Sociation Today

Beliefs about Drinking Problem Causation

Susan Bullers

University of North Carolina Wilmington

and

Carol A. Prescott

University of Southern California

Background

   Individuals' perceptions about the causes of behavioral and health problems can have an impact on whether or not they participate in treatment programs as well as on their beliefs about the efficacy of various treatment methods (McKirnan 1984; Harris and Fennell 1988). In a study of college students in Chile, Didier (2001) found that respondents believed the most effective treatments for drinking problems would be those corresponding most directly to their causes. However, Harris and Fennell (1988) found that problem drinkers who attributed their heavy drinking to workplace stress were less likely to participate in an Employee Assistance Program than those who had other attributions for their drinking.

    The Health Belief Model, originally developed by Irwin Rosenstock, uses psycho-social attitudes about one's health to understand and predict health behavior and treatment compliance. This model was further refined (Rosenstock et al. 1988) to include a component of self-efficacy; individuals who believe that they have some control over the course or outcome of a health problem will be more likely to seek treatment. McKirnan (1984) agrees that beliefs about the causes of any abnormal behavior are important to the course and treatment of the problem and specifically argues that help-seeking behavior requires at least some degree of belief in personal efficacy.

    Cultural beliefs about drinking problem causation have changed over time; away from moral weakness explanations toward social influence or disease model beliefs. For example, in 1977 Hoy found that causal beliefs were consistent with a personality disorder concept, yet in 1984, Furnham found social influences such as friends, family and media were the dominant causal beliefs. Sigelman et al. (1992) studied a sample of students (6th grade through college) and found that disease model explanations constituted the most common causal belief explanations for problem drinking but that the majority of respondents also believed that curing drinking problems was a personal responsibility. In 1978, Fisher discussed the limitations of the emerging disease model trends for alcohol and in 1994, Cunningham et al. suggested that there was still conflict about these causal beliefs. Schomeru et al. (2011) recently reviewed population studies regarding the stigma of alcohol dependence and found that alcoholics were significantly more likely to be held responsible for their conditions than were those with other mental disorders, such as depression or schizophrenia.

    One distinction made in causal attribution research is whether causes are perceived to be "internal" or "external."   Internal causation reflects a belief in which individuals have some degree of control over their drinking and are at least partially responsible for their own drinking problems. Internal attributions include factors such as personal decisions, morality, judgment, and choice of friends. In contrast, external causal beliefs attribute drinking problems to factors beyond the individual's control. External attributions include factors such as exposure to stressors, family upbringing, genetic or biological predisposition, and bad luck.

Demographic correlates

Several studies have looked for demographic correlates of causal beliefs. Although a long history of research has found correlations between generalized internal/external attributional styles and gender, age, cohort, income, health and education (Pearlin and Schooler 1978; Kohn, 1989; Mirowsky and Ross1989; Bullers and Prescott 2001), alcohol-specific attributions may not be associated with generalized attributional styles. In preliminary work using a subset of the current dataset (Bullers and Prescott 2000) failed to find a significant association between a generalized perceived control construct and any of several internal and external problem drinking causal beliefs in a sample of college students.

Some studies have found associations between demographic characteristics and drinking-specific causation beliefs. McKirnan (1984) found that higher SES individuals tend to believe in more internal (personal responsibility) causes whereas lower SES individuals were more likely to attribute alcohol abuse to external causes (environment, stressful situations). Genetic and/or biological factors were not included in this study. Kauffman et al. (1997) did include genetic and disease model variables and found that endorsement of biological attributions was higher among individuals with higher levels of employment and income.

    Findings regarding gender differences in drinking causation beliefs are contradictory; in 1988 Harris and Fennell found that females were more likely than males to attribute alcohol problems to moral weakness and emotional and interpersonal problems but in 1997, Kauffman et al. found that females were more likely than males to make external (in this case biological or environmental) attributions. More recently, Lambert et al. (2006) found males more likely than females to make internal (character) attributions.

    In addition to cultural shifts in beliefs about drinking problems, cross-sectional studies suggest there are age effects. Lambert et al. (2006) found that disease model attributions were higher and "character" attributions lower with older age among a sample of college students. Sigelman et al. (1992) also found that attributions to "moral weakness" were lower among older respondents but found no age differences for any of the other causes, including disease, bad environment, normative motives, or personal responsibility. However, in separate analyses about curing drinking problems they found that beliefs about personal responsibility were higher with older age.

