On this week’s Strength In Numbers podcast, G. Elliott Morris and David Nir explain why polls could underestimate either the Democrats or the Republicans this November. You can make a good case either way.

Plus, will Democrats win a majority of governors’ mansions for the first time since 2009?

This video is free to watch, but it was not free to produce. If you value this podcast and want to help keep it going, please consider becoming a premium subscriber to Strength In Numbers. Paid subscriptions keep the show ad-free, fund the newsletter, and ensure we have the time it takes to do original data-driven analysis of politics and elections. Paying subscribers also gain the ability to send in questions during our livestreams, so you can directly shape our conversations on the podcast.

Upgrade to support Strength In Numbers

A few big takeaways from the conversation:

• Polls underestimated Republicans in 2016, 2020, and 2024. Why did they do that? Is it possible we’re seeing the same biases now? Or have pollsters overcorrected for their past misfires and are now undershooting Democrats?

• Pollsters could be increasing the weight on very pro-Trump segments of the population. Due to possible error from something pollsters call “recall bias,” there might be too many MAGA Republicans in surveys—especially if disaffected Trump voters are claiming they did not in fact vote for him in the past.

• Or, super-engaged respondents are flooding polls with pro-Democratic attitudes. Elliott finds that “high-effort” respondents lean Democratic. If they’re overrepresented in polls (as they have been in the past), this could be costing Republicans about a point or 2 on average.

• Democrats look favored to net at least two governorships, and could win back a majority of states’ executive posts for the first time since 2009. But there are a ton of extremely close races, so if Democrats finish strong, that number could grow.

Thanks to everyone who joined us for the weekly livestream on Substack. We’ll be back next week!

As a reminder, paid subscribers to Strength In Numbers get to participate in our live Q&A! If you missed our video livestream, you can watch it by clicking play on the web version of this post at gelliottmorris.com.

You can also subscribe to us on your favorite podcast app to listen on your own time. And if you do listen via one of those apps, please drop us a five-star rating and review if you feel we’ve earned it — it really helps people discover the show!

Subscribe now

You can read the transcript of our conversation by clicking the headline of this article to take you to the web version of the podcast, then clicking the button just below the byline that looks either like a piece of paper or is labeled “Transcript,” like so:

If you’re a reader of Strength in Numbers and haven’t yet subscribed to David’s newsletter, head to the-downballot.com.

Subscribe to The Downballot

And if you’re coming from David’s site, subscribe here to get the numbers behind the news.

Subscribe to Strength In Numbers

Readers, an exciting announcement below from David and me…

It’s our pleasure to invite you to a pre-election briefing exclusively for Founding Members of The Downballot and Strength In Numbers! The event will be hosted by David Nir and G. Elliott Morris.

On Tuesday, Oct. 27, at 6:30 PM ET, we’ll host a private Zoom call to walk through the key metrics we’re using to make sense of the midterms in the closing week — from special elections to primary turnout to the generic congressional ballot and more. We’ll also drill down into what the forecast models predict on Elliott’s sister site, FiftyPlusOne, and give a candid assessment of uncertainty in the polls heading into the final week of the race.

After our presentation, we’ll open up the floor for a Q&A and discussion on any and all topics related to the election. We’ll wrap at 7:45 PM.

This event is open only to Founding Members of either site, but if you’d like to bring a guest, just let us know when you fill out our RSVP form below.

If you’re not a Founding Member of SIN, you can become one using this link:

Become a Founding Member

Then come back to this page on the Strength In Numbers website and scroll down to enter your contact information so we can invite you directly to the call.

Thanks, and we look forward to seeing you!

Elliott & David

Read more

As I showed in this article and this one, while polls likely have some error in them (this is inherent to how polls work), it is a fool’s errand to try to predict the direction of error ahead of time.

But that does not mean that we are without tools to explore and explain how our polls might be off. Strength In Numbers is doing some pretty sophisticated stuff on this front, since we know that how pollsters adjust their data is now as important as, or potentially more than, how they gather it.

And in this week’s Deep Dive, I want to present one of the new tools I’m using to detect potential bias in the polls ahead of November’s elections.

See, the age-old problem in polling — all polling, not just the Strength In Numbers/Verasight poll — is that the people who will take our polls are often different from those who won’t. In 1936, polls had too many high-income voters. In 2016, too few non-college-educated whites answered their phones.

This problem is called “non-response bias” — since the non-responders are biased compared to the responders. The good news is that pollsters can mostly weight away non-response for different demographic groups by forcing samples to match Census totals (the Census being, of course, a giant poll with a super high response rate!).

The bad news is that weighting can’t fix partisan non-response — supporters of one side quietly opting out of polls at a higher rate than the other, in a way no demographic correction can undo. That’s because there’s no official national tally of Democrats and Republicans to weight to, so pollsters usually find out they had a problem only after the votes are counted.

But what if you could undo bias in a different metric that serves as a proxy for partisan non-response, and that you can detect entirely within the survey data? No more weighting to guesses at what the country’s party ID breakdown is this month!

We know, for example, that our freaky group of survey-takers tends to be super politically and socially engaged. They vote more, volunteer more, donate more to campaigns, and even see more movies in theaters than poll non-responders.

Can we find a way to see if some of our poll respondents are more engaged than the rest? If they are, are they also more Democratic or Republican leaning?

That’s exactly what I did in our August Strength In Numbers poll. I ran an experiment with free-response text questions so I could grade each respondent’s “effort” in answering the poll. This let me compare high- and low-effort respondents and compare their politics.

In other words, in the hypothetical where polls are over-representing high-engagement individuals, can we proxy that with our in-survey measure of survey effort?

It turns out that high-effort respondents are significantly more favorable to one party over the other. And if this measure of effort is at all predictive of the voting behavior of people who don’t answer the polls (who can be thought to have an effort of zero), the polls could be off in terms of vote margin by up to 3.5 points favoring one party this November.

You’ll have to read on to figure out which one. The upcoming midterm elections will serve as an out-of-sample test of this analysis, which has predicted polling errors in recent elections, but only after the fact.

Low-engagement survey-takers are significantly less ___ than high-engagement ones

Read more

If you’re new here, or on a monthly plan, I want to ask you to support this independent publication by becoming a yearly member. More than ask you, I want to incentivize you: that’s why I’ve enabled a discount code for 20% off annual subscriptions here at SIN from now until Election Day. Redeem it with the button below.

Strength In Numbers is not an election-year-only publication, and I have big plans for our work in 2027 and 2028 — including the monthly poll and data dashboard, podcast, and collaborations with other publications — that you won’t want to miss. Not only do I think that $72/year is a really great deal for informed, data-driven politics coverage delivered straight to your inbox 3x a week, I think we’re building something special here that you can’t get anywhere else in the media. And, frankly, SIN has had a very good track record since launching full-time in May 2025, with a lot of Value Above Replacement compared to the pundits you’re used to hearing from/about.

So please join us. If you believe that there is strength in numbers, you’ll feel right at home here.

Upgrade to support Strength In Numbers!

1. Polls might overstate Democrats, but they roughly align with the fundamentals

There is a lot of debate online right now about whether the polls can be believed. Democrats are doing suspiciously well, the logic goes, in red states where polls have overstated support for Democrats before. On Wednesday, CNN’s Harry Enten pointed out that polls overestimated Democrats in Iowa, Michigan and Ohio in 2016, 2020 and 2024. Two years ago, Republicans beat their September polling in those states by 9 points in Iowa, 6 in Ohio and 5 in Michigan, according to Enten.

It is worth noting that polls were unbiased on average nationally in 2018 and 2022, though they overstated Democrats again in those same Midwestern states.

President Trump reposted the clip to Truth Social, and by Thursday the National Republican Senatorial Committee had written up a whole press release, warning Democrats (and GOP donors) against relying on “rosy polling” that historically overstates Democrats. Last week, the Senate Leadership Fund pulled out of the race in North Carolina — independent actions that seem to affirm, not contradict, the polls.

It is understandable that the Republicans would want to deny the polls. But now Enten’s segment is being used as evidence by the broader pundit class that the polls are not to be trusted. As blogger Matthew Yglesias put it, “the midwestern polling is probably fake.”

Do I think Yglesias actually means that pollsters are faking their data in these states, as he says? No, I don’t think he really believes that. But I’m not going to put words in his mouth. That’s a pretty serious allegation!

It’s true that polls overestimated Democrats in these states recently. By my count, September polling averages overstated the Democrats in all 24 Senate, presidential, and governor races in Iowa, Ohio, and Michigan since 2016, by about 6 points on average. Not great, Bob!

But as I wrote in August, past patterns of polling error are not historically predictive of future misses. If you expand the analysis from 2016-2024 to 1998-2024, the results from Enten’s analysis reverse. Using our database of polls at FiftyPlusOne — as far as I know, the most comprehensive collection of surveys up and down the ballot — I have shown that the direction of polling bias bounces around from election to election. Knowing what the polls did in the previous four cycles tells you almost nothing about the next one.

For example, from 1998 to 2014, polls in the Midwest leaned the other way, understating Democrats by about 5 points in Michigan, 2.5 in Iowa, and a point in Ohio. If you used those values to adjust the polling averages in 2016, you would have ended up with even worse predictions than the polls gave.

More recently, those trying to “unskew” the polls have failed badly. In 2022, RealClearPolitics subtracted its estimate of polling bias (from the past three cycles) from its Senate averages and turned a tied race in Pennsylvania into “Oz +2.2.” Fetterman won by 4.9. RCP’s adjusted averages had Republicans ahead in Arizona, Georgia, and Nevada, and its final map projected a 53-seat Republican Senate. But after runoffs took place, Democrats won 51. I wrote about that at the time.

And who can forget in 2012, when Dean Chambers’s UnskewedPolls.com (yes, that was its real name) adjusted surveys for an assumed GOP party ID lead and said Mitt Romney was winning? Barack Obama won by nearly 4 points.

Finally, in the most recent election cycle in 2025, the polls under-shot Democrats by almost TEN PERCENTAGE POINTS in the New Jersey and Virginia governor’s races that year. That is an error twice as big as the state polling misfires in 2016, 2020, and 2024. Some people even used early pro-Republican partisan polls to predict the GOP would carry New Jersey!

Yes, the polls today could be different from the final results. There are 30 days left, and measurement error is inherent to polling. But that doesn’t mean you should expect bias in any direction. On average historically, assuming surveys are unbiased is empirically the right decision. It’s also possible that polls are underestimating Democrats, and if so, all the predictions of pro-Democratic bias are going to look really silly.

I’ll have more analysis of potential Democratic bias in Tuesday’s exclusive Deep Dive for paying subscribers. I deploy a new test of “effortful responding” in our own polls to look for signs of partisan non-response that weighting can’t fix. To maybe spoil this a bit, yes, I do think there are some things to be worried about. Make sure you’re subscribed to get that in your inbox.

But enough about that for now; the ground has been pretty well-trodden at this point. What I want to point out today is that these state-level polls aren’t the only positive sign for Democrats. The electoral “fundamentals” show a big pro-Democratic year, too.

Subscribe for the weekly Deep Dive

What the fundamentals say

At FiftyPlusOne, our Senate forecast starts every race from a “fundamentals” estimate that uses no state polls at all. There are four components to this projection:

First, the national environment. The model uses our House popular-vote forecast, which is anchored on the generic ballot polls and stood at about D+9 in the latest run on Saturday night, October 3.

Yes, there is some circularity in using the generic ballot polls as a component in the fundamentals forecast. But the good news is that the D+9 reading from the generic ballot (which is by far the best predictor we have of the House popular vote) is in line with the other indicators of a Democratic year, from Democrats’ edge in special elections to the signals David Nir and I walked through in this interview with Paul Krugman published yesterday. And see this regression analysis based on presidential approval from Joel Wertheimer:

So the polling and non-polling national predictors are all in basic agreement that we’re at D+8-11 in the national House vote.

Second, the state’s partisan lean. This is a measure of how much redder or bluer a state voted than the nation in 2024 and 2020, weighted 80/20 toward the more recent election. This value gets multiplied by 0.83, our forecasting model’s estimate of how much of a state’s presidential lean carries over into a future Senate race.

Third, the candidates: Incumbency, political experience, and each nominee’s share of the two candidates’ individual fundraising historically predict Senate elections.

Fourth, the “personal vote”, which measures electoral over- or under-performance for incumbent Senators, relative to all the factors mentioned above. More about these four categories of predictors in our model write-up.

If you do the math on these predictors in most states with competitive Senate seats, you get a result that looks an awful lot like the polls. Here, for example, is how the fundamentals add up in Iowa, the race the skeptics are most focused on:

The upshot of this math is that before you even read a single Iowa Senate poll, the fundamentals say it is a one-point Republican race. The FiftyPlusOne polling average in Iowa today is D+1. When you consider that the fundamentals are not perfect predictors of election results (they have a historical standard deviation of out-of-sample error of 7 percentage points), the 2-point difference between the polls and fundamentals is not statistically significant.

If you run the same exercise on, for example, Ohio, you get a predicted election-day margin of D+1.5; Texas is D+2.0; and Alaska is D+2.0. From very first principles, a state that voted ~12 points redder than the country is almost a coin flip in a D+9 year. Again, you get a pretty Democratic leaning environment before even looking at the polls!

Here’s the full table of fundamentals vs. the polls in every competitive Senate seat. The last column measures how different the polls are from the fundamentals, considering the uncertainty in said fundamentals, using something called a Z-score. Generally speaking, values lower than 1 represent pure noise.

A real boost for Democrats is that they have recruited really good candidates. The model’s personal-vote term says Dan Sullivan has run about 5 points behind Alaska’s partisan baseline over his career — worth 2.7 points to Mary Peltola — and that Roger Marshall has run about 2 points behind Kansas’s. Susan Collins has run 12 points ahead of Maine’s, which is why that race is closer than the state’s lean.

In our testing, a candidate who leaves office and comes back does not get much of a boost for past overperformance, so, e.g., Sherrod Brown gets nothing for carrying Ohio by 12 points in 2006 and 7 in 2018, in years when the national House vote was D+8 and D+9, environments a shade smaller than today’s. This may end up being a blind spot for our model, so if you give Brown some of that back, Ohio’s polls, 2.8 points above its fundamentals, look about right.

In every one of the red states skeptics are focused on, the polls sit within about half a standard deviation of what the fundamentals predict: Ohio at 0.4, Iowa at 0.3, Alaska at 0.3, Texas at 0.2, Kansas at 0.7. This is a very different situation from 2016 and 2024, or even 2020, when polling leads for Democrats in places like Iowa and Ohio were very out of line compared to the fundamentals.

The polls on average are about half a point bluer than the fundamentals.

