Longitudinal Patterns of Secure Institutional Placement Among Serious Adolescent Offenders

Abstract

This study identifies longitudinal patterns of institutional placement to understand experiences in the juvenile justice system. We used monthly calendar data from the Pathways to Desistance study (N = 1,354), which focuses on understanding how serious adolescent offenders desist from antisocial activity. Youth between 14 and 18 years of age were followed for 7 years. We used group-based trajectory modeling to identify longitudinal patterns of institutional placement. We also examined bivariate and multivariate associations between our identified groups and demographic, legal, and extralegal factors. We chose the 4-group solution, which reflected a pattern of steady time in the community (33.3%), and three patterns of youth spending varying (22.5%), declining (24.4%), and steady high (18.8%) time in placement. Significant differences between groups suggest that youth from the most disadvantaged contexts and those who were most likely to have trouble in school and live in disorganized neighborhoods spent the most time in placement.

Keywords

trajectories juvenile delinquency juvenile offenders sanctions incarceration

While the use of institutional placement for serious delinquent youth has plummeted over the last decade (Hockenberry, 2016), this trend does not reduce the need to understand how institutional placement affects positive youth behaviors. Institutional placement is still a fixture in the juvenile justice system, and its impact on youth is an important and underexamined issue. Knowing the effects of institutional placement on youth productivity and development is vital to develop informed policies and practices.

We generally lack a systematic understanding of how juvenile justice interventions can facilitate a process of desistance (Mulvey et al., 2004). While the majority of adolescent offenders eventually desist from crime (Mulvey et al., 2010; Piquero, Monahan, Glasheen, Schubert, & Mulvey, 2013), the level and type of involvement in juvenile justice services does not appear to be highly predictive of which adolescents do or do not desist (Mulvey et al., 2010). Of particular relevance, we have little knowledge of how the powerful and expensive practice of long periods of institutional placement is related to less recidivism (Loughran et al., 2009; Winokur, Smith, Bontrager, & Blankenship, 2008) or whether these experiences have an iatrogenic effect (Gatti, Tremblay, & Vitaro, 2009; Nguyen, Loughran, Paternoster, Fagan, & Piquero, 2016).

This study begins to build our knowledge base of the effects of institutional placement by identifying patterns of placement; decisions about juvenile justice intervention should be informed by a systematic understanding of the implications of these decisions. This study takes the first step toward developing an understanding of how juvenile justice intervention can contribute to positive adult outcomes by identifying longitudinal patterns of institutional placement. This study uses group-based trajectory modeling (GBTM) to identify patterns of secure institutional placement among adolescents for 7 years following a serious offense. This analytic method uses a life course approach to understanding system intervention by taking into account both the timing of events and individual trajectories (Elder, 1994; Macmillan & Copher, 2005). It is possible that certain subgroups of youth are more likely to experience certain patterns of institutional placement. Thus, this study also examines how extralegal factors such as family status, mental health, and substance use are related to the different patterns of institutional placement (Fader, Harris, Jones, & Poulin, 2001; McCoy, Walker, & Rodney, 2012; White, 2016). We explore how demographic, legal, and extralegal factors distinguish among these groups.

Experiences of Institutional Placement

In 2013, there were 173 juveniles in placement for every 100,000 juveniles, with the largest percentage in placement for person offenses (Hockenberry, 2016). But there is variation and serious adolescent delinquents have a different pattern of incarceration than youth overall. Most serious adolescent offenders have at least one institutional stay (residential treatment, detention, etc.), while about half of those processed in juvenile courts received some community-based treatment (Mulvey, Schubert, & Chung, 2007). And the experiences of serious adolescent offenders are somewhat dependent on the locale. A recent study of serious adolescent offenders found that in Philadelphia County, 66% were placed in institutional care after disposition for an average of 15 months. In Maricopa County, 24% were placed in institutional care after disposition for an average of 20 months, and in both locales, youth returned to institutional placement an average 6 months later (Mulvey et al., 2007). While experiences of institutional placement have been described in the aggregate, it would be useful to identify whether there are common trajectories of institutional placement, such as a trajectory where individuals spend long periods of time in placement or a trajectory where individuals may cycle in and out of placement.

Theoretical Rationale

The life course perspective focuses on examining individual development over time within the individual’s larger social context (Elder, 1994). Social roles unfold over adolescence and emerging adulthood; they “take place over an extended period of time and index temporal involvement in major institutions through schooling, paid employment, marriage, and parenthood” (Macmillan & Copher, 2005, p. 859). For youth involved in the juvenile justice system, they are playing the role of “delinquent” while involved with the system, but that role becomes primary while they are in a secure placement. The trajectories that people follow are shaped by the opportunities and societal constraints prevalent in peoples’ societies and cultures (Bynner, 2005; Elder, 1998; Shanahan, 2000). Examining patterns of institutional placement among adolescents over time can be informative about how the juvenile justice system may play a role in the opportunities or constraints some youth experience, and how institutional placement may be related to later outcomes.

The transition to adulthood within the current historical context has become more prolonged and heterogeneous compared with previous generations, and arguably more challenging (Arnett, 2006; Fussell & Furstenberg, 2005; Stanger-Ross, Collins, & Stern, 2005). Increasing heterogeneity has occurred along with the broad shift from a manufacturing to service-oriented economy where jobs are lower paying, less stable, and coming with fewer opportunities for career growth (Fussell & Furstenberg, 2005). At the same time, housing and health care costs have risen (Auerbach & Kellermann, 2011; Stanger-Ross et al., 2005). Individuals require more time after high school to explore possibilities and to establish themselves financially. The acquisition of human capital has become a central task of late adolescence and emerging adulthood. Thus, advantages or disadvantages that may accumulate during childhood and adolescence may have stronger implications for an individual’s ability to make the transition into adulthood successfully (Lee, 2014). Consequently, juvenile justice interventions, especially prolonged institutional placement that limits a youth’s ability to acquire positive human capital, may have more far-reaching ramifications for a youth today than in previous generations.

Potential Difference in Institutional Placement

There are likely to be group differences in patterns of institutional placement, because juvenile justice decisions take into account factors unrelated to the youth’s reason for being involved in the system (Fader et al., 2001). These factors include characteristics external to the youth’s control, including fixed individual characteristics, such as gender and race/ethnicity, as well as family, neighborhoods, and other characteristics of the youth’s social contexts. For example, there are racial disparities in juvenile justice involvement, where youth of color are involved with the juvenile justice system at disproportionately higher rates (Rodriguez, 2013). In 2013, 68% of youth in residential placement were minority youth, and Black youth were detained almost 6 times and committed more than 4 times the rate of White youth (Hockenberry, 2016). In addition, females represent a growing proportion of youth entering the juvenile justice system (Sickmund & Puzzanchera, 2014) and comprised 14% of the youth in residential placements (Hockenberry, 2016). The juvenile justice system appears to respond differently to males and females, with some evidence that males experience more severe sanctions than females (Espinosa, Sorensen, & Lopez, 2013). Males tend to stay in residential facilities longer, and minority youth were detained longer than White youth (Hockenberry, 2016). We would expect to see patterns of cumulatively higher rates of institutional placement among males and youth of color.

