levels of engagement in gainful activities throughout the study period. There are two groups that start with a medium-high level of gainful activities. One group dips initially and then ends the study period with the highest levels of gainful activities, which we call dips-then-rises (17.3%). The other group is the largest group with 33.0% of the sample, and reports an overall medium level of engagement in gainful activities with a steady and slow decline until month 60 (5 years after the index offense) followed by a slight increase at the end.

Figure 1.

Trajectories of gainful activities.

There were significant bivariate associations between the four trajectory groups and various characteristics, which are reported in Table 2. We highlight the significant differences based on post-hoc tests. There were significant differences in race/ethnicity by group, where the low group was differed from the dips-then-rises and high groups, and included more black and fewer white youth. This is consonant with the racial characteristics of the Philadelphia data collection site, from which a higher percentage of youth in this trajectory group originated. Additionally, this group differed significantly from the dips-then-rises and high groups in reporting higher average number of prior petitions, lower average IQ score, and higher scores on both social and physical neighborhood disorganization. Additionally, this group differed significantly from the high group in family factors, reporting higher maternal warmth, lower parental socioeconomic status, and higher rates of being from a single parent household. This group also differed significantly from the high group in school factors, reporting lower rates of being enrolled in school, average teacher bonding, and school orientation scores. These youth also report the highest rates of ever being expelled, which was significantly different from all three of the other groups. In terms of risk scores, these youth reported significantly different total, offenses, and education risk scores in comparison to the low group, and reported significantly different attitudes scores compared to the medium group.

Table 2.

Descriptive Statistics and Gainful Activities Group Differences.

	Overall			Group 1: low		Group 2: medium		Group 3: dips-then-rises		Group 4: high
	(n = 1,110)			(n = 293)		(n = 388)		(n = 175)		(n = 254)
	N (%)		χ²	N (%)		N (%)		N (%)		N (%)
Male	941 (.85)		1.03	247 (.84)		325 (.84)		152 (.87)		217 (.85)
Female	169 (.15)		1.03	46 (.16)		63 (.16)		23 (.13)		37 (.15)
Age
14–15	388 (.35)		6.64	95 (.32)		152 (.39)		60 (.34)		81 (.32)
16	339 (.31)			93 (.32)		113 (.29)		48 (.27)		85 (.34)
17–19	383 (.35)			105 (.36)		123 (.32)		67 (.38)		88 (.35)
Race/ethnicity
White (ref)	233 (.21)		79.04***	34 (.12)^a		68 (.18)^a		55 (.31)		76 (.30)
Black	495 (.45)			172 (.59)		192 (.49)		64 (.37)		67 (.26)
Hispanic	335 (.30)			76 (.26)		114 (.29)		49 (.28)		96 (.38)
Other race	47 (.04)			11 (.04)		14 (.04)		7 (.04)		15 (.06)
	Mean	SD	F	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Charge grade	4.53	2.14	2.85*	4.45	2.21	4.34	2.22	4.80	2.06	4.73	1.97
Prior petitions	3.06	2.09	5.03**	3.43	2.23^a	3.03	2.03	2.83	1.98	2.82	2.02
Risk assessment scores
Total	8.59	4.98	3.64*	9.40	4.95^b	8.48	4.87	8.28	5.15	8.06	5.00
Offenses	1.17	1.07	3.62*	1.35	1.13^b	1.12	1.01	1.13	1.08	1.06	1.08
	Mean	SD	F	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Family	0.66	0.88	1.02	0.70	0.92	0.68	0.90	0.56	0.82	0.65	0.83
Education	1.72	0.88	4.99**	1.78	0.83^b	1.79	0.91^b	1.72	0.90	1.53	0.86
Peers	0.68	0.94	1.31	0.76	0.98	0.70	0.95	0.62	0.90	0.61	0.91
Substance abuse	2.08	1.54	0.53	2.18	1.62	2.04	1.51	2.02	1.45	2.09	1.56
Leisure	0.31	0.46	0.55	0.34	0.47	0.31	0.46	0.30	0.46	0.29	0.45
Personality	1.11	1.45	1.83	1.28	1.53	1.07	1.44	1.00	1.37	1.03	1.43
Attitudes	0.87	1.19	3.01*	1.03	1.27^c	0.77	1.15	0.93	1.20	0.80	1.11
	Mean	SD	F	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Individual factors
WASI full scale IQ	84.00	13.15	23.35***	80.22	12.77^a	82.83	12.88^b	85.57	13.12^b	89.01	12.36
PTSD symptoms	1.52	3.96	0.41	1.51	3.92	1.61	4.15	1.23	3.36	1.60	4.10
MDD symptoms	0.84	2.37	0.38	0.96	2.48	0.82	2.37	0.78	2.32	0.77	2.28
Substance use	3.95	4.24	0.46	4.06	4.13	3.77	4.22	3.93	4.34	4.12	4.34
Family factors
Maternal warmth	3.21	0.69	2.80*	3.29	0.67^b	3.22	0.70	3.16	0.74	3.13	0.67
Parent index of social position	51.23	11.87	3.97**	52.89	11.95^b	51.40	11.19	50.66	12.17	49.46	12.36
	N (%)		χ²	N (%)		N (%)		N (%)		N (%)
Single parent household	373 (.34)		7.64	116 (.40)^b		126 (.32)		58 (.33)		73 (.29)
School factors
Enrolled in school	828 (.75)		8.79*	203 (.69)^b		291 (.75)		130 (.74)		204 (.80)
Ever expelled	430 (.39)		9.18*	135 (.46)^d		141 (.36)		61 (.35)		93 (.37)
	Mean	SD	F	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Teacher bonding	3.33	0.84	3.70*	3.23	0.82^b	3.31	0.88	3.33	0.84	3.47	0.81
School orientation	3.58	0.74	2.54	3.53	0.71	3.57	0.76	3.51	0.73	3.69	0.72
Neighborhood factors
Socially disorganized	2.31	0.80	21.76***	2.54	0.78^a	2.39	0.78^a	2.16	0.78	2.05	0.76
Physically disorganized	2.38	0.77	15.62***	2.54	0.79^a	2.45	0.74^b	2.28	0.74	2.13	0.75
	N (%)		χ²	N (%)		N (%)		N (%)		N (%)
Study site
Philadelphia	649 (.58)		80.63***	211 (.72)^a		257 (.66)^a		81 (.46)		100 (.39)
Maricopa county	461 (.42)		80.63***	82 (.28)		131 (.34)		94 (.54)		154 (.61)
Year of interview
2000	49 (.04)		9.67	19 (.07)		15 (.04)		4 (.02)		11 (.04)
2001	620 (.56)			153 (.52)		222 (.57)		103 (.59)		142 (.56)
2002	440 (.40)			454 (.41)		151 (.39)		68 (.39)		100 (.39)
2003	1 (.00)			0 (.00)		0 (.00)		0 (.00)		1 (.00)

