- Poor public health was significantly associated with the additional Republican presidential votes cast in 2016 over those from 2012
- A substantial association was seen between poor health and a switch in political parties in the last election
- Although not necessarily causal in nature, poor public health may sway election results more than is appreciated and understood
The percentage shift in votes for Republican presidential candidates in 2016 from those cast in 2012 was significantly associated with poor county community health status, per a Massachusetts General Hospital study led by Jason Wasfy, MD.
Subscribe to the latest updates from Cardiovascular Advances in Motion
In an ecological study of data from 3,009 counties, the researchers linked both 2016 and 2012 presidential voting data to the following health measures:
- Physically unhealthy days
- Mentally unhealthy days
- Percent food insecure
- Teen birth rate
- Age-adjusted mortality per 100,000 people per year
- Percent diabetic
- Percent obese
To ensure rigorous data analysis, the team removed variables with information gaps between counties, such as the rate of drug overdose deaths. A principal component analysis stripped away variables that were too highly correlated, including primary care physician rate, violent crime rate and overall health care costs. The remaining unhealthy measures were normalized to a scale and put into a linear regression model with the primary outcome being net voting shift, as a percentage change, between votes cast for Donald Trump in 2016 and those for Mitt Romney in 2012.
87.8% of counties had a positive net voting shift to Trump. A higher normalized unhealthy score was associated with a positive net voting shift (22.1% shift per unit unhealthy with p<0.0001). Together the unhealthy measures accounted for 68% of total data variance. Even when adjusted for demographic variables in a generalized linear model, each unhealthy measure was associated with a net voting shift of 4.1%.
While previous scholars unveiled an association between Republican voting and better health, this trend was not found, even though there was a county-level shift to Republicanism in the 2016 election.
To reach these results, researchers adjusted for gender, education level, race, rural and urban location, total population, percent without English proficiency, income and total health care costs. They also controlled for state-level fixed effects of voter clustering and the effect of conducting the same analysis on individual-level data, as well as the confounding effect between county community health and county voting turnout in both years. To this last point, after adjusting for demographic variables and state-level effects, results revealed no association between unhealthy scores and aggregate voting turnout (p=0.836).
Sensitivity analyses done through repeated regression analyses using the same data across regions, indicated the Northeast showed no effect on the unhealthy score and net voting shift. The Midwest showed more net voting shift in relationship to the unhealthy measures relative to the South and West.
In regards to geography, net voting shift in health status was the strongest in states that changed political parties from 2012 to 2016.
While the causes for the shift are not known, the findings do suggest public health plays a larger role than expected in overall election outcome.