Abstract
The current study draws on two cohorts of African American youths from the Project on Human Development in Chicago Neighborhoods, Longitudinal Cohort Study to examine whether perceived racial discrimination directly and indirectly affects juvenile delinquency. The analyses reveal that racial discrimination may foster offending by increasing (1) the likelihood that African American youths will drop out of school and (2) the degree to which they associate with delinquent peers. Evidence supporting the pathway between racial discrimination, associating with delinquent peers, and offending was found after introducing controls for demographic, social, and individual trait factors. In a society that remains racialized, it thus appears that a full explanation of African Americans’ offending should take into account the ways in which racial subordination may place African American youths on pathways that lead toward criminal involvement.
An emerging body of research recognizes that an understanding of African American offending must be grounded in the everyday experiences of being Black in a racialized society. Notably, the overwhelming majority of African Americans report facing racial discrimination running the gamut from omnipresent “microaggressions” (i.e., daily life hassles that are attributed to racial discrimination) to poignant encounters such as being racially profiled by the police or being denied unemployment because of their skin color (Geller, Fagan, Tyler, & Link, 2014; Sellers, Copeland-Linder, Martin, & Lewis, 2006; Wong, Derthick, David, Saw, & Okazaki, 2014). The ubiquity of these experiences means that African Americans as young as 6 years of age understand and apply racist stereotypes and expect to experience racial discrimination by the age of 8 (Hope, Skoog, & Jagers, 2015; Rowley, Burchinal, Roberts, & Zeisel, 2008).
The available research is also replete with evidence that racial discrimination has deleterious consequences for African Americans, including physical health-related issues (e.g., higher blood pressure), mental health issues (e.g., substance abuse), increases in anger and depression, internalizing and externalizing behaviors, and elevated incidences of offending (Burt & Simons, 2015; Gibbons et al., 2012; Parham & McDavis, 1987; Riina, Martin, Gardner, & Brooks-Gunn, 2013; Simons et al., 2002; Unnever, 2014). Indeed, a meta-analysis of 134 samples found that perceived discrimination has a significant negative effect on both mental and physical health and is related to participation in unhealthy behaviors and nonparticipation in healthy behaviors (Pascoe & Richman, 2009). Thus, the research reveals that the more African Americans experience racial discrimination, the greater their likelihood of exhibiting negative behaviors such as criminal activity.
Although scholars recognize that a significant relationship exists between racial discrimination and offending, the reasons for such an association remain less clear (Unnever, Cullen, Mathers, McClure, &, Allison, 2009). A systematic attempt to address this issue, however, is present in Unnever and Gabbidon’s (2011) race-specific theory of African American offending. This theory begins with the observation that racial discrimination is not situational but systematic within racially stratified societies such as the United States. Because it is systemic, African Americans have developed a racialized understanding of what it means to be Black. A component of this racialized understanding is the awareness that they will encounter racial discrimination because of their skin color. Thus, exposure to racial discrimination is chronic: It emerges in childhood, is experienced throughout adolescence, and persists into adulthood.
A core component of Unnever and Gabbidon’s (2011) theory of African American offending is the hypothesis that experiencing racial discrimination weakens ties (i.e., social bonds) with conventional institutions. This hypothesis flows from their argument that African Americans perceive that racial discrimination emanates from racist institutional arrangements. This means that African Americans will encounter both individual acts of racism and those that have become institutionalized. Consequently, African Americans should have weaker bonds with those institutions.
In this article, we test a core component of Unnever and Gabbidon’s (2011) theory of African American offending, which asserts that weak social bonds are a key pathway between racial discrimination and misconduct among African American youths. Analyzing a longitudinal sample of African American youths from the Project on Human Development in Chicago Neighborhoods, Longitudinal Cohort Study (PHDCN-LCS), we assess whether perceived racial discrimination is related to youth misconduct because it increases the likelihood that African American youths drop out of high school. The analysis also explores whether racial discrimination results in higher offending by leading African American youths to associate with deviant peers, a known risk factor for criminal involvement.
Racial Discrimination and Pathways to African American Offending
Racial Discrimination and Offending
Scholars have explicitly tested whether racial discrimination is a salient cause of African American offending. Analyzing the Family and Community Health Study’s (FACHS) longitudinal data, Simons et al. (2006) selected only those boys who reported no involvement in delinquency in 1997 (Wave 1) and then examined the extent to which racial discrimination predicted offending in 1999 (Wave 2). Their results showed that racial discrimination is associated with the onset of violent delinquent behavior among males. Furthermore, in a previous study analyzing the FACHS data, Simons, Chen, Stewart, and Brody (2003) tested for the possibility of reverse causation. That is, whether offending causes African American youths to perceive racial discrimination or whether racial discrimination precedes the onset of delinquent behavior. Using the first two waves of the FACHS data for both girls and boys, Simons et al. found that there was a significant path from racial discrimination to problematic behaviors but not for the reverse pathway (Simons, Chen, Stewart, & Brody, 2003, p. 842). In short, the evidence suggests African American youths who perceive racial discrimination are at risk for subsequent offending.
