Adult Outcomes of Justice Involved Indigenous Youth

Abstract

Juvenile arrest serves as a critical turning point in the life-course that disrupts the successful transition to adulthood and carries numerous consequences including diminished socioeconomic status. Despite their disproportionately high rates of contact with the criminal justice system (CJS), Indigenous people’s experiences remain largely invisible in extant research. Further, colonization has left them in an extremely marginalized position in terms of social, economic, and political power, which is compounded by CJS involvement. In the current study, we apply propensity score matching to investigate whether being arrested in adolescence impacts early adult socioeconomic outcomes (i.e., education, employment, and income). Data come from the Healing Pathways project, a longitudinal, community-based participatory study of North American Indigenous young people that includes eight waves of data in adolescence and three waves in early adulthood. We find that being arrested at least once in adolescence is associated with significantly less education and income, and lower rates of full-time employment in young adulthood (mean age = 26.2 years). Criminal justice system involvement widens existing socioeconomic disparities, and remedying these consequences requires changes in how CJS policies are enacted as well as larger structural changes to address significant inequities in income, education, and employment for Indigenous people.

Involvement in the criminal justice system (CJS) impacts numerous parts of an individual’s life. It results in diminished educational and employment opportunities, lower socioeconomic status (SES) (Pettit & Western, 2004; Western, 2002), negative mental and physical health outcomes (Massoglia & Pridemore, 2015; Schnittker et al., 2012), family instability (Wakefield & Wildeman, 2014), and substance use/abuse (Bath et al., 2018). Most research has focused on the harmful outcomes of CJS involvement among adult populations; however, CJS involvement has large consequences for youth as well. This is an important point as the United States continues to incarcerate a larger proportion of its youth population in comparison to other developed countries (Barnert et al., 2016). Because adolescence is an important developmental phase in the life-course, it is imperative to better understand the impact that CJS involvement in adolescence has on subsequent outcomes in adulthood.

CJS involvement in the United States is not equally distributed. Large inequities exist with racial and ethnic minorities having disproportionate contact and involvement with the CJS in comparison to white individuals (Barnert et al., 2016; Hockenberry & Puzzanchera, 2018; Puzzanchera, 2019; Rodriguez, 2010; Wakefield & Uggen, 2010). In 2001, Garland introduced the term “mass imprisonment” (often interchanged with mass incarceration), and identified two distinct features about this phenomenon. The first was the overall size of the imprisoned population, and the second was the concentration of imprisonment among specific populations. For example, he noted that “imprisonment becomes mass imprisonment when it ceases to be the incarceration of individual offenders and becomes the systematic imprisonment of whole groups of the population” (Garland, 2001, p. 6). Garland noted the overrepresentation of Black and Latinx individuals within the CJS at the time of his writing; however, North American Indigenous individuals also experience disproportionately high rates of involvement with the CJS (Armstrong et al., 2009), and should be situated within this era of mass imprisonment as well.

The structure of the CJS greatly contributes to inequalities seen throughout the era of mass imprisonment. Existing research shows that inequality is present at all stages of the criminal justice process such as policing and arrests (Daly & Tonry, 1997; Jacobs et al., 2012), sentencing (Pettit & Western, 2004; Rios, 2006; Steffensmeier et al., 1998), incarceration (Wakefield & Uggen, 2010; Western, 2006), and community reentry (Petersilia, 2003). Inequalities seen throughout the CJS are a direct outcome of how policies and laws are written and implemented by criminal justice officials. Particular groups, such as individuals who have lower levels of income and members of minority groups, are often targeted and therefore, have an increased presence within the CJS (Pettit & Western, 2004; Wacquant, 2009), leaving some scholars to argue that mass imprisonment is rooted in classism and racism (Brewer & Heitzeg, 2008). Overall, the disproportionality among low income minority groups within the CJS is not a result of inherent criminality, but a reflection of a system that is designed and functions in a discriminatory and biased manner.

The theme of disproportionality is consistent through each stage of the CJS and within the juvenile justice process (Williams, 2009), and is especially true for Indigenous individuals. Existing studies have varied findings depending on the sample and which stage of the juvenile justice process and offense type is being examined; however, Indigenous youth consistently experience overrepresentation within each stage in comparison to their overall population (Bachman et al., 2009; Sittner Hartshorn et al., 2015). For example, Baek, Roberts, and Higgins (2018) state that “although Indigenous youths represented almost 1% of the national youth population in 2010, they were grossly over-represented in juvenile justice systems and youth arrests” (p. 57). Rountree (2015) goes on to note that even though Indigenous youth only represent 1% of the youth population, tribal youth represent almost half of those youth within the federal system. Furthermore, Cross (2008) highlights that Indigenous youth in comparison to white individuals, are 50% more likely to be subjected to the harshest punishment. Additionally, one consistent theme indicates that Indigenous youth experience higher rates of delinquency adjudication in comparison to other racial and ethnic groups (Freiburger & Burke, 2010; Hockenberry & Puzzanchera, 2018). Freiburger & Burke (2010) found that while adjudication rates were similar among their sample of white, Black, Hispanic, and Native Americans, their Native American participants experienced adjudication at the highest rate of 44%. Indigenous individuals also made up 53% of juveniles that were in the custody of the Federal Bureau of Prisons (BOP) from 1999–2008 (Sickmund & Puzzanchera, 2014). Years of colonialization have left Indigenous individuals around the world socially, economically, and politically marginalized, which increases their risk on encountering the CJS (Cunneen & Tauri, 2019). Their unique position within Anglo-settler-colonial society warrants their direct experiences being the focus of research attention and each of these examples highlights the importance of examining those experiences within criminal justice research.

