Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Aug 1.
Published in final edited form as: Addiction. 2022 Apr 3;117(8):2225–2234. doi: 10.1111/add.15882

Education Matters: Longitudinal Pathways to Midlife Heavy Drinking in a National Cohort of Black Americans

Nina Mulia 1,*, Jane Witbrodt 1, Katherine J Karriker-Jaffe 2, Libo Li 1, Camillia K Lui 1, Tamika Zapolski 3
PMCID: PMC9506987  NIHMSID: NIHMS1798321  PMID: 35320593

Abstract

Aims:

To estimate longitudinal pathways from childhood socioeconomic position (SEP) to educational attainment and midlife heavy drinking in Black Americans in order to identify potential points of early intervention to reduce risk for alcohol-related problems in adulthood.

Design, Setting, Participants:

Data are from 1,299 Black Americans in the US National Longitudinal Survey of Youth, followed from 1979 (ages 15-19) through 2012. Given gender differences in factors related to education and alcohol outcomes, gender-stratified path models were analyzed.

Measurements:

Youth socioeconomic indicators included parental education (approximating childhood SEP) and adolescent poverty duration. Education-related measures included high-poverty school, perceived school safety, academic problems, suspension from school, educational expectations, and educational attainment. Adulthood measures included repeated unemployment, poverty duration, and mean frequency of heavy drinking (6+ drinks/day) in young adulthood and midlife. Covariates included age, dual-parent household, marital status, early drinking onset, and family history of alcohol problems.

Findings:

For both genders, two main pathways originating from low childhood SEP flowed to educational attainment through (1) educational expectations and (2) suspension, and from educational attainment to midlife heavy drinking (total indirect effect = 0.131 (95% confidence interval [CI]: .072-.197) for women, and 0.080 (.035-.139) for men). For both genders, adolescent poverty (standardized βs ≥ 0.139, academic problems (βs ≥ 0.221) and school suspension (βs ≥0.166) were significantly (ps <.05) related to lower educational expectations. In adulthood, educational attainment was indirectly protective against midlife heavy drinking through its significant effects (ps <.05) on young adult heavy drinking for both genders (βs ≤ −.204) and economic hardships for women (βs ≤ −0.372).

Conclusions:

Low childhood socioeconomic position among Black Americans appears to be associated with subsequent, adverse socioeconomic and school experiences that lead to lower educational attainment and, ultimately, greater heavy drinking at midlife. Interventions that mitigate these earlier, adverse experiences might have indirect effects on midlife heavy drinking.

Keywords: Alcohol use, Heavy drinking, Racial and ethnic disparities, Black American, Racial and ethnic minority, Socioeconomic, Education, Life course, School suspension

INTRODUCTION

Black Americans in the US have high rates of childhood socioeconomic disadvantage, low rates of upward socioeconomic mobility 1, and elevated alcohol risks in adulthood. Despite greater abstinence rates, alcohol use disorder rates are similar to or exceed those of White adults 2, and Black Americans bear a greater burden of alcohol-related health harm, morbidity, and mortality than other racial/ethnic groups 3,4. Black drinkers also experience more alcohol-related problems than White drinkers 5–7. Yet, they remain understudied in US alcohol research.

Racial disparities in health become more pronounced in middle age 8 , and thus harmful health behaviors in midlife are important to address. Our prior US longitudinal studies of Black Americans and other racial/ethnic groups suggest enduring effects of early socioeconomic disadvantage on midlife heavy drinking but also a universal protective effect of higher education on heavy drinking trajectories for both women and men 9,10, consistent with research on harmful drinking and alcohol problems 11–14. Specifically, each additional year of education beyond 11th grade has been shown to correspond to a reduction in adult heavy drinking frequency, with the greatest decrease found for 16 or more years of education, the equivalent of a 4-year college or Bachelor’s degree 15.

Although health benefits of education have been ascribed to greater cognitive skills and wages resulting from higher education 15,16, research has increasingly highlighted the role of noncognitive skills present earlier in life – such as behavioral and socioemotional adjustment, attention, motivation, and perseverance – in educational success, and suggested that these skills might help explain the education-health gradient observed in adulthood 17–20. Importantly, new studies suggest that children’s experiences of their school environment, teachers, and activities can affect noncognitive skills linked to educational success 21,22. Thus, to the extent that school environments and experiences shape noncognitive skills, the former might hold relevance for educational attainment and heavy drinking years later.

The current study examined pathways to educational attainment and midlife heavy drinking in a national cohort of Black Americans. Given US educational inequalities by race, we were interested in school contexts and experiences that affect Black youth in particular (e.g., high-poverty or unsafe schools, suspension from school which excludes the student from classroom learning) and socioeconomic conditions that might affect noncognitive skills and educational attainment and, indirectly, adult heavy drinking. This longitudinal study extends our earlier research on racial and ethnic differences in effects of early disadvantage on midlife heavy drinking (see 10) by examining whether and how such educational contexts and school experiences link early disadvantage with educational attainment and adult socioeconomic conditions and heavy drinking in Black Americans. Recognizing that educational factors could be linked to family socioeconomic position and that parental education can influence children’s own educational attainment and adult health 23, we chose parental education — a key indicator of childhood socioeconomic position (SEP) — as the starting point for this study. In addition, given reports of gender differences in factors related to education and alcohol outcomes 24,25, gender-specific analyses were conducted to identify important predictors for each group.

