Discrimination, Psychological Functioning, and Substance Use Among U.S. Young Adults Aged 18–28, 2017

Abstract

This study aimed to test whether (a) discrimination is associated with past 30-day/current alcohol, cigarette, e-cigarette, alcohol, marijuana, and other illicit drug use among Black and White U.S. adults aged 18–28, (b) psychological distress (PD) and positive well-being (PW) are mediators of the discrimination–substance use relationships, and (c) the associations are moderated by race and sex. Using data from a 2017 U.S. nationally representative survey we conducted multiple-group moderated mediation analyses among 2,192 young adults aged 18–28 (508 Black males, 594 Black females, 533 White males, 557 White females). Black males had higher discrimination, Whites had higher PW, and females had higher PD scores. Discrimination was positively associated with PD and negatively associated with PW. Among all groups, discrimination was positively associated with other illicit drug (direct and indirect), and marijuana use through PD. Indirect effects were stronger among White males for other illicit drugs and Black males for marijuana. The indirect effect of discrimination and alcohol use through PW was positive for Black females and negative for all other groups examined. Among Black males only, discrimination was positively associated with cigarette and alcohol use through PD (positive) and cigarette smoking through PW (negative). This study highlights the negative influence of perceived discrimination on current licit and illicit substance use among Black and White young adults. Our results suggest that this relationship may be partially mediated by PD and PW, especially among Black male young adults. Future discrimination and substance use studies should consider potential mediation effects of poor mental health and differences by race and sex.

Substance use rates (Park et al., 2018) are similar among Black and White adults, but among youth, Black Americans have lower rates (Park et al., 2018) and this pattern has persisted (Center for Behavioral Health Statistics and Quality, 2018). However, Black Americans are more likely to experience substance use problems, mortality, and morbidity (Center for Behavioral Health Statistics and Quality, 2018; Henley et al., 2016; Roberts et al., 2016; Zapolski et al., 2014) and a deeper understanding of contributors to substance use is needed to eliminate disparities. Discrimination has emerged as a substance use risk factor (Gerrard et al., 2012; Hunte & Barry, 2012; Stock et al., 2011; Williams & Mohammed, 2009) linked to increased risk of licit (Bennett et al., 2005; Hunte & Barry, 2012; Madkour et al., 2015; Parker et al., 2016) and illicit substance use (Carliner et al., 2016; Hunte & Barry, 2012; Hunte & Finlayson, 2013) and substance use problems among Black Americans (Hunte & Barry, 2012; Kendzor et al., 2014; Zapolski et al., 2014).

Those who experience discrimination may use substances to cope with discrimination-related stress (Williams & Mohammed, 2009). Discrimination is linked to mental health conditions, including psychological distress (PD; Lee & Ahn, 2013; Mouzon et al., 2016; Stevens-Watkins et al., 2014), depression (Banks et al., 2006; Clark, 2014), and anxiety (Banks et al., 2006; Levine et al., 2014; Mouzon et al., 2016), that are independently linked to substance use (Clark, 2014). The strong relationship between mental health and substance use has been well documented (Conway et al., 2017; Esmaeelzadeh et al., 2018; Luger et al., 2014) and literature suggests that mental health is an important mediator of the discrimination–substance use relationship (Williams & Mohammed, 2009). For example, in a study of Black young adults, depressive symptoms fully mediated the discrimination–substance use relationship and discrimination explained one-quarter of the variance in substance use and depressive symptoms (Clark, 2014).

Emerging adulthood (ages 18–28) is associated with the onset of several mental health conditions (Henin & Berman, 2016), increased responsibilities and stress (Arnett, 2003, 2011), and the development of unhealthy behaviors and coping mechanisms like substance use (Arnett, 2005; Sussman & Arnett, 2014) that can influence later health. Unsurprisingly, emerging adults have a higher risk of psychological disorders (Arnett et al., 2014; Henin & Berman, 2016), substance misuse/disorders (Center for Behavioral Health Statistics and Quality, 2017; Johnston et al., 2016; Sussman & Arnett, 2014), and problems with substance use treatment (Bergman et al., 2016). Additionally, differences in the prevalence of mental health conditions and substance use among emerging adults by age, sex, and race have been found.

For example, Twenge et al. (2019) found that rates of past 30-day PD in the U.S. were higher among young adults aged 18–25 (v ≥ 26) and between 2008 and 2017 rates of PD increased 71% among young adults aged 18–25 while weaker and less consistent increases were observed among adults aged ≥26. Moreover, the cohort increase in PD was higher among females (vs. males) over time but fluctuated among Black and White Americans. However, interactions with sex and race were also found by age cohort such that the cohort increase in PD was larger among women (vs. men) and White (vs. Black) Americans. Similarly, prior studies have consistently found that emerging (vs. established/older) adults and males (vs. females) have higher rates of substance use (Center for Behavioral Health Statistics and Quality, 2018; Substance Abuse and Mental Health Services Administration, 2018a, 2018b); however, interactions exist by age, race, and sex among emerging adults. For example, in 2017, the prevalence of past-month cigarette smoking among young adults aged 18–25 was 28.8% among White males, 25.0% among White females, 22.7% among Black males, and 11.5% among Black females. Conversely, the prevalence of past-month cigarette smoking among young adults aged 26–34 was 31.4% among White males, 27.1% among White females, 36.0% among Black males, and 22.2% among Black females (Substance Abuse and Mental Health Services Administration, 2018b).

