Why Trump's Winning

Just the boring material cause, though, not the efficient or final causes.

Yes, I’m still typing the second part of the Electoral College post! Sorry! It’ll be out soon! Darn Supreme Court, it threw off my groove! However, short post today, in part to prove I can still write short posts.

---

Whatever you think of him (and you know what I think of him), there’s no question that, today, right now, Donald Trump is ahead in the 2024 presidential election. Trump is the first Republican presidential candidate since 2004 to lead in the polling average for more than a week or so.¹ And Trump hasn’t just led for a week.

He’s led for five months.

Bush did that in 2000, I think,² but, since most people don’t start paying attention to politics until at least their sixteenth birthdays, some quick back-of-the-envelope math tells me that about 40% of the electorate does not remember a time when a Republican consistently led in the presidential polls. That includes me! I started getting really into politics during Recount 2000. (Although I do remember the lockbox.) It feels very strange to see a Republican actually ahead. It is even stranger for that Republican to be Donald Trump.

Better still for Trump, he seems to be leading by a slightly larger (or at least more stable) margin in the swing states which will decide the election. In other words, once again, the Republican has lucked into the electoral college advantage. The electoral college advantage does not consistently favor one party over the other (it favored Obama in 2008 & 2012), but, in close races, it matters. This looks close.

Of course, a polling lead in March is hardly the end of the election. Presidential polls in March miss the final November result by an average of 8 points (though only 3.5 points in the past 3 presidential elections). Even the final polls in November are off by 2 points on average (2.4 points in the last 3 presidential elections).³ Trump leads today by only around 2 points. In other words, Biden is just a normal polling error behind Trump (a phrase that fills 2016 progressive and 2022 conservative veterans with dread). If the election were being held tomorrow, Trump would have about a 70-75% chance to win. (Back-of-the-envelope math alert!) Since the election is actually being held in eight months, the polls today indicate Trump has maybe a 55-60% chance of winning the election.⁴

Nevertheless, Trump is ahead today, and that is interesting in itself, since it hasn’t happened in a generation.

For a lot of people, it’s pretty confusing. Heck, it’s confusing to me! How can Trump have gained on Biden? I’ll bet we all know someone who voted for Trump and is sticking by him today. Call this Trump’s base. I’ve got plenty of friends in this camp. We also all know someone who voted for Trump once or twice but who refuses to do it again (often because of January 6). Call these the Trump-Biden voters. And we all know people who have opposed Trump implacably from the get-go. Call that Biden’s base. I’ve got plenty of friends there, too.

But who the heck are these people who’re changing their votes from Biden to Trump? Wouldn’t these people have to believe something like, “Boy, Trump was bad news in 2020, but, 91 criminal charges and a failed insurrection later, choo choo all aboard the Trump Train!”? What would be the logic here? Plus, logic or not, where are they? The polls imply that these people exist, but I’ve never met one! If discussions with my readers are any indication, neither have you!

The weird fact that there must be Biden-to-Trump voters, but we don’t know any, is leading people in various places (e.g., reddit, my living room) to generate all kinds of theories that kinda sorta maybe fit our experience: Shy Trump Voters! Inflation! Elections are always a referendum on the incumbent! Something something Israel! Elections are overdetermined, so all these theories probably have some truth to them.

However, the main reason I don’t know any Biden-Trump voters seems much simpler.

It’s ‘cause I’m White.

Joe Biden won the 2020 popular vote by a convincing 4.5 points. He is now losing by ~1.7 points. We can examine Biden’s 2020 supporters by looking at Pew Research’s “validated voter” study of 2020. We can see his 2024 supporters using the March 2, 2024 poll from the New York Times/Siena, an A+ polling outfit with a track record of noticing huge trends before anyone else. Comparing Pew to Siena, we see that, between 2020 and 2024, Joe Biden…

• …has gained 1 point among Whites (-12% 2020 vs. -11% today)

• …has lost 33 points among Blacks (+84% 2020 vs. +51% today)

• …has lost 23 points among Hispanics (+21% 2020 vs. -2% today)

• …has lost 27 points among Asians/Others (+20% 2020 vs. -7% today)⁵

• …has lost ~30 points with non-Whites overall (+46% 2020 vs. +14% today)⁶

I don’t live in a political bubble. I regularly talk to, and am friends with, people all over the spectrum, from QAnons to polygamous communists. However, I did some research on myself for this post, and I was surprised and a little distressed to discover that I live in a pretty strong racial bubble. In a country that is only around 55% White, the people with whom I regularly interact are well over 90% White.