Child functioning is likely to be related to a youth’s level of involvement with the juvenile justice system. Mental health need and past traumatic experiences have been found to be associated with more severe juvenile justice sanctions (Espinosa et al., 2013; White, 2016). Juvenile justice youth in detention report higher rates of mental health disorders (Abram, Teplin, McClelland, & Dulcan, 2003; Teplin, Abram, McClelland, Dulcan, & Mericle, 2002), while youth with unmet social service needs were more likely to be involved in the juvenile justice system than in other public systems (Hazen, Hough, Landsverk, & Wood, 2004; Maschi, Hatcher, Schwalbe, & Rosato, 2008). Youth with unmet service needs are likely to demonstrate poorer individual functioning, while youth who receive quality, needed services are more likely to reflect improved functioning. Thus, service receipt is also likely to be related to patterns of secure detention.

Risk assessments are used to classify youth into high, medium, and low levels of risk for reoffending and are correlated with recidivism among juvenile offenders on probation (Baglivio, 2009; Onifade, Davidson, Campbell, et al., 2008). However, there is evidence that specific patterns of risk may be more informative than cumulative risk (Onifade, Davidson, Livsey, et al., 2008). It is also likely that assessed risk levels, and some of the specific risk domains, would be associated with patterns of institutional placement.

Family functioning is also likely to be related to a youth’s experiences with the juvenile justice system (Fader et al., 2001; McCoy et al., 2012), as well as the youth’s neighborhood (Rodriguez, 2013). Child welfare involvement, a marker of poor family functioning, appears to be related to decisions about sanctions, more specifically related to a higher likely that the youth will receive probation or be placed in a detention center (Ryan, Herz, Hernandez, & Marshall, 2007). Fader et al. (2001) found that the juvenile court took a child welfare approach to first-time offender by considering extralegal factors such as child and family functioning, but shifted for repeat offenders to focus on crime control. This suggests that patterns of involvement are likely to shift over time. Thus, we would expect to see factors related to family functioning have a weakening association with secure detention, and more severe sanctions regardless of extralegal factors as an individual’s number of offenses increase.

Current Study

In the current study, we used 7 years of monthly data collected in the Pathways to Desistance study to identify patterns of secure institutional placement over time among serious adolescent offenders. The Pathways to Desistance study was designed with the goal of examining the process of desistance with attention to developmental processes, social context, and sanctions and interventions (Schubert et al., 2004). We focused on secure institutional placement, situations in which the juvenile or adult court required a time period in a secure facility that did not provide access to the community. Secure institutional placement is the most extreme juvenile justice intervention, and thus it is useful to develop a more nuanced understanding of this type of juvenile justice intervention. While we were interested in allowing small or unexpected group to emerge from the analyses, we identified several patterns that we hypothesized would emerge. While a majority of serious adolescent offenders have experienced time in placement, not all do, and so we expected to find a pattern that reflected minimal or no time in placement. Given that many youth return to placement an average 6 months later, there is likely to be a pattern where youth cycle in and out of placement. In addition, some youth (i.e., males and youth of color) experience longer stays, so there is likely to be a pattern of long periods of time in placement.

Next, we examined factors related to individual, child functioning, family functioning, school, and neighborhood correlational differences between trajectory groups. Finally, we identified predictors of group membership using a multivariate approach. We expected to see patterns of institutional placement related to a number of individual and social contextual factors, including gender, race/ethnicity, individual functioning, and family functioning. This study seeks to empirically define institutional pathways and to allow unanticipated pathways to emerge.

Method

Data

Baseline data for the Pathways to Desistance study were collected in 2000-2003 for 1,354 youth (654 in Maricopa County [Phoenix, Arizona] and 700 in Philadelphia County, Pennsylvania) who were between 14 and 18 years of age and had committed a serious offense and were found guilty. The majority of the participants are male (86.4%), and the sample is diverse (20.2% White, 41.4% Black, 33.5% Hispanic). The last interview was conducted in March 2010.

All youth were followed for 7 years, with interval periods of 6 months for the first 3 years, and then annual follow-ups until the seventh year when participants were 21 to 25 years old. Over 90% of the sample was interviewed at each follow-up interview. A subsample of youth who experienced a residential placement were also interviewed shortly before or after their release to provide information about their experiences and services during the recent institutional stay. The data used for the current analyses were downloaded from the Interuniversity Consortium of Political and Social Research (Mulvey, 2013, 2017), with the addition of risk/need variables supplied by the authors generating these scores for a prior publication (Mulvey et al., 2016). In addition to measures at each of the interviews, the Pathways to Desistance study also includes monthly calendar data over 87 months about time in secure placement, which enabled us to estimate patterns of secure placement.

Measures

Secure Institutional Placement

The measure of the proportion of monthly time the youth spent in a secure facility with no community access was taken from the monthly calendar data set. This variable could have values from 0 (no time spent in secure placement) to 1 (the whole time spent in secure placement) or any range of values in between. These values over repeated observations were used to determine trajectory group memberships.

Individual Demographic Characteristics

Individual demographic characteristics, age, gender, and race/ethnicity, were taken from the baseline interview. Both gender and race/ethnicity have been previously found to be correlated with different experiences of institutional placement, in that males and youth of color experienced higher rates of placement or longer stays (Hockenberry, 2016). 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 with older adolescents ages 17 to 19 (= 0). This categorization aggregated the small numbers of younger and older offenders with their closest, and larger, age cohort. This provided a general test of the effect of age, one that is more operationally relevant to disposition decision making about whether to impose placement on young, modal, or older adolescent offenders. 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 the Charge and Prior Petitions

Information on the category and grade of the most serious charge, based on official records, were also examined. Public safety concerns would suggest that more serious offenses would be associated with more time in placement. There were six categories of charges: person, property, weapons, drug, sex, and other. The most serious charge, which was used to determine study eligibility, was based on whether the youth was charged with a felony or misdemeanor, where the most serious offenses had the lowest values. We included a measure of the number of petitions including the current petition, also based on official records. The number of petitions ranged from 1 to 15.

Mental and Behavioral Health

Measures of mental and behavioral health were taken from the baseline interviews. Youth involved in the juvenile justice system report high levels of mental health and unmet service needs, so understanding whether mental and behavioral health differentiates patterns of institutional placement has implications for service delivery. The Brief Symptom Inventory (BSI; Derogatis & Melisaratos, 1983) Global Severity Index is the mean of the subscale scores, which consists of somatization (α = .813), obsessive-compulsive (α = .798), interpersonal sensitivity (α = .661), depression (α = .805), anxiety (α = .779), hostility (α = .746), phobic anxiety (α = .685), paranoid ideation (α = .685), and psychoticism (α = .638). Two additional 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). Finally, we included a measure of the consequences of substance use, which was a modified version of the Substance Use/Abuse Inventory (Chassin, Rogosch, & Barrera, 1991). The lifetime total substance use consequences scale had values that ranged from 0 to 16.