    Few studies have looked at the role of ethnicity in beliefs about drinking problem causation. Aggregate findings have shown that Native Americans drink more than Whites, who in turn, drink more than African Americans. Although there is a vast literature on ethnicity and alcohol use, there are also several critiques of this literature arguing that the relationship between drinking and ethnicity is fraught with methodological problems including confounding socioeconomic factors and lack of cultural homogeneity among common ethnic groupings (Caetano et al. 1998; Fisher 1987). Of studies that have looked at ethnicity and attributions regarding problem drinking, Lambert et al. (2006) found that whites were more likely to attribute substance abuse problems to internal (character) causes, whereas non-whites were more likely to make disease model attributions. These results may reflect the disproportionate representation of non-whites in lower SES ranges; findings regarding the association between generalized external attributions and lower SES are fairly consistent (Pearlin and Schooler 1978; Kohn 1989; Mirowsky and Ross 1989; Bullers and Prescott 2001).

    Religion may also affect drinking problem causation beliefs. Some religious denominations explicitly prohibit or object to alcohol consumption. The degree of identification with one of these denominations could have a strong effect on one's causal beliefs. Viewing alcohol consumption as a "sin" constitutes an internal attribution, suggesting that an individual has control over the course and outcome of the problem. Religions in which doctrine explicitly prohibits alcohol use are referred to as "objector religions." The association between membership in objector Religions and lower drinking rates may also account for some ethnicity effects: African Americans have the highest rates of membership in objector religions in the US (Kosmin and Keysar 2009).

Relation of drinking experience to causal beliefs

    Personal drinking history and exposure to others with drinking problems may also affect beliefs about the causes of alcohol problems. Research has shown that heavy drinkers "normalize" their drinking problem beliefs to exclude themselves from problem drinking categories (Furnham and Lowick 1984; Oei et al. 1990; Didier 2001; Wild 2002). Because heavy drinkers define problem drinking at much higher levels of consumption than do light or non-drinkers, the relationship between heavy drinking and attribution may be more complex than a simple linear association. McKirnan (1984) found an interaction effect in which heavy drinking was associated with more internal attributions among higher SES individuals but was associated with more external attributions among lower SES individuals. In essence, the internal/external attribution patterns of SES were accentuated as drinking increased.

    In summary, previous research has found that individuals' beliefs about the causes of drinking problems can affect treatment-seeking and beliefs about the efficacy of treatments. Some SES, gender, ethnicity, and age correlates of alcohol abuse causation beliefs have been observed in the literature, but the gender findings are inconsistent and most studies have not included genetic or disease model causal options. In addition, few of these studies considered the respondents' personal experiences with heavy drinking. We hypothesized that one's own drinking and one's exposure to others' problem drinking could influence causation beliefs.

    Using data from a sample of 170 University students (over-sampled for ethnic and age diversity) we examine the strength of attributions for ten possible causes of drinking problems: personal decision, drinking friends, bad judgment, weak morals, family upbringing, stress/life circumstances, genetics, physiology, bad luck and low intelligence. A factor analysis of cause beliefs is conducted to test if these causes represent distinct causal dimensions that can be further analyzed as composite variables. Finally, using OLS multiple regression we explore the independent effects of demographic and drinking experience variables on each causal factor.

Methods

Sample

This analysis consists of data from two different survey samples. The first data set was originally collected to explore the issues concerning perceived control and drinking. The second survey was intended to explore ethnic differences in perceptions of drinking problem causation. Because these data sets both included information concerning perceptions of drinking problem causation, they were combined for this study. This pooling of data sets created a sample with an adequate numbers of cases in all three ethnic groups of interest. However, because these two data sets had inconsistent measures of some variables, these variables had to be re-constructed in a manner that would allow for consistent and valid measures across both data sets.

The first data set was collected in 2000 and consisted of 91 respondents from a large Southern university in a metropolitan area. Respondents filled out a seven-page questionnaire in non-randomly selected university classes. Purposive class selection included day and evening extension programs to increase the age diversity. A purposive range of subject areas was sampled to avoid over-reliance on the social sciences. The sample had mean age of 36.3 (SD=18.4) and was 63% female. The second data set was collected in 2003 and included 79 students at a traditionally Native American University located in a smaller Southern city. These respondents filled out questionnaires in non-randomly selected sociology daytime classes. The age of this sample was M=27.3 (SD=10.0) and the sample was 69% female. Six respondents did not answer the religion items and five respondents did not answer the race item reducing the sample size for this study to 159. An additional seven respondents answered the race item with responses other than white, Native American or African American. The small number in these ethnic categories did not allow for meaningful analysis so these cases were also dropped resulting in a final sample size of 152. Missing values in the variables used to construct the causation factors further reduce sample sizes in some analyses and will be noted where appropriate.