The effects of colonization have created unique risk factors that place Indigenous individuals at an increased possibility of coming into contact with the CJS, including substance use and mental health issues, delinquent peers, disadvantaged communities (i.e. high rates of poverty and increased community violence), and familial involvement in the CJS (Sittner & Gentzler, 2016). Despite the proliferation of studies that examine involvement in the CJS and the negative impacts on Black and Latinx individuals, there remains sparse research that explicitly focuses on CJS involvement of Indigenous adolescents and resulting adult outcomes (Ulmer & Bradley, 2019). The intersection of unique risks for criminal justice involvement and a discriminatory and biased system create a socially unjust situation for Indigenous individuals. Furthermore, it reinforces the need to place Indigenous individuals at the center of analysis in order to address concerns specific to this population.

In the current study, we investigate Indigenous youths’ involvement in the CJS and the long-term influences on adult outcomes. We begin by presenting two theoretical frameworks, the life-course perspective and Indigenous criminology, to give context and the lens with which we frame the current study. Next, we review the literature related to the negative socioeconomic outcomes of CJS involvement. We then focus on the unique positions that Indigenous people occupy in the United States and Canada to situate our study within their experiences. The main research question guiding the study is: how is involvement in the CJS in adolescence related to adult socioeconomic outcomes among Indigenous youth? We specifically examine outcomes related to socioeconomic status (i.e. educational attainment, employment and income). Furthermore, as Indigenous youth already experience higher levels of socioeconomic inequality within society, does CJS involvement widen the existing disparities? We address these questions utilizing data from the Healing Pathways project, a longitudinal, community-based participatory study of Indigenous youth from the northern Midwest of the United States and southern Canada followed into early adulthood.

Theoretical Framework

Life-Course Perspective

Elder, Modell, and Parke (1993) state “the life-course refers to the age-graded life patterns that are embedded in social institutions and subject to historical change. These patterns are defined by trajectories, which extend across much of the life-course, such as family and work, and by transitions, or short-term changes such as leaving home for school, getting a full-time job, and marrying” (p. 4). While the life-course is subject to historical change as well as variation among different racial and ethnic groups (Graber et al., 1996), there remains consistency in what the life-course should look like, and the roles that one should take on to transition into adulthood. The criteria for adulthood varies, but it is interconnected. For example, Arnett and Tanner (2006) note that accepting responsibility for oneself, making independent decisions, and becoming financially independent are the main benchmarks for adulthood. Graber et al. (1996) cite more specific examples of achieving adulthood such as completing one’s education, gaining full-time employment, supporting and living on one’s own, getting married, and starting a family.

Incarceration can impact the life-course and the transition to adulthood in many negative ways (Huebner, 2005; Loeffler, 2013). The life-course perspective emphasizes the importance of timing of transitions, and disruptions to this timing has the potential to alter trajectories. For example, Gilman, Hill, and Hawkins (2015) state “it could be that youth who might otherwise desist from criminal behavior as they transition to adulthood face a turning point when they are incarcerated, which may alter their opportunity structures during the critically important transition to adulthood and produce negative consequences in both criminal and non-criminal domains” (p. 34). In terms of criminal consequences, involvement in the juvenile justice system is related to later offending in adulthood (Gilman et al., 2015; Sittner & Gentzler, 2016), resulting in more contact with the CJS. As for non-criminal consequences, socioeconomic factors, lack of employment and lower lifetime wages impact the life-course in making it more challenging and/or longer for someone to become financially stable and establish an independent household on their own. Lack of financial stability due to lower rates of employment and lifetime wages as the result of incarceration also influence marriage patterns (Huebner, 2005). Arguably, individuals with criminal records are less desirable in terms of marriage, which results in diminished marriage rates, indicating another facet of how CJS involvement impacts the transitions to adulthood.

Indigenous Criminology

Indigenous individuals in Anglo-settler-colonial societies possess a complex and unique position (Cunneen & Tauri, 2019; Ulmer & Bradley, 2019); therefore, it is imperative that scholars utilize a framework to more adequately understand and situate their experiences. Indigenous criminology provides this conceptual framework to more thoroughly and accurately examine the experiences of Indigenous people within the CJS. The framework centers the experiences of Indigenous people as a focus, and highlights how the historical and continuous colonization of Indigenous people is critical for understanding their current encounters and overrepresentation within the CJS (Cunneen & Tauri, 2016).

The long history of colonization has resulted in social, economic, and political marginalization among Indigenous people which directly impacts their involvement in the CJS, both in terms of issues related to crime and criminalization (Cunneen & Tauri, 2019). The CJS continues to be utilized to reproduce marginalization of Indigenous individuals by removing them from their families, communities, and cultures (Tauri & Porou, 2014), and exposing them to the negative outcomes associated with CJS involvement such as lower levels of educational attainment, full-time employment, and lifetime earnings. Additionally, stereotypes about Indigenous individuals, which are conceived in colonial thought, shape their interactions with the CJS and result in higher rates of stops, searches, and arrests along with harsher sentences (Cunneen & Tauri, 2016). Once in custody, Indigenous individuals often experience more severe uses of force by CJS officials, such as being pepper sprayed, restrained, or isolated (Cross, 2008) which results in higher rates of death among Indigenous individuals in police and correctional custody (Cunneen & Tauri, 2019). Overall, as it is no longer appropriate to be overtly racist and assimilationist, colonization must be deployed in more covert forms, such as removing and incarcerating large amounts of Indigenous individuals from their communities, subjecting them to extremely harsh treatment within custody of the CJS (Tauri & Porou, 2014), while failing to provide adequate services that are culturally competent (Primm et al., 2005; Shepherd & Phillips, 2016).

