The Demographics of a Trump Victory—Or Defeat

A peek underneath the hood.

Next week, the election campaign will (finally) be over. It’s still basically a coin toss race. But it’s also clear that Trump is in a reasonably good position to win. The national margin for Harris has narrowed significantly; in Nate Silver’s average, it’s gone from 3.4 points around the time of the Walz-Vance debate to 1.1 points today. The same pattern can be seen in Silver’s swing state averages, where Harris’s margin has declined or Trump’s margin has increased over the time period. Or, in the critical case of Pennsylvania, flipped from a narrow Harris lead to a narrow one for Trump.

All this is good for Trump, even leaving aside the possibility that the polls are underestimating his support as they famously did in 2020 and 2016. And Silver’s forecast model currently leans slightly toward Trump (other models are closer to a flat 50-50 assessment). But as Silver himself emphasizes, even a 55-45 probabilistic assessment for Trump is closer to a coin toss than what people traditionally think of as a “favorite.”

So that’s where we are. How did this happen? How did Trump, widely-loathed and dramatically flawed candidate that he is, wind up with a coin-toss chance of winning his second presidential election? Put another way, how are Democrats falling short not just of recreating the Obama coalition but even the Biden coalition of four years earlier?

Examining the current demographics of the Harris coalition and comparing them to the demographics of Biden’s 2020 coalition provides a window into understanding how Trump has positioned himself for a possible victory. Here are four key points of demographic comparison, using the gold standard Catalist data from 2020 and crosstabs from the New York Times/Siena survey (rated A+ in Silver’s pollster ratings) and from the running demographic averages maintained by Cook Political Report (CPR).

(1) It’s still a working-class election. As I have previously noted, the key demographic to keep track of is the working class (noncollege) vote. How these voters move will likely determine the outcome of the 2024 election. They will be the overwhelming majority of eligible voters (around two-thirds) and, even allowing for turnout patterns, only slightly less dominant among actual voters (around three-fifths). Moreover, in all seven key swing states—Arizona, Georgia, Michigan, Nevada, North Carolina, Pennsylvania and Wisconsin—the working-class share of the electorate, both as eligible voters and as projected 2024 voters, will be higher than the national average.

In 2020, according to the Catalist data, Trump carried working-class voters overall by four points but lost college-educated voters by 18 points. In the latest Times survey, Trump is carrying the working class by 13 points, a 9-point improvement over his 2020 performance. He is also losing the college-educated by 21 points, representing a 3-point slippage relative to 2020. But the net of these two changes is clearly in his favor. Indeed, given the preponderance of working-class voters in the electorate, to truly set off widening deficits among the working class Democrats would need margin gains among the college-educated that are 50 percent larger than their margin losses among working-class voters. That is not happening.

The Times data allow working-class performance to be broken down between whites and nonwhites. Among whites, Trump is carrying white working-class voters by 30 points, a 4-point improvement over his already-large 26-point margin in 2020 but losing college whites by 16 points, a 7-point deterioration relative to 2020 (CPR data show the same pattern but more muted). White college graduates are the major demographic where Democrats have consistently improved election-over-election since 2012. They look set to do the same in this election. If Harris is, in the end, able to overcome deteriorating working-class support it is likely to come from spiking support among these voters.

Looking at nonwhites, it is here that declining working-class support is most dramatic. Among nonwhite working-class voters, according to the Times data, Harris is currently leading by 26 points. That may sound like a lot but Biden carried these voters by 49 points in 2020. And Obama carried these same voters by 67 points in 2012! Thus Harris is running an astonishing 41 points behind Obama among nonwhite working class voters, an absolutely core demographic for Democrats.

As I have previously observed:

Since the latter part of the 20^th century, the left has had a plan. Well, not really a plan, it just kind of….happened. Call it, to use Thomas Piketty’s term, the Brahmin Left. That is his characterization of Western left parties increasingly bereft of working-class voters and increasingly dominated by highly educated voters and elites. The Brahmin Left has evolved over many decades and certainly includes today’s Democratic Party…

For Brahmin Left parties, the temptation is great to lean into their emerging strengths and just hope they can retain enough of their working-class base to make the political arithmetic work. That is the natural inclination of the elites and activists who now dominate the parties. But these parties have been increasingly battered by right populist competitors who are bleeding off more and more of the left’s working-class support. That calls the viability of the Brahmin Left model into question. There is a point beyond which the loss of working-class voters cannot be plausibly balanced by increased support among college-educated and professional voters and the model is fatally undermined.

We shall see if this is the election where that model finally breaks.

(2) The rise of racial depolarization. There was already significant racial depolarization in the 2020 election, where Trump substantially improved his performance among nonwhites, especially Hispanics and, as noted above, the working class. We are seeing more of that in this election cycle, which should net out to Trump’s advantage. The Times data show Trump doing slightly worse among whites as a whole (an 11-point lead vs. 12 points in 2020) but much better among nonwhites (a 28-point deficit vs. 48 points in 2020).