Moreover, Black emerging adults may be particularly at risk for discrimination-related issues and substance use (Assari et al., 2017; Bennett et al., 2005; Gerrard et al., 2012; Stock et al., 2011). Black American emerging adults reported similar rates of discriminatory experiences relative to Black Americans aged 30–49 and 50–64 (Robert Wood Johnson Foundation, 2017a) and recent data indicate that Black (vs. White) Americans are more likely to report experiences of discrimination (Menasce Horowitz et al., 2019). Recent data also indicate that certain discriminatory experiences (i.e., being unfairly stopped by police, people acting as if they are suspicious of you) are more common among Black males (vs. females; Anderson, 2019). Conversely, Borrell et al.’s (2007) longitudinal examination of the relationship between discrimination and substance use among Black adults (aged 18–30 at baseline) found no interaction between discrimination and sex. Recent, publicized social and political events, such as the 2016 U.S. Presidential election and police shootings of unarmed Black youth and adults, have increased racial tensions and been linked to poor health among Black Americans (Alang et al., 2017; Williams & Medlock, 2017). Young adults may be acutely aware of such racial and sociopolitical issues due to ubiquitous use of the internet, mobile devices, and social media and the visibility of such issues on these platforms (Daniels, 2013; Pew Research Center, 2016) potentially contributing to poor mental health and substance use (Leventhal et al., 2018). Thus, discrimination may be an especially relevant stressor for young adults, but more research is needed to understand how experiences with discrimination are related to substance use in this population.

There is also a Black–White mental health paradox where Black Americans are less likely to be diagnosed with mental health conditions despite poorer physical health and greater social–economic stressors (Barnes & Bates, 2017; Mezuk et al., 2010, 2013). Limited research suggests that nonspecific PD may be a better measure of poor mental health among Black Americans relative to diagnosable conditions like depression due to differential item-functioning (Barnes & Bates, 2017; Mouzon et al., 2016; Watkins & Johnson, 2018). PD is associated with increased risk of mortality (Forman-Hoffman et al., 2014), chronic diseases (Weissman et al., 2015), and substance use (Forman-Hoffman et al., 2014) and a previous study found that having three or more (vs. 0) experiences of racial discrimination was associated with nearly seven times higher odds of PD (Krieger et al., 2011). Conversely, Keyes posited that mental health and mental illness are two distinct continuums (Keyes, 2005) with mental health representing social, emotional, and psychological well-being as opposed to the absence of mental illness (Keyes, 2007). Given the Black–White mental health paradox and mental health–substance use comorbidity (Cougle et al., 2010; Schroeder & Morris, 2010), PD and Keyes’ mental health/positive well-being (PW) could provide unique insights into the discrimination–substance use relationship.

Unfortunately, few studies have examined the discrimination–PD relationship (Kessler et al., 1999; Schulz et al., 2000; Williams & Mohammed, 2009) and none have examined the discrimination–PW relationship, as most have focused on diagnosable mental illnesses (Mays et al., 2007; Williams & Williams-Morris, 2000). Similarly, few studies have examined the relationships between discrimination, measures of psychological functioning, and substance use among young adults overall and by race and sex. Gibbons et al.’s (2004) longitudinal study of Black parents and their children found that the relationship between discrimination and substance use was mediated by PD. Several differences were found by sex such that at Time point 1 young females reported more experiences with discrimination and at Time 2 reported more PD and substance use. However, interactions between discrimination and sex were not significant in the final models and parent–children dyads did not include young adults.

Several other studies have examined the relationship between discrimination, mental health, and/or substance use and interactions with race, sex, and sexual orientation. However, these studies utilized data collected in the early to mid-2010s and often focused on different types of discrimination (i.e., discrimination attributed to race vs. sex vs. sexual orientation) or stratified results by sexual orientation (Evans-Polce et al., 2020; Slater et al., 2017; Vu et al., 2019), as opposed to examinations of the discrimination–psychological functioning–substance use relationship overall and by race–sex groups. For example, Vu et al. (2019) found that compared to White heterosexual women, Black sexual minority women had a higher risk of tobacco and marijuana use and a lower risk of alcohol use but no differences by race were found among Black and White men. Similarly, women who reported experiencing discrimination attributed to both race and sexual orientation (vs. no discrimination) had greater depressive symptoms, alcohol use, tobacco use, marijuana use but no associations were found among men. While such analyses are warranted and needed, additional research is also needed to understand the between discrimination, psychological functioning, and substance use among young adults overall and generally by race and sex.

Given the current sociopolitical climate, more recent examinations of these relationships are needed, particularly among young adults and by both race and sex. We aimed to provide an updated, novel assessment of the discrimination–mental health–substance use relationship among young adults and answer the following research questions: (a) Is discrimination directly associated with substance use (alcohol, cigarette, e-cigarette, marijuana, and other illicit drug use) among our overall sample of Black and White emerging U.S. adults aged 18–28, (b) Are PD and PW mediators of the discrimination–substance use relationship among the overall sample, and (c) Are these associations moderated by both race and sex?

Method

Data

Data were from the 2017 early release Transition into Adulthood Study (TAS), a supplement to The Panel Study on Income Dynamics (PSID) and follow-up to the PSID Child Development Study (CDS; Institute for Social Research). PSID is a nationally representative, longitudinal survey of U.S. families (Institute for Social Research) and CDS is a cohort of parents and children aged 0–12. In 2017, TAS was relaunched to include all young adults aged 18–28 in PSID regardless of CDS participation (Institute for Social Research). Given our goal of examining Black–White comparisons, analyses were restricted to those who self-identified as Black or White with complete data on discrimination, PD, and PW (excluded n = 330), yielding 2,192 participants. Given the use of public, secondary data, this study was determined to be not human subjects’ research by the Georgia State University Institutional Review Board (approval #H19202, 10/17/18).

