What the heck just happened?
An iron law of off-year turnout, the Jersey Hispanic vote, low-propensity voters are obsessed with affordability, Dems win economy voters 2-to-1, NYC divided, how AI is ruining the job market
No. 381 | November 7th, 2025
🇺🇲 Turnout and persuasion in Tuesday’s Democratic wins
Republicans got walloped in Tuesday’s off-year elections. There are two stories here: a turnout story and a persuasion story.
In Virginia, red counties dropped off more compared to their turnout in 2024, while blue counties dropped off less. In 2021, it was exactly the opposite. In fact, the trendlines from both years look like mirror images of the other. This complicates the story that it’s just Republicans that have problems turning out their low-propensity voters. Democrats had the exact same problem in 2021. In fact, this pattern of turnout hurting the party in the White House is pretty systematic across cycles, making it almost an iron law.
Most of the polls missed this turnout shift, and as a result, candidates, the media, and other observers were blind as to what election they were running in. In 2024, the equivalent might have been thinking you were running in a race in Pennsylvania, only to find out on election night that you were running in Minnesota, a state just a few points to the left, but much harder to win.
In Virginia, the electorate likely shifted left by 4 to 6 points, meaning a Harris +10 to +12 electorate. That means a 3 to 5 point underperformance for Winsome Sears and a 4 to 6 point overperformance for Jason Miyares, who would have needed a 2024-like electorate to overcome the Jay Jones texting scandal.
In the Virginia House of Delegates, 13 districts flipped to the Democrats. In 2017, Democratic gains in the House of Delegates proved to be a good barometer for their gains in the 2018 midterms. In 2025, these gains reached into districts won by Trump by up to 3 points, but districts won by Trump by 5 points or more held. Both the 2022 and current mid-decade redistricting cycles limited the number of competitive seats in this range, so even a big swing to the Democrats may not mean that many House gains. And without a serious candidate quality differential, it would not put states like Ohio, Iowa, and Texas in play in the Senate.
In New Jersey, Mikie Sherrill won back many Latino voters that Trump won in 2024. Steve Kornacki compiles the data for Hispanic-heavy North Jersey towns:
But both persuasion and turnout certainly played a role in this shift. Nate Cohn reports that New Jersey Hispanics’ self-reported 2024 vote in 2025 was Harris +25, a universe that the 2024 exits pegged at Harris +9. Right off the bat, this suggests an electorate shifted 16 points left on turnout alone.
That’s not the whole story, though. Sherrill also won these voters by 37 points, a further 12 point shift based on the 2024 vote. A combination of turnout and persuasion created this huge shift.
This is a clear warning sign to Republicans who may have counted on all of Trump’s gains to stick, particularly in redistricting Texas based on the 2024 results. And it should cause Republicans to double down on their new 2024 voters, since it’s now clear they will swing back and forth much more than the rest of the electorate. The New Jersey exits found that 18 percent of Hispanic Trump voters defected on Tuesday, compared to just 5 percent of white Trump voters.
But a bit of perspective is in order too. Downballot Republicans were already performing double digits worse than Trump in 2024 in many of these places. Before the election, I warned that Republicans could see losses compared to the 2024 presidential, and that we should also be comparing the results to past election cycles.
And that was the case in New Jersey. Hispanic North Jersey towns held steady from 2021 even as the state shifted 11 points left. Since 2017, which saw an identical margin in the Governor’s race, these towns have shifted 18 points points to the right. Downballot and presidential elections each move on their own timelines.
💳 Affordability now looms even larger
Comments from the White House signal a course correction with more of a focus on the cost of living in 2026.
Our polling shows that this is still the issue that matters the most… to voters overall… to low-propensity voters… and most of all, to low-propensity Trump voters. Those are exactly the voters most at risk of not showing up next year, and if they do, of going Democratic.
In all three of Tuesday’s races, exit polls showed that voters who listed the economy as their main issue favored Democrats by nearly 2 to 1 — a flip from 2024.
What candidates focus on and the words they use matter. That’s highlighted by a Searchlight Institute study of the words used by the New York City mayoral candidates, with words related to prices and affordability appearing in more than twice as many (78%) of Mamdani’s messaging compared to Cuomo’s (32%). And Mamdani was much closer to voter priorities on affordable prices.
🗽 Mamdani’s win reveals a divided New York City
In the end, the New York City mayor’s race was closer than the polls were saying, with much more of the anti-Mamdani vote consolidating around Andrew Cuomo than was expected.
Compared to the primary, Mamdani did better among Black and Hispanic voters. Cuomo did better in Asian, traditionally conservative, and wealthier neighborhoods.
The New York Times’ precinct map lets you drill down on specific precinct types by demographics.
The three-way contest did not map cleanly onto party lines, but on an otherwise good night for Democrats, the New York results showed a far more spirited opposition to Mamdani’s progressive platform than anticipated. Few would have expected New York City to be the closest of the major contests on Tuesday.
One way of visualizing this is by looking at Mamdani’s margin compared to Harris last year. He lost the most among Outer Borough Blacks and among wealthy Manhattanites.
New York City makes for aesthetically pleasing precinct maps, but sometimes what you really need is something showing all the different neighborhoods represented by Wojaks.
🇺🇲 More election shoutouts
DDHQ, State Navigate, and VoteHub proved to be indispensable Election Night resources. The era of the live precinct map has well and truly arrived.
Chaz Nuttycombe’s predictions may have looked like Democratic hopium in the run up to the election, but proved exactly right.
🇺🇲 Redistricting war goes into overdrive
The victory for California’s Prop 50 may prompt Democratic states to go even further in redistricting states, including amending their state’s Constitutions, something that California has successfully done and Virginia may do next. Nate Cohn now says this means Democrats could outright win the redistricting war.
The Times’ rundown before the election, combined with this helpful chart, points to a slight Republican edge.
For decades, Section 2 of the Voting Rights Act has served as the sole federal limit on partisan gerrymandering, and should it be struck down there would be no federal law against drawing extremely partisan congressional maps.
📚 It’s not just going to college. It’s what you study that makes you more left-leaning.
Studies show majoring in social sciences and humanities makes students more left-leaning, implying that if all students majored in business, the college-noncollege ideological gap would shrink by 1/3.
🫏🐘 The ways that both parties have gotten more extreme over time.
💻 How AI is actually ruining the job market
There’s now clear evidence that applicants using LLMs to write cover letters changes who get hired, allowing applicants to disguise previously weak applications and leading to weaker hires overall.
