Why Polls Are Real and Why They Really Matter.

Marist, January 12-13: 38%¹

CNN/SSRS, January 16: 39%²

AP-NORC, January 8-11: 40%³

Economist/YouGov, January 9-12: 40%⁴

Emerson College, January 2026: 41%⁵

Reuters/Ipsos, January 24-26: 42%⁶

When I shared these poll numbers earlier today, I made the mistake of including Rasmussen. The first response I got called that out. And fair enough. Rasmussen has earned its reputation.

Here’s what makes a poll biased. Rasmussen uses automated phone calls that only reach landlines, which skews older. They weight their samples in ways that consistently produce results more favorable to Republicans than actual election outcomes. In 2018, Rasmussen predicted Republicans would win the generic ballot by one point; Democrats won by nearly nine, a miss of ten percentage points, the largest error among major pollsters that cycle.⁷ In 2024, FiveThirtyEight dropped Rasmussen from its polling averages entirely, citing failure to meet methodological standards.⁸ We know all this because we can check. Every major pollster releases a final survey before Election Day, and then we get the real numbers. Over years and cycles, Rasmussen misses in the same direction, by roughly the same margin, every time. When a thermometer always reads high, you stop trusting the thermometer.

So yes, throw Rasmussen out. I shouldn’t have included it.

But here’s what caught my attention. The reaction wasn’t “good point about Rasmussen, what about the others?” The reaction was “Rasmussen is trash, so I don’t believe any of them.” And I understand that impulse. I feel it too. These numbers are miserable. Forty percent of our neighbors look at children in cages, at pardoned war criminals, at open corruption, and say: more of this, please. That’s a hard thing to sit with.

But let’s slow down, because this is where we have to be careful. If we dismiss evidence because we don’t like what it tells us, we become the thing we criticize.

Let me explain how the legitimate polls actually work, because if we can’t explain why we trust something, we’re not trusting it. We’re taking it on faith. And faith is their game, not ours.

Imagine you have a jar filled with 330 million jellybeans, some red and some blue, and you want to know the ratio without counting every single one. You could grab a handful and look at what you’ve got. But if you only grab from the top, you might get a bad read; maybe the reds settled to the bottom. So you shake the jar first. You reach in blind. You pull from different depths. You do this a thousand times.

That’s random sampling. The key word is random. Every jellybean has to have an equal chance of getting picked, or the handful you pull won’t reflect the jar.

Now here’s where it gets interesting. A thousand jellybeans, properly sampled from a jar of 330 million, will tell you the true ratio within about three percentage points, nineteen times out of twenty. That mathematical relationship holds whether you’re polling a county or a country. The jar can be almost any size; what matters is the randomness of your sampling and the size of your handful.

Pollsters like Pew, Gallup, and the AP-NORC Center at the University of Chicago build their entire methodology around protecting that randomness. They use techniques designed to reach people who don’t answer phones, who work night shifts, who don’t have landlines. Then they weight the results. If their sample came back with too many college graduates compared to the census, they adjust so the final numbers reflect the actual population. Every choice they make, every weight they apply, gets published. We can read it. We can critique it. We can check their final pre-election polls against the vote counts and see if their methods actually work.

And here’s the thing: they do. Not perfectly, but consistently within their stated margins of error, across decades, across hundreds of elections. That track record is public. FiveThirtyEight maintains ratings of every major pollster based on exactly this kind of historical accuracy check.⁹ The firms I’m talking about have exposed themselves to that scrutiny and passed.

Now, someone might say: maybe they’re all coordinating. Maybe Pew and Gallup and AP-NORC and YouGov are all in on it together, producing the same numbers to manufacture consent.

Think about what that would require. These are competitors. They serve different clients, use different methods, employ different people. A conspiracy to produce identical fake results would require hundreds of statisticians across multiple organizations to secretly coordinate while publishing their methodologies for anyone to critique. And for what? The numbers they’re producing aren’t flattering to anyone in power. They’re telling us the country is split almost in half on the most basic questions of decency. That’s not propaganda. Propaganda would look much better.

When six independent organizations using different methods land on the same ugly number, the simplest explanation is that they’re measuring the same reality.

So why does it feel wrong?