Consistent with this, the Times data finds Trump trailing among Hispanics by just 10 points, 13 points better than his 23-point deficit in 2020 (which, in turn, was down from 39 points in 2016). Among blacks, Trump is being thumped by 69 points—but that is actually 12 points better than his 81-point gap in 2020. The CPR averages confirm this general pattern of gains from 2020, with Trump doing not quite as well among Hispanics but better among blacks.

(3) A declining age gap? A persistent feature of this election cycle has been relatively good performance for Trump among younger voters. In the Times poll, Trump is behind by 12 points among voters under 30 (this is identical to the CPR running average), an 11-point improvement over his 23-point deficit in 2020.

It is interesting to note that the 18-29 year old age group is now essentially a Gen Z group. Millennials are now almost all in the 30-44 year old age group. And here also Trump seems to be doing better. He is behind by only 5 points among this Millennial-dominated age group, compared to a 14-point deficit in 2020.

But among those 65 and over—now heavily dominated by Baby Boomers—Trump seems to have lost some ground, though not drastically. In the Times data, Trump is behind by a point among these voters, compared to a 4-point lead in 2020. The CPR average is slightly better for him, giving him a half-point lead, but still indicating a fall-off from 2020.

No matter which way you look at it, the data do seem to indicate a declining age gap. Comparing 18-29 year olds to those 65 and over, the age gap in 2020 was 27 points. Today in the Times data, it’s 11 points (13 points in the CPR data).

(4) A widening gender gap? There has been much talk about a widening gender gap in this election as women flock to support Harris’s candidacy and men seem to move toward Trump. This has generally been interpreted as a factor that favors Harris but that does not appear to be true. The simple math of a widening gender gap is that its political effect is determined by the relative movement of women and men in widening that gap.

In this case, the movement of men toward Trump is widening the gap not the movement of women toward Harris. In the Times data, Harris is carrying women by 12 points, actually slightly less than Biden’s 13-point advantage in 2020. But Trump is carrying men by 14 points, 8 points better than his 6-point advantage in 2020. Thus the gender gap has widened from 19 points in 2020 to 26 points today but this is entirely due to Trump doing better than before among men not a surge of support for Harris among women.

The CPR averages tell a similar story: a 12-point lead for Harris among women and an 11-point lead for Trump among men. Again: a widening gender gap (23 points) but entirely driven by increasing support for Trump among men. Indeed people seem to have forgotten that a key to Biden’s victory in 2020 was doing better among men while holding Clinton’s advantage among women. In that election, the gender gap was compressed but benefited Democrats. In this election, we may see the reverse, a widening gender gap that benefits Republicans.

This may seem strange in light of the extensive media coverage of very high margins for Harris among younger women, who appear to be leaning ever more strongly to the left. But it is not clear that women under 30 are giving margins to Harris that are much bigger than those they gave to Biden in 2020. And there is significant evidence that men under 30 may be poised to vote much more pro-Trump this election than they did in 2020. In any event, however the trends net out among young men and women they do not appear to be enough to change the overall story of a widening gender gap driven not by a pro-Democratic trend among women but rather a pro-Republican trend among men.

---

These data make clear how Trump may win. However, they do not mean he will win. As noted, the race is still basically a coin toss. If Harris wins, it would be no surprise if some of the demographic trends noted above turned out to be more favorable to her than they currently look: less deterioration among the working class; even higher support among white college graduates; a return of black and Hispanic support margins to close to 2020 levels and so on. This in turn could be driven by perhaps the Democrats’ best hole card: turnout. An excellent article by Nate Cohn lays out the contours of the Democrats’ potential turnout advantage:

As we’ve reported all cycle, Democrats excel among high-turnout voters, while Donald J. Trump is strong among relatively low-turnout voters. He’s made his biggest gains among low-turnout demographic groups like young men and nonwhite voters….but almost all of that strength is contained among those who sat out the midterms.

This is not simply about education: Even the college graduates who sat out the midterms were far likelier to say they backed Mr. Trump.

Of course, just because Mr. Trump leads among irregular voters does not necessarily mean he will win the irregular voters who decide to show up. In the midterms, Democrats managed to draw a disproportionately Democratic group of voters out of the pool of voters who didn’t vote in primaries. This time, it’s possible they could draw a disproportionately Democratic group out of the Republican-leaning pool of those who didn’t vote in the midterms.

Imagine, for instance, that the infrequent Black or young voters who say they back Mr. Trump in the polls generally don’t show up, while those who back Ms. Harris really do come to the polls.

This is a plausible story about how Trump may lose. But it does not mean he will lose. That will be determined by, as they say, the only poll that really counts. Stay tuned and don’t forget to vote.