The one race where the polls tell a story that is fundamentally (ha) different from the fundamentals is Nebraska, where independent Dan Osborn polls a point behind Pete Ricketts in a state the fundamentals put at R+11. Our model mostly refuses to believe this and gives Osborn an 18% chance of winning. I think the fundamentals are probably wrong in Nebraska and over-indexing on Trump’s 2024 margin there.

Share

Here’s a final thought: Pundits seeking attention are highly incentivized to predict polling errors before they happen so they can claim “I called this!” after the votes are counted. I think this is silly and a waste of readers’ attention. Silly because this exercise is historically fraught, and mostly the naysayers are just saying “these polls violate my priors” but not otherwise offering any useful/insightful/original commentary. And a waste because there’s a lot of very good journalism out there — from, for example, Strength In Numbers, Split Ticket, Nate Cohn at the Times, and others — about the mechanics of error that promote a real understanding of surveys. That is key to unlocking the potential of polls to positively shape our democracy.

So you should be prepared for polling error. There will probably be some. But nobody can tell you with any confidence ahead of time which direction it’ll run. This election looks like a blue wave whether you’re looking at fundamentals, state polls, or other indicators. If the current map shocks you, the thing to reconsider may not be the polls, but your own personal priors.

2. What Strength In Numbers published last week

It was a six-article week here at Strength In Numbers!

In Last Sunday’s roundup, I showcased results from our September Strength In Numbers/Verasight poll that found Trump’s promise to send a check worth $5,000 to every American if the Republicans hold the House and Senate at a net 18-point lead unpopular, and 71% of adults say that kind of promise is inappropriate for a president to make:

On Monday, I unveiled a new database of news headlines stretching back to 2016 that I scraped online. I showed that Iran, prices, ICE and Epstein have been the biggest stories of 2026, and that the “Trumpapalooza” GOP Midterm convention didn’t move the news agenda the way a real convention does:

Tuesday’s Deep Dive covered results on election administration from our September poll. Voters said they oppose sending ICE agents to polling places by 30 points, and several vulnerable groups said they’d be less likely to vote if ICE agents were there:

On the podcast Thursday, David Nir and I talked about how much worse (or better?) things could get for the GOP over the last (at that point) 33 days of this election. We also worked out why the House looks so lopsided in the forecasts, how Democrats got to a Senate edge through long-shot red-state races, and spoke frankly about how much of this you should believe:

Friday’s free Chart of the Week covered that for the first time in our tracking poll: a majority of Republicans and Republican-leaning independents — 52% — disapprove of Trump’s handling of prices, a net rating of -8, as diesel hit a record high. This is probably hurting Republican candidates in red states, too. If Republicans lose the Iowa and Ohio Senate races, Trump’s war in Iran will be one of the biggest causes.

And on Saturday, Paul Krugman had David Nir and me on to talk through four indicators pointing to a Democratic November. We tried to wrap our heads around why the polls might be wrong, and also why the media whiffed so badly on Trump in early 2025. Free video with a transcript:

If you’re a frequent reader of Strength In Numbers, I’m confident you will get a lot of value out of a paid subscription to the site. You’ll get the Tuesday Deep Dive every week, access to a private community for paying members on Discord, and an early look at our monthly Strength In Numbers/Verasight poll before public release.

More than that though, you’ll be supporting original journalism about American politics that is rooted in data that we collect, together, as a community devoted to the importance of public opinion.

Upgrade to support Strength In Numbers

3. Even more numbers!

• In an era with almost no ticket-splitting, Henry Olsen finds a model of House election outcomes based solely on the final generic ballot average came within 11 seats of the actual House majority in each of the last three midterms, and within two seats in 2014 and 2018. Run the same method on an 8-point Democratic average, and you get a Democratic majority of 225 to 232 seats. How many House seats will the Democrats win? The polls may already know.

  • I think Olsen is probably understating democrats here. I made the same model as part of a 2026 redistricting analysis back in May, and one of the things I learned is that this model badly underestimates Democratic seats in 2020 and 2024. That’s because Democratic House candidates have generally been running ahead of presidential candidates by a couple of percentage points. My best guess is he is about 10 seats off what an adjusted model (adjusted, in this case, to make it more accurate in 2020, 2022, 2024, and 2026) would show.

• Pew has an analysis of AI “digital twins” that is pretty damaging for the synthetic sampling crowd. Pew finds that AI-imputed answers for their panelists missed real data from their human respondents by 12 percentage points on average, and the models exaggerated differences between groups. Pew says none of its demographic groups averaged under 12 points of error. How well synthetic samples replicate public opinion

  • Related: My whitepaper analysis on the same subject for Verasight, published this summer, found much the same thing Pew does here, with an average error of synthetic data vs benchmarks of 11 points.

• The University of Maryland’s Program for Public Consultation surveyed more than 20,000 adults across 11 battleground states and 28 battleground House districts on taxes. An annual tax of 2% on wealth over $50 million and 3% over $1 billion is popular with a wide swath of Americans, including a majority of Republicans. Common Ground in the Battlegrounds: federal taxes

• Michael Tesler at Good Authority: Google searches for “testosterone booster” track Trump’s 2024 state vote share at r = .85, and a July YouGov poll finds Republicans call boosting men’s testosterone a good thing, 39% to 8%. Republicans’ fascination with testosterone in three charts

• Dan Drezner starts gaming out Democrats’ worst cases for November, as a form of “exposure therapy.” Related to the main contents of this newsletter. Exposure Therapy for Democrats, Part One

Share

And that’s it for this week! Thanks for reading. Strength In Numbers will be back in your inbox on Tuesday.

Comments are OPEN on this post. If you have a question or comment you want answered/addressed in next week’s podcast or another article, leave it below!

Leave a comment

Related Articles

Watch this video on gelliottmorris.com.

Paul Krugman had David Nir, who co-hosts the SIN podcast with me, and me on his weekly interview show this Thursday, and the video is out this morning. Paul has graciously agreed to let us repost the video interview and transcript here.

In this recording, we talk about (obviously) the state of the midterms — particularly the now-quite-rosy odds of the Democrat taking the House and Senate in November. We also talk about (go figure) the economy, and the counterintuitive finding from historical midterms that economic growth and other objective indicators typically don’t predict midterm outcomes well. What matters far more than any indicator is the label of the party in control of the White House. Of course, when everyone is saying they're mad about prices and only ~20% say the incumbent party is handling the issue well, that tends to have an effect electorally too.

We also get into the chance of a large polling miss this year, like in some notable recent elections. I give a reminder that tools like election forecasting models, when done well, can represent uncertainty in the election given measurement error in the indicators.

Thanks to Paul and David for a great discussion. Here are some links, and I’ve pasted the transcript below for easy reading.

Elliott

Subscribe to Paul Krugman

Subscribe to The Downballot

Upgrade to support Strength In Numbers

Transcript

Paul Krugman: Like almost everybody I know, I spend about 17 hours a day thinking about the midterms, and I’ve got G. Elliott Morris and David Nir from Strength in Numbers and The Downballot here to talk to me about: What do we know? What’s going to happen? What are the indicators? And then maybe some broader issues about what the heck we think is going on in U.S. politics. So, hi, guys. Welcome.

Elliott Morris: Hey, thanks, Paul.

David Nir: Paul, it’s a pleasure.

Krugman: So, yeah, as we record this, I guess it’s one month and two days before the midterms, and if I believe you guys, the polling and other indicators, which David will talk about, seem to be almost kind of shocking right now. Elliott, I just looked at your Fifty Plus One post, and I believe you now have 99% odds of Democrats controlling the House.

Morris: Yeah, well, it’s October, Paul. So we’re in what I consider spooky probability territory. 99% is when you start to go, “I hope these models are parameterized correctly.” Look, the way to unpack this: Our model at 50 Plus One shows a 99% chance that the Democrats will win 218 seats or more. 99% seems very large. As the forecaster, I don’t want to be wrong on the 99, but it makes sense when you look at the numbers on paper.

So, let’s take 2018 as a comparison case. That’s when Democrats won 235 seats in the House. That year, the generic ballot—which is a question that asks people how they will vote in their local congressional district, for which party they will vote—that has Democrats up somewhere between 8 and 9 points. Democrats won in 2018, leading the polls by about 8 or 9 points. They actually won the House popular vote by seven and a half points or so, if you account for some weirdness with, like, uncontested seats. So we are in a more favorable environment, if the polls are right, than 2018. And we also have a lot of other district-level indicators pointing the same way.

Krugman: But, I mean, the gerrymandering was fairly extreme before this, and if we go back to the beginning of this year, it was looking a little iffy, if I recall.

Morris: Yeah. So gerrymandering seems to have shifted maybe a handful of seats at most to the Republicans, accounting for some pretty rosy polling, for example, in South Texas and in Florida for the Democrats, where there’s been a larger shift than the national shift, which is about 10 or 11 percentage points. So if we are around eight points, our model says Democrats need to win the House popular vote by about three and a half points. So there’s room for the polls to be wrong by about five.

The average expected error for the polls at this point, 30 days before the election, is about three and a half percentage points. So we’re looking at a larger-than-average error in the polls. And then if you account for other indicators, district-level surveys, the race raters like we do, that’s how we get up to that 99. But yeah, I acknowledge this is a very, very large probability.

Nir: And Paul, just to address your point about the gerrymandering, which did heavily favor the GOP, it’s important to remember that Republicans were not doing that from a position of strength. They were doing it from a position of weakness, precisely because they knew that they were facing a very difficult midterm election. And they definitely—I can say this with certainty—definitely will not pick up, they will not flip, all of the seats that they targeted in Texas and Florida and elsewhere. And that has a lot to do with the quality of candidate recruitment on both sides.

And also, in particular in Texas and Florida as well, has a lot to do with Latino voters moving back against Trump after moving heavily for him in 2024. So, 2024 in a lot of ways was a high watermark. And so when you’re looking at how Trump performed in these districts two years ago, in a lot of cases, expect really a lot of fallback for the GOP.

Krugman: Tell me if I’m being underinformed here, but my understanding is that the gerrymandering was based, at least partially, on the assumption that the Hispanic shift towards Trump that took place in ‘24 was an enduring feature of the landscape, and that it all kind of goes wrong if, as now appears to be the case, it’s not enduring. Right?

Nir: That’s definitely what it appears to be. Because if you look at these redrawn districts, and instead of looking at how they went in 2024, if you go back a few years earlier and look at how they performed in 2020, when Joe Biden did much better with Latino voters than Kamala Harris did in many parts of the country, they show a much more competitive picture. And so you have Republicans talking about, you know, “Oh, these districts are Trump plus five, Trump plus ten, Trump plus 15.” But that’s 2024. You dial back to 2020, and suddenly we’re talking about districts that Joe Biden won, that Democrats have a real chance of holding on to in 2026.

Krugman: Okay. And I want to talk about other indicators, and polling in general. But just a question about the Senate: So now, we’re up into the 70s again for the Senate, right? And that’s a lot less. I mean, the Senate is sort of inherently gerrymandered. States that have about the population of some neighborhoods in the Bronx have as many seats. So, yeah. But that’s also looking pretty strong, right?

Morris: Well, speaking of state gerrymandering, don’t get me started about the drawing of the state border of Nevada. I mean, that was a gerrymander, right? So, look, the Democrats have 47 seats right now. They need 51 to have a majority and to overcome JD Vance’s tiebreaker as president of the Senate. So they need four pickups. They look likely, according to our model, to get a pickup in North Carolina, where the former governor, Roy Cooper, is running against the current chair of the Republican National Committee. So that’s what’s on my mind.

They look then likely as well to somewhere between a sort of toss-up and a lean Democratic seat in Texas and Ohio, where Democrats have got some good recruits in James Talarico and Sherrod Brown, a former incumbent senator of Ohio. And I should say Democrats are also very likely to hold on to their seat in Georgia, just for what it’s worth. And then we get into the seats that are, like, a lot, a lot tighter. That’s Iowa, currently Republican-held, but where the Democrat, Josh Turek, is leading the polls; Maine, where the polls were somewhat, you know, incredibly wrong in 2020, showing Susan Collins losing to—

Nir: Sarah Gideon.

Morris: See, this is why we do it together, Paul, right?

Krugman: Yeah, I know, it’s okay.

Morris: And then there’s Michigan, which seems like it should be a Democratic win in an environment where Democrats are ahead by eight points on the generic ballot. But the race is surprisingly close between Abdul El-Sayed and Mike Rogers. And there’s like a bunch of other reach seats that might be competitive. We can talk about Kansas, where Democrats currently have a polling deficit of under half a point in Kansas. Like, what’s the matter with Kansas?

Krugman: Yeah. I mean, objectively, Trump policies, especially a war that’s driven up diesel prices and fertilizer prices and all of that, has been really pretty bad for farm country. But I don’t know to what extent we’re seeing that, to what extent we’re just seeing really just bad Republican recruitment or just general Trump fatigue.

Morris: I’ll say one thing about the numbers and then hand it over to David. But the other Plains states—Nebraska, Iowa, and Kansas are the ones I’m thinking of—are all much more competitive than you would expect, given the fact that Donald Trump won these places by, you know, between 10 and 28 percentage points. I think in the case of Nebraska, if I’m remembering correctly. These are all within a point or so in the polls today, and back in 2025, when I was looking at Trump’s approval rating and all of the polling we do at Strength in Numbers.

So these are individual interviews from these Plains states. Donald Trump was also much less popular than you would expect, given his margins, and deportations and, and tariffs especially were very unpopular, or even more unpopular than Trump overall in these places. So, I think you’ve triangulated maybe there’s more economic pain in these places. Donald Trump has certainly very publicly pushed for these policies. So I can imagine voters are putting more blame on him directly because of the press conferences for tariffs in the early part of last year.

Nir: And Paul, I would say you’re exactly right to bring up diesel prices, because as bad as gas prices have gotten, you know, diesel has gotten much more expensive. And an attack on diesel prices is almost perfectly an attack on Trump’s base, because diesel is what’s used to carry goods in these Plains states that Elliott is talking about. It’s so important to the agricultural world, and Trump is really assaulting his own most loyal voters as hard as possible.

But it’s not just on that front, as Elliott said. You know, Republicans are upset about the stepped-up ICE raids in Kansas affecting the cattle industry. And so I think it’s just so many factors all combined together. You probably can’t pull just one out of them. And I think that they combine to make a lot of people feel like they’re just not being heard in general at all.

Krugman: Yeah. I mean, it’s one of those things where, for once, being an effete, liberal northeasterner makes me understate the amount of trouble that Trump is in, connected to diesel, because there’s not a whole lot of people driving tractors down the New Jersey Turnpike.

Nir: Not usually.

Krugman: But yeah, it’s actually kind of shocking to realize that gas sales are about two-thirds gasoline, one-third diesel, and the price of diesel is way up. So the actual pump shock out of the Iran war is more diesel than it is gasoline. And that’s hitting the states really hard.

