Children of Incarcerated Parents: Developmental Trajectories of Externalizing Behavior Across Adolescence

Abstract

Growth mixture modeling was used to identify distinct trajectories of externalizing behavior for youth (N = 647) across the period 10 to 16 years of age. Four trajectory classes were identified: Low-Stable, Mid-Increasing, Borderline-Stable, and Chronic-High. Relations of the identified trajectories with parental incarceration, parent–child relationships, trauma, and parenting as well as future substance use and criminality were then examined. Children of incarcerated parents were underrepresented in the Low-Stable trajectory and overrepresented in the Mid-Increasing group. However, nearly 60% of the children of incarcerated parents were best represented by the low-risk trajectory. The trajectory classes differed significantly on many of the preadolescent measures, such as parent–child relationships and trauma, as well as on adolescent delinquency, adult criminality, and substance use. The Mid-Increasing, Borderline-Stable, and Chronic-High trajectory groups showed significantly higher levels of early risk factors and problematic outcomes than the Low-Stable trajectory group. Implications for practice are discussed.

Keywords

parental incarceration children growth trajectories externalizing prevention

Over the past five decades, the number of incarcerated adults within the United States has increased dramatically, and the United States now leads the world in terms of both the largest prison population and the highest rate of imprisonment (Maruschak & Mumola, 2010; Walmsley, 2015). Because most incarcerated men and women are parents, this situation has had major impacts on the population of youth in the United States. By 2007, more than 1.7 million minor children (2.3% of all minors) had an incarcerated parent (Maruschak & Mumola, 2010). The number of children who have experienced parental incarceration over their lifetime is estimated at nearly 5 times this amount (Kjellstrand & Eddy, 2011a; Western & Wildeman, 2009).

Although the impact of this societal level of imprisonment on crime continues to be debated (e.g., King, Mauer, & Young, 2005), the impact on the children and families of incarcerated individuals is worrisome (e.g., Eddy & Poehlmann, 2010). Many families struggle with not only the loss of a parent but also increased financial insecurity and its aftermath, as well as the stigma of having an incarcerated family member. Published findings point to problematic outcomes that can arise for children, including mental health problems, delinquency, crime, and substance abuse (e.g., Geller, Garfinkel, Cooper, & Mincy, 2009; Kjellstrand & Eddy, 2011a, 2011b; Muftic & Smith, 2018; Murray & Farrington, 2005). Of special concern is the link between adolescent externalizing behavior and future substance abuse, delinquency, and crime (Geller et al., 2009; Kjellstrand & Eddy, 2011a, 2011b; Wildeman, 2009). A meta-analysis of the most rigorous research on children of incarcerated parents found that children with a history of parental incarceration were at a higher risk of displaying externalizing (i.e., covert and overt antisocial) behavior, pooled effect size of odds ratio (OR) = 1.4, p < .01, than children without incarcerated parents (Murray, Farrington, & Sekol, 2012). This risk persisted even when controlling for other established risks of such problems.

Notably, although many quantitative studies estimate an average negative effect of parental incarceration on child well-being, there appears to be great variability in how children of incarcerated parents fare (Johnson, Arditti, & McGregor, 2018; Murray et al., 2012; Turney, 2017; Turney & Wildeman, 2015). Focusing on the average effect of parental incarceration on child well-being ignores the heterogeneity of children, experiences, and contexts surrounding parental incarceration. Several qualitative studies also have found heterogeneous effects of parental incarceration on children’s behavior (Johnson & Easterling, 2015; Luther, 2015; Turanovic, Rodriguez, & Pratt, 2012). Clearly, more research is needed that advances our understanding of the differential impacts of parental incarceration across individuals, families, and contexts.

Growth mixture modeling (GMM; Muthén & Muthén, 2000) is a statistical technique that helps examine variation by using individual observed histories to classify individuals into specific homogeneous trajectory groups and to model heterogeneity over time (Muthén & Muthén, 2000). In terms of the variable of interest, individuals within a trajectory class exhibit more similar response patterns than individuals between trajectory classes. By modeling distinct trajectories, and investigating how these trajectories are associated with specific developmental influences, a better understanding can be gained on what factors promote or impede child development and for whom.

In the current study, we use GMM to identify and examine specific developmental trajectories of externalizing behaviors. Understanding different patterns of emergence and persistence of this disruptive behavior over time; the factors that minimize, maintain, or elevate problems; and how growth patterns in behavior are related to adult outcomes is critical in the development of appropriate prevention and intervention strategies. Although the examination of developmental pathways for externalizing behaviors is becoming increasingly prevalent, no study has examined how these pathways relate to children of incarcerated parents. Understanding these pathways may be key in preventing intergenerational cycles of disruptive and criminal behavior for corrections-involved families.

Models of the Development of Externalizing Behavior Problems

Developmental research has demonstrated that exposure to one risk factor, such as having an incarcerated parent, does not destine a child to a life of problems. Rather, child development is influenced by the cumulative effect and interactions of a variety of risk and protective factors at the individual, family, and community levels (e.g., Gutman, Sameroff, & Cole, 2003; Rutter et al., 1997). Furthermore, research suggests that there are a variety of developmental pathways, some of which are problematic and place a youth at risk of different types of problems in late adolescence and adulthood, and others that are not (Moffitt, 1993; Patterson, DeBaryshe, & Ramsey, 1989).

Theorists on the development of externalizing behavior have posited three general types of trajectories: an “early starter” or “life-course persistent” trajectory where children display a chronic pattern of externalizing behaviors starting early in childhood and continuing into adolescence and adulthood; a “late starter” or “adolescent-limited” trajectory characterized by increasing antisocial behavior across childhood into adolescence; and a low-risk, normative group who display little or no externalizing behavior over time (e.g., Moffitt, 1993; Patterson et al., 1989). These hypotheses have influenced a shift in research activity from variable-centered analyses focused on average trends toward person-centered approaches focused on individual developmental differences and influences.

Major advances in methodology in modeling developmental trajectories (e.g., Muthén & Muthén, 2000; Nagin, 1999) have led to an increase in research on longitudinal patterns of externalizing behaviors across childhood (e.g., Malti, Averdijk, Ribeaud, Rotenberg, & Eisner, 2013; Maughan, Pickles, Rowe, Costello, & Angold, 2000; Nagin & Tremblay, 1999; Nivard et al 2017; Schaeffer, Petras, Ialongo, Poduska, & Kellam, 2003). Whereas studies have focused on a variety of age spans and subpopulations, Nivard et al. (2017), Maughan et al. (2000), and Nagin & Tremblay (1999) focus specifically on the growth of externalizing through adolescence in population-based samples, also the focus of the current study. Nivard et al. (2017) used population-based data from the Avon Longitudinal Study of Parents and Children conducted in Bristol, England. Data from youth aged 7 to 15 years (N = 7,212) were analyzed to examine externalizing trajectories. Five unique growth trajectories were identified: two low externalizing trajectories (low representing 54.3% of the youth and very low representing 27.9%), which represented the majority (82.2%) of the youth; a high, decreasing trajectory (7.0% of the youth); an increasing trajectory (8.4% of youth); and finally, a high, stable trajectory (2.4% of youth). Maughan et al. (2000) used population-based data from the Great Smoky Mountains Study in rural southeastern United States to examine growth trajectories of youth aged 9 to 16 years (N = 1,419). Similar to Nivard et al. (2017), they identified a stable, low problem trajectory, which represented the majority of youth (accounting for 70%-80% across groups). They also identified a trajectory with a declining level of problems (accounting for 15%-34% of the youth), and a stable high problem trajectory (accounting for 3%-7% of the youth). They did not identify an increasing group. Finally, Nagin & Tremblay (1999) examined a sample of boys from low socioeconomic areas in Montreal, Quebec, Canada, from ages 6 to 15 years (N = 1,037). They identified four trajectories for aggressive behaviors: a chronic high (4%, youth who consistently display high levels of aggression), a high desister (28%, youth with initially high aggression that decline), a moderate desister (52%, youth with initially moderate levels of aggression that decline), and a low group (20%, youth who rarely display any aggression).