Our study is informed by cumulative disadvantage theory and the chain of risks model from lifecourse epidemiology 26 which highlight how early disadvantage can lead to further disadvantages affecting subsequent socioeconomic and behavioral determinants of health across the life course. Based on the literature, we developed a conceptual model shown in Figure 1 in which low childhood SEP (low parental education) was hypothesized to be associated with adolescents’ duration of poverty and exposure to adverse school contexts (high student poverty rates and lower school safety). In the US, high-poverty schools have less school funding and resources, substandard facilities, and less experienced and underpaid teachers 27; graduation rates are more than twenty percentage points below those of low-poverty schools 28. Also, perceived school safety is strongly, positively associated with academic achievement 29. We expected adverse school contexts and adolescent poverty duration, as a marker of chronic stress, would be associated with academic difficulties, school suspension (which is associated with poor academic performance 30), and lower educational expectations, and that all three would be associated with lower educational attainment 31–35. In turn, lower educational attainment was expected to be associated with greater heavy drinking in young adulthood, adult economic hardships, and heavy drinking in midlife.

Fig. 1.

Fig. 1

Open in a new tab

Conceptual Model of Pathways from Low Childhood Socioeconomic Position to Midlife Heavy Drinking

METHODS

Data Source

This secondary analysis used survey data collected over a 30-year period from 1979 - 2012 from the National Longitudinal Survey of Youth 1979 (NLSY79), a nationally representative panel study of civilian, non-institutionalized, US-born youth with oversamples of racial/ethnic minority youth born in the late 1950s and early 1960s (for more information, see 36). The response rate (RR) in 2012 was 79% for eligible, non-deceased individuals and retention was comparable across race/ethnicity. Use of NLSY survey weights adjusted the multi-wave survey dataset to be representative of the US population at baseline 37. This analysis included 1,299 Black respondents (654 females and 645 males) ages 15 through 19 in 1979. The final path model included 1,140 respondents (see Supplemental Table S1 for sample inclusion, missing data, and sample sizes.) This study analyzed respondents’ survey data, their high school’s characteristics (the High School Survey was completed by school principals in 1980; RR=77%) and their high school transcripts (collected 1980-1983; RR=81%). This analysis of publicly available data was exempt from review by the Institutional Review Board of the Public Health Institute (Oakland, California). As our primary research question and analysis plan were not pre-registered on a publicly available platform, results should be considered exploratory.

Measures

Because the analytic sample ranged in age from 15 through 19 at baseline, data were obtained from varying survey years to create measures for a specific age or age range (see Supplemental Figure S1).

Average midlife heavy drinking frequency (ages 42-52), our key outcome variable, is the number of days drinking 6+ drinks/day in the past month based on respondent self-reports averaged across up to 4 surveys conducted biennially 2006-2012. Average young adulthood heavy drinking frequency (ages 24-30) is the drinking frequency averaged across the annual 1988 and 1989 surveys. Following prior NLSY studies, ordinal response categories were coded using the weighted empirical means of the midpoints for each category as 0.0, 1.8 (mean of once and 2-3 times per month), 5.1 (mean of 4-5 times and 6-7 times per month), and 11.0 (mean of 8-9 times and 10 or more times per month) and treated as a continuous variable 9,10,38 . Both heavy drinking measures were log-transformed to account for skewness.

Low childhood SEP, our main predictor, was operationalized as parental education less than a high school diploma (for both parents if a dual-parent household) reported at the 1979 baseline . Adolescent poverty duration was the proportion of years from ages 15-19 that parent-reported, past-year family incomes were at or below the federal poverty level (survey years 1979-1983; 15-year-olds had data from up to five years; 19-year-olds had data from 1979).

Five measures of the respondent’s school context and experiences were generated from their High School Survey (HSS), high school transcript (HST) and respondent surveys (RS). High-poverty school (1980 HSS) was defined as having at least 75% low-income students, consistent with the National Center for Education Statistics (NCES) definition 28. Unsafe school (RS 1979) was indicated by “I don’t feel safe at this school,” where responses were coded from 1 “not at all true” to 4 “very true”. Academic problems (HST) were indicated by enrollment in remedial math or English classes 10. High school suspension/expulsion (HST), hereafter referred to as suspension, indicated any incident (yes/no). Educational expectations (RS 1979-82) were captured by the question, “As things stand now, what is the highest grade or year you think you will complete?” The most recent available response was used and dichotomized to indicate lower expectations (completing less than 16 years of education).

Educational attainment (RS 1988-2012) was a categorical measure indicating highest education level completed: did not receive a high school diploma by age 19, graduated high school “on time” by age 19, or received a Bachelor’s degree by age 25. “On time” high school completion is associated with greater educational attainment, higher-wage employment, and health-promoting behaviors 39.

Adult economic hardship was indicated by repeated unemployment and adult poverty duration. Repeated unemployment was operationalized as unemployed at 2 or more interviews at ages 26-38 (RS 1990-1998, asked biennially). Recognizing potential effects of repeated unemployment on subsequent household income, adult poverty duration was assessed at older ages (36-42, RS 1996-2006) and was constructed like adolescent poverty duration.

Control variables

Early onset of alcohol use by age 14, reported retrospectively (R 1982+), is a predictor of heavy drinking in adolescence 40 and young adulthood, particularly for Black Americans 41. To assess family history of alcohol problems, respondents were asked whether any relatives were “alcoholic or a problem drinker at any time”, and specified their relationships to up to seven alcoholic relatives (RS 1988). Following Chartier and colleagues (2017), our family history predictor was coded using a mean weighted score for all first- and second-degree relatives with alcohol problems 42. Additional control variables included baseline age, dual-parent household at baseline, and respondent marital status. Because Mplus drops individuals with missing values on predictors, missing values on categorical predictor variables were dummy coded and included as additional control variables. Missing values for family history (continuous) were not replaced.