Significantly different from groups 3 and 4.

Significantly different from group 4.

Significantly different from group 2.

Significantly different from groups 2, 3, and 4.

p < .05. **p < .01. ***p < .001.

When there were statistically significant differences, the medium trajectory group was significantly different from the high and/or the dips-then-rises group. Similar to the low group, the medium group was significantly different from the dips-then-rises and high groups in terms of race/ethnicity, and included more black and fewer white youth. Again, a higher percentage of youth in this trajectory group also originated in Philadelphia. Similar to the low group, this group also reported statistically significant differences in IQ compared to the high group, reporting lower average IQ scores. This group differed from the low group in terms of reporting a lower rate of ever been expelled. This group also differed from the high group, reporting higher levels of neighborhood physical disorganization, and higher levels of neighborhood social disorganization than both the dips-then-rises and high groups. Finally, in terms of risk scores, this group differed from the high group reporting higher average education risk, and differed from the low group reporting lower average attitudes risk.

The dips-then-rises group differed from the low and medium groups in race/ethnicity, reporting a higher percentage of white youth and lower percent of black youth. This reflects that fewer of these youth are from Philadelphia. This group reported lower average number of prior petitions than the low group. This group also reported higher average IQ scores than the low group and lower average IQ scores than the high group. Additionally, this group reported lower rates of ever expelled than the low group. This group reported lower scores on both measures of social disorganization (social and physical) than the low group, and also differed from the medium group in terms of social disorganization. As far as risk scores, this group scored in the middle and was not statistically different from any of the other groups on any of the scales.