One of the questions we all have now is, you know, polling. But David, you have this other indicator, which is kind of a cross-check on the polling, right? I want to hear about that, because I think it’s important.

Nir: It’s really fascinating, and I’m glad you brought it up. So at The Downballot, my colleague Daniel Donner published a new study recently, and we’re pretty sure that no one has actually looked at what Daniel looked at in precisely this way. So folks often ask, does turnout in primaries have anything to say about general elections? And they usually look at it on a state-by-state or case-by-case basis. Democrats, for instance, had incredibly strong turnout in the Texas primaries this year, better than Republicans, which is extremely unusual for Texas. So, does that actually mean anything for November? And when you look at these individual data points to compare primary turnout and general election performance, things kind of go all haywire.

But what Daniel did was he decided to look at primary turnout across the entire country. And he didn’t just do it for one year. He went back more than 100 years. And it turns out that there is a very close relationship in what we’re calling this national primary, when you add up primary votes for all 50 states, and then you look at the national House popular vote in the subsequent general election.

And what Daniel did, his real innovation here, it’s almost like he took the first derivative, because instead of looking at what primary turnout is in a given year, he looked at the change from midterm to midterm. And so if you look at the 2022 turnout, Democrats didn’t do particularly well that year. In fact, they are doing in 2026, 18 points better in this national primary that Dan essentially created synthetically. And what that extrapolates to, is that Democrats are looking like they’re doing 16 points better than in 2022, when you modify things a little bit, because, you know, Kamala Harris lost the popular vote in 2024.

But what this points to is a Democratic margin in the House popular vote of 13 points, which is considerably higher than what the current congressional ballot, the generic congressional ballot that Elliott was talking about, is pointing to. That’s around D+8. But what’s so important is that these metrics are all pointing in the same direction. And when we back-test this, when we look at how these perform historically, if these metrics are pointing in the same direction, then they tend to collectively be pretty accurate in predicting the House popular vote, which points to a real blowout for Democrats.

Krugman: I’m supposed to pretend to be objective and say “just the facts,” but actually I do care about how this election turns out. And what’s a little bit reassuring is these are actual votes. It’s sort of like the two things that we have that are actual votes, which are primary turnouts, and then special elections are not just based upon: Do people actually pick up the phone or actually respond online?

But how worried should we be? Response rates on polling are so low now, and how well are the pollsters managing to deal with this environment where people [are so overwhelmed with robocalls]? I think two-thirds of the time when my phone rings, it’s from my good friend, Spam Risk?

Nir: Hey, I’m buddies with them too.

Morris: Look, I really don’t have a whole lot of comfort for people who just want the polls to be right, because we just don’t know. Like, we just don’t know if the decisions pollsters are making to combat non-response bias—and I will define that—if their methodology, their decisions are right. Because, you know, the capital-P problem in polling is that the people that answer the phone are very often not like the people who don’t answer the phones. They’re not representative of the whole population.

That was true in 1936, the first big test of scientific polling, when there were too many high-income, Republican-leaning voters. It was the same case in 1948. It’s just been true for the entire history of this instrument. And pollsters have a lot of really fancy things they can do to try to make the poll look like the population in terms of demographics, because we do have a national poll funded by the U.S. government called the census every ten years. So we know how many white or Black or young or poor people there should be. We don’t know how many Democrats there should be in a poll, or how many Democratic voters for the upcoming election there should be, because that election hasn’t happened yet.

So we’re just left looking for signs that the polls are wrong. And I’ll say a couple words about that, and also just exploring what potential outcomes could happen given the fact that polls have been wrong before. That’s what our election forecast does. But the polls are +8 right now. As I said earlier, average error in the generic ballot is three points. Let’s simulate a lot of different elections given that distribution of error, the three-point standard deviation on error. What could that election look like? And we answer that question through the election forecast.

So then the thing that’s left is checking—checking the polling data. One thing we do at Strength in Numbers is to take a predictive model of how people voted in 2024, trained on 2024 data, not our current polling data, and predict the Democratic or Republican lean of the people who are actually answering our survey. And there’s some potential bias in that. But the trendline is really important, because if the types of people who are answering your poll in 2025, for example, look like a Donald Trump plus-one electorate based off of your predictive models, and then that drifts over time to be, you know, Harris +1, Harris +4, Harris +5, then you have some good indications that the types of people you’re getting are too Democratic. But we do not see that. It’s a totally flat line.

That’s one thing we’ve done. And there’s some other things we can do to look for red flags of non-response ahead of time, and basically they’re all fine right now. So I’m as confident as I ever am that the polls are right. I’m never really all that confident. That’s why we simulate the election.

Krugman: And in both your guys’ modeling, how much are you also doing fundamentals? For instance, state of the economy. And also, economic surveys, which may be less politicized.

Morris: The only fundamental that matters in a midterm election is the president’s approval rating. That might not be a fundamental for an economist. But historically, you know, growth in GDP, change in real disposable income, change in employment rate—these things have very little correlation to midterm election outcomes.

Krugman: Okay.

Morris: The only real structural factor is who’s the party in the White House. And to some extent, how many seats do they hold right now? There tends to be a slightly larger punishment historically when the party in the White House holds a bunch of seats. But that’s maybe a gerrymandering factor.

So, I do think the economic discontent in the country matters, but it is probably being proxied through the polling right now. It’s not the type of thing that is historically predictive of a midterm election in the way that these things are predictive of presidential elections.

Krugman: That’s interesting. I’ve been maybe a little off on that. I knew about the historically very strong predictive power of the economic variables, but for presidential elections. And so are you saying there’s no independent effect, or really just that it doesn’t help you at all with midterms, even if you didn’t have the polling?

Morris: Yeah. There’s a very weak correlation between some of these economic fundamentals and the outcome of the election. If you then look out of sample at your predictive accuracy of a model that is trained on previous elections using these fundamentals and test out-of-sample, it’s even worse. And yes, once you start accounting for some other factors, like literally not even the president’s approval rating, just which party’s in the White House, then any relationship totally falls away.

Krugman: Which does kind of raise a bit of a question, right? If you drop the presidential approval rating, then you still don’t get very much, right?

Morris: So, yeah, I mean, there is a positive correlation, but I think we’re looking at a +0.4 correlation here.

Krugman: Yeah. But then there are the obvious fundamentals: there’s the price of diesel, there’s just generally really bad vibes about the economy. But how do we reconcile that with the fact that the midterms look at this point so much like a blowout? What’s driving that?

Morris: Well, I think the economic discontent is a big part of the equation here. You know, Paul, we’ve talked about the “vibe session” on your podcast previously. So, there might be some other residual, just disapproval or poor rating on the economy that people map onto politics. Just because this isn’t a historical explanation for why midterms go the way they go doesn’t necessarily mean that it doesn’t matter this time around.

The thing I’ve identified in U.S. politics over the last year and a half of writing my Substack is just a very endemic, anti-status-quo, anti-incumbent bias. In our polling, like 70% of people say that they face significant hardship from inflation even this year. So I think these are all jumbled together.

Nir: And also we have seen some pretty remarkable correlations. In particular, when you drill down on Donald Trump’s approval rating about the economy and prices in general. Elliott, we have these wild graphs that when you look at Trump’s approval rating and overlay it with the change in gas prices and just shift it just by a couple of weeks, because the polls always lag a little bit, then they match up almost perfectly. Gas prices really do seem to dictate how people view Donald Trump’s success on prices. And so there’s no question that this anger is feeding into his low approval rating.

Krugman: During the Bush years we used to talk about that correlation a lot. And then, it seemed to go away. And now it’s kind of back. We used to say it was really terrible that so much seemed to depend upon gas prices over which the president really has no control. Which is not exactly true at this point, though, right?

Morris: Yeah. Historically I think that is right, yeah. This president has done a remarkable job of very publicly taking very unpopular stances on the policies that Americans notice the most, which is a very unique combination. And that might be the most simple explanation for why this midterm is going the way it looks like it’s going to go.

Krugman: Yeah. I mean, a couple of questions related to that. How much do you think Trump is paying a price for all of those extravagant promises he made in 2024? Do people remember that? It’s always a question in my mind. For political pundits, we’re always saying, “You said this, and now you say that.” But does the public have a memory of these things?

Morris: I don’t have any quantitative data on this, so I will get into anecdotal territory here. I watched a segment on CNN—yes, I do watch cable news sometimes—earlier this week, with a reporter just interviewing people at a gas station in Florida and interviewing former Trump voters. And there’s some sentiment of dissatisfaction. People say, “Oh, things aren’t going well.” And maybe two out of the ten people they interviewed said, “I feel betrayed.” Someone said “hoodwinked.”

So I think that is definitely an undercurrent we could identify, but it would be very hard, putting on my quantitative hat here, to identify how many voters the Republican Party has lost over the last year and a half due to Trump making such large promises. There’s definitely, in hindsight, I think for him, a bit of an own goal to be promising prices are going to go down on day one. But maybe he doesn’t actually care about the midterms or 2028. He just wanted to be president again, in which case maybe it was worth it for him.

Nir: And we’ve seen the same thing with his immigration enforcement policy. How many articles have we read featuring quotes from ordinary Americans, many of whom voted for Trump and said, “Oh, I voted for him to deport the violent criminals. I didn’t vote for him to deport the lovely fellow who’s been living and working in my community for 20 years,” and we might all smack our foreheads or get very angry about what feels like a lot of naivete about that.

But I am struck, even to this point, we’re now in October of 2026, how many times voters say, “This is what I thought Trump was going to do and he did X.” So two years later, people are still saying, yeah, “This was what we thought the promise was and he’s breaking it.”

Krugman: Yeah. Again, you guys are tracking this systematically. I’m not. But there does seem to be an abnormally large number of people willing to say, “I made a mistake in 2024.”

Morris: Yeah, about 15%, I believe, of Trump’s voters on average, according to some polls. A Navigator Research poll came out last month, in September, that showed 20% of Trump’s 2024 voters say they regret—they will agree with a poll question using the words “regret” their vote in the last election.

I don’t have comparable numbers, but just in absolute terms, if Donald Trump loses a quarter of his vote, or even if he loses a quarter of that 20%, then the 2024 election is like a Dukakis-style electoral defeat for the Republican Party in 2024. That’s a meaningful slice of the electorate, even 4 or 5%.

Krugman: I am still thinking about “vibe sessions” and all of that stuff. And there’s one thing I noticed, and maybe it’s just one poll, but in the new AP-NORC poll, one of the things that was true consistently through the whole Biden era was that if you asked people, “How is the national economy doing?” they said, “Terrible.” You asked them, how is your local economy doing? And it was much better. And that gap is now gone.

It’s an interesting thing. I don’t know exactly where this fits in all this. But it is interesting that we’re no longer in a period where people are, you know, hearing bad stuff, but personally experiencing better stuff. Just generally, everyone is really pretty mad at the state of things now, which probably has something to do with all of these numbers. I think that was a complete non-sequitur given our previous conversation, but whatever.

Morris: No, I think there’s something to triangulate here. And in that same poll, I think Donald Trump’s approval rating on the economy among Republicans is at an all-time low in this new survey out this morning from AP-NORC.

But, I’ll back up from first principles here. Much of the conversation about the inflated or deflated consumer expectations and consumer sentiment over the past year has been about this idea that people are overindexing on the national economy, which may or may not be true. This survey seems to indicate maybe some movement towards convergence. If that’s the case, Republicans are, I hasten to say, dramatically more exposed to inflation in diesel and gas prices.

Krugman: Yeah.

Morris: And so I think we’re at the point in this cycle where after 15 months of inflationary statistics and inflationary news, Republican voters would be now coming around to the idea that this isn’t working out for them the way that they wanted.

In our Strength in Numbers September poll, we also found Trump at an all-time low approval among Republicans. So, our mental model is that maybe independents are super rational and everyone else isn’t. And so Democrats and Republicans wouldn’t be reacting to economic news as much as independents. That’s just the traditional understanding. So it would take a pretty cataclysmic-like event for Republican voters to then be saying, “I disapprove of this Republican president.” About 20, 25% of them do now. So I think these are all sort of related in my mind and my sort of mental model of voter psychology at this point.

Krugman: Okay. So backing up from there, we think we have a pretty good idea of what’s going to happen in the midterms, maybe—definitely in the House, and maybe the Senate—but Elliott, you’ve written a lot about this, and I think David as well—the extent to which people treated 2024 as a real seismic moment, defining a fact that American politics is never going to be the same. That’s really awesomely wrong, isn’t it?

Nir: Yeah. I mean, there’s this old saw in baseball that the good teams win close games, and that’s just not true. The good teams win blowouts, and the pundit class and so much of the media and frankly, a lot of politicians, including politicians in the Democratic Party, they failed to grasp that Trump won by one and a half points in 2024. That’s really about as close as it gets.

And I think that because the stakes of Trump winning felt so extreme and so dramatic, and you knew it was going to be so much worse than his first term, that kind of mentally or maybe even emotionally translated into something that felt like an absolute drubbing. But electorally, it simply wasn’t. And anyone at the time who was crowing about some Republican conservative era of cultural dominance, a massive sea change, wound up being wrong. And there were a lot of people who were saying, no, this is totally wrong. People like Elliott. Elliott was outspoken on this. He said it was a close election. It doesn’t mean what you think it means. And folks who hew to that view were totally vindicated.

Morris: Well, lo and behold.

You know, I actually give people a lot less credit than David did, given my previous employment situation. I think that there’s just an endemic bias in Washington, D.C. Well, there’s a couple of them. One is to make every story the biggest story. If you want to run like three or four days of TV coverage of Donald Trump because people are tuning in, you can’t just say, “Oh, he won a small election.” He has to have won a big election so that you can do that TV coverage that is performing well so you can justify seeking the ratings journalistically. I think that’s part of it.

And maybe you saw this in your past lives as well—a tendency for pundits to draw patterns out of one or two data points for the sake of narration. And I think that’s something that’s going on here, too. But I think the real meat of it was just that a lot of journalists write stories to seem smarter than everyone else. And one way to seem smarter than everyone else in Donald Trump’s victory is to just say that the liberal class of journalists were all wrong: “Look at this. Look at this huge victory. We now have to adapt to this new world we’re in where Donald Trump is winning huge victories so that we continue to not be—not be like a Democratic liberal herd.” That’s an instinct that is especially strong in the more rationalist, empirical publications, say, and I think it breeds a lot of bad storytelling.

The other thing that was a product of this was the constant humdrum about illiberalism on college campuses, starting in 2010 and right up to 2022, 2023. Well, what’s the real illiberalism on college campuses after 2025? What’s the real threat to free speech on college campuses after 2025? It’s probably not students preventing certain speakers from coming. It’s probably more the attacks on freedom of collegiate education writ large. So I think that that is another example of journalists missing the broader story going on for the sake of trying to appear smart to their peers.