Measures

Discrimination

Discrimination was assessed using 7 Everyday Discrimination (EDD) scale items (Williams et al., 1997). Respondents reported the frequency of discriminatory experiences in their day-to-day life, specifically being treated with less courtesy; less respect; as if they are not smart; or receiving poorer service at restaurants/stores than others and others acting as they are afraid of them; dishonest; are better than them. Responses ranged from 1 = never to 6 = almost every day. A one-factor confirmatory factor analysis (CFA) demonstrated adequate fit, χ²(14df) = 510.486, p < .001; RMSEA = 0.127; CFI = 0.969, TLI = 0.954, and reliability (Cronbach’s α = 0.85) to the EDD items. Responses were summed and ranged from 7 to 42 with higher scores representing more frequent experiences of discrimination.

Substance Use

Cigarettes and E-Cigarettes.

Participants who answered “yes” to “Do you smoke cigarettes?” were considered current cigarette smokers. Those who reported vaping (i.e., used vape pen/e-cigarette/e-hookah/e-vaporizer to inhale a mist into the lungs) ≥1 day in the past 30 days were considered past-month e-cigarette users.

Alcohol.

Participants reported ever drinking alcoholic beverages (beer/wine/liquor) and ever drinkers were asked, “In the last year, on average, how often did you have any alcohol to drink?.” Those who responded, “about once a month,” “several times a month,” “about once a week,” “several times a week,” or “every day” were considered past-month drinkers.

Marijuana and Other Illicit Drugs.

Participants reported medicines and drugs they ever tried, even if only once, including marijuana/hashish, cocaine, heroin, narcotics/pain relievers, amphetamines, barbiturates, tranquilizers, hallucinogens, and inhalants. Those who used ≥1 occasion in the past 30 days were considered past-month users. Past-month marijuana use was assessed independently while other drugs were combined.

Psychological Functioning

PD.

Past-month PD was assessed using the Kessler-6 (K6) scale (Kessler et al., 2002) which screens for serious mental illness. Participants were asked, “During the past 30 days, about how often did you feel: “nervous,” “hopeless,” “restless or fidgety,” “so depressed that nothing could cheer you up,” “that everything was an effort,” and “worthless.” Responses ranged from 0 = none of the time to 4 = all of the time. A one-factor CFA demonstrated adequate fit, χ²(9df) = 167.663, p < .001; RMSEA = 0.089; CFI = 0.988, TLI = 0.980, and reliability (Cronbach’s α = 0.79) to the K6 items. Responses were summed and ranged from 0 to 24 with higher scores indicating greater PD.

PW.

Past-month PW was assessed using the 14-item Mental Health Continuum Short Form (MHC-SF), which measures emotional, social, and psychological well-being. Participants reported how often they felt several states in the past month including “happy” and “that our society is becoming a better place.” Responses ranged from 0 = never to 5 = every day and a one-factor CFA demonstrated adequate fit, χ²(77df) = 3790.137, p < .001; RMSEA = 0.148; CFI = 0.833, TLI = 0.803, and reliability (Cronbach’s α = 0.86) to the MHC-SF items, while the subscales fit the data less well. Thus, the MHC-SF items were summed, and responses ranged from 0 to 70 with higher scores indicating greater PW.

Participant Characteristics

Race/ethnicity was from 2017 TAS and age and sex were from 2017 PSID.

Analyses.

Discrimination was the primary independent variable, PD and PW were mediators, five substance use measures (alcohol, marijuana, other illicit drugs, cigarettes, e-cigarette) were dependent variables, and sex and race were moderators (Figure 1). Given the multiple pathways and categorical moderators, we employed multiple-group moderated mediated structural equation modeling (SEM). SEM was conducted to assess whether the proposed mediation effects of PD and PW varied by the proposed moderators on the discrimination-mediator, mediator-substance use, or discrimination–substance use pathways (Preacher et al., 2007).

Figure 1. Path Diagram of Hypothesized Moderated Mediated Discrimination–Psychological Functioning–Substance Use Relationship.

Note. The a paths represent the association between perceived discrimination and a given mediator (psychological distress or positive well-being), the b paths represent the association between a given mediator and a given substance use outcome (past-month/current alcohol, marijuana, other illicit drugs, cigarettes, or electronic nicotine delivery systems; ENDS use), and the c′ paths represent the association between perceived discrimination and a given substance use outcome.

Mediation Analyses

Descriptive and bivariate analyses were conducted in SAS 9.4 and CFA, mediation, moderation, and multiple-group moderated mediation analyses in Mplus 8. Path analyses were conducted using a variance-adjusted weighted least squares (WLSMV) estimator which is appropriate for categorical and continuous data and may be more accurate for categorical/dichotomous outcomes (MacKinnon, 2008). As such, the SEM models used probit regression, which models the inverse of the standard normal cumulative distribution.

Path analyses tested the mediating effects of PD and PW and the moderating effects of race and sex on the discrimination–substance use relationship. Final models controlled for age, mediators were allowed to covary, substance use outcomes were allowed to covary, and both standardized and unstandardized path coefficients were estimated. For robust standard errors for model parameter estimation, bootstrapping was used to test for statistical significance of mediation and moderated mediation effects by producing 95% confidence intervals (CI) from 10,000 resamples of the data (Hayes, 2013). Measures of overall model fit were not presented as the path models were saturated and the model fit perfectly to the data as shown in Figure 1, all variables share pathways in the model. First, mediation models were run with only one mediator and one outcome, next with both mediators and each outcome, and finally with all mediators and outcomes.