Variables

    Age is measured in years. Since all respondents were undergraduate students, education level did not offer enough variability for valid inclusion in the analysis. Household income was also a problematic indicator of socioeconomic status (SES) for this population because about half of the respondents are young college students living independently whose household income is not likely to represent their social class. Although income and education of family of origin may be better indicators of social class for the younger students in this sample, we expressly sampled a substantial number of non-traditional students. Income or education of family of origin for the older students is equally unlikely to represent current SES for these respondents so was not included in analyses. In addition, the sample is somewhat homogenous regarding socioeconomic status, simply by virtue of being college students.

    Ethnicity for this analysis was categorized as: White, African American, and Native American. Other ethnic categories included too few respondents (6) for meaningful analysis and were removed from the sample to avoid un-interpretable results. Ethnic identification in this community is complicated because the Native American community has struggled for full official tribal recognition for decades. Although at the time of data collection for this study full recognition had not been passed, the latest effort in 2009 included a bill for recognition that was approved by the United States Senate Committee on Indian Affairs. In addition to official tribal recognition, a large number of residents have both Native American and African American heritage and their primary identity may be either African American or Native American. Although ethnic categorization is problematic under normal circumstances, it is especially difficult here, where ethnic self-identity may differ from socially and legally determined ethnicity. We use a self-defined measure of ethnicity here.

    The religion items differed for the two samples; the urban campus was asked only for broad religious category membership with a write-in denomination option. Many of the write-in denomination responses did not actually represent recognized religious denominations; many used broad categories such as "Christian," resulting in an unusable denomination variable for the urban campus. The item was re-worded into a separate variable for the rural campus sample; asking whether their religious beliefs prohibited alcohol consumption. One religion variable was common to both samples and included the following categories; Protestant, Catholic, Jewish, other, and none. Sixty-five percent of the sample were Protestant, 12.2% were Catholic (not an "objector" religion), and 21.3% were "other"; which likely included those who did not identify with any religion. Although this measure is not ideal, "objector" Christian denominations do constitute the largest of any religious affiliation category in this "bible belt" state (Kosmin and Keysar, 2009).

    The "causes of drinking problems" questions asked respondents: "How much influence do you think each of the following has in causing drinking problems?" Response options were: 0="no influence", 1="weak influence", 2="strong influence" and 3="complete influence". The causes included: family upbringing, personal physiology/chemistry, friends with drinking problems, bad judgment, low intelligence, genetic/inherited causes, personal decisions, weak moral character, bad luck, and life circumstances/stress. We also asked an item about advertising and cultural messages, but this was inadvertently formatted differently on the surveys administered to the two different samples. In preliminary analyses it was not found to have equivalent measurement properties in the two groups and so was eliminated from further analysis.

    For own drinking we used a variable constructed from items that asked different questions in each sample. The older, more urban campus sample was asked, "During the time in your life when you used alcohol the most, how much would you usually drink?" Response options were: never used alcohol, 1-2 drinks, 3-4 drinks, 5-6 drinks 7-9 drinks, and 10 or more drinks. This scale was re-coded to mid-points resulting in scores of 0, 1.2, 3.5, 5.6, 8, and 10. For the younger rural sample, respondents were asked, "How many drinks do you typically have on an occasion?" Although this inconsistency is problematic, traditional college age is typically the life stage at which people drink the most. Responses to this item for both younger and older respondents are likely to reflect their heaviest drinking experiences so far. As with most alcohol measures, this variable produced a non-normal distribution with over 20% reporting "never used alcohol." We re-coded the consumption data to reflect non-drinkers, light/moderate drinkers and heavy drinkers. Heavy drinking was defined as four or more drinks per drinking occasion. The recoded drinking scale for older urban sample placed those above the 3-4 drinks category into the heavy drinking category. There are several ways to measure and define heavy drinking but this cutoff point corresponds roughly to the National Institute of Alcohol Abuse and Alcoholism's (NIAAA) definition of problem drinking (NIAAA, 2009).