Finally, the high gainful activities group generally reported the most positive scores and values. This group consisted of a highest percentage of white and Hispanic youth and lowest percentage of black youth in comparison to the low and medium groups. Again, this group reported a lower percent of youth from Philadelphia than the low and medium groups. Youth in this group reported lower average prior number of petitions compared to the low group. They reported higher average IQ score than the other three groups. This group reported significant differences in family characteristics than the low group: lower maternal warmth, higher parental socioeconomic status, and lower rate of being from a single parent household. This group also reported significant differences in school characteristics than the low group: higher rates of being enrolled in school; lower rates of ever expelled; and higher teacher bonding. This group reported lower social disorganization scores than the low group, and lower physical disorganization scores than the low and medium groups. They also reported the lowest average scores on the risk scales, and were significantly different from the low group in total score and offenses subscale, and the low and medium groups in the education subscale.

The Joint Trajectory Model

Table 3 presents the conditional probabilities from the joint trajectory model, which is the probability a youth will be in each of the four gainful activities groups conditioned on their assignment to a specific institutional placement trajectory. Although we report conditional probabilities, where secure placement precedes gainful activities, our interactional theory approach does not assume a linear causality model where placement leads to gainful activities. Rather, these are concurrent developmental processes. We report probabilities conditioned on the patterns of secure placement because policy implications for the juvenile justice system can be derived from understanding how patterns of secure placement are related to gainful activities.

Table 3.

Joint Trajectory Conditional Probabilities.

Gainful activities conditional on institutional placement	Low gainful activity	Medium gainful activity	Dips-then-rises gainful activity	High gainful activity
Steady in community	0.134	0.201	0.158	0.508
Declining time in placement	0.236	0.246	0.253	0.266
Varying time in placement	0.369	0.257	0.221	0.154
Steady high in institutional placement	0.478	0.202	0.165	0.155

Youth assigned to the steady in community trajectory had about a .51 probability of being in the high levels of gainful activities trajectory and a .16 probability of being in the dips-then-rises gainful activity group. Youth assigned to the declining time in placement had approximately one-fourth probability of being in each of the four gainful activities group. Youth assigned to the varying time in placement had the highest probability (.37) of being in the low levels of gainful activity trajectory, and the lowest probability (.15) of being in the high gainful activities trajectory. Finally, youth assigned to the steady high in institutional placement group had the highest probability (.48) of being in the low gainful activity trajectory and the lowest probabilities of being in the high and dips-then-rises gainful activities trajectory groups (.16 and .17, respectively).

Discussion

While studies have identified trajectories of delinquency or offending over time, much less attention has been paid to identifying trajectories of gainful activities among youth found guilty for a serious offense. To our knowledge, this is the first study to identify four groups with distinct trajectories of prosocial activities. While it is clear that youth involved with the juvenile justice system report poor educational and employment outcomes, this study advances our knowledge of outcomes by providing insight into which of these youths developed those outcomes. Overall, about 40% of the sample ended the study period very likely to be engaged in gainful activities—and thus poised to successfully transition into adulthood. About one-third appear to be somewhat involved in gainful activities, while a little more than one-fourth was very unlikely to be engaged in gainful activities at the end of the study period. In comparison, about 80% of youth from this sample eventually desist from self-reported offending (Piquero et al., 2013). Although we do not know the overlap, even if we assumed that the 20% who do not desist are among those who report low gainful activities throughout the study period, that leaves about 40% of the sample that does not reflect clearly prosocial trajectories of gainful activities (i.e., some from the low and medium gainful activities groups). This could be a missed opportunity for intervention if these youth who desist are not adequately prepared for the transition to adulthood.

The differences between those in the low gainful activities and high gainful activities groups are not surprising—those in the low gainful activities group reported more disadvantages across multiple domains, especially compared to those in the high gainful activities group. Not unexpectedly, the low gainful activities group reported the highest average number of petitions and this may indicate that the youth’s cumulative risk has not been addressed adequately through their system involvement. The youth may not be receiving needed services that address factors related to recidivism, and consequently have not been engaging in gainful activities during a period when their crucial task is to acquire human capital. This suggests that ongoing juvenile justice involvement may only compound their struggles.

From an interactional theory perspective, for these youth in the low gainful activities group, structural adversity has been “intensely coupled” with multiple factors across multiple domains. For example, these youth report the most disorganized neighborhoods, which may provide few employment and prosocial opportunities and prosocial adult models. Additionally, these youth report higher rates of school expulsion than the other three groups which indicates that exclusionary school policies do not redirect trajectories, but rather reinforce trajectories. Notably, this group differs from the high gainful activities group on the total risk score, which appears to be driven by differences in offenses and education. These findings support Thornberry and Krohn’s (2005) premise of bidirectional causality and the mutually reinforcing nature of interactions between the youth and their social environment, especially with respect to their schools. More specifically, this suggests that school expulsions are a maintaining response to the individual, which contributes to interactional continuity (Sampson & Laub, 1997). As Aizer and Doyle (2015) found, youth who were incarcerated were unlikely to return to school with their peers. This suggests that developing programs to return these youths to their schools may be an important aspect of facilitating their prosocial development.