It therefore makes perfect sense for me to be confused about Donald Trump’s current standing in the polls! Among White people, Trump’s support since 2020 has gone down! Going by the Pew+Siena data, he’s lost 1 point among White people. Going by the exit polls+Siena data, he’s lost 6!⁷ The indictments and the insurrection do seem to have hurt Trump… with Whites.

Now, De Civitate doesn’t see color. (I mean that literally! I have a facial-recognition disorder that makes it difficult for me to identify people’s races. Playing Guess Who with my kids is harrowing!) Moreover, Substack doesn’t give me demographic data on subscribers. However, I am willing to lay odds that a great many of you are White and are, like me, hanging out in largely White spaces.

When I see something happening politically in White America—that is, among my friend group—I tend to assume that it’s happening in wider America, too. That’s nearly always true. After all, Whites and other races move together. For example, right after a terrorist attack, Whites, Blacks, and Hispanics all become more concerned about terrorism.

However, when they don’t move together, White opinion is just so massive that it dwarfs everything else happening in the electorate. Whites make up about 70% of the electorate.⁸ The next largest racial voting bloc, Blacks,⁹ are only 11 or 12% of the electorate. Thus, if Whites move a mere 3 points toward Candidate A—a move so slight it’s hard for polls to separate it out from statistical noise—Blacks would have to move an epochal 17.5 points toward Candidate B at the same time to offset that slight shift in White opinion. If you belong to a smaller racial group (Asian, Indian, Vulcan)? Fuhgeddaboudit.

On top of that, for at least the past few decades, Whites have been “swingier” than other racial groups. Elections during my lifetime have pretty much been about indecisive White people trying to make up their minds. Our cultural image of “undecided voters” matches that experience. Check out the SNL 2012 Town Hall Debate (parodying undecided voters), and notice how many of them are non-White (spoilers: zero). So of course we pay disproportionate attention in politics to Whites!¹⁰ Indeed, winning Whites is how Republicans have remained competitive, even though they’ve consistently lost every other group (usually by huge margins) for longer than I’ve been alive.

What we are seeing in the 2024 polls, then, cuts completely against everything we’ve experienced during my lifetime. Trump gaining 33 points among Blacks, 23 points among Hispanics, and 27 points among “Asians/Other” is stunning. A shift this large in any one of these racial categories would spell a revolutionary change in American politics—and the polls are showing it happening in all non-White racial categories at once! If these numbers hold up, every single White person could vote exactly the way they did in 2020—or even tack a little bit more toward Biden—and Trump would still be the next President of the United States.

So, naturally, the first time I saw numbers like this, I guffawed out loud. They are impossible. Some poor pollster got an unlucky sample. Clearly an outlier! Then another poll showed a similar result. Then another. At this point, all the polls are showing this. I love and trust the NYT/Siena poll, but everyone gets outliers, so, to verify it before I wrote this post, I double-checked its results against a second recent national poll, plus two recent state battleground polls.¹¹ I got the same answers. They vary in the details,¹² but the story’s the same: Trump is stronger with non-Whites overall than any Republican in at least 35 years.

…at least, he is in polls. Let’s pour some cold water on all this: It remains to be seen whether this vast non-White mobilization for Trump is real or a mirage. It’s common for a poll to miss big on some specific small subgroup (like, say, Asians, who were less than 5% of the 2020 electorate) and publish some absurd result. This is just because it’s hard to get sufficiently large samples of small groups, especially when 99% of people phone by a pollster deny the call. Sometimes you get a weird sample and absurd results. True, it’s much less common for all pollsters to miss systematically on a small subgroup, but something very like that happened in the 2015 U.K. general election. Besides, it’s undeniable that, while Republicans have done pretty well in a lot of recent polls, Democrats keep winning all the actual elections.

