Excessive Drinking Among African American Men: Individual and Contextual Correlates

Abstract

In this paper we explored associations of multiple domains with regular drinking and getting drunk among adult African American men. Questionnaire-based, computer-assisted interviews were conducted with 484 men in Atlanta, Georgia. Data analysis involved multivariate logistic regression analyses. Findings show that being older increased the odds of both drinking behaviors. Sensation seeking increased the odds of regular drinking and having experienced childhood sexual and physical abuse increased the odds of getting drunk. Having health insurance reduced the odds of both outcomes. Insurance coverage and the heterogeneity among adult African American men must be considered in risk reduction efforts.

INTRODUCTION

This paper focuses on alcohol use, specifically regular drinking and having been drunk, among adult African American men. Alcohol is a legal recreational drug in U.S. society and its consumption is accepted as long as people control their intake so that there is no harm to the drinker and there are no negative social and health consequences. Such consequences to the user and society at large are well known (Hasin, Stinson, Ogburn, & Grant, 2007; Hedden, Malcolm, & Latimer, 2009; Kuramoto, Martins, Ko, & Chilcoat, 2011; Substance Abuse and Mental Health Services Administration, 2010). Regular drinking and having been drunk both symbolize excessive, therefore socially unacceptable drinking (Gusfield, 1996).

One the most valuable population-based surveys on alcohol consumption in the United States is the National Survey on Drug Use and Health (NSDUH). According to the most recent NSDUH, 51.9% of the US population aged 12 or older reported current alcohol use, and approximately one-half of current drinkers indicated binge drinking (SAMHSA, 2010). The survey results also reveal that prevalence rates for illicit drug use are only slightly higher than those for heavy drinking (9% and 7%, respectively).

Alcohol use rates are lower among African-Americans than among whites (Hasin, et al., 2007; Hedden, et al., 2009). NSDUH data on alcohol use for 2009 show a prevalence rate of 57% for whites as compared to 47% among African Americans (SAMSHA, 2010). Studies on alcohol use often have included members of racial and ethnic minority populations as a comparison group to whites (Kerr, Patterson, & Greenfield, z2009; Mulia, Ye, Greenfield, & Zemore, 2009). Those that do focus on minority populations tend to provide a negative picture. For example, many of the studies on the use of alcohol and other drugs among African Americans focus on the use of illicit substances alone (Operario, Smith, Arnold, & Kegeles, 2009) or combined with alcohol (Reisner et al., 2010), thereby ignoring the unique impact of the use of licit substances such as alcohol. This is especially true when the focus is on those who reside in inner-city neighborhoods. This study has an urban focus with the study area consisting of neighborhoods that lack social capital and residents that face a continued decline of the physical infrastructure coupled with social chaos as well as neighborhoods that are thriving or undergoing urban renewal (Reference Removed Blind Review).

Research that focuses on adults who use alcohol and other drugs is less prevalent than studies involving adolescents and young adults. Findings from studies among adults do show that men who have not developed unhealthy drinking behaviors by the time they transition from young adults to adulthood are unlikely to acquire them as they grow older (Fillmore, 1988). On the other hand, men who do acquire a drinking problem before reaching adulthood are likely to develop chronic drinking problems.

In addition to filling the void in knowledge about adult African American men and alcohol consumption, in this paper we also provide a theoretical contribution. Past research has identified several key socio-demographic characteristics (e.g., age, educational attainment, and relationship status) (Hedden, et al., 2009; Kuramoto, et al., 2011; Mares & Rosenheck, 2010) and social-psychological characteristics (e.g., sensation seeking, depression, and self-esteem)(Byers, Raven, Hill, & Robyak, 1990; Hedden, et al., 2009; Pekala, Kumar, Maurer, Elliott-Carter, & Moon, 2009; Quinn & Fromme, 2010) that are linked to alcohol use patterns. In addition, having health insurance has been shown to impact seeking treatment for alcohol (Kuramoto, et al., 2011). Among the more distal factors explored, childhood abuse experiences have been associated with a wide variety of subsequent problems and dysfunctions in people’s lives, including unhealthy alcohol use patterns (Briere, 1992; Herman, Susser, Struening, & Link, 1997; Klein, Elifson, & Sterk, 2007; Odonne-Paolucci, Genuis, & Violato, 2001). Studies on the role of childhood abuse experiences on the lives of adults are more common among women than men (Genuis, Thomlison, & Bagley, 1991). However, in one sample of men that was primarily African American, having experienced unwanted sexual activity during childhood was associated with a 23% increase in the odds of reporting problems with alcohol (DiIorio, Hartwell, & Hansen, 2002).

