Catalist shows a durable partisan realignment
How the polls were bad in 2024, a more diverse electorate, low Dem ratings may not matter, Latino partisanship, community types, AI snitching
No. 361 | May 23rd, 2025
🇺🇲 2024
Probably the most trusted account of how different groups voted in the election, Catalist’s “What Happened” report from 2024 is now out. It confirms the broad realignment of key groups in 2024, with Hispanics shifting 18 points right, Asians 9 points, and Black voters 6 points. But on the gender gap, it contradicts the narrative from the post-election exit polls.
Both the Edison exit poll and AP VoteCast found no real widening of the gender gap in 2024, but Catalist has it ballooning to 26 points—men for Trump by 16 points, women for Harris by 10. Because everyone is in agreement that nonwhite men swung the most strongly to Trump, the biggest differences between the exits and Catalist are among whites. To take one example, where AP VoteCast had white college women swinging 7 points to Trump and white college men 2 points to Harris, Catalist has this flipped: white college men moving 10 points to Trump, and white college women 2 points.
That result better matches pre-election intuitions about the role of gender in the campaign. It shows the Harris campaign’s focus on abortion wasn’t a total loss—that is, if you don’t count the big swings against her among men. As an explanation, it’s more internally consistent: why would the gender gap widen among nonwhites, but tighten among whites? But alas, any claims about how the two genders voted are virtually unfalsifiable. Unlike race, you can’t compare them to precinct-level results because the ratio of men and women is pretty similar throughout the country. On the other hand, I buy the standout results for Hispanics, since it matches up well with precinct data.
I’ve condensed all of the 2020-24 group swings into a single big chart:
I now want to examine an overlooked part of the report: the vote for the U.S. House and how it compares to the presidential vote. Post-election, many have claimed that the Realignment doesn’t really extend beyond Donald Trump. If that were true, it would show up in the House vote, with massive underperformances for Republican House candidates among realigning groups.
The report shows this easy narrative to be largely unfounded: downballot Republicans have gotten the lion’s share of Trump’s gains with minorities, though not all of them.
Because Catalist keeps records of the House vote over several cycles, we can compare the House swing to the presidential swing. And it turns out that the 2020-24 Presidential and House swings right are highly correlated and positive:
It’s true that House Republicans did not get as big a national swing as Trump in 2024, due largely to outperforming him in 2020. And they tended to underperform more with Latinos and other realigning groups. But they still got the bulk of Trump’s gains from 2020, with a 14 point swing right in the House compared to an 18 point swing to Trump. That shows this is not a Trump-only phenomenon. And it changes the calculus in a House battleground that is heavily Hispanic.
We’re also seeing a gradual, long-term secular shift in the makeup of the GOP coalition, where nonwhites have made up more of the coalition in each successive presidential and midterm cycle since 2012. Not only were the Trump and House coalitions functionally the same in 2024: 18.2% nonwhite vs. 17.7%, but these figure represent a near doubling of nonwhite representation in the party compared to 12 years ago, when nonwhites were just 9.9% of Romney voters. All of these gains far outpace the nonwhite growth in the electorate—and they were happening before 2024.
The Cook Political Report also outlines some of Catalist’s key findings; Harris lost because of a drop in support from 2020 voters and her failure to win over new and infrequent voters.
Ruy Teixeira also argues the Realignment is here to stay — and the Obama coalition is not coming back.
“As Democrats dig out from their debacle, it’s important for them to understand just how far away they now are from the salad days of the Obama coalition. In 12 short years, they have lost two of three elections to Donald Trump and huge chunks of support from key demographics, including most of their rising constituencies. They need to face the uncomfortable fact that not only did the Obama coalition not come back, it’s likely never coming back.”
Analysts at Sabato’s Crystal Ball compared Catalist’s findings with the two exit polls across a range of different demographics.
The Atlantic’s Marc Novicoff spotlighted the polling industry’s repeated underestimation of Donald Trump and noted some experts’ opinions on how to overcome the nonresponse bias that can taint pollsters’ work.
“This is where some of the art and science get a little mixed up,” Michael Bailey, a Georgetown professor who studies polling, told me. If you weight a sample to be 30 percent Republican, 30 percent Democrat, and 40 percent independent—because that’s roughly how people self-identify when asked—you are making an assumption about how the three groups will behave, not merely matching a poll to population demographics such as age, gender, and education.
These assumptions vary from pollster to pollster, often reflecting their unconscious biases. And for most pollsters, these biases seem to point in the same direction: underestimating Trump and overestimating his opponent. Most pollsters, like most other people in the expert class, are probably not huge fans of Trump,” the election-forecasting expert Nate Silver told me. This personal dislike may not seem to matter much—after all, this should be a science—but every decision about weighting is a judgment call.”
The Working Class Project, a large-scale Democratic project aiming to reconnect the party with the working-class, released findings showing Democrats still in trouble among Hispanics in South Texas.
“It’s important for Democrats to listen to voices like those we heard in the Rio Grande Valley. They help illustrate the challenges the Democratic Party is facing with working class Latino voters, who continue to feel like the party doesn’t understand them and isn’t prioritizing the issues they care most about.
On immigration, the voters we heard from in McAllen viewed the issue not through a lens of bias or prejudice, but through worries about their own economic security and a sense of fairness. Democrats clearly need to re-examine their position on immigration, with stronger emphasis on fairness, border security, and real reform. On the economy, these voters feel squeezed – like many other working class voters we’ve heard from across the country – and they think Democrats are prioritizing what they see as niche and liberal social and cultural issues over real ideas to make life more affordable. Focusing on clear, simple ideas like raising the minimum wage could help earn back trust.”
Tom Wood uses 2024 ANES data to illustrate the continued realignment among whites by income, an update of a chart I cited in Chapter 1 of Party of the People.
Michael McDonald’s analysis of Census Bureau data found that the 2024 electorate was slightly more diverse than the 2020 electorate, despite declining turnout rates in minority communities.
🇺🇲 2026
Two partners at Split Ticket argue that Democrats’ current negative ratings may not matter much in the midterms, because the electorate has a tendency to sour on the incumbent party.
📊 Public Opinion
Research from Pew found that most Americans think numerous ethnic and racial groups face discrimination, although that number has declined over the last year.
👫 Demographics
The Liberal Patriot published an analysis of the 15 community types that compose the United States, as defined by The American Communities Project.
A new study on Latino partisanship confirms a move away from Democratic identification, to the benefit of Independent and Republican support.
🗺️ Data Visualization
The shift to a service economy and away from manufacturing hits all economies in the same way, as a function of output per worker.
🤖 Artificial Intelligence
Claude 4 can catch companies engaging in unethical behavior and report them to the authorities. (This was a real-life testing scenario at Anthropic, though it didn’t ship.)
Great analysis here! Very concise data to share with some of my friends.
FWIW: The Party That Cannot Tell You What A Woman is, Mad Men Will Not Vote For Them. https://torrancestephensphd.substack.com/p/the-party-that-cannot-tell-you-what