However, I tend to take polls at face value, as vital soundings from a portion of the country with which I have little or no direct contact. Whenever I accuse the polls, I end up having to repent in dust and ashes. So, I’m just gonna stop hemming and hawing: it’s probably not a mirage. It might be somewhat smaller than the polls say (or somewhat bigger). Its precise magnitude will probably decide the 2024 election. But I think we have to conclude that non-Whites are indeed moving quickly toward Trump.

Why, then, is non-White America suddenly embracing Donald J. Trump, especially after they rejected him in 2020? “Do they just love indicted rapist criminal frauds?” my frustrated White friends sometimes ask, before realizing they’re talking about Black and Hispanic people, biting their own tongues off, and letting themselves bleed out so that they can be buried in the White Guilt Cemetery. The softer version of this stereotype, which can be (and is) suggested in polite company, is that non-Whites, especially Hispanics, have a “machismo” culture that explains their attraction to Trump. (One study of the machismo theory failed to find evidence for it, but that hasn’t stopped the theory from circulating at parties.)

I think this is looking in the wrong place. I don’t think the non-White shift has anything to do with Trump. Look at the NYT/Siena poll numbers again. Trump puts up amazing, historic numbers with non-Whites. But Nikki Haley does even better. She not only cleans up with Whites; Haley also wins Hispanics by 10 points and gets 5% more Black support than Trump does. (She does a lot worse with Asians, though.) Haley is about as close to a Generic Republican as you can get in these weird, Trumpy times. Yet she does better with minorities than Trump. This suggests that what’s attracting minorities to the GOP isn’t Donald Trump.

It’s conservatism.

That actually makes a lot of sense. In a big thread that took Election Data Twitter by storm this week, John Burn-Murdoch of the Financial Times made a simple observation: among White people, essentially 100% of conservative voters vote for Republicans, and essentially 100% of progressive voters vote for Democrats. It has been this way since at least 2012. But, among Asian, Latino, and especially Black voters, many conservative voters have spent years voting for Democrats, not Republicans. (Non-White progressives also voted for Democrats.) That’s changing:

As the nation became more and more polarized along ideological lines in the 2000s and 2010s, there remained a reservoir of non-White conservatives who nevertheless voted for Democrats. This buoyed Democratic fortunes throughout the 2010s. However, as the Civil Rights generation (which was deeply committed to the Democratic Party) dies off, younger Black voters are getting less bonded to the Democratic party. For most Americans, young people are abandoning the Republican Party. For young Blacks, it’s the opposite:

As you can see, Black youth are still a far cry from a Republican stronghold! Yet their shift in the GOP’s direction is starting to shake the foundations of the Democratic coalition, affording even a profoundly weak candidate like Donald Trump a good shot at the White House.

Why now? Burns-Murdoch and Lakshya Jain both suggest it’s because the nation is diversifying, and racial minorities are more likely to mix socially with people outside their racial group. Racial minorities have a strong group loyalty to the Democratic Party, and their culture is thus able to “pull” even conservative minorities into the Democrats’ orbit. However, as minorities mix more with Whites and even with other minorities, that cultural “pull” weakens, which “frees” conservative minorities to defect to the Republican Party. The more people in the group ignore this cultural norm, the weaker its “pull” becomes, “freeing” more people in the group to follow their true political preferences, which in turn weakens the “pull” even more, “freeing” even more people, and so on, faster and faster. This is called a “preference cascade.” (You may know preference cascades from their previous work, the Arab Spring.) When the dust settles, we should expect Black, Hispanic, and Asian conservatives to end up voting Republican at the same rates as White conservatives, and non-White moderates to be swing voters instead of loyal Democrats. This ideological Big Sort would shatter the foundation of the Democratic Party’s current winning popular coalition.¹³

Megan McArdle further wonders (via Richard Lucas) whether sudden social changes might have spurred a sudden loosening and re-evaluation of group norms. After all, what was 2020 but a blizzard of sudden, dramatic, and lasting social changes? Covid split the nation in ways we hadn’t imagined, schools were closed sometimes for years, the George Floyd murder set off major turbulence in our racial politics. Maybe the chaos of 2020 is culminating in some non-Whites telling the Democratic Party, “I think we should try seeing other people.”

