Perceived Discrimination and Longitudinal Increases in Adolescent Substance Use: Gender Differences and Mediational Pathways

Abstract

Objectives. This study was designed to test hypotheses about the prospective association of adolescents’ perceptions of discrimination with increases in substance use and the processes that mediate this association.

Methods. African American youths residing in rural Georgia (n = 573; mean age = 16.0 years) provided longitudinal data on their experiences with discrimination, substance use, school engagement, and affiliations with substance-using peers.

Results. For male youths, perceived discrimination was significantly related to increases in substance use, and, as hypothesized, this association was mediated by the contributions of perceived discrimination to decreases in school engagement and increases in affiliations with substance-using peers. Analyses also indicated that discrimination influences substance use rather than vice versa.

Conclusions. Results are consistent with the hypothesis that high levels of discrimination are linked to increases in substance use for African American male adolescents.

Historically, rural residence has protected African American adolescents from high-risk behaviors prevalent in urban areas. Recent epidemiologic data, however, indicate that African American adolescents in rural areas are engaging in substance use at rates equal to or exceeding those of youths who live in densely populated inner cities.^1–3 Substance use is a leading cause of accidents, injuries, and disability among African Americans aged 15 to 24 years.⁴ It predicts the likelihood of infection with HIV/AIDS and other sexually transmitted infections; affects future educational attainment, behavior problems, depressive symptoms, unintended pregnancies, involvement with the criminal justice system, ability to find and keep employment, establishment and maintenance of family relationships; and leads to drug abuse and dependence during adulthood.^5–8 The experience of discrimination has been identified as a stressor with the potential to increase African American youths’ vulnerability to a host of problems during adolescence, including substance use.⁹ The primary purpose of this study was to test hypotheses regarding the influence of perceived discrimination on substance use and the psychosocial processes that account for these effects.

Research has established that the experience of unfair treatment based on race is common among African American adults¹⁰ and adolescents.¹¹ Associations have been documented between self-reported discrimination and various forms of substance use, including smoking,¹² alcohol consumption,¹³ and use and abuse of other drugs.¹⁴ The stress-coping model that has framed much of the research on the effects of discrimination^15,16 posits that frequent experiences with discrimination deplete coping resources and increase the attractiveness of avoidant coping strategies, such as substance use, because drug use offers temporary respite from discrimination-induced stress. Almost all of these studies, however, were cross-sectional (see Gibbons et al.¹⁷ for an exception), which limits their ability to determine whether substance use is a consequence or a cause of discrimination and to examine the intermediate processes that account for the influence of discrimination on substance use. The primary purposes of this study were to test the hypothesis that perceived discrimination will forecast increases in substance use across adolescence rather than the reverse. Recent research suggests that racial discrimination may affect male and female youths differently.^9,18 For example, Brody et al.⁹ found that discrimination was a more powerful predictor of conduct problems for male adolescents than for female adolescents. Given the comorbidity of substance use and conduct problems, we hypothesized that the influence of discrimination would be more apparent among male than female youths.

A secondary purpose of this study was to investigate the processes through which perceived discrimination results in increases in substance use. We propose that African American adolescents who feel devalued and demoralized by perceived discrimination become less inclined to accept conventional values and pursuits; hence, they come to view school, a major social institution, as irrelevant and gravitate toward peers who also reject conventional values. This is consistent with findings that youths who experience racial discrimination report more negative beliefs about the usefulness of school, lower academic efficacy,¹⁹ and lower grade point averages.²⁰ These youths subsequently become more prone to affiliate with like-minded peers who sanction and encourage nonconventional and risky behavior.²¹ Because declines in school engagement and affiliations with substance-using peers are proximal risk mechanisms known for onset and escalation of substance use,²¹ we expected an indirect effect of perceived discrimination on increases in substance use mediated through its effect on decreasing school engagement and increasing affiliations with substance-using peers.^22,23

