Analysis: The polarization express

Political  junkies continue to pore over the 2020 presidential election maps featured in the New York Times looking for keys to electoral trends: the maps graphically display the vote shift from 2016 to 2020 in every county in the United States. 

The setup is simple: counties with larger vote margins for Donald Trump in 2020 than in 2016 are pinned with a red arrow pointing to the right; counties with larger margins for Joe Biden than Hillary Clinton have a blue arrow pointing left. The size of the arrow is based on the size of the margin between 2020 and 2016. 

There are no significant outliers in Tennessee. The map confirms that Trump’s popularity grew in rural areas and shrank in counties near Tennessee’s four main urban centers.

In an election that the victor has defined as a referendum on unity, this map does not paint a picture of Tennesseans coming together. 

Rather, it says that the trenches are deeper than they were four years ago. On average, where Trump won in 2020, he won by more than he did in 2016 — 53.5% to 51.4% — and where he lost, he lost by more. 

This trend extends beyond the presidential election. Local elections featured Republican candidates increasingly strident in their fealty to Trump and their Democratic opponents who hoped to capitalize on a uniquely unpopular incumbent president failed to bridge the gap. Democrats flipped only one seat in the state legislature. That lone Democratic victory was against Steve Dickerson, one of the few remaining moderate Republicans in the General Assembly.

Tennessee has little to show for its rich history of political centrism. The Howard Bakers of the world belong to a bygone era, and despite some significant movement towards the middle in a handful of counties, purple is not in our color wheel. What remains are historic levels of partisanship.  Tennessee, the 16th most populous state in the country, delivered Trump the largest raw vote margin of any state in the nation. 

Our political ecosystem has become defined by its polarity. Geographically speaking, where do we appear to be the least politically aligned? When placed side by side, which counties have the most dissimilar political preferences? Which are moving most swiftly in opposite directions?

The election left a wealth of voting data in its wake, so the answers are well within reach. Consider this my own personal data dump, comprising what I think are the seven most relevant observations regarding the election in Tennessee. 

1) The opposite ends of the political spectrum

Below are the 10 counties where Trump performed best and the 10 counties where Biden performed best. Numbers in red represent Trump’s margin of victory; blue, Biden’s. They are the counties with the most dissimilar political preferences in Tennessee in 2020. They represent the poles.

The eight counties painted purple are not actually purple. They are still majority Republican by a comfortable margin. They are painted purple in this map only to demonstrate that while Republican, they voted more Democratic than the other 85 counties in the state.

2) The counties shifting furthest to the right and left

Next are the 10 counties that shifted most dramatically towards Trump from 2016 to 2020 and the 10 counties that shifted most dramatically away from him. Calculating the vote shift involved measuring the difference between Trump’s 2016 margin of victory (or defeat) and his 2020 margin. A negative result indicates a shift to the left. For example, Rutherford county’s vote shift was -10.44%.

3) The rural and urban gap: population as a predictor of Trump’s margin of victory

Joe Biden’s victory was cemented by overwhelming wins in the nation’s largest cities and suburbs. The so-called “blue wall” that Democrats worked so hard to rebuild is really a series of blue outposts in an ocean of red—urban centers towering over small-town rural America. The urban and rural gap has become the basis of electoral politics in this era of American history. 

To predict either Biden’s or Trump’s popularity in any given county in Tennessee, you can do no better than consult the census population estimates. Where the population is higher in Tennessee, Biden’s favorability increased. Where it is lower, Trump dominated. This trend is particularly striking when observing the counties with the most dissimilar political preferences. The total population in the 10 most left-leaning counties is 3,421,489; in the 10 most right-leaning counties, it is only 197,570. The 10 most left-leaning counties are almost 18 times larger than the 10 most right-leaning.

What happens if you broaden the sample size? If you list every county by Trump’s margin of victory in descending order, the total population of the counties above the median (indicating a higher margin of victory—63.83% on average) is 1,315,757. Total population below the median (indicating a lower margin of victory—37.79% on average) is a whopping 5,560,051. That population gap is larger than the entire state of Oregon. Even if you take out Shelby and Davidson counties, the two most populous counties and most formidable Democratic strongholds in the state, the total population below the median (42.43% on average) is still 3,924,956.

The graph below illustrates this trend. It plots Trump’s margin of victory in 2020 for every county in Tennessee against population density (people per square mile) in ascending order. The scatter plot and its corresponding linear trendline show that Trump’s margin of victory tends to decrease as population density increases. That tremendous dip on the far left end of the scatter plot is Haywood County. It is the only rural county in Tennessee that voted Democratic. Trump recorded a negative margin of victory there. In other words, he lost.

