Last week I mentioned that there were something like a million and a half votes still to be counted in New York. Those results are now in, and as expected they ran very strongly for Joe Biden. His lead over Donald Trump grew by about 200,000 votes in Manhattan, 150,000 in Brooklyn, 125,000 in Queens, and he erased a 74,000 vote deficit in Suffolk county.
There are apparently some more to count in Maryland, also a safe Democrat state, and possibly a few stray parcels elsewhere, but at most we’re talking tens, not hundreds, of thousands. For practical purposes the election is over. [See addendum below]
Biden finishes with a little over 81 and a quarter million votes, seven million ahead of Trump – the largest two totals on record. The gap between them on the raw (i.e. not two-party) vote is 4.4 percentage points: last month, when that figure was more like two points, Nate Silver projected that it would finish at 4.3. I thought it would get above five, but I was wrong and he was right.
In two-party terms, Biden has 52.3% of the vote, an improvement of 1.2 points on Hillary Clinton’s 51.1%. That’s the second-smallest swing recorded since the nineteenth century.
I’ve made this point before, but the ossification of American voting patterns is so striking that’s it’s worth making again. In the 32 years to 1992 there were nine presidential elections; only one of them (1984) produced a two-party swing of less than five points. In the seven elections since then, not once has the swing been more than five points: 4.9% in 2008 was the biggest.
In what now seems like the remote past, it wasn’t unusual for a candidate to win with more than 58% of the two-party vote. Three Republicans in succession (Harding, Coolidge and Hoover) did it in the 1920s; Franklin Roosevelt did it twice; Lyndon Johnson, Richard Nixon and Ronald Reagan did it once each. But no-one has won more than 54% since Bill Clinton in 1996, and even that was a quite modest 54.7%.
Regional differences seem to be disappearing as well. The movement to the Democrats varied very little by region: a median swing of 1.2% in the midwest, 1.5% in the south, 2.5% in the west and 2.8% in the north.* Compare 2016, when the median swing to Trump was 4.9% in the midwest but only 1.1% in the west.
This is bad news for pundits. Volatility sells; they want us to think that voters are highly sensitive to passing trends, making up their minds at the last minute and swinging decisively one way or the other. But the evidence is quite the opposite: that the large majority are set in their ways and that even the appearance of the most unorthodox Trump was not enough to unsettle them.
That fact is all the more remarkable given the change in the number of people voting. Turnout in 1996 was only just over half (51.7%) of the eligible population; this year it is estimated to hit two-thirds. Yet it seems to make no net difference. The story of voting behavior can pretty much be told without any reference to turnout at all.
But no-one knows how this story ends. Will something happen to shake the system out of its rigidity? If so, what could it be? Or will every administration just have to live with the fact that almost half the population will be hostile, whatever it does – and can democracy really cope with such a state of affairs?
* I’m counting Delaware and Maryland as north, Kentucky and West Virginia as south and Missouri as midwest, but changing them around makes hardly any difference.
ADDENDUM Thursday morning: Overnight the New York Times upped its Maryland result from 96% counted to 100% counted, without actually adding any new figures. So Biden’s 81.28 million can for all practical purposes now be taken as final.