Building on Bronfenbrenner’s ecological model, the perspective of this study considers individual, network, and social and contextual factors as perceived by the individual (Bronfenbrenner, 1979). In addition to the individual level factors described above, among adults it has been shown that earlier increased participation rates in social activities such as parties, picnics, or playing card games demonstrated associations with alcohol consumption while earlier alcohol consumption related to later peer approval for alcohol use, suggesting a reciprocal relationship between social influence and drinking (Moos, Brennan, Schutte, & Moos, 2010). An inverse association between having a higher proportion of non-substance users in one’s social network and alcohol consumption, including the amount, frequency and problems related to drinking, has also been reported (Lau-Barraco & Collins, 2010). Being a user of illicit drugs is associated with alcohol use, at times to augment the effects of the illicit drug or as a substitute when drugs or the means to purchase these are unavailable (Gossop, Manning, & Ridge, 2006; Sterk, 1999). Exposure to community violence has been associated with greater use of alcohol and other drugs (Fowler, Ahmed, Tompsett, Jozefowicz-Simbeni, & Toro, 2008). Finally, fear of violent crime has been linked to alcohol availability, but not as predictive of substance use (Stockdale et al., 2007).

The multi-domain approach of this study, including socio-demographic, social-psychological, social network, situational, and contextual factors elucidates diverse influences on African American males’ drinking patterns, specifically regular drinking and getting drunk as early indicators of problematic alcohol consumption. It also contributes to the development and implementation of tailored prevention and interventions efforts (Rakowski, 1999) that consider the heterogeneity of African American men.

METHODS

One of the main aims of Be Healthy, an ongoing prospective longitudinal study, is to gain an understanding of risk behaviors such as the use of alcohol and other drugs through accounting for more than individual risk factors. This study includes baseline data collected between January 2010 and January 2011 in Atlanta, Georgia. The 484 participants were recruited through active (e.g., outreach) or passive (e.g., flyers, referral) efforts in Atlanta, Georgia. Eligibility criteria were self-identifying as African American or black, being 18 years or older, and having lived in the study area for 12 months prior to the interview. Following consent procedures, questionnaire-based, computer-assisted interviewers, taking on average 70 minutes, were conducted at a community-based research site. Participants were compensated $30 for their time. The Emory Institutional Review Board approved the study protocol.

Measures

Dependent variables

Respondents were asked how many days they had consumed alcohol in the last 90 days. The variable was dichotomized into regular drinking: (0) no - drinking less than 60 out of 90 days and (1) yes -drinking 60 days or more days during the past 90 days. Ever drunk was measured with the question “Have you ever had so much to drink that you became drunk?”

Independent variables

Socio-demographic characteristics included age at baseline (years), education (years), relationship status (single including divorced, separated or widowed; married or living as married; and having a partner and not living with them), employment status (full or part-time legal employment versus all others), home-ownership (yes/no), and homelessness experience (ever/never). Health insurance was measured as having had health insurance or Medicaid coverage for all 12 months of the past year (yes/no). Physical health and mental health involved separate self-rankings of the respective status, using a 5-point Likert scale that ranged from “Poor” to “Excellent”.

Social-psychological characteristics included self-esteem, depression, and sensation seeking. Self-esteem was based on 10 items from the Rosenberg Self Esteem scale (Rosenberg, 1965), measuring general feelings of self worth. Scores ranged from 1 to 40, with higher scores representing higher levels of reported self-esteem (Cronbach’s alpha = 0.78). Depression was measured using the 20-item Centers for Epidemiologic Depression Scale (CES-D) (Radloff, 1977) that assesses symptoms of depression (Cronbach’s alpha=.86). Sensation seeking was measured with nine categorical items from the Zuckerman Sensation Seeking scale (Zuckerman, Kolin, Price, & Zoob, 1964) (False (0), True (1)). Overall scores ranged from 0 to 9, and higher scores represented higher levels of sensation seeking (Cronbach’s alpha = .71).