4) The rural and urban gap: population as a predictor of the vote shift

Margin of victory is significant, but it’s two-dimensional. It says only that rural areas lean more Republican and urban areas more Democratic. It doesn’t provide any insight into which direction urban and rural areas are heading—one of the most pressing questions for our two-party system in the post Trump era. What will happen when the most divisive American political figure of our lifetime is no longer on the ballot? Theories abound. In the meantime, observe the vote shift.

If the next four years are anything like the last four, we may be in trouble. The average shift in the 10 least dense counties was 6.05%  to the right; in the 10 densest, it was 6.16%  to the left or a near mirror image. And just how stark is the population gap? The total population in the 10 counties that shifted furthest to the left is 2,954,745; in the 10 counties that shifted furthest to the right, 121,252. 

When you broaden the sample size, the trend is obvious. The graph below plots the vote shift from 2016 to 2020 against population density in ascending order. Note that any shift above 0% means that the county shifted to the right, and any shift below 0% means the county shifted to the left. The scatter plot and its corresponding linear trendline show that the vote shift tends to decrease as population density increases. This is to say, rural areas are becoming more Republican, and urban and suburban areas more Democratic.

5) The rural and urban gap: population growth as a predictor of the vote shift

To add another dimension, consider what is happening to the population in counties shifting to the right or left. Are they growing or shrinking?  The graph below plots the vote shift against population growth rates from 2010 to 2020 in ascending order. The scatter plot and its corresponding linear trendline show that the vote shift tends to decrease as population growth rates increase. So when counties shrink, they tend to shift more to the right. Alternatively, as they grow, they tend to shift more to the left. It is slight, and the downward slope of the trendline small, but it is significant.

 The picture is clearest at the poles. The average population change over the last 10 years among the 10 counties that shifted furthest to the left was 18.33%. This group is headlined by Williamson and Rutherford counties, which grew by 32.16% and 29.43%, and shifted to the left by 8.86% and 10.44%. This is compared to the 10 counties that shifted furthest to the right, which grew by a much smaller 5.2%. Furthermore, if you remove Trousdale County, which grew by an almost unfathomable 43.84% in 10 years (the largest such figure in the state), the average population change was a measly 0.91%. But regardless of whether we include Trousdale or junk it as an outlier, the counties that have been moving most swiftly to the right have also been shrinking relative to the state of Tennessee, which as a whole grew by about 7.5% over the same time span.

6) Economic distress in rural counties: population density as a predictor of economic well-being

So what if urban and rural communities are diverging politically? So what if Republicans seem to perform better where the population has been more stagnant and worse where it has grown? Ultimately voters don’t consider population data when they enter the voting booth. People don’t decide to vote for Republicans because they live in a small town, in the same way that hurricanes don’t decide to hit the Gulf Coast. Hurricanes hit the Gulf Coast because the water is warmer in the Gulf of Mexico, the Atlantic Ocean rotates clockwise and low air pressure zones hover over the Caribbean. Choices at the ballot box are made for a similarly long list of complex cultural and economic factors. The question then is which factors can help explain diverging political preferences in urban and rural areas? Judging only by some very basic measures of economic well-being, the most rural areas in Tennessee appear to be in worse shape than their more urban counterparts.

*Note: the figures in the first two columns are averages. For example, to find the poverty rate for the 10 least population dense counties, I calculated the average poverty rate among each of those 10 counties. This ensures that each county is weighted equally.

These figures are startling, not only because the 10 most dense counties perform better by every single metric, but because of how far the 10 least dense counties lag behind statewide figures. The median household income is $16,000 lower than the rest of the state. Nearly one in five people live in poverty. The economy is completely stagnant at a 0.3% GDP growth rate. 

These places have been discarded from our civic consciousness. Meanwhile, Tennessee has amassed the largest Temporary Assistance for Needy Families (TANF) surplus in the nation. We are currently sitting on $741 million in reserve funds, with no plan to use it.

What if you broaden the scope? Unfortunately, doing so doesn’t inspire any more joy. For each of the seven metrics listed in the table above, as population density increases, the numbers improve. Rural counties are worse off in every metric compared to urban counties. 

• By plotting the poverty rate for every county in the state in ascending order of population density, it appears that as population density increases, the poverty rate decreases. That big spike on the left end of the graph, by the way, is Lake County. It is one of the 10 counties that shifted furthest to the right over the past four years. The poverty rate in Lake county is 36.5%, the highest in Tennessee.

