This is the 5th post in my series on County Health Rankings variables and Trump’s vote in Pennsylvania. Previously, I posted on the univariate relationship between some college, median income, and Trump’s vote in Pennsylvania counties. When I entered these and other variables into a multiple regression model, median income fell out of the model. The percentage of the county with at least some college and 6 other variables remained in the model accounting for 92.9% of the variability. The model states that for every 1% increase in the county with at least some college education, there is a predicted 0.416% decrease in Trump’s vote in PA. This variable has the strongest association with Trump’s vote.
The graph above shows the univariate relationship between the college educated population and Trump’s vote. This variable accounts for 55.8% of the variance in Trump’s vote. Forest and Philadelphia Counties were outliers in this relationship. Forest county is a rural county with 32.3% with some college and 71.8% of the vote for Trump. Philadelphia County is of course urban with 68.8% having some college and 20% of the vote for Trump.
The second graph shows the relationship when Forest and Philadelphia counties are excluded. The univariate model accounts for 63.2% of the variance. This shows the effect outliers can have in a regression model. With the relationship between some college and Trump’s vote in PA, we can certainly see why Trump loves the poorly educated. Next up I will look at county level census participation and Trump’s vote.
Allegheny Independent Media is affiliated with Nonprofit Partners. It is a subsidiary of the Community Foundation for the Alleghenies. Through the site you can donate to the website. It is essential to allow us to provide coverage to the critical issues facing the Alleghenies. You may donate to our site here.