    Exposure to any close others' drinking problems was ascertained by asking, "Have you ever lived with anyone who had an alcohol problem?" This item will encompass the residents of any households that the respondent has lived in, including parents, siblings, other relatives, roommates, partners, spouses, or children. We were attempting to capture beliefs arising from daily interactions with and observations of the problem drinker, rather than asking about family history of alcoholism which may have occurred at a time other than when the respondent was in residence.

Analysis

    First, sample means, standard deviations and modes for individual cause beliefs and Cause Factors will be calculated and presented to examine the rank ordering of the individual causal beliefs. Internal/external distinctions will be indicated. Next, sample descriptives will be computed giving frequencies and percentages for all demographic, drinking, and drinking exposure groups.

    We hypothesized that we would find evidence for three factors, including a factor of personal characteristics identified by bad judgment, personal decisions, weak moral character, and to a lesser extent, low intelligence; and a second factor of external influences, including family upbringing, life circumstances/stress, friends with drinking problems, and bad luck. We hypothesized that the two biological causes, genetics and physiology, would form a separate factor. We tested this structure using confirmatory factor analysis (CFA) and indexed fit using RMSEA, with "good fit" represented by e<.08 (e.g., Kline, 2011). We also compare the fit of the CFA models to the best possible fit from 3-factor and 4-factor exploratory models.

    Finally, using OLS regression analysis, each of the cause factors will be regressed on the demographic, drinking, and alcohol exposure variables. The factor analyses were conducted using the Mplus structural modeling program (version 5.0, Muthen & Muthen, 2005). All other analyses were conducted using SPSS version 17.

Results

Descriptive statistics (Table 1) show that 64.5% of the sample was female, 76.3% were white; 11.8% were African American, and another 11.8% were Native American. Protestants comprised 69.1% of the sample. Over-sampling of older students resulted in a relatively even distribution of students in the four age categories; 18-21, 22-25, 26-39, and 40+. Respondent drinking status was distributed as 20.4% non-drinkers, 45.4% light/moderate drinkers, and 34.2% heavy drinkers. These are relatively low rates for college students and are likely affected by the over-sampling of older and minority students. Over one-third (36.2%) of the respondents reported having lived with a problem drinker.

Table 1: Sample Descriptives by Cause Factor Means

	N	%	Personal Cause	Biological Cause	Social Cause
Age 18-21	44	28.9	1.56	1.27	1.71
Age 22-25	38	25.0	1.43	1.38	1.86
Age 26-39	32	21.1	1.70	1.75	2.04
Age 26-39	32	21.1	1.70	1.75	2.04
Age 40+	38	25.0	1.43	1.79	1.67
Gender: Female	98	64.5	1.50	1.52	1.84
Gender: Male	54	35.5	1.58	1.54	1.76
Non-Drinker	31	20.4	1.75	1.42	2.02
Light Drinker	69	45.4	1.56	1.65	1.84
Heavy Drinker	52	34.2	1.35	1.43	1.65
Lived with Problem Drinker	55	36.2	1.58	1.66	1.88
Did Not Live with Problem Drinker	97	63.8	1.49	1.45	1.78
Protestant Faith	105	69.1	1.58	1.54	1.88
Other Faith	47	30.9	1.39	1.53	1.64
White	116	76.3	1.44	1.62	1.72
African American	18	11.8	1.74	1.37	2.09
Native American	18	11.8	1.82	1.17	2.04
Total Mean			1.53	1.53	1.82
N(a)	152	100	141	143	144

(a)Missing values in the cause factor variables account for decreases in n's in cause means.

Figure 1 shows the mean endorsement scores for each of the ten causes. Personal decisions received the highest rating as a cause of drinking problems, followed by family upbringing, and stress/circumstances. There was very little support for the low intelligence cause. There was not a clear pattern of preference for either internal or external causes.

The results from our initial confirmatory factor analysis indicated that the bad luck item did not load on the hypothesized factor (with the other external causes), having a loading of <.30. An additional analysis freeing its loading indicated that it did not load better on the other factors, so we opted to eliminate it from the structure and use the remaining 9 causes. The resulting structure is shown in Figure 2.