While the evidence base shows that racial discrimination places African Americans at risk for offending, the possible pathways by which it elevates the likelihood of African American offending are less well understood (Simons & Burt, 2011; Unnever, 2014). Below we outline a pathway by which racial discrimination may lead to African American youth offending that, to the authors’ knowledge, has not been previously examined. We test whether racial discrimination predicts dropping out of school and associating with deviant peers, which in turn predicts delinquent behavior.
Racial Discrimination, Dropping Out of High School, and Associating With Deviant Peers
A basic assumption of Unnever and Gabbidon’s (2011) theory of African American offending is that racial discrimination weakens African Americans’ ability to develop strong bonds with conventional institutions. A large segment of their theory focuses on how racial discrimination, in all of its variegated forms, undermines the ability of African American youths to build strong bonds with their schools and education. They focus on school failure because the available research reveals that it is a robust predictor of offending (Noble, 2006; Wang & Fredricks, 2014). Western (2006) notes that roughly 60% of African American males who drop out of high school spend time in prison by age 34 (Western, 2006, p. 27). Thompson and Gregory (2011) add that the problem of dropping out of high school is acute for African American youths because they state that nearly 50% of African Americans entering high school do not graduate with a traditional high school diploma. In short, Unnever and Gabbidon (2011) posit that racial discrimination is a significant antecedent risk factor for the school-to-prison pipeline among African American youths (Coggshall, Osher, & Colombi, 2013; see generally, Rocque & Paternoster, 2011).
Unnever and Gabbidon (2011) contend that it is difficult for African American youths to bond with their schools and education when they perceive that those institutions have historically and currently discriminate against them. Cross (2003, p. 72) captures this systemic racism as he argues that “southern society, with the cooperation of the North, made extraordinary efforts to blunt, mangle, and where possible, destroy Black achievement motivation.” Previous studies show that racial discrimination is associated with (lower) school performance, (less) engagement, and (lower) long-term educational goals (Dotterer, McHale, & Crouter, 2009; Sellers & Shelton, 2003; Smalls, White, Chavous, & Sellers, 2007; Thompson & Gregory, 2011; Wang & Huguley, 2012). Unexpectedly, though, we found no available research that specifically investigates whether racial discrimination increases the likelihood that African American youths will drop out of high school. This is one of the gaps in the literature that will be addressed by the present study.
Unnever and Gabbidon (2011) draw on the classic study by Steele and Aronson (1995) on stereotype threats—which they argue constitute a form of racial discrimination—to illustrate how racial discrimination undermines the ability of African American youths to bond with their schools and why they associate with delinquent peers. The awareness of racist stereotypes—stereotype threats—is just one of the ways African Americans are forced to confront the realization that their race matters in a pejorative way. Steele and Aronson (1995, p. 797) define stereotype threat as “… being at risk of confirming, as self-characteristic, a negative stereotype about one’s group.” Thus, African Americans must forever be aware that anything they say or do may be used to confirm deleterious stereotypes about what it means to be Black (Brown & Lee, 2005; Brown & Pinel, 2003; Inzlicht, McKay, & Aronson, 2006). Gates and Steele (2009) argue that African Americans cannot escape from the noxious consequences of stereotype threats because they are as omnipresent as if they were “in the air.”
Steele (1997, p. 613) argues that African American youths develop strong bonds with their school when school achievement is a part of their self-definition, a self-identity to which they hold themselves accountable. This accountability translates into sustained achievement motivation. That is, African American youths are likely to be motivated when they perceive that they have the opportunities, resources, interests, and skills necessary to succeed in school. Their motivation—the strength of their bonds—is also related to the belief that African Americans are accepted, valued, and belong in school.
However, Steele (1997) argues that deleterious depictions of African American youths threaten their ability to perform well in school. Furthermore, he argues that poor performances because of stereotype threats spiral into sustained patterns of failures, especially among those who initially desired to develop strong bonds with their schools. That is, successive failures over time undermine the enthusiasm that African American students initially had toward their schools. Steele (1997, p. 614) defines this process as “disidentification,” whereby African American youths reconceptualize their self and their values in order to remove performing well in school as a basis for self-evaluation (Woodcock, Hernandez, Estrada, & Schultz, 2012). Steele (1997) argues that even though it may be protective, such disidentification can undermine sustained motivation and weaken their bonds to their school. He further posits that this new self-identity can be costly when the domain is as important as doing well in school. Indeed, researchers argue that racial differences in educational achievement can be accounted for by the relationship between stereotype threats and the underperformance of African American youths rather than any alleged differences in lack of ability or incompetence (Haslam, Salvatore, Kessler, & Reicher, 2008). 1
Unnever and Gabbidon (2011) argue that racial discrimination, whether experienced (e.g., being racially profiled by the police) or specifically encountered within the context of schools—for example, toxic stereotypes about their intelligence—causes African Americans to manifest anger and hostility toward historically White-dominated institutions. These negative emotions undermine their ability to bond with those institutions, particularly those they perceive as affiliated with the specific acts of racial discrimination they have experienced. These weakened social bonds have profound consequences. Steele (1997, p. 623) argues that disidentified African American youths—those with weak social bonds—engage in behaviors that allow them to create a self-identity within “domains in which their prospects are better” rather than within school-related activities. Unnever and Gabbidon (2011) assert that one of the domains that disidentified African American youths are likely to choose is delinquency—a behavior that most often does not demand a high level of competence (Gottfredson & Hirschi, 1990). Thus, delinquent behavior offers an opportunity for disidentified African American youth—those who have weak bonds with their school or those who have dropped out—to gain a sense of self-importance—that is, an identity that is not diminished by their tenuous relationship with their school. This assertion is consistent with Steele’s (1997) thesis that as African American youths retreat from their attachment, commitment, and involvement in their schools, they may engage in compensatory “grandiose” behaviors in order to bolster their sense of self-worth.