Or maybe no cause needs to be discovered. Any number of people have suggested that this cascade may have been “scheduled” to begin much earlier. After all, the rest of America polarized around 2004-2008. The only reason non-Whites didn’t follow suit (according to this theory) was because Barack Obama arrived on the scene in 2007. Perhaps, through his singular charisma and organizing prowess, Obama was able to forestall the polarization of non-Whites for a time. However, the long shadow of Obama is finally dissipating.

No doubt, the causes of this racial realignment are overdetermined, like everything else in politics, so there’s likely at least some truth to all these ideas. However, I’d like to suggest just one other possible reason for a sudden change in non-White voting behavior.

In a different (as-yet unpublished) De Civitate post, I’ve written:

Before the fall of Roe v. Wade, polling pretty consistently showed that about a third of Democrats were pro-life and about a third of Republicans were pro-choice. Since Roe, the fundamental pro-life political problem has turned out to be quite simple: the pro-choice Republicans were telling the truth, and the pro-life Democrats were lying.

Since Roe, Pro-choice Republicans have proved willing to vote for abortion ballot measures and, to a lesser extent, for Democratic candidates. Self-identified pro-life Democrats, by contrast, have not shown up to vote down abortion ballot measures, and they have not been voting for Republican candidates. In 2022, abortion likely cost the GOP about 1.5 points of national vote share (maybe a little less), and thus about half a dozen House seats. On ballot referenda, pro-lifers have suffered loss after loss, without interruption, since Roe.

Yet I think it’s interesting to note that pro-life Democrats pre-2022 were disproportionately Black and Hispanic. For example, roughly 80% of White Democrats were pro-choice in Pew’s 2020 study, but only 60% of Black and Hispanic Democrats were. The only remaining pro-life Democrat in Congress, Rep. Henry Cuellar of Texas, comes from TX-28, a 75% Hispanic district. Rep. Monica de la Cruz, likewise, became the first Republican since 1903 elected from TX-15 after the fall of Roe. The district is 81% Hispanic, and the race was not even close.

In a 2022 poll cleverly designed to reveal lying to pollsters, Hispanics and Asians were far and away the most likely to lie to pollsters about their (dishonest) support for Roe v. Wade. As every 2020 Democratic primary voter recalls, it was mainly Black voters who rejected all the more progressive (more pro-choice) presidential candidates in 2020 and put Biden on the throne. Biden, the ex-pro-lifer, the last candidate to renounce the Hyde Amendment (which bars taxpayer funding for abortion), remains clearly troubled by abortion. (Of course, it probably helped that he was also Obama’s VP!) Meanwhile, Data For Progress has put out a preliminary analysis of voters who’ve changed their mind since 2020, and it’s interesting: when you convert the percentage points into counts of people, it turns out that, yes, some voters defected from Trump because of (probably) abortion: about 15. Some voters also defected from Biden because of (probably) abortion. How many? Also 15!¹⁴

Clearly I suspect that Roe and the Democratic Party’s full-throated embrace of abortion-on-demand may have helped shake loose some non-White pro-lifers from loyalty to the Democrats. Yet, if my theory is true, where were these non-White pro-lifers in the abortion referenda and special elections over the past several years?

Well, by coincidence, every state with an abortion referendum to date has been minimum 80% White. We therefore haven’t had much of a chance to see these putative non-White pro-lifers in action. Even so, in Ohio, where Whites and Blacks are the only substantial minority groups, I notice that a Black community which tilted 91-8 for Biden in 2020 was only 83-17 for abortion in 2023. That’s in the ballpark of the 66-18 figure Biden is getting from Black Ohioans in 2024.¹⁵

Meanwhile, it’s a well-known, long-standing fact that, on average, and for a variety of reasons, Hispanic and Black adults are lower-propensity voters than White adults. In the special elections where Democrats have been cleaning up recently, the electorate is whiter than it is in a November presidential election. The Ohio 2023 abortion referendum drew 50% turnout, which is very impressive for an off-year election. However, the Ohio 2020 presidential election drew 74% turnout. Maybe we haven’t seen pro-life non-White ex-Democrats impact elections because they haven’t shown up yet.