The model obtained c2=48.3 with 24 df, and fit was good: CFI=.94, RMSEA=.077. This compared to c2=419 with 36 df for a baseline model of no association for these 9 items. Although the loadings are not all high, they are consistent with our hypothesized structure. As a further check, we ran an exploratory factor analysis using the EFA option in Mplus. This identified the 3-factor solution as the most parsimonious for these data (the first four eigenvalues were: 3.01, 1.56, 1.10 and 0.89) and the fit of a minimally structured three factor model (with 12 df) was not substantially better: c2=21.1, i.e., a difference of 27.2 on 12 df. Furthermore, examination of the loadings of the EFA solutions did not suggest the fit could be improved by moving any items to a different factor. We thus conclude that our slightly-revised hypothesized three factor solution provides a good summary of the way participants were responding to these items. We therefore created weighted factor scores for each participant by summing the items loading on the factor and dividing by the number of items (i.e., 3 for social, 2 for biology, 4 for personal) so that the scores for each would be in comparable units.

The group average scores on the factors are shown with the causes in Figure 1. Overall, there was higher endorsement of the social cause factor than for either the personal or biological cause factors; which had similar average scores. Table 1 shows the mean scores on the three cause factors for the sample divided by demographic and drinking variables. The 26-39 age group gave higher scores to both personal and social causes than did the other age groups, whereas the biological cause scores were higher across age. Both personal and biological cause attribution were negatively associated with drinking level, and Protestants scored social causes higher than did non-Protestants. Ethnicity results suggest that whites differed from both African Americans and Native Americans, who made similar attributions on all three cause factors. Preliminary analyses found minimal or no differences by gender on causal beliefs, so this variable was not included in the regression models.

The results from the regression analyses are shown in Table 2. Of the three models, the demographic and alcohol exposure variables explained the least amount of variation in the personal cause factor (Adj. R2= .084), and the most amount of variation in the social cause factor (Adj. R2= .144). All three cause factors were significantly associated with age, but in different ways: among older respondents, personal and social causation beliefs were lower but biological causation beliefs were higher. Consistent with the results in Table 1, Protestants were significantly more likely to attribute drinking problems to personal and social causes than were non-protestants, but there was no association between religion and biological causation beliefs. Heavy drinkers were less likely than non- and light/moderate drinkers to attribute drinking problems to personal or social causes, but the drinking status groups did not differ in biological cause beliefs. Those who had lived with a problem drinker were more likely than others to attribute drinking problems to personal and biological causes but did not differ for social causes.

Table 2: Regression of Cause Factor Scores on
Demograhic, Own Drinking and Exposure Variables

	Personal Cause	Biological Cause	Social Cause
Constant	1.590 (.159)	1.209 (.188)	1.861 (.135)
Age in Years	-.007 (.003)*	.013 (.004)**	-.008 (.003)**
Native American(a)	.244 (.165)	-.314 (.190)	.194 (.134)
African American	.168 (.165)	-.286 (.194)	.176 (.139)
Protestant(b)	.221 (.109)*	-041 (.128)	.279 (.092)**
Non-Drinker(c)	.072 (.120)*	-.198 (.160)	.097 (.114)
Heavy Drinker	-.264 (.120)*	-.131 (.141)	-2.32 (.101)*
Lived with Problem Drinker	.080 (.103)*	.256 (.121)*	.093 (.087)
Adjusted Rsquare	.084	.125	.44
N	141	143	144

(a)Reference category is white.
(b)Reference category is non-Protestant
(c)Reference category is light/moderate drinker.
*p<.05
**p<.01

Discussion

    Despite suggestions in the health literature that many behavioral problems have become "medicalized," attributions among this sample favored social and personal rather than biological explanations. The strength of endorsement of causal beliefs about problem alcohol use shows no clear preference for either internal or external causes.   The factor analysis supported the existence of three factors. A personal cause factor is generally consistent with internal causation beliefs; bad judgment, personal decisions, and morality are generally seen as characteristics over which individuals do have control. However, the low intelligence component is somewhat ambiguous. It can hold an internal meaning similar to bad judgment if viewed as a situational state, or it can reflect an external attribution, similar to that of genetics or physiology. In either case, support for this particular cause component was consistently low.

     A Biological cause factor comprised of genetics and physiology, most clearly reflects deterministic external attributions. It received relatively low support overall, but was more strongly endorsed by older participants. Whether this is an age or cohort effect is unclear, but it seems inconsistent with increasing societal trends toward medical explanations.