School and Intelligence

We had several measures related to school and academics from the baseline interview. Juvenile justice youth may struggle in school, reporting high levels of special education needs and low academic achievement (Leone & Weinberg, 2012). In addition, youth in the juvenile justice system who experience exclusionary disciplinary measures, such as school expulsions, arguably may be caught in the school-to-prison pipeline (Wald & Losen, 2003). A dichotomous variable indicated whether the youth was enrolled at school (= 1), and a second dichotomous variable indicated whether the youth had ever been expelled from school (= 1). A measure of intelligence, from the baseline interview, was based on the Wechsler Abbreviated Scale of Intelligence (WASI; Wechsler, 1999) score, which is based on two subtests, vocabulary and matrix reasoning with values ranging from 55 to 128.

Family and Neighborhood Context

There were two measures of the youth’s family context taken from the baseline interview, as family characteristics have been found to be a factor in dispositional decisions (Fader et al., 2001). The measure of parental socioeconomic status (SES), parental index of social position, was computed using the Hollingshead (1975) formula, based on both parent education and occupation. Values ranged from 11 to 77. There was also a dichotomous variable that indicated whether the youth was from a single-parent household (= 1).

There was one total measure from the Neighborhood Conditions Measure (Sampson & Raudenbush, 1999) included in the youth’s baseline interview. The Neighborhood Conditions Measure is a scale that includes ratings of both social and physical conditions in the neighborhood. This measure was included because neighborhood conditions appear to be taken into consideration by court officials (Rodriguez, 2013). The scale of physical disorder is comprised of 12 questions about the physical disorder in the neighborhood, such as graffiti or cigarettes on the street. The scale of social disorder includes nine questions about social disorder in the neighborhood, such as whether adults fight or there are people using needles or syringes. The total scale (α = .94) was the calculated mean of the 21 items from both scales, which took on values between 1 and 4 where higher values indicated greater disorder.

Control Variables

Site was coded as a dichotomous variable, where Philadelphia County = 1 and Maricopa County = 0. Year was included to take into account potential variation or trends in sanctions. If the youth’s baseline interview was in either 2000 or 2001, the year variable was coded as 1, and if the baseline interview was in either 2002 or 2003, the variable was coded as 0.

Risk Scales

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 constructed risk scores derived from information collected in the Pathways interview mirrored, as much as possible, the Youth Level of Service/Case Management Inventory (YLS; CMI). Risk scores predict offending behavior, and thus we would expect to be correlated with patterns of institutional placement. In addition to a total risk score, the subscales covering eight domains were also examined: offense history, family characteristics/parenting, education/employment, peer relations, substance abuse, leisure/recreation, personality/behavior, and attitudes/orientation. For details on how the scales were created, please see Mulvey et al. (2016).

Analytic Approach

We used GBTM to identify patterns of institutional placement. GBTM assumes clusters or groupings of developmental trajectories that may reflect distinct etiologies, in contrast to latent growth analysis, which assumes a common process of growth or decline with individual variability (Nagin, 2005). GBTM uses maximum likelihood estimation to jointly estimate the shape of trajectories and the proportion of the sample in each trajectory, and is useful for identifying rare or unexpected trajectories (George, 2009). Group-based methods focus on identifying different trajectory shapes, and on examining how groups differ and identifying factors that distinguish group membership (Nagin, 2005). Juvenile justice intervention is likely to cluster in that individuals from different subgroups would be more or less likely to experience certain types of sanctions, rather than having sanctions continuously distributed throughout the population.

We estimated models based on study time, where month 1 represented the first month the youth was in the study. We were interested in the process that unfolds when an adolescent becomes involved in the juvenile justice system, and thus the start of the trajectory is when the youth has already been charged and found guilty by the court for a serious offense. Using the percent of time each month an individual was in secure institutional placement over the course of 87 months, we used GBTM to estimate trajectories of institutional placement. We used fit statistics, the Bayesian information criterion (BIC) and Akaike information criterion (AIC), to evaluate model fit. We also took into account model parsimony and what made substantive sense to select the best solution (Nagin, 2005). Once we selected the best solution, we examined other indicators to evaluate the suitability of the selected solution. These other model fit 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). In addition, we report the odds of correct classification (OCC) for each group, which compares the maximum probability classification rule to the random assignment of individuals, where an OCC greater than 5 for all groups is considered a good model (Nagin, 2005).

Once we identified the best solution, we tested bivariate associations between the trajectory groups and demographics, mental health indicators, offense history, site, year, school, family, neighborhood characteristics, and risk scores. We then proceeded to estimate group-based trajectory models that included predictors of trajectory group membership by adding individual variables to the models. These multivariate models that predicted trajectory group membership were multinomial logistic regression models, with the trajectory groups as the outcome variable. Given that many of the variables included in the models were likely related, we estimated two versions of the models predicting group membership because of the possibility of multicollinearity—one with the set of risk scales, and the other with the original measures. In the final models, there were no indications that multicollinearity might be a problem: None of the standard errors were particularly high, and none of the bivariate correlations were close to the 0.9 typically considered problematic. In our final models, we report key contrasts between successful and unsuccessful trajectory groups.

Results

The model fit statistics (BIC, AIC, and log likelihood [LL]) are presented in Table 1. The fit statistics steadily decreased in magnitude with the addition of groups until Group 4, and then started to increase in magnitude, suggesting the 4-group solution provided the best fit. The group probability and the proportion of individuals assigned to the group were comparable, and the average posterior probabilities were over .95 for all four groups, much higher than the 0.7 criteria considered to indicate good fit. The average posterior probabilities ranged from a low of .333 in Group 1, but were closer to .5 for the other three groups, to a high of 1 for all four groups. The OCC ranged from 49 for Group 1 to 228 for Group 4, much higher than the suggested 5.0 criteria.

Table 1:

Model Fit Statistics

Number of groups	BIC (n = 82,876)		BIC (n = 1,093)		AIC	LL
1 group	–67,355.6		–67,344.8		−67,332.3	−67,327.3
2 groups	–54,336.8		–54,315.2		−54,290.2	−54,280.2
3 groups	–51,233.6		–51,201.1		−51,163.7	−51,148.7
4 groups	–49,302.9		–49,259.6		−49,209.7	−49,189.7
5 groups	–50,577.5		–50,523.4		−50,461.0	−50,436.0
Number of groups	Group name	n	π	P _j	Ave PP (SD) [range]	OCC
Group 1	Steady community	380	.333	.342	.961 (.148) [.333, 1]	49.356
Group 2	Varying time	250	.228	.225	.969 (.093) [.528, 1]	105.839
Group 3	Declining time	271	.245	.244	.971 (.092) [.487, 1]	103.182
Group 4	Steady placement	209	.193	.188	.982 (.069) [.538, 1]	228.116

Note. BIC = Bayesian information criterion; AIC = Akaike information criterion; LL = log likelihood; π = estimated probability of group membership; P_j = proportion assigned to group; Ave PP = average posterior probability; OCC = odds of correct classification.