• By plotting educational attainment (the percent of the population with a high school degree or higher) for every county in the state in ascending order of population density, it appears that as population density increases, educational attainment increases.

• By plotting median household income for every county in the state in ascending order of population density, it appears that as population density increases, median household income increases.

7) Economic distress in rural counties shifting furthest to the right: vote shift as a predictor of economic well-being

Economic shortcomings in sparsely populated areas of Tennessee do not necessarily explain their gravitation towards the Republican party, right? Unfortunately, in Tennessee, it does to some extent. The clues are all right there.

First, unlike population density, voters do carry things like personal finance, age, and education into the voting booth. These are the sorts of things that chart the hurricane’s course. If you and your community are struggling to make ends meet, you might be more likely to vote for the candidate who promises to bring jobs back from China. If you don’t have a high school degree, you might be more likely to vote for the candidate that disparages higher education and basic science. And if you are over the age of 65 and anxious about the rapid pace of societal and demographic change, you might be more likely to vote for the candidate who wants things to go back to the way things were, or dare I say, make America great again. In rural communities, which are generally poorer, older, and less educated, the choice has become easy. 

“Rural areas have been losers in the globalization of trade,” says Kent Syler, a political science professor at Middle Tennessee State University. “We watched many, many factories and opportunities move out of these rural counties. It has happened across the country, and it dramatically changed the standard of living. Republicans have done a good political job of capitalizing on that [economic] distress. Especially with president Trump basically vilifying certain people as being responsible for it.”

Syler’s explanation meshes with the data uncovered thus far, chiefly that rural areas vote more Republican and that rural areas are worse off economically. It would be statistical malpractice to suggest proof of causation on those grounds, but there does appear to be a relationship between the dilapidated state of rural economies and Republican party preference. Let’s take a look at the same seven economic indicators, but instead of reporting on the most rural and urban counties, reintroduce the group of counties that shifted furthest to the right and left. 

The counties where Republican enthusiasm has grown most dramatically perform worse in every category. This table is effectively identical to the one evaluating the most rural and urban counties. These communities, which not only voted for Trump by huge margins, but doubled down on that support over the past four years, are all operating under some degree of economic distress. Republicans, and Trump especially, have clearly figured out a way to appeal to that sense of distress.

This trend is not limited to the poles. For each of the seven metrics listed in the table above, as the vote shift increases, the numbers get worse. Counties that recorded a rightward vote shift tend to be worse off than those with a leftward vote shift.

• By plotting the poverty rate for every county in the state in ascending order of vote shift, it appears that as the vote shift increases, the poverty rate increases. In other words, poverty rates tend to be higher in counties that have shifted further to the right.

• By plotting educational attainment (the percent of the population with a high school degree or higher) for every county in the state in ascending order of vote shift, it appears that as the vote shift increases, educational attainment decreases. In other words, educational attainment tends to be lower in counties that have shifted further to the right. 

• By plotting median household income for every county in the state in ascending order of vote shift, as the vote shift increases, median household income decreases. In other words, median household income tends to be lower in counties that have shifted further to the right. 

Is it the chicken or the egg? Did Trump and the Republican party capitalize on economic disillusionment in rural communities, or did economic disillusionment in rural communities lead voters to Trump and the Republican party. They aren’t mutually exclusive events, and in all likelihood it was a bit of both. But it does call into question the incentive structure facing American political parties, and the responsibility they bear for creating an environment in which economic distress and geographic isolation are catalysts for political polarization.

Politics is about identifying a set of issues, and building a coalition around some plan to address them. Political parties are just vehicles to building those coalitions, which give elected officials the ability to crowd source momentum for policy solutions to the issues that matter most. My concern is not that we are divided on those solutions, which we are. If it were only the case that people in rural areas tended to vote for Republicans because they are married on policy, then I would sleep better at night. Voting for the candidate or party that best represents your worldview is a normal incentive. And when worldviews in urban and rural areas differ, you can see how we ended up where we are.

My concern is that our political parties and the politicians they feed have an incentive to keep divisions hot and worldviews incompatible—to stoke the flames between rural and urban Americans rather than actually fix the economic divisions that seem to have brought us here. 

After all, there’s a reason more people voted in the 2020 election than ever before. It’s the same reason that the trenches are deeper now than they were four years ago, and why our politics appear to exist at opposite ends—both economically and geographically. When people are angry and their opposition is well defined, they vote. It will be hard to put Pandora back in the box.