Across these studies, the types of trajectories are similar to those posited in the aforementioned theories, namely, a normative class of children with low levels of externalizing behavior across time, a smaller group of children with consistently high levels of externalizing behaviors, and a “late starter” or “adolescent-limited” group, which exhibits an initial increase of externalizing, which is often followed by a decrease in externalizing as the youth approach adulthood. Although the normative group tends to have a lower risk of future problematic outcomes, the high externalizing trajectory groups are typically related to such adverse outcomes as substance use, delinquency, and suicidal ideation during adolescence and early adulthood (e.g., Nagin & Tremblay, 1999; Petras et al., 2004; Schaeffer et al., 2003).

Theorists suggest that such factors as coercive parenting, low academic achievement, concentration problems, peer rejection, family adversity, and parental monitoring play key roles in the development of externalizing behaviors across the life course (Moffitt, 1993; Patterson et al., 1989). In turn, growth analyses have shown significant relations between these factors and outcomes. For example, Petras et al. (2004) examined boys’ development, and demonstrated that low academic achievement, poverty, poor parent monitoring, deviant peer association, and neighborhood criminality were correlated with higher levels of youth externalizing behaviors. Similarly, Maughan et al. (2000) found that youth with persistently high levels of externalizing behavior were more likely to have experienced family adversity, poor parenting, and parental criminality.

Goals

In the current study, we have three main goals. The first is to examine developmental trajectories of externalizing behaviors from late childhood through adolescence. We then investigate how distinct (and possibly divergent) trajectories are related to parental incarceration as well as hypothesized key developmental risk factors during this period. Finally, we examine how these trajectories are related to future problem behavior during young adulthood.

Based on previous theoretical and empirical work, we expect to identify at least three distinct trajectories, namely, a large group of youth who demonstrate low levels of externalizing behaviors across time, a small group of youth who demonstrate high levels of externalizing behaviors across time (“early starters”), and a small group of youth who experience an initial increase in externalizing behaviors followed by a decrease (“adolescent limited”). A fourth trajectory also might be identified: a small group of youth who experience an increase in externalizing behaviors across time (“late starters”). In addition, we anticipate that the trajectories that lead to displays of high externalizing behaviors in the later adolescent years will be associated with higher levels of risky activities during the early adult years than the low or declining externalizing trajectories. Finally, we expect to find children with incarcerated parents to be overrepresented on the higher risk trajectories and underrepresented on the lower risk trajectories by virtue of the increased family vulnerabilities and challenges present before and after a parent is incarcerated.

Method

Data from the longitudinal Linking Interests of Families and Teachers (LIFT) randomized controlled trial were used for the current study (see Eddy, Reid, & Fetrow, 2000; Reid, Eddy, Fetrow, & Stoolmiller, 1999). The trial, which began in 1991, was designed to examine the impact of the LIFT school-based multimodal preventive intervention within higher risk neighborhoods in a moderately sized metropolitan area in the Pacific Northwest. Neighborhood risk was defined solely by a high rate (top 50% in the local metro area) of police contact with juveniles due to suspected delinquent behavior.

Participants

The entire first- and fifth-grade classes of students and their families from 12 elementary schools were invited to participate in the study. The majority (85%) chose to participate. Participants reflected the demographics of the region at the time, with most being White and from the lower to middle socioeconomic classes. Data reported in this study were from assessments given when the youth were roughly 10, 12, 14, and 16 years old (fifth through 10th grades). Of the 671 participants originally recruited for the study, 16 first graders were no longer involved in the study by fifth grade. Analyses here are based on the remaining 655 youth (51% girls). Based on an earlier analysis completed on the LIFT sample (Eddy, 2003), missing data appeared to be missing at random (MAR).

Measures

Externalizing Behaviors

Using the Child Behavior Checklist (CBCL; Achenbach, 1991), mothers, fathers, and teachers rated the target child on 30 overt and covert antisocial behaviors (e.g., argues a lot, physically attacks people) using a 3-point scale (0 = not true, 1 = somewhat/sometimes true, 2 = very or often true). Cronbach’s alphas for the externalizing subscales at the various time points were comparable with those in the original normative sample (i.e., .80-.90). Scores of the raters showed high levels of correspondence and were combined to create an observed estimate of an average “externalizing” construct score, as described and employed in Patterson (1982) and Patterson, Reid, and Dishion (1992; see also Dishion & Snyder, 2016). Mean T scores, computed from CBCL national normative data, were used in the analyses. T scores that are less than 60 are considered in the normal range. Scores from 60 to 63 represent borderline scores, whereas scores greater than 63 are in the clinical range (Achenbach, 1991).

Parental Incarceration

Whether or not a child had experienced parental incarceration was based on information from two sources: official records from local correctional departments and youth reports when participants were aged 20 to 25 years old. A dichotomous variable was created to indicate whether a child’s parent had been in jail or prison for at least 1 day at any time when the child was 10 years old or younger. If either source indicated that a parent was incarcerated, the parent was coded as “incarcerated.” Both the definition of incarceration and the age range of interest to have experienced incarceration were chosen to replicate the parameters employed in Murray and Farrington (2005). Of the 655 boys and girls in the LIFT sample, 10.2% (n = 67) had experienced parental incarceration by age 10 years.

Parent–Child Relationship

The parent–child relationship was assessed through the Parent Interview (Oregon Social Learning Center, 1990) when the child was 10 years old. It was based on two 5-point Likert-type scaled questions focusing on how well the parent got along with the child (from 1 = not well to 5 = very well), and how enjoyable were the activities with the child (1 = not enjoyable to 5 = very enjoyable). Cronbach’s alpha for this scale was .62.

Inconsistent Discipline

Inconsistent discipline was assessed using the Parent Interview (Oregon Social Learning Center, 1990) when the child was 10 years old. This scale comprises 11 5-point Likert-type scaled items (1 = never to 5 = always) in which parents report their use of different discipline behaviors (e.g., How often does a child get away with things you feel the child should have been punished for? If you have asked the child to do something, how often do you give up trying to get him or her to do it?). The final score is a mean response of all the items with higher numbers reflecting less consistent discipline. Cronbach’s alpha was .75.

Harsh Disciplinary Practices

Harsh disciplinary practices were assessed through the Parent Interview (Oregon Social Learning Center, 1990) when the child was 10 years old. This summative score is based on the number of harsh disciplinary practices (e.g., yell, slap, spank) that parents suggest using for specific scenarios (e.g., child argued or talked back to a parent, hit another child, stole from a store). For each item, the parents can report up to two possible discipline techniques they might use. A higher score indicates harsher disciplinary practices. Cronbach’s alphas were not calculated because the score was a sum of items that were not necessarily correlated with one another.

Parent Depression

Each parent’s level of depression was assessed using the Center for Epidemiological Studies–Depression Scale (CES-D scale) when the child was 10 years old. The CES-D scale (Radloff, 1977) is a self-reported, 20-item measure, which focuses on feelings and symptoms of depression (e.g., bothered by things, felt like life was a failure). Cronbach’s alpha for this scale was .85.