Literature Review

Criminal Justice System Involvement and Adult Outcomes

Involvement in the CJS has many negative consequences that represent an immediate turning point for individuals and alter their life-course trajectory in the longer term. One of the most pronounced and studied consequences is socioeconomic status, generally understood through educational attainment and employment opportunities that then manifest through income. CJS involvement and SES characteristics such as education, employment, and income are intertwined in complex ways. First, due to discriminatory application of policies and laws, individuals who have lower levels of education, employment, and income are more likely to encounter the CJS (Wakefield & Uggen, 2010). Second, once an individual has been involved in the CJS, numerous consequences occur that often negatively impact their education, employment, and income opportunities (Pettit & Western, 2004; Western, 2002).

In terms of education, the majority of individuals who come into contact with the CJS have lower levels of education in comparison to the rest of the general population (Wakefield & Uggen, 2010). And due to consistent budget cuts, once incarcerated, individuals are less likely to participate in educational programs (Petersilia, 2003). CJS involvement also disrupts their current educational trajectory (Sittner & Gentzler, 2016) by removing individuals from their present educational institutions and placing them in facilities that have limited educational opportunities. This disrupts the transition to adulthood by potentially lengthening the time it takes one to complete their education or by increasing the risk of dropping out of school. Furthermore, education is closely linked to employment and income, as those with higher levels of education typically earn more income through their employment.

Similar findings exist when examining the impact of incarceration on employment. First, those with lower levels of participation in the labor market are at an increased risk of coming in to contact with the CJS (Wakefield & Uggen, 2010). Additionally, incarceration disrupts current employment status, by removing an individual from a job they presently occupy. Second, once incarcerated, they are less likely to participate in job training programs (Petersilia, 2003). And finally, once released they have difficulty securing employment (Pager, 2003). In combination, lack of education and employment lead to a reduction in lifetime earnings (Western, 2002).

While youth and adults can both experience the aforementioned negative consequences of CJS involvement, juveniles are impacted in unique ways. Incarcerated youth have a right to obtain some form of education while incarcerated (Gagnon et al., 2009); however, the quality of education provided may be limiting and their educational trajectories are still disrupted. Being involved in the juvenile justice system has been shown to decrease rates of high school graduation which impact employment opportunities and income (Robles-Ramamurthy & Watson, 2019). Lack of employment and income lead to other negative consequences such as homelessness and eviction (Remster, 2019), which juvenile justice involved-youth experience at higher rates. And finally, rates of recidivism are higher among individuals who have been detained, which results in increased likelihood of further involvement in the CJS and compounded negative consequences (Robles-Ramamurthy & Watson, 2019; Sittner Hartshorn et al., 2015).

Numerous studies have highlighted that CJS involvement is largely unequal, with racial and ethnic minorities experiencing overrepresentation. However, most studies focus on white/Black comparisons or white/Black/Latinx comparisons; fewer studies include a direct examination of Indigenous individuals (Ulmer & Bradley, 2019). Additionally, despite the proliferation of studies on the consequences of CJS involvement, there remains sparse research that explicitly focuses on Indigenous youth. Furthermore, colonization among Indigenous individuals has resulted in lower levels of education, and full-time employment which impacts lifetime wages, each of which increases the risk of encountering the CJS as we previously mentioned (Cunneen & Tauri, 2019). This once again highlights the importance of situating Indigenous experiences at the center of analysis.

Unique Experiences of Indigenous Individuals

Race impacts CJS involvement, with large disparities among racial and ethnic minorities. Disparities accumulate throughout the juvenile justice process indicating that larger levels of inequality are seen as an individual moves through the system (Rodriguez, 2010). Some scholars have noted that disparities within the juvenile justice system are a result of racial and ethnic minority individuals being treated differently in comparison to white individuals. This “differential treatment” is a direct consequence of stereotypes and assumptions that CJS officials have about racial and ethnic minority individuals (Rodriguez, 2010). Due to continuous colonization, stereotypes and assumptions related to Indigenous individuals manifest in unique ways impacting how they are treated by CJS officials (Cunneen & Tauri, 2016). Additionally, colonization has left Indigenous individuals in an extremely marginalized position in terms of social, economic, and political power (Cunneen & Tauri, 2019) making it less likely for them to be able to resist discriminatory CJS practices (Bachman et al., 2009).

Indigenous experiences within the CJS remain largely invisible in extant research. However, the history and position of Indigenous peoples in society make their experiences unique in comparison to other racial and ethnic groups (Ulmer & Bradley, 2019). The long history of colonization has resulted in high rates of poverty, mental health issues, and substance use/abuse (Cunneen & Tauri, 2019), all of which have been shown to increase the likelihood of coming into contact with the CJS. Additionally, Indigenous individuals experience higher rates of violent victimization, which also increases contact with the CJS. For example, violence against Indigenous individuals is over twice the U.S. national average (Perry, 2004). Therefore, colonization, oppression, and marginalization subject Indigenous people to higher rates of contact with the CJS, which then leads to further disadvantage because of the collateral consequences associated with CJS involvement.