If disidentification increases the likelihood that African American youths will engage in delinquency, it also stands to reason that disidentification will affect peer network structures. Social networking scholars have long noted the etiology of a peer group is driven by the complex interaction between selection and propinquity (Gottfredson & Hirschi, 1990; McPherson, Smith-Lovin, & Cook, 2001; Young, Rebellon, Barnes, & Weerman, 2014). In other words, disidentified youths are likely to gravitate (i.e., selection) to other similarly disidentified youths who live in or near their neighborhood (i.e., selection facilitated by propinquity). Social learning theorists have recognized these points and made explicit attempts to account for the clustering of similarly situated youths in neighborhoods (Akers, 2009). Modern social learning theories recognize the important role that propinquity can play in bringing youths together, but once a social network has formed, behaviors shared by group members will be sanctioned and exacerbated (Akers, 2009; Harris, 1998). In essence, the peer group will act as an insulator that will extend youths’ feelings of disidentification, while simultaneously accelerating their involvement in delinquent activity.
It is important to note, however, that no research—of which we are aware—has explored the role of racial discrimination on peer group formation. Unnever and Gabbidon’s (2011) theory would suggest that youths who experience racial discrimination are more likely to find solace in companionship with one another. Thus, to the extent that discrimination experiences play out in one’s neighborhood or school, where other similarly situated youths may be found, peer groups might be expected to form. Once the social network has formed, the delinquent behavior of individual members might be heightened as a result of typical social learning processes (Akers, 2009).
Recent research by Unnever, Cullen, and Barnes (2016) lends support to Unnever and Gabbidon’s (2011) assertions. Analyzing the PHDCN-LCS and the same two cohorts included in the present study, they found that the more African American youths perceived being discriminated against, the less they were attached to their teachers and were less committed to their education. Unnever et al. (2016) also report that racial discrimination undermined the strength of these school-related bonds while controlling for the degree to which the African American youths were attached to their mother and father and other covariates. As further support for Unnever and Gabbidon’s (2011) theory, their analyses revealed that the influence of racial discrimination on changes in externalizing behavior from Wave 1 to Wave 3 was both direct and indirect via the measures of school bonds while controlling for parental affection and the other covariates. Unnever et al. (2016) conclude that racial discrimination can increase externalizing behaviors directly among African American youths, and it can increase their level of externalizing behaviors by weakening their attachment to teachers and their commitment to their education.
Our analyses will build on the findings reported by Unnever et al. (2016) in two important ways. First, we examine whether racial discrimination is related to self-reported offending behavior. Unnever and colleagues analyzed externalizing behaviors. Second, we test for another possible racialized pathway of offending that operates through the school. Specifically, Unnever and Gabbidon (2011) posit that racial discrimination may lead to disidentification from the school, raising the probability that the youth will drop out. Moreover (or perhaps in addition to dropping out), disidentified youth should be more likely to associate with youths who are also in the process of disengaging from their school. Unnever and Gabbidon argue that disidentified youths are likely to turn to criminal behavior as a way to feel “successful,” as crime can be readily and easily accomplished (Gottfredson & Hirschi, 1990). In sum, Unnever and Gabbidon’s (2011) theory highlights a pathway to crime that posits that the African American youths who are most likely to offend are those who (1) drop out of school and (2) associate with deviant peers because of their greater exposure to racial discrimination.
Current Study
Several core assertions made by Unnever and Gabbidon (2011) are tested. Specifically, we test whether perceived racial discrimination is related to the probability that African American youths will drop out of high school. Moreover, we consider whether perceived racial discrimination predicts the frequency of contact with delinquent peers (Roberts et al., 2012; Simons, Whitbeck, Conger, & Conger, 1991). Once these relationships are tested, we examine whether perceived racial discrimination predicts involvement in criminal activity directly and/or indirectly through intervening variables measuring whether the youth had dropped out and the rate of delinquency the youth is exposed to in his or her peer group. More specifically, we test five hypotheses:
Hypothesis 1: Perceived racial discrimination increases African American offending.
Hypothesis 2: Perceived racial discrimination increases the probability that African American youths will drop out of school.
Hypothesis 3: Dropping out of school increases African American offending.
Hypothesis 4: Perceived racial discrimination increases the rate at which African Americans associate with deviant peers.
Hypothesis 5: Associating with deviant peers increases African American offending.
