2020 Considered: Part 1, the Results

In the first of three posts on the fallout of the 2020 election, I consider the results themselves.

Magnitude of the Win

The final results are still not quite in (New York again sets the standard for failure to tally ballots), but we know Biden won 306 electoral votes, will have won more than 81 million votes overall, and achieved a margin that will end up somewhere between 4.5–5% . That’s a good win but no landslide and as a result, partisans on both sides have been trying to characterize it as more or less than it was. Let’s look at it in historical perspective. Taking elections from 1968 forward (14 in all, the modern era of elections), here’s where Biden’s performance places him.

Electoral College (538 total)
270–300: 3
301–350: 4 ← Biden
351–400: 3
401–450: 1
451–500: 1
500–538: 2

Popular Vote Margin
-2.1 to 0%: 2
.1 to 2%: 2
2.1 to 4%: 2
4.1 to 6%: 2 ← Biden
6.1 to 8%: 2
8.1 to 10%: 2
10% and over: 2

A couple other notable facts. Both candidates won 25 states exactly, yet the popular vote margin, with a million votes outstanding, is around 7 million. Much has been made of the widening gap between rural and urban voters, and this is a further example. Nevertheless, had 44,000 people voted differently in Wisconsin, Arizona, and Georgia, the result would have been an electoral vote tie, 269–269. Finally, the GOP has been remarkably consistent in the six elections this century. In five of those, they received between 45.7 and 47.9% of the vote. Only once did they exceed that window, in 2004, when an incumbent president was running during wartime. (The Dems, for their part, have never won fewer than 48% of the vote in any election this century.)

Finally, despite the incredible loyalty some Trump voters exhibit, he did underperform relative to other Republicans. He will end up losing to Biden by around 4.5%. House Republicans, by contrast, received just 3% fewer votes collectively than Dems. It’s not a huge disparity, but Trump being Trump hurt his candidacy. (Romney got a larger percent of the vote than Trump did in either of his elections.)

 

Analysis, extremely Briefly

As final results emerge, we’re getting a more granular sense of how the 2020 election differed from 2016. The upshot: not very much. Educated, particularly suburban whites shifted a bit toward Biden, the gender gap widened, and some Latino and Black men voted Trump this time. Some of these patterns may reveal minor truths about the electorate. (I’m particularly interested to see how toxic masculinity may have helped Trump.)

But 2020 isn’t about White suburban trends in the exurbs or the fracturing of the Latino vote. Rather, the election revealed a major truth about voters in the United States. An incumbent president who grossly mishandled a raging pandemic, who was impeached for uncontested abuse of power, who lied at a staggering rate while shattering norms of governance and behavior, who didn’t bother to hide his visible corruption, and who shared his scheme to steal a national election months in advance managed to get more votes than any other candidate except Joe Biden in an American election. In the end, 75 million Americans voted for Donald Trump because he was a White ethno-nationalist authoritarian. Trying to parse the meaning of the election in any terms that fail to reckon with this disturbing reality mean little.

 

Mean Reversion and State Leans

2020 did reveal how states continue to change — or not. Some states, like Florida and Ohio, seem to be swing states yet turn out, in the recent realignment, not to be. Others, like Georgia and Arizona, have really moved. Still others — Pennsylvania and Wisconsin — may have become true swing states. Have a look at the results in 2000, 2012, and 2020:

• Georgia. 2000: R +11.7%, 2012: R +6.2%, 2020: D +.2%
• Florida: R +.1%, D +2.8%, R +3.3%
• Wisconsin: D +.2%, D +14.1%, D +.7%
• Arizona: R +6.3%, R +9%, D +.3%
• Pennsylvania: D +4.2%, D +6.4%, D+1.2%

 

Obama muddies the water in these trends a bit because his wins were substantial, big enough to push him to a win in FL (which has only voted D when he was on the ticket), and run up the score in WI and PA. No idea what was happening in Arizona in 2012, honestly. But following that election there was a lot of talk of abandoning it by Dems — which goes to show how these things are hard. Instead, Arizona moved left in 2016 and again in 2020, following Nevada’s pattern of turning blue. Georgia, similarly, moved steadily blue throughout the century. Pennsylvania and Wisconsin seem to be center-left but up for grabs. North Carolina, incidentally, seems to fall somewhere between Georgia and Florida — moving left, but very slowly.

Since states tend to follow national results, we would expect to see some swing in elections where the winner had a big majority versus toss-up elections. A state that typically votes +3 for the GOP might balloon up to +6 in a wave election for a GOP candidate, or be a toss-up in a wave election won by a Dem. In non-wave years, we would expect mean-reversion (that is, such a state returning to +3 GOP). The states like GA and AZ that continually move in one direction illustrate real change. (Worth noting than in a national election in which Biden won by only +2, we’d expect him to lose GA and AZ).

There’s some of this in terms of voting blocs as well. Dems have been winning enormous percentages of Black and Latino voters. That Biden won “only” 88% of Black voters and 70% of Latinos may just represent mean-reversion. In 2016, Trump only won 18% of Latinos — a number without historic precedent — so his 27% in 2020 isn’t so surprising. It’s exactly what Romney won in 2012.

 

Polling Errors, Trump Voters

Pollsters had yet another bad cycle, and it looked a lot like 2016. The error varied, and in some states polling was quite accurate. But again, the error was all in one direction, underestimating Trump’s strength. In order to account for the disparities here, we can’t look at structural problems with polls like voter access. If those were the culprits, we’d see similar errors across the board. The starkest examples of the weirdness were the outcomes in Minnesota and Wisconsin.

• Minnesota. Polls showed Biden up 9.2%. He won by 7.1%.
• Wisconsin. Polls had Biden up 8.4%. He won by just .7%, an even larger error than in 2016.

These two states are very similar in nearly every way. Minnesota is the more educated of the two, but Wisconsin’s voters are more educated than voters nationally. It’s hard to see how that accounts for the differences. So what gives?

The 2020 vote was gigantic — the largest in a hundred years. Two things appeared to happen. Biden managed to flip a lot more Trump voters than vice-versa. However, Trump again won a majority of the late-deciders and seemed to get a lot of non-voters to cast ballots. If so, two factors contributed to the error. First, pollsters’ likely-voter screen missed Trump voters. Second, pollsters’ weighting failed to account for Trump voters.

In two elections, Trump has reversed a long-standing dynamic that once favored Democrats in which low-propensity voters are induced to cast a ballot in big elections. Those same voters didn’t show up in 2018 when Trump wasn’t on the ballot, and polls then were a lot more accurate.

A final, plausible explanation is that GOP voters, whom Trump fed a consistent message that polls are fake, disproportionately refused to participate in them.

 

The Weirdest, Trumpiest Election

Despite Trump’s incredibly transgressive, unprecedented term, his approval never escaped a narrow window. Neither events nor other actors affected his approval at all. Americans knew who Trump was and either liked him or didn’t. No amount of chaos or accomplishment was going to change that. It doesn’t seem possible that something as profoundly destabilizing as a pandemic would have no affect on the election, but it didn’t — or not much. Given how close the election was, small changes might have altered the outcome. Had Trump handled the pandemic better, he might have pulled out a repeat of 2016. It barely mattered who the Dem was, though it’s possible to imagine one tipping the scales to Trump just enough. But effectively, nothing Trump could do would lose him the support of about 46% of the electorate. After it was all done, he got 46.1% against Clinton, and he’ll end up getting 46% and change against Biden.

* * *

This is the Trump age in the smallest of nutshells: the American people knew exactly who Donald J. Trump was, and 46.5% liked what they saw.