Analysis

Preliminary analyses included polychoric correlations of indicators and careful inspection of model assumptions 43,44. Gender-stratified longitudinal path models conducted in Mplus 45 tested hypothesized relationships (see Figure 1), following a model generating approach whereby a theoretically driven path model was specified and tested 43,44. School context indicators were allowed to correlate, as were school problem indicators (curved arrows in model figures). Based on recommended practices, preliminary analyses, model fit statistics 46–48, and modification indices 49, reduced models that were theoretically sound and consistent with our conceptual model were subsequently estimated to determine a parsimonious model 50,51 (see Figures 2 and 4 notes for model fit statistics). For each path in the overall model, non-significant control variables were trimmed to preserve degrees of freedom. The final model was re-estimated using a bias-corrected bootstrap re-sampling method to derive standard errors and confidence limits and non-symmetrical bootstrap confidence intervals for model parameter estimates 43,52. Models were estimated using the robust weighted least squares (WLSMV) estimator 53,54 which accommodates missing data, handles covariance among outcomes and mediators, and provides fit indices for model evaluation 55.

Fig. 2.

Fig. 2

Open in a new tab

Standardized Estimates for Black American Women

Notes: All modeled standardized direct paths are displayed. Solid lines and * indicate p <.05. Observations = 583 (71 observations were excluded from the final path model due to missing values on either family history of alcohol problems or low childhood SEP), df = 105, CFI = 0.995, TLI = 0.992, RMSEA estimate = 0.008 (p<=.05), R2 midlife heavy drinking (0.223), total effect (β = 0.131, CI= 0.072-0.197).

Fig. 4.

Fig. 4

Open in a new tab

Standardized Estimates for Black American Men

Notes: All modeled standardized direct paths are displayed. Solid lines and * indicate p <.05. Observations = 557 (88 observations were excluded from the final path model due to missing values on either family history of alcohol problems or low childhood SEP), df = 111, CFI = 0.983 , TLI = 0.976, RMSEA estimate = 0.011 (p<=.05), R2 midlife heavy drinking (0.200), total effect (β=0.080, CI=0.035-0.139)

Results

Table 1 shows the proportion or mean of key indicators and correlations by gender (females above the diagonal; variances are shown on the diagonal).Youth experienced substantial socioeconomic disadvantage (e.g., 43.4% had low childhood SEP and repeated poverty was common). Large proportions (43.6%) attended a high-poverty school or had lower educational expectations (56.7%). School suspension was experienced by 39.5%, which is considerably higher than the NLSY sample overall (<25% suspended) and consistent with the US racial disparity in school suspension 56. Two-thirds (67.2%) graduated high school on time but did not attain a 4-year college degree; 22.3% did not graduate high school. Overall, 37.8% (26.7% females, 49.1% males) drank heavily in young adulthood, and 25.0% (16.4% females, 34.0% males) drank heavily at midlife (percentages not shown in Table).

Table 1.

Variable proportions, means, and correlations by gender (females n=654 and males n=645)1

All F M 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
1 Low childhood SEP (%) 43.4 45.7 41.2 -- .37 .19 .06 .42 .26 .10 −.39 .06 .24 .22 .03 −.04 −.08 −.10 −.06 −.09
2 Adol. poverty dur. (m) 40.1 0.41 0.39 .29 .17
.16 		.20 .09 .36 .21 .24 −.33 .09 .22 .23 .09 −.03 .09 .01 −.32 −.03
3 High-poverty sch. (%) 43.6 43.7 43.5 .04 .19 -- .14 .17 .02 .03 −.13 −.07 .04 .12 .01 .07 −.09 −.04 −.11 −.07
4 Unsafe sch. score (m) * 1.62 1.71 1.54 .09 .02 −.08 .79
.81 		.16 .13 −.02 −.19 −.02 −.13 .08 .06 −.05 .27 −.02 −.08 −.02
5 Lower educ. expect. (%) 56.7 54.7 58.7 .33 .22 .05 .09 -- .38 .31 −.72 .07 .17 .36 .32 .01 .17 .01 −.20 −.07
6 Academic probs. (%) 34.7 30.9 38.4 .15 .22 .11 .15 .32 -- .24 −.43 .18 .22 .28 .06 −.06 .07 −.02 −.14 −.07
7 Suspend/expel (%) * 39.5 29.6 49.5 .19 .15 −.01 .13 .19 .12 -- −.39 .20 .34 .32 .13 .02 −.09 .07 −.30 −.16
8 Educ. attain. HS (%)2 * 67.2 71.5 63.0 −.35 −.25 −.16 −.08 −.64 −.35 −.35 -- −.18 −.25 −.43 −.12 −.05 −.04 −.10 .22 .02
9 YA HD freq. (m)3 * 1.09 0.57 1.63 .06 .03 .07 .05 .11 .07 .23 −.19 1.87
6.58 		.15 .24 .28 .06 .14 .08 −.10 −.22
10 Repeated unempl. (%) 4.49 4.37 4.61 −.03 .10 .33 .13 .20 .16 .05 −.24 .01 -- .32 .14 −.07 −.04 −.08 −.00 −.34
11 Adult poverty dur. (m) * 0.26 0.31 0.22 .13 .08 .06 .03 .21 .28 .24 −.37 .07 .31 .16
.12 		.17 .01 .16 .08 −.23 −.27
12 Midlife HD freq. (m)4 0.54 0.29 0.79 .09 .07 .14 .04 .06 .18 .11 −.13 .37 .12 .09 .98
2.67 		−.00 −.05 −.02 −.02 −.16
13 Age at interview (m) 17.1 17.1 17.0 .03 −.05 −.08 −.09 −.01 −.13 .08 −.12 .04 −.02 .01 −.04 1.86
1.94 		−.08 .09 −.18 −.02
14 Alc. onset age ≤14 (%)* 6.98 2.42 11.1 .04 .06 −.05 −.02 .07 −.12 .11 −.16 .08 −.30 . 08 .11 .00 -- −.15 −.22 −.04
15 Fam. alc. history (m) * 0.07 0.08 0.06 −.00 −.09 −.11 −.01 .10 −.11 .13 .09 −.06 −.13 .01 −.04 .02 .04 .01
.01 		−.13 −.13
16 Two-parent HH (%) * 54.4 51.1 57.6 −.02 −.29 .11 .06 −.04 −.02 −.15 .06 .01 .02 −.14 .03 −.10 −.21 −.21 -- .16
17 Married at age 25 (%) * 28.4 32.5 24.2 −.01 −.14 −.12 −.01 −.04 −.00 .03 .13 −.11 −.41 −.21 −.12 −.12 −.13 .12 .24 --
Open in a new tab

NOTES.