Data
Sample
We draw on data from the PHDCN-LCS) to analyze our hypotheses. The PHDCN-LCS is a cohort study of 6,207 children and their caregivers based on a representative sample drawn from 80 neighborhood clusters (NCs) in the city of Chicago in 1995. A two-stage sampling procedure was conducted. U.S. census data were first used to find 343 NCs—groups of two to three census tracts containing approximately 8,000 people who were relatively homogeneous with respect to racial–ethnic mix, socioeconomic status, housing density, and family structure. A random sample of 80 of the 343 NCs was drawn within 21 strata defined by racial–ethnic composition (seven categories) and socioeconomic status (high, medium, and low). Second, children within seven age cohorts (0 [birth], 3, 6, 9, 12, 15, and 18) were sampled within the sampled 80 NCs from randomly selected households based on a screening of over 35,000 households at roughly 3-year age intervals (Perkins & Sampson, 2015).
The first wave of data was collected from 1995 to 1997 and the third wave was collected approximately 5 years later. A simple random sample of households drawn from these neighborhoods generated a total sample of 587 African American youths in the 9-year-old and 12-year-old cohorts (at Wave 1) and at Wave 3. The response rates were 77.5 for the 9-year-olds and 74.9 for the 12-year-olds (Perkins & Sampson, 2015). The African Americans included in our analyses ranged from ages 13 to 18 at the third wave of data collection. This age bracket is a period of the life course that is characterized by increasing rates of problem behavior (Gottfredson & Hirschi, 1990; Laub & Sampson, 1993; Moffitt, 1993; Sampson & Laub, 1993; Simons et al., 2002).
We selected these two cohorts of African Americans because they were the only respondents who completed the questionnaire on racial discrimination and the school interviews and focus on the third wave because it was the only time the racial discrimination questionnaire was administered. Our analyses only include respondents who self-reported being Black. Each respondent was shown a list of racial/ethnic groups and asked, “Would you say that you are a member of only one or more than one of these racial or ethnic groups?” (emphasis in original). Respondents who indicated being a member of “one” group were asked a follow-up question: “Which of these would you choose?” A response option was “Black/African American.” We analyze respondents who selected this response (n = 368).
Measures
Dependent variable
Offending
The dependent variable was generated from the Self-Report of Offending questionnaire. This questionnaire was administered at each wave of data collection to determine whether the youths had committed a wide variety of offenses in the year preceding the interview. Scholars argue that self-reports of offending capture a broader range of behaviors than official measures of crime (Thornberry & Krohn, 2003). Our dependent variable, Offending W3, is the self-report measure from Wave 3. The index included 18 questions tapping into the respondent’s involvement (0 = has not committed the act and 1 = has committed the act) in criminal behavior such as “carried/hidden a weapon,” “broke into building to steal,” “sold cocaine/crack,” and “attack with weapon” in the past year. Fifty-one percent of the African American youths reported at least one offense at Wave 3. The Offending W3 measure was created by summing across the 18 items, and its α coefficient was .741.
Intervening variables
Dropped out
Three questions on the School Interview questionnaire completed at Wave 3 were used to determine whether the youth had dropped out or had been expelled from school. Specifically, each youth was asked whether they were currently going to school. The large majority of African American youths answered “yes” (95.56%, coded as 0). Those who responded “no” were asked two follow-up questions. The first asked whether they had attended school in the past year. The second question inquired about the reason the youth was not in school. All respondents who answered no to the initial question, no to the first follow-up question, or who said they had “dropped out” or were “expelled” (second follow-up question) were coded as 1 for the Dropped Out variable.
Delinquent Peers
Peer delinquency scales were created from the Deviance of Peers questionnaire adapted from Huizinga, Esbensen, and Weiher (1991). The African American youths were asked how many of their friends (1 = none, 2 = some, 3 = most, or 4 = all of them) engaged in behaviors such as “purposely damaging or destroying property that did not belong to them” and “attacking someone with a weapon.” The Deviance of Peers questionnaire was included in the data collection for both Wave 1 and Wave 3. We created two indices, Delinquent Peers W1 and Delinquent Peers W3. Delinquent Peers W1 was created by summing across items completed at Wave 1, and its α coefficient is .714. Delinquent Peers W3 was created by summing across items completed at Wave 3, and its α coefficient is .771.
This strategy for measuring peer delinquency is often referred to as an indirect and/or perceived measure because it asks the target respondent to report on his or her peers’ delinquency. Some have argued these measures conflate personal and peer delinquency (e.g., Gottfredson & Hirschi, 1990), but recent reports have shown that they identify distinct constructs albeit with a high degree of covariance (Young, Rebellon, Barnes, & Weerman, 2015). Moreover, the finding that much of adolescent delinquency is committed in groups (Reiss, 1988; Warr, 2002) further confounds the methodological distinction. With these points in mind, we note that the present measure of peer delinquency may include projection effects, but the group nature of adolescent delinquency suggests these influences are not enough to completely discount the argument that our measure represents a separate and unique construct. As evidence of this point, we note that the correlation between Offending W1 and Delinquent Peers W1 was r = .382 and the correlation between Offending W3 and Delinquent Peers W3 was r = .397. As can be seen, neither of these correlations is strong enough to suggest that our dependent variable is completely confounded with the measure of peer delinquency. Finally, we should note that we use the only available measure of peer delinquency in the PHDCN-LCS. The results, of course, should be viewed with appropriate caution, given these concerns.