Moderated Mediation Analyses

To determine whether mediation paths were equal among racial groups and sexes, we utilized a multiple-group approach rather than interaction terms. Specifically, multiple-group analyses were conducted among Blacks and Whites, males and females, and finally Black males, Black females, White males, and White females. Thus, differential effects were parameterized as group-specific effects. To determine whether moderation was present by race, sex, and race by sex, direct and indirect effects of the a, b, and c’ pathways were compared across groups (Figure 1). Global differences across all groups were assessed using multiparameter Wald χ² tests conducted at the α = 0.05 level using the Model Test option in Mplus’ model constraint command. To identify statistically significant differences in specific pathways, terms representing the product of coefficients (i.e., a*b) were defined and assessed through 95% bootstrap CIs. If there was no evidence of a difference across groups from the multiparameter Wald test, the pathway was constrained to be equal across groups. All nonsignificant interaction terms were specified as model constraints.

Results

Descriptive Analyses

Descriptive analyses by sex and race are in Table 1 and by past-month/current substance use are in Table 2. The average discrimination score was 16.3 (SD = 7.1), PD score was 4.9 (SD = 4.0), PW score was 50 (SD = 10.9) and age was 22.7 (SD = 3.1). The sample had slightly more females but was similar by race and substance use rates ranged from 6.2% (cigarettes) to 28.2% (marijuana; Table 1). On average, Black males had higher discrimination scores, White Americans had higher PW scores, and females had higher PD scores (Table 1). White Americans, especially males, were more likely to drink alcohol and males were more likely to use marijuana. Black females and White Americans were more likely to use other illicit drugs. Black males had notably higher rates of cigarette smoking, while White males had notably higher rates of e-cigarette use (Table 1). Finally, cigarette smokers and marijuana, other illicit drug, and e-cigarette users were more likely to have higher discrimination and PD scores and lower PW scores (Table 2).

Table 1.

Discrimination, Mental Health Mediators, Substance Use Outcomes, and Age by Race and Sex Among U.S. Young Adults Age 18–28, 2017

	Overall	Black male	Black female	White male	White female
Measure	n = 2,214 M (SD) or % (n)	n = 516 M (SD) or % (n)	n = 600 M (SD) or % (n)	n = 536 M (SD) or % (n)	n = 560 M (SD) or % (n)
Predictor and mediators
Everyday discrimination	16.3 (7.1)	17.2 (8.2) a	15.6 (6.9) b	16.6 (6.8) a	15.8 (6.2) b
Psychological distress	4.9 (4.0)	4.5 (4.0) a	5.1 (3.9) b	4.8 (4.1)a,b	5.2 (4.0) b
Positive well-being	50.0 (10.9)	49.8 (11.2)a,b	49.2 (10.9)b	50.5 (11.1)a,b	50.7 (10.5)a
Substance use measures
Past-month alcohol use***	23.6 (522)	22.1 (114)	18.0 (108)	30.2 (162)	24.5 (137)
Past-month Marijuana use***	28.2 (625)	36.6 (189)	22.5 (135)	33.4 (179)	21.8 (122)
Past-month other illicit drug use***	14.4 (319)	9.7 (50)	14.2 (85)	17.7 (95)	15.9 (89)
Cigarette smoking status***	6.2 (138)	11.6 (60)	3.2 (19)	6.2 (33)	4.3 (24)
Past-month e-cigarette use***	7.5 (167)	6.8 (35)	3.8 (23)	13.4 (72)	6.6 (37)
Demographic characteristics
Age	22.7 (3.1)	22.6 (3.2)	22.7 (3.0)	22.6 (3.1)	22.8 (3.1)
Sex
Male	47.6 (1,054)	—	—	—	—
Female	52.4 (1,160)	—	—	—	—
Race/Ethnicity
White, NH	49.5 (1,096)	—	—	—	—
Black, NH	50.5 (1,118)	—	—	—	—

Note. M = Mean; SD = Standard deviation; NH = Non-Hispanic.

^***Indicates p < .001 based on t or χ². Differences in superscripts indicate statistically significant differences in means across groups (i.e., groups with ^a and ^b have statistically significant different means).

Table 2.

Discrimination, Mental Health Mediators, and Demographic Characteristics by Current Substance Use Status Among U.S. Young Adults Age 18–28, 2017

	Alcohol		Marijuana		Other illicit drugs		Cigarettes
	Yes	No	Yes	No	Yes	No	Yes	No	Yes	No
Measure	n = 521	n = 1,691	n = 625	n = 1,588	n = 319	n = 1,895	n = 136	n = 2,076	n = 167	n = 2045
	M (SD) or % (n)
Predictor/mediator
Everyday discrimination	16.7 (7.1)	16.1 (7.1)	18.3 (7.3)***	15.5 (6.9)	18.3 (7.2)***	15.9 (7.0)	18.7 (8.9)***	16.1 (6.9)	18.7 (8.2)***	16.1 (7.0)
Psychological distress	5.0 (4.1)	4.9 (4.0)	6.1 (4.3)***	4.5 (3.8)	6.5 (4.4)***	4.6 (3.9)	6.5 (5.1)***	4.8 (3.9)	6.4 (4.8)***	4.8 (3.9)
Positive well-being	50.8 (10.6)	49.8 (11.0)	47.6 (11.2)***	51.0 (10.6)	47.2 (12.1)***	50.5 (10.6)	48.1 (11.1)*	50.1 (10.9)	48.3 (12.1)*	50.2 (10.8)
Demographic characteristics
Age	22.0 (2.6)***	22.9 (3.2)	22.6 (3.0)	22.7 (3.2)	23.0 (3.0)*	22.6 (3.1)	21.9 (3.0)	22.7 (3.1)	21.6 (3.1)***	22.8 (3.1)
Sex
Male	53.0 (276)**	45.9 (776)	58.9 (368)***	43.1 (684)	45.5 (145)	47.9 (907)	68.4 (93)***	46.2 (959)	64.1 (107)***	46.2 (945)
Female	47.0 (245)	54.1 (915)	41.1 (257)	56.9 (903)	54.5 (174)	52.1 (986)	31.6 (43)	53.8 (1,117)	35.9 (60)	53.8 (1,100)
Race/ethnicity
Black, NH	42.6 (222)***	52.9 (894)	51.8 (324)	49.9 (792)	42.3 (135)**	51.8 (981)	58.1 (79)	50.0 (1,037)	34.7 (58)***	51.7 (1,058)
White, NH	57.4 (299)	47.1 (797)	48.2 (301)	50.1 (795)	57.7 (184)	48.2 (912)	41.9 (57)	50.1 (1,039)	65.3 (109)	48.3 (987)

