For many, the 2016 election revealed the large degree of geographic and cultural separation within America. While coastal areas and big cities moved more Democratic than they already were, more rural areas and those in the so-called “heartland” of the U.S. turned politically redder than before. Political bubbles have become much clearer in the aftermath of this past election, as it seems as though Americans increasingly interact with and live among others that share the same worldview and political inclinations–creating a sort of political homophily.
I wanted to check this idea of growing political bubbles–which entails people living in more solidly Democratic and Republican areas and fewer living in mixed or “purple” regions–with geographic voting data. I turned to county-level vote totals in the 2008, 2012, and 2016 elections to form margins in favor of either the Democrat or Republican. I used these margins to describe the political environment–the type of bubble–in which voters live.
For example, in this first graph, I did the following for each of the last three elections:
- Calculated the percentage point margin separating Hillary Clinton and Donald Trump in each county
- Classified every county in 10-point margin increments–Clinton winning by more than 80 points, winning by 70-80 points (expressed as “+70”), all the way to Trump winning by more than 80 points–to create different groups/levels of political bubbles
- Summed the total number of voters living in each type of bubble (using total votes cast)
As a guiding example, 13.73 percent of 2012 voters lived in counties where the presidential vote margin was between 10 and 20 points in favor of the Democratic candidate, Barack Obama.
The most purple types of bubbles–ones in which the margin between the Democrat and Republican is small and thus have many people of opposite parties living in proximity of each other–are located in the center of the above distributions. Where the Democratic margin of victory was between 0 and 10 points, 15.9 percent of 2008 voters, 15.1 percent of 2012 voters, and 12.1 percent of 2016 voters resided. For counties that saw between a 0-10 point close Republican victory, 10.1 percent of 2008 voters, 11.2 percent of 2012 voters, and 9.8 percent of 2016 voters lived there. Thus, fewer voters are living in the most politically diverse counties in the country, with the biggest decrease occurring in 2016–an election with Clinton and Trump as candidates saw fewer counties with mixed vote preferences.
The next most heterogeneous categories in Democrat +10 margin (15.6 to 13.7 to 11.4 percent share of all voters) and Republican +10 margin (10.9 to 10.5 to 8.8) all have fewer voters living in them, with 2016 introducing the largest decrease. So where are all these voters now living? In much more homogeneous counties. Relative to 2012, the Dem +20, Dem +30, Dem +50, Dem +60, and Dem +70 bubbles–areas where the vote is much more homogeneous and befitting of a “political bubble” label–all see increases in number of voters in the 2016 election. On the other end of the spectrum, the same growth of bubbles occurs for every group from Rep +20 through Rep +80. To put this in terms easier to understand, 1.7 percent of 2008 voters lived in counties where McCain won by 50-60 points (+50), 2.5 of 2012 voters lived in counties where Romney won by 50-60 points, and 4.9 of 2016 voters lived in counties where Trump won by 50-60 points. In general, the distribution of voters is becoming flatter across degree of political bubbles–more and more voters now live in counties with an increasingly one-sided blue or red vote.
To better visualize what’s going on (and reduce the number of bubble buckets), below is the same graph but simplified to 20-point margin increments.
The same story of growing politically homogeneous bubbles materializes. Eight years ago, 26.1 percent of all voters lived in counties where the margin between Democratic and Republican vote shares was less than 10 percent; in 2016, that number has dropped to 21.9. Fewer voters are now living in counties where Democrats won by 10-30 points and 30-50 points, but more voters now live counties that are decidedly blue (by more than 50 points). While 7.5 percent of voters lived in counties that saw a 30-50 point Republican victory in 2008, 13.1 percent of voters now live in this territory. The portion of electorate living in the most Republican counties–50 points or more red–has increased by more than five percentage points as well.
Finally, to measure just total homogeneity and not whether the bubble has a Democratic or Republican bent, the below graph shows where voters live based on absolute differences in major party support (smaller vote margins–such as “0-10 % Diff.”–indicate more heterogeneous political areas and larger vote margins–such as “50+ % Diff.”–indicate more homogeneous areas).
Slight pluralities of voters still live in more politically mixed counties, but it’s changing. 26.1 percent of 2008 voters and 26.4 percent of 2012 voters lived in counties where the margin between the Democratic and Republican candidates was less than 10 points–only 21.91 of voters did so in 2016. The percentage of voters living in counties with a 10-20 point margin (e.g. where the Democrat won 55 percent and Republican won 40 percent) also dropped from 26.5 in 2008 to 20.2 in 2016. The share of voters living in 30-40 point margin counties grew slowly, but the biggest change occurs at the extreme end: the percentage of voters living in counties where the Democrat or Republican candidates won by more than 50 points grew from 10 in 2008 to 16.8 to 2016. In other words, many more voters are now living in counties that contain very homogeneous vote preferences–either one-sidedly Democrat or Republican. Political bubbles were already starting to grow a bit, but the 2016 election accelerated this process, contributing a clearly observable increase in county-level geographic sorting.
My pieced was posted over at the Decision Desk (about the same content, but you’ll find graphs you can zoom in on there).
Also, check out the The Crosstab’s post that dives deeper into this topic.