I offer this not as a certainty, but as something to monitor. While abortion has probably won the Democratic Party some real inroads with White people, especially young White women, it may be the case that the party’s full-throated embrace of abortion has played some role in pushing some minority members of their coalition to re-evaluate their partisan loyalty. How large a role, if any, I can’t say. But keep an eye on it. If it is there, the Media will not report it, because it violates the narrative that America Loves Abortion. (Don’t look now, but the Media is largely driven by young White progressive women.)

Whatever the case there, though, it certainly seems that something is pushing minorities to re-evaluate their partisan loyalty! Not all of them are switching. Not even most. (Biden’s definitely going to win an overall majority of the Black vote, no matter what!) So far, though, enough non-Whites are moving red in large enough numbers to overcome the White movement away from Trump over the course of Biden’s presidential term.

In other words: non-Whites are the reason Trump is currently ahead.

That makes the non-White vote the most important thing to watch in this election right now. If this apparent realignment is borne out at the ballot box, and if it’s as large as the polls make it look (which… I haven’t bet any money on it, that’s for sure!), it would be the most dramatic shift in the American political landscape since the failure of Bush Conservatism and the rise of Obama. If no race is monolithically supporting one particular party anymore, a world of fascinating possibilities opens up in the American politics of the near future.

---

…shoot, that wasn’t a short post after all, was it? Blast!

1. I don’t like RealClearPolling’s dumb-as-a-brick polling average, which uses weird selection criteria, fails to take into account house effects, and is generally less reliable than the 538 polling average. However, as of 2023, 538.com is dead, the new boss running the shell of 538 is a bit of a hack, and they don’t even have their 2024 polling average online yet, so! RealClearPolling it is.

2. “I think“ because polling averages weren’t really a thing until 2004, and I don’t have time to go build one myself. (Short post, remember.) But Bush led in March and, if I recall correctly, pretty much stayed in the lead for most of the race, with Gore sloooowly gaining on him. It was close enough at the end that a few people people openly wondered whether Gore might win the electoral vote while losing the popular vote, setting off a constitutional convulsion. It was all the more shocking, then, when the exact opposite happened!

P.S. For a fun time, get in your time travel pod back to RCP’s primitive 2000 polling average, where Bush is blue and Gore is red. Until the year 2000, blue was generally the Republican color and red the Democrats’ color. This was consistent with international usage, where red represents socialist/communist/progressive parties (see also: the flags of Communist China and Communist Russia) and where blue represents conservative parties (e.g. the British Tories). However, usage was very fluid in the United States. The idea of red and blue divisions became locked in after the 2000 election, in part because that election was so close and so bitterly divided on state lines. The notion of “red states” and “blue states” (and corresponding “tribes”) soon became so analytically important that the colors will never change again.

3. Here is my data to support that. Sorry it’s an image instead of a table. Substack still does bad tables.

The 1968-2012 data is from FiveThirtyEight’s notorious article. The 2016 data and 2020 data are my calculation based on the FiveThirtyEight popular vote projection one week prior to the election. In 2016, it’s the polls-only projection. In 2020, the polls-only projection does not appear to be available, so I used the default projection (which, by a week before the race, was pretty close to polls-only anyway). I should have rounded to 2.0, but I forgot.

I think it’s best to think of the polls a week out as the “final polls” of an election, because pollster herding kicks in big time during the final week.

4. In reality, since Trump may be convicted of crimes between now and November, and since voters keep telling pollsters that they would be much less likely to vote for Trump if he got a criminal conviction, I think he has more like a 40-50% chance of winning. This is still remarkably good for someone with Trump’s baggage.

5. Pew, NYT/Siena, and the 2020 Exit Poll all divvy up Asian/Other slightly differently. Pew excludes “Other” completely. NYT/Siena treats Asians as “Other.” Only the 2020 Exit Poll lists Asian and Other separately. So I used the 2020 Exit Poll (not the Pew survey I used in the other bullet points) to calculate a weighted average of the Asian + Other vote in 2020 (which was easy, since Asian and Other were the same size in that sample), then compared that result to the NYT/Siena’s “Other” category.

In both the 2020 Exit and the NYT/Siena samples, this combined group added up to 8% of the electorate, so I feel pretty good about the accuracy of this measurement. But it was enough of a kludge that I thought I’d better footnote it.