Note. M = Mean; SD = Standard deviation; NH = Non-Hispanic.

^*p < .05.

^**p < .01.

^***p < .001 based on t or χ².

Mediation Analyses

Overall mediation results are found in Table 3 and unstandardized path coefficients (a*b) are reported below. In the overall model, there were positive, indirect effects of discrimination on cigarette smoking (0.008, 95% CI [0.002, 0.014]), marijuana (0.007, 95% CI [0.004, 0.011]), other illicit drugs (0.010, 95% CI [0.006, 0.014]), and e-cigarette use (0.008, 95% CI [0.003, 0.012]), through PD. There were also direct effects of discrimination on cigarette smoking, marijuana, other illicit drug, and e-cigarette use, wherein higher discrimination predicted substance use. No direct effect of discrimination on alcohol use was found. However, there was a negative, indirect effect of discrimination on alcohol use (−0.003, 95% CI [−0.006, −0.001]), and a positive, indirect effect on marijuana use (0.003, 95% CI [0.001, 0.006]), through PW. That is, greater discrimination was associated with less PW and a higher likelihood of drinking but a lower likelihood of marijuana use. In summary, in the overall model, PD mediated the effect of discrimination on all substances but alcohol, while PW only mediated the effect of discrimination on alcohol and marijuana.

Table 3.

Decomposition of Direct and Indirect Effects From the Overall Model: Unstandardized and Standardized Probit Regression Coefficients, N = 2,192

	Overall
Pathway	Unstandardized coefficient	SE	p	95% LL	CI UL	Standardized coefficient
Discrimination → Distress	0.22	0.01	<.001	0.20	0.25	0.39
Discrimination→ Well-being	−0.40	0.03	<.001	−0.47	−0.34	−0.26
Distress → Alcohol	0.01	0.01	.39	−0.01	0.03	0.03
Well-being → Alcohol	0.01	0.00	.01	0.00	0.01	0.08
Discrimination → Distress → Alcohol	.002			−0.002	0.006
Discrimination → Well-being → Alcohol	−0.003			−0.006	−0.001
Discrimination → Alcohol	0.01	0.00	.14	−0.00	0.02	0.05
Distress → Marijuana	0.03	0.01	<.001	0.02	0.05	0.13
Well-being → Marijuana	−0.01	0.00	.01	−0.01	0.00	−0.09
Discrimination → Distress → Marijuana	.007			0.004	0.011
Discrimination → Well-being → Marijuana	0.003			0.001	0.006
Discrimination → Marijuana	0.02	0.00	<.001	0.01	0.03	0.16
Distress → Illicit drugs	0.04	0.01	<.001	0.03	0.06	0.17
Well-being → Illicit drugs	0.00	0.00	.30	−0.01	0.00	−0.04
Discrimination → Distress → Illicit drugs	0.010			0.006	0.014
Discrimination → Well-being → Illicit drugs	0.001			−0.001	0.004
Discrimination → Illicit drugs	0.02	0.01	<.001	0.01	0.03	0.11
Distress → Smoking	0.04	0.01	<.001	0.01	0.06	0.14
Well-being → Smoking	0.00	0.01	.67	−0.01	0.01	0.02
Discrimination → Distress → Smoking	0.008			0.002	0.014
Discrimination → Well-being → Smoking	−0.001			−0.005	0.003
Discriminationà Smoking	0.01	0.01	.03	0.00	0.03	0.10
Distress → E-cigarette	0.03	0.01	<.001	0.01	0.06	0.13
Well-being → E-cigarette	0.00	0.00	.79	−0.01	0.01	0.01
Discrimination → Distress → E-cigarette	0.008			0.003	0.012
Discrimination → Well-being → E-cigarette	0.000			−0.004	0.003
Discrimination → E-cigarette	0.02	0.01	.01	0.00	0.03	0.11

Note. Boldface indicates statistical significance at the p < .05 level.

Moderated Mediation Analyses

Prior to conducting multiple-group moderated mediated SEM by both race and sex, models were run to test two-way interactions by race and sex alone. In models assessing moderation by sex, among males there was evidence of a positive indirect effect of discrimination on marijuana (0.014, 95% CI [0.009, 0.020]), other illicit drug (0.011, 95% CI [0.005, 0.017]), cigarette (0.012, 95% CI [0.004, 0.020]), and e-cigarette use (0.008, 95% CI [0.002, 0.015]), through PD. Among males there was evidence of a negative indirect effect of discrimination on alcohol use (−0.005, 95% CI [−0.009, −0.002]), through PW. Among females, there was evidence of a positive indirect effect of discrimination on other illicit drug (0.008, 95% CI [0.002, 0.014]), and e-cigarette use (0.009, 95% CI [0.002, 0.017]), through PD and on marijuana (0.007, 95% CI [0.003, 0.011]), and other illicit drug use (0.005, 95% CI [0.001, 0.009]), through PW.