METHODS

Participants were African American adolescents who lived in a rural region of Georgia in which poverty rates are among the highest in the nation.²⁴ Data were collected from 2002 to 2010 in the context of a family based substance use prevention study.²⁵ Schools (n = 23) in 9 rural Georgia counties provided lists of 11-year-old students, from which youth participants were selected randomly.²⁵ Families were contacted and enrolled in the study by African American community liaisons who resided in the counties in which the participants lived; 70% of eligible families screened agreed to participate. Youths randomly assigned to the intervention group took part in a 7-week family centered program. Study hypotheses were tested with the entire sample; however, assignment to the prevention or control condition was controlled in all analyses. Youths participated in 6 waves of data collection at ages 11, 12, 14, 16, 17, and 18.5 years. The perceived discrimination measure was placed into the longitudinal protocols when the youths were 16 years old; hence, study hypotheses were tested with data obtained from the latter 3 waves when youths were 16 (Wave 1: mean = 16.00; SD = 0.38), 17 (Wave 2: mean = 17.00; SD = 0.51), and 18.5 years old (Wave 3: mean = 18.48; SD = 0.46). Of the 667 participants who provided data at baseline (mean age = 11.65 years; SD = 0.34), 573 participants provided data at age 16 years (Wave 1), and 524 participants provided data at age 18.5 years (Wave 3). No differences in study variables emerged based on attrition status from baseline to age 16 years or from age 16 to age 18.5 years.

Trained African American field researchers conducted interviews in participants’ homes at each of the 3 waves of data collection. Field researchers read items preprogrammed into a laptop computer. For sensitive items, youths were given a key pad with which to enter their responses privately. Youths were interviewed individually and privately; they were told that their answers were strictly confidential and would not be disclosed to anyone within or outside the family. Youths received $40.00 at each wave for their participation. The university institutional review board approved all study protocols.

Measures

Measures were selected for their relevance to the lives of rural African American youths. They were derived from previous research, which included focus group meetings and pilot testing followed by construct validation of the instruments.²⁵

Control variables.

Maternal education, household income adequacy, and maternal substance use were controlled in the analyses. Mothers reported their educational attainment on a scale ranging from 1 (grades 1–4) to 10 (doctorate or professional degree). Mothers reported the adequacy of their income to meet their needs on the following item: “How adequate do you feel your income is in meeting your needs?” Response scale ranged from 1 (much less than adequate to meet even our basic needs) to 5 (more than adequate to meet all of our needs and wants). Maternal substance use involvement was assessed with an index based on 3 items from the Monitoring the Future Survey.²⁶ Mothers reported their frequency of cigarette smoking, consumption of 4 or more drinks at one time during the previous month and their frequency of marijuana use during the previous 3 months on a 6-point ordinal scale. Items were recoded to form dichotomous indicators representing the presence or absence of daily smoking, weekly heavy drinking, and monthly marijuana use. Dichotomous scores were summed to form an index ranging from 0 to 3 of maternal substance use involvement.

Perceived discrimination.

At Wave 1, target youths completed 9 items from a version of the Schedule of Racist Events (SRE)¹⁰ revised for use with adolescents.^9,17,27,28 Items in the revised SRE assessed the frequency during the previous year, ranging from 0 (never happened) to 3 (happened a lot), with which the respondent perceived specific discriminatory behavior events. Events included racially based slurs and insults, disrespectful treatment from community members, physical threats, and false accusations from business employees or law enforcement officials. Responses to items were summed to form the perceived discrimination scale, ranging from 0 to 27. Coefficient α for the scale was 0.85.

School engagement.

At Waves 1 and 2, adolescents completed a 6-item scale developed for the Family and Community Health Study⁹ to assess school engagement among rural youths. On a response set ranging from 1 (strongly disagree) to 5 (strongly agree), adolescents rated their attitudes toward school, homework, teachers, and grades. Responses were summed to form a scale ranging from 6 to 30, with higher scores indicating a stronger attachment to school activities. Cronbach's α for the scale was 0.70.

Peer affiliations.

Five items on substance use experiences among youths’ friends assessed at Waves 1 and 2 the numbers of peers who had smoked cigarettes; drunk beer, wine, or wine coolers; or smoked marijuana. Responses were reported on scales ranging from 1 (none) to 3 (4 or more). Items were summed to form a composite score that ranged from 5 to 15, with higher scores indicating more substance-using friends. Cronbach's α was 0.81.

Substance use.