    The social cause factor presents an interesting case with respect to internal/external attributions. Although family and stress are generally not under one's control, they are not as deterministic as biological causes. Although social factors may lead to problem drinking, it is generally believed that one can overcome detrimental social influences and learn to cope with stress in a non-destructive manner. This factor then, provides an attribution that is neither distinctly internal nor distinctly external. The implications of this attribution for treatment-seeking and treatment efficacy are unclear but it does suggest some level of individual agency.

    Consistent with the literature, heavier drinkers have lower endorsement of personal and social attributions of drinking problems. Individuals who had lived with a problem drinker were more likely than those who had not to attribute drinking problems to personal and biological causes. This effect may be associated with the first-hand, day-to-day experiences of observing the struggles and dealing with the frustrations of someone who is addicted to alcohol.

    Protestants were more likely than non-protestants to attribute drinking problems to personal and social causes. Although specific denomination data were not available for this analysis, these findings may reflect the effects of "objector" religious doctrine which is found predominantly in protestant denominations and is especially prevalent in the U.S. South. Interpreting drinking problems as a breach in piety or as a deviation from religious doctrine would account for the attributions to both personal and social cause beliefs. These religious interpretations suggest that drinking problems can, and should, be addressed and overcome.

    There were no significant ethnicity effects. As discussed earlier; categorizing ethnic identity in this community is complicated. Overlap between African American and Native American social identities may result in similar effects among the two groups. However, post-hoc analyses using a combined "ethnic minority" category also failed to detect any significant ethnic differences versus Whites. As noted earlier, the effects of ethnicity on internal/external attributions in the literature may have been due to unmeasured SES effects; specifically education. The homogeneity of educational attainment in this sample may have controlled for these effects to a great extent.

Conclusions

    This study found that older respondents were more likely than younger respondents to attribute drinking problems to biological causes and were less likely to attribute to personal causes or social causes.   Heavy drinkers were less likely than others to attribute to internal causes, suggesting they may also be at increased risk regarding treatment access. Protestants as opposed to non-Protestants, showed a pattern of internal attribution, suggesting beliefs about personal efficacy in controlling drinking.

    The emergence and support of the social cause factor suggests that current drinking problem causation beliefs may not reflect a clear an internal/external dichotomy. The social cause belief suggests that external factors, such as family or stress, may cause drinking problems but these causes are not insurmountable or deterministic. The effect of this attribution on treatment efficacy is unknown, but it does suggest an element of individual agency. Protestants were more likely to attribute drinking problems to the social cause factor, whereas older respondents and heavy drinkers were less likely to attribute to social cause factors.

    Although the literature suggests that some internal attribution is necessary for effective treatment outcomes, over-reliance on personal attribution may also hamper treatment efforts. Blaming all treatment setbacks or difficulties on personal shortcomings could negatively impact motivation in treatment efforts. Knowledge of the biological mechanisms involved may inform the treatment process and balance self-blaming and reproach. In addition, beliefs about the genetic risk for alcoholism often prevent or limit drinking for at-risk individuals.

    Limitations of this study include a non-random sample and inconsistent measurement of religious denomination and current drinking across the two samples. Because these data were originally collected to address other issues compromises had to made in the construction of some of these meaures in order realize the benefits of the greater ethnic and age variation and a larger sample size in the combined data. In addition, the difficulties inherent in measuring social class among college students also prohibited a rigorous SES measure. Future data collection could easily overcome these measurement issues and could also allow for an analysis of changes in these attitudes over time. Ideally, future research would look at these drinking problem attributions in a representative population with more educational and SES variability. Finally, findings from this line of inquiry can be used to examine the effects of these causal beliefs on treatment-seeking and treatment efficacy.

References

Bullers, Susan and Carol A. Prescott. 2001. "The Independent Contributions of Genetics, Shared Family Environment, and Adult Roles to Adult Perceived Control." Sociological Inquiry 71(2):145-163.

Bullers, Susan and Carol A. Prescott. 2000. "Beliefs about the Causes of Drinking Problems, Exposure to Drinking Problems and Perceived Control." American Journal of Medical Genetics, Special Issue: Abstracts of Presentations: Eighth World Congress on Psychiatric Genetics, Versailles, France, 7 August 2000, 96(4): 451–572.

Caetano, Raul, Catherine L. Clark, and Tammy Tam. 1998. "Alcohol Consumption Among Racial/Ethnic Minorities." Alcohol Health and Resarch World 22(4):233-238.