The final four trajectory groups, depicted in Figure 1, were estimated with a cubic functional form. We named our groups: steady in the community, varying time in placement, declining time in placement, and steady high in placement. The first trajectory, steady in the community, was the largest with 34.2% of the sample assigned to this group. These youth were likely to remain in the community throughout the study period. The fourth trajectory, steady high in placement, was the smallest consisting of 18.8% of the sample, and reflected consistently high rates of experiencing secure placement. The two remaining groups between the two extremes represented approximately half of the sample. The declining time in placement trajectory consisted of about one fourth of the sample (24.4%), and spent less and less time in secure placement over the course of the first 4 years, and then remain in the community for the last 3 years of the study. The varying time in placement trajectory had 22.5% of the sample assigned to this group and appeared to experience “churn” where they exit from and then return to secure placement. The steady in the community and declining time in placement can be considered system successes, because both groups end the study period spending all their time in the community, while the varying time and steady high in community groups spend a considerable amount of the study period in institutional placement.

Figure 1:

Four-Group Solution—Trajectories of Secure Institutional Placement

Descriptive statistics and bivariate associations are presented in Table 2. Overall, there were group differences by individual factors, school, family, neighborhood, and offense and risk scores. The steady in the community trajectory had the highest percentage of females, followed by the declining time in placement trajectory, with the remaining two trajectories reporting almost no females. The steady in community group reported the highest percentage of White and Hispanic youth and only 30% Black youth compared with the remaining groups that reported at least 50% Black youth. There were differences by study site, with a higher percentage of youth in Arizona classified in the steady in community group.

Table 2:

Descriptive Statistics and Bivariate Relationships Between Trajectory Groups and Individual, Family, and Neighborhood Factors

	Overall			Steady in community		Declining time in placement		Varying time in placement		Steady high in placement
	(N = 1,110)			(n = 380)		(n = 271)		(n = 250)		(n = 209)
Measures	n	%	χ²	n	%	n	%	n	%	n	%
Male	941	.85		268	.71	226	.83	239	.96	208	1.00
Female	169	.15	118.08***	112	.29	45	.17	11	.04	1	.00
Age			5.48
14-15	388	.35		124	.33	99	.37	82	.33	83	.40
16	339	.31		118	.31	75	.28	83	.33	63	.30
17-19	383	.35		138	.36	97	.36	85	.34	63	.30
White (ref)	233	.21	53.77***	102	.27	55	.20	43	.17	33	.16
Black	495	.45		115	.30	142	.52	133	.53	105	.50
Hispanic	335	.30		140	.37	65	.24	69	.28	61	.29
Race, Other	47	.04		23	.06	9	.03	5	.02	10	.05
Philadelphia County	649	.58	72.07***	159	.42	198	.73	160	.64	132	.63
Maricopa County	461	.42		221	.58	73	.27	90	.36	77	.37
Year
2000/2001	669	.60	5.25	246	.65	158	.58	141	.56	124	.59
2002/2003	441	.40		134	.35	113	.42	109	.44	85	.41
Category of charge			38.10**
Person	400	.36		157	.42	86	.32	91	.37	66	.32
Property	302	.27		96	.25	71	.26	73	.29	62	.30
Weapons	110	.10		28	.07	34	.13	23	.09	25	.12
Drug	203	.18		79	.21	49	.18	41	.16	34	.16
Other	48	.04		10	.03	10	.04	16	.06	12	.06
Sex	42	.04		7	.02	21	.08	5	.02	9	.04
	Overall			Steady in community		Declining time in placement		Varying time in placement		Steady high in placement
	(N = 1,110)			(n = 380)		(n = 271)		(n = 250)		(n = 209)
Measures	M	SD	F-statistic	M	SD	M	SD	M	SD	M	SD
Grade of charge (1-9)	4.53	2.14	8.40***	4.96	1.98	4.20	2.34	4.46	2.09	4.26	2.11
Prior petitions	3.06	2.09	54.20***	2.11	1.50	3.26	1.98	3.40	2.07	4.11	2.43
Total risk score	8.59	4.98	19.20***	7.28	4.68	8.43	4.72	9.28	4.82	10.37	5.26
Offenses risk score	1.17	1.07	40.36***	0.73	.91	1.27	1.07	1.33	1.03	1.65	1.11
Family risk score	0.66	0.88	1.13	0.60	0.86	0.69	0.89	0.64	0.87	0.74	0.90
Education risk score	1.72	0.88	2.06	1.64	0.91	1.70	0.90	1.76	0.84	1.82	0.86
Peer relationships risk score	0.68	0.94	10.48***	0.51	0.86	0.63	0.90	0.79	0.98	0.93	1.03
Substance abuse risk score	2.08	1.54	4.72**	1.91	1.50	1.99	1.45	2.22	1.56	2.36	1.67
Leisure risk score	0.31	0.46	2.29	0.26	0.44	0.33	0.47	0.33	0.47	0.35	0.48
Personality risk score	1.11	1.45	5.47**	0.91	1.22	1.05	1.43	1.24	1.53	1.38	1.68
Attitudes risk score	0.87	1.19	5.97***	0.73	1.13	0.78	1.15	0.95	1.15	1.14	1.32
BSI—Global Index	0.54	0.51	0.44	0.53	0.53	0.57	0.55	0.52	0.49	0.52	0.45
PTSD symptoms	1.52	3.96	1.60	1.38	3.77	1.93	4.54	1.23	3.52	1.59	3.96
MDD symptoms	0.84	2.37	3.14*	1.13	2.75	0.77	2.26	0.70	2.11	0.57	1.99
Lifetime substance use consequence	3.95	4.24	3.76*	3.66	4.27	3.51	4.04	4.47	4.32	4.43	4.26
Parent index of social position	51.23	11.87	1.90	50.39	11.94	50.70	11.73	52.23	11.63	52.22	12.11
Single-parent household	0.34	0.47	10.07*	105	0.28	95	0.35	97	0.39	76	0.36
WASI Full-Scale IQ	84.00	13.15	12.91***	87.09	13.32	83.12	12.53	83.14	13.50	80.52	12.08
Enrolled in school	0.75	0.44	5.76	292	0.77	209	0.77	183	0.73	144	0.69
Ever expelled	0.39	0.49	21.24***	117	0.31	109	0.40	100	0.40	104	0.50
Neighborhood conditions	2.35	0.75	14.77***	2.16	0.74	2.39	0.72	2.44	0.73	2.54	0.76

Note. BSI = Brief Symptom Inventory; PTSD = posttraumatic stress disorder; MDD = major depressive disorder; WASI = Wechsler Abbreviated Scale of Intelligence.

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

There were statistically significant differences by individual factors: IQ score, MDD symptoms, and lifetime substance use consequences. For the IQ scores and MDD symptoms, the steady in community group reported the highest average scores and steady high in placement reported the lowest average scores. For the lifetime substance use consequences, the steady high in placement and varying time in placement groups reported the higher average scores.

Statistically significant variations also occurred in family, school, and neighborhood characteristics. The steady in community group reported the lowest percentage of single-parent households, the lowest percentage of having been expelled, and lowest average neighborhood disorganization scores. In contrast, the steady high in placement reported the highest percentage of youth who had been expelled and the highest average neighborhood disorganization scores. Varying time and declining groups reported average scores in between the two extremes.

Offense history also varied significantly within the trajectories. The steady in community reported the lowest average number of prior petitions, while steady high in placement reported the highest average number of prior petitions. The steady in community group also reported the highest percentage of person and drug charges, while steady high in placement and declining time in placement reported higher percentages of weapons and sex charges. The steady in community group also reported the highest average charge (least severe), followed by varying time, steady high in placement, and finally declining time in placement groups.