Social network characteristics included friends and people who lived with the respondents. Characteristics of friends were assessed with a series of questions that began with the stem “Of the people you regularly hung out with in the past year, how many of them …” with the following response options: were employed, used drugs, and were involved in illegal activities other than drug use. Characteristics of people lived with began with the stem “Of the people you regularly lived with this past year, how many of them [were] . . .” with the same response options. The questions were recoded such that (0) corresponded to no friends or people lived with that were employed, used drugs, or were involved in illegal activities other than drug use and (1) corresponded to any friends or people lived with that were employed, used drugs, or were involved in illegal activities other than drug use.

Situational characteristics included the use of cocaine (crack or powder form) in the last 90 days (yes/no) and childhood abuse experiences, including emotional, physical and sexual abuse. Questions addressing abuse during childhood were asked after the prompt: “These next questions deal with your childhood before you turned 18. Please tell me for each statement if it was never true, rarely true, sometimes true, often true, or very often true.” Child physical abuse was measured with the statement “Someone in my family hit me so hard that it left me with bruises or marks.” Child sexual abuse was measured with the question “Someone tried to touch me in a sexual way or tried to make me touch them.” Child emotional abuse was measured with the question “I believe that I was emotionally abused.”

Contextual factors included variables assessing perceptions of crime and observed crime. Perceived fear was assessed using 16 items from a modified version of the Perceived Fear of Victimization Scale (Warr & Stafford, 1983) with responses that ranged from (0) not at all afraid to (4) very afraid, differentiating between perceived fear of non-violent crimes (e.g., burglary while away, burglary while home, car theft, and fraud) and perceived fear of violent crimes (robbery, being raped, and murder). The items for each scale were summed in a measure for which higher scores indicated increased fear. The alphas for perceived fear of non-violent crimes and that of violent crimes were .79 and .93, respectively. Recent neighborhood criminal activity was measured by asking if the respondent had heard about any of the following occurring in the neighborhood during the past 6 months: the use of a weapon, sexual assault or rape, and a robbery or mugging. Responses ranged from (0) Never to (3) Often (Cronbach’s alpha=.72). Observed violence was measured with 7 questions from the Community Experiences Questionnaire (Schwartz & Proctor, 2000). The items addressed having observed in the neighborhood during the past year events such as “somebody got threatened,” “somebody got hit, punched, or slapped,” and “somebody carried a gun, or other weapon (besides police, military, and security guards)”. Response options ranged from (1) Seldom to (3) Often. Higher scores corresponded to greater perceived violence (Cronbach’s alpha=.86).

Analysis

Descriptive statistics were computed using SAS 9.2. All inferential statistics were computed using survey statistical procedures that account for the correlation of individuals within the same census block group. Bivariate associations between the outcome variables and continuous independent variables were tested using biserial Pearson correlations. Bivariate associations between the outcome variables and categorical independent variables were tested using chi-square. Given the exploratory nature of the analysis and the large number of independent variables tested, potential predictors needed to be associated with the outcome at the level of p < .05 for inclusion in multivariate logistic regression analysis in order to maximize power. The full models were then reviewed and predictors that demonstrated statistical significance at the level of p < .10 were retained for reduced models. There was minimal missing (< 5%) and case wise deletion was applied during multivariate analyses.

RESULTS

The descriptive statistics for the study sample are shown in Table 1. The mean age of the men in this sample was 39.6 years (SD=13.7, Range: 18–71) and their mean educational attainment was 12.4 (SD=2.3). A little over half (51%) of the men reported being single, 22% reported being employed, and 30% reported having health insurance, including Medicaid, for all of the last 12 months. Of those who were insured, 22% reported private health insurance. Seventy percent of the men reported “hanging out” with at least one person whom they knew to be a drug user and 35% reported living with at least one drug user. Almost two-fifths (38%) of the men reported cocaine use in the past 90 days, thereby reflecting the prevalence of alcohol as well as cocaine availability in the study area. Childhood abuse experiences were not uncommon, with almost one-third of the men having reported childhood physical abuse, 13% reported sexual abuse, and 16% reported emotional abuse. In terms of the outcome variables, 31% reported regular alcohol consumption, defined as having drunk at least 60 out of the last 90 days, and 77% reported ever having been drunk.