At each wave, adolescents reported the numbers of times during the previous month that they drank alcohol, had 3 or more drinks of alcohol at one time, or smoked marijuana. Responses to these 3 items were summed to form a previous month substance use index ranging from 0 to 104. Items were significantly intercorrelated; all P < 0.01. This index has been used widely in previous research.^29,30

Plan of Analysis for the Study Hypotheses

We first examined the main effects and gender interactions of reported discriminatory experiences assessed at Wave 1 on increases in substance use at Wave 3, controlling for Wave 1 levels. Because of the skewed distribution of the substance use variable (skew > 6.5 across waves), Poisson regression was used. We then used a cross-lagged structural equation model (SEM) to determine whether perceived discrimination led to increases in substance use or vice versa. Meditational hypotheses were also investigated using SEM, specifying a Poisson distribution for the substance use outcome. All analyses were implemented in Mplus software,³¹ using the maximum likelihood estimator, which allows testing of hypothesized models against all data present, obviating the need for list-wise deletion.

RESULTS

Table 1 presents descriptive statistics for male and female youths.

TABLE 1—

Characteristics of Male and Female Adolescents Who Took Part in the Strong African American Families Program in Rural Georgia; Longitudinal Data Collected from 2002 to 2010.

Characteristic	Male Participants (n = 252), No. or Mean (SD)	Female Participants (n = 286), No. or Mean (SD)	t (P)
Wave 1 (n = 538)
Target age, y	16.00 (0.37)	16.02 (0.39)	1.02 (.31)
Primary caregiver educationa	4.65 (1.35)	4.64 (1.38)	0.13 (.9)
Income adequacyb	3.13 (1.18)	3.14 (1.08)	−0.07 (.94)
Primary caregiver substance usec	0.44 (1.03)	0.53 (1.15)	−1.04 (.3)
Perceived discriminationd	3.67 (3.51)	3.57 (3.55)	0.33 (.74)
School engagemente	13.63 (2.63)	12.81 (2.62)	3.58 (< .001)
Substance-using peersf	7.04 (2.36)	6.68 (2.00)	1.90 (.06)
Substance useg	0.028 (0.55)	0.32 (0.63)	0.72 (.47)
Wave 2 (n = 528)	246	282
Target age, y	17.03 (0.49)	17.07 (0.52)	0.68 (.5)
School engagemente	13.81 (2.77)	12.78 (2.44)	4.49 (< .001)
Substance-using peersf	7.30 (2.55)	6.83 (2.21)	2.27 (.02)
Substance useg	3.31 (8.34)	1.64 (3.27)	3.10 (< .001)
Wave 3 (n = 553)	256	297
Target age, y	18.45 (0.47)	18.50 (0.45)	1.28 (.2)
Substance useg	4.36 (9.25)	2.51 (5.37)	2.92 (< .001)

^aMothers reported educational attainment on a scale ranging from 1 (grades 1–4) to 10 (doctorate or professional degree).

^bMothers reported the adequacy of their income to meet their needs on the following item: “How adequate do you feel your income is in meeting your needs?” Response scale ranged from 1 (much less than adequate to meet even our basic needs) to 5 (more than adequate to meet all of our needs and wants).

^cAssessed with an index based on 3 items from the Monitoring the Future Survey.²⁶ Mothers’ frequency of cigarette smoking, consumption of ≥ 4 drinks at one time during the previous month and their frequency of marijuana use during the previous 3 months reported on a 6-point ordinal scale were recoded to form dichotomous indicators representing the presence or absence of daily smoking, weekly heavy drinking, and monthly marijuana use. Dichotomous scores were summed to form an index ranging from 0–3 of maternal substance use involvement.

^dTarget youths completed 9 items from a version of the Schedule of Racist Events (SRE)¹⁰ revised for use with adolescents.^9,17,27,28 Items in the revised SRE assessed the frequency during the previous year, ranging from 0 (never happened) to 3 (happened a lot), with which the respondent perceived specific discriminatory behavior events. Responses to items were summed to form the perceived discrimination scale, ranging from 0–27.

^eOn items from the Family and Community Health Study,⁹ from a response set ranging from 1 (strongly disagree) to 5 (strongly agree), adolescents rated their attitudes toward school, homework, teachers, and grades. Responses were summed to form a scale ranging from 6–30, with higher scores indicating a stronger attachment to school activities.