At the same time, this group surprisingly reported higher maternal warmth than the high gainful activities group and no differences in their school orientation. These youth appear to have positive attachments to their families. Yet, this attachment does not appear to be able to compensate for the other challenges in their lives. This suggests that interventions that target parenting practices may be inadequate. Rather, interventions should be targeted at some of the other sources of disadvantage within the youth’s social context, including providing resources to schools so that there are resources to enable schools to minimize or avoid the use of expulsion as a disciplinary response, and neighborhoods to reduce both social and physical disorganization.

The more interesting contrast is between the two groups with diverging trajectories, since this may suggest points of intervention. Yet, there were few statistical differences between these two groups. Other than differences in racial composition and locale, these two groups differed in their reported neighborhood social disorganization score, but not physical disorganization score. Social disorganization refers to the activities of other adults in the youth’s neighborhoods, and reinforces Elder’s (1994) principle of linked lives. This serves as a reminder that these youth do not exist in isolation, and are impacted by the relationships and interactions between the adults in their lives—parents, aunts and uncles, neighbors, etc. This suggests that investing in communities, including the adults in their communities, may have collateral benefits for these youth. Given that social problems cluster by neighborhood (Sampson et al., 2002), targeting neighborhoods for intervention may be more efficient than providing individualized services for specific youth. Yet, less is known about how these neighborhood factors can be addressed—more research should explore how best to intervene at the neighborhood level in order to improve youth outcomes.

Findings from this study suggest that, in addition to measuring risk for recidivism, efforts should be made to consider promise for prosocial outcomes among youth involved in the juvenile justice system. We previously found that the family, education, and leisure risk scores were not related to longitudinal patterns of secure placement (Lee et al., 2018). In this study, peer relationships, substance abuse, and personality risk scores additionally were not related to gainful activities while the education risk score was related to gainful activities. While it may not be surprising that some of the risk scales that predict recidivism do not also predict positive outcomes, this warrants further examination and consideration. Risk scales are used to inform sanctions. While the sanction may be a consequence of the youth’s past actions, it also contributes to the youth’s future outcomes. Incarcerating a youth during this critical developmental period may have additional long-term costs if youth who would otherwise engage in prosocial behaviors receive sanctions that interfere with their ability to adequately prepare for adulthood.

Attending to a youth’s prosocial promise aligns with the movement to incorporate Positive Youth Development (PYD) strategies into the juvenile justice system (Butts et al., 2005). The PYD movement encourages a shift toward building skills and competencies across multiple domains (Butts et al., 2005). These domains are: work, education, relationships, community, health, and creativity (Butts et al., 2010). PYD is a developmental approach, and views the youth as a resource to be cultivated with an emphasis on the individual’s future (Butts et al., 2010). Since many of these youth have not received needed services elsewhere, the juvenile justice system has an opportunity to meet these needs and facilitate the development of these youths. This may require greater collaboration with the other systems involved in the youth’s lives, which may look like a case management approach that coordinates the youth’s ongoing involvement in school, work, rehabilitation and treatment services, and other prosocial activities, rather than disconnecting the youth from their environments completely such as when youth are placed in secure placement.

The conditional probabilities we estimated in the joint trajectory model are not surprising, yet showed that membership in certain trajectories of secure placement do not determine membership in gainful activities trajectories. For example, youth who were assigned to the steady high in placement trajectory were most likely to be assigned to the low gainful activities trajectory group (47.8%), but they still had was a small probability of being assigned to the high gainful activities group (15.5%) while they were in the community.

The joint trajectory model also suggests that the pattern of placement matters. Youth who spent varying amounts of time in the community had a comparable likelihood of engaging in a high level of gainful activities throughout the study period to youth who spent a majority of the 7 years in institutional placement. Moreover, they were most likely to have spent the 7 years engaged in low levels of gainful activities. In contrast, youth who spent declining time in placement had equal probabilities of being assigned to each of the four gainful activities trajectory groups. Thus, youth who cycle in and out of institutional placements have a smaller chance of prosocial trajectories than those who spend declining time in institutional placement and a comparable chance to those who spend most of their time in placement. This probably reflects how difficult it is to maintain a job and progress educationally when experiencing repeated removals from the community. With additional removals, it may become more and more difficult to find a new job, and missing an assignment or two in school may quickly snowball into falling behind by a semester or a year—and getting caught up may become increasingly daunting. Future studies should continue to examine and unpack these associations between patterns of institutional placement and gainful activities during the transition to adulthood.