*

p<.05 when comparing males with females; correlational coefficients for females shown above the diagonal and males below the diagonal. Variances are shown on the diagonal with female values displayed above male values. Low childhood SEP = parent did not graduate high school; m = mean value; adol.=adolescent; dur.=duration; sch.=school; educ.=educational; HS= high school graduate; expect.= expectations; probs.=problems; YA= young adult; HD = heavy drinking; unempl.=unemployment; freq.=frequency; alc=alcohol; Fam. alc. history=Family alcohol history score; HH= household;

1

Educational attainment HS = High school diploma by age 19 but no Bachelor’s degree. Other categories not shown in table: Educational attainment < HS degree: total=22.3%, female=17.1%, male=27.6%; Bachelor’s degree or higher by age 25: total=10.5%, female=11.5%, male=9.5%

2

Inclusion criteria: Black American (n=2,805), ages 15-19 (n=1,951) and NLSY school survey data available (n=1,299), resulting in females n=654 and males n=645.

3

% who engaged in YA HD (not included in table): 37.8%

4

% who engaged in midlife HD (not included in table): 25.0%

For both genders, socioeconomic disadvantage, academic problems, suspension, and lower educational expectations were negatively correlated with educational attainment (all r < −.30). Educational attainment was negatively correlated with adult economic hardships (r’s < −.23). Correlates of midlife heavy drinking included young adult heavy drinking for both genders (r’s > .27), academic problems for males (r =.18), and lower educational expectations for females (r = .32).

Females

Figure 2 shows standardized coefficients and standard errors from the final adjusted model for the sample of women. The model achieved good fit (see Figure 2 notes) and most of the hypothesized direct effects were upheld. Low childhood SEP was strongly associated with adolescent poverty (β=.607), which was significantly associated with high-poverty school (β=.175), academic problems (β=.200), suspension (β=.195), and lower educational expectations (β=.146). Notably, suspension (β=.222) and academic problems (β=.221) were significantly related to lower educational expectations for Black adolescent girls, and two factors – lower educational expectations (β=−.525) and suspension (β=−.264) – were significantly associated with educational attainment. As expected, educational attainment was inversely associated with young adult heavy drinking frequency (β=−.254), repeated unemployment (β=−.372) and adult poverty duration (β=−.513), and these three factors were associated with midlife heavy drinking frequency (β’s = 0.359, 0.173, and 0.145, respectively). The adjusted model explained 22% of variance in Black women’s midlife heavy drinking.

Significant indirect pathways from low childhood SEP to educational attainment and midlife heavy drinking were observed (total indirect effect = 0.131, 95% CI= 0.072-0.197, see Figure 3 notes). One set of pathways originating in low childhood SEP flowed through educational expectations (directly and indirectly via adolescent poverty) and educational attainment, and onward through young adult heavy drinking or adult economic hardships. Another major pathway flowed through adolescent poverty and school suspension to lower educational attainment and either young adult heavy drinking or adult poverty.

Fig. 3.

Fig. 3

Open in a new tab

Significant Indirect Paths for Black American Women

Total indirect effect: β=.131, 95% CI=.072-.197. Standardized bias-corrected bootstrap coefficients and 95% confidence intervals for significant indirect pathways:
Low SEP --> low expectn --> educ attainmt --> YA heavy drkg --> Midlife heavy drinking (β=.023, CI=.010-.045)
Low SEP --> low expectn --> educ attainmt --> adult poverty --> Midlife heavy drinking (β=.019, CI=.004-.043)
Low SEP --> low expectn --> educ attainmt --> unemplymt --> Midlife heavy drinking (β=.016, CI=.002-.046)
Low SEP --> adol poverty --> low expectn --> educ attainmt --> YA heavy drkg --> Midlife heavy drinking (β=.004, CI=.001-.011)
Low SEP --> adol poverty --> suspension --> educ attainmt --> YA heavy drkg --> Midlife heavy drinking (β=.003, CI=.001-.008)
Low SEP --> adol poverty --> suspension --> educ attainmt --> adult poverty --> Midlife heavy drinking (β=.002, CI=.001-.007)
Open in a new tab

Males

Figure 4 shows results for the final model for men, with fit indices showing good model fit (see Figure 4 notes). Many hypothesized, direct effects held for males. Low childhood SEP was strongly associated with adolescent poverty (β=.463); the latter was significantly associated with high-poverty school (β=.172), academic problems (β=.193), and lower educational expectations (β=.139). For males, adolescent poverty was not associated with perceived school safety or suspension. As found for females, both academic difficulties (β=.252) and suspension (β=.166) were significantly associated with lower educational expectations, and lower expectations (β=−.491), suspension (β=−.264), and academic problems (β=.−187) were related to educational attainment. In turn, educational attainment was negatively associated with young adult heavy drinking (β=−.204), repeated unemployment (β=−.295), and adult poverty (β=−.401). Young adult heavy drinking (β=.380) was significantly associated with midlife heavy drinking. The adjusted model explained 20% of variance in midlife heavy drinking frequency among men.

Compared to women, men evidenced fewer significant, indirect pathways from low childhood SEP to educational attainment and midlife heavy drinking (total indirect effect = .077, 95%CI=.038-.123, see Figure 5 notes). One main pathway flowed through lower educational expectations to educational attainment while another flowed through school suspension to educational attainment, and from there to young adult heavy drinking and midlife heavy drinking. Unlike women, men showed significant, residual effect of low childhood SEP on educational attainment (β=−.250).

