Trump’s numbers are not what you think

It’s the myth that just won’t die; this week’s election has only given it extra life. Donald Trump, everyone says, is a “disruptor”, a man with exceptional electoral impact, who has rewritten the political map and won a crushing victory. No-one wants to dispute it: the media love it because it sells, the right have stopped caring about truth and falsity, and the left are revelling in their own masochism.

But it simply isn’t true. You can start by reading two previous exposures – this one from the 2016 election and this one four years later. In this post I’ll try to update the argument in the light of this week’s results.

Most of this week’s punditry, particularly the amateur sort that populates my Facebook feed, has been woefully misleading as regards numbers. There are four interrelated problems:

• Because in some states it takes a long time for all (or even almost all) the votes to be counted, people are working with incomplete figures.
• Because there is no centralised tally, people are relying on different sources, which, being updated at different times, will sometimes disagree.
• Because the minor party candidates have no effect on the result (unlike in Australia, where preferences are counted), people don’t know what to do with them; logically they should be factored out before calculating percentages, just like the informal vote, but most pundits follow American practice and leave them in.
• For no obvious reason, except perhaps that they’ve been bamboozled by the previous three points, people often use raw numbers rather than percentages, which obscure comparisons and (perhaps deliberately in some cases) spread confusion.

The first problem is the most serious. The slow-counting states are not a random sample of the whole, and late counting doesn’t always go the same way as early counting, so the figures are not just incomplete but unrepresentative. You can see, for example, in the 2020 post that I cited above, that at that stage of the count Joe Biden had about 51.5% of the two-party vote, but he actually finished with 52.3%.

This year has been shifting the same way. As of midday today our time (8pm Thursday in New York), the Associated Press tabulation has Trump on 51.54% of the two-party vote, a lead of a little under 4.4 million votes – already noticeably down from the widely quoted figure of five million. Eight states are reported to have more than 15% of their vote still to count; two of them are small Trump-voting states (Alaska and Utah), another is Arizona, a swing state narrowly won by Trump,* but the others are all medium-to-large states being won by Kamala Harris.

The largest state, California, is also (as usual) the slowest to count and one of the safest for the Democrats. With only 58% of its vote so far counted, if the remainder flows as strongly to Harris she will net about another 1.35 million votes. Nothing else will have anywhere near that impact, but Maryland and Washington should each net her more than another 100,000, Colorado and Oregon a bit less.

On my rough calculation that would bring Trump down to around 50.9% of the two-party vote. Let’s call it 51-49 as a round figure.

That sounds close. Indeed it is, but as I’ve pointed out many times, close elections have become the norm in America. Here are the results of the popular vote from the last seven presidential elections:

Year	Winner	Vote (2-party)
2000	Gore	50.2%
2004	Bush	51.2%
2008	Obama	53.7%
2012	Obama	52.0%
2016	Clinton	51.1%
2020	Biden	52.3%
2024	Trump	51.0% (est.)

Only 2008 could be called at all decisive, and even that was narrow by historical standards. Yet the media give a completely different impression: this year’s swing of around 3% is being presented as something huge and unprecedented, whereas it is actually quite modest. (In the 1960s and ’70s, four elections in succession produced swings in double digits.)

The electoral college of course is a different story, at least to some extent. Of those seven winners, two failed to take the presidency due to the vagaries of the college (and, in Gore’s case, the Supreme Court). Trump will have 312 electoral votes to Harris’s 226, slightly better than the 306-232 that Biden won with last time. But that too is very close historically: in the twentieth century, presidents frequently racked up more than 400 in the college (three of them hit 500).

On a uniform swing, the Democrats will require a swing of about 1% to retake the presidency in 2028 (Pennsylvania will almost certainly be the median state). Hardly an impossible target, yet on that basis they are now being urged to overturn their whole approach to politics – just as the Republicans were four years ago, with equally little statistical reason. (Fortunately for them, if not for the country, they ignored the calls.)

Not only have recent swings been very small, but (as I pointed out on Tuesday) they have been very uniform. Again, Trump has in no way disturbed the pattern. We’ll look at this in detail when counting has finished, but the six states he picked up were exactly the ones he would have on a uniform swing, and there are no big outliers elsewhere on the pendulum. Voting is continuing to gradually realign along (broadly speaking) educational rather than class lines, just as it’s been doing for more than two decades. There’s no evidence that the result would have looked different if, say, Nikki Haley had been the nominee, although presumably her margin would have been a bit larger.

None of this is to say that Trump is not important: he has changed the character of the Republican Party and through it much of the national conversation. He may yet do so in more dramatic ways still. But in electoral terms he is just another Republican, and a rather unpopular one.

.

* AP and some other sources have not officially called Arizona or Nevada (which has 91% counted), but it is clear that Trump has won both of them.