Limitations

These findings should be interpreted with caution for several reasons. First, the juvenile justice system is operated locally, and this study used data from two locales. Thus, findings may not be generalizable to all locales. Future studies should identify longitudinal patterns in other jurisdictions. Second, this is a correlational study. Group-based modeling technique is a descriptive approach that can provide useful insights, but does not identify causal relationships. Moreover, our analyses examined contemporaneous trajectories, so without a time-lag, there is no causal ordering in our analyses. Finally, our analyses may seem to verify that youth who spend more time in the community simply have more opportunity to be engaged in gainful activities. Although there are more rigorous techniques for controlling for exposure time, we viewed secure placement as a consequence of the youth’s activities. Thus, we only considered a youth’s gainful activities if they were in the community, and thus capture this interaction, or bidirectional causality, between individual agency and social context. Many of these youths may reside in contexts where their opportunities are limited, and thus there may be differences in the amount of agency a youth has to make prosocial choices. Future research should continue to identify patterns of gainful activities among adolescent offenders with other samples and integrate such findings with data on more localized effects of changes.

Conclusions

This study advances our understanding of positive youth behaviors as a consequence of being involved in the justice system, even for individuals with a serious offense. The associations between longitudinal patterns of institutional placement and gainful activities are unsurprising; yet, this is the first study to provide these estimates. While research often focuses on predicting recidivism, such a focus obscures the potential that these youth have for becoming productive adults, especially since correlates of productivity are rarely reported. Without attention to prosocial outcomes, juvenile justice interventions may not only interfere with emotional and social development, but may also miss opportunities to positively impact youth involved with the juvenile justice system. Given that nearly half of the youth who were serious offenders benefited from gainful activities, and these were useful in reducing recidivism, the current emphasis on support reforms in the juvenile justice system to emphasize positive development is warranted.

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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding to support this study was received from the National Institute of Justice, award number 2015-JF-FX-0144.

ORCID iD

JoAnn S. Lee

Ethics Approval and Consent to Participate

This study received approval by the George Mason University Institutional Review Board, project 722899-6.

Availability of Data and Material

The datasets (Research on Pathways to Desistance) analyzed during the current study are available in the Inter-university Consortium for Political and Social Research (ICPSR) repository. The subject measures are available here: https://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/29961/summary. The calendar data are restricted data, which were used under license for the current study, and are available upon reasonable request and with permission here: .

Author Biographies

JoAnn S. Lee, PhD, is an associate professor in the Department of Social Work at George Mason University. Her research seeks to improve outcomes during the transition to adulthood for marginalized youth with a focus on the child welfare and juvenile justice systems. She frames her work within the life course perspective, recognizing the interplay between human agency and larger social contexts throughout an individual’s life.

Faye S. Taxman, PhD, is a university professor at George Mason University and Director of Center for Advancing Correctional Excellence (ACE!). Her expertise is in organizational change, systems of care, and implementation of evidence-based practices. She has received numerous awards for her contributions to the field including the 2017 Joan McCord Award from the Division of Experimental Criminology and the 2015 Rita Warren and Ted Palmer Differential Association Award from the Division of Corrections and Sentencing from the American Society of Criminology.

Edward P. Mulvey, PhD, is professor of Psychiatry Emeritus at the University of Pittsburgh School of Medicine. Dr. Mulvey has directed numerous funded research studies on the link between mental illness and violence, the development of juveniles in the justice system, and the impact of sanctions and interventions for young people who have committed serious crimes. He received his Ph.D. in Psychology from the University of Virginia, and post-doctoral training at Carnegie-Mellon University.

Carol A. Schubert, M.P.H., is a researcher for the Law and Psychiatry Program at the University of Pittsburgh. She has coordinated several large-scale projects regarding the assessment of violence risk among mentally ill individuals as well as factors related to desistance from crime as serious adolescent offenders make the transition to adulthood. Her recent focus has been on the provision of services to and experiences of youth in the juvenile justice system.