Fig. 5.

Fig. 5

Open in a new tab

Significant Indirect Paths for Black American Men

Total indirect effect: β=.077, 95% CI=.038-.123. Standardized bias-corrected bootstrap coefficients and 95% confidence intervals for significant indirect pathways:
Low SEP --> educ attainmt --> YA heavy drinking --> Midlife heavy drinking (β=.014, CI=.002-.034)
Low SEP --> low expectatn. --> educ attainmt --> YA heavy drinking --> Midlife heavy drinking (β=.016, CI=.006-.031)
Low SEP --> suspension --> educ attainmt --> YA heavy drinking --> Midlife heavy drinking (β=.006, CI=.002-.014)
Open in a new tab

DISCUSSION

This study investigated socioeconomic and education pathways from low childhood SEP to midlife heavy drinking for a US national cohort of Black Americans. Low childhood SEP characterized more than 40% of the sample and was associated with a sequence of adverse socioeconomic and school experiences that led to lower educational attainment and, ultimately, greater heavy drinking at midlife. For both genders, pathways to midlife heavy drinking occurred via young adult heavy drinking and, for women, also through repeated unemployment and greater adult poverty duration.

Lower educational expectations and school suspension emerged as important factors in pathways from low childhood SEP to lower educational attainment and midlife heavy drinking. For both genders, adolescent poverty, academic problems, and suspension were associated with lower expectations of educational attainment. For females, lower expectations were also associated with attending an unsafe school, and for males expectations were independently associated with low childhood SEP. In contrast to educational expectations, there were few paths to school suspension, which was associated with greater adolescent poverty duration for females and low childhood SEP for males.

Our findings are consistent with studies of racially/ethnically diverse groups indicating strong associations between educational expectations and educational attainment, and between early socioeconomic disadvantage and lower educational expectations 57–59. Unlike educational aspirations, which are hope-based, expectations reflect a student’s perception of likelihoods 58. In our study, academic problems and suspension (whereby students are excluded from classroom learning) were associated with lower educational expectations. Importantly, academic problems and suspension have both been linked to implicit bias against students of color 60. School-level strategies such as implicit-bias trainings and culturally relevant curricula might help to mitigate this (e.g., see 61). In addition, programs that provide academic support, mentorship by adults from similar backgrounds, experiential opportunities such as college campus visits, job shadowing and training programs, and student financial aid could help broaden educational expectations of disadvantaged youth. The Harlem Children’s Zone is one prominent example of a comprehensive, multi-level program from birth to college that has increased academic achievement of disadvantaged youth and reduced racial disparities 62.

Our finding that suspension comprises a key pathway from low childhood SEP to lower educational attainment and midlife heavy drinking warrants special attention. Highlighting long-term consequences of suspension, a recent study matched suspended and non-suspended youth on 60 pre-suspension variables including parental SEP and youth academic problems, delinquency, and substance use. Suspended Black youth were 94% less likely to earn a Bachelor’s degree in contrast to a 24% lower likelihood for suspended youth in the sample overall 63. Thus, Black youth are both differentially exposed to and impacted by school suspension practices. Notably, a study of elementary school children in 20 US cities indicated that Black-White disparities in suspension were largely due to differential treatment by school personnel, followed by differences in school environments and, to a much lesser extent, behavioral differences 64. Some school districts have replaced exclusionary discipline with restorative justice practices that hold promise for reducing disparities in both suspension and academic performance 65. Our study supports the importance of such efforts by highlighting long-term consequences of school suspension on midlife health risk behavior for Black Americans.

Independent of these two main pathways involving educational expectations and suspension, there remained a residual, direct effect of low childhood SEP on educational attainment for men. Research on gender differences in effects of lower SEP on children’s educational achievement suggests low-SEP boys may be more vulnerable than low-SEP girls to lower achievement later in childhood 24. Such gender differences have been found as early as elementary school 24,66 .

A final notable finding concerns pathways from educational attainment to midlife heavy drinking via young adult heavy drinking and also, for women, economic hardships. Our results point to the importance of early alcohol screening and brief intervention with young adults, and suggest that interventions to reduce poverty and help sustain employment during the transition out of young adulthood might reduce Black women’s heavy drinking risk later in life.

The finding that repeated unemployment and chronic poverty were significant paths to heavy drinking for Black women only is consistent with our prior studies showing that adult poverty and cumulative socioeconomic disadvantage are risk factors for heavy drinking among Black women but not Black men 9,25. This might reflect differences in acceptable (unmeasured) drinking norms for women and men; that is, men’s heavy drinking might be generally more influenced by permissive drinking norms. Additionally, (unmeasured) experiences of racial discrimination might significantly influence Black men’s drinking (e.g., see 7,67). If true, this might help explain why paths from acute economic hardships to heavy drinking were weaker in men than women.

This study has notable strengths, including the use of multiple data sources and prospective analysis across more than three decades. But several study limitations should be borne in mind. WLSMV assumes missing at random with respect to covariates and also excludes cases with missing values on predictors and covariates. Importantly, sensitivity analysis indicated our path model results were robust for women and men, although possibly conservative in male estimates of adverse effects of low childhood SEP on unsafe school, and unsafe school on academic problems. Additionally, findings may not generalize to more recent birth cohorts of Black Americans. Data were not available on neighborhood conditions such as alcohol availability and neighborhood poverty, although attendance at a high-poverty school might be a proxy for neighborhood poverty. Available data did not permit analysis of reciprocal effects between adolescent drinking and educational achievement reported in prior studies 68–70 so this bears replication in future research. However, to estimate independent effects of educational attainment on young adult and midlife heavy drinking, we accounted for early onset of drinking which is associated with both escalation of drinking among Black adolescents 41 and, potentially, educational attainment. Thus, path coefficients from educational attainment to heavy drinking represent effects independent of early drinking onset. Because available heavy drinking data captured 6+ drinks/day, which exceeds the US National Institute on Alcohol Abuse and Alcoholism’s threshold of 5+/4+ drinks/day for men and women, respectively, observed associations with heavy drinking might be conservative. Finally, future research on heavy drinking pathways could investigate the roles of neighborhood conditions, cannabis and other drug use, and protective and resilience factors that support paths to higher educational attainment and lower-risk drinking.