6. As in the previous footnote, I couldn’t use the Pew survey because they didn’t include “Other Non-White” in their chart, so this figure is from the 2020 Exit Poll and doesn’t quite add up to a weighted average of the three bullet points above it like it should.

7. There are good reasons to interpret the 2020 exit polls with caution, especially since it happened during a pandemic.

8. The national exit polls said only 67%, but Pew says 72% and the Census data says 71%.

9. Hispanic voting power is very nearly equal to Black voting power, even though there are more Hispanics than Blacks in this country. If Hispanic voting power hasn’t overtaken Black voting power already, it soon will.

10. Of course, we don’t pay all Whites the same attention, or the same kind of attention. Until the 2016 election delivered Trump on the backs of Rust Belt Whites without a college education, non-college Whites were almost completely invisible to the Indigo Blob. Now, non-college Whites are a major topic of the Conversation—although they still aren’t usually present at the table during the Conversation, and the Conversation hasn’t figured out how to respect them (best it can do is pity, although it still tries scorn plenty, too).

This is also a problem, but their opinion of Trump has barely moved (+35 Trump in the 2020 exits, +32 Trump in the 2020 Pew survey, +31 Trump in the NYT/Siena), so non-college Whites do not help explain Trump’s rise in the polls between 2020 and 2024, which means they don’t figure much into this post.

Not that I can cast many stones here. I’m on the far right edge of the Indigo Blob, but I’m still very much a part of it. (My friends list is not only overwhelmingly White, but also overwhelmingly college-educated. Another bubble to work on.)

11. Specifically, the YouGov/Economist March 12 national poll, the Emerson College/The Hill 14 March poll of Pennsylvania, and the Fox News Feb 14 poll of Michigan.

12. YouGov/Economist in particular shows Trump stronger with Whites and weaker with non-Whites compared to the other polls. Nevertheless, he’s leading in their poll, and it’s largely thanks to his newfound non-White support.

13. It’s important to note that Democrats have been making major inroads into traditional Republican territory during the past few years, too. The winning Republican coalition is traditionally based on ironclad margins among college-educated suburban whites who turn out to vote in a big way. Those voters have moved decisively leftward in the age of Trump. If a Republican candidate were benefitting from the big racial shift and the level of suburban support Mitt Romney enjoyed in 2012, this race would be a blowout.

…and, indeed, if the GOP were running Nikki Haley, the polls tell us that’s exactly what would be happening. Biden would be getting blown out. Instead, we ran Trump. That’s on us, we set the bar too low.

14. Technically, multiplying the weighted N’s by the reported percentages, Trump lost 15.74 voters who rated abortion their highest priority and Biden lost 15.4 such voters. I assume that the abortion voters who defected from Trump were pro-choice and the abortion voters defecting from Biden were pro-life, but that’s just an assumption.

Don’t read too much into this. 30 people in a 12,000-person sample is just not a lot, the data released did not have a lot of precision (all figures rounded to the nearest integer), and this wasn’t quite the question they were asking anyway, so it’s risky to assume that this is the correct interpretation of DfP’s data.

Even if accurate, it would be a mistake to think that abortion is a political winner, or even politically neutral, for Trump right now. The same data shows far more abortion enthusiasm on the Left right now, which is historically weird, because the pro-life movement was more energized throughout the Roe era. Moreover, Biden is clearly benefiting from abortion voters among “New/Other” voters, where he picked up (by my math) 26.46 abortion-first voters to Trump’s 14.56. The pro-life movement needs to maintain its defensive posture for the time being. Things still seem pretty toxic nationally.

However, that doesn’t negate the likelihood that some of the political benefits Biden is reaping from his abortion rhetoric is offset by driving some fraction of non-Whites out of his coalition. The question is how large that fraction is, and how much of the larger non-White exodus it accounts for.

15. OTOH, Latinos trended toward abortion in the abortion referendum. OTOH (again) they were statistically a very small group—small enough that SurveyUSA doesn’t even have 2024 crosstabs for them. OTOH, I probably wouldn’t have added that caveat if the Latino results had supported my hypothesis, which makes me sus.