In models assessing moderation by race, among Black Americans, there was evidence of a positive, indirect association between discrimination and marijuana (0.007, 95% CI [0.003, 0.011]), other illicit drug (0.009, 95% CI [0.004, 0.014]), cigarette (0.007, 95% CI [0.001, 0.014]), and e-cigarette use (0.010, 95% CI [0.003, 0.016]), through PD. Among White Americans, there was evidence of a positive, indirect association between discrimination and marijuana (0.008, 95% CI [0.002, 0.015]), and other illicit drug use (0.012, 95% CI [0.005, 0.019]), through PD and on marijuana use (0.005, 95% CI [0.001, 0.010]), through PW. Among White Americans, there was evidence of a negative, indirect association between discrimination and alcohol use (−0.007, 95% CI [−0.013, −0.003]), through PW (data not shown).

A summary of the multiple-group moderated mediation SEM results by both race and sex is presented in Table 4 based on unstandardized path coefficients (a*b). There was evidence of several group differences per multiparameter Wald tests, Wald χ² (3 df), p, including the effect of discrimination on alcohol use, directly or indirectly through PD, by race and sex (8.83, p = .0316), by race among males (15.37, p = .0015) and females (18.15, p = .0004). Additionally, there were group differences in the effect of discrimination on alcohol use, directly or indirectly through PW, by race among males (10.97, p = .0119) and females (9.91, p = .0193). There were group differences for discrimination, PD, and cigarette smoking (17.73, p = .0005), marijuana (16.59, p = .0009), and other illicit drug use (16.51, p = .0009) by race among females. Finally, there were group differences for discrimination on e-cigarette use, directly or indirectly through PD by race among females (21.15, p = .0001) and directly or indirectly through PW by race among males (8.46, p = .0374).

Table 4.

Summary Table of SEM Results for Overall and Multiple-Group Moderated Mediation Model

Pathway	Overall	Black males	Black females	White males	White females
Discrimination → Distress	+	+	+	+	+
Distress → Alcohol		+
Discrimination → Alcohol
Discrimination → Distress → Alcohol		+
Distress → Marijuana	+	+	+	+	+
Discrimination → Marijuana	+	+	+		+
Discrimination → Distress → Marijuana	+	+	+	+	+
Distress → Illicit Drugs	+	+	+	+	+
Discrimination → Illicit Drugs	+	+	+	+	+
Discrimination → Distress → Illicit Drugs	+	+	+	+	+
Distress → Smoking	+	+
Discrimination → Smoking	+
Discrimination → Distress → Smoking	+	+
Distress → E-cigarette	+	+	+		+
Discrimination → E-cigarette	+
Discrimination → Distress → E-cigarette	+	+	+		+
Discrimination → Well-being	−	−	−	−	−
Well-being → Alcohol	+	+	−	+	+
Discrimination → Well-being → Alcohol	−	−	+	−	−
Well-being → Marijuana	−		−	−	−
Discrimination → Well-being → Marijuana	+		+	+	+
Well-being → Illicit Drugs
Discrimination → Well-being → Illicit Drugs
Well-being → Smoking		+
Discrimination → Well-being → Smoking		−
Well-being → E-cigarette
Discrimination → Well-being → E-cigarette

Note. Plus and minus signs indicate positive and negative associations/pathways that are statistically significant at the α < 0.05 level. Discrimination was assessed by the Everyday Discrimination scale, Distress or psychological distress was assessed by the K-6 scale, and Well-being or positive well-being was assessed by the Mental Health Continuum Short Form (MHC-SF) scale.

Overall, there was evidence of several group differences as indicated by single parameter Wald tests, Wald χ² (1 df), p. Of the a paths, there were group differences by race among females between discrimination and PD (16.21, p < .001) and males between discrimination and PW (5.41, p = .0200). Of the b paths, through PD, there were group differences by sex among Black Americans for cigarette smoking (6.66, p = .0099), by race and sex for alcohol (4.12, p = .0423), and by sex among Black Americans (7.14, p = .0075) and White Americans for marijuana use (4.35, p = .0370). Of the b paths, through PW, there were group differences by race among females for alcohol use (9.04, p = .0026), by sex among Black Americans for cigarette smoking (9.42, p = .0021), alcohol (4.51, p = .0337), and marijuana use (5.44, p = .0196), and among White Americans for other illicit drug use (3.91, p = .0480). Finally, there was a sex difference in the c’ path between discrimination and marijuana use among White Americans (5.56, p = .0184).

The multiple-group moderated mediation results by each race–sex subgroup are in Tables S1a–1d. Among all groups, discrimination was positively associated with PD, negatively associated with PW, and there were positive, direct, and indirect effects between discrimination, PD, and other illicit drug use (Tables S1a–d). There was evidence for moderated mediation such that the indirect effect of discrimination on other illicit drug use was strongest for White males. Among all groups, there were positive, indirect effects of discrimination on marijuana use through PD and on alcohol use through PW. There was an interaction with sex among Black Americans and White Americans for the indirect effect on marijuana use wherein the effect was stronger among males. Conversely, there were interactions with race and sex for the indirect effect on alcohol use among females and Black Americans wherein there was a positive association between discrimination and alcohol use through PW among Black females but negative association among Black males and White Americans.