A variety of explanations have been advanced to explain the higher rates of involvement of Indigenous people with the CJS. Broadly speaking, community structure may largely contribute to these higher rates. Community structure in this respect consists of the concentration of disadvantage in many Indigenous communities, including high levels of poverty and low levels of education and employment (Armstrong et al., 2009; Sarche & Spicer, 2008). Many Indigenous communities also experience high rates of violent crime (Morgan & Truman, 2018), which have been linked to increased CJS involvement. Due to concentrated disadvantage within Indigenous communities, individuals are more likely to have a friend, family member, or know someone who has experienced incarceration, which can contribute to being involved in the CJS (Farrington et al., 2001). Additionally, Indigenous communities often experience high rates of mental health issues, along with substance use/abuse (Sarche & Spicer, 2008), both of which increase the likelihood of coming in to contact with the CJS system. Furthermore, stereotypes created in colonial thinking shape how Indigenous individuals are policed and handled within the CJS (Bachman et al., 2009; Cunneen & Tauri, 2016). For example, Indigenous youth have the highest rates of liquor law violations (Hockenberry & Puzzanchera, 2018), despite self-reported levels of drinking by high school seniors that are similar among various racial and ethnic groups (Sickmund & Puzzanchera, 2014)

Overall, Indigenous individuals start out in a marginalized position due to the historical colonization they have experienced and as a result have diminished opportunities and encounter situations that put them at an increased risk of experiencing CJS involvement. Moreover, they continue to be treated more harshly by the CJS all of which contributes to their overrepresentation within the CJS. In combination, this results in Indigenous individuals being subjected to compounding effects of CJS involvement. Therefore, there continues to be a cycle of inequality (Wakefield & Uggen, 2010) among Indigenous individuals that is perpetuated and effectively solidified through the CJS.

The Current Study

Although a growing body of literature supports the claim that justice involvement leads to worse outcomes, claims of cause and effect can be problematic. Making causal arguments that involvement in the CJS, and arrest more specifically, leads to negative outcomes is difficult to do with observational data because those factors that place individuals at risk of being arrested likely also increase the chances of negative outcomes compared to those who are not at risk of arrest. Furthermore, adolescents who are arrested may be qualitatively different than those who are not arrested. In other words, the association between arrest and negative outcomes is potentially due to a selection effect (i.e, who is more likely to be arrested) and to confounding variables associated with both arrest and negative outcomes. This bias could be addressed using experimental data, but studies in which adolescents are randomly assigned to the arrest and control conditions are widely regarded as unethical. Propensity score matching models have been commonly used in criminological research (Apel & Sweeten, 2010) and help, in part, to overcome the limits of observational studies by mimicking the randomization process and minimizing the effects of selection and confounding.

The current study makes two main contributions to the existing literature on criminal justice involvement inequities and socioeconomic status for North American Indigenous people. First, we examine links between arrest in adolescence and SES in early adulthood using longitudinal data spanning eight years in adolescence with outcomes in early adulthood. Arrest serves as a critical turning point in the life-course that can disrupt the successful transition to adulthood. Indigenous youth experience unequal treatment in the justice system as well as unique and disparately numerous risk factors for being arrested. Because of the substantial socioeconomic disparities experienced in Indigenous communities, it is important to investigate how criminal justice involvement contributes to these disparities. Finally, we apply propensity score matching to address selection bias in who is likely to be arrested and to assess the causal role of arrest in adolescence on adult SES.

Data and Methods

Procedures

The data used in the current study come from the Healing Pathways (HP) project, a longitudinal, community-based participatory research (CBPR) study of Indigenous youth and their caregivers. The youth were from one of the largest Indigenous cultural groups in North America and resided in the upper Midwest of the United States and Ontario, Canada (the name of the cultural group is withheld to respect confidentiality agreements with the communities). The first phase of the study, from 2002–2010, consisted of eight waves of interviews conducted annually. At the first wave (n=735), adolescents were between 10 and 12 years of age (mean age =11.1 years). In the second phase (2017-present), the original participants, now young adults (mean age = 26.3 years at Wave 9) were re-contacted and interviewed for three more years, resulting in 11 waves altogether. For both phases, tribal resolutions from the eight communities were obtained in which the study goals, interview procedures, and sampling protocol were approved. Community research councils (CRCs) were created at all sites and were involved in establishing study goals, adapting and developing measures, and disseminating results back to their communities. The CRCs were also responsible for hiring and managing interview staff. All interviewers were enrolled members or spouses of enrolled members of the tribal communities, and all underwent intensive training annually in personal interviewing procedures and data protections. CBPR principles informed all aspects of the study. A more detailed description of the study and procedures is available in (Whitbeck et al., 2014).

The original study included 735 adolescents, with a response rate of 74%, and high retention rates throughout the study, ranging from 96% at Wave 2 to 82% at Wave 8. Nearly all of the adolescents (98%) participated in at least three waves, and 94% participated in at least four. At baseline, the sample was 50.2% female with a mean per capita family income of US $5,725. Approximately 86% lived on reservation or reserve land, and 9.5% resided in more remote locations. In the follow-up study, retention rates in waves 9 and 10 were 61.6% and 69.8%, respectively. Over three-quarters (77%, n=566) completed both follow-up waves, which is the focus of the current study. The young adult sample demographics changed somewhat from baseline, but the only significant difference was that slightly more women than men were interviewed, with 54.1% of the follow-up Wave 9/10 sample being female.