And, statistically significant differences occurred in the average overall risk scores at baseline, as well as the average scores for several subscales: offenses, peer relationships, substance abuse, personality, and attitudes. The pattern remained consistent across scales: steady in the community reported the lowest average risk scores, followed by declining time in placement, varying time in placement, and steady high in placement reporting the highest average risk scores.

The multivariate models allowed us to identify how specific groups differed while controlling for other variables, and are presented in Table 3. The comparison group for the multinomial model was the steady in the community trajectory for the first three columns and declining time for the last two columns, because both can be considered system successes. Adolescents from Philadelphia were more likely to be in the declining time (odds ratio [OR] = 4.95, p < .001), varying time (OR = 2.36, p = .005), and steady placement (OR = 2.66, p = .003) than steady in the community group. Males were also more likely to be in the declining time (OR = 2.56, p < .001), varying time (OR = 14.15, p < .001), and steady placement (OR = 100.48, p < .001) than steady in the community group. Finally, youth who were 14 or 15 years old (in comparison with youth who were 17-19) were also more likely to be in the declining time (OR = 2.05, p = .004), varying time (OR = 2.14, p = .004), and steady placement (OR = 3.63, p < .001) than steady in the community group. Youth who were Black (in comparison with youth who were White) were more likely to be assigned to the varying time (OR = 2.23, p = .017) and steady placement (OR = 2.23, p = .030) than steady in the community group. Similarly, youth who were 16 were more likely to be assigned to the varying time (OR = 1.68, p = .044) and steady placement (OR = 2.29, p = .004) than steady in the community group.

Table 3:

Multinomial Logistic Regression Predicting Group Membership

	Declining vs. steady community		Varying vs. steady community		Steady placement vs. steady community		Varying vs. declining		Steady placement vs. declining
Measures	B	SE	B	SE	B	SE	B	SE	B	SE
Constant	−3.53	0.69***	−6.61	0.81***	−10.12	1.30***	−3.08	0.78***	−6.59	1.26***
Philadelphia County	1.60	0.29***	0.86	0.31**	0.98	0.33**	−0.73	0.29*	−0.61	0.31*
Maricopa County (ref)
Year 2000 or 2001	−0.56	0.21**	−0.60	0.22**	−0.54	0.24*	−0.04	0.20	0.03	0.21
Year 2002 or 2003 (ref)
Female (ref)
Male	0.94	0.25***	2.65	0.39***	4.61	1.03***	1.71	0.40***	3.67	1.02***
White (ref)
Black	0.09	0.30	0.80	0.33*	0.80	0.37*	0.71	0.31*	0.71	0.34*
Hispanic	−0.43	0.28	−0.22	0.30	−0.07	0.34	0.21	0.31	0.36	0.33
Race, Other	−0.40	0.54	−0.45	0.59	0.45	0.58	−0.05	0.62	0.85	0.59
Ages 14 or 15	0.72	0.25**	0.76	0.26**	1.29	0.29***	0.04	0.25	0.57	0.27*
Age 16	0.34	0.25	0.52	0.26*	0.83	0.29**	0.19	0.25	0.50	0.27
Ages 17-19 (ref)
Prior petitions	0.48	0.06***	0.47	0.07***	0.61	0.07***	−0.01	0.05	0.14	0.05**
Grade of charge (1-9)	−0.10	0.05*	−0.08	0.05	−0.11	0.05*	0.02	0.04	−0.01	0.05
BSI—Global Index	0.38	0.20	0.07	0.23	0.27	0.24	−0.31	0.21	−0.11	0.22
Substance use consequence	0.05	0.03	0.13	0.03***	0.09	0.03**	0.08	0.03**	0.04	0.03
Parent index of social position	0.01	0.01	0.02	0.01*	0.02	0.01*	0.01	0.01	0.01	0.01
Single-parent household	0.14	0.22	0.49	0.22*	0.23	0.25	0.35	0.21	0.09	0.22
Ever expelled from school	0.36	0.21	0.20	0.22	0.40	0.24	−0.16	0.20	0.04	0.22
Neighborhood conditions	−0.01	0.15	0.21	0.16	0.47	0.17**	0.21	0.15	0.48	0.16**

Note. BIC = −45,684.38 (N = 76,200), BIC = −45,537.28 (n = 1,007), AIC = −45,370.18, LL = −45,302.18. BIC = Bayesian information criterion; AIC = Akaike information criterion; LL = log likelihood.

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

The coefficient for prior petitions was positive and significant, indicating that more prior petitions was associated with a higher likelihood of being in the declining time (OR = 1.62, p < .001), varying time (OR = 1.60, p < .001), and steady high in placement (OR = 1.84, p < .001) groups compared with the steady in community group. Youth with less serious charges were less likely to be in the declining time (OR = 0.90, p = .026) and steady high in placement (OR = 0.90, p =.039) groups than in the steady in community group. Youth who reported more consequences from lifetime substance use were more likely to be in the varying time (OR = 1.14, p < .001) and steady high in placement (OR = 1.09, p = .006) groups than the steady in community group. In addition, youth from higher SES families were more likely to be in the varying time (OR = 1.02, p = .021) and steady high in placement (OR = 1.02, p = .043) groups than the steady in community group, and youth from single-parent households were more likely to be in the varying time (OR = 1.63, p = .030) group than the steady in community group.

The last two columns of Table 3 show that youth from Philadelphia were associated with a lower likelihood of being in the varying time (OR = 0.28, p = .011) and steady placement (OR = 0.54, p = .046) versus declining time group. Male youth were more likely to be in the varying time (OR = 5.53, p <.001) and steady placement (OR = 39.25, p < .001) versus declining time group. Youth who were Black were also more likely to be in the varying time (OR = 2.03, p = .023) and steady placement (OR = 2.03, p = .035) versus declining group. Youth with more substance use consequences were more likely to be in the varying time (OR = 1.08, p = .004) versus the declining time group, while youth with more prior petitions (OR = 1.15, p = .005) and from more disorganized neighborhoods (OR = 1.62, p = .003) were more likely to be in the steady placement than declining time group.

Table 4 presents results from the multinomial logistic regression models using study controls, demographics, and baseline average risk scores, because the risk scores overlap with many of the variables included separately in the models reported in Table 3. The results for the demographics, study site, and study year are similar to the models presented in Table 3. In terms of risk scores, holding all other variables constant, the risk scores for offenses at baseline were significantly associated with a higher likelihood of being in the varying time (OR = 1.77, p < .001), declining time (OR = 1.87, p < .001), and steady high in placement (OR = 2.24, p < .001) groups compared with the steady in community group. In addition, higher scores on peer relationships at baseline were also associated with a higher likelihood of being in the steady high in placement (OR = 1.25, p = .028) versus the steady in community group. The contrasts between both varying and steady placement versus the declining time group were not differentiated by risk scores, but rather study site, gender, and race.

