US Presidential Election, Part II: Sociodemographic characteristics and party affiliations

In a previous post, I outlined the terms of the analysis this Blog and you are going to make to the 2016 US Presidential election. In summary, we are going to adventure into trying to understand voters across counties in USA.

The first thing to do is to take you, the reader, for a journey into demographic and economical data, and describe those counties won by Trump and Clinton, without venturing into statistical significance just yet.

In the next part of the election analysis we are going to describe this data and separate counties into those won by Trump/Clinton by a small and by a large margin, and try to describe the population of these counties by their characteristics.

Next, we are going to group counties by some characteristics, and using a statistical analysis called cluster analysis, we are going to classify these counties.

For clarity and to make the analyses simpler, we are going to exclude Alaska and Hawaii from the maps.

Characteristics of counties across the US

We must start with the counts of population per county:

 

The latter graph shows the population in counties won by Trump (red) and Clinton (blue).


Now, to analyze and try to understand the 2016 election, let’s look at the unemployment rate in each county for 2016 and 2012:

 

In the next figure, the vertical axis are counties, over the horizontal axis oshowing unemployment rate:

 

Keeping in mind we are not looking for statistical significance and are merely looking for effect size and differences, the figures suggests that the unemployment rate did not differ in counties won by Trump and Clinton: 36% of Clinton counties had unemployment below the national rate, while unemployment in 39% of Trump counties were below the national rate.

This data means that lack of employment did not had an effect in Trump or Clinton votes in the 2016 election.


Let’s look now to income inequality:

Income inequality within the county, measured as the ratio of the mean income of the top 20% divided by the mean income of the bottom 20% of earners, was 30% higher in Clinton counties in 2015, compared to counties won by Trump (again keeping in mind we are not looking for statistical significance).

 


 

 

We turn our attention now to per-capita income in 2016:

The per-capita income in counties won by Clinton was 27% larger, a difference of $11,529 in a year. This is a large difference, and for now, it is unknown why this difference happens.

 

Let’s outline some possible takeaways for the data:

  • The DEM counties are more populous and economically developed, with more income inequality.
  • GOP counties won by Trump have lower per-capita income compared to counties won by Clinton.
  • The unemployment rate was not a driver of 2016 presidential election results in the United States.

We cannot suggest anything further without more in-depth statistical analyses, which I will try to do in the next few weeks, and post it in this blog.

 


Please submit any comment you may have to angel.paternina @ gmail.com. I will reply to each comment made to this address, and post in the Blog the most interesting ones.

Source for county data: Federal Reserve Bank of St. Louis, FRED.

Source for Source for election data: https://github.com/stoneyv/presidential_general_elections