TABLE 1.

Descriptive Statistics

Variable	Alpha	Number or Mean	Percentage or Standard Deviation
Drinking behavior
Alcohol use in past 90 days
Regular (60 days)		150	31%
Less than 60 days		334	69%
Became drunk
Yes		371	77%
No		113	23%
Sociodemographic
Age (y)		39.63	13.66
Education level		12.44	2.32
Work status
Employed		107	22%
Unemployed		377	88%
Living situationa
Owns home		36	8%
Does not own home		426	92%
Homeless
Ever		212	44%
Never		272	56%
Relationship statusa
Singleb		245	51%
Partner, not living with		133	28%
Married or living as married		102	21%
Had health insurance in past 12 monthsa:
Yes		143	30%
No		340	70%
Physical health		2.14	1.04
Mental health		2.59	1.08
Social-psychological
Self-esteem	0.78	38.85	5.19
Depression	0.86	28.34	10.05
Sensation seeking	0.71	3.85	2.38
Social networks
Characteristics of people hanging out witha:
Are legally employed		337	70%
Are drug users		333	70%
Particpate in illegal activities		161	34%
Characteristics of people living witha:
Are legally employed		216	45%
Are drug users		167	35%
Participate in illegal activities		40	8%
Situational
Crack/cocaine 90 days		185	38%
Experienced childhood physical abuse		147	30%
Experienced childhood sexual abusea		61	13%
Experienced childhood emotional abusea		75	16%
Contextual factors
Perceived fear of non-violent crime	0.78	4.58	3.99
Perceived fear of violent crime	0.93	21.55	13.57
Recent neighborhood criminal activity	0.72	4.38	2.54
Witnessed violence	0.86	12.54	5.55

^aThese variables do not total to N = 484 due to missing data.

^bIncludes separated, widowed, and divorced.

Bivariate associations for the outcome and predictor variables are reported in Table 2. For the socio-demographic predictors, older age, ever having been homeless, lack of health insurance, and lower ratings of physical and mental health were found to be significantly associated with regular drinking as well as with ever having been drunk. Among the psychosocial variables, depression was significantly related to regular drinking as well as ever having been drunk. Self-esteem was significantly negatively related to regular drinking, but only marginally (p=.06) related to ever having been drunk. Sensation seeking had a significant positive relationship with regular drinking but not to having ever been drunk. In terms of the social network variables, the percent of regular drinkers was also significantly higher for those who reported “hanging out” with individuals who were unemployed compared to those who were legally employed. The percent of regular drinkers was significantly higher for those who reported living with a drug user compared to those who did not and for those who reported living with someone who engaged in illegal activities. The percent of regular drinkers was significantly lower for those who reported living with someone who was legally employed compared to those not living with someone employed. In terms of having been drunk, a significant positive association was noted for hanging out with drug users and those who engage in illegal activities. A similar significant relationship was noted for living with someone not legally employed. For the situational variables, significant positive associations were noted for crack/cocaine and both regular drinking and having been drunk. The three child abuse variables (physical, sexual, emotional) were all significantly associated with having been drunk but not with regular drinking.

TABLE 2.

