Photo credit: Carandoom

On November 3, Joe Biden was elected to be the 46th president of the United States. This was the expected outcome, according to most mainstream polls. Yet political polling was even further off than it was in 2016, with predictions particularly wrong among white voters without college degrees, Midwestern swing states, and miscalculating the early voters. The highly anticipated "blue wave" failed to materialize: despite winning the presidency, Democrats didn't manage to flip a single state chamber and lost at least 13 seats in the House.

What happened? Official voter data from the U.S. Census won't be available for at least another year, but exit polls can provide some clues. According to a New York Times analysis of 15,590 voter interviews we know quite a bit about the kind of person who tended to vote for Joe Biden or Donald Trump. The people most likely to vote for Biden are Black, young, LGBT, urban, unmarried women, and holders of graduate degrees. By contrast, the populations most likely to have voted for Trump are evangelical or born-again Christian, living in the rural Midwest and South, white and middle-aged, noncollege graduate, and male.

Politicos continue to debate the existence of the 'shy Trump voter': a supposed group of eventual Trump voters that reported to pollsters that they would vote for Biden. Why lie? In some cases, people did not feel comfortable disclosing their actual candidate, while others were actually still undecided. Others may have deliberately lied to pollsters to sabotage polls. Myth or not, this supposed group of Trump voters was flagged in the 2020 election (as they were in 2016) to help pollsters, researchers, and news media explain voters' unexpectedly enduring support for Trump.

Do progressives live in a bubble?

Political scientists would say that polls were never meant to be prophetic, but rather a snapshot of one group at one point in time. But talk of the "shy Trump voter" and the failure of the "blue wave" to materialize are also representative of a kind of collective surprise felt in liberal circles at Trump's enduring popularity with his voters.

Some commentators have decried Trump's "cult-like" support, encouraged by Trump's own assertion in a 2016 campaign rally that he could "stand in the middle of 5th Avenue and shoot somebody" and not lose any votes. But many Trump voters simply felt that the Trump campaign's promises best reflected their needs, in ways that were often reduced by commentators to simple racism (and correctly so, in some cases), but for many was just plain old political differences.

But the foreignness of the Trump voter psyche to Biden supporters is also emblematic of a growing separation between liberal and conservative voters that extends well beyond elections. The United States has become sufficiently divided -- in shorthand, between the progressive, coastal cities and the conservative, rural inner-America -- that genuine calls are made to split the country, to "get rid of" liberal states like California, even to have a second Civil War.

Of course, Democrats are a diverse bunch: many are poor, many own guns, many serve in the military, and they are to be found in cities, towns, and farms across the country. And progressives are certainly under no obligation to take responsibility for the beliefs of the Republican party, nor to shoulder the responsibility for the hyper-partisan divide. It is, however, incumbent upon both parties to reach beyond whatever bubble they inhabit to genuinely understand the mindsets, beliefs, and drivers of fellow citizens with vastly different political ideologies.

And the first step is to understand, and acknowledge, whatever kind of bubble each one of us is living in.

What's your bubble?

In the 2020 presidential election, educational attainment emerged as one of the defining divisions between voters. According to the New York Times analysis, 54% of voters that never attended college voted for Trump, while 46% voted for Biden. By contrast, 62% of voters with a graduate degree were Biden voters (while just 37% voted for Trump).

Notably, educational level does not have the same predictive power among nonwhite voters: about 70% of nonwhite voters supported Biden, regardless of whether they graduated from college.

Source: The U.S. Census. 2019. Educational attainment is among adults 25 years and older.

Geography and place of residence were also an important predictor of votes in the 2020 election. Urban voters, for example, were most likely to vote Biden: 60% of people living in cities with more than 50,000 residents voted Biden, while 57% of those in small cities or rural areas voted for Trump.

And it's not just urban density, which state voters live in matters too. In 2020, voters in the coastal East and West tended to be Biden voters (58% and 57% voted for Biden, respectively), while those in the Midwest and South went for Trump (51% and 53% voted for Trump).

Source: The U.S. Census. Rurality data are from Census's urban-rural classification of counties, and counties are defined as 'coastal' if they contain an ocean coast.

In the 2016 election, Hillary Clinton's failure with blue-collar workers in "rust belt" states like Michigan and Ohio played an important role in Trump's election.

In the 2020 election, this voting bloc was again in center stage, driving debates about Trump's trade war with China, Biden's historic support for the free-trade agreement NAFTA, and both candidates' proposed policies impacting union workers and farmers.

Source: The Bureau of Labor Statistics. Employment figures are calculated from major occupational groups (May 2019) and industry. White collar sectors are defined as government, sales, information, finance and insurance, real estate, professional and technical services, management, education, health care and social assistance, arts, entertainment, and recreation, accommodation and food services, and part of administrative and support and waste management. "Professional" white collar jobs include occupational titles of management, business and financial operations, computer and mathematics, architecture and engineering, sciences, community and social services, legal, education and library, arts, media, design, sports, healthcare practitioners, and sales.

Want to see for yourself? All our data are on GitHub.

Copyright © 2020 Data, Et cetera