References

Aaltonen

Kivivuori

Martikainen

(2011). Social determinants of crime in a welfare state: Do they still matter? Acta Sociologica, 54(2), 161–181. https://doi.org/10.1177/0001699311402228

Aizer

Doyle

J. J.

(2015). Juvenile incarceration, human capital, and future crime: Evidence from randomly assigned judges. The Quarterly Journal of Economics, 130(2), 759–803. https://doi.org/10.1093/qje/qjv003

Apel

Sweeten

(2010). The impact of incarceration on employment during the transition to adulthood. Social Problems, 57(3), 448–479. https://doi.org/10.1525/sp.2010.57.3.448

Aud

Fox

KewalRamani

(2010). Status and trends in the education of racial and ethnic groups (NCES 2010-015). U.S. Department of Education, National Center for Education Statistics.

Barrick

(2014). A review of prior tests of labeling theory. In Labeling theory: Empirical tests (Vol. 18, pp. 89–112). Transaction Publishers.

Bernburg

J. G.

Krohn

M. D.

(2003). Labeling, life chances, and adult crime: The direct and indirect effects of official intervention in adolescence on crime in early adulthood. Criminology, 41(4), 1287–1318. https://doi.org/10.1111/j.1745-9125.2003.tb01020.x

Butts

J. A.

Bazemore

Meroe

A. S.

(2010). Positive youth justice. In Framing justice interventions using the concepts of positive youth development. Coalition for Juvenile Justice.

Butts

Mayer

Ruth

(2005). Focusing juvenile justice on positive youth development. Chapin Hall Center for Children at the University of Chicago.

Castaneda

M. B.

Levin

J. R.

Dunham

R. B.

(1993). Using planned comparisons in management research: A case for the Bonferroni procedure. Journal of Management, 19(3), 707–724.

10.

Cernkovich

Giordano

(1992). School bonding, race and delinquency. Criminology, 30(2), 261–291.

11.

Chassin

Rogosch

Barrera

(1991). Substance use and symptomatology among adolescent children of alcoholics. Journal of Abnormal Psychology, 100(4), 449–463.

12.

Chetty

Hendren

Katz

L. F.

(2015). The effects of exposure to better neighborhoods on children: New evidence from the Moving to Opportunity experiment (Working Paper No. 21156). National Bureau of Economic Research.

13.

Conger

R. D.

Elder

G. H.

Lorenz

F. O.

Simons

R. L.

(1994). Economic stress, coercive family process, and developmental problems of adolescents. Child Development, 65(2), 541–561.

14.

Danziger

Ratner

(2010). Labor market outcomes and the transition to adulthood. The Future of Children, 20(1), 133–158.

15.

Elder

G. H.

(1994). Time, human agency, and social change: Perspective on the life course. Social Psychology Quarterly, 57(1), 4.

16.

Elder

G. H.

Jr. (1985). Perspectives on the life course. In Life course dynamics: Trajectories and transitions, 1968-1980 (pp. 23–49). Cornell University Press.

17.

Gilman

A. B.

Hill

K. G.

Hawkins

J. D.

(2015). When is a youth’s debt to society paid? Examining the long-term consequences of juvenile incarceration for adult functioning. Journal of Developmental and Life-Course Criminology, 1(1), 33–47. https://doi.org/10.1007/s40865-015-0002-5

18.

Hackman

D. A.

Farah

M. J.

(2009). Socioeconomic status and the developing brain. Trends in Cognitive Sciences, 13(2), 65–73. https://doi.org/10.1016/j.tics.2008.11.003

19.

Hazel

(2008). Cross-national comparison of youth justice. Youth Justice Board for England and Wales.

20.

Hjalmarsson

(2008). Criminal justice involvement and high school completion. Journal of Urban Economics, 63(2), 613–630. https://doi.org/10.1016/j.jue.2007.04.003

21.

Hockenberry

(2020). Juveniles in residential placement, 2017. US Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention.

22.

Hollingshead

A. B.

(1975). Four factor index of social status [Unpublished manuscript].

23.

Huebner

B. M.

(2005). The effect of incarceration on marriage and work over the life course. Justice Quarterly, 22(3), 281–303. https://doi.org/10.1080/07418820500089141

24.

Kirk

D. S.

Sampson

R. J.

(2013). Juvenile arrest and collateral educational damage in the transition to adulthood. Sociology of Education, 86(1), 36–62. https://doi.org/10.1177/0038040712448862

25.