Table 4:

Multinomial Logistic Regression Predicting Group Membership (Risk Scores)

	Declining vs. steady community		Varying vs. steady community		Steady placement vs. steady community		Varying vs. declining		Steady placement vs. declining
Measures	B	SE	B	SE	B	SE	B	SE	B	SE
Constant	−2.56	0.43***	−4.39	0.55***	−7.63	1.13***	−1.83	0.58**	−5.07	1.14***
Philadelphia County	1.76	0.25***	1.12	0.27***	1.24	0.29***	−0.64	0.27*	−0.52	0.28
Maricopa County (ref)
Year 2000 or 2001	−0.55	0.20**	−0.65	0.21**	−0.53	0.23*	−0.10	0.20	0.02	0.22
Year 2002 or 2003 (ref)
Female (ref)
Male	0.84	0.24*	2.57	0.38***	4.49	1.02***	1.74	0.39***	3.66	1.02***
White (ref)
Black	0.08	0.29	0.84	0.32**	0.92	0.35**	0.76	0.30*	0.83	0.33*
Hispanic	−0.29	0.26	0.03	0.28	0.14	0.31	0.32	0.29	0.43	0.32
Race, Other	−0.42	0.54	−0.22	0.57	0.58	0.55	0.20	0.64	0.99	0.60
Ages 14 or 15	0.60	0.24*	0.40	0.25	0.98	0.28***	−0.20	0.24	0.39	0.27
Age 16	0.26	0.24	0.31	0.25	0.58	0.28*	0.04	0.25	0.32	0.28
Ages 17-19 (ref)
Risk scales at baseline
Offenses	0.62	0.11***	0.57	0.11***	0.81	0.12***	−0.06	0.10	0.18	0.11
Family	0.19	0.12	0.06	0.12	0.20	0.13	−0.13	0.12	0.01	0.12
Education	−0.05	0.11	0.01	0.11	0.03	0.13	0.05	0.12	0.08	0.13
Peer Relationships	0.18	0.13	0.25	0.13	0.30	0.14*	0.07	0.12	0.12	0.13
Substance Abuse	−0.02	0.08	0.11	0.08	0.08	0.09	0.12	0.08	0.09	0.08
Leisure	0.25	0.22	0.22	0.22	0.21	0.24	−0.03	0.21	−0.05	0.22
Personality	0.03	0.08	0.07	0.08	0.04	0.09	0.04	0.08	0.01	0.09
Attitudes	−0.10	0.10	−0.07	0.10	0.06	0.10	0.03	0.10	0.16	0.10

Note. BIC = −46,222.73 (N = 77,717), BIC = −46,075.50 (n = 1,023), AIC = −45,907.87, and LL = −45,839.87. BIC = Bayesian information criterion; AIC = Akaike information criterion; LL = log likelihood.

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

Discussion

In this study, we identified four patterns of secure institutional placement in the 7 years after an adolescent received a charge for a serious offense. Three of the four trajectory groups confirmed our hypothesized patterns. About one third (34.2%) of the adolescents spent most of their time in the community, and almost one fifth (18.8%) spent a majority of the 7 years in secure institutional placement. More than one fifth (22.5%) of the group spent varying amounts of time in institutional placement. The fourth group that emerged, consisting of about one fourth (24.4%) of the group, was not hypothesized, but spent declining amounts of time in secure institutional placement, spending almost all their time in the community during approximately the last half of the study period. This group appears to reflect an effective system intervention, where the youth spent some initial time in institutional placement, but were able to stay out of the system during the last few years of the study.

The picture that emerged in examining the four groups revealed that there were clear and consistent contrasts between the steady in community group and the steady high in placement group. The steady high in placement group, as expected, appears to be the most troubled group across multiple domains, with higher scores on lifetime substance use consequences, half reporting having ever been expelled from school, and with higher neighborhood disorganization scores. The steady in community group appears to be the least troubled group, reporting the lowest neighborhood disorganization scores, lowest rates of single-parent households, and lowest percentage of youth expelled from school. Yet this group is not without their own challenges, as almost one third of this group still reported that they had been expelled from school and reported the highest average MDD symptoms.

The two remaining groups appear to fall between the two extreme groups and are similar to each other in some ways, but at other times appear to be more similar to the extreme groups. The declining time group had more females and lower lifetime substance use consequences, and more similar average risk scores to the steady in community group. Interestingly, the declining time group reported the most serious charge, followed by the steady high in placement, varying time in institutional placement, and finally the steady in community group. Thus the declining time group, while initially committing the most serious offense, may be a group of youth who have been “reformed” throughout the study period—and certainly were system successes. At the same time, studies have shown that the seriousness of the charge may not be the strongest predictor of recidivism (Cottle, Lee, & Heilbrun, 2001).

These bivariate associations suggest that the juvenile justice system seems to be effectively sorting serious adolescent offenders. Not only are the youth with higher risk/needs being placed in institutional placements within the first 2 years of their charge (Mulvey et al., 2007), but this study shows that those who spend the most time in the system (i.e., steady high in placement) appear to be from the most disadvantaged contexts at baseline across multiple domains (families, schools, neighborhoods), while those who spend the least amount of time in the system (i.e., steady in community) appear to be from the least disadvantaged contexts. It is possible that these youth who are spending the most time in placement during the 7-year study period are being charged again. Still, this supports other studies that have found that concentrated disadvantage play a role in more serious sanctions in the juvenile justice system (Rodriguez, 2013). Youth with unmet social service needs often end up in the juvenile justice system (Maschi et al., 2008), perhaps indicating that other systems have not served these youth adequately. Yet the association between disadvantage in multiple contexts and involvement in the juvenile justice system is unclear and could reflect discrimination or that their needs require more intensive intervention. Still, this raises the question as to whether these adolescents are receiving the most beneficial services while in institutional placement, and whether that is the best environment for these youth to receive those services, especially if these youth will be transitioning back into those disadvantaged environments. Judges have been found to shift their approach to youth away from a treatment orientation toward a more punitive approach with subsequent charges (Fader et al., 2001). Yet prior studies have not found an association between more time in institutional care and positive outcomes (Loughran et al., 2009; Winokur et al., 2008). This suggests a process of accumulating disadvantage; these results also suggest that the juvenile justice system may have an opportunity to disrupt the accumulating disadvantage by providing more services in the community. Developing our understanding of how to best address these youth’s unmet needs warrants further study.

The multivariate models indicate that individual lifetime substance use consequences and family characteristics are robust in differentiating groups when controlling for other variables. Surprisingly, school expulsions and neighborhood disorder were not significant in the multivariate models. This may reflect that a youth’s neighborhood or school records may be more difficult for the judge to ascertain compared with the youth’s substance use or family characteristics, and thus play a larger role in their decisions. Alternatively, it is possible that school expulsion is correlated with family and neighborhood factors, so there is no difference by trajectory group when taking into account those multiple factors. While controlling for other variables, the average number of prior petitions differentiates the steady in community group from the other three groups, and the seriousness of the charge also differentiates the declining time and steady high in placement group from the steady in community group. However, not only is the contrast between steady in community and other groups informative, but the contrast between declining time and varying time groups is also informative. The steady in community and declining time in placement may be considered system successes, because both groups end the study period spending all their time in the community. The varying time and steady high in community end the study period spending a considerable amount of time in institutional placement. What differentiates the declining time and varying time groups were demographic characteristics (gender and race) and substance use consequences, but not offense history or seriousness of the charge. This suggests that by providing effective substance use services, juvenile justice system may be able to shift more serious adolescent offenders from the varying time to declining time trajectory group.