Krugman: Yeah. When I was more part of the sort of conventional, journalistic milieu—which I never was very much, but somewhat—the idea was everything had to be counterintuitive. And then being against liberal pieties, and so precisely because Trump was so horrible from the point of view of any kind of conventional liberal pespective, therefore it seemed important to talk up how epoch-making and transformative he was. That was part of being counterintuitive.

And in some sense, the fact that somebody so self-evidently harmful to have won had to mean that there was something supernatural about his appeal, when in fact, the reality is that most normal people don’t actually pay that much attention.

Nir: You know, Paul, when I first started reading you, it was back when you were at Slate. And that reminds me of the infamous Slate pitch. Not that you were ever guilty of that, but the contrarian, counterintuitive, anti-liberal piety article that Slate made famous.

Krugman: Yeah, exactly. Most people won’t even remember, but the Slate pitch was, roughly speaking, “Here’s why running down cyclists is a good thing.”

Morris: It’s the “Um, actually...” article, right?

Krugman: Right, yeah.

Morris: And this goes back to that sort of anti-incumbency bias that stems from economic hardship. There’s just a tendency for voters to want to throw the bums out, to borrow that old phrase. There’s just an endemic negativity in the American public right now. And voters really want some big changes. I mean, in our survey, like 55% of voters every month—this percentage barely changes—say that major disruptive changes are needed to the nation’s political system and economy, whereas like 7% of people say things are going well for them. There is a middle option, for what it’s worth. But like those extremes, there’s a very wide imbalance in those extremes. And you would expect from that a tendency to punish the incumbent party, or pretty much whoever is in charge.

Paul, in one of our earlier conversations last year, we also talked a lot about the low-information voters who I know you’ve written about. This is a good part in our conversation to remind all the high-information listeners out there: Most Americans are not like us, not like you. About 50% of Americans say they do not get news on a weekly basis.

Krugman: Isn’t your definition of the information voter—you think it’s some kind of insult, but you just mean people who don’t know which party controls Congress, right?

Morris: Right. So this is sort of the low engagement voter. And then there is the actual low-information people, which is not really a putdown. Like we all have stuff going on. It takes a lot of energy to know what’s going on in politics.

On our survey only about 80% of people can say who controls both houses of Congress. We will be asking this question on a recurring basis in an upcoming project that is yet to be announced, a weekly survey at Strength in Numbers, which is exciting, and I imagine this number will be really helpful for us to track Democratic and Republican vote share, Trump approval and such by those voters who are or are not high or low information, or high or low engagement.

Those are the people who decide elections, and this is my broader point. Those are the people who decided the 2024 election: People who sort of remembered in the folds in their brain that things used to be better, so let’s go back to those things. Now, they’re in the sort of find-out stage of the F.A.F.O. And I think they will inevitably vote Democrats in charge and expect big changes.

And the three of us can probably forecast the number of major transformative changes that will come from Democratic Party leadership in Congress under the Donald Trump presidency, or even under a unified Congress in 2028, which is probably not a whole lot, let’s be honest. I imagine the American people will then punish the Democratic Party just as harshly.

Krugman: Yeah, well, certainly the next Democratic president, by inheriting reality, will simply be at a disadvantage.

I’m going to ask one last thing: the AI data centers. This is the political storm that I did not see coming. The apparent intensity of opposition is a surprise. And Trump—I’m sure that he has absolutely no idea about how the models work, but—

Morris: “Let data reign!”

He had a tweet where he said, “Let data reign,” right?

Krugman: Yeah, but is this going to matter? I’ve been seeing some people saying that at least in a couple of swing states, it might actually be an important factor.

Nir: I think so. Like you said, Paul, this issue almost feels like it came out of nowhere. The intensity of the anger. When you look at the polls now, I mean, huge majorities say that they oppose, for instance, building new hyperscale data centers in their neighborhoods or in their areas. And even six months ago, the numbers were much less opposed. And there is no question in my mind that so many voters are using this as an opportunity to take a stand against so many of the things that they don’t like: you know, the tech bro oligarchs, AI being shoved into every last corner of our lives. And these data centers are a stand-in for that.

And, boy, there have been so many ads run on them, Paul. And most of the ads are being run by Democrats. It is definitely one of the top issues that we are seeing coming up this year. It’s not just gas prices. It’s not just Trump’s mass deportation policy. Data centers are definitely playing a real role in some of these races. And what we’re going to see is a lot of races, especially governorships in Midwestern states that have leaned red in recent years, they’re probably going to be pretty close. And I think that when you’re on the wrong side of an issue like this, that you could lose a close race as a result.

I look at the example of Governor Greg Abbott, who has never had to sweat a general election. Now he’s running against Democrat Gina Hinojosa. Polls show it very, very close. And all of a sudden, Greg Abbott, this ultraconservative team-player guy for the MAGA movement, is coming out in favor of data center moratoriums and issuing executive orders to that effect. And so I think there’s no way he’d be doing that if he wasn’t afraid that this was a real issue that has lit a huge fire under voters.

Morris: I mean, he was approving the Starbase project for Sam Altman and Oracle, I believe in Texas, just this time last year. And now he says you have to go through “special approval,” which just means delays, right?

One place this might show up in our polling data from September is our question: which party do you think cares more about people like you? This was really bad for the Democrats in 2024, for example. Now they have a pretty large lead. About 38% of people say Democrats care more about them. 26% of people say the Republican Party cares more about them. It’s very interesting. There’s the next category of people, the 7% who say both parties care equally about them, and I guess people could be reading that as both parties care a lot about me, or neither party cares about me at all. 22% of people say neither party cares most about people like me. It’s kind of a comparative question. So maybe we’ll change that wording now that I’m thinking about it.

But I imagine getting up at his press conferences with the tech billionaires, hosting them at state dinners, tweeting—or I guess, truth posting via Truth Social? Truthing? Truthers? Yeah. Truthing that “Let data reign” comment and saying communities are going to be backwards if they reject data centers, it makes sense to me that they would say, “Oh, that guy, the Republican Party guy who’s on the wrong side of the biggest land use issue right now, which is dramatically unpopular—that party doesn’t care about me. Maybe the Democrats do now.” But there is a notable seesawing in that question across the elections, too. It just goes back to the sort of anti-incumbency undercurrent.

Nir: Yeah. And Elliott, I think, to that point, and Paul, getting back to what you said earlier about how the punditry just totally misunderstood 2024, how much talk did we see about polls showing that the Democratic Party’s brand was in poor shape? Those polls are absolutely true, and those voters’ feelings were absolutely understandable, but it simply missed the fact that people have to actually vote for one or the other. And even if they dislike both parties, these so-called double haters, they have to make a choice or stay home altogether.

And so I think ignore the polling on political parties’ standing and look at the polling of voter intentions. And that’s going to tell you a totally different story. People can still vote for Democrats even if they’re not happy with them.

Krugman: Yeah. Neither of you, happily for your own sanity, is in the business of modeling the mind of Donald Trump.

Morris: Thank God.

Krugman: But to choose this of all issues to really go on. I guess I thought maybe he doesn’t care at this point, but I would have thought that this would be a good time to pretend to be worried about data centers and AI and all of that.

Morris: Well, we have to consider maybe—I don’t want to do too much putting myself in the mind of Donald Trump because, oh, God. But we can think about the information environment he’s operating under, which is like the CNN headlines: about 99% of self-identified MAGA voters still approve of Donald Trump. You say, okay, well, that’s tautological. Whatever. He’s just insulated by these tech billionaires in and out of the White House. That’s practically his social circle at this point.

Nir: Right.

Morris: And then to the extent he gets information, it’s either fed to him by staffers who spend all day on X or Truth Social or whatever. Streaming news, or whatever cable news service currently exists for the right. But especially X, which is very endemic with super right-wing tech bros on there. I mean, Elon Musk being the number one example. If you’re operating any type of business and that’s the information environment you’re in, maybe you just don’t believe all the polls for the midterms. Maybe you don’t believe that people see these moves as unpopular.

Not to give him any credit, right? But I think the “X factor,” as David has called it, I believe, is a real problem in American politics. And you can see it operating chiefly in the chief executive.

Krugman: And there we are. Okay. Well, let’s hope that they’re in the bubble and we aren’t. Thanks a lot. And we’ll probably want to have a conversation after the midterms and figure out now what.

Morris: I’ll tell you if they happen.

Krugman: Yeah, if they happen.

Morris: That’s a joke. Hoping they happen.

Nir: Oh, Elliott! Don’t go there.

Morris: Okay, I’m kidding.

Krugman: No, I’m quite sure the elections will proceed. And then we’ll have a coup that will overturn it. But never mind. Sorry.

Nir: Guys! Let’s pull it together!

Krugman: Ok, ok. Well, thanks so much for speaking with me.

Morris: Always a pleasure.

It’s election season. Here’s my pitch to you to support this work with a paid subscription.

The next four weeks are like an extended Super Bowl for election analysts. It’s when our work to root your understanding of politics in the hard data matters most. There will be a flood of polls, narratives, and hot takes between now and Election Day that draw attention and take our eyes off the ball. My goal is to help you separate the real signals from the noise with independent, data-driven analysis.

If you support that mission, consider becoming a paying subscriber to Strength In Numbers. You’ll get access to our private community on Discord and get additional premium analysis at least once weekly. Paying members of this community keep Strength In Numbers running by supporting the time and data collection it takes to do analyses like this one.

Because election season is also when press and search-engine referral traffic to independent sites like SIN is the highest, Oct-Nov is also our best chance to recruit new members to this community. If you find value in this work, a paid subscription makes it sustainable over the long term.

Paid subscriptions are what fund our surveys, data analysis, and simply the time it takes for me to do this work carefully. If you’ve been thinking about joining, now is a particularly good time.

Upgrade to support Strength In Numbers

Here’s a headline from our September poll that got overlooked:

For the first time in our monthly polling here at Strength In Numbers, a majority of Republicans and Republican-leaning independents disapprove of Donald Trump’s handling of prices.

Today I want to share two charts. Well, actually four charts, but the piece is mostly about the chart below. It shows, on the top, the retail price of diesel since Donald Trump became president; and, on the bottom, Donald Trump’s approval rating on prices among people in our monthly Strength In Numbers/Verasight poll who self-identify as Republicans.

Since May 2025, the price of diesel fuel has risen from an average of $3.50 across the whole country to $6.29 as of September 28, 2026 (the next weekly update from the U.S. Energy Information Administration comes out October 6).

In September 2026, the price of a gallon of diesel hit an all-time high.

As I wrote in my write-up of our September poll, an overall 6-point net drop in Trump’s approval from August was driven almost entirely by Republicans. Among Republicans and Republican-leaning independents, net approval of Trump fell from +54 in August to +43 in September. That’s an 11-point drop in a month, to the lowest level in 16 waves of our monthly poll. A year ago, Republicans were at +68!

Last month, Donald Trump’s approval rating for handling prices and inflation also hit a new low among self-identified Republicans. The chart above shows his rating at −8.

You know who uses a lot of diesel? Farmers and truckers. You know which occupations have a large share of Republican voters? Farmers and truckers!

This map, which I found via Google at https://valverideflow.com/diesel-prices, shows diesel is particularly expensive in several states holding Senate elections this fall, particularly Ohio, Iowa, and Michigan. Kansas and Nebraska too, though to a lesser extent. (Diesel is also expensive on the West Coast, but so is everything else.)

My case in this week’s Friday column is that prices are a unique vulnerability for Trump and Republicans with their base. In fact, prices are the only issue Republicans rate Trump net-negative on.

Subscribe now

Republicans rate Trump net-negative on prices

This graph shows Trump’s net approval rating — the percent approving of him minus the percent disapproving — in August and September 2026 among Republicans on 12 different issues. From August to September, the largest decline in Trump’s ratings was on prices, going from +5 to −8. That move is outside the surveys’ respective margins of error (about 5 points on net approval).

This chart suggests Trump and Republicans are uniquely vulnerable on prices. The diesel price hikes make inflation even more salient for everyday Americans.

But the president’s weakness extends beyond his ratings on prices. My state-level model also estimates that Trump’s overall job approval is underwater in many states he won comfortably.

My MRP model has Trump net-negative in some of the most deep-red states in the country. From May 2025 to September 2026, Trump’s approval rating has fallen the most in...

1. Wyoming (+22 in May 2025 to −3 in September 2026)

2. Utah (0 to −23).

3. Kentucky (+17 to −5).

4. Mississippi (+2 to −20)

5. West Virginia (+31 to +9)

6. Alabama (+15 to −6)

7. Arkansas (+17 to −4)

8. Idaho (+19 to −1)

9. North Dakota (+19 to −2)

10. Tennessee (+16 to −3)

Here’s the data from the graph above if you are just looking at states with Senate races, sorted by Trump’s approval rating among his own 2024 voters:

1. Iowa (+42 among Trump’s 2024 voters; −24 among all voters; open seat)

2. Mississippi (+47; −20; Hyde-Smith)

3. Wyoming (+50; −3; open)

4. South Carolina (+51; −17; appointed, Darline Graham)

5. Texas (+51; −21; open, Paxton nominee)

6. Louisiana (+52; −14; open)

7. Kentucky (+54; −5; open)

8. Kansas (+55; −16; Marshall)

9. North Carolina (+56; −21; open)

10. Montana (+56; −12; open)

Note the placement of Iowa and Texas here.

It is unlikely that anxiety about prices, and backlash against Trump’s handling of the issue, would only impact the president. This all may be hurting Republican nominees for the Senate.

In the chart above, I’ve bolded states that have competitive Senate races this year. Many of these are the states on that map above showing average diesel prices, especially Ohio and Iowa.

Trump won every one of those six states by double digits in 2024. Yet the Republican Senate candidates are now tied or trailing in the polling averages.

Trump won every one of those six states by double digits in 2024. Yet the Republican Senate candidates are now tied or trailing in the polling averages.

These elections are not over. And issues other than diesel matter. But it is notable that a majority of Republicans and Republican-leaning independents now disapprove of the president’s handling of prices.

That has the potential to turn soft partisans into Democratic voters, and to turn off strong partisans from voting.

Republican candidates can hope that voters care more about other policies, where Trump scores better ratings. But on the cost of living, the number-one issue among voters of all parties, the president and his party have something more basic to do: They have to win back their own voters.

With charts like the above showing all-time highs for diesel prices across the country, that’s going to be hard. And with only 32 days before Election Day, time is running out.

This post was free for everyone. But paying subscribers pay our bills and make work like this possible. If you find this kind of data-driven analysis valuable, consider supporting the work with a paid subscription.

