The Internet vs. real life, quantified

Political Tribes Quiz results, Realignment wakeup call, why Texas isn't a dummymander, moderation still wins, Nate Silver's blistering takedown, a key difference in Trump's approval, politics at work

Nov 14, 2025

No. 382 | November 14th, 2025

🗳️ Takeaways from the first 46,000 Political Tribes quizzes

The Political Tribes quiz is inching up to 50,000 responses. Take it here.

Quizzes like this are never representative of the broader public. And that’s baked into the design: everything about this quiz is designed to compare people on the Internet to a representative sample of voters who answered the same questions. Here’s how they’re different:

Quiz takers are more progressive and less populist than people in real life.

An old-school left Labor Party in our multiparty democracy test does about as well on the Internet as in real life, but more niche left factions like the neoliberal Acela Party and the far-left Green Party do better. A traditional Conservative Party does better than a MAGA-like Nationalist Party than it does in real life.

Along these lines, moderate middle voters who have a mish-mash of right and left opinions are far less well represented online than in real life. That’s seen via our Tribes cluster analysis. Three key groups, representing key 2024 swing groups, are far more common in the electorate than in the X-heavy quiz taker base. These include the Young and Disillusioned (think Starbucks barista) who are left-leaning but anti-establishment, the Middle American Optimists who are, low-information moderates, and the New Republican Populists, the largely rural, now very Trumpy voters who are extremely left-leaning on economics but culturally very right-leaning. Combined they are just 8% of quiz takers but 33% of the electorate.

A substantial share of right-leaning voters in our quiz did not vote for Trump, a constituency that practically doesn’t exist in the real world. Social issues and trust in institutions, as represented by our (anti) Establishment index, divide right-leaning voters who voted for Trump from those who didn’t.

On the issues, quiz takers lean to the left of voters on every dimension but the biggest social issue gap is on gay marriage (favored by a net 55 points among quiz takers, and 20 points in the public). On the flip side, online respondents are far more likely to take a pro-free speech stance (+44) than voters (+8).

On economics, the gap isn’t as big, but there’s a notable divide here. Voters are far more likely to emphasize redistribution (higher taxes on the rich, minimum wage hikes), while quiz takers place more faith in regulators, both overall and on the environment.

But the biggest splits are on trust in institutions. Quiz takers are overwhelming in their trust in experts (vs. ordinary people). Among Harris voters, net support for this is +96 points, second only to gay marriage (+98) and higher than abortion rights (+91). That’s a yawning gap with Harris voters IRL, who are only +50 on this question.

Because many right-leaning voters in our quiz didn’t vote for Trump, those who remain are much more right-leaning across social and economic issues (but not trust in institutions). See the chart below on social issues. If you live and breathe politics on X, you likely haven’t encountered a moderate or swing voter who voted for Trump — a nagging problem for the political and media elite since 2016.

We’re running some correlations on the data to figure out potential vote drivers. You can check those all out in a dedicated section of the results. But one big takeaway here is that faith in traditional institutions like the media, politicians, multilateral organizations, represented here by our establishment index, is far more important as a vote driver online than it is in the real world. Online, it competes with social issues as a vote driver for Harris voters, while social issues are clearly most more correlated with vote choice IRL.

🇺🇲 Realignment wakeup call

The Democrats’ recovery with minority voters in Virginia and New Jersey should be a wake up call for Republicans, particularly on affordability and the cost of living that were top of mind in both 2024 and 2025. The Center for Politics and Zachary Donnini each highlight the numbers.

The shifts take Republicans back to where they were in 2020 — which nevertheless represents a dramatic improvement from where they were in earlier cycles but nowhere near the frothiness of 2024. One bright spot for Republicans was the hydrogen bomb of new votes in Lakewood, NJ — the dark red area of Ocean County — where an endorsement from the Orthodox Jewish Vaad shifted that community dramatically rightward.

If you haven’t gotten enough charts on the 2025 elections, here are 10 more from DDHQ.

🗺️ Is Texas a dummymander?

Not really, says Jacob Rubashkin at Inside Elections. Even under the best of circumstances, Democrats would stand to gain just two of the five new seats expected to go Republican.

And Florida is waiting in the wings with its own R+5 remap.

🫏 For the umpteenth time… moderation still wins

The WAR truthers just won’t give up, but the data is clear that moderation still wins elections.

If you created the perfect Democratic candidate in a lab, what would that look like? Blueprint has the answer.

Speaking of “analysts” who deny reality and twist data to support whatever their audience wants to hear, I highly recommend this clip from Nate Silver’s appearance on the new Central Air podcast. I promise, you won’t be disappointed.

📊 The difference in Trump approval now vs. the first term

YouGov highlights what I think is a Rosetta Stone to understanding public approval of Trump between his two terms: in his first term, Trump hadn’t fully consolidated the Republican base yet, whereas now Trump’s underperformance now comes from Independents.

It’s a positive for Trump that the Republican base is now larger and more supportive of him. But it’ll also be harder for Republicans to win over the marginal Trump approving Independent this time around.