Bivariate Associations

Variable	Regular Drinking		Been Drunk
	No	Yes	No	Yes
Sociodemographic
Age (y)		0.29***		0.28***
Education level		−0.01		−0.00
Work status
Employed	72%	28%	29%	71%
Unemployed	68%	32%	22%	78%
Living situation
Owns home	69%	31%	28%	72%
Does not own home	69%	31%	23%	77%
Homeless
Ever	60%	40%***	16%	84%***
Never	76%	24%	29%	71%
Relationship status
Single	66%	34%	21%	79%
Partner, not living with	75%	25%	29%	71%
Partner, living with	70%	30%	23%	78%
Had health insurance for past 12 months
Yes	65%	35%**	31%	69%*
No	78%	22%	20%	80%
Physical health		−0.10*		−0.19†
Mental health		−0.12***		−0.10*
Social-psychological
Self-esteem		−0.19***		−0.10†
Depression		0.24***		0.16***
Sensation seeking		0.16***		−0.05
Social networks
Characteristics of people hanging out with:
Are legally employed	72%	28%*	23%	77%
Are not legally employed	62%	38%	23%	77%
Are drug users	66%	34%†	20%	80%**
Are not drug users	74%	26%	31%	69%
Particpate in illegal activties	63%	37%†	17%	83%**
Do not particpate in illegal activities	72%	28%	27%	74%
Characteristics of people living with:
Are legally employed	75%	25%**	30%	70%**
Are not legally employed	64%	36%	18%	82%
Are drug users	59%	41%***	21%	79%
Are not drug users	75%	25%	25%	75%
Particpate in illegal activities	43%	58%**	15%	85%
Do not particpate in illegal activities	72%	28%	24%	76%
Situational
Crack/cocaine 90 days
Yes	49%	51%***	14%	87%***
No	82%	18%	29%	71%
Experienced childhood physical abuse
No	69%	32%	27%	73%**
Yes	70%	30%	15%	85%
Experienced childhood sexual abuse
No	70%	30%	25%	75%**
Yes	64%	36%	12%	89%
Experienced childhood emotional abuse
No	70%	30%	25%	75%*
Yes	65%	35%	15%	85%
Contextual factors
Perceived fear of non-violent crime		0.06		−0.04
Perceived fear of violent crime		0.08		−0.05
Recent neighborhood criminal activity		0.09*		0.07
Witnessed violence		0.13**		0.06

Note. Values not presented as percentages are Pearson correlation coefficients.

^†p < .10,

^*p <. 05,

^**p < .01,

^***p < .001.

The results for the full and reduced logistic regression models are reported in Table 3. For the regular drinking outcome, the full model with all 16 predictor variables resulted in an overall statistically significant model (Wald χ2(16)=160.62, p<.0001, Cox and Snell R-Square = .21). In this model, having had health insurance in the past 12 months significantly reduced the odds of regular drinking. However, being older, sensation seeking, and having used cocaine significantly increased the odds of regular drinking. Living with people who used drugs or with people who participated in illegal activities were marginally significant (p <.10). When variables meeting the criteria of p < .10 were entered into a reduced model (Table 3), the overall model remained statistically significant (Wald χ2(6)=139.04, p<.0001, Cox and Snell R-Square = .20). In this reduced model, having had health insurance reduced the risk of regular drinking. Being older, having used cocaine, and higher levels of sensation seeking were associated with increased odds of regular drinking. Living with someone who participated in illegal activities remained marginally significant (p<.10).

Table 3.

Full and reduced logistic regression models for regular drinking and been drunk

Variable	Full Models OR (95% CI)		Reduced Models OR (95% CI)
	Regular Drinking	Been Drunk	Regular Drinking	Been Drunk
Sociodemographic
Age (y)	1.05*** (1.03, 1.07)	1.05*** (1.03, 1.08)	1.05*** (1.03, 1.07)	1.06*** (1.04, 1.08)
Never homeless	1.10 (.67, 1.81)	0.94 (.57, 1.54)	–	–
Had health insurance for past 12 months?	0.53** (.35, .80)	0.47** (.28, .80)	0.50***(.33, .75)	0.46** (.28, .76)
Physical health	1.01 (.77, 1.33)	0.81† (.64, 1.02)	–	0.85 (.69, 1.06)
Mental health	1.06 (.84, 1.34)	1.16 (.90, 1.48)	–	–
Social-psychological
Self-esteem	1.00 (.94, 1.07)		–	–
Depression	1.02 (.99, 1.05)	1.02 (.99, 1.05)	–	–
Sensation seeking	1.15* (1.02, 1.28)	–	1.18*** (1.08, 1.30)	–
Social networks
Hangs out with those who are legally employed	0.88 (.51, 1.51)	–	–	–
Hangs out with drug users	–	1.89** (1.27, 2.81)	–	1.84** (1.25, 2.70)
Hangs out with those who participate in illegal activities	–	1.48† (.94, 2.33)	–	1.52† (.96, 2.40)
Lives with those who are legally employed	0.82 (.47, 1.41)	0.77 (.48, 1.22)	–
Lives with drug users	1.75† (.94, 3.26)	–	1.64 (.91, 2.96)	–
Lives with those who participate in illegal actitivies	2.35† (.86, 6.43)	–	2.26† (.94, 5.45)
Situational
Crack/cocaine 90 days	2.10** (1.21, 3.66)	0.94 (.51, 1.73)	2.42*** (1.45, 4.04)	–
Experienced childhood physical abuse	–	1.87† (.99, 3.54)	–	2.03* (1.12, 3.71)
Experienced childhood sexual abuse	–	1.94† (.92, 4.08)	–	2.07* (1.02, 4.18)
Experienced childhood emotional abuse	–	1.29 (.58, 2.85)
Contextual Factors
Perceived fear of non-violent crime	1.04 (.97, 1.11)	–		–
Recent neighborhood criminal activity	0.98 (.89, 1.09)	–
Witnessed violence	1.03 (.97, 1.09)	–		–