Conclusion

Education across the life course matters in various ways for Black Americans’ heavy drinking at midlife. Observed pathways to educational attainment highlight intergenerational effects of parents’ own education which affects children’s exposure to adolescent poverty. High school students’ experiences of suspension and lower educational expectations also strongly influence educational attainment. The latter, in turn, affects risk for subsequent young adult heavy drinking and, for women, repeated unemployment and chronic poverty during the transition out of young adulthood. Our findings suggest several possible intervention targets for disrupting pathways to midlife heavy drinking among Black Americans. Multisector efforts to reduce early poverty, alongside school-based strategies to address implicit racial bias, replace exclusionary school discipline practices, and elevate educational expectations and academic performance, could help increase educational attainment of Black Americans. If successful, an array of social and health benefits could follow 71, including population-level reductions in harmful drinking.

Supplementary Material

Mulia et al. 2022 Education Matters Suppl Table S1
Mulia et al. 2022 Education Matters Suppl Fig S1

Acknowledgments

The authors gratefully acknowledge Edwina Williams, MPH, for contributions to the construction of the analytic data set and Tammy Tam, PhD, for developing our project’s original data structure charts (modified for Supplemental Figure S1). This study was supported by funding from the National Institutes of Health (NIH) (National Institute on Alcohol Abuse and Alcoholism (NIAAA) R01AA022668 and R01AA026956, N. Mulia, PI; National Institute on Drug Abuse (NIDA) K01DA043654, T. Zapolski, PI). Content is solely the responsibility of the authors and does not represent official views of NIAAA, NIDA or the NIH, which had no role in the study design, collection, analysis or interpretation of the data, writing of the manuscript, or the decision to submit the manuscript for publication.

Footnotes

Conflicts of Interest: None.