Social Economic Status (SES)

Hollingshead’s (1975) four-factor index of social status was used to calculate each parent’s SES based on the individual parent’s education and occupational level when the child was 10 years old. For two parent families, SES was a mean of the parent’s individual SES.

Academic Achievement

Academic achievement when the child was 10 years old was based on the mother’s rating of the student’s academic achievement in terms of reading/English, writing, math, and spelling at each time point (1 = failing to 4 = above average). If three of the four items were present, the mean of these items was calculated to form an indicator of academic achievement. Cronbach’s alpha for this scale was .86.

Total Trauma Experienced by Child

Total trauma was based on the parent’s report of his or her child’s exposure to 12 highly stressful life events over the past year (e.g., robbery, physical assault, serious car accident). The summative score represents the total number of stressful events the child at age 10 has been exposed to over the past year.

Substance Use at Age 16

Substance use was assessed using self-reports of alcohol and drug use where participants estimated the amount of alcohol use over the past 6 months (0 = none to 8 = 2 to 3 times per day). Regular substance use for youth was defined as using at least once every 2 to 3 months.

Delinquency at Age 16

Youth delinquency was based on self-reported data from the Elliott Delinquency Scale (Elliott, 1983) where youth estimated how often (0 = not at all to 5 = over 15 times the past year) they participated in 42 different delinquent acts (e.g., purposely damaged or destroyed property, stole money or things). Serious delinquent acts (e.g., stole or tried to steal something worth more than US$50, sold hard drugs) were identified in the Major Offense subscale of the Elliott Delinquency Scale (Elliott, 1983). Minor delinquency consisted of the 31 remaining delinquent acts on the scale. The final scores for total, major, and minor delinquency were sums of the responses to the items within each category. Cronbach’s alpha was .67.

Deviant Peers at Age 16

Deviant peers were assessed using self-reports (12 items) of the number of friends the youth had who participated in a variety of problem behaviors (e.g., cheated on school tests, ruined or damaged something on purpose, stolen something worth over US$5). The final score was a sum of the number of friends participating in the problem behaviors. Cronbach’s alpha was not calculated because the score was a sum of items.

Substance Use

Substance use was assessed using self-reports of alcohol and drug use where participants estimated the amount of alcohol use (1 = never to 8 = daily). Repeated alcohol and drug use as an adolescent was defined as at least 4 to 6 times per year. Regular substance use as adults was defined as two or more times per week with the exception of mushrooms, cocaine, methadone, and sleeping pills, which was defined as a few times or more over the year.

Arrest as Young Adult

Information on arrests was obtained yearly from court data from the jurisdiction in which the participant was living. This variable was recoded as dichotomous variables with a value of 1 indicating the participant had been arrested at least once before the age of 21.

Incarceration as Young Adult

Similar to the arrest variable, information on incarcerations was obtained yearly from court data from the jurisdiction in which the participant was living. This variable was recoded, with a value of 1 indicating the participant had been incarcerated at least once before the age of 21, and a value of 0 indicating this had not occurred.

Analysis Plan

One of the primary aims of the current study was to better understand the severity and chronicity of externalizing symptoms across adolescence in relation to parental incarceration. To do this, we used GMM to identify distinct and divergent developmental trajectories of externalizing behaviors based on the parent’s and teacher’s reports of a child’s behavior from ages 10 to 16 years. We then examined the occurrence of children of incarcerated parents within these different trajectories. Finally, we examined the relations between the developmental trajectories and a variety of childhood predictors (e.g., SES, trauma, parenting) and emerging adult outcomes (e.g., criminality, substance use).

GMM was conducted using externalizing behavior observed at four time points (i.e., 10, 12, 14, 16 years of age) and a summary score representing the number of times the child’s average T score fell within the borderline/clinical range (greater than or equal to 60) for externalizing behaviors. The externalizing scores satisfied the normality assumption. The summary score for the four time points (range = 0-4) was modeled as a Poisson distribution. Estimates were obtained using maximum likelihood in Mplus, version 6.11 (Muthén & Muthén, 2012). The analysis assumes a common variance–covariance matrix of the externalizing measure at four time points across classes and conditional independence of the Poisson variable given class membership. This method uses all available data. Participants with missing scores for all four times points were excluded from the analysis (n = 8). All planned analyses were conducted using likelihood methods valid under MAR assumptions. This approach accounts for study attrition while most efficiently utilizing all the available data, as participants without complete data will be included fully in the analyses. Across all four time points in the remaining sample, the total missing externalizing behavior data were 1%. We created indicator variables for missingness at each time point and tested for associations between the missingness pattern at each time point for both the summary score (the number of time points that reach clinical levels of externalizing behavior ≥60) and the growth curve classes using a chi-square test of association at the 5% significance level. To satisfy the assumption of MAR, there should be no significant associations.

The model was assessed and compared by partitioning the data into different number of groups and, based on model convergence, the optimal number of classes was produced. The following fit indices were used to determine the optimal number of classes: Bayesian information criterion (BIC), Akaike information criterion (AIC), Bootstrap Likelihood Ratio Test (BLRT), and Lo, Mendell, and Rubin Likelihood Ratio Test (LMR-LRT; Nylund, Asparouhov, & Muthén, 2007). Analyses were stopped after reaching a maximum of 10 classes, which would limit the qualitative usefulness of attaching descriptive labels to each cluster. We then examined the different fit indices along with issues related to parsimony, theoretical justification, and interpretability to determine the best fitting model (Bauer & Curran, 2003; Muthén, 2003). After choosing an optimal number of classes, qualitative descriptions of the resulting patterns of childhood externalizing behaviors were assigned based on an in-depth examination of increases or decreases in the mean levels of the time point scores within each class as compared with the overall sample average.

In the second stage, we conducted a series of planned comparisons of relevant adolescent and adult characteristics between the trajectories. One-way between-groups ANOVA with planned comparisons was used for continuous variables, and chi-square tests were used for categorical variables. The low-risk, normative class was used as the comparison group.

Results

Descriptives

Table 1 displays the mean T scores and standard deviations for externalizing behavior across the four time points. Using GMM, subgroups for distinct developmental trajectories of externalizing behavior from 10 to 16 years were identified. Across all four time points, eight adolescents were missing externalizing behavior data (1.22% [8/655]). We created indicator variables for missingness at each time point (1 if the externalizing behavior was measured and 0 if the externalizing behavior was missing), and found no statistically significant associations between the missingness pattern at each time point and the five levels of the summary score (zero to four time points that reach clinical levels of externalizing behaviors) using a chi-square linear trend test with 1 degree of freedom and the four-class solution using a chi-square test of association with 3 degrees of freedom ([2 rows − 1] × [4 classes − 1]) at the 5% significance level.

Table 1:

Means, Standard Deviations, and Sample Size for Externalizing Behaviors Across Waves

Age	Externalizing M (SD)	Externalizing behavior N
10	49.05 (9.43)	647
12	50.19 (8.51)	587
14	50.78 (8.80)	570
16	49.48 (9.17)	536

As described in the “Analysis Plan” section, we examined the several fit indices along with issues related to parsimony, theoretical justification, and interpretability to determine the best fitting model. The model fit improved as more latent classes of trajectories were included up to four classes (see Table 2). The four-class solution resulted in four unique, interpretable classes, which were in line with theory. There were discrepancies among the fit indices when a fifth class was included and the additional class in the five-class solution was small (n = 9), did not add unique information, and nearly overlapped with one of the other high disruptive groups. For reasons of parsimony and interpretability, the four-class model was selected as the best fitting model.