We also examined the association between risk scores at baseline and the four patterns of institutional placement. While risk factors change over time, and more current risk assessments are more strongly related to offending (Baglivio, Wolff, Piquero, Howell, & Greenwald, 2017; Mulvey et al., 2016), we found robust associations between the offenses scale at baseline and patterns of institutional placement. While bivariate associations were significant between several subscales (offenses, peer relationships, substance abuse, personality, and attitudes), when controlling for other variables, only the offense scale was significant and robust in differentiating the steady in community group from the other three groups. However, there is no such clear distinction between the declining time and both varying time and steady high in placement groups. While this may appear to suggest that the system is appropriately targeting youth at highest risk, there is evidence that a youth’s offense is not a good predictor of whether they will eventually persist or desist from crime (Sickmund & Puzzanchera, 2014). This suggests that care should be taken in using risk scores to inform dispositional decisions, because they are not necessarily related to later system success.

Limitations

These analyses only examined serious adolescent offenders from two locales, so these results cannot be generalized to the general adolescent population. Also, we could not distinguish from types of secure institutional placement, and whether the placement was before or after the youth’s disposition. In addition, we used GBTM, but the groups we identified are not necessarily fact—Rather, the distinct patterns we identified are only approximations that are useful in describing the phenomenon of institutional placement among our sample of serious adolescent offenders (Nagin, 2005). Thus, future research should continue to identify patterns of institutional placement, as well as other sanctions, with other samples. Finally, in this study we only examined baseline characteristics and did not take into account events that may have been occurring during the study period. For this study, we were interested in whether the predictors known about a serious adolescent offender entering the system are predictive of their longitudinal experiences in placement and can thus help inform the decision-making process. At the same time, examining what might be transpiring in the lives of the youth over time, including potential reoffending, new charges, and service receipt, may further illuminate differences between groups.

Implications and Conclusion

In spite of these limitations, these findings add to the discussion of how institutional placement may be related to positive youth behaviors. While prior studies provide static descriptions of experiences in institutional placement (Hockenberry, 2016; Mulvey et al., 2007), this study extends current knowledge by identifying longitudinal patterns of institutional placement. The four patterns of institutional placement we identified suggest that the juvenile justice system successfully filters youth in most need, in that youth who were consistently spending time in secure placement were from the most disadvantaged contexts—They were the most likely to have trouble in schools and live in disorganized neighborhoods. While juvenile justice intervention can be a window of opportunity to create the turning points discussed by Laub and Sampson (1993), that is not clearly reflected in altered trajectories. Prior studies suggest that more time in placement does not necessarily translate to better outcomes (Loughran et al., 2009; Schubert, Mulvey, & Pitzer, 2016). This suggests a reimagining of the services provided to juvenile justice youth, one that balances accountability and service delivery. Restorative justice, for example, might be an approach that could allow youth to remain in their communities while addressing difficulties in schools and investing in positive opportunities in their neighborhoods (van Wormer, 2003). Providing effective treatment for substance use services in the community may increase the likelihood these youth will be a system success (i.e., in the declining time rather than varying time group).

The youth from the varying time and steady high in placement groups are spending a good fraction of their time in secure placement, and these disruptions may challenge their ability to become independent and acquire the human capital necessary to participate positively in society. The difficulty here is figuring out what the counterfactual might be; what the outcomes might have looked like if these adolescents had spent less time in institutional care. Consequences from substance use was robust in differentiating groups, and the sole dynamic risk factor that differentiated the varying time and declining groups, suggesting that a focus on providing effective substance use treatment to youth involved in the juvenile justice system may increase the likelihood that youth who struggle will become system successes as they make the transition to adulthood.

This study identifies four patterns of institutional placement among serious adolescent offenders over the course of 7 years. This is an important preliminary step in developing a systematic understanding of the impact of juvenile justice intervention on the lives of youth. These patterns raise questions about what services these youth receive and their outcomes in terms of both delinquency (or crime, when they become adults) and becoming productive citizens. It will be important to understand whether any of these patterns of institutional placement are more or less related to a successful transition to a productive adulthood.

Footnotes

This research was supported by a grant from the National Institute of Justice.

ORCID iD

JoAnn S. Lee

JoAnn S. Lee, PhD, is an assistant 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 a professor of psychiatry and director of the Law and Psychiatry Program at the University of Pittsburgh School of Medicine, where he has been on the faculty since 1983. He 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 also works with practitioners and policy makers on the provision of services to individuals at risk for involvement in violence.

Carol A. Schubert, MPH, 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

Abram

K. M.

Teplin

L. A.

McClelland

G. M.

Dulcan

M. K.

(2003). Comorbid psychiatric disorders in youth in juvenile detention. Archives of General Psychiatry, 60, 1097-1108.

Arnett

J. J.

(2006). Emerging adulthood: Understanding the new way of coming of age. In Arnett

J. J.

Tanner

J. L.

(Eds.), Emerging adults in America: Coming of age in the 21st century (pp. 3-19). Washington, DC: American Psychological Association.

Auerbach

D. I.

Kellermann

A. L.

(2011). A decade of health care cost growth has wiped out real income gains for an average US family. Health Affairs, 30, 1630-1636. doi:10.1377/hlthaff.2011.0585

Baglivio

M. T.

(2009). The assessment of risk to recidivate among a juvenile offending population. Journal of Criminal Justice, 37, 596-607. doi:10.1016/j.jcrimjus.2009.09.008

Baglivio

M. T.

Wolff

K. T.

Piquero

A. R.

Howell

J. C.

Greenwald

M. A.

(2017). Risk assessment trajectories of youth during juvenile justice residential placement: Examining risk, promotive, and “buffer” scores. Criminal Justice and Behavior, 44, 360-394.

Bynner

(2005). Rethinking the youth phase of the life-course: The case for emerging adulthood? Journal of Youth Studies, 8, 367-384. doi:10.1080/13676260500431628

Chassin

Rogosch

Barrera

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

Cottle

C. C.

Lee

R. J.

Heilbrun

(2001). The prediction of criminal recidivism in juveniles: A meta-analysis. Criminal Justice and Behavior, 28, 367-394.

Derogatis

Melisaratos

(1983). The Brief Symptom Inventory: An introductory report. Psychological Medicine, 13, 595-605.

10.

Elder

G. H.

(1994). Time, human agency, and social change: Perspective on the life course. Social Psychology Quarterly, 57, 4-15.

11.

Elder

G. H.

(1998). The life course as developmental theory. Child Development, 69, 1-12. doi:10.1111/j.1467-8624.1998.tb06128.x

12.

Espinosa

E. M.

Sorensen

J. R.

Lopez

M. A.

(2013). Youth pathways to placement: The influence of gender, mental health need and trauma on confinement in the juvenile justice system. Journal of Youth and Adolescence, 42, 1824-1836. doi:10.1007/s10964-013-9981-x

13.