^fNumber of peers who had smoked cigarettes; drunk beer, wine, or wine coolers; or smoked marijuana. Responses were reported on scales ranging from 1 (none) to 3 (≥ 4). Items were summed to form a composite score that ranged from 5–15.

^gAdolescents reported the numbers of times during the previous month that they drank alcohol, had ≥ 3 drinks of alcohol at one time, or smoked marijuana. Responses to these 3 items were summed to form a previous month substance use index ranging from 0–104.

Sample Characteristics

Substance use increased significantly each year from Wave 1 (age 16 years) to Wave 3 (age 18.5 years) for male youths (P < .05) and increased significantly (P < .05) from Waves 1 to 3 and Waves 2 (age 17 years) to 3 for female youths. Participants’ mean household gross monthly income was $2577 (SD = $2608) and mean monthly per capita gross income was $670 (SD = $745). Although 72.6% of the mothers were employed outside the home and worked an average of 39.7 hours per week, 46.4% of the families lived below federal poverty standards, and another 20.5% of families lived within 150% of the poverty threshold. The study participants were representative of families living in rural Georgia counties.³²

Perceived Discrimination and Increases in Substance Use

Poisson regression of discrimination on substance use across Waves 1 and 3, controlling for Wave 1 discrimination, indicated that gender significantly interacted with discrimination to predict substance use (P < .001). Separate analyses by gender indicated no influence of discrimination on substance use for female youths (odds ratio [OR] = 1.00; P = .998) but a significant influence for male youths (OR = 1.11; P < .001). Given the lack of a main effect for female youths, remaining analyses were conducted with the male subsample only.

Cross-lagged analyses (Wave 1 perceived discrimination to Wave 3 substance use; Wave 1 substance use to Wave 3 perceived discrimination; Figure 1) were conducted, and the 2 lagged pathways differed significantly (Wald = 4.16; P = .041). The path from perceived discrimination to increases in substance use was significant (b = 0.11; P < .001), but the path from substance use to increases in perceived discrimination was not (b = 0.02). These data support the study's premise that perceived discrimination leads to increases in substance use across 2 years and not vice versa.

FIGURE 1—

Cross-lagged analysis of the influence of discrimination on substance use.

*P < .05. ***P < .001.

Mediational Hypotheses

The next set of analyses was designed to test the hypothesis that the contribution of perceived discrimination to increases in substance use is mediated by the contributions of discrimination to decreases in school engagement and increases in affiliations with substance-using peers. Figure 2 shows results of the model used to test this hypothesis. The model shows that perceived discrimination is associated with decreases in school engagement from Wave 1 to Wave 2 (b = −0.16; P < .01) and increases in affiliations with substance-using peers from Wave 1 to Wave 2 (b = 0.17; P < .01); these changes, in turn, are associated with changes in substance use from Wave 1 to Wave 3. The indirect effect of perceived discrimination on increases in substance use is significant through decreases in school engagement (b = 0.11; P < .05) and increases in affiliations with substance-using peers (b = 0.09; P < .05). The presence of the mediators reduced the direct path between perceived discrimination and increases in substance use from 0.56 (P < .05) to 0.20 (not significant).

FIGURE 2—

School engagement and affiliations with substance-using peers as mediators of the effect of perceived discrimination on substance use.

Note. Wave 1 substance use, adequacy of income, parental education, and intervention condition were controlled in the model. The main paths in the model are depicted with thick lines.

*P < .05. ***P < .001.

DISCUSSION

Results of this study showed that for male youths, perceived discrimination was linked longitudinally with increases in substance use through the hypothesized mediators. This finding underscores the importance of addressing discrimination in prevention efforts for African American male youths. These efforts include public health campaigns designed to reduce discrimination in schools and communities as well as preventive interventions that focus on coping with the sequelae of discrimination. Because data were obtained from a representative sample of adolescents from rural Georgia and findings were obtained with demographic characteristics controlled, we expect the results to be generalizable to similar areas in the rural south. Although researchers have found main effects of discrimination on substance use,^17,33 we found this effect only for male adolescents. Little previous work, however, has considered gender differences in the association between discrimination and substance use. Findings from a recent study that examined gender differences in the link between discrimination and smoking were consistent with our finding of a stronger effect on male adolescents.³⁴ From a public health perspective, this has important implications because it adds to a growing body of evidence indicating that frequent experiences with discrimination during adolescence shift the alcohol and drug consumption curve so that more male adolescents in rural populations become frequent users. This shift can lead not only to driving fatalities but also to increases in other threats that substance use poses to adolescents’ mental health, school achievement, and family relationships.