Table 2:

Fit Indices of Growth Mixture Modeling for Externalizing Behaviors

Latent classes	BIC	AIC	BLRTp	LMR-LRTp	Entropy
2	16,082.765	15,988.846	<.001	<.001	.889
3	16,038.752	15,917.999	<.001	.104	.893
4	16,016.695	15,869.107	<.001	.085	.891
5	16,006.031	15,831.609	<.001	.318	.892
6	16,032.651	15,831.395	.088	.190	.911

Note. BIC = Bayesian information criterion; AIC = Akaike information criterion; BLRT = Bootstrap Likelihood Ratio Test; LMR-LRT = Lo, Mendell, and Rubin Likelihood Ratio Test.

The growth curves for the four externalizing classes (named Low-Stable, Mid-Increasing, Borderline-Stable, and Chronic-High based on the observed pattern) are shown in Figure 1. The Low-Stable group (n = 488) comprised youth who had low levels of externalizing behaviors across adolescence. Youth in the Borderline-Stable group (n = 75) consistently had middle to borderline high scores in externalizing behaviors across all time points. The Mid-Increasing group of youth (n = 63) comprised youth who exhibited middle to lower levels of externalizing behaviors at the age of 10, but had steady growth of externalizing behaviors through the age of 16. Finally, the Chronic-High group of youth (n = 21) exhibited chronically high levels of externalizing across development.

Figure 1:

Externalizing Growth Curves for Children From 10 to 16 Years

Descriptives of Trajectories

After identifying the appropriate class solution, we examined the associations between the trajectories and different adolescent characteristics and emerging adult outcomes. Of primary interest was the relation between parental incarceration and the trajectory classes. The chi-square test for independence indicated a significant association between these two elements, χ²(3, N = 647) = 19.7, p < .001, φ = .18. As shown in Table 3, children of incarcerated parents were underrepresented in the Low-Stable trajectory and overrepresented in the Mid-Increasing trajectory. Although children of incarcerated parents were also overrepresented in the Borderline-Stable group, this difference was not statistically significant. Tables 4 and 5 present the results of the associations of the trajectory classes and additional preadolescent characteristics and emerging adult outcomes. Descriptions of each of the classes follow.

Table 3:

Percentages of Children in Each Trajectory Class

Trajectory class	Total percent of youth(n = 647)	Percent of youth withparental incarceration(n = 67)	Percent of youth withno parental incarceration(n = 580)
Low-Stable (n = 488)	75.4	59.7_a	77.2_b
Borderline-Stable (n = 75)	11.6	17.9_a	10.9_a
Mid-Increasing (n = 63)	9.7	22.4_a	8.3_b
Chronic-High (n = 21)	3.2	0.0_a	3.6_a

Note. χ²(3, n = 647) = 19.7; p < .001; φ = .18. Column proportions of youth who have and have not experienced parental incarceration were compared using a z test. If a pair of values is significantly different from one another, the values have a different subscript letter (a, b) assigned to them.

Table 4:

Preadolescent and Family Characteristics: One-Way Between-Groups ANOVA With Planned Comparisons

Characteristics	Low-Stable(n = 488)	Borderline-Stable(n = 75)	Mid-Increasing(n = 63)	Chronic-High(n = 21)
Preadolescent and family measures
SES	34.8 (10.7)	33.2 (10.7)	33.2 (12.0)	35.6 (10.8)
Mother–child relationship^a	4.5 (0.6)	4.0 (0.8)***	4.3 (0.7)	3.9 (1.0)**
Father/child relationship	2.8 (0.8)	2.9 (0.8)***	2.8 (0.7)	3.3 (0.7)
Maternal depression^a	9.5 (8.4)	15.4 (9.7)***	11.8 (8.8)	16.1 (11.5)**
Paternal depression^c	8.5 (6.6)	15.3 (8.6)	9.5 (6.2)	9.9 (7.1)
Academic achievement^c	3.3 (0.5)	3.2 (0.6)	3.2 (0.6)	3.1 (0.7)
Total trauma^a	0.7 (0.8)	0.9 (0.9)	1.2 (1.2)***	1.6 (1.3)***
Inconsistent discipline^a	2.1 (0.5)	2.5 (0.5)***	2.2 (0.5)	2.7 (0.6)***
Harsh discipline^a	2.7 (1.0)	3.0 (1.0)	3.4 (0.9)***	3.5 (1.1)**
Late adolescent measures
Deviant peers^a	18.4 (6.4)	24.8 (7.8)***	24.0 (8.6)***	25.3 (8.1)
Delinquency total^a	6.4 (10.9)	13.8 (15.5)***	14.7 (17.5)***	21.7 (25.5)***
Delinquency major^b	0.5 (1.9)	1.0 (2.4)***	1.4 (2.6)***	3.7 (6.2)***
Delinquency minor^a	5.9 (9.6)	12.8 (14.2)***	13.2 (15.8)***	18.0 (20.2)***
Tobacco^a	1.0 (2.2)	2.6 (3.2)***	4.7 (3.5)***	3.9 (3.9)***
Beer use^a	1.0 (1.4)	1.9 (2.0)***	2.0 (2.0)***	2.4 (2.3)**
Wine use	0.5 (1.0)	0.9 (1.6)*	0.6 (1.2)	0.8 (1.7)
Liquor use^a	0.6 (1.2)	1.8 (2.2)***	1.7 (1.9)***	1.5 (1.8)**
Any alcohol^a	1.2 (1.6)	2.4 (2.2)***	2.3 (2.0)***	2.8 (2.4)**
Marijuana use^a	0.8 (1.7)	2.2 (2.7)***	2.5 (2.9)***	2.4 (3.0)**
Other drugs use^b	0.2 (0.7)	0.4 (1.3)	0.8 (1.9)***	0.2 (0.8)

Note. Reference category is the Low-Stable class for the planned comparisons. SES = social economic status.

F statistic significant at p < .001. ^bF statistic significant at p < .01. ^cF statistic significant at p < .05.

t < .05. **t < .01. ***t < .001.

Table 5:

Percentages and Odds Ratios of Criminality and Substance Use for Each Class

Measures	Trajectories				Odds ratios
Measures	Low-Stable(A)% (n = 488)	Borderline-Stable(B)% (n = 75)	Mid-Increasing (C)% (n = 63)	Chronic-High(D)% (n = 21)	B/A	C/A	D/A
Late adolescent measures
Criminality
Ever arrested as juvenile	21.8	52.4	56.6	68.8	3.94* (2.28-6.81)	4.67* (2.58- 8.44)	7.88* (2.67-23.26)
Repeated alcohol use^a
Beer***	20.8	39.3	48.1	52.9	2.47* (1.40-4.34)	3.53* (1.97-6.33)	4.28* (1.60-11.41)
Wine	8.8	16.4	16.7	17.6	2.03 (0.95-4.33)	2.07 (0.94-4.57)	2.21 (0.61-8.07)
Liquor***	14.0	36.1	35.2	41.2	3.47* (1.92- 6.28)	3.34* (1.79- 6.24)	4.31* (1.58-11.78)
Any alcohol***	26.0	49.2	55.6	58.8	2.76* (1.59-4.77)	3.56* (1.99-6.36)	4.07* (1.51-10.96)
Repeated substance use^a
Tobacco***	16.7	36.1	68.5	52.9	2.81* (1.57- 5.04)	10.85* (5.78- 20.38)	5.61* (2.09-15.05)
Marijuana***	15.2	42.6	48.1	47.1	4.15* (2.33-7.37)	5.18* (2.85-9.43)	4.96* (1.84-13.35)
Other drugs***	2.5	8.2	14.8	5.9	3.55* (1.17-10.78)	6.92* (2.60-18.42)	2.49 (0.30-20.63)
Young adult measures
Criminality
Ever arrested	41.7	61.9	66.0	81.3	2.27* (1.32-3.92)	2.72* (1.49-4.97)	6.06* (1.70-21.60)
Ever arrested as adult	28.3	38.1	37.7	50.0	1.56 (0.90-2.71)	1.54 (0.85-2.79)	2.54 (0.93-6.92)
Ever incarcerated	9.0	20.0	15.9	23.8	2.52* (1.32-4.81)	1.90 (0.91-4.00)	3.15* (1.10-9.02)
Regular substance use
Tobacco^b ***	32.1	44.3	80.4	64.3	1.68 (0.97- 2.91)	8.67* (4.20-17.87)	3.81* (1.25-11.59)
Marijuana^b *	11.3	18.0	9.8	35.7	1.73 (0.84-3.58)	0.86 (0.32-2.27)	4.38* (1.40-13.67)
Beer^b	25.3	26.2	23.5	35.7	1.05 (0.57-1.94)	0.91 (0.46-1.80)	1.64 (0.54-5.01)
Wine^b	4.7	4.9	2.0	14.3	1.05 (0.30-3.67)	0.41 (0.05-3.10)	3.38 (0.70-16.24)
Liquor^b	9.1	11.5	5.9	14.3	1.29 (0.55-3.05)	0.62 (0.18-2.10)	1.66 (0.36-7.71)
Mushrooms^c ***	1.0	9.8	2.0	—	10.31* (2.82-37.69)	1.89 (0.21-17.25)	N/A
Cocaine^c	5.5	8.2	2.0	14.3	1.54 (0.56-4.25)	0.35 (0.05-2.62)	2.87 (0.60-13.68)
Methadone^c	0.8	—	2.0	—	N/A	2.53 (0.26-24.83)	N/A
Sleeping pills^c	3.7	8.2	5.9	—	2.35 (0.81-6.77)	1.64 (0.46-5.93)	N/A

At least 4 to 6 times per year. ^bTwo or more times per week. ^cA few times or more over the year.

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

Low-Stable Trajectory

Most of the youth (n = 488, 75.4%) were in the Low-Stable trajectory in which the youth displayed low levels of externalizing behaviors across adolescence. Compared with the other three trajectories, this group was characterized by a strong mother–child relationship, low likelihood of maternal depression, low levels of trauma, and consistent and appropriate parenting. At age 16, this group was the least likely to participate in delinquent activities (both minor and major), had fewer deviant peers, and had the least amount of tobacco, alcohol, and drug use of the four classes. In terms of criminality, this group of youth was the least likely to be arrested (as a juvenile or adult) or incarcerated. As adults, individuals in this group were the least likely to use tobacco products regularly; however, their usage pattern for alcohol and other drugs was not significantly different from the other groups.

Borderline-Stable Trajectory

The Borderline-Stable trajectory was the next largest group of youth (n = 75, 11.6%) and had consistently higher levels of externalizing behavior across time (although rarely over the “clinical” cut point). The prevalence of children of incarcerated parents in this group (17.9%) was greater than children without a history of parental incarceration (10.9%). However, this difference was not significant. Compared with the Low-Stable group, this trajectory class was characterized by weaker parent–child relationships, higher levels of maternal depression, and less consistent parenting. The greater levels of trauma observed was marginally significant. On average, the youth participated in more delinquent acts (mostly minor), had more deviant peers, and were more likely to use tobacco, all types of alcohol, and marijuana. Compared with youth in the Low-Stable group, these youth were nearly 4 times more likely to have been arrested as a juvenile, 2.5 times more likely to use beer repeatedly, 3.5 times more likely to use liquor repeatedly, nearly 3 times as likely to use tobacco repeatedly, and 4 times as likely to use marijuana and drugs repeatedly at the age of 16. As adults, they were more than 2 times more likely to have ever been arrested and 2.5 times more likely to have been incarcerated. Alcohol and drug use as adults was not significantly different from the Low-Stable group except around the use of mushrooms where individuals were more than 10 times as likely to have tried mushrooms.

Mid-Increasing Trajectory

The Mid-Increasing group of youth was the third largest class of youth (n = 63, 9.7%). This group had middle to low level of externalizing behaviors when they were 10 years old but the levels gradually increased across the time points. The prevalence of children of incarcerated parents (22.4%) was significantly greater than children without a history of parental incarceration (8.3%). Compared with the low-risk class, this class experienced harsher parenting and more trauma. By 16 years, youth in this group were more likely to be participating in more delinquent acts (both major and minor) and have more peers who are deviant. They were drinking more alcohol and using more marijuana and other drugs. In terms of criminality, they were more than 4.5 times more likely to have been arrested as a juvenile, 3.5 times more likely to drink beer and liquor repeatedly. They were nearly 11 times more likely smoke repeatedly, 5 times more likely to smoke marijuana repeatedly, and nearly 7 times more likely to have used drugs repeatedly. As an adult, they were nearly 3 times more likely to have ever been arrested and nearly 9 times more likely to use tobacco.

Chronic-High Trajectory

The Chronic-High group of youth had the smallest number of youth (n = 21, 3.2%). There were no children of incarcerated parents in this group. Youth in this group had consistently clinically high levels of externalizing behavior across all four time points. Compared with the Low-Stable group, the Chronic-High group was characterized by a weaker mother–child relationship, higher levels of maternal depression, higher levels of trauma, and less consistent and harsher parenting. Of all the groups, they had the highest levels of reported delinquency both major and minor offenses, and the highest levels of beer consumption. In terms of criminality, they were nearly 8 times more likely to have been arrested as a juvenile, more than 4 times more likely to be a repeat drinker of beer and liquor. They were nearly 6 times more likely to smoke repeatedly, nearly 5 times more likely to smoke marijuana repeatedly, and 2.5 times more likely to have used drugs repeatedly. As an adult, they were nearly 6 times more likely to have ever been arrested, 3 times as likely to have been incarcerated, and nearly 4 times more likely to use tobacco and marijuana regularly.

Overall, the trajectory classes differed significantly on many of the preadolescent measures (e.g., the mother–child relationship, maternal depression, total trauma, inconsistent and harsh parenting) as well as adolescent delinquency, adult criminality, and substance use. In general, the Mid-Increasing, Borderline-Stable, and Chronic-High trajectory groups showed significantly higher levels of early risk factors and problematic outcomes than the Low-Stable trajectory group. Although these three trajectories were all problematic, each varied in its relation to specific negative outcomes of interest.

Discussion

Our findings support earlier work demonstrating multiple different developmental trajectories of externalizing behaviors. The best fitting model had four trajectory classes: a Low-Stable trajectory and three more potentially problematic trajectories (Mid-Increasing, Borderline-Stable, and Chronic-High). These trajectories were similar to earlier theorized growth patterns including a low-risk group, “late starters,” and “early starters” (Moffitt, 1993; Patterson et al., 1989). The majority of youth (75%) fell into the Low-Stable group. In this group, youth displayed low levels of externalizing behaviors across adolescence. The Mid-Increasing class was similar to the “late starters” group, where youth had lower levels of externalizing at age 10, but these levels gradually increased to clinically high levels by the time the youth were 16 years old. This group represented roughly 10% of the youth in the study. Finally, we found two unique early starter groups: a Chronic-High group and a Borderline-Stable group. Both groups demonstrated a relatively high level of externalizing behavior over time (with a slight decline toward age 16); however, the Chronic-High group (3.2%) displayed clinically high levels of externalizing behaviors, whereas the Borderline-Stable group (11.6%) exhibited high levels of externalizing that hovered slightly below a clinical level. By extending analyses beyond age 16, we might have seen a continued decrease in externalizing as identified by other researchers.