References

  • 1.Chetty R, Hendren N, Jones MR, Porter SR. Race and economic opportunity in the United States: an intergenerational perspective. Q J Econ. 2020;135(2):711–783. [Google Scholar]
  • 2.Grant BF, Goldstein RB, Saha TD, et al. Epidemiology of DSM-5 Alcohol Use Disorder: results from the National Epidemiologic Survey on Alcohol and Related Conditions III. JAMA Psychiatry. 2015;72(8):757–766. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Chartier KG, Vaeth PAC, Caetano R. Focus On: Ethnicity and the social and health harms from drinking. Alcohol Res. 2013;35(2):229–237. [PMC free article] [PubMed] [Google Scholar]
  • 4.Shield KD, Gmel G, Kehoe-Chan T, Dawson DA, Grant BF, Rehm J. Mortality and potential years of life lost attributable to alcohol consumption by race and sex in the United States in 2005. PLoS ONE. 2013;8(1):e51923. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Grant JD, Vergés A, Jackson KM, Trull TJ, Sher KJ, Bucholz KK. Age and ethnic differences in the onset, persistence and recurrence of alcohol use disorder. Addiction. 2012;107(4):756–765. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Witbrodt J, Mulia N, Zemore SE, Kerr WC. Racial/ethnic disparities in alcohol-related problems: differences by gender and level of heavy drinking. Alcohol Clin Exp Res. 2014;38(6):1662–1670. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Zapolski TCB, Pedersen SL, McCarthy DM, Smith GT. Less drinking, yet more problems: understanding African American drinking and related problems. Psychol Bull. 2014;140(1):188–223. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Adler NE, Stewart J. Health disparities across the lifespan: meaning, methods, and mechanisms. Ann N Y Acad Sci. 2010;1186(1):5–23. [DOI] [PubMed] [Google Scholar]
  • 9.Mulia N, Karriker-Jaffe KJ, Witbrodt J, Bond J, Williams E, Zemore SE. Racial/ethnic differences in 30-year trajectories of heavy drinking in a nationally representative U.S. sample. Drug Alcohol Depend. 2017;170:133–141. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Karriker-Jaffe KJ, Witbrodt J, Mulia N. Differential consequences: racial/ethnic and gender differences in the enduring impact of early disadvantage on heavy drinking in midlife. Prev Sci. 2019;20(7):1009–1020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Grant BF. Prevalence and correlates of alcohol use and DSM-IV alcohol dependence in the United States: results of the National Longitudinal Alcohol Epidemiologic Survey. J Stud Alcohol. 1997;58(5):464–473. [DOI] [PubMed] [Google Scholar]
  • 12.Droomers M, Schrijvers CTM, Stronks K, van de Mheen D, Mackenbach JP. Educational differences in excessive alcohol consumption: the role of psychosocial and material stressors. Prev Med. 1999;29(1):1–10. [DOI] [PubMed] [Google Scholar]
  • 13.Casswell S, Pledger M, Hooper R. Socioeconomic status and drinking patterns in young adults. Addiction. 2003;98(5):601–610. [DOI] [PubMed] [Google Scholar]
  • 14.Karlamangla A, Zhou K, Reuben D, Greendale G, Moore A. Longitudinal trajectories of heavy drinking in adults in the United States of America. Addiction. 2006;101(1):91–99. [DOI] [PubMed] [Google Scholar]
  • 15.Cutler DM, Lleras-Muney A. Understanding differences in health behaviors by education. J Health Econ. 2010;29(1):1–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Galobardes B, Shaw M, Lawlor DA, Lynch JW, Smith GD. Indicators of socioeconomic position (part 1). J Epidemiol Community Health. 2006;60(1):7–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Conti G, Heckman J, Urzua S. The education-health gradient. Am Econ Rev. 2010;100(2):234–238. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Conti G, Hansman C. Personality and the education-health gradient: A note on “Understanding differences in health behaviors by education”. J Health Econ. 2013;32(2):480–485. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Khine MS. Non-cognitive skills and factors in educational success and academic achievement. In: Khine MS, Areepattamannil S, eds. Non-cognitive Skills and Factors in Educational Attainment. Vol 9. Rotterdam: Sense Publishers; 2016:3–9. [Google Scholar]
  • 20.McIntyre JR, Vecchione R. Considering non-cognitive factors in the preparation and selection of educators. In: Khine MS, Areepattamannil S, eds. Non-cognitive skills and factors in educational attainment. Rotterdam: Sense Publishers; 2016. [Google Scholar]
  • 21.Shubert J, Wray-Lake L, McKay B. Looking ahead and working hard: How school experiences foster adolescents’ future orientation and perseverance. J Res Adolesc. 2020;30(4):989–1007. [DOI] [PubMed] [Google Scholar]
  • 22.Liu A. Can non-cognitive skills compensate for background disadvantage? -- the moderation of non-cognitive skills on family socioeconomic status and achievement during early childhood and early adolescence. Soc Sci Res. 2019;83:102306. [DOI] [PubMed] [Google Scholar]
  • 23.Ross CE, Mirowsky J. The interaction of personal and parental education on health. Soc Sci Med. 2011;72(4):591–599. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Mensah FK, Kiernan KE. Gender differences in educational attainment: influences of the family environment. British Educational Research Journal. 2010;36(02):239–226. [Google Scholar]
  • 25.Mulia N, Tam T, Bond J, Zemore SE, Li L. Racial/ethnic differences in life-course heavy drinking from adolescence to midlife. J Ethn Subst Abuse. 2018;17(2):167–186. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Kuh D, Ben-Shlomo Y, Lynch J, Hallqvist J, Power C. Life course epidemiology. J Epidemiol Community Health. 2003;57(10):778–783. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.United States Commission on Civil Rights. Public Education Funding Inequity in an Era of Increasing Concentration of Poverty and resegregation. Washington, DC: United States Commission on Civil Rights;2018. [Google Scholar]
  • 28.Aud S, Hussar W, Planty M, et al. The Condition of Education 2010 (NCES 2010–028). In: Vol 2021. Washington, DC: National Center for Education Statistics, Institute of Education Sciences, U.S.; 2010: https://nces.ed.gov/pubs2010/2010028.pdf. [Google Scholar]
  • 29.Milam AJ, Furr-Holden CDM, Leaf PJ. Perceived school and neighborhood safety, neighborhood violence and academic achievement in urban school children. Urban Rev. 2010;42(5):458–467. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Arcia E. Achievement and enrollment status of suspended students: outcomes in a large, multicultural school district. Educ Urban Soc. 2006;38(3):359–369. [Google Scholar]
  • 31.Owens A. Neighborhoods and schools as competing and reinforcing contexts for educational attainment. Sociol Educ. 2010;83(4):287–311. [Google Scholar]
  • 32.Chen P, Vazsonyi AT. Future orientation, school contexts, and problem behaviours: a multilevel study. J Youth Adolesc. 2013;42(1):67–81. [DOI] [PubMed] [Google Scholar]
  • 33.Hintsanen M, Hintsa T, Merjonen P, Leino M, Keltikangas-Järvinen L. Family- and school-related factors in 9- to 15-year-olds predicting educational attainment in adulthood: a prospective 27-year follow-up study. Electronic Journal of Research in Educational Psychology. 2011;9(2):523–540. [Google Scholar]
  • 34.Skiba RJ, Chung C-G, Trachok M, Baker TL, Sheya A, Hughes RL. Parsing disciplinary disproportionality: contributions of infraction, student, and school characteristics to out-of-school suspension and expulsion. Am Educ Res J. 2014;51(4):640–670. [Google Scholar]