Among Black males only, there was a positive, indirect effect of discrimination on alcohol use (0.010, 95% CI [0.003, 0.017]), and cigarette smoking (0.015, 95% CI [0.006, 0.024]), through PD and a negative, indirect effect of discrimination on cigarette smoking through PW (−0.006, 95% CI [−0.012, 0.000]). This indicated a three-way interaction with race and sex wherein the indirect effects were only present among Black males. Meanwhile, among Black males, Black females, and White females (but not White males), there were positive, direct effects between discrimination and marijuana use (Tables S1a, b, d), as well as indirect effects of discrimination on e-cigarette use through PD (0.011, 95% CI [0.005, 0.019]). This indicated an interaction by sex among White Americans wherein there were direct and indirect effects among White females but not White males. Finally, among Black females, White males, and White females (but not Black males) there were positive, indirect effects of discrimination on marijuana use through PW (0.005, 95% CI [0.002, 0.008]), which indicated an interaction by sex among Black Americans wherein there was an indirect effect among Black females but none among Black males.

Discussion

We found that everyday discrimination is directly associated with current licit and illicit substance use among emerging U.S. adults and these associations were also mediated by past-month PD and lower PW. The present study is one of the first to examine the relationship between discrimination, PW, and substance use and we found evidence of different mediational effects for PD and PW and fewer associations through PW. Similarly, this study is one of the first to examine the relationship between discrimination and e-cigarette use and we found several novel associations across substances. Overall, our moderation findings suggest that the discrimination–mental health–substance use relationship varies by both race and sex such that discrimination is a salient predictor of substance use especially among young Black males.

In the present study, there was no difference in the effect of discrimination on PD across all race–sex groups. Byrd (2012) found that although Black (vs. White) Americans reported significantly more experiences of discrimination, the relationship between discrimination and PD was moderated by race, such that the effect of discrimination on PD was stronger among White (vs. Black) Americans. Similarly, Williams et al. (1997) found that Black (vs. White) Americans reported lower levels of PD following adjustment for discrimination. However, neither study considered differences by sex. Conversely, Kessler et al. (1999) found interactions between everyday discrimination with race and sex on PD such that the effect of discrimination on PD was stronger among females (vs. males) and Black Americans (vs. other racial/ethnic groups but not White Americans). In Mossakowski’s (2018) diverse sample of Hawaiians, discrimination was associated with greater PD among women relative to men; however, Black Americans were not included in the study and substance use was not considered. Additional research is needed by race and sex to clarify whether the discrimination–psychological functioning relationship differs by race and sex.

We found that discrimination was associated with cigarette smoking through PD and PW only among Black males. These findings are consistent with prior research that linked discrimination with cigarette smoking among Black men (Parker et al., 2016). Purnell et al. (2012) found that PD mediated the relationship between discrimination and cigarette smoking regardless of race as the interaction between discrimination and race was not significant). However, Purnell et al.’s (2012) study utilized data from 2004 to 2008 and did not examine interactions by sex or race and sex. The present study is also one of the first to examine the relationship between discrimination and e-cigarette. A recent study found that more frequent discrimination was directly associated with higher odds of combustible tobacco use in the overall sample and among Black Americans, but higher odds of e-cigarette use in the overall sample and among White Americans (Rogers et al., 2018). Moreover, we previously found that persons with PD and lifetime mental health conditions were more likely to be current or daily e-cigarette users (Spears et al., 2017, 2018, 2019). Interestingly, discrimination was directly associated with e-cigarette use only in the overall sample and indirectly associated through PD among all race–sex groups except White males. Although White males generally had higher rates of e-cigarette use relative to other groups this finding suggests that e-cigarette use among White males is not driven by coping with discrimination and/or psychological functioning. Furthermore, the indirect effect of discrimination on e-cigarette use through PD was the same among Black males, Black females, and White females due to model constraints. Additional research is needed to better understand the relationship between discrimination and e-cigarette use.

We found relatively consistent associations across groups for marijuana and illicit drug use such that discrimination was associated with marijuana and other illicit drug use through PD among all race–sex groups. Discrimination was also directly associated with marijuana use among all groups except White males and indirectly associated with marijuana through PW among all groups except Black males. Notably, indirect effects were stronger among White males for other illicit drugs and Black males for marijuana. Our findings suggest that (a) discrimination and PD contribute to marijuana and illicit drug use among young males and females alike, (b) marijuana use may be used to cope with discrimination-related stress among Black males especially, and (c) other illicit drug use may be used to cope with discrimination-related PD among White males especially. Few studies have examined the relationship between discrimination, psychological functioning, and marijuana or discrimination, and other illicit drug use, with most of the available studies focusing on sexual minorities (Vu et al., 2019) or stigma associated with illicit drug use (Ahern et al., 2007; Couto et al., 2019). Assari et al.’s (2019) longitudinal study of Black Americans found an interaction between discrimination and sex with marijuana use such that discrimination at baseline (during emerging adulthood) was associated with marijuana use over time only among males. Unfortunately, this study did not consider measures of psychological functioning. Conversely, Parker et al. (2017) did not find an association between everyday discrimination and marijuana use among Black males following adjustment for major stress, depressive symptoms, and other sociodemographic characteristics. Interestingly, an association was found with major discrimination and Black males who used marijuana almost every day had a decreased risk for major discrimination suggesting that marijuana use in particular may be used as a coping mechanism among Black males. Hunte and Finlayson (2013) found that discrimination was associated with illicit drug use following adjustment for race/ethnicity, gender, and other sociodemographic characteristics and among a nationally representative sample of Black Americans (Hunte & Barry, 2012). However, interactions with race and/or sex were not considered.

