Influences of Alcohol Consumption in the United States

Overview

In this project, I explored the relationship between alcohol consumption and various socio-economic and health-related factors using data from the IPUMS Health Survey, which draws on the National Health Interview Survey (NHIS)—the leading source of health data on the U.S. civilian noninstitutionalized population. I built a linear regression model that, while statistically significant overall (low p-value on the F-test), explained a modest portion of the variance in drinking behavior (R² ≈ 0.05).

Despite the low R², the model identified several statistically significant predictors of alcohol consumption, including sex, race, hours worked, education level, BMI, smoking status, and diabetes. Notably, men reported ~33% more drinking days than women, and white respondents drank most frequently on average. Higher education was associated with increased drinking frequency, while smokers drank ~29% more often than non-smokers. Conversely, individuals with diabetes reported ~42% fewer drinking days.

These findings reveal clear and consistent demographic patterns in alcohol use, suggesting that factors like gender, race, work habits, and health status are strong predictors of consumption. The results have practical implications for tailoring public health messaging, informing community-level screening, and guiding education campaigns. Future models that integrate geographic, social, and psychological variables could enhance predictive power and better inform targeted interventions.