Upgrade to support Strength In Numbers

Related Articles

On this week’s Strength In Numbers podcast, Elliott Morris and David Nir walk through the current electoral state of play just over one month out from the 2026 midterm elections.

This video is free to watch, but it was not free to produce. If you value this podcast and want to help keep it going, please consider becoming a premium subscriber to Strength In Numbers. Paid subscriptions keep the show ad-free, fund the newsletter, and ensure we have the time it takes to do original data-driven analysis of politics and elections. Paying subscribers also gain the ability to send in questions during our livestreams, so you can directly shape our conversations on the podcast.

Upgrade to support Strength In Numbers

Share

A few big takeaways from the conversation:

• The House at 99% is in “spooky probability territory™,” but the seat-level math adds up. In Elliott’s forecast model at FiftyPlusOne, Democrats need only about a 3.5-point national margin to win the majority, and they're polling 5 to 6 points better than that. For Republicans to hold on, the polls would have to move about five times more than they historically do in the last month, or miss by roughly twice the usual error.

• Democrats have a 73% chance at the Senate because of the long shots, not the toss-ups. If you only look at the competitive seats, the Senate looks like a coin flip. But election forecasting models also pick up on Democratic upsets in the 20-to-30% zone (Florida, South Carolina, Kansas, Nebraska) that drag their median up to 52 seats.

• David makes the case that red-state Republicans have forgotten how to campaign. In states like Kansas and Texas, GOP candidates are making big missteps that would be unlikely for someone accustomed to running competitive races. Can you lose your general election campaign muscles?

• Elliott and David answer listener questions about cross-over voting and modeling polling errors.

Thanks to everyone who joined us for the weekly livestream on Substack. We’ll be back next week!

As a reminder, paid subscribers to Strength In Numbers get to participate in our live Q&A! If you missed our video livestream, you can watch it by clicking play on the web version of this post at gelliottmorris.com.

You can also subscribe to us on your favorite podcast app to listen on your own time. And if you do listen via one of those apps, please drop us a five-star rating and review if you feel we’ve earned it — it really helps people discover the show!

Subscribe now

Share

You can read the transcript of our conversation by clicking the headline of this article to take you to the web version of the podcast, then clicking the button just below the byline that looks either like a piece of paper or is labeled “Transcript,” like so:

If you’re a reader of Strength in Numbers and haven’t yet subscribed to David’s newsletter, head to the-downballot.com.

Subscribe to The Downballot

And if you’re coming from David’s site, subscribe here to get the numbers behind the news.

Subscribe to Strength In Numbers

Every month, Strength In Numbers lets subscribers to this newsletter suggest questions for us to run in our recurring poll with Verasight. This month, several readers suggested we poll about the popularity of the federal government sending agents of Immigration and Customs Enforcement and other agencies to local polling places this November under the guise of “election security.”

News reports indicate that recently uncommon levels of federal “oversight” could take multiple forms. The Justice Department said in August that it is sending a record 1,000 federal monitors to polling places this November, up from 714 in 2024 and 289 in 2022. Homeland Security Secretary Markwayne Mullin has said federal agents would show up “if there’s a threat to a polling place or to serve a warrant.” This is a retreat from earlier claims by the president that ICE could “surround the polls.”

More extreme actions are possible. A draft executive order reported in February would let Trump declare a national emergency and assert federal control over voting machines nationwide. Back in February, Trump said in an interview with the New York Times that he should have ordered the National Guard to take possession of voting machines in 2020 in order to find evidence of fraud.

Among people I know in Washington, DC, and local election officials, there is a lot of anxiety about the potential outcomes above. So I decided to include this subscriber’s question in this month’s poll.

Before we get to the results, here’s a very short pitch to you to support Strength In Numbers.

Original polling like this is funded entirely by readers, who are devoted, as I am, to the causes of democracy and to using polling data to give a voice to the people in Washington. If you believe in our mission/mantra- that there is strength in numbers - become a paid subscriber to support the site.

Upgrade to support independent polling

In keeping with the usual Deep Dive rules, I have paywalled this piece about halfway down, but left the topline findings in front of the paywall because of their significance to the public.

We asked Americans whether they would support or oppose five things the federal government could do around this November’s elections — each carrying very different stakes.

1. Sending federal Justice Department monitors to polling places. This is already happening. It used to be a routine, bipartisan practice — but it carries more risk under a DOJ whose civil rights division has lost most of its career staff and strictly follows the orders of the president.

2. Having federal officials take custody of voting machines or ballots after Election Day. This isn’t legal. States control the chain of custody for ballots, not Washington. This is a democracy-ending scenario.

3. Sending National Guard troops to polling places to monitor the election. Historically rarer than DOJ monitors, and it carries real tail risk given how freely Trump has deployed the Guard domestically this term.

4. Having the military take custody of voting machines or ballots after Election Day. Another outcome that isn’t merely “bad” but a legitimate constitutional crisis.

5. Sending immigration enforcement (ICE) agents to polling places. Homeland Security keeps hedging its “no plans” promise, but this one could easily happen.

Here is net support overall and by political party for each of these potential actions:

In our poll, the DOJ monitors basically split even at net +1, then things fall apart. Federal agents taking custody of voting machines is underwater by −10, National Guard troops at −17, military custody at −27, and ICE agents going to the polls at −30. Additionally, as we see with Trump’s approval rating overall, 49% strongly oppose ICE agents at polling places* and 44% strongly oppose the military taking custody of ballots.

Political independents align closely with the average American. While Republicans support most of the above measures, they are comparatively cold on the idea of the military seizing ballots.

Subscribe now

An ICE chilling effect on turnout

We also asked respondents to tell us whether, if ICE agents were present at your polling place, that would make them more or less likely to vote.

Those results are presented in the chart below:

Read more

Upgrade to support Strength In Numbers

From time to time, Strength In Numbers accepts sponsors for free posts sent out to all readers (never paywalled articles). These funds help offset subscriber churn and pay for things like data acquisition and staff members, including people who help with the podcast and survey research. This post contains one such sponsorship, clearly denoted after the introductory portion of the article.

Three weeks ago, Republicans held the first televised midterm convention in American history. For two nights in Dallas, the president and his allies attempted to upend historical political gravity and turn around what is currently a losing campaign to keep their House and Senate majorities.

On Sunday I showed how voters are reacting to the most-covered takeaway from the convention, an announcement* (*attempted bribe) by the president that he will send $5,000 to every American if the Republicans keep the House and Senate. In our September Strength In Numbers/Verasight poll, we found that sending money to voters was generally popular, but once respondents were told the funds were conditional on Republicans winning the midterms, they turned against the proposal.

In today’s article, I want to answer a narrower question: Did the convention work as a news event at all, as normal presidential nominating conventions do? The whole point (from the media point of view) of a party gathering normally is to get the press covering the party, on the party’s terms, for a sustained period of time. Did the “Trumpapalooza” midterm convention accomplish that, or not?

The answer is no. According to my analysis, the Dallas gathering accounts for a small number of front-page stories the political press produced since the convention was held, and coverage completely faded within a week. Compared to press attention to past conventions, the first-ever midterm convention was a failure. And it failed to displace other stories that are more problematic for the president and his party.

Today’s article is a bonus article, written on top of the usual 4x weekly schedule I run here at Strength In Numbers.

If you are a devoted reader of SIN, please become a paying subscriber to keep funding work like this and growing our operation.

Subscribe now

An archive of major headlines going back to 2016

For the last couple of years I’ve been tinkering with a constantly updated archive of the news stories that are popular with the political press. I have set up a fleet of computer robots to collect the contents of the front page of the news aggregator memeorandum.com every five minutes. Memeorandum calls itself “an auto-generated summary of the stories that US political commentators are discussing online right now.” Presumably they accomplish this by scraping social media and other news aggregators and extracting the links, but I’m not really sure.

I’ve also backfilled the site’s own archive pages, so the database now covers all memorandum .html updates from 2016 on. This dataset contains more than 200,000 links to unique front-page stories, 24,988 of them from 2026 alone. (I say “unique” because sometimes a story will lead memorandum.com for a couple of days.)

To sort the headlines, I wrote keyword rules for 18 of the year’s major stories — Iran, ICE and immigration, the Epstein files, the Supreme Court, the midterm campaign, the convention, and so on — and audited every one of them by hand. A headline can match more than one subject, and plenty match none. I also score the tone of every headline with a pre-trained sentiment model. (A more sophisticated version of this categorization might use LLMs, and I’ll consider updating the program if there is enough interest.)

What the archive measures is what political media were writing and linking about, not what readers read. And that is exactly the thing a convention is supposed to move.

As far as I know, nobody else in independent data journalism is running a news-monitoring program like this — and it only exists because paying subscribers to Strength In Numbers fund the unglamorous work of building the underlying tech. If you’re a paying subscriber, thank you for making work like this possible! If you’re a free reader, consider supporting the site with the button below. You’ll be supporting independent, data-driven political analysis, and you’ll also get an ad-free version of this newsletter.

If you are trying to stay on top of the news like me, free tools like memeorandum are a good start, but for serious analysis you need Ground News.

Ground News is a website and app I use constantly that helps me identify when stories are being pushed by one side of the ideological spectrum. Ground News gathers news articles from over 50,000 sources worldwide and shows you a measure of each outlet’s factuality, ownership, and political bias. You can also filter news by whether an outlet is owned by private equity or a corporate conglomerate — or is an independent outlet, like this Substack!

Ground News is a smart news aggregator to help you deal with the challenges of our modern information ecosystem. The constant news monitoring and feature filtering help me keep my own news diet attuned to facts, not spin. I think these tools would be incredibly useful to my audience — informed readers trying to find the facts in the news. Today, they’re offering readers of Strength In Numbers 40% off their “Vantage plan,” which gives you a personalized feed as well as data on factuality, partisan bias, and funding information for every news outlet.

Get 40% OFF Ground News today

The top stories from 2026 are ICE and Iran

Here are the top categories of stories in 2026 so far, by subject and headlines per day. Immigration and ICE led the first two months, then the Epstein files took over, then the war in Iran has carried most days since. The late-term Supreme Court decisions popped up in June and July, and the midterms finally started showing up in late August.

To put some numbers behind this: With 2,481 front-page stories and 9.9% of everything the political press linked to in 2026, Iran is the story of the year. ICE and immigration are second at 2,113 stories and 8.5%. The Justice Department, the FBI, and Epstein come third at 1,447. The midterm campaign itself managed 805, barely ahead of Supreme Court coverage at 791.

The midterm convention is the green line along the bottom. While there is a visible spike on Sep 9-10, it never led the news with just sixty-five stories linked at memeorandum.com, 16th of the 18 subjects I tracked in 2026. If you count Trump’s $5,000 promise as convention news, you get 91 total stories and 0.4% of the 2026 news agenda.

But we shouldn’t expect the convention to overshadow events that happened earlier in the year. A more useful statistic is the percent of all stories that are devoted to each topic, on the day coverage of that topic peaked.

Subscribe now

The convention was a minor story even while it happened

The convention’s single best day of coverage was Sept. 10, when it took 11% of the front page in isolation and 23% when you include headlines about Trump’s $5k promise. Here’s how that compares with every other subject’s biggest day of the year:

For comparison, the invasion of Venezuela to capture Nicolás Maduro was 76% of the front page on Jan. 3, Iran’s biggest day was 58%, and on the day ICE agents shot and killed Renee Good in Minneapolis, immigration was 57% of all stories that showed up at memeorandum.com. That makes it between the eighth and twelfth most covered story of the year.

Subscribe now

Convention coverage usually lasts two weeks. Trumpapalooza faded in six days

Because we have data going back to 2016, we can also compare coverage of the Trumpapalooza “convention” against the presidential nominating conventions in the archive, each aligned on its opening day. Trump’s convention is the thick green line at the bottom of the chart. (I excluded 2020 because of disruptions to the usual convention proceedings due to COVID-19.)

In the two weeks before and after the 2026 midterm convention, coverage of the event made up 1.8% of front-page stories. The previous low was the 2024 Republican convention in Milwaukee, at 3.7%. Philadelphia in 2016 got 3.8%, Chicago in 2024 got 5.0%, and Cleveland in 2016 got 7.0%.

And the chart shows how fast everyone moved on from Trump-fest. The last front-page story about the convention ran on Sept. 14 (”’Trump Dividend’ is the new Trump Steaks”), and from Sept. 15 through Sept. 25, eleven full days, not one convention story hit the front page of memeorandum.com in any five-minute snapshot I archived. (Two stories about the checks did, on Sept. 15 and 17.) Over the same eleven days after their own conventions, the 2016 RNC in Cleveland got 3 stories, 2024 DNC 7, 2024 RNC 8, and 2016 Democratic convention in Philadelphia got 11.

Now there is an obvious objection to this comparison: Nominating conventions have multiple days of real programming and a presidential nominee; we shouldn’t expect a similar scale of coverage from a midterm convention with less news potential. But that’s partly the point: Without potential big news events, the expected earned media bump from the gathering should have been proportional to the expected impact of a normal Trump rally.

And as our September Strength In Numbers/Verasight poll showed, the one big headline that was generated by the convention is dramatically unpopular with voters. 62% of adults say Trump promised his $5k victory dividend mainly to win votes, and 71% say it’s inappropriate for a president to condition payments on his party winning elections (including 53% of Republicans).

This, while the president and his surrogates failed to defang the affordability issue that’s dragging down his party. While the speeches delivered at the convention were obviously geared to shift the national political conversation toward other issues — in his remarks, J.D. Vance spoke at length about transgender sports and “radical left,” and spent 10 minutes addressing a feud with Michigan Democratic Candidate Abdul El-Sayed over a throwaway line about J.D. Vance’s “papaw.”

Subscribe now

Stories about affordability have only increased since Trumpapalooza

On the convention’s first night, Trump told the crowd, “Our costs are coming way down.”

Here’s the proportion of all news stories devoted to major topics this year. The convention is the blue bump in September (Trump’s 5k pledge is in orange). The red band on top is the cost of living — including coverage of inflation, the Fed’s actions, prices, groceries, rent, gas and diesel — and it’s the one subject currently growing in coverage:

Across June, July and August 2026, cost-of-living headlines accounted for an average of 1.8 stories per day at memeorandum.com. Through Sept. 25 they’re running at 3.4 a day, an 88% jump. September has carried at least one front-page story about prices for 25 of the past 28 days.

Most of that increase in coverage of inflation/prices is about gas/diesel. Energy headlines have more than quadrupled since August, from 0.45 a day to 2.0, the highest since the oil shock in March. This includes headlines like “Diesel jumps above $6 a gallon, adding to inflation pressure.” “A ‘sleeper issue’ in red states this fall: Diesel prices.”

There have been sixteen front-page headlines about diesel on memeorandum.com in five days. Meanwhile, the convention itself and Trump’s $5k bribe haven’t made the front page since Sept. 14.