^aThe variable was coded as follows: 1 = never homeless and 0 = ever homeless.

^†p < .10,

^*p < .05,

^**p < .01,

^***p < .00.

In terms of ever having been drunk, the full model with all 13 predictor variables resulted in an overall statistically significant model (Wald χ2(13)=164.95, p<.0001, Cox and Snell R-Square = .16). In this model, being older and hanging out with drug users significantly increased the odds of having been drunk. However, having insurance significantly decreased the odds of ever having been drunk. Several variables were found to be marginally significant (p <.10) including hanging out with those involved in illegal activities, childhood physical and sexual abuse and the self-rating rating of physical health. When just these variables were entered into a reduced model (Table 3), the overall model remained statistically significant (Wald χ2(7)=57.97, p<.0001, Cox and Snell R-Square = .15). In this reduced model, having health insurance reduced the risk of ever having been drunk while being older, hanging out with drug users, and reporting childhood physical and sexual abuse were associated with increased odds of ever having been drunk.

DISCUSSION AND CONCLUSION

The aim of this paper was to explore associations of regular alcohol consumption and ever having been drunk among African American men within multiple domains, including individual socio-demographic and social-psychological characteristics, social network features, and contextual and situational factors. One of the most surprising findings was the role of having health insurance in the past year. Being covered was associated with a 50% reduction in the odds of regular drinking and a 54% reduction in ever having been drunk. Having some form of health insurance appears to serve as a key correlate of lifestyle factors such as regular drinking and getting drunk. It is of note that a large majority of those who were covered by insurance in this sample were covered by public sources. An analysis of a nationally representative sample found that among people who were in the top half of spending based on total expenditures, those without public or private insurance reported that alcohol comprised a larger percentage of their budget than those in the top half of spenders with insurance (Levy & DeLeire, 2008). This is especially important to note as more and more insurance policies provide incentives for people that live healthy and engage in preventive health behaviors (e.g., blood pressure and cholesterol screening, working out, and healthy nutrition) and implement penalties for those who engage in unhealthy behaviors. For example, many insurance policies ask higher premiums for cigarette smokers, with some excluding cigarette smokers. As many African American men may already fall into this vulnerable group, these requirement could create an another path to health disparities in this population. In addition, the association of being insured with a reduction in the odds of excessive drinking behaviors among African American men is significant given the continued policy challenge of access to care, particularly for substance abuse or dependence (McCarty, McConnell, & Schmidt, 2010).

There is also evidence that African Americans are less likely than whites to seek treatment services for unhealthy alcohol use (Wells, Klap, Koike, & Sherbourne, 2001). As health insurance covers limited treatment for alcohol and other drug use, health disparities are likely to become more visible. Other factors that could impede treatment seeking among the men in this study include a combination of cultural beliefs, such as distrust in the health care system (Ornelas et al., 2009), but also policies that limit support and access (Satcher & Higginbotham, 2008), highlighting the importance of structural factors such as insurance coverage.