The theoretical rationale for this study is based on the premise that perceptions of racial discrimination lead to youth substance use rather than the reverse. To verify this rationale, we conducted a cross-lagged analysis that simultaneously examined the evidence for causal effects from discrimination to substance use and from substance use to discrimination. This analysis detected an effect from discrimination to substance use but not vice versa. This finding provides important evidence that discrimination in adolescence is a causal element in the etiology of substance use and not a consequence of youth involvement in problem behavior. Similar findings have emerged in analyses of youth depression and conduct problems.⁹

Tests of mediating process have suggested 2 pathways through which perceived discrimination affects male adolescents’ substance use. Results show independent paths from perceived discrimination to decreases in school engagement and to increases in affiliations with substance-using peers. These paths fully accounted for the effect of discrimination on substance use. Findings are consistent with research that suggests that the perception of prejudice against one's group can result in reductions in identification with and persistence in academics.^19,20 The findings are also consistent with research sponsored by differential peer affiliation models of the etiology of substance use, which suggest that alienated youths are inclined to affiliate with like-minded peers who hold similar values.³⁵

Our study did not detect discrimination effects for female adolescents. This may have resulted from the lower substance use and discrimination rates reported by female youths compared with those reported by male youths. This may have reduced our power to detect effects. Alternatively, recent research has suggested that female adolescents are more likely than are male adolescents to respond to stressful events such as discrimination with internalizing symptoms rather than externalizing problems such as substance use.³⁶ Female youths also may have had access to unique protective processes that shielded them from the influence of discrimination. For example, in a prospective analysis, Brody et al.⁹ found that the influence of discrimination on youths’ depressive symptoms and conduct problems was weakened among youths who received nurturant-involved parenting. Further research is warranted to examine differential responses to discrimination by gender.

Our findings indicate that public health efforts designed to prevent substance use among male African American youths should include a focus on discriminatory experiences and strategies for coping with them. Results also indicate that prevention efforts must interrupt discrimination-induced negative trajectories such as affiliation with deviant peers and disengagement from school. An example of such a program is the Strong African American Families (SAAF) intervention.^25,37 A 7-week, 14-hour family skills training program, SAAF is designed to enhance individual coping both by teaching youths anger management skills and by teaching parents to support youths’ disclosure of discriminatory events and efforts to deal with them. Youth coping is further bolstered by racial socialization practices designed to increase racial pride. The Flint Fathers and Sons program is designed to provide nonresident African American fathers and their sons with structured opportunities for youths to develop parent-supported skills for coping with racial discrimination.³⁸ Families spend 32 hours in 15 intervention sessions and 13 hours completing homework assignments and participating in community events. Given the high prevalence of discriminatory experiences and the vulnerability of male African American youths to substance use and other behavior problems, wider implementation of such programs is warranted.

Limitations

Some aspects of the present study constitute possible limitations. It is not known if the results can be generalized to rural African American youths outside of Georgia or to youths in other environments. The discrimination measure addresses interpersonal discrimination only, rather than organizational or institutional discrimination; thus, it provides a limited assessment of the range of discriminatory experiences that adolescents in the rural South encounter. The use of a quantity or frequency measure may be subject to recall bias. Biological measures as well as daily diary or time line follow-back methods are known to increase reports of substance use.³⁹ Finally, other processes of interest were not controlled in this study, such as maternal mental health, stressful life experiences, sibling substance use, and community level processes that might influence rural African Americans’ substance use.

Conclusions

These cautions notwithstanding, the present study demonstrates the ways in which perceived exposure to racial discrimination affects increases in substance use among male African American adolescents. An important direction for future research involves epidemiological studies of the predictors of growth in substance use in rural versus nonrural areas and the role discrimination may play in these trends.