At the time of writing, the U.S. presidential election will continue to count, and hopefully it will continue until every vote is counted, despite President Donald Trump’s incomprehensible attempt to prevent it. While it looks like Joe Biden is about to wrest the presidency from Trump, Democrats will nonetheless be disappointed. The victory will be less forceful than the polls suggest (although the national polls confirm the leadership lead better than many state polls in swing states). And most importantly, even if he loses, Trump has increased his share of the vote compared to 2016. This wasn’t the overwhelming rejection of Trumpism that its opponents had hoped voters would deliver.
It is all the more interesting to examine who voted and how, and to try to gain insight into what influences voting behavior. Let’s take a look at the votes while we wait for the votes.
Quit polls should now be treated with caution – even more so this year. At best, exit surveys are difficult to correct and can be replaced with more thorough but less immediate surveys. This year’s record of early votes and mail-in votes makes it even more difficult to get a representative sample of voters. On the other hand, it could be argued that changes in survey responses from 2016 to 2020 are less biased by issues that affect the initial query at any given time. Even so, all of the dates given below come with a strong health warning.
However, what preliminary picture can we draw from the data from the 2020 baseline survey if we compare it with the (now much better) data we have on voter behavior in 2016?
Some observers saw a sign in the initial polls that the polarization is becoming less pronounced. The finding on which this interpretation is based is that within demographic groups of voters, the percentage point spread between the Republican and Democratic vote has shrunk. As can be seen in the table below from my FT colleagues, the support share for each of the two major parties has moved closer to 50-50 in most voter groups, defined by gender, ethnic or educational background.
There are other ways to crop the voter data that shrink the margins. In the polls on the exit reported by the New York Times, the urban-rural split appears to have narrowed somewhat from 2016 to 2020. Hillary Clinton’s 24-point advantage in cities with populations over 50,000 fell slightly to 23 for Biden; Trump’s lead in rural areas fell from 28 to nine points. Trump has also lost almost all of his leadership among male voters.
All of this shows that the voting groups are becoming less one-sided. However, this does not mean that voters are less polarized.
First, because the balance within the constituencies is not universal. For example, according to CNN’s analysis of the same output elections, young voters have become overwhelmingly more democratic than they were in 2020. (But Trump’s advantage among the oldest voters has all but disappeared, possibly due to his handling of the pandemic.) In terms of income, the democratic lead has increased strengthened among low-wage earners, while Trump’s narrow advantage among the highest paid has grown to a lead of 11 points in 2016.
We can also look at the actual voting results of the states in which almost all the votes were counted. That shows a shift toward Democrats in many states that Trump won in 2016 – but that shift appears to be larger on average in the states Clinton won last time. So this is at best a one-sided realignment.
Second, groups of voters could in principle become more balanced at the national level, but more polarized at the local level. For example, in the three battlefield states of Florida, Georgia, and North Carolina, surveys show that the rural advantage of Republicans has increased compared to 2016.
Finally, however, the bigger point is that realignment does not in itself mean a less polarized electorate. My colleague Ed Luce points out in his Swamp Notes exchange with Rana Foroohar (sign up to get Swamp Notes in your inbox): “We may be shocked that after all this happened, about 48 percent of America voted for Trump , especially on Covid. But again they would be shocked that 52 percent of America would vote for what they see as aging prisoners of the left. If that choice has taught us anything (so far), then Americans live in separate universes. “
All that can happen is that these separate universes now split each voting group closer in half, while in the past some groups flopped more collectively for one or the other. This seems to be the reason, for example, for Trump’s entry into the “Latino vote”, as my colleague John Paul Rathbone describes. Political tribalism is becoming a separate dimension of polarization rather than just exacerbating old divisions – but I don’t realize this is something to celebrate.
A recent study by Andrés Rodríguez-Pose, Neil Lee, and Cornelius Lipp of the LSE has a finding that is useful to ponder. They note that the crucial electoral mobilization for Trump in 2016 came from places that combined long-term economic decline with strong social capital – that is, high levels of social cohesion. Regarding the influential thesis of political scientist Robert Putnam that Americans are less closely connected and increasingly “bowls alone,” the authors write: “Places in the United States that have remained connected but experienced permanent decline are no longer alone Bowling, you play golf with Trump. “
What does social capital help you? It helps you develop a sense of shared identity – and especially one that comes with economic adversity for the group you belong to, if not yourself. With that in mind, it shouldn’t be that surprising Trump has managed to consolidate his support in the face of the economic crisis. Indeed, an economically threatening environment can help him expand the identity with which his constituents seem to identify. The lesson for the Democrats is that identity campaigns are well suited to those that can appeal to the largest identity-based groups. You can do better by leaving identity politics behind and focusing fully on politics.