Older age was associated with an increase in the odds of the drinking behaviors in this study. This is in contrast with the results of national surveys that indicate drinking declines with age (Kerr, Greenfield, Bond, Ye, & Rehm, 2009). However, it has been shown that prevalence of alcohol abuse and dependence peaks between the ages of 30 and 44 and does not drop substantially until closer to age 60 (Kessler et al., 2005). Sixty-five percent of the sample was between the ages of 30 and 59, possibly accounting for the lack of a decreasing effect. It is also possibly that alcohol disorders function differently in this population with respect to age. For example, an analysis of multiple years of epidemiological surveys of alcohol use revealed declining percentages among white men but the same pattern did not emerge for black men (Grucza, Bucholz, Rice, & Bierut, 2008).

Findings from our study also revealed associations with drug use, both with the individual’s usage and usage within a person’s social network. For example, similar to previous research that has demonstrated the protective effects of non-users in one’s social network with respect to excessive drinking and associated problems (Lau-Barraco & Collins, 2010), our study found that hanging around with drug users increase the odds of having been drunk but not with regular drinking. This may be because associating with people who do drugs may create an environment where there is acceptance for drinking so much that one becomes drunk. However, personal cocaine use was associated with regular drinking. This may be because of the confluence of alcohol and cocaine in the study area (DePadilla & Wolfe, 2010; Rothenberg et al., 2000).

The only socio-psychological variable to demonstrate a significant association was sensation seeking, and this was only evident for regular drinking. Although previous evidence has indicated a link between alcohol abuse and lifetime depression (McCarty, et al., 2010), the CES-D is used to assesses symptoms of psychological distress in the last seven days and may not have captured the length and degree of depression that would be predictive of alcohol use or abuse. Previous work that has examined self-esteem in early life as a predictor of substance use in later life concluded that the role of self esteem is reduced when the psychosocial context in which self esteem was developed is considered, for example, childhood abuse experiences (Boden, Fergusson, & Horwood, 2008) In this study, both child sexual abuse and child physical abuse demonstrated significant associations with having been drunk, suggesting that these situational factors outweigh or perhaps pre-date psychosocial characteristics such as self-esteem.

It is of note that perceived fear, recent neighborhood criminal activity, and observed violence were not associated with either regular drinking or having been drunk. This is consistent with previous research examining the association between crime and violence and substance use disorders that did not find a significant association between violence and substance use in multivariate analysis (Stockdale, et al., 2007). It may be that personal experiences of violence such as physical and sexual abuse are more relevant to drinking habits than a person’s perceptions of the environment. For example, previous work has identified stressful life events that were defined by personal experience (e. g. illness, assault, and robbery) as being associated with drug use (Boardman, Finch, Ellison, Williams, & Jackson, 2001).

There are a number of limitations for this study. The data are cross-sectional and causal inferences cannot be made. Additionally, the self-reported responses of the participants may be subject to bias caused by social desirability. Moreover, the sample was purposive and limited to one metropolitan area and any generalizations should be made with caution.

The findings of our inquiry support associations between four of the five domains we explored and regular drinking and getting drunk among African American men. Factors from the socio-demographic (age, health insurance) were associated with increases in the probability of regular drinking and having been drunk, suggesting that these kinds of individual variables function similarly across different operationalizations of excessive drinking. However, the social psychological factor of sensation seeking was only associated with regular drinking while the social network factor of associating with drug users was only associated with having been drunk. It may be important to assess how much alcohol is being consumed regular in order to understand why these factors operate differently across the two outcomes. Additionally, current cocaine use was associated with regular drinking while child sexual abuse was associated with having been drunk and it may be that incorporating the timing and frequency of getting drunk is important to understanding how it is impacted by more recent events. Finally, there were no associations between the contextual neighborhood factors and drinking outcomes in the study and it may be important to consider crimes that have personally impacted people as opposed to how they perceive crime to be happening around them when exploring alcohol use and abuse. Further research on the multiple domains that impact regular drinking and getting drunk is needed. The findings also show the need for continued prevention and intervention efforts that consider factors beyond the individual, including efforts that assist racial minority communities in preventing or reducing the use of alcohol and other drugs as well as the associated negative social and health consequences. For example, future research must address the impact of neighborhood characteristics more extensively, including personal experiences of violence. Moreover, the findings show the importance of recognizing the heterogeneity among African American men residing in stable, decaying, and emerging urban neighborhoods.