How Hillary Lost by Focusing on Cities — and Trump Won By Focusing on Rural Areas
Author’s Note: I wrote this article in December 2016 and never published it, primarily because it became clear at the time that there were other important factors to consider in the 2016 election, including foreign interference. However, I think this is likely relevant now as campaigns are making decisions about where and how to allocate resources — and as far as I know no one else spotted this in the data. If I were to rewrite this article today, I’d include more contemporary analysis, especially regarding swing states. Still, as an artifact from late 2016, I think it has intrinsic value today.
Over the last few years, many have remarked that our presidential elections have increasingly reflected a divide between rural and urban America. I wrote my own analysis of this phenomenon in 2012 to which several other prominent commentators referred.
In this 2016 election, the phenomenon has been accepted as orthodoxy: America is truly a country divided. But I thought it would be useful to do some new analysis that attempts to understand this in more detail.
Let’s take a look at voting behavior by population density for the last five elections, back to Bush vs. Gore in 2000.
This graph offers three significant lessons. First, the parties have been attenuated around population density for some time, and each election cycle, the margins become more extreme. Second, the crossover point between Republican and Democratic primacy has always been between around 1,000 people per square mile. Third, Trump’s margin in low-density counties was higher than for any Republican candidate in the last five elections; likewise, Clinton’s margin in low-density areas was lower than for any recent Democratic candidate.
We also see that Clinton’s performance in high density areas was about the same, but just slightly lower, than Obama’s in 2012. However, in the highest density county (Manhattan) Clinton outperformed him significantly. (This should perhaps not be surprising, as Trump is deeply unpopular in Manhattan — look at his own low margin there. She also campaigned there heavily.)
Was this Strategic?
It’s hard to know whether this was something that either campaign pursued intentionally. Arguably, Trump has few real “friends,” and his campaign appearance strategy was based on his instincts gained through his work in television, and catering to areas he felt he would be well-received. Clinton, by contrast, has many celebrity supporters in New York, Los Angeles, and other major population centers. It would have been tempting to do fundraisers in these areas rather than spend time in lower density areas.
To try to get an answer to this question, I conducted an analysis of each candidate’s campaign appearances by location.
Clinton’s campaign conducted 52% of their appearances in counties with more than 1,000 people per square mile. By contrast, 79% of Trump’s events were in counties with fewer than 1,000 people per square mile.
Another startling fact: Clinton’s campaign held 120 events in New York City. Trump held none.
In Los Angeles County, Clinton held 63 events. Trump? Again, zero.
In fact, the highest density place that Trump held a campaign appearance was in Denver, Colorado (with about 4,400 people per square mile, vs. 49,000 in New York City.)
So whether by intention or instinct, the strategies of the two campaigns were very, very different, and focused on distinctly different geographies.
Clinton held at least 1,665 events, while Trump held just 321.
Arguably, had Clinton simply spent more time in the hinterlands, it may have had some effect; but on the other hand, these differences may just be a reflection of how deeply this country is now divided — between the high density coasts, and the low-density exurbs and rural states. Each zone has distinctly different concerns for the future, and a distinctly different economic reality.
Which Way Forward?
Arguably there are two ways forward. Parties can either recognize the division between rural and urban that exists in this country, and decide to cater to both ends of the spectrum.
Or they can double down, further divide, and sharpen rhetoric that appeals to their respective constituencies. I’d argue that human history suggests that this latter approach will be the tendency. Nature abhors a compromise, once popularly-held views coalesce to a point where any reversal is perceived as defeat. We have likely passed this point.
Given that America’s highest density areas have the upper hand with regards to wealth production, it’s probably time to worry less about bringing the country’s lowest-density areas along, and create bottom-up alliances that bring together the country’s highest-density cities and states.
The rest of the country wants to go in a different direction. Maybe it’s time to let them go.
Dave Troy is a technologist, data analyst, and student of human networks based in Baltimore, Maryland. Watch his TED talk here. He can be reached at dave@410labs.com.