One of our primary interests was the relation between parental incarceration and the different trajectory classes. Our study demonstrated that children of incarcerated parents were underrepresented in the Low-Stable group and overrepresented in the Mid-Increasing group. However, one of our key findings is that most children of incarcerated parents (60%) were best represented by the low-risk trajectory (the Low-Stable group). Although this prevalence of children of incarcerated parents was significantly less than children without a history of parental incarceration (77%), parental incarceration in and of itself does not destine all children to problematic pathways. Many youth with a history of parental incarceration exhibited low levels of externalizing behaviors across adolescence.

This low-risk group was characterized by a strong mother–child relationship, a low likelihood of maternal depression, less trauma, and consistent and appropriate parenting. In later adolescent years, this group was the least likely to participate in delinquent activities (both minor and major), had fewer deviant peers, and used the least amount of tobacco, alcohol, and drugs. In terms of criminality, youth in this group were the least likely to be arrested (as a juvenile or an adult) or incarcerated of any of the four trajectory groups and were the least likely to use tobacco on a regular basis. Their usage pattern for alcohol and other drugs was not significantly different from the other groups. These results are consistent with earlier research that demonstrates the protective nature of a strong parent–child (especially mother) relationship and effective parenting, and harmful effects of trauma and maternal depression in terms of future delinquency, criminality, and substance use (e.g., Murray & Farrington, 2005; Nagin & Tremblay, 1999; Schaeffer et al., 2003).

Although children of incarcerated parents were underrepresented in the low-risk trajectory, another key finding is that they were overrepresented in the Mid-Increasing group. This finding is consistent with other research on children of incarcerated parents, which demonstrates increase risks of several problematic outcomes (e.g., Murray & Farrington, 2005). The Mid-Increasing group exhibited the largest difference in prevalence between youth who had experienced parental incarceration (22.4%) and those who had not (9.7%). Youth experiencing parental incarceration were also overrepresented in the Borderline-Stable group. However, this difference was not significant. Compared with the low-risk class, youth in these classes exhibited weaker mother–child relationships, parenting problems, and more trauma early in adolescence. The increased likelihood of delinquency, criminality, and regular use of tobacco, alcohol, and various drugs for both these groups is disconcerting.

Interestingly, there were no children of incarcerated parents in the final group: the Chronic-High class of youth. As a whole, this group had the smallest number of youth (n = 21, 3.2%). Compared with the Low-Stable group, the Chronic-High group was characterized by a weaker mother–child relationship, higher levels of maternal depression, higher levels of trauma, and less consistent and harsher parenting. Of all the groups, youth in this group had the highest regular use of beer and liquor as adolescents. In terms of delinquency and criminality, this group is the most worrisome. Youth in this group had the highest levels of reported delinquency, both major and minor offenses, and had the greatest odds of being arrested as a juvenile. As adults, they had the greatest odds of being arrested and incarcerated.

Overall, the results suggest that there are several developmental pathways of externalizing behaviors—one that represents a low-risk pathway and three that are more problematic and are related to increased delinquency, criminality, and certain types of substance use in later adolescence and emerging adulthood. More than half of the children of incarcerated parents fell in a low-risk group. Although this finding highlights the resilience of many children of incarcerated parents, the underrepresentation of children in this group combined with the overrepresentation of children in some of the problematic trajectories underscores the urgency to develop and test programs, which help support children affected by parental incarceration.

The results also provide support to the idea that there is heterogeneity in the impact of parental incarceration on children, at least as it relates to the development of externalizing behaviors. Although an individual risk factor can be detrimental to a child’s development, the combined effect of multiple influences (both helpful and harmful) and their interactions can lead to a variety of outcomes over time. As a group, children of incarcerated parents face many of the same risks and challenges. However, individually, they are exposed to a different array of risk and protective factors at the individual, family, and community level. As such, their development varies. The present study highlights these issues, particularly the association between youth development and parenting, parental health, and trauma. Those children with strong parent–child relationships, consistent and appropriate parenting, healthy parental mental health, and low levels of trauma were more likely do well over time and were at lower risk of later substance use and criminality, whereas those who exhibited higher levels of parenting and family dysfunction were less likely to fare as well during the late adolescent and early adult years.

Limitations

The current study offers new contributions to our understanding of the children of incarcerated parents. Each of these is in need of examination in other data sets. All such examinations, such as this one, have limitations. This study has six key limitations. First, most participants in the sample were White, and all were living in a medium size urban area in the Pacific Northwest. It is uncertain whether findings from this study are relevant to other populations within other geographic locations. Given the disproportionality of people of color within corrections systems, as well as differential concentration of incarcerated individuals within and between regions within the United States, future research should examine the impact over time of parental incarceration within specific groups and geographical areas.

Second, the number of children in the sample who had experienced parental incarceration was small (i.e., 10.2%, n = 67), and this may limit the meaningfulness of our findings. However, this sample is well in the range of other peer-reviewed and published samples on the topic of interest. For example, in a recent in-depth analysis of longitudinal data from four countries on outcomes for the children of incarcerated parents (Murray, Bijleveld, Farrington, & Loeber, 2014), the percentage of affected children ranged from 2% to 12%. The two samples in that work that were from studies with characteristics most similar to the one reported here had samples sizes of 397 (with 6% of boys, or n = 23, who had experienced parental incarceration by age 10) and 1,009 (with 12% of boys, or n = 122, who had experienced parental incarceration by age 18). All samples have their problems and limits, and, thus, all findings from this and any other analysis need to be replicated, and multiple times. As more studies are conducted of samples that include children of incarcerated parents, findings will emerge that replicate and appear robust.

Third, the main and interactive effects of pubertal timing within genders were not included in this study. Research has demonstrated that changes in externalizing during the adolescent years do not happen in isolation. Rather, the interactions of factors such as pubertal timing, gender, and stressful events can affect externalizing problems across adolescence (e.g., Ge, Conger, & Elder, 2001; Graber, Seeley, Brooks-Gunn, & Lewinsohn, 2004; Negriff & Susman, 2011). For many children, parental incarceration is a major stressor and, as such, is an event during adolescence, which when interacting with pubertal timing and gender could moderate their externalizing trajectories. Future studies need to examine the complex interrelationships between parental incarceration, pubescent development within genders, other potential risk factors, and externalizing trajectories.

Fourth, several of the predictor variables (e.g., parent–child relationship factors, parental discipline practices) may change over the course of adolescence. This study was somewhat underpowered in terms of the capacity to model this. Future research with a larger sample could include these variables as time-varying covariates.

Fifth, the incarceration variable lacks information. The variable did not capture several elements related to the parents’ incarceration that could be relevant to the youth’s development including the age of the child at the time of the incarceration, the relationship of the child with the incarcerated parent, the level of disruption the incarceration caused, and the frequency and length of the incarceration. Each of these elements could affect a youth differentially over time. Unfortunately, more detailed information such as this was not available in the present sample. Collecting information on and then analyzing these different aspects of incarceration in future studies will help to not only illuminate the mechanisms through which parental incarceration affects children but also better inform future policy and practice interventions.