Measures

Dependent variables.

We measured SES in early adulthood (Waves 9 and 10) with income, education, and employment. Highest level of education was recoded into three binary variables: less than high school, high school or GED, and any post-secondary education. Current employment status was used to construct two binary variables: employed full-time (FT) = 1 and unemployed = 1. Low income was measured first by asking the young adults to indicate their current income in increments of $5,000 if under $24,999, and increments of $10,000 if above $24,999, and then dichotomizing it into lowest quartile = 1, else = 0. For all SES outcomes, the most recent interview was selected (i.e., if interviewed in both waves 9 and 10, wave 10 was selected; if interviewed in only one wave, that wave was selected).

Treatment.

Adolescents were asked if they had ever been arrested (Wave 1) or arrested in the past year (Waves 2–8). To ensure proper temporal ordering with the Wave 1 covariates used to calculate the propensity score, we only used past-year arrest data from Waves 2–8, and excluded from the analysis any adolescents who reported ever being arrested at Wave 1 (n=7). Adolescents who reported being ever arrested in at least one wave were coded as 1; adolescents who never reported being arrested were coded as 0.

Analytic Strategy

We applied propensity score matching (PSM) using psmatch2 (Leuven & Sianesi, 2003) in Stata 15 to investigate the impact of being arrested during adolescence on early adult socioeconomic status, specifically education, income, and employment status. PSM was used to match adolescents who were arrested during adolescence with adolescents who had not been arrested based on a broad selection of Wave 1 covariates theoretically and empirically thought to be associated with arrest (Sittner & Gentzler, 2016) and with SES outcomes. Following Apel and Sweeten (2010), all of the covariates temporally precede treatment (i.e., adolescent arrest in Waves 2–8) and the early adult SES in Wave 9. These covariates were drawn from multiple domains of life, including family (parent education, family per capita income, single parent household, parental monitoring, parent arrest, exposure to criminal victimization), community (residing on/off reservation, remote location, exposure to community violence), school (attachment to school), and individual (age, sex, frequency of drinking alcohol and using marijuana, deviant peers, delinquent behavior). A full list of the covariates and their descriptive statistics are shown in Table 1.

Table 1.

Descriptive Statistics for Treatment, Outcomes, and Covariates

	Mean	s.d.	Range
Treatment
Arrested	0.46		0,1
Early Adult Outcomes
Did not finish high school	0.25		0,1
High school graduate	0.36		0,1
Postsecondary education	0.39		0.1
Employed FT (n=558)	0.43		0.1
Unemployed (n=558)	0.37		1,0
Low income (n=507)	0.26		0,1
Community a
Off Reservation/Reserve (n=555)	0.14		0,1
Remote	0.10		0,1
Community Violence (n=540)	0.95	0.54	0–2
Family a
Parent Education (n=555)	2.48	0.86	1–5
Per Capita HH Income in Thousands of $ (n=548)	5.73	4.94	.25–75
Welfare (n=555)	0.38		0,1
Single Parent	0.27		0,1
Parental Monitoring (n=557)	1.51	0.32	.5–2
Parent Arrest	0.10		0,1
Criminal Victimization	0.18		0,1
School a
Attachment to School (n=557)	6.00	1.46	0–7
Personal a
Female Sex	0.54		0,1
Age	11.05	0.81	9–13
Drinking Frequency (n=554)	0.07	0.33	0–3
Marijuana Frequency (n=556)	0.10	0.57	0–5
Deviant Peers (n=555)	0.62	0.61	0–3
Delinquent Behavior	2.75	3.40	0–19

Note: n = 559 unless indicated otherwise.

^aBaseline covariates.

PSM matches each control group case to a treatment group case using the smallest differences in propensity scores. The control group represents a counterfactual for the expected SES of the treatment group had they not been arrested during adolescence. Following the recommendations of Stuart and Rubin (2008), we first selected the matched samples without the outcomes to ensure that the propensity scores were not biased regarding the outcomes we wanted to investigate. We examined different matching methods and caliper adjustments including nearest neighbor with and without replacement (results available upon request). Success of the matching strategy was assessed using Rubin’s B <25% and Rubin’s R between 0.05 and 2 (Rubin, 2001), and standardized bias <10% (Morgan, 2018). Rubin’s B is the standardized difference of the means and Rubin’s R is the ratio of the variances between the arrested and not arrested groups. Epanechnikov kernel matching with bandwidth =.05 was selected because it had the lowest mean bias (3.7%) and lowest Rubin’s B (21.3%), and Rubin’s R within the desired range (1.83) of those methods examined, as well as the largest number of cases on common support (467 out of 474 cases). The next-best fitting model used three nearest-neighbor matching and no caliper, but Rubin’s B exceeded the threshold (i.e., >25%), although the results were substantively identical to the selected model. Using the selected matching criteria, 85% of the sample was matched (treatment group n = 222; control group n = 252). Propensity scores in our analysis ranged from 0 to 1; in this case, 0 = certainty of not being arrested and 1 = certainty of being arrested. We then repeated the matching with the three outcomes included to assess the average treatment effect (ATE) and the average treatment effect of arrest on the treated (ATT). ATE is the expected effect of adolescent arrest on adult SES for any individual, whereas ATT is the expected effect of adolescent arrest on adult SES for individuals who reported being arrested.