Leaw

J. N.

Ang

R. P.

Huan

V. S.

Chan

W. T.

Cheong

S. A.

(2015). Re-examining of Moffitt’s theory of delinquency through agent based modeling. PLoS One, 10(6), e0126752. https://doi.org/10.1371/journal.pone.0126752

26.

Lee

J. S.

Tajima

E. A.

Herrenkohl

T. I.

Hong

(2017). Effects of formal and informal deviant labels in adolescence on crime in adulthood. Social Work Research, 41(2), 97–110. https://doi.org/10.1093/swr/svx003

27.

Lee

J. S.

Taxman

F. S.

Mulvey

E. P.

Schubert

C. S.

(2018). Longitudinal patterns of secure institutional placement among serious adolescent offenders. Criminal Justice and Behavior, 45, 762–782.

28.

Maschi

Hatcher

S. S.

Schwalbe

C. S.

Rosato

N. S.

(2008). Mapping the social service pathways of youth to and through the juvenile justice system: A comprehensive review. Children and Youth Services Review, 30(12), 1376–1385. https://doi.org/10.1016/j.childyouth.2008.04.006

29.

McDaniel

DiPrete

T. A.

Buchmann

Shwed

(2011). The Black gender gap in educational attainment: Historical trends and racial comparisons. Demography, 48(3), 889–914. https://doi.org/10.1007/s13524-011-0037-0

30.

McLoyd

V. C.

(1998). Socioeconomic disadvantage and child development. American Psychologist, 53(2), 185–204.

31.

Moffitt

T. E.

(1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. 100(4), 674–701.

32.

Mulvey

E. P.

(2013). Research on pathways to desistance [Maricopa County, AZ and Philadelphia County, PA]: Subject measures, 2000-2010 (ICPSR29961-v2). Inter-University Consortium for Political and Social Research [distributor]. https://doi.org/10.3886/ICPSR29961.v2

33.

Mulvey

E. P.

(2017). Research on pathways to desistance [Maricopa County, AZ and Philadelphia County, PA]: Calendar data, 2000-2010 [Restricted] (ICPSR32282-v3). Inter-university Consortium for Political and Social Research [distributor]. https://doi.org/10.3886/ICPSR32282.v3

34.

Mulvey

E. P.

Schubert

C. A.

Chung

H. L.

(2007). Service use after court involvement in a sample of serious adolescent offenders. Children and Youth Services Review, 29(4), 518–544. https://doi.org/10.1016/j.childyouth.2006.10.006

35.

Mulvey

E. P.

Schubert

C. A.

Pitzer

Hawes

Piquero

Cardwell

(2016). An examination of change in dynamic risk of offending over time among serious juvenile offenders. Journal of Criminal Justice, 45, 48–53. https://doi.org/10.1016/j.jcrimjus.2016.02.008

36.

Mulvey

E. P.

Steinberg

Fagan

Cauffman

Piquero

A. R.

Chassin

Knight

G. P.

Brame

Schubert

C. A.

Hecker

Losoya

S. H.

(2004). Theory and research on desistance from antisocial activity among serious adolescent offenders. Youth Violence and Juvenile Justice, 2(3), 213–236. https://doi.org/10.1177/1541204004265864

37.

Mulvey

E. P.

Steinberg

Piquero

A. R.

Besana

Fagan

Schubert

Cauffman

(2010). Trajectories of desistance and continuity in antisocial behavior following court adjudication among serious adolescent offenders. Development and Psychopathology, 22(2), 453. https://doi.org/10.1017/S0954579410000179

38.

Murray

Blokland

Farrington

D. P.

Theobald

(2014). Long-term effects of conviction and incarceration on men in the Cambridge Study in Delinquent Development. In Labeling theory: Empirical tests (Vol. 18, pp. 209–236). Transaction Publishers.

39.

Nagin

D. S.

(2005). Group-based modeling of development. Harvard University Press.

40.

Piquero

A. R.

Monahan

K. C.

Glasheen

Schubert

C. A.

Mulvey

E. P.

(2013). Does time matter? Comparing trajectory concordance and covariate association using time-based and age-based assessments. Crime & Delinquency, 59(5), 738–763.

41.

Rodriguez

(2013). Concentrated disadvantage and the incarceration of youth: Examining how context affects juvenile justice. Journal of Research in Crime and Delinquency, 50(2), 189–215. https://doi.org/10.1177/0022427811425538

42.