The Republicans staged a first-ever midterm convention this year in an attempt to activate low-turnout GOP voters ahead of likely defeat this November. I have explained why the strategic logic here was flawed, and shown that the midterms landscape has only moved toward Democrats since the gathering ended.

In this piece, I have shown that the midterm convention also failed to earn media coverage in the way past midterms do. While Trump and other GOP leaders clearly aimed to move the press off the affordability story, coverage of the convention itself and the president’s $5k offering was short-lived — and stories about prices have only circulated more in the two weeks after.

Aside from prices, the notable feature of the other lead stories in 2026 — Iran, ICE, and Epstein — is that they all divide the Republican coalition, Iran in particular. This is why I have said the GOP is effectively being torn apart from two ends, both the middle and the far right.

In the end, this is a good reminder that the midterm rout for the party in the White House is as close to a law of physics as we get in politics. With a president at a 35% approval rating, it was foolish to think that giving him more time in front of the average voter would help his cause. According to our election forecast at FiftyPlusOne.news, the Democrats are almost a sure bet to win back control of the House, and are now favored (at a 7-in-10 chance) to flip the Senate.

If you like the free version of SIN, please consider becoming a paying subscriber to support our work. In exchange for your currency, you will get access to our private community and additional premium analysis at least once weekly. Paying members of this community keep Strength In Numbers running by supporting the time and data collection it takes to do analyses like this one.

Upgrade to support Strength In Numbers

Related Articles

Upgrade to support Strength In Numbers

Hi everyone, Elliott here with a short programming note: I have opened up comments on the Sunday roundups so both paying and free readers can leave questions and recommended links directly to me and the whole community.

If you have a question or comment you want answered/addressed in any Strength In Numbers output, including the podcast, Deep Dives, or free Friday and Sunday articles, leave it below!

1. This is why a midterm convention is a dumb idea, part one

Earlier this month, as “Trumpapalooza” opened in Dallas, I argued the strategic logic of this first-ever midterm “convention” was entirely backwards. Republican strategists like Liam Donovan and Patrick Ruffini said a big Trump-centric gathering would help activate low-propensity Republicans the party needs in November.

The problem with this strategy (and this was obvious in the pre-convention data) is that Republican voters who are not committed to voting in November actually disapprove of Trump more than the Republicans who were already planning to vote. And polls show the issue they care most about is the cost of living — which Trump called a hoax on TV, and no other speakers mentioned. I covered this data on the podcast with David Nir two weeks ago and again in that Friday’s Chart of the Week.

Now our September Strength In Numbers/Verasight poll, fielded Sept. 16–21, gives us a few more data points to assess the success of Trump’s convention. These results have not been released previously.

I have also compiled historical data on news headlines to see if coverage of the convention broke through other stories, much in the way presidential nominating conventions tend to. I’m saving that data for a standalone article to be published tomorrow.

The data here is almost uniformly bad for the president. And in the weeks since the convention, his polls have only gotten worse.

The $5,000 offer backfired

The big story out of Dallas was Donald Trump’s pledge to send every American adult a $5,000 check if and only if Republicans keep control of Congress. This type of thing is tricky to poll about, since people generally support transfers, but we had a reasonable belief the conditioning on GOP victory would turn voters off the idea. So we ran a survey experiment where respondents were broken into three groups: One third of respondents were asked whether they support “sending $5,000 payments to American households,” full stop. A third were told Donald Trump proposed it. And a third got the deal as he conditioned it in Dallas: the checks go out if Republicans win in November.

When the payments were described as a neutral policy, 53% of adults said they supported them while 35% opposed them, a net of +18. When we told voters it was Trump’s idea, support ticked down to +12, a difference inside the margin of error. But when we described the checks as he actually offered them, voters were opposed, at 35% support and 54% oppose, net −18.

And among the voters who didn’t say they were definitely voting this fall — the kind of low-propensity voters the convention was supposed to reach — the swing is bigger still, from +39 to −5.

Voters aren’t confused about the motive, either. 62% say Trump made the promise mainly to win votes in November, and only 13% think he genuinely intends to send the money. Moreover, 71% of adults say it is inappropriate for a president to promise government payments that depend on his party winning an election, including 53% of Republicans. Only about a quarter think the checks would actually go out even if Republicans win.

We also asked voters whether Trump’s promise makes them more or less likely to vote for Republican candidates for Congress. 28% said less likely, 13% more likely, and 53% said it makes no difference. But many voters in that 13% already told us they were voting Republican in November. Fewer than 4% of adults say the checks make them more likely to vote Republican, and aren’t already there. This offsets with large losses among independents, who said by an 11-point margin Trump’s scheme turned them off.

These are bad numbers for what was essentially the only major story from the entire event. In part 2 of this article, publishing tomorrow, I dig into a new database on news headlines shared by the press on social media to show the failure of the convention to act as a forcing mechanism for earned media the way conventions normally do.

Share

For more on this, see the great piece by Seth Masket published last Monday. He shows that the numbers for Trump and Republicans have only decreased since Trumpapalooza (though he uses RealClearPolitics instead of FiftyPlusOne for some reason). Masket makes the correct point that the convention was illogical from the start; vulnerable Republicans need room to run away from Trump this fall, just like blue-dog Democrats tried to distance themselves from Obama in the 2010 red wave.

But a convention built around Trump naturally makes the midterm all about Trump. If you’re a Republican running in a competitive district, this hurts you, because it makes the election all about the guy with a 34% approval rating.

More on this tomorrow, in a standalone piece on this new media headlines dataset I have assembled.

Share

2. What Strength In Numbers published last week

In last Sunday’s roundup, I walked through how to update your beliefs about an election given a new poll, using the Cook Political Report’s battleground-district survey as the worked example. A rough heuristic about how much a single survey is worth is to calculate the distance between that poll result and the current average in a race, and move it toward the new poll by difference * 1/(number of polls in average + 1).

On Wednesday, I released the results of our new September Strength In Numbers/Verasight poll. Trump’s approval in that poll fell to 33% approve, 63% disapprove, his worst reading in the 16 waves of our tracker, driven by opinion among Republicans dropping 11 points in a month. Democrats lead the House vote 52–42 among likely voters:

On Thursday, David Nir and I did a “vibe check” about the blue wave and asked whether this is what past midterm elections have felt like. Are the polls biased again, or can we be reassured in their accuracy by other indicators, like turnout and special elections?

And Friday’s free Chart of the Week looked at our new poll results showing 58% of Americans say the country is worse off today than it was four years ago, including a quarter of Trump’s own 2024 voters. By 56% to 26%, voters say they want a Congress that checks President Trump rather than one that helps pass his agenda:

If you’re a frequent reader of Strength In Numbers, I’m confident you will get a lot of value out of a paid subscription to the site. You’ll get the Tuesday Deep Dive every week, access to a private community for paying members on Discord, and an early look at our monthly Strength In Numbers/Verasight poll before public release.

More than that though, you’ll be supporting original journalism about American politics that is rooted in data that we collect, together, as a community devoted to the importance of public opinion.

Upgrade to support Strength In Numbers

3. Even more numbers!

• Bryan Bennett compiles 60 years of ANES data on trust in government and shows how skepticism is a core threat to Democratic power in 2028/30. Trust fell from 48% in 2004 to 30% in 2008 and, unlike after Vietnam or Watergate, it never bounced back: it was 16% in 2024. His warning to Democrats is that any majority they win this fall is at risk in 2030 unless they break a two-decade streak of distrust. Democrats Need to Confront the Length and Depth of Distrust in Government and Institutions

• The Wall Street Journal reports Republicans are panicking five weeks out (as we covered on Thursday). The Journal’s own poll puts Trump’s approval at 37%, the lowest for any president heading into a midterm in its polling going back to 1990. Trump Is Predicting Midterms Success, but Republicans Are Starting to Panic

• Consumer sentiment sank again this month. The University of Michigan’s final September index came in at 48.1, down from 51.7 in August and 55.1 a year ago, and year-ahead inflation expectations jumped from 4.0% to 4.6% on anxieties from fuel prices and new tariffs. University of Michigan Surveys of Consumers

• A new Politico poll shows a crack in the GOP base over deportations. The poll shows 31% of non-MAGA Trump voters say the deportation campaign is too aggressive, against 14% of MAGA voters, and 36% of the MAGA base says it isn’t aggressive enough. Poll: A crack is emerging in Trump’s coalition over deportations

• Julia Azari shows there’s a contradiction in the political science models of the election. The fundamentals point to big Republican losses (like 40+ House seats), but because polarization “calcifies” voters, the swing is likely to be smaller. Studying for the midterms

  • Our model at FiftyPlusOne has a median forecast of +25 seats for Democrats.

• Rural voters haven’t turned on Trump, but about half of them tell the AP-KFF that the economy is worse than when he took office. They put groceries, gas, and health care at the top of their list of problems. Only half of rural voters approve of Trump’s job as president, well above the national number but below his vote share. Economic frustration tests Trump’s standing with rural voters, new AP-KFF poll finds

Share

And that’s it for this week! Thanks for reading. Strength In Numbers will be back in your inbox on Tuesday.

Comments are OPEN on this post. If you have a question or comment you want answered/addressed in next week’s podcast or another article, leave it below!

Leave a comment

Related Articles

This old line from Ronald Reagan’s 1980 campaign for president encapsulates a surprising breadth of political science on American elections. Unpopular presidents and their parties tends to perform poorly; this is worse when the economy isn’t growing or inflation is high; and in midterms, there is a nearly unavoidable pattern by which voters en masse decide on rhe need to throw the bums out.

In our September Strength In Numbers/Verasight poll, a plurality of voters say they are not doing better than they were four years ago. The new poll interviewed just over 1,500 adults from September 16-21. Overall, 46% of Americans say their own life is worse off than four years ago, and just 25% say it’s better. When we asked respondents the same question about conditions in the country as a whole, 58% say they are worse off: 40% much worse, 18% somewhat. Just 24% say better.

The poll also finds voters want Congress to act as a check on President Trump.

Full toplines and a methodology description are available here, as are crosstabs. More numbers and charts for this week’s Friday column below.

Here’s my pitch to you to support this work with a paid subscription.

Strength In Numbers pioneered independent polling on Substack in 2025, when I noticed many media outlets weren’t conducting surveys outside of the election season, and their pollsters don’t have the time or the freedom to try new ways of measuring public opinion. The SIN/Verasight polling partnership was formed to help fulfill this publication’s mission of making knowledge and analysis of public opinion part of the republican (note the lower-case “R”) infrastructure of our modern democracy.

If you support that mission, consider becoming a paying subscriber to Strength In Numbers. You’ll get access to our private community on Discord and get additional premium analysis at least once weekly. Paying members of this community keep Strength In Numbers running by supporting the time and data collection it takes to do analyses like this one.

And a yearly subscription is really a great value. While a lot of election analysis websites fall off after Election Day, through this project and others, SIN promises to be here publishing new polls and analysis in both the on- and off-years for many cycles to come!

Upgrade to support Strength In Numbers

Better off than 4 years ago?

In my first write-up of our new poll results on Tuesday, I showed how Donald Trump’s approval rating has taken a particularly severe dip among self-identified Republican voters over the last 30 days. His net approval with his own base is +43 today — a much softer embrace than Democrats’ opposition, which sits at about -90.

This malaise among his base extends to the “better off/worse off” question too. In our new poll one in four Republicans, and one in four of Trump’s own 2024 voters, say their own lives are worse off than they were four years ago. 29% of Republicans say the same about the country.

The chart below shows lower income voters in particular feel that they have fallen behind under Trump. Among families making less than $50,000 a year, opinion is divided between 17% better off to 53% worse. Above $150,000, the options are even at 34% each.

Among those already worse off, 43% expect their lives to be worse still a year from now. Among Trump’s own worse-off voters, 38% do.

I’ve argued before that Democrats lost the 2024 election because they lost the affordability issue. Polls since Trump took office show that he won most of his marginal voters because of backlash to inflation, not for support of ideological goals.

In 2026, as in 2024, voters want cheaper groceries, housing, electricity, and a feeling of economic security, or at least a lack of precariousness. But after a year of inflationary tariffs and foreign wars, they rightly don’t feel like they’re getting that.

Source: Macrotrends.net

Subscribe now

Voters want Congress to act as a check on Trump

Another new polling question we asked this month is whether voters want Congress to be a check on President Trump, or to ally with him to pass his agenda. 56% of respondents say it’s more important to them to elect a Congress that will act as a check and balance on Trump. In comparison, just 26% want a Congress that helps Trump pass his agenda, and 18% pick neither or aren’t sure.

Even Trump’s own coalition is split on this. Republicans and Republican-leaning independents divide 57% for helping his agenda against 25% for checking him — a quarter of his own party wants Congress to hold him back. Independents lean toward a check, 46% to 9%, with 45% landing on neither. Democrats are nearly unanimous: 88% want Congress to be a check on the president, while 8% don’t know and 4% want it to help Trump.

Trump won in 2024 in part because voters wanted relief from high prices and a feeling that they were getting ahead. Our new poll suggests many are still waiting. And this is not just Democrats expressing their usual opposition to a Republican president: One in four of Trump’s own voters say their lives are worse off than four years ago.

That dissatisfaction does not automatically turn a Trump voter into a Democratic voter. But Democrats do not need everyone to embrace their party’s entire agenda to benefit from a backlash against the president. And these numbers present a more fundamental story about democratic satisfaction that has significant implications beyond partisan electoral outcomes.

A majority of Americans already say they want Congress to check Trump — today, not in January 2027, today. While, yes, that is a clear warning to Republicans about the electoral consequences they will face for standing by, it is also a descriptor of how democratic government ought to be working today.

In 2024, Republicans benefited from the desire to “throw the bums out” and shake things up. Now, they’re the ones who have to defend the status quo. For the vast majority, that is a losing battle.

Subscribe now

Related Articles

We also share previously unreleased polling on Trump's promise to send every American a $5,000 check if his party wins the midterms (teaser: it doesn’t poll well!); answer your questions about the new New York Times/Siena poll in Maine; and respond to a recent piece arguing that polling has become a playground for grifters, liars, and frauds. Elliott thinks social media is the problem, not polling.

This video is free to watch, but it was not free to produce. If you value this podcast and want to help keep it going, please consider becoming a premium subscriber to Strength In Numbers. Paid subscriptions keep the show ad-free, fund the newsletter, and ensure we have the time it takes to do original data-driven analysis of politics and elections. Paying subscribers also gain the ability to send in questions during our livestreams, so you can directly shape our conversations on the podcast.