The final stage of analysis was to further explicate the chain of causality between arrest and SES by accounting for the role of the different SES indicators. In particular, education contributes to both employment status and income level, and employment in turn determines income. Thus, because the indicators of SES are themselves causally related, we conducted a series of regression analyses to assess whether arrest remains associated with subsequent SES after also accounting for separate adult SES indicators.

Results

The descriptive statistics for the outcome and treatment variables are provided in Table 1. Less than one-half (46%) of the adolescents reported being arrested at least once between waves 2 and 8 (lifetime arrests at wave 1 were excluded to establish temporal ordering between covariates and the treatment). Socioeconomic outcomes were measured at waves 9 and 10. One-quarter of the young adults had not finished high school, 36% graduated or received their GED, and 39% had at least some postsecondary education. Forty-three percent of the young adults were employed full-time, but 37% were unemployed.

Covariate balance statistics are shown in Table 2. Means and proportions of the covariates were compared pre- and post-match for the arrested and not arrested groups using t-tests or χ² tests. In the unmatched sample, the group that did reported being arrested in adolescence was less likely to have more educated parents and had lower per capita household incomes, reported less parental monitoring and poorer attachment to school. That group also was more likely to have a parent who was arrested, to use alcohol and marijuana more frequently, to have more deviant peers, and to engage in more delinquent behavior than the non-arrested group. After matching, none of the comparisons reached statistical significance and the SB was below 10% for all of the covariates, indicating that the two groups were sufficiently similar on the covariates.

Table 2.

Covariate Balance Statistics for Unmatched and Matched Samples

	Unmatched (n=559)			Matched (n=474)
Baseline Variables	Arrested (n=259)	Not Arrested (n=300)	SB (%)	Arrested (n=222)	Not Arrested (n=252)	SB (%)	% Bias Reduction
Community
Off Reservation/Reserve	0.12	0.15	−7.4	0.13	0.13	0.0	99.6
Remote	0.07	0.11*	−14.0	0.07	0.09	−8.5	39.0
Community Violence	0.97	0.94	6.6	0.96	0.96	−0.2	97.2
Family
Parent Education	2.31	2.63***	−37.6	2.35	2.32	3.5	90.6
Per Capita HH Income	4.72	6.24***	−39.2	4.82	5.08	−6.7	83.0
Receive welfare	0.51	0.27***	50.5	0.49	0.48	1.1	97.8
Single Parent	0.31	0.25	12.6	0.31	0.30	2.7	78.7
Parental Monitoring	1.46	1.54*	−23.8	1.47	1.46	4.0	83.2
Parent Arrest	0.15	0.07**	27.6	0.14	0.11	7.9	71.3
Criminal Victimization	0.21	0.16	11.5	0.21	0.22	−2.1	81.5
School
Attachment to School	5.85	6.20*	−24.8	5.91	5.78	9.2	63.0
Personal
Female Sex	0.52	0.59*	−13.9	0.51	0.50	2.4	82.6
Age	11.15	10.94*	26.5	11.11	11.04	8.2	69.1
Drinking Frequency	0.12	0.02**	27.9	0.09	0.10	−2.6	90.6
Marijuana Frequency	0.19	0.03**	27.7	0.09	0.11	−3.4	87.7
Deviant Peers	0.76	0.48***	51.9	0.68	0.71	−7.1	86.4
Delinquent Behavior	3.74	1.94***	53.8	3.35	3.37	−0.5	99.0
Rubin’s B (%)	90.4			21.3
Rubin’s R	1.83			1.08
Mean SB (%)	25.4			3.7

Note: Mean/proportion comparisons pre- and post-match conducted with t-tests/chi-square tests (* p<.05; ** p<.01; *** p<.001). Matching was assessed using Rubin’s B <25%, Rubin’s R 0.05–2, and Standardized Bias (SB) <10%. Epanechnikov kernel matching with bandwidth =.05.

Table 3 displays differences in the risk of experiencing the early adult SES outcomes for the treatment (arrested) and control (not arrested) groups. Bootstrapped ATT and standard errors were estimated to reduce bias. Adolescent arrest had a significant treatment effect on all of the outcomes except for high school graduation/GED. A history of arrest in adolescence made it more likely that these Indigenous young adults would not finish high school and less likely they would have any postsecondary education. Adolescent arrest also made it less likely they would be employed full-time in early adulthood and more likely they would be unemployed. Further, adolescent arrest made it more likely that the young adults would be in the lowest quartile of income.

Table 3.

Risk Comparisons of SES in Early Adulthood by Arrested Versus Not Arrested (n=474)

	Arrested	Not Arrested	Difference (ATT)	s.e.	z	ATE
Did Not Finish High School	0.34	0.21	0.13	0.07	2.64	0.13
HS Graduate/GED	0.37	0.31	0.05	0.09	−0.77	0.03
Any Postsecondary Education	0.28	0.47	−0.19	0.07	−1.78	−0.16
Employed FT	0.32	0.52	−0.20	0.09	−1.83	−0.18
Unemployed	0.52	0.29	0.24	0.07	3.62	0.22
Low Income	0.38	0.17	0.22	0.06	4.14	0.20

Note. ATT = Average treatment effect on the treated; ATE = Average treatment effect. Bootstrapped standard errors were estimated for the matched ATT sample statistics using Epanechnikov kernel matching with bandwidth =.05.