Key independent variable
Racial discrimination
The Racial Discrimination index was constructed from items included in the Personal Identity questionnaire, which was completed at Wave 3. Respondents were first asked: Now I have a few questions about discrimination. Sometimes people feel they are discriminated against or treated badly or differently because of their race, ethnicity, color, language, or the country they or their family came from. Please tell me if you have felt discriminated against for this reason IN THE PAST YEAR at any of the places listed on this card. (emphasis in original)
Control variables
Based on prior research and theory, we control for several potential confounding influences. We include Offending W1, which is the self-report measure of offending collected at Wave 1, to help rule out a host of potential confounding influences that have a stable impact on offending behaviors over time. Age is a quantitative measure of the respondent’s age at the time of the Wave 3 interview. Male is a binary measure (0 = female, 1 = male). Parents Married is a dichotomous measure of whether the primary caretaker was married (0 = not married, 1 = married). Parents High School is a dichotomous measure of whether the primary caregiver finished high school (0 = did not finish high school, 1 = finished high school). Family Income is the total household income and its values range from 1 (less than US$5,000) to 11 (more than US$90,000). Parents in Prison is based on two separate questions asking the male and female primary caregivers whether either had ever been arrested and sent to prison (0 = neither caregiver was sent to prison, 1 = one of the primary caregivers was sent to prison, and 2 = both primary caregivers were sent to prison). The intergenerational transmission of crime represents one of the most important correlates of offending, meaning it is critical to control for these influences whenever possible (Farrington, 2011; Miller & Barnes, 2013).
Two measures were included to control for the African American youth’s religious involvement. Our first measure, Religious Salience, was created from a question included in the Personal Identity questionnaire completed at Wave 3. African American youths were asked how important their religious beliefs were, with responses ranging from 1 (not at all important) to 4 (very important). The second measure, Church Attendance, was created from a question that asked how often the youths attended religious services. Responses ranged from 1 (never) to 6 (nearly every day).
We control for four attitudes or personality traits that the prior research has found to be related to African American offending (Gibbons et al., 2012; Simons et al., 2002; Unnever, 2014). First, we control for the youth’s level of impulsivity. Impulsivity is also included to account for individual-level factors associated with selection into offending, delinquent peer associations, and dropping out of high school (Gottfredson & Hirschi, 1990; Young, Barnes, Meldrum, & Weerman, 2011). Impulsivity was a measure created by the PHDCN-LCS research team. It combines information from five subscales of the Emotionality, Activity, Sociability, and Impulsivity Temperament Survey (Buss & Plomin, 1975), which was administered to the primary caregivers at Wave 1 (see Gibson, Sullivan, Jones, & Piquero, 2009, for a discussion of this scale). Second, we control for the youth’s temper. Temper is based on an item included in the Child Behavior Checklist, which was completed by the African American youth’s primary caregiver at Wave 3. The primary caregiver was asked whether the youth “has temper tantrums or a hot temper.” The responses are based on a 3-point scale (0 = not true to 2 = often true). Third, we control for the youth’s extent of recent depression symptoms. Our measure, Depressive Symptoms, is based on the Center for Epidemiological Studies Depression Scale protocol, which was self-reported by the youths at Wave 3. Finally, we include a control for the child’s Verbal IQ that was measured at Wave 1 via the Wechsler Intelligence Scale for Children–Revised (WISC-R). We utilized the “scaled” version of the WISC-R vocabulary scores that are provided by the PHDCN staff.
Analytical Strategy
Model Estimation and Missing Data
The primary dependent variable for this analysis, Offending W3, was coded as a count of offenses committed by the respondent, meaning the Poisson distribution is most appropriate for fitting a predictive model. Model fit indices revealed the negative binomial model (an augmented version of the Poisson model) provided a better fit than the Poisson model to the conditional distribution generated from an equation using Racial Discrimination as the only predictor (Bayesian Information Criterion [BIC] improvement = 206.02). Thus, the negative binomial model was estimated.
Multiple imputation (MI) was utilized to retrieve missing information due to item nonresponse, which allowed for the analysis of all n = 368 cases. All analyses were estimated in Stata 14. MI was carried out with the “ice” package, and all regression models were estimated using the “mi estimate” prefix. Table 1 presents the descriptive statistics for the variables included in the analyses.
Descriptive Statistics for Selected PHDCN Study Variables.
Note. PHDCN = Project on Human Development in Chicago Neighborhoods; HS = high school.