Cunningham, John A., Linda Sobell, Jonathan L. Freedman, and Mark B. Sobell. 1994. "Beliefs About the Causes of Substance Abuse: A Comparison or Three Drugs." Journal of Substance Abuse 6:219-26.

Didier, Marcelo. 2001. "Attributions in the Evaluation of Alternative Solutions to Social Problems." The Journal of Social Psychology 127 (4): 357-65.

Fisher, Jonathan A.D. 1987. "Alcoholism and Race; The Misapplication of Both Concepts to North American Indians." Canadian Review of Sociology and Anthropology 24:81-98.

Ford, Julie and Charles Kadushin. 2002. "Between Sacral Belief and Moral Community: A Multidimensional Approach to the Relationship Between Religion and Alcohol Among Whites and Blacks." Sociological Forum 17(2):255-279.

Furnham, Adrian, and Victoria Lowick. 1984. "Attitudes to Alcohol Consumption: The Attribution of Addiction". Social Science and Medicine 18(8):673-681.

Harris, Michael M., and Mary L. Fennell. 1988. "Perceptions of an Employee Assistance Program and Employees' Willingness to Participate." The Journal of Applied Behavioral Science 24(4): 423-38.

Hoy, Richard M. 1977. "Some Findings Concerning beliefs About Alcoholism." British Journal of Medical Psychology 50: 227-235.

Kaufman, Stephen E., Paula Silver, and John Poulin. 1997. "Gender Differences in Attitudes toward Alcohol, Tobacco, and Other Drugs." Social Work 42(3):231-41.

Kline, Rex B. (2011). Principles and Practice of Structural Equation Modeling (3rd Ed). New York: Guilford Press.

Kohn, Melvin L. 1989. "Social Structure and Personality." A Quintessentially Sociological Approach to Social Psychology." Social Forces 68(1):26-33.

Kosmin, Barry A. and Keysar Ariela. 2009. "American Religious Identification Survey. (ARIS) 2008". http://b27.cc.trincoll.edu/weblogs/AmericanReligionSurvey-ARIS/reports/ARIS_Report_2008.pdf.
Hartford, Connecticut, USA: Trinity College. Retrieved 2011-08-05.

Lambert, Eric G., Lois A. Ventura, David N. Baker, and Morris Jenkins. 2006. "Drug Views: Does Race Matter ?" Journal of Ethnicity in Criminal Justice 4:93-111.

McKirnan, David J. 1984. "The Identification of Alcohol Problems: Socioeconomic Status Differences in Social Norms and Causal Attributions." American Journal of Community Psychology 12(4):465-84.

Mirowsky, John, and Catherine E. Ross. 1989. Social Causes of Psychological Distress. Aldine de Gruyter, New York.

Muthén, Linda K., and Bengt O. Muthén. 2005. Mplus, The Comprehensive Modeling Program for Applied Researchers, User's guide. Los Angeles, CA: Muthén & Muthén

National Institute on Alcohol Abuse and Alcoholism (NIAAA), "How To Screen For Heavy Drinking"http://pubs.niaaa.nih.gov/
publications/Practitioner/PocketGuide/
pocket_guide5.htm Accessed October 12, 2009

Oei, Tian Po, D. Hokin, and Ross McD. Young. 1990. "Differences Between Personal and General Alcohol-Related Beliefs." The International Journal of the Addictions 25(6):641-651.

Pearlin, Leonard. I., and Carmi Schooler. 1978. "The Structure of Coping." Journal of Health and Social Behavior 19:2-21.

Rosenstock, Irwin, Victor Strecher, and Marchall Becker. 1988. "Social Learning Theory and the Health Belief Model." Health Education & Behavior 2(15): 175-183.

Schomerus, Georg, Michael Lucht, Anita Holzinger, Herbert Matschinger, Mauro G. Carta, Mathias C. Angermeyer. 2011. "The Stigma of Alcohol Dependence Compared with Other Mental Disorders: A Review of Population Studies". Alcohol 46: 105-112.

Sigelman, Carol K., Stacy Gurstell, and Amanda K. Stewart. 1992. "The Development of Lay Theories of Problem Drinking: Causes and Cures." Journal of Adolescent Research. 7(3):292-312.

Wild, Cameron T. 2002. "Personal Drinking and Sociocultural Drinking Norms: A Representative Population Study." Journal of Studies on Alcohol 63(4):469-475.