Upgrade to support Strength In Numbers

Share

A few big takeaways from the conversation:

• Republicans continue to slide in the polls. According to Elliott’s election model, Democrats now have about a 70% chance of winning complete control of Congress this November. Trump’s net approval (-28 on average) is nearly three times worse than it was before the 2018 midterms, and Republicans are sweating races like Kansas Senate and Oklahoma’s 5th District (the latter of which Trump won by 18 points in 2026).

  

• Our new poll finds that voters saw Trump’s $5,000 checks as a bribe. Trump’s approval is 33% with 63% disapproving in our September Strength In Numbers/Verasight poll, and Democrats lead the generic ballot by 10 among likely voters. Trump’s promise to send every American adult $5,000 if Republicans win the midterm is underwater by 18 points generally, with 66% of U.S. adults (64% of Republicans) calling it inappropriate. A plurality of Republicans believe Trump made his promise just to win more votes and doesn’t intend to actually send checks.

  

• One poll shouldn’t move your priors much, but the race can still move a lot by November. A new New York Times/Siena poll out today showing Susan Collins up 3 points in Maine moves the average in the state by about half a point in isolation. Senate polling averages have historically shifted about 4.5 points in the final six weeks. There is a correct balance between updating your beliefs the right amount given new data (a small amount, IF you have a lot of data) and being open to the possibility of big changes in the future.

• Fake “polls” are a problem with social media, not polling itself. Elliott responds to a piece arguing that polling has been polluted by grifters and frauds. But fake and AI-generated polls spread mostly through low-quality aggregator accounts on X, and web panels from quality pollsters have been at least as accurate as phone polls in recent years. The fix for this is better gatekeeping, and critics have to be careful not to discredit polls more broadly.

Thanks to everyone who joined us for the weekly livestream on Substack. We’ll be back next week!

As a reminder, paid subscribers to Strength In Numbers get to participate in our live Q&A! If you missed our video livestream, you can watch it by clicking play on the web version of this post at gelliottmorris.com.

You can also subscribe to us on your favorite podcast app to listen on your own time. And if you do listen via one of those apps, please drop us a five-star rating and review if you feel we’ve earned it — it really helps people discover the show!

Subscribe now

Share

You can read the transcript of our conversation by clicking the headline of this article to take you to the web version of the podcast, then clicking the button just below the byline that looks either like a piece of paper or is labeled “Transcript,” like so:

If you’re a reader of Strength in Numbers and haven’t yet subscribed to David’s newsletter, head to the-downballot.com.

Subscribe to The Downballot

And if you’re coming from David’s site, subscribe here to get the numbers behind the news.

Subscribe to Strength In Numbers

Paying subscribers to Strength In Numbers have access to additional visuals and a full archive of crosstabs, and can suggest questions for future polls in the comments section below!

Forty-two days before Election Day, and as tens of thousands of early voters have already cast their ballots, our new Strength In Numbers/Verasight poll finds President Donald Trump and the Republican Party at their lowest point in the last 18 months.

In our new poll, 33% of Americans approve of the job Donald Trump is doing as president and 63% disapprove, a net rating of −30 (using unrounded approval figures) and the worst reading in the 16 waves of this poll since May 2025. The previous floor was −26 in April.

While our past surveys documented that the president has lost significant ground with non-college-educated white voters, independents, non-voters, Hispanics, and young people, this month’s survey reveals Trump slipping significantly with his own party. The president’s net approval rating among Republicans and Republican-leaning independents fell from +54 in August to +43 this month, with 71% approval to 27% disapproval. That’s an 11-point drop and the lowest we have recorded.

Democrats (−90) and independents (−49) barely moved, suggesting GOP holdouts may be responding more negatively to recent crises like the escalating war in Iran, resurgent increase in gas-price inflation, or the president’s pledge to give $5,000 to every U.S. adult if his party wins the midterms.

In this poll, Democrats have a 10-point lead among likely voters in the “generic congressional ballot,” which asks voters which party’s candidate for U.S. House they’ll vote for in their local House district.

Below, I report full results from our monthly Strength In Numbers/Verasight political tracking survey. That includes a record low for Trump on handling prices and findings on voters’ trust in each party to handle various issues, now including artificial intelligence. We also asked Americans whether they are better off than they were four years ago, what they think of Trump's $5,000 check attempted bribe, and who should fund new electrical infrastructure needed for AI data centers.

Share

Headline poll findings

• Job approval: 33% approve / 63% disapprove (net −30 before rounding), down from −24 in August and below April’s −26 floor. Among adults, 50% strongly disapprove vs 15% who strongly approve — a record 35-point intensity gap in the president’s approval.

• Republicans: net approval among Republicans and Republican leaners fell from +54 to +43, the lowest in the tracker. Among 2024 Trump voters, strong approval is 35%, down from 53% in May 2025.

• Prices: The president now has a net approval on prices of −54, a new record low for any issue, down from −48 in August. 21% approve, 75% disapprove.

• Generic ballot: Democrats lead 50–42 among registered voters (D+9 unrounded), 52–42 among likely voters (D+10), and 55–42 (D+12) among registered voters who say they will definitely vote. Democrats have led in all 16 of our polls since May 2025.

• Party trust: Democrats lead on 10 of 13 issues, set series highs on five, and pull ahead of Republicans on trust to handle immigration (D+1) for the first time. They lead by 11 on handling of each voter’s own most important problem (45–35).

• Top problem: 38% name prices/inflation the single most important problem facing the country; 59% put it in their top three.

• Better off than you were four years ago? 26% say they are, versus 45% who say they’re doing worse. 24% of adults say the country is better off, 58% worse.

• Direction of the country: 54% say things are going poorly and major, disruptive changes are needed; just 7% say things are going well.

Strength In Numbers conducts a monthly poll with Verasight asking a mix of political tracking questions and questions sent to us by readers. Subscribe to Strength In Numbers to get future poll results in your inbox, and to be able to submit questions we may use in future polls.

Upgrade to support Strength In Numbers

Trump’s approval hits a new low, thanks to Republican defection

In our poll, 33% of adults approve, and 63% disapprove of the job Trump is doing as president, a net of −30 (using unrounded figures), down from −24 in August. That is a new low on all three measures: the first wave under 35% approval, the first over 62% disapproval, and the worst net since our poll began last year.

The month-to-month move on the topline is outside our margin of error, as are several important differences by subgroup. Among Republicans and Republican leaners, net approval of the president fell from +54 to +43 (71% approve, 27% disapprove), an 11-point drop outside the confidence interval in movement across previous months in the tracker. Trump’s prior approval among Republicans was +51 in April. Democrats went from −89 to −90 and independents from −48 to −49.

Updating a chart we made for the first time last month, we see there is an increasingly large gap between the share of American adults who strongly approve and disapprove of the president’s job performance. About half of Americans strongly disapprove, a sign of major discontent among the American public.

Trump hits yet another record low on prices

Trump’s net approval on prices and inflation fell to −54 (21% approve, 75% disapprove), the worst number we have recorded for any issue, from −48 in August, which had been the previous record. Five of the 12 issues we have tracked all year are at series lows: prices, health care (−33), jobs and the economy (−28), trade (−27) and deportations (−15).

The one cluster that improved is immigration. Border security went from −5 to −3, immigration from −14 to −11, and crime from −13 to −10, all within the margin of error. Our survey was out of the field before the latest shooting by Immigration and Customs Enforcement in Austin, Texas, this past weekend.

This month is the first time we have asked about approval for Turmp on artificial intelligence. The president starts with a −27 (30% approve, 56% disapprove) approval rating on the issue, with 14% unsure.

Subscribe now

Democrats lead the generic ballot for a 16th straight poll, and by 10 among likely voters

Democrats lead the generic congressional ballot 50–42 among registered voters (D+9 unrounded) and 49–41 among all adults in our September poll, up from 49–43 in August and the second-widest lead in the series behind February’s D+10 among registered voters. Across 16 polls since May 2025, Democrats have never trailed the Republicans on the generic ballot. The GOP’s 42% in this month’s survey is a series low.

When registered voters are weighted by their probability of turning out to vote in November, Democrats lead Republicans 52 to 42 (D+10.4), up from D+7.5 in August and the highest since February. The likely-voter screen has added between a quarter of a point and two points to the Democratic margin in every wave since we began asking the turnout question a year ago.

The tightest screen tells the same story. Among registered voters who say they will definitely vote in November, about 73% of registered voters, Democrats lead 55–42, a 12-point margin (D+12.3 before rounding), up from the 9-point lead the same group gave them in August. Just 3% of these voters are undecided.

Our LV model translates a respondent’s stated vote intention to a probability of turnout based on the responses and behavior of validated voters in the 2024 Cooperative Election Study.

Subscribe now

Democrats lead on 10 of 13 issues and pull even on immigration for the first time

Our new poll also reveals Democrats are trusted more to handle most issues Americans rate as their most important problems.

The party is trusted more than Republicans to handle health care by 25 points (a series high), government funding and social programs by 21, education by 17, prices by 14 (a high; the gap was one point in May 2025), elections and democracy by 13, trade by 12 (high), jobs by 11 (high), and foreign policy by 8 (high).

In our first survey including artificial intelligence in the issue battery, Americans trust Democrats over Republicans on the issue by 7 points.

On immigration, Democrats are at D+1 (42–41), the first reading in the tracker that isn’t a Republican advantage; it was R+12 in May 2025. Republicans lead only on deportations (R+1), crime (R+5) and border security (R+11).

On whichever issue each respondent named as the country’s most important problem, Democrats lead 45–35, an 11-point edge; that lead has run between 9 and 13 points all year.

Democrats have gained ground on almost every issue, but especially high-salience issues, since last May. Prices, the country’s top problem by a wide margin, went from a one-point Democratic edge to D+14; jobs and the economy from D+2 to D+11; health care from D+13 to D+25. Immigration, R+12 when we started asking in May 2025, is now even.

If you prefer a time-series chart for just one variable, the graph below shows the party trust numbers for each of the 13 issues we track going back to September 2025.

Prices are still the top problem for most Americans — making Democrats’ continued lead on the consequential in the upcoming midterms.

Overall, 38% of U.S. adults name prices and inflation the single most important problem facing the country, up a point from August and a point shy of May’s record. Jobs and the economy and elections and democracy are tied for second at 13%, then health care at 7% and crime at 6%. Artificial intelligence, asked for the first time, is named by 5%, and 16% put it in their top three, roughly level with crime and a point behind immigration.

Subscribe now

Most say they are worse off than four years ago

Our new poll also finds 45% of Americans say they are worse off than they were four years ago, compared to 26% who say they are better off. Assessments of the country as a whole are even more negative: 58% say the country is worse off and 24% say it is better off. Most Americans connect these changes to the president, with 78% saying Trump’s policies have a great deal or a fair amount to do with them. Among those who say they are worse off, 63% say President Trump’s policies have had “a great deal” to do with it.

Americans are somewhat more optimistic about the year ahead: 34% expect their own lives to get better and 23% expect them to get worse.

There is also significant demand for changes to the political system. 54% of Americans say things are going poorly and major, disruptive changes are needed, up from 52% in August and 49% in July, and just one point below the high we recorded in April. Another 36% say things could be going better, while just 7% say things are going well. Here, too, we see Republicans becoming less satisfied: the share who say things are going well has fallen from 21% in January to 13% this month.

Reader-suggested questions

Each month, paying subscribers to Strength In Numbers can suggest questions for us to pick from and include in a designated community question block. I usually field at least a few questions from readers. This month we asked three:

• Which party cares more about people like you? Democrats 38%, Republicans 26%, both 7%, neither 22%, statistically unchanged since May.

• What motivated your 2024 vote? Among 2024 voters, 31% say agreement with their side’s priorities, 25% keeping the other side out of power, and 37% both equally. More results on this question, breaking down by crosstabs, next week.

• Who should pay for AI data centers? 63% of adults say the technology companies building AI data centers should pay for the power plants and grid upgrades necessary to run them, unchanged from July’s 64%. Agreement is bipartisan, with 69% of Democrats agreeing, along with 66% of Republicans.

I’m holding two blocks from this wave for later in the week. On Friday, results from an experiment we ran on Trump’s $5,000-check promise, and what they say about the convention it came from. And next Tuesday, a Deep Dive on what Americans think of federal troops and ICE agents at polling places, and whether either side would accept losing in November. Make sure you’re signed up to get those in your inbox!

And if this type of work is useful to you or you believe in supporting it for the public good, consider becoming a paid member of this Substack. Paying subscribers to Strength In Numbers make it possible for me to spend time on this original, independent, data-driven political journalism that’s not beholden to any corporation or billionaire donor. Support my work and get access to articles like this one for $10/month or $90/year.

Upgrade to support Strength In Numbers

Methods statement: Verasight collected the data for this survey from Sept. 16–21, 2026. The sample consists of 1,528 United States adults ages 18 and above. The data are weighted to match the August 2026 Current Population Survey on age, race/ethnicity, sex, income, education, region, and metropolitan status, as well as to a running three-year average of partisanship distributions from the Pew Research Center NPORS benchmarking surveys and population benchmarks of 2024 vote. The margin of sampling error is ±2.6%.

While Strength In Numbers has input on question wording, all other methodological decisions were made and carried out by Verasight to ensure independence of the data-generating process behind these results. Verasight reviewed the questionnaire for typos and completeness.

You can download a full topline file, key crosstabs, and full methodology statement at the Strength In Numbers/Verasight polling portal. The paywalled section containing full crosstabs and more interactive graphics is updated as of publication.

If you have any questions about this poll or the release, please email polling[AT]gelliottmorris.com.

Have a suggestion for next month’s poll? Leave it in the comments below!

Related Articles

As I sometimes do during poll week, I’m pushing back the normal Tuesday Deep Dive to next week to focus on the monthly Strength In Numbers/Verasight survey release. We got data back from the field mid-day Monday, so I used my work block on Monday night and am using most of Tuesday this week to put together several different analyses, some of which will be public and one paywalled, on what are some surprising and insightful new numbers. There are some pretty striking results and charts to share, including new records for Donald Trump’s second term on multiple questions.

If you want to get the poll results and crosstabs early, take out a premium subscription to SIN and refresh this page around midnight Wednesday morning. If you are a reporter, I will also share a .pdf copy of the toplines and tabs with you under embargo; just email pollingATgelliottmorrisDOTcom.

The monthly poll with Verasight is the big recurring data project we do here at Strength In Numbers and something that really distinguishes this Substack from the crowd. The philosophy behind the poll is to ask questions in new ways, with new formats, on topics that are more consequential to the average voter than what you often find in media surveys, all with the high-minded goal of preserving polling’s role as part of our democratic infrastructure. We do lots of survey experiments and statistical modeling that make the partnership more sophisticated than most other polls, and we also seek to create knowledge via new and fun question wording. As a reminder, subscribers get to submit topics/questions for consideration on future surveys.

