All analyses were population weighted and tests for the key assumptions of this analysis were undertaken.57 The data breached the linearity of log-odds assumption for AUDIT-C; therefore, an exploration of higher polynomial terms for AUDIT-C was undertaken. This indicated that AUDIT-C had a quadratic relationship with the dependent variable; consequently, a linear and quadratic term for AUDIT-C was included in the model. There was no evidence of multicollinearity among independent variables using variance inflation factors (online supplemental Table 2). The discriminative power of the primary model was assessed using receiver operating characteristic area under the curve (AUC). Alcohol consumption has frequently been linked to sociodemographic factors including gender, ethnicity, and age (Cahalan et al., 1969; Clark & Midanik, 1982; Hilton, 1987).
A diary or in-depth interview study would allow for the use of more sensitive measures than were possible in the large-scale survey described here. Qualitative research which examines the complex interplay of individual, social, and situational factors would be of particular value. This study’s results suggest that a strong belief in either coping or social reasons for drinking alcohol puts individuals at risk for abusing alcohol, especially when the appropriate environmental circumstances arise. In the second preliminary step, the interactions between each of these three sociodemographic factors and each motive for drinking alcohol (six interaction terms) were entered as a block of predictor variables. This determined if motives for drinking interacted with demographic factors in a consistent way. For heavy alcohol consumption, the interaction between age and coping motives was a significant predictor variable.
Descriptive analyses illustrate the proportions of respondents consuming no/lo drinks at least monthly for low and high endorsers of each alcohol drinking motive. Quasibinomial logistic regression models, including drinking motives as continuous variables, tested for associations between regular no/lo consumption (dependent variable) and alcohol drinking motives. This method is a robust approach for binary outcomes when overdispersion is present,55 56 which was a concern given the low base rate of at least monthly no/lo consumption (21%) in our sample. While negative binomial or zero-inflated regression models are valuable for addressing overdispersion, they are primarily designed for count data rather than the binary (yes/no) outcome capturing no/lo consumption in this study. The quasibinomial approach, which models a dispersion parameter, was thus the most appropriate method to account for overdispersion while maintaining the binary nature of our dependent variable. There is also a large literature on people’s self-reported alcohol expectancies.
A major focus of this study concerned the role of physical availability in alcohol purchase and consumption decisions. Consequently, participants were required to be of the legal drinking age so that their purchase habits (by legal means) could be assessed (Abbey, Scott, & Smith, 1993). A 2024 report from the American Association for Cancer Research concluded that more than 5% of all cancers in the U.S. are attributable to alcohol use.
The median household income for study participants fell in the range of $15,000 to $24,999. As an indicator of heavy alcohol consumption, study participants were asked to rate how often in the past month they had consumed five or more alcoholic drinks on one day (Cahalan et al., 1969; Hilton, 1987). This question was answered using a 5-point Likert-type scale with response options ranging from “never” to “nearly every time or every time” they drank.
All “reasons” items were answered using 4-point Likert-type scales with response options ranging from “not at all important” to “very important” as reasons for drinking alcohol. Moderate drinking is typically defined by public health agencies as up to one alcoholic drink per day for women and up to two for men. A standard drink is 12 ounces of beer, 5 ounces of wine or 1.5 ounces of distilled spirits. Factors including age, genetics, body size and existing health conditions all influence how alcohol affects a person. This shift in understanding is particularly significant because it challenges deeply ingrained cultural beliefs about the healthfulness of certain alcoholic beverages. It’s becoming clear that the potential risks of alcohol consumption, even in moderate amounts, may outweigh any perceived benefits.
In some of the data analyses reported in this paper, frequency and quantity were multiplied to produce an indicator of total monthly alcohol consumption. Studies using similar (but not identical) measures of alcohol consumption found high reliability in self-reports (Russell, Welte, & Barnes, 1991; Williams, Aitken, & Malin, 1985). In this study, the four types of alcoholic beverages (beer, wine, wine coolers, and liquor) were mentioned in each question, and study participants were asked to take a minute to think before giving their answers. This is the first study to quantitatively explore associations between the reasons adults drink alcohol and the consumption of no/lo drinks. Thirty-five-minute telephone interviews were conducted with 781 Michigan residents who had consumed alcohol in the past 30 days. In order to examine ethnic differences in predictors of alcohol consumption, the sample was restricted to White and Black individuals, and Blacks were oversampled.
On average, 2% of additional variance was explained when these interaction terms were included. When coping motives were high as compared to low, individuals experiencing moderate or high levels of stress engaged in more heavy alcohol consumption. When social motives were high as compared to low, individuals whose friends were high-frequency drinkers engaged in the most heavy drinking. The unadjusted regression model included drinking motives and no/lo consumption. The adjusted model controlled for sociodemographic characteristics (gender, age, education, social grade, and IMD) and hazardous drinking (AUDIT-C).
Study participants ranged in age from 21 (the minimum legal drinking age in Michigan) to 86 years, with a median age of 37 years. Thirty-three percent of study participants resided in a city, 14% lived in a suburb, 18% lived in a town, and 35% lived in a rural area. Eighty-seven percent of study participants had at least a high school education.
Understanding these patterns can inform alcohol policies, public health messaging, and responses to future crises. Ultimately, the decision to drink alcohol remains a personal choice, but it’s one that should be made with a full understanding of the potential risks and benefits. As the scientific community continues to investigate alcohol’s effects on health, we can expect further refinements in our understanding, potentially leading to more targeted and effective public health strategies in the future. One of the most striking findings is that the previously observed benefits of moderate drinking have essentially vanished under closer scrutiny. The health halo that once surrounded alcoholic beverages, particularly red wine, is rapidly dissipating as more rigorous and unbiased studies come to light. Overall, 52% of Americans ages 21 and older say they’ve heard about studies showing that drinking alcohol can increase a person’s risk of cancer.
For over sixty years, the Alcohol Research Group (ARG) has been actively engaged in critically needed alcohol- and other drug-related public health research. We study drinking and other drug use and how these and other factors such as gender, race/ethnicity, sexual identity, socioeconomic disparities, and environmental differences affect health. ARG is also home to the NIAAA-funded National Alcohol Research Center and training program. A preliminary set of subgroup hierarchical multiple regression analyses were conducted.
Drinking for social reasons was assessed with four items which asked study participants the extent to which they drank alcohol in order to be sociable, to enhance the enjoyment of social situations, because the people they knew drank, and to celebrate social occasions. why alcoholics drink research insights Age, social grade and education were treated as factors, whereas IMD was treated as a continuous variable. Ethnicity is reported descriptively (white, black, Asian, mixed heritage, other, table 1) but was not included in the regression model due to small numbers of black, Asian and other ethnically diverse groups in the sample population. Finally, we’re learning more about the impact of alcohol on women and older adults. Women have begun to catch up to men in alcohol consumption and alcohol-related harms.
We measured frequency of no/lo consumption as a single item.39 Participants were asked, “How often do you have an alcohol-free or low-alcohol drink (beer, wine, cider, spirits or other type of alcoholic drink under 1.2% ABV)? Participants responded on an 8-point scale, ranging from never to nearly every day. Due to low numbers responding at higher frequencies, responses were recoded as a binary variable–less than monthly/at least monthly, to capture whether respondents were a regular consumer of no/lo drinks or not.