We conducted supplemental logistic regression analyses to examine the associations between adolescent arrest and young adult SES while controlling for the baseline covariates, shown in Table 4, M1. Because of sample size requirements in logistic regression (Hosmer & Lemeshow, 2000; Peduzzi et al., 1996), we selected a reduced list of covariates with the largest standardized bias pre-matching. Models with all baseline covariates were also estimated and the results were virtually unchanged. These results were substantively very similar to the PSM analyses: being arrested in adolescence was associated with higher odds of not finishing high school, being unemployed, and having low income, and lower odds of postsecondary education and full-time employment. In neither the PSM analyses nor the logistic regressions was adolescent arrest associated with finishing high school.

Table 4.

Adjusted and Unadjusted Odds Ratios of Early Adult SES

	Did Not Finish High Schoola	Any Postsecondary Educationa	Employed FTb		Unemployedb		Low incomec
	M1	M1	M1	M2	M1	M2	M1	M2
Did Not Finish High School d				0.54*		1.68*		1.21
Any postsecondary Educationd				1.94**		0.50**		0.47*
Employed FTe								0.14***
Unemployede								2.16*
Ever Arrested	2.17**	0.49**	0.53**	0.61*	2.47***	2.16***	2.83***	2.04*
Parent Education	0.64	1.59***	1.26*	1.14	0.75*	0.83	1.17	1.60**
Per Capita HH Income	0.92*	1.08**	1.05*	1.04	0.96	0.97	0.88**	0.89**
Receive welfare	1.26	0.62*	0.87	0.95	1.04	0.96	1.42	1.29
Parental Monitoring	0.55	1.97*	1.53	1.36	1.20	1.38	1.26	1.63
Parent Arrest	2.51**	0.52	0.98	1.21	0.85	0.70	0.83	0.73
Attachment to School	0.97	1.04	0.96	0.95	1.00	1.01	0.99	0.95
Age	0.78	1.26	1.43**	1.38**	0.89	0.94	1.09	1.34
Drinking Frequency	1.00	0.81	0.73	0.74	1.45	1.45	1.86*	1.74
Marijuana Frequency	0.83	0.97	0.95	0.93	1.00	1.02	0.83	0.76
Deviant Peers	1.03	0.91	0.92	0.94	1.35	1.34	1.05	1.01
Delinquent Behavior	1.02	1.00	0.97	0.98	1.01	1.00	0.99	0.96

Note:

^an=539.

^bn=538.

^cn=489.

^dHigh school graduate/GED is reference category.

^eOther employment status is reference category.

^*p<.05

^**p<.01

^***p<.001

We also examined the linkages between the early adult SES outcomes, in particular the relationship between education and employment, and education, employment, and income to better understand how a history of arrest affects later SES. After controlling for adolescent arrest, education was significantly associated with employment in early adulthood, and both education and employment were significantly associated with low income. However, the effect of arrest on employment was weaker after accounting for education, and the effect of arrest on low income was diminished after accounting for the other early adult SES outcomes. For example, without considering education or employment status, the odds of being in the lowest quartile of income almost three times higher for those with a history of arrest (O.R. = 2.83, p<.001), but were only two times higher when controlling for education and employment (O.R. = 2.04, p<.05).

Discussion

Although Indigenous youth are understudied in research examining the consequences of criminal justice system involvement, available evidence suggests that justice involvement for Indigenous youth carries significant consequences for later health, family stability, substance use/abuse, and especially relevant for the current study, socioeconomic status. The sample of Indigenous young people in the current study, as a group, reported substantial socioeconomic disadvantage, similar to Indigenous peoples across the United States and Canada (Ogunwole, 2006; Statistics Canada, 2006). In addition, 45% of the sample in the study reported being arrested at least once during the first eight waves of the study, spanning ages 10 to 19 years. The primary goal of the current study was to understand how arrest during adolescence affects subsequent SES for our sample. Using propensity score matching, we found that being arrested at least once in adolescence was associated with significantly less education and income and lower rates of full-time employment in young adulthood (mean age = 26.3 years). Our results are mostly consistent with a large and growing body of research on the collateral consequences of involvement in the CJS. Although we only examined arrest, which is just one facet of justice involvement, we demonstrated that being arrested contributes to widening the socioeconomic inequities already experienced by many Indigenous young people.

Our findings also contribute to research that focuses on the life-course perspective and Indigenous criminology in many ways. First, our findings show that youth arrested in adolescence were less likely to achieve certain markers of adulthood such as full-time employment. Arrested youth also had lower levels of education and income which can impede the achievement of other indicators of adulthood. For example, with lower levels of education and income it can be more challenging for individuals to become financially stable and establish an independent household on their own. Very few studies include a sample of Indigenous peoples, and even fewer only focus on Indigenous peoples. Therefore, the uniqueness of our sample facilitates the use of a distinctive framework when analyzing our research questions. We employed Indigenous criminology to critically examine the overrepresentation of Indigenous peoples within the CJS. Our sample further shows that almost half (45%) of participants had been arrested a least one time during the study period, suggesting that certain tenants of Indigenous criminology manifest within our sample, such as over policing and harsher punishment (Cunneen & Tauri, 2016).