Hypothesis Tests
We relied on a series of multiple regression models to test the five hypotheses outlined above. 3 The first step was to establish the relationship between Racial Discrimination and Offending W3 (Hypothesis 1) after adjusting for the influence of the control variables. We refer to this model as the baseline model because it removes the confounding influences of the control variables, but it does not consider the role of Dropped Out nor the influence of Delinquent Peers. The second step to the analysis was to establish the relationship between Racial Discrimination and Dropped Out (Hypothesis 2) after adjusting for the influence of the control variables. Because Dropped Out is a dichotomous variable, a logistic regression model was estimated. The third step was to reestimate the impact of Racial Discrimination on Offending W3 after including Dropped Out as a covariate along with the other control variables. This step will test Hypothesis 3. The fourth step tests Hypothesis 4. Here, Delinquent Peers W3 is the outcome variable and Racial Discrimination is the key covariate of interest.
Testing Hypothesis 5 was the fifth step to the analysis. To do so, the regression equation estimated in Step 1 was augmented to include the effect of Delinquent Peers W3 on Offending W3. This will allow for the estimation of any intervening role that Delinquent Peers W3 may have on the link between Racial Discrimination and Offending W3. The sixth step to the analysis included both Dropped Out and Delinquent Peers in the model simultaneously, which will allow for the estimation of the intervening effects of both variables. Finally, the seventh step to the analysis was to reestimate the previous model, this time including a lagged version of the dependent variable from Wave 1. Including a lagged measure of the dependent variable provides a conservative test of Hypotheses 1, 3, and 5 because it will rule out many time-stable influences on the dependent variable.
Based on the theoretical backdrop provided by Unnever and Gabbidon (2011), we anticipate a positive coefficient for the Racial Discrimination variable to emerge for each of the five hypothesis tests. In general, two-tailed tests of statistical significance are preferred for exploratory analyses, but one-tailed tests (i.e., placing all of the critical regions in one tail of the distribution rather than splitting it across both tails) are justified whenever the researcher has theoretical reasons to expect a specific directional relationship. That being said, we report both one- and two-tailed tests of statistical significance.
While we are able to control for offending and peer group associations at Wave 1, we are unable to do so with the Racial Discrimination measure because it was only included at Wave 3 of the PHDCN-LCS. Thus, the PHDCN-LCS’s design cannot fully satisfy the criteria of causal ordering. With this in mind, though, it is worth pointing out that Unnever et al. (2016, pp. 150–152) argued that efforts to falsify theories can be undertaken even when available data are not perfectly suited to the focal theory. In short, scholars can build prediction models based on the theoretical statements. If those predictions hold in a data set, even an imperfect data set, then the theory is not falsified and becomes a candidate for future inquiry (Popper, 1959). Moreover, recall that our analysis controls for the past behavior (offending at Wave 1) and a range of individual traits (verbal IQ, impulsivity, temper, depressive symptoms, and parental criminality) in an effort to rule out many sources of bias. Finally, prior work using the FACHS data suggests racial discrimination precedes delinquent behavior and not the reverse (Burt, Simons, & Gibbons, 2012; Evans, Simons, & Simons, 2016; Simons et al., 2006, 2003), lending preliminary support for the pathways being tested here.
Results
Reported in the first column of Table 2 are estimates from a negative binomial regression model where Offending W3 was the dependent variable. This column represents the baseline model, which tested Hypothesis 1. Net of the effects of the control variables (i.e., Age W3 through Verbal IQ W1), Racial Discrimination was positively associated with Offending W3. The coefficient of relationship reveals that the log of the expected count on the Offending W3 Scale increases by 0.164 (p < .05, two-tailed) for each 1-unit increase in Racial Discrimination. The incident rate ratio (IRR) for the impact of Racial Discrimination on Offending W3 was 1.178, meaning that a 1-unit increase in Racial Discrimination increased the expected rate of Offending W3 by 17.8% (i.e., 100 × [IRR − 1]).
The second model in Table 2 indicates that Racial Discrimination was positively associated with the odds of dropping out of high school (b = .270, p = .056, two-tailed). The odds ratio was 1.310, revealing that a 1-unit increase in Racial Discrimination was associated with a 31% increase in the odds of dropping out of high school. It is important to note, however, that there were relatively few respondents who had dropped out of school by Wave 3 of the PHDCN. Thus, statistical power was weakened for this analysis and for the subsequent model testing whether Dropped Out might intervene in the relationship between Racial Discrimination and Offending W3. Readers are encouraged, therefore, to interpret findings from the dropout analysis with due caution.
The Impact of Racial Discrimination on Offending W3 and Intervening Variables.
Note. HS = high school.
*p < .10. **p < .05 (two-tailed).
The results from the test of Hypothesis 3 are provided in the third column of Table 2. This model estimates the impact of Racial Discrimination on Offending W3 after including the Dropped Out variable as an intervening variable. The results of this model indicate that Dropped Out was positively associated with Offending W3, but the effect was not statistically significant. The relatively small number of respondents who were coded as having dropped out likely contributed to this nonsignificant effect because the effect size was substantively large (b = .215, IRR = 1.240, p = .528). These estimates do not provide support for Hypothesis 3. Also of note is that the impact of Racial Discrimination maintained a positive and statistically significant relationship with Offending W3. In other words, the inclusion of the Dropped Out variable does not appear to affect the association between Racial Discrimination and Offending W3. 4
The next step to the analysis was to test the association between Racial Discrimination and Delinquent Peers W3. The results from the ordinary least square (OLS) regression of Delinquent Peers W3 on Racial Discrimination are presented in column 4 of Table 2. OLS regression was used because the distribution of the Delinquent Peers W3 variable was approximately normal. The OLS parameter estimates indicated that Racial Discrimination was positively associated with the Delinquent Peers W3 variable. These findings provide support for Hypothesis 4.
