The Effect of Alcohol-Use Status on Self-Reported Stress Levels:
A Statistical Analysis
A Statistical Analysis
This was one of my favorite independent research projects that I've done, it was the final project for a class titled 'Multivariate Statistics for the Social Sciences'. I used an open source dataset from the National Institute of Health (NHIS) and answered the question: what effect does alcohol use have on self-reported stress levels?
The statistical analysis was comprised of one linear regression model, and two multivariate regression models, and I used R to clean the data and conduct the analysis.
My hypothesis was that current drinkers will have higher self-reported stress levels than former drinkers or abstainers. I used abstainers as the reference variable, and concluded that former drinkers have the highest stress levels, followed by current drinkers. This was a surprising finding, and it was extremely satisfying to have concluded on an analysis which did not align with my 'gut feeling,' overall this project really enhanced my love for data analysis.
Written Research Report
annotated-SOC469 Final Paper-1.pdfDataset: The National Health Interview Survey (NHIS)
The NHIS is a nationally representative survey of Americans aimed at understanding the health of the American population and how it changes over time. The dataset is cumulative of data acquired from 2014-2017 and contains almost 150,000 Americans. The NHIS is focused on health and demographics.
2. Research Question:
How do non-drinkers (abstainers) compare to current drinkers, and former drinkers, in their self-reported stress level ratings?
3. Hypothesis:
Current drinkers will report higher stress levels than both abstainers, and former drinkers.
4. Other Predictors:
How does income category (SES), race, and gender affect one's self-reported stress level?
Results
Former drinkers have significantly higher stress levels than the other two groups.
Abstainers have the lowest reported stress level.
Findings are consistent even when controlling for race, income, and gender.
Income is the greatest predictor of stress levels of all of the tested predictors.
Alcohol use is the second greatest predictor of stress levels, even greater than gender or race.
Non-Hispanic white people report more stress than other races, with black people reporting the least amount of stress.
Females report much greater stress than males.
The “most stressed” risk group from the few select variables would be white females, who earn under $35k per year, and are former drinkers.
Script in R: Regression (Statistical Modeling)
final_script.pdfOral Research Report with Q&A
This presentation was presented in only 5 minutes, with additional time to answer questions and defend my analysis. It gives a very brief overview of the final paper.
Emma H Flash Slides.pptx