Fader

J. J.

Harris

P. W.

Jones

P. R.

Poulin

M. E.

(2001). Factors involved in decisions on commitment to delinquency programs for first-time juvenile offenders. Justice Quarterly, 18, 323-341. doi:10.1080/07418820100094921

14.

Fussell

Furstenberg

F. F.

Jr. (2005). The transition to adulthood during the twentieth century: Race, nativity, and gender. In Settersten

R. A.

Furstenberg

F. F.

Rumbaut

R. G.

(Eds.), On the frontier of adulthood: Theory, research, and public policy (pp. 29-75). Chicago, IL: University of Chicago Press.

15.

Gatti

Tremblay

R. E.

Vitaro

(2009). Iatrogenic effect of juvenile justice. Journal of Child Psychology and Psychiatry, 50, 991-998. doi:10.1111/j.1469-7610.2008.02057.x

16.

George

L. K.

(2009). Conceptualizing and measuring trajectories. In Elder

G. H.

Giele

J. Z.

(Eds.), The craft of life course research (pp. 163-186). New York, NY: Guilford.

17.

Hazen

A. L.

Hough

R. L.

Landsverk

J. A.

Wood

P. A.

(2004). Use of mental health services by youths in public sectors of care. Mental Health Services Research, 6, 213-226.

18.

Hockenberry

(2016). Juveniles in residential placement, 2013. Washington, DC: Office of Juvenile Justice and Delinquency Prevention, Office of Justice Programs, U.S. Department of Justice. Retrieved from https://www.ojjdp.gov/pubs/249507.pdf

19.

Hollingshead

A. B.

(1975). Four factor index of social status (Unpublished manuscript). New Haven, CT: Yale University.

20.

Laub

J. H.

Sampson

R. J.

(1993). Turning points in the life course: Why change matters to the study of crime. Criminology, 31, 301-325.

21.

Lee

J. S.

(2014). An institutional framework for the study of the transition to adulthood. Youth & Society, 46, 706-730.

22.

Leone

Weinberg

(2012). Addressing the unmet educational needs of children and youth in the juvenile justice and child welfare systems. Washington, DC: Center for Juvenile Justice Reform. Retrieved from https://cjjr.georgetown.edu/wp-content/uploads/2015/03/EducationalNeedsofChildrenandYouth_May2010.pdf

23.

Loughran

T. A.

Mulvey

E. P.

Schubert

C. A.

Fagan

Piquero

A. R.

Losoya

S. H.

(2009). Estimating a dose-response relationship between length of stay and future recidivism in serious juvenile offenders. Criminology, 47, 699-740. doi:10.1111/j.1745-9125.2009.00165.x

24.

Macmillan

Copher

(2005). Families in the life course: Interdependency of roles, role configurations, and pathways. Journal of Marriage and Family, 67, 858-879. doi:10.1111/j.1741-3737.2005.00180.x

25.

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, 1376-1385. doi:10.1016/j.childyouth.2008.04.006

26.

McCoy

Walker

J. T.

Rodney

H. E.

(2012). Predicting preadjudication detention decisions: An examination of family status and race. Journal of Ethnicity in Criminal Justice, 10, 87-107. doi:10.1080/15377938.2012.669652

27.

Mulvey

E. P.

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

28.

Mulvey

E. P.

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

29.

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, 518-544. doi:10.1016/j.childyouth.2006.10.006

30.

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. doi:10.1016/j.jcrimjus.2016.02.008

31.

Mulvey

E. P.

Steinberg

Fagan

Cauffman

Piquero

A. R.

Chassin

. . . Losoya

S. H.

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

32.

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, 453-475. doi:10.1017/S0954579410000179

33.

Nagin

D. S.

(2005). Group-based modeling of development. Cambridge, MA: Harvard University Press.

34.

Nguyen

Loughran

T. A.

Paternoster

Fagan

Piquero

A. R.

(2016). Institutional placement and illegal earnings: Examining the crime school hypothesis. Journal of Quantitative Criminology, 33, 207-235. doi:10.1007/s10940-016-9291-z

35.

Onifade

Davidson

Campbell

Turke

Malinowski

Turner

(2008). Predicting recidivism in probationers with the Youth Level of Service Case Management Inventory (YLS/CMI). Criminal Justice and Behavior, 35, 474-483. doi:10.1177/0093854807313427

36.

Onifade

Davidson

Livsey

Turke

Horton

Malinowski

. . . Wimberly

(2008). Risk assessment: Identifying patterns of risk in young offenders with the Youth Level of Service/Case Management Inventory. Journal of Criminal Justice, 36, 165-173. doi:10.1016/j.jcrimjus.2008.02.006

37.

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, 738-763.

38.

Rodriguez

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

39.

Ryan

J. P.

Herz

Hernandez

P. M.

Marshall

J. M.

(2007). Maltreatment and delinquency: Investigating child welfare bias in juvenile justice processing. Children and Youth Services Review, 29, 1035-1050. doi:10.1016/j.childyouth.2007.04.002

40.

Sampson

Raudenbush

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

41.

Schubert

C. A.

Mulvey

E. P.

Pitzer

(2016). Differentiating serious adolescent offenders who exit the justice system from those who do not. Criminology, 54, 56-85.

42.

Schubert

C. A.

Mulvey

E. P.

Steinberg

Cauffman

Losoya

S. H.

Hecker

. . . Knight

G. P.

(2004). Operational lessons from the Pathways to Desistance project. Youth Violence and Juvenile Justice, 2, 237-255. doi:10.1177/1541204004265875

43.

Shanahan

M. J.

(2000). Pathways to adulthood in changing societies: Variability and mechanisms in life course perspective. Annual Review of Sociology, 26, 667-692.

44.

Sickmund

Puzzanchera

(2014). Juvenile offenders and victims: 2014 National Report. Pittsburgh, PA: National Center for Juvenile Justice.

45.

Stanger-Ross

Collins

Stern

M. J.

(2005). Falling far from the tree: Transitions to adulthood and the social history of twentieth-century America. Social Science History, 29, 625-648.

46.

Teplin

L. A.

Abram

K. M.

McClelland

G. M.

Dulcan

M. K.

Mericle

A. A.

(2002). Psychiatric disorders in youth in juvenile detention. Archives of General Psychiatry, 59, 1133-1143.

47.

van Wormer

. (2003). Restorative justice: A model for social work practice with families. Families in Society, 84, 441-448.

48.

Wald

Losen

D. J.

(2003). Defining and redirecting a school-to-prison pipeline. New Directions for Youth Development, 2003(99), 9-15.

49.

Wechsler

(1999). Wechsler Abbreviated Scale of Intelligence. New York, NY: The Psychological Corporation.

50.

White

(2016). Incarcerating youth with mental health problems: A focus on the intersection of race, ethnicity, and mental illness. Youth Violence and Juvenile Justice, 14, 426-447. doi:10.1177/1541204015609965

51.

Winokur

K. P.

Smith

Bontrager

S. R.

Blankenship

J. L.

(2008). Juvenile recidivism and length of stay. Journal of Criminal Justice, 36, 126-137. doi:10.1016/j.jcrimjus.2008.02.001

52.

World Health Organization. (1990). Composite International Diagnostic Interview. Geneva, Switzerland: Author.