Hypothesis 5 was tested by regressing (using the negative binomial model) Offending W3 on Racial Discrimination and the Delinquent Peers W3 Scale. Results from this model are presented in the fifth column of Table 2. These estimates indicate that Delinquent Peers W3 had a positive and statistically significant effect on Offending W3. The IRR revealed that a 1-unit increase in Delinquent Peers W3 was associated with an 11.2% increase in the expected rate of Offending W3. This result provides support for Hypothesis 5. Notably, there is some evidence to suggest the impact of Racial Discrimination on Offending W3 may partially operate through the Delinquent Peers W3 variable. Indeed, a user-written bootstrap mediation test revealed the indirect effect to be positive (b = .032, e (b) = 1.033, standard error [SE] =.015, 95% confidence interval = [.003, .062], z = 2.16, p = .031, two-tailed). 5
The sixth column of Table 2 provides the estimates from the negative binomial regression of Offending W3 on Racial Discrimination, Dropped Out, and Delinquent Peers W3. As can be seen, Delinquent Peers W3 maintained a positive and statistically significant influence on Offending W3. Dropped Out did not have a statistically significant influence. Finally, the last model presented in Table 2 (i.e., Model 7) included all of the variables in the analysis, along with a lagged version of the dependent variable (Offending W1) and a lagged measure of Delinquent Peers (from Wave 1). The impact of Racial Discrimination and the effect of Delinquent Peers W3 remained positive and statistically significant. In short, the results from this model—which include the potential confounding influences that are captured by Offending W1 and Delinquent Peers W1—provide support for Hypothesis 1 and Hypothesis 5 under a conservative testing condition.
Discussion
Evidence is accumulating that the everyday experiences of being African American, such as those with the innumerable forms of racial discrimination, impact offending (Burt et al., 2012; Riina et al., 2013; Unnever, 2014). The current article extends the prior research first by showing that in the context of Chicago, perceived racial discrimination is associated with offending. Notably, this relationship was found with controls for a host of theoretically relevant variables. This finding reinforces the conclusion that a full understanding of African American youths’ misconduct should take into account their lived experiences with racial discrimination.
Beyond assessing the link between racial discrimination and offending, the current project illuminates pathways that may partially account for this relationship. We contextualized our analysis within the framework posited by Unnever and Gabbidon’s (2011) theory of African American offending. A core thesis within their theory is that the experience of racial discrimination disrupts the ability of African American youths to build strong bonds with conventional institutions. A large segment of their theory is devoted to explaining how this process unfolds within schools (Unnever, 2015). Extending the classic work of Steele (1997), Unnever and Gabbidon (2011) argue that experiences with racial discrimination weaken the ties that African American youths have with their schools.
Our analyses lend partial support for Unnever and Gabbidon’s (2011) pathway perspectives. The analyses revealed that the more African American youths experienced racial discrimination, the greater their likelihood of dropping out of school and the more they were exposed to delinquent behavior in their peer group. Thus, as predicted by Unnever and Gabbidon, the results reveal that racial discrimination may be a significant reason why African American youths choose to drop out of high school and why they may be more likely to associate with delinquent peers. Recall, however, that our results did not support Hypothesis 3 which anticipated a positive relationship between dropping out of school and offending. As was noted above, this finding was likely driven by the small number of respondents who had dropped out of school by the Wave 3 interviews. Indeed, fewer than 20 respondents had dropped out, meaning we lacked the statistical power to truly unpack whether the association of perceived racial discrimination with offending operates through dropping out of high school. 6 Evidence did emerge, however, to suggest that associating with delinquent peers explained a portion of the relationship between perceived racial discrimination on offending.
Two additional analyses addressed whether the results are robust. First, we examined whether racial discrimination is associated with offending at Wave 3 after controlling for offending at Wave 1. To do so, we included a lagged measure of offending as a control variable in the last model presented in Table 2. This analysis represents a conservative model in that many stable factors that influence offending at both time points will be controlled. The results show that racial discrimination is significantly associated with African American youths’ offending at Wave 3 even in this model. Second, we controlled for peer group affiliation at both Wave 1 and Wave 3. The results showed that racial discrimination remained a statistically significant predictor of offending at Wave 3 even when peer group affiliation at two time points was controlled. All told, as posited by Unnever and Gabbidon (2011), the results suggest that turning points across the lives of African American youths—that is, choosing to drop out of school, to affiliate with deviant peers, and to offend—may be related to their experiences with racial discrimination.