  • 35.Bertrand M, Pan J. The trouble with boys: social influences and the gender gap in disruptive behavior. Am Econ J Appl Econ. 2013;5(1):32–64. [Google Scholar]
  • 36.Rothstein DS, Carr D, Cooksey E. Cohort profile: The National Longitudinal Survey of Youth 1979. Int J Epidemiol. 2019;48(1):22–22e. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Bureau of Labor Statistics. National Longitudinal Surveys: Custom weighting program documentation [Accessed: 2016-04-20. Archived by WebCite® at http://www.webcitation.org/6guiInSY0]. Washington, DC: 2016. [Google Scholar]
  • 38.Williams E, Mulia N, Karriker-Jaffe KJ, Lui CK. Changing racial/ethnic disparities in heavy drinking trajectories through young adulthood: a comparative cohort study. Alcohol Clin Exp Res. 2018;42(1):135–143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.APHA’s Center for School Health and Education. Chronic Stress and the Risk of High School Dropout. American Public Health Association; February 2018. [Google Scholar]
  • 40.Hingson RW, Heeren T, Winter MR. Age of alcohol-dependence onset: associations with severity of dependence and seeking treatment. Pediatrics. 2006;118(3):e755–e763. [DOI] [PubMed] [Google Scholar]
  • 41.Horton EG. Racial differences in the effects of age of onset on alcohol consumption and development of alcohol-related problems among males from mid-adolescence to young adulthood. J Ethn Subst Abuse. 2007;6(1):1–13. [DOI] [PubMed] [Google Scholar]
  • 42.Chartier KG, Thomas NS, Kendler KS. Interrelationship between family history of alcoholism and generational status in the prediction of alcohol dependence in US Hispanics. Psychol Med. 2017;47(1):137–147. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Barbeau K, Boileau K, Sarr F, Smith K. Path analysis in Mplus; A tutorial using a conceptual model of psychological and behavioral antecedent of bulimic symptoms in young adults. The Quantitative Methods for Psychology. 2019;15(1):38–53. [Google Scholar]
  • 44.Ullman JB, Bentler PM. Structural Equation Modeling. In: Weiner IB, ed. Handbook of Psychology, Second Edition. Vol 2. Wiley; 2012. [Google Scholar]
  • 45.Mplus Version 7.2 [computer program]. Los Angeles, CA: Muthén & Muthén; 2013. [Google Scholar]
  • 46.Steiger JH. Understanding the limitations of global fit assessment in structural equation modeling. Pers Individ Dif. 2007;42(5):893–898. [Google Scholar]
  • 47.Cangur S, Ercan I. Comparison of model fit indices used in structural equation modeling under multivariate normality. Journal of Modern Applied Statistical Methods. 2015;14(1):152–167. [Google Scholar]
  • 48.Weston R, Gore PA Jr.. A brief guide to structural equation modeling. Couns Psychol. 2006;34(5):719–751. [Google Scholar]
  • 49.Byrne BM. Structural equation modeling with Mplus: Basic concepts, applications, and programming. New York: Routledge; 2012. [Google Scholar]
  • 50.Raykov T, Marcoulides GA. On multilevel model reliability estimation from the perspective of structural equation modeling. Struct Equ Modeling. 2006;13(1):130–141. [Google Scholar]
  • 51.Bentler PM. Comparative fit indexes in structural models. Psychol Bull. 1990;107(2):238–246. [DOI] [PubMed] [Google Scholar]
  • 52.MacKinnon DP. Introduction to statistical mediation analysis. New York: Laurence Erlbaum Associates; 2008. [Google Scholar]
  • 53.Hu L-t Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct Equ Modeling. 1999;6(1):1–55. [Google Scholar]
  • 54.Muthén LK, Muthén BO. Mplus User’s Guide, Version 8. Los Angeles, CA: Muthén & Muthén; 2017. [Google Scholar]
  • 55.Asparouhov T, Muthen B. Weighted Least Squares Esimtation with Missing Data. 2010. https://www.statmodel.com/download/GstrucMissingRevision.pdf. Accessed 10-07-2020.
  • 56.United States Government Accountability Office. K-12 Education: Discipline Disparities for Black Students, Boys, and Students with Disabilities [GAO-18-258]. 2018. https://www.gao.gov/products/GAO-18-258.
  • 57.Ou S-R, Reynolds AJ. Predictors of educational attainment in the Chicago longitudinal study. School Psychology Quarterly. 2008;43(2):199–229. [Google Scholar]
  • 58.Beal SJ, Crockett LJ. Adolescents’ occupational and educational aspirations and expectations: links to high school activities and adult educational attainment. Dev Psychol. 2010;46(1):258–265. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Boxer P, Goldstein SE, DeLorenzo T, Savoy S, Mercado I. Educational aspiration-expectation discrepancies: relation to socioeconomic and academic risk-related factors. J Adolesc. 2011;34(4):609–617. [DOI] [PubMed] [Google Scholar]
  • 60.Chin MJ, Quinn DM, Dhaiiwal TK. Bias in the air: a nationwide exploration of teachers’ implicit racial attitudes, aggregate bias, and student outcomes. Educ Res. 2020;49(8):566–578. [Google Scholar]
  • 61.Dee TS, Penner EK. The causal effects of cultural relevance: evidence from an ethnic studies curriculum. Am Educ Res J. 2017;54(1):127–166. [Google Scholar]
  • 62.Dobbie W, Fryer RG Jr.. Are high-quality schools enough to increase achievement among the poor? Evidence from the Harlem Children’s Zone. Am Econ J Appl Econ. 2011;3(3):158–187. [Google Scholar]
  • 63.Rosenbaum JE. Educational and criminal justice outcomes 12 years after school suspension. Youth Soc. 2020;52(4):515–547. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Owens J, McLanahan SS. Unpacking the drivers of racial disparities in school suspension and expulsion. Soc Forces. 2020;98(4):1548–1577. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65.Morgan H. Restorative justice and the school-to-prison pipeline: a review of existing literature. Education Sciences. 2021;11(4):159. [Google Scholar]
  • 66.Driessen G, van Langen A. Gender differences in primary and secondary education: are girls really outperforming boys? International Review of Education. 2013;59:67–86. [Google Scholar]
  • 67.Kogan SM, Bae D. Racial discrimination, protective parenting, and binge drinking among emerging adult Black men. Alcohol Clin Exp Res. 2020;44(11):2343–2349. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68.Bradley BJ, Greene AC. Do health and education agencies in the United States share responsibility for academic achievement and health? A review of 25 years of evidence about the relationship of adolescents’ academic achievement and health behaviors. J Adolesc Health. 2013;52(5):523–532. [DOI] [PubMed] [Google Scholar]
  • 69.Patrick ME, Schulenberg JE, O’Malley PM. High school substance use as a predictor of college attendance, completion, and dropout: a national multicohort longitudinal study. Youth Soc. 2016;48(3):425–447. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 70.McLeod JD, Uemura R, Rohrman S. Adolescent mental health, behavior problems, and academic achievement. J Health Soc Behav. 2012;53(4):482–497. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 71.Freudenberg N, Ruglis J. Reframing school dropout as a public health issue. Prev Chronic Dis. 2007;4(4):A107. [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Mulia et al. 2022 Education Matters Suppl Table S1
NIHMS1798321-supplement-Mulia_et_al__2022_Education_Matters_Suppl_Table_S1.pdf (314.9KB, pdf)
Mulia et al. 2022 Education Matters Suppl Fig S1
NIHMS1798321-supplement-Mulia_et_al__2022_Education_Matters_Suppl_Fig_S1.pdf (434.4KB, pdf)

RESOURCES