We found several novel differences for alcohol use by race and sex: There was a positive and negative relationship between discrimination and alcohol use through PW among Black females and all other race–sex groups, respectively. Additionally, discrimination was associated with alcohol use through PD only among Black males.

Hurd et al. (2014) conducted a latent growth model of perceived discrimination, mental health (e.g., depressive symptoms and anxiety symptoms), and substance use (cigarette smoking and alcohol use) among Black emerging adults. Although discrimination was associated with alcohol use, depressive symptoms, and anxiety symptoms, no associations were found between discrimination or mental health with cigarette smoking and mental health was not associated with substance use. Additionally, results were similar by sex. Notably, Hurd et al.’s (2014) study utilized data collected in the early 2000s, a past-year discrimination measure, past-week mental health measures, and past 30-day substance use measures. Other studies have found that Black males are equally or more likely to be current drinkers than White Americans and Black females (Center for Behavioral Health Statistics and Quality, 2018) and linked discrimination with alcohol and substance use disorders among U.S. Black adults (Hunte & Barry, 2012).

Higher levels of PW, under Keyes’ conceptualization, captures positive functioning psychologically, emotionally, and socially. Conversely, PD captures mental illness. Limited research suggests that problem drinking may be more strongly associated with greater PD compared to less problematic drinking patterns (Foulds et al., 2013). Given our focus on past-month alcohol use (vs. problem drinking), it is possible that the mediational effect of PW reflects discrimination contributing to poor overall well-being and alcohol use wherein a mediational effect with PD may have reflected problematic drinking. Nevertheless, our findings implicate discrimination and poor mental health/well-being as contributors to substance use and may provide insights into the Black–White substance use age crossover (Caraballo et al., 2016; Zapolski et al., 2014). Similarly, alcohol and marijuana were the most commonly used substances across groups, are associated with positive expectancies (Jones et al., 2001; Skenderian et al., 2008), and are less stigmatized in the U.S. (Carliner et al., 2017) although some studies suggest that alcohol is more stigmatized in the Black community (Keyes et al., 2010) even among those with lifetime alcohol dependence (Smith et al., 2010). Thus, it follows that emerging adults subjected to discrimination may experience PD or less PW and view marijuana or alcohol use as a less problematic way to cope. Future studies on discrimination and substance use may benefit from considering PW.

Discrimination attribution was not included in path analyses, but discrimination is understood as a stressful experience across racial/ethnic groups and regardless of attribution (Paradies et al., 2015; Priest et al., 2013; Williams et al., 2003). However, most associations in the present study were observed among Black Americans. Additionally, young adults may experience discrimination and mental health challenges while undergoing major life changes. Thus, it is possible that those in emerging adulthood, especially Black Americans and Black males in particular, are more likely to encounter certain psychosocial stressors that contribute to their substance use or establishing a pattern of coping-related substance use. Future preventative interventions and treatment programs, especially those targeted toward Black Americans, may benefit from considering discrimination and nonspecific mental health conditions.

Many prior discrimination studies focus exclusively among Black Americans or other racial/ethnic minorities but in the present study several associations were also found among White Americans. In a 2017 study, 55% of White Americans endorsed the belief that discrimination against White Americans exists in America today, and those who endorsed the belief were more likely to report institutional and interpersonal discrimination. However, less than 20% of White Americans reported personal experiences of discrimination (Robert Wood Johnson Foundation, 2017b). Future research should consider racial comparisons of the discrimination–mental health and discrimination–mental health–substance use relationships to better understand underlying mechanisms across groups.

Limitations

Due to the TAS relaunch, we utilized one wave of TAS data and could not assess causality or directionality of associations. We were also limited to the variables included in the TAS survey. Importantly, there is a temporal ordering to the measures used. Specifically, discrimination was modeled with more frequent occurrences corresponding with higher discrimination scores, which would likely precede (and was modeled with) past-month psychological functioning and substance use. However, although the everyday discrimination scale captures the frequency of discriminatory experiences in respondent’s day-to-day life, the scale does not assess discriminatory experiences within a specific time frame. Conversely, the examined mediators and substance use outcome measures were framed as past 30-day/current occurrences. Prior studies indicated that discrimination precedes poor psychological functioning and substance use (Borrell et al., 2007; Brody et al., 2012; Gibbons et al., 2004); however, longitudinal data are needed to establish causality and the direction of the discrimination–psychological functioning–substance use associations.

Future studies would benefit from longitudinal data and biochemically verified substance use. Additionally, analyses were age adjusted and future studies should consider other demographic characteristics including sexual orientation. Differences observed in the present study and prior studies could be attributed to changes in the discrimination–psychological functioning–substance use relationship over time (i.e., our use of more recent vs. older data) and/or our consideration of both race and sex. Regardless, the present study is a novel examination of the discrimination–substance use relationship due to its focus on emerging adults and examining PW, e-cigarettes, and illicit drug use, which are understudied.

Conclusion

Among U.S. emerging adults aged 18–28, in 2017, higher everyday discrimination was associated with current use of alcohol, marijuana, other illicit drugs, cigarettes, and e-cigarette directly and through PD and PW. Future discrimination research should not only consider the effects of race and sex but the role of mental health outcomes as mediators to understand for which group’s discrimination is a strong social determinant of substance use and the underlying mechanisms of the discrimination–substance use relationship.

Public Health Significance.

This study suggests that discrimination was associated with past 30-day use of other illicit drugs (directly and indirectly) and marijuana use through past 30-day psychological distress (PD) among White and Black young adults aged 18–28 regardless of sex. Moreover, our findings suggest that discrimination may only be associated with current cigarette smoking and alcohol use through past 30-day PD only among Black young adult males.