Finally, there was a lack of reliability and validity information on some of the variables of interest. Specifically, the reliability for two study measures (i.e., parent–child relationship, delinquency) were slightly below the typical threshold for adequate internal consistency (i.e., α = .70). However, the measures were retained for three main reasons. First, both measures were deemed important covariates in terms of their relation to child externalizing. Second, even with somewhat low alpha values, the constructs operated as theoretically expected in terms of their relation to externalizing. Finally, the Cronbach’s alpha for delinquency was quite close to .70, and the parent–child relationship included only two items (which tends to decrease internal consistency). In terms of the validity of variable, this issue is certainly not unique to this study but is a general problem where reliable measures are typically used but which have only face validity. As a field, continued work in this area will help improve the quality of knowledge on life course development.

Implications for Research and Practice

In light of the large number of youth affected by parental incarceration and the associated risks that incarceration can have on youth and their families, additional studies are needed. We encourage researchers to examine the diverse web of interacting influences that can affect children over time, and, rather than grouping all children of incarcerated parents together into one group, to take an individualized approach to better identify the specific strengths and risks that can promote or impede positive development. High-quality quantitative and qualitative information is critical to guiding the development of targeted, tailored preventive and clinical interventions that support youth and families affected by parental incarceration.

In terms of practice, there are a variety of interventions to support children of incarcerated parents and their families. However, few have been rigorously tested (e.g., Kjellstrand, 2017). Interventions have tended to focus on one of three areas: child adjustment (e.g., mentoring), parenting/family functioning (e.g., parenting education), or the broader personal/community context (e.g., job training). Work in each of these areas may be helpful for a given child, but with the number of issues at play for many children of incarcerated parents, it is clear that a “one-size-fits-all” approach to addressing the child and family needs is likely going to be quite limited in terms of its usefulness, both for individuals and for the population at large. Rather a tailored, multilevel approach, which accounts for individual differences, is needed.

As a first step, social workers, school counselors, teachers, mentors, and others who work with youth who have experienced parental incarceration need to be aware of the multiple challenges that can come into play for these families as well as the corresponding strengths that different youth and families possess. As such, completing a thorough assessment of the child’s and family’s situation across levels (individual, family, and community) is critical. Our findings indicate that, within this assessment, special attention should be given to issues related to parenting, the parent–child relationship, parental mental health (specifically depression), exposure to trauma, and history of problem behavior. Such an assessment will help guide the development of a tailored intervention strategy for the child and family that builds on strengths while targeting challenges. Such a strategy will require combining multiple relevant interventions as well as a coordinated effort across systems. Despite the challenges in this regard, this type of strategy seems best suited to address the needs facing this population, and, over the long term, seems most likely to result in better outcomes for the subset of the children of incarcerated parents who are most at risk of problematic outcomes.

Footnotes

Acknowledgements

We would like to express our deep appreciation and thanks to the participating organizations, personnel, principals, teachers, parents, and youth in the LIFT trial. We are also thankful to the other staff members who have worked on LIFT over the years.

Authors’ Note:

Support for this project was provided by Grant R01 MH 65553 from the Prevention and Behavioral Medicine Research Branch, Division of Epidemiology and Services Research, NIMH, NIH, U. S. PHS; Grant R01 MH 054248 from the Prevention Research Branch, NIDA, NIH, U.S. PHS; and Grant 2013-JU-FX-0007 from the Office of Juvenile Justice and Delinquency Prevention, U.S. DOJ.

Jean Kjellstrand, MSW, PhD, is an assistant professor in the Department of Counseling Psychology and Humans Services within the College of Education at the University of Oregon. Her research focuses on the impact of parental incarceration on child development and the effectiveness of interventions which strive to support families both during and after incarceration.

Gary Yu, MPH, DrPH, is an associate research scientist at the New York University Rory Meyers College of Nursing. He is also a biostatistician affiliated with the Center for Drug Use and HIV/HCV Research (CDUHR).

J. Mark Eddy, PhD, is a senior research scientist and director of community-based research with the Family Translational Research Group at New York University. His research focuses on developing and rigorously testing targeted psychosocial interventions designed with and for established systems of care within communities.

Charles R. Martinez, Jr., PhD, is the Philip H. Knight Professor in the Department of Educational Methodology, Policy, and Leadership at the University of Oregon, where he also serves as founding director of the Center for Equity Promotion. His substantive interests center on identifying malleable factors within education and social systems that promote healthy adjustment for families and children, especially those from culturally diverse populations.

References

Achenbach

T. M.

(1991). Manual for the child behavior checklist/4-18 and 1991 profile. Burlington, VT: Department of Psychiatry, University of Vermont.

Bauer

D. J.

Curran

P. J.

(2003). Distributional assumptions of growth mixture models: Implications for overextraction of latent trajectory classes. Psychological Methods, 8, 338-363. doi:10.1037/1082-989X.8.3.338

Dishion

T. J.

Snyder

J. J.

(Eds.). (2016). The Oxford handbook of coercive relationship dynamics. New York, NY: Oxford University Press.

Eddy

J. M.

(2003). LIFT continuation grant application. Eugene: Oregon Social Learning Center.

Eddy

J. M.

Poehlmann

(2010). Children of incarcerated parents: A handbook for researchers and practitioners. Washington, DC: Urban Institute Press.

Eddy

J. M.

Reid

J. B.

Fetrow

R. A.

(2000). An elementary school-based prevention program targeting modifiable antecedents of youth delinquency and violence: Linking the interests of families and teachers (LIFT). Journal of Emotional and Behavioral Disorders, 8, 165-176. doi:10.1177/106342660000800304

Elliott

D. S.

(1983). Interview schedule, National Youth Survey. Boulder, CO: Behavioral Research Institute.

Conger

R. D.

Elder

G. H.

(2001). Pubertal transitions, stressful life events, and the emergence of gender differences in adolescent depressive symptoms. Developmental Psychology, 37, 404-417. doi:10.1037/0012-1649.37.3.404

Geller

Garfinkel

Cooper

Mincy

(2009). Parental incarceration and child well-being: Implications for urban families. Social Science Quarterly, 90, 1186-1202. doi:10.1111/j.1540-6237.2009.00653.x

10.

Graber

J. A.

Seeley

J. R.

Brooks-Gunn

Lewinsohn

P. M.

(2004). Is pubertal timing associated with psychopathology in young adulthood? Journal of the American Academy of Child & Adolescent Psychiatry, 43, 718-726. doi:10.1097/01.chi.0000120022.14101.11

11.

Gutman

L. M.

Sameroff

A. J.

Cole

(2003). Academic growth curve trajectories from 1st grade to 12th grade: Effects of multiple social risk factors and preschool child factors. Developmental Psychology, 39, 777-790. doi:10.1037/0012-1649.39.4.777

12.

Hollingshead

A. B.

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

13.

Johnson

Arditti

McGregor

(2018). Risk, protection, and adjustment among youth with incarcerated and non-resident parents: A mixed-methods study. Journal of Child and Family Studies, 27, 1914-1928. doi:10.1007/s10826-018-1045-0

14.

Johnson

Easterling

(2015). Coping with confinement: Adolescents’ experiences with parental incarceration. Journal of Adolescent Research, 30, 244-267. doi:10.1177/0743558414558593

15.

King

Mauer

Young

(2005). Incarceration and crime: A complex relationship. Washington, DC: The Sentencing Project.