If you don’t believe 40% of the country supports this, go find them. Not online. In person. Walk onto a construction site. Sit down in a biker bar in upstate New York. Go to a gun show. Stop at a truck stop. Attend most churches in any zip code, including yours. These people are everywhere, and they’re not hiding. They have flags on their trucks and signs in their yards. They’re your coworkers and your cousins and the guy who fixes your furnace. Even in blue states, there are plenty of towns where 80% of the room believes in this.

We have sorted ourselves so thoroughly that the 40% might as well live on another planet. We might work with them, but we don’t talk politics. We don’t drink with them. We don’t hear them. And because we rarely encounter their views directly, their existence starts to feel like a lie.

The polls aren’t broken. Our experience of the country is just incomplete. We’re looking at our handful of jellybeans and assuming the whole jar matches.

I know this is uncomfortable. It would be comforting to believe that the country has turned a corner, that the support for this regime is a mirage manufactured by bad actors with bad methods. But when I look at the evidence, when I actually examine how these polls work and check them against outcomes, I can’t make that case. And if I told you I could, I’d be doing the thing we accuse them of doing. I’d be choosing my conclusion first and hunting for justifications after.

We’re supposed to be better than that. We’re supposed to be the people who follow evidence even when it leads somewhere we don’t want to go. That’s the whole point of being on this side. Not that we’re always right, but that we’re willing to be wrong. That we check our assumptions instead of defending them. That we can tell the difference between inquiry and faith.

The 40% is real. And that fact has strategic implications we cannot afford to ignore.

This is why the polls matter. Not as horse-race entertainment but as intelligence about the country we actually live in. If 40% is real, federal solutions become much harder to believe in. And in places where that number crosses 50%, where the majority actively wants what’s happening, we face a harder truth. That we cannot share a seat at the table with them.

We don’t cut them off because there aren’t good people there who need help. We cut them off because sharing governance with them will get us all killed. These are not people who disagree with us about tax policy. These are people who watched the last decade and approved. They see us as evil. As terrorists. As enemies to be destroyed. They are glad when we suffer. They will use every lever of shared government to hurt us, and they will call it patriotism.

You cannot share a democracy with people who do not believe in democracy. You cannot hand power to people who will use it to end you. This is a reasonable response to address an existential threat. 40% of the country has gleefully handed over our nation to a transnational criminal organization that aims to ensure we never have another free and fair election.

For information on how to fight back, check out the Intro to Soft Secession and Oppositional Federalism booklets at TheExistentialistRepublic.com and free versions of everything at the shop on BuyMeACoffee.com/TheER

References

¹ Marist Poll. (2026, January 13). U.S. foreign policy, January 2026. https://maristpoll.marist.edu/polls/u-s-foreign-policy-january-2026/

² CNN. (2026, January 16). CNN poll finds majority of Americans say Trump is focused on the wrong priorities. https://www.cnn.com/2026/01/16/politics/trump-economy-first-year-cnn-poll

³ AP-NORC Center for Public Affairs Research. (2026, January). Fewer want the U.S. to take an active role in global affairs. https://apnorc.org/projects/fewer-want-the-u-s-to-take-an-active-role-in-global-affairs/

⁴ The Economist/YouGov. (2026, January 9-12). Poll of 1,602 U.S. adults. Cited in Newsweek. https://www.newsweek.com/donald-trump-approval-hits-rock-bottom-11373941

⁵ Emerson College Polling. (2026, January). January 2026 national poll. https://emersoncollegepolling.com/january-2026-national-poll/

⁶ Reuters/Ipsos. (2026, January 24-26). Poll on Trump approval. Cited in Al Jazeera. https://www.aljazeera.com/news/2026/1/27/trumps-approval-on-immigration-falls-to-record-low-new-poll-finds

⁷ Rasmussen Reports. (n.d.). In Wikipedia. https://en.wikipedia.org/wiki/Rasmussen_Reports

⁸ Sotomayor, M. (2024, March 8). 538 drops Rasmussen Reports from its analysis. The Washington Post. https://www.washingtonpost.com/politics/2024/03/08/rasmussen-538-polling/

⁹ FiveThirtyEight. (n.d.). Pollster ratings. https://projects.fivethirtyeight.com/pollster-ratings/