Culturally relevant and informed prevention programs, particularly those developed through partnerships with Indigenous communities, show promise for reducing substance use and suicide (Noe et al., 2003; Whitbeck et al., 2012); however, such programs for Indigenous people in the CJS are few in number, lack rigorous evaluation, and are predominantly located in Australia and New Zealand. The available evidence focused on existing programs points to lower recidivism rates for those enrolled, as well as better retention and more program engagement (Clough et al., 2008; Gutierrez et al., 2018). As Gutierrez et al (2018) noted in their meta-analysis of treatment programs for Indigenous offenders, “it seems plausible that treatment programs that are more culturally compatible – in terms of language, format, and regional or historical context, for example – with their target group are indeed more accessible and applicable, and therefore more effective” (p. 342).

Moreover, much more work is needed to address the biases within the CJS that contribute to the overrepresentation of marginalized youth. Considerable effort has been taken to identify the causes of overrepresentation of Indigenous people in the CJS in Canada and, to a lesser extent, the United States, but solutions are inconsistently applied (Office of the Correctional Investigator, 2016). Bias in the CJS compounds the intergenerational effects of colonization, and itself is a consequence of colonization. Remedying these consequences requires changes in how CJS policies are enacted, in this case at the stage of arrest, as well as larger structural changes to address significant inequities in income, education, and employment for Indigenous people. Cunneen (2011) argues that the system itself needs to be changed by decolonizing the relationship between the CJS and Indigenous people. Efforts should be made to reduce the use of arrest as a primary means of addressing adolescent misbehavior. Given the high rates of re-offending that have been documented elsewhere in the literature (Jones et al., 2006), the potential deterrent effect of arrest on future misbehavior must be considered with respect to the consequences arrest has on future socioeconomic status.

Limitations and Future Directions

The strengths of the current study should be evaluated with a few limitations in mind. First, we utilized self-reported measures of arrest and did not have access to official arrest records. However, self-reports are widely used and generally regarded as valid (Thornberry & Krohn, 2000). Relatedly, the adolescents were not asked for information about incarceration and we were unable to examine whether further penetration into the justice system is more impactful for later SES, as suggested by other studies (Apel & Powell, 2019; Dennison & Demuth, 2018).

A second limitation concerns the generalizability of our results. The sample represents one of the largest Indigenous cultural groups in North America, but the results of the current study should be applied cautiously to young people from different cultural groups. Because Indigenous people are largely absent from criminological research, more studies with different cultural groups is critical. The sample also resides on or near rural reservations, and generalizing to urban Indigenous youth from the same cultural group should also be done cautiously.

A final limitation concerns the use of propensity score analysis. PSM methods overcome some of the weaknesses of traditional, regression-based tests of causality, yet caution is still warranted because “the strength of the methodology to allow for causal inference rests on the availability of the data to include all key covariates driving selection” (Loughran et al., 2015, p. 635). PSM cannot account for unobserved variables, and, although we included a large variety of self- and parent-reported covariates that have previously been examined as risk factors for arrest, there may be important explanatory variables that were not available. For example, we do not have community-level covariates such as neighborhood disadvantage or other contextual variables associated with CJS involvement. Although covariate balance between the control and treatment groups was achieved, Shadish (2013) cautions that “balance is necessary, but not sufficient, for bias reduction” (p. 135). Further, the covariates associated with adolescent arrest may change over the course of adolescence (Sittner & Gentzler, 2016). The covariates in the current study were drawn at Wave 1 and temporally preceded the arrest treatment condition, but future research on time-varying predictors of arrest propensity would be valuable.

In addition to research on time-varying predictors in calculating propensity scores, more attention should be paid to more outcomes. The consequences of CJS involvement for mental, physical, and behavioral health of Indigenous young people are understudied. Our study focused on SES, which is inarguably an important aspect of the social determinants of health and wellbeing. However, CJS involvement for Indigenous people is an important social determinant that likely exacerbates many existing inequities. Further, there are other indicators of adulthood that are also important, such as family formation and community involvement that may be deleteriously affected by early arrest. Future research should also consider the multiple contexts of justice involvement. For example, as noted in the Indigenous criminology literature (Ulmer & Bradley 2019), Indigenous individuals can be subjected to tribal, state, or federal jurisdictions depending on tribal agreements with justice agencies and aspects of the case (i.e. crime committed, location, and identity of defendant and victim(s)).

One final limitation concerns the ethnic homogeneity of the sample. Because this study sought to understand how arrest impacts later SES for Indigenous early adults and the sample only included Indigenous people, we could not isolate the effect of ethnicity on CJS involvement. Future research could utilize PSM methods with a sample of Indigenous and non-Indigenous youth to examine how arrest differentially impacts SES by ethnicity and contributes to widening SES disparities observed in many Indigenous communities.

Conclusion

The effects of colonization continue to be felt in Indigenous communities, including the increased likelihood of contact with the CJS. Once contact has occurred, numerous consequences follow that can ripple across the life-course. Here we demonstrated the consequences of CJS involvement in adolescence on socioeconomic status in early adulthood. We found that being arrested, in comparison to not being arrested, resulted in lower levels of education, employment, and income, all of which can impact other aspects of health and wellbeing. Given that Indigenous people are over-represented in the justice system, both in adolescence and in adulthood, the consequences of justice system involvement can attach to large portions of the Indigenous population in many communities. Therefore, it is critical to examine not only how Indigenous individuals are negatively impacted by CJS policies and procedures, but work to alleviate the social harms caused by the structure of the CJS.