16.

Kjellstrand

(2017). Building a tailored, multilevel prevention strategy to support children and families affected by parental incarceration. Smith College Studies in Social Work, 87(1), 112-129. doi:10.1080/00377317.2017.1248203

17.

Kjellstrand

J. M.

Eddy

(2011a). Mediators of the effect of parental incarceration on adolescent externalizing behaviors. Journal of Community Psychology, 39, 551-565. doi:10.1002/jcop.20451

18.

Kjellstrand

J. M.

Eddy

(2011b). Parental incarceration during childhood, family context, and youth problem behavior across adolescence. Journal of Offender Rehabilitation, 50(1), 18-36. doi:10.1080/10509674.2011.536720

19.

Luther

(2015). Examining social support among adult children of incarcerated parents. Family Relations, 64, 505-518. doi:10.1111/fare.12134

20.

Malti

Averdijk

Ribeaud

Rotenberg

K. J.

Eisner

M. P.

(2013). “Do you trust him?” Children’s trust beliefs and developmental trajectories of aggressive behavior in an ethnically diverse sample. Journal of Abnormal Child Psychology, 41, 445-456. doi:10.1007/s10802-012-9687-7

21.

Maruschak

L. M.

Mumola

C. J.

(2010). Incarcerated parents and their children: Findings from the Bureau of Justice Statistics. In Eddy

J. M.

Poehlmann

(Eds.), Children of incarcerated parents: A handbook for researchers and practitioners (pp. 33-54). Washington, DC: Urban Institute Press.

22.

Maughan

Pickles

Rowe

Costello

Angold

(2000). Developmental trajectories of aggressive and non-aggressive conduct problems. Journal of Quantitative Criminology, 16, 199-221. doi:1023/A:1007516622688

23.

Moffitt

T. E.

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

24.

Muftic

Smith

(2018). Sex, parental incarceration, and violence perpetration among a sample of young adults. Journal of Interpersonal Violence, 33, 316-338. doi:10.1177/0886260515605123

25.

Murray

Bijleveld

Farrington

Loeber

(2014). Effects of parental incarceration on children: Cross-national comparative studies. Washington, DC: American Psychological Association.

26.

Murray

Farrington

D. P.

(2005). Parental imprisonment: Effects on boys’ antisocial behaviour and delinquency through the life-course. Journal of Child Psychology and Psychiatry, 46, 1269-1278. doi:10.1111/j.1469-7610.2005.01433.x

27.

Murray

Farrington

D. P.

Sekol

(2012). Children’s antisocial behavior, mental health, drug use, and educational performance after parental incarceration: A systematic review and meta-analysis. Psychological Bulletin, 138, 175-210. doi:10.1037/a0026407

28.

Muthén

(2003). Statistical and substantive checking in growth mixture modeling: Comment on Bauer and Curran (2003). Psychological Methods, 8, 369-377. doi:10.1037/1082-989X.8.3.369

29.

Muthén

L. K.

(2000). Integrating person-centered and variable-centered analyses: Growth mixture modeling with latent trajectory classes. Alcoholism: Clinical and Experimental Research, 24, 882-891. doi:10.1111/j.1530-0277.2000.tb02070.x

30.

Muthén

L. K.

Muthén

B. O.

(2012). Mplus User’s Guide (Version 7) [Software]. Los Angeles, CA: Muthén & Muthén.

31.

Nagin

Tremblay

R. E.

(1999). Trajectories of boys’ physical aggression, opposition, and hyperactivity on the path to physically violent and nonviolent juvenile delinquency. Child Development, 70, 1181-1196. doi:10.1111/1467-8624.00086

32.

Nagin

D. S.

(1999). Analyzing developmental trajectories: A semiparametric, group-based approach. Psychological Methods, 4, 139-157. doi:10.1037/1082-989X.4.2.139

33.

Negriff

Susman

E. J.

(2011). Pubertal timing, depression, and externalizing problems: A framework, review, and examination of gender differences. Journal of Research on Adolescence, 21, 717-746. doi:10.1111/j.1532-7795.2010.00708.x

34.

Nivard

Lubke

Dolan

Evans

Pourcain

Munafo

Middeldorp

(2017). Joint developmental trajectories of internalizing and externalizing disorders between childhood and adolescence. Development and Psychopathology, 29, 919-928. doi:10.1017/S0954579416000572

35.

Nylund

K. L.

Asparouhov

Muthén

B. O.

(2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14, 535-569. doi:10.1080/10705510701575396

36.

Oregon Social Learning Center. (1990). Linking the interests of families and teachers assessment [Measurement instrument]. Eugene, OR: Author.

37.

Patterson

DeBaryshe

B. D.

Ramsey

(1989). A developmental perspective on antisocial behavior. American Psychologist, 44, 329-335. doi:10.1037/0003-066X.44.2.329

38.

Patterson

G. R.

(1982). Coercive family process. Eugene, OR: Castalia.

39.

Patterson

G. R.

Reid

J. B.

Dishion

T. J.

(1992). Antisocial boys. Eugene, OR: Castalia.

40.

Petras

Schaeffer

C. M.

Ialongo

Hubbard

Muthén

Lambert

S. F.

. . . Kellam

(2004). When the course of aggressive behavior in childhood does not predict antisocial outcomes in adolescence and young adulthood: An examination of potential explanatory variables. Development and Psychopathology, 16, 919-941. doi:10.1017/S0954579404040076

41.

Radloff

L. S.

(1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385-401. doi:10.1177/014662167700100306

42.

Reid

J. B.

Eddy

Fetrow

R. A.

Stoolmiller

(1999). Description and immediate impacts of a preventive intervention for conduct problems. American Journal of Community Psychology, 27, 483-517. doi:10.1023/A:1022181111368

43.

Rutter

Dunn

Plomin

Simonoff

Pickles

Maughan

. . . Eaves

(1997). Integrating nature and nurture: Implications of person-environment correlations and interactions for developmental psychopathology. Development and Psychopathology, 9, 335-364. doi:10.1017/S0954579497002083

44.

Schaeffer

C. M.

Petras

Ialongo

Poduska

Kellam

(2003). Modeling growth in boys’ aggressive behavior across elementary school: Links to later criminal involvement, conduct disorder, and antisocial personality disorder. Developmental Psychology, 39, 1020-1035. doi:10.1037/0012-1649.39.6.1020

45.

Turanovic

Rodriguez

Pratt

(2012). The collateral consequences of incarceration revisited: A qualitative analysis of the effects on caregivers of children of incarcerated parents. Criminology, 50, 913-959. doi:10.1111/j.1745-9125.2012.00283.x

46.

Turney

(2017). The unequal consequences of mass incarceration for children. Demography, 54, 361-389. doi:10.1007/s13524-016-0543-1

47.

Turney

Wildeman

(2015). Detrimental for some? Heterogeneous effects of maternal incarceration on child wellbeing. Criminology and Public Policy, 14, 125-156. doi:10.1111/1745-9133.12109

48.

Walmsley

(2015). World prison population list. Retrieved from http://www.prisonstudies.org/sites/default/files/resources/downloads/world_prison_population_list_11th_edition_0.pdf

49.

Western

Wildeman

(2009). The black family and mass incarceration. The Annals of the American Academy, 621, 221-242. doi:10.1177/0002716208324850

50.

Wildeman

(2009). Paternal incarceration and children’s physically aggressive behaviors: Evidence from the fragile families and child wellbeing study. Social Forces, 89, 285-310. doi:10.1353/sof.2010.0055