A core hypothesis of Unnever and Gabbidon’s (2011) theory is that racial discrimination can directly and indirectly increase the likelihood that some African American youths will offend. These scholars argue that it can directly increase offending by causing some African American youths to become angry, depressed, and hostile, which in turn increases their probability of committing crime. Research supports this hypothesis (Unnever, 2012). They also argue that racial discrimination can indirectly enhance the likelihood of offending through multiple racialized pathways. For example, Unnever and Gabbidon (2011) argue that racial discrimination should weaken the bonds that African American youths have with their school, which in turn should increase their likelihood of offending. Research supports this contention (Bryd, 2014). Unnever et al. (2016) report that racial discrimination weakens the attachment that African American youths have to their teachers and it diminishes their commitment to their education, which in turn escalated their likelihood of exhibiting externalizing behaviors. The current research reveals another racialized pathway toward offending posited by Unnever and Gabbidon’s (2011) theory. Our findings indicate that racial discrimination increases both the likelihood that African American youths will drop out of school and affiliate with delinquent peers. The latter—affiliating with delinquent peers—may go on to predict elevated levels of offending. Taken together, these studies support Unnever and Gabbidon’s (2011) core assertion that racial discrimination is a debilitating force in the lives of African American youths, compelling some down racialized paths that lead to offending behaviors.
It is notable that the effects of the variables we focused on (particularly racial discrimination and delinquent peers) were significant but were smaller in magnitude than the direct effects of some of the control variables (e.g., Parents Married, Impulsivity). These findings are not unforeseen, as Unnever and Gabbidon’s (2011) theory was developed as a supplement to other explanations of offending. Consequently, it is expected that measures of the racial subordination of African Americans and other key covariates should both predict offending. Another consideration is that our analyses may underestimate the true effects of racial discrimination on youth misconduct, dropping out of high school, and associating with delinquent peers. The PHDCN-LCS includes standard global measures of racial discrimination. However, these items do not fully measure the lived reality of experiencing racial discrimination. Unnever and Gabbidon (2011) suggest that racial discrimination should be measured along some of the same axes as child abuse. Thus, scholars should measure who the perpetrators were, the age of onset, the frequency of the experiences, its chronicity, and its severity. The true effect may also have been underestimated because the PHDCN-LCS does not capture the full range of experiences with racial discrimination (see also Unnever, Cullen, & Barnes, 2016, p. 158). For example, the PHDCN-LCS only allows for the consideration of relatively profound expressions of racial discrimination at the exclusion of measures tapping into insidious everyday experiences with racist microaggressions (Nadal, Griffin, Wong, Hamit, & Rasmus, 2014). In addition, Unnever and Gabbidon (2011) argue that the deleterious consequences of racial discrimination can manifest through direct and vicarious experiences.
Unfortunately, the temporal ordering of the racial discrimination and offending measures in the PHDCN-LCS data precludes the possibility of definitively testing whether racial discrimination is a contributing cause for African American youth misconduct. Our analyses show a substantive correlation between youth offending and racial discrimination. However, correlation analyses leave open the possibility that African American youth who are engaging in disruptive behavior are more likely to encounter racial discrimination or that they are more likely to perceive and self-report experiences of discrimination compared to youth who are not engaging in offending behavior. Future scholars collecting longitudinal data may wish to further investigate the complex relationships between youth misconduct and perceptions of racial discrimination among African American youth.
Future research may also want to explore whether the relationships between racial discrimination, youth misconduct, dropping out of high school, and delinquent peer group affiliation are mediated by positive racial socialization experiences and having a positive racial identity (Burt & Simons, 2015; Hughes et al., 2006; Unnever & Gabbidon, 2011). Prior studies have found that positive racial socialization experiences coupled with an affirming positive racial identity can attenuate the toxic consequences of racial discrimination on academic skills and delinquency among African American youths (Burt et al., 2012; Burt & Simons, 2015; Caughy & Owen, 2015; Neblett, Chavous, Nguyên, & Sellers, 2009; Neblett, Terzian, & Harriott, 2010; Wang & Huguley, 2012). Furthermore, scholars may find it insightful to replicate our analyses with more contemporary data. Such an analysis could, for example, assess whether the recent series of high-profile deaths of African Americans by the police have exacerbated the relationships among racial discrimination, dropping out of school, delinquent peer group affiliation, and offending.
Lastly, these racialized pathways suggest that race-centered school reforms may be needed to address the unique racialized experiences that African American youths encounter within their schools. Unnever (2015) outlines numerous approaches that educators may wish to consider to enhance the bonds that African American youths build with their teachers, peers, and school. For example, Unnever suggests that educators must openly acknowledge the racist history of schooling and confront the racist practices (e.g., lower expectations that teachers may have for African American youths) and stereotypes (e.g., the “criminalblackman”) that pervade schools. This requires that faculty and administrators initiate dialogues within the classrooms that discuss the nature of racial prejudice and discrimination. This dialogue also can be facilitated by African American youths being exposed to a curriculum that embraces racial diversity, promotes a positive racial identity (e.g., by celebrating African American achievements), and being exposed to more African American teachers. Research indicates that African American youths report less stress if they attend historically Black schools rather than White–dominated institutions (Greer & Brown, 2011). In short, the bonds that African American youths have with their teachers, peers, and schools should be strengthened if educators diligently work to eliminate the multitude of ways that racial discrimination can manifest itself within the educational system.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.