Sampson

R. J.

Laub

J. H.

(1990). Crime and deviance over the life course: The salience of adult social bonds. American Sociological Review, 55(5), 609–627. https://doi.org/10.2307/2095859

43.

Sampson

R. J.

Laub

J. H.

(1997). A life-course theory of cumulative disadvantage and the stability of delinquency. Developmental Theories of Crime and Delinquency, 7, 133–161.

44.

Sampson

R. J.

Morenoff

J. D.

Gannon-Rowley

(2002). Asessing “neighborhood effects”: Social processes and new directions in research. Annual Review of Sociology, 28(1), 443–478. https://doi.org/10.1146/annurev.soc.28.110601.141114

45.

Sampson

Raudenbush

(1999). Systematic social observation on public spaces: A new look at disorder in urban neighborhoods. American Journal of Sociology, 105(3), 603–651.

46.

Schubert

C. A.

Mulvey

E. P.

Hawes

S. W.

Davis

(2018). Educational and employment patterns in serious adolescent offenders with mental health disorders: The importance of educational attainment. Criminal Justice and Behavior, 45(11), 1660–1687. https://doi.org/10.1177/0093854818784330

47.

Schubert

C. A.

Mulvey

E. P.

Steinberg

Cauffman

Losoya

S. H.

Hecker

Chassin

Knight

G. P.

(2004). Operational lessons from the pathways to desistance project. Youth Violence and Juvenile Justice, 2(3), 237–255. https://doi.org/10.1177/1541204004265875

48.

Sickmund

Sladky

T. J.

Kang

Puzzanchera

(2019). Easy access to the census of juveniles in residential placement. OJJDP.

49.

Sirin

S. R.

(2005). Socioeconomic status and academic achievement: A meta-analytic review of research. Review of Educational Research, 75(3), 417–453. https://doi.org/10.3102/00346543075003417

50.

Sweeten

(2006). Who will graduate? Disruption of high school education by arrest and court involvement. Justice Quarterly, 23(4), 462–480. https://doi.org/10.1080/07418820600985313

51.

Tanner

Davies

O’Grady

(1999). Whatever happened to yesterday’s rebels? Longitudinal effects of youth delinquency on education and employment. Social Problems, 46(2), 250–274.

52.

Thornberry

T. P.

(1987). Toward an interactional theory of delinquency. Criminology, 25(4), 863–892. https://doi.org/10.1111/j.1745-9125.1987.tb00823.x

53.

Thornberry

T. P.

Krohn

M. D.

(2005). Applying interactional theory to the explanation of continuity and change in antisocial behavior. In Integrated developmental and life-course theories of offending (pp. 183–210). Transaction Publishers.

54.

Thornberry

T. P.

Lizotte

A. J.

Krohn

M. D.

Farnworth

Jang

S. J.

(1991). Testing interactional theory: An examination of reciprocal causal relationships among family, school, and delinquency. The Journal of Criminal Law and Criminology, 82(1), 3. https://doi.org/10.2307/1143788

55.

Ttofi

M. M.

Farrington

D. P.

Piquero

A. R.

Lösel

DeLisi

Murray

(2016). Intelligence as a protective factor against offending: A meta-analytic review of prospective longitudinal studies. Journal of Criminal Justice, 45, 4–18. https://doi.org/10.1016/j.jcrimjus.2016.02.003

56.

Waithaka

E. N.

(2014). Family capital: Conceptual model to unpack the intergenerational transfer of advantage in transitions to adulthood. Journal of Research on Adolescence, 24(3), 471–484. https://doi.org/10.1111/jora.12119

57.

Wechsler

(1999). Wechsler abbreviated scale of intelligence. The Psychological Corporation, Harcourt Brace & Company.

58.

Widdowson

A. O.

Siennick

S. E.

Hay

(2016). The implications of arrest for college enrollment: An analysis of long-term effects and mediating mechanisms. Criminology, 54(4), 621–652.

59.

Wilcox

R. R.

(1987). New designs in analysis of variance. Annual Review of Psychology, 38(1), 29–60.

60.

Wodtke

G. T.

Harding

D. J.

Elwert

(2011). Neighborhood effects in temporal perspective: The impact of long-term exposure to concentrated disadvantage on high school graduation. American Sociological Review, 76(5), 713–736. https://doi.org/10.1177/0003122411420816

61.

World Health Organization. (1990). Composite international diagnostic interview. World Health Organization.