And, by the way, the survey will continue after the election, so don’t go anywhere on November 4th!

More tomorrow,

Elliott

Upgrade to support Strength In Numbers

Upgrade to support Strength In Numbers

Before we get going, an important programming note! The September Strength In Numbers/Verasight poll will be coming out later this week. If you are a reporter and would like to be added to our email embargo list for receiving toplines and crosstabs early, email me at pollingATgelliottmorrisDOTcom.

1. How to update your beliefs with new polling data, the Strength In Numbers way

The Cook Political Report released its first Battleground District Project poll of the fall last Thursday, and the write-up was the subject of some conversation in the Strength In Numbers Discord on Friday-Saturday. I’m explaining some of the math I did there for paying subscribers here because I think it will be useful to a broader audience.

Cook writes that Democrats hold a small 2-point lead in seats they have identified as Lean Democrat, Lean Republican, or Toss-up, a swing of 7 points since 2024 (Trump won the average seat in these buckets by about 5 points, and 5 - -2 = +7). Carrie Dann and Amy Walter write:

Democrats’ small lead on the generic ballot among voters in these red districts suggests that Democrats’ gains this fall will mostly be limited to districts Trump carried by 10 points or fewer. While that’s cold comfort for Republicans running in districts that narrowly went red in 2024, it’s also a reminder of the compactness of the House playing field — 86% of seats held by Republicans are those Trump carried by more than 10 points.

Here are the poll details. Cook, GS Strategy Group, and New River Strategies surveyed 1,052 likely voters from September 8-11 across the 37 districts Cook rates as competitive. Those seats went for Trump by 4.8 points on average in 2024. Today, the generic ballot in them is Democrats 49, Republicans 47, and Trump’s job approval is 42-58. That’s how they get that 7-point swing (okay, 6.8 points, whatever).

These numbers may help explain why Cook, as well as other race raters, are more bearish than the statistical modelers in assessing Democrats’ prospects for the November midterm elections. As I reported Friday, even if you account for the historical tendency of expert handicappers to push toss-up and lean ratings toward the out-party in midterm elections, they’re still forecasting 229-230 seats for the Democrats, which is about 5 seats less than the quantitative forecasters on average (and 7-8 less than the median at 50+1). If you take Cook’s ratings at face value — eg, you split “toss-up” races 50-50 — the organization is “predicting” Democrats to win a bare 218-seat majority.

In reality, they know most of these seats will go Dem; they just haven’t called it yet. One expert described what the forecasters are doing to me as “trying to skate to where the puck is going” on Election Day, not forecasting outcomes as forecasters do.

Okay, but those are the ratings, and this new data from Cook is just one poll. Yes, it’s certainly less blue than the forecasters, which are officially now in Big Blue Wave territory. So how can you reconcile these different measurements? If you’re starting from the position that the election will be around D+7 to D+9, how do you square that with the Cook poll that shows just D+5? (I get to D+5 by adding 7 to -2.3, since Republicans won the House vote by 2.3 points in 2024.)

Well, what you should do is treat the poll as you would every other poll: one survey that is subject to random sampling error and potential non-response bias, and which shouldn’t shape your whole worldview. How does the Cook poll compare to other polls in the districts Cook sampled?

To answer, I pulled every non-partisan, non-internal district poll we have compiled at FiftyPlusOne.news in the 37 seats Cook targeted. We have 13 such surveys, covering 13 districts (the other 24 districts have no qualifying independent non-partisan public poll at all). Those 13 polls average Democrats +4.8 in seats that Trump carried by 6.9 points in 2024, for a swing of D+11.7 points, with a median of 13.1. So we have two readings of the same 13 battlegrounds that are roughly 5 points apart.

If you were to treat each poll as an independent read of the political environment, via laws of sample size, the district polls end up having about six times as much weight as Cook’s single survey. They have more respondents in total, and come from nine different pollsters whose house effects partly cancel out. Cook is one survey, with roughly 28 respondents per district and a margin of error of 3-4 percentage points. (All these polls also carry a shared potential for cycle-wide polling error, but since that is shared I’m ignoring it here.)

If the independent 13 polls are showing a D+11.7 swing, with a standard deviation of the mean of 2.2 points, and you get a new reading at a D+6.8 swing (Cook), with a standard error of 4.5 points (sampling error plus the design effect and potential house effect from Cook’s pollsters), then you’d naively update the +11.7 (+/- 2.2) to +10.8 (+/- 2.0). (The math for this is given in Bayes theorem via the conjugate normal update.)

This is a good start, in that you’re now thinking of the data as a Bayesian and in an additive sense — not Cook vs Everyone Else but what all the data is saying in combination. But there’s also another factor that I’ve been hiding from you.

Cook didn’t ask about voting for individual candidates, but voting for parties. They asked voters in these districts the generic ballot, not support for candidates with their names. I showed last week that Democrats typically do better when they are named instead of asked about generically, just with party labels, so this is an important factor to take into account if you’re trying to forecast an election... where candidates have names.

The seats that have been polled by independent pollsters include districts Democrats are holding that Trump won by double digits — Henry Cuellar in TX-28, Kathy Castor in FL-14, and Don Davis in NC-01 — and the independent polls show a swing of 6.5 points in Lean D seats, 14.2 in Toss Ups, and 16.3 in Lean R seats. The redder the seat, the further Democrats run ahead of the “generic Democrat” label.

If you assume that Democrats get a 5-point boost for being named — maybe the party recruited well, but Democrats also have an incumbency advantage in competitive districts — the two sources reconcile almost exactly. Frankly, 5 points seems high to me, but that’s just a vibe check; and as shown above, if you assume the data is in conflict, you still get to a swing of about 9 points and a national environment of about D+7.

Our polling nowcast at FiftyPlusOne for these same districts shows a +9.8 swing, which is a D+7.5 national environment. That’s what you’d expect from a model already mixing the above non-partisan district polls with district-level campaign internals and the national generic ballot. So it’s not that far off.

Cook’s poll should move your estimate of the 2024-2026 swing from +9.8 to roughly +9.2. This corresponds to a national environment of D+6.9, down from D+7.5. That’s a real update, but still firmly in blue wave territory — especially once you control for likely additional Dem gains in the generic ballot, candidate quality differences, and fundraising favoring Democrats.

What D+7 in votes means in seats

For some context, I have translated the national margins above to naive seat estimates by counting up the number of Republican-held seats that Trump lost or won by less than (6, 8, 10, 12) points. If Cook is at D+5, that works out to about 221 seats for the Democrats:

With the models around D+8 today (a 9.5-point swing), you can see why they are giving a seat-level picture that looks so optimistic (for Democrats) when compared to Cook. There are a lot of seats up for grabs in the D+5 to D+8 range, and at D+10 nationally, Democrats are favored in 243 districts. Gains accelerate even more beyond D+10, which is possible but not necessarily likely.

And this brings me back to a line in Cook’s report about gains being “limited to districts Trump carried by 10 points or fewer.” This sounds small (Cook points out 86% of GOP-held seats are redder than Trump +10), but it’s not. A wave that reaches Trump +10 seats is a national environment of about D+8.5. On this map, that’s 234 seats — one seat shy of the Democratic total from 2018, when Trump’s approval rating was higher than it is today, and before pandemic-era inflation increased anti-incumbent sentiment significantly.

Hopefully this helped you understand how we approach the topic of updating our beliefs with new data here at Strength In Numbers. One principle that comes out of the math here is that individual polls should not alone carry that much weight in our forecasts. While Cook is prolific and this is a survey across many different districts, at the end of the day, it’s just one poll. Treat it appropriately.

Share

2. What Strength In Numbers published last week

Last Sunday’s used polling data from YouGov to show that Democrats are doing better in named-candidate polling than the generic ballot:

Tuesday’s Deep Dive addressed the under-discussed potential that polls could be understating support for Democrats, rather than overstating it as many analysts assume. My simulations show the way pollsters are weighting their samples is potentially costing Democrats up to one point in the generic ballot:

On Thursday, David Nir and I discussed this on the podcast, and I made the case that the current political environment was very predictable by the political science “fundamentals”. We also did a state-by-state tour of the current FiftyPlusOne election forecast:

And Friday’s free Chart of the Week looked at the expert handicappers’ ratings for November. Across the last four midterms, when Cook, Inside Elections, and Sabato said a race was a “tossup,” those seats broke for the out-party about two-thirds of the time. This is pretty convincing evidence that their current seat totals for either party are understating Democrats.

If you’re a frequent reader of Strength In Numbers, I’m confident you will get a lot of value out of a paid subscription to the site. You’ll get the Tuesday Deep Dive every week, access to a private community for paying members, and first access to our monthly Strength In Numbers/Verasight poll. More than that though, you’ll be supporting original, data-driven journalism about American politics.

Upgrade to support Strength In Numbers

3. Even more numbers!

• Pew’s Scott Keeter and Courtney Kennedy have the definitive explainer on weighting by past vote. They explain (like I did this week!) how past-vote weighting is risky, but why it still helps polling on average so pollsters should probably do it anyway. Weighting by past vote, I explained last week, is a bet that partisan non-response bias in polling is larger than potential recall bias, which is probably (but not definitely) true in 2026. Much bigger question mark about 2028, though.

• Nate Cohn digs into this week’s Times/Siena poll, which has Democrats up 8.5 on the generic ballot among likely voters. His spends a lot of time on the question of whether Republicans can still hold the Senate by painting Democrats as too far left. A ‘Blue Wave’ May Be Building, but How High?

• UMass Amherst went into the field in late August and found just 11% of Americans would support an AI data center being built in their community, while 65% oppose it. Even Republicans are against it nearly 3-to-1 (a rare bipartisan consensus these days), and Trump is 34 points underwater on his handling of AI. Just 1 in 9 Americans Support AI Data Center Construction in their Local Communities

• Annenberg’s Constitution Day survey, fielded in August, found that 35% of American adults can’t name all three branches of government, just 9% can name all five First Amendment rights, and 25% think the right to bear arms is in the First. I have written previously about low-information voters.

• Survey methodologist (and SIN subscriber!) Raphael Nishimura built a nice interactive app for teaching how sampling actually works. It’s in Portuguese, but your internet browser can translate that. Raphael Nishimura post on his web app.

Share

And that’s it for this week! Thanks for reading. Strength In Numbers will be back in your inbox on Tuesday.

Got more for next week? Email your links or add to the comments below!

Leave a comment

Related Articles

As of publication, the three big election handicappers have the race for U.S. House as a pure toss-up and the Senate leaning Republican. The Cook Political Report currently has 206 seats at least leaning Democratic, 208 at least leaning Republican, and 21 tossups. Inside Elections has 200 Democratic, 202 Republican, and 33 tossups. Sabato’s Crystal Ball has 210 Democratic, 203 Republican, and 22 tossups.

Read literally, that implies Democrats need to win the lion’s share of seats rated “tossup” to carry the House: 12 of Cook’s 21, 18 of Inside Elections’ 33, 8 of Sabato’s 22. By “read literally,” I mean if you assume “tossup” means a 50-50 chance of going either way, Lean means something around 70-30 for the favored party, Likely 90-10, and Safe 97%+. The Democrats’ climb to a majority is even steeper if you consider they have more seats in the Lean Coulmn.

This is despite Democrats being up by 7 points in the House generic ballot among likely voters, and posting even larger overperformance in district polls and special elections. Our model at FiftyPlusOne has Democrats at a median of 238 House seats and a 98% chance of the majority.

What do the race raters know that we don’t?

I’ve been taking a deeper dive into race ratings for an analysis yet to come at FiftyPlusOne soon. But for this Chart of the Week, I wanted to share an important finding here based on the historical performance of the race raters. And that is this: since 1990, race raters have badly overestimated support for the incumbent party in midterm elections. When they say “tossup,” historically, those races break overwhelmingly for the out-party. This equates to Democrats being likely to win about 70% of current tossups.

If you like the free version of SIN, consider becoming a paying subscriber to get access to our private community and additional premium analysis at least once weekly. Paying members of this community keep Strength In Numbers running by supporting the time and data collection it takes to do analyses like this one.

Upgrade to support Strength In Numbers

Raters have a durable pro-White House bias in midterms

I took the historical race ratings database that powers part of the FiftyPlusOne election forecast, subset it to ratings given in midterm years as of 45 days before Election Day, and analyzed how often races in each rating bucket were won by the party in control of the White House. These ratings show a durable pro-White House bias.

For example, 45 days before the 2018 election, the Cook Political Report rated 28 House contests in Republican-held seats as tossups. Democrats (the out-party that year) went on to win 23 of them — a win rate of 82%. Of tossups!

That year, of the 27 seats Cook called Lean Republican, Democrats won 5, and of the 26 Likely R seats they won 2. Every one of the 22 seats Cook rated Lean D or Likely D stayed Democratic.

This is a big calibration problem. If you read the probabilities generated by the raters literally, seats rated as tossups should go to the parties at roughly equal rates; win probabilities of about 50%. But over the long term, this is not the case.

Here is Cook’s House record across the 2010, 2014, 2018, and 2022 midterms, using each seat’s rating as of 45 days before the election and recoding every label as favoring the “in-party” (the president’s party) or the “out-party”:

• Safe in-party: the in-party won 100% of 311 seats.

• Likely in-party: 88% of 69.

• Lean in-party: 78% of 86.

• Tossup: the in-party won 33% of 131. The out-party won 67%.

• Lean out-party: the in-party won 5% of 37. The out-party won 95%.

• Likely out-party: 0% of 73.

• Safe out-party: 0% of 395.

Two things stand out here. First, we should expect Democrats to win a lion’s share of the races Cook currently labels as tossups. And second, there is a pronounced asymmetry in the ratings curve; A “Lean” label that holds 78% of the time when it favors the president’s party and 95% of the time when it favors the opposition is not a symmetrical probabilistic statement.

And I don’t mean to pick on Cook; other raters have the same bias. Inside Elections’ Lean in-party seats have held 69% of the time, Sabato’s 72%. All three give the out-party 67% of tossups. All three hold their Lean out-party seats 95% to 97% of the time.

You can see that all the race raters have the same problem. The out-party wins the lion’s share of races that, at 45 days out, experts say are toss-ups or only “lean” toward them.

And this is mostly not a problem that the quantitative models have. Will write more about that next week.

Share

Debiased expert ratings suggest Democrats are 83% to win the House

So we can be reasonably sure the race raters are underestimating the Democrats — at least in the House, if not the Senate too. A useful exercise is to compare the picture of the national House race that is implied by a literal, symmetrical reading of the race raters’ current seat-level forecasts, with the one that you get if you adjust for their endemic underestimation of the out-party.

Today that literal reading gives Democrats 217 seats on Cook’s ratings, 217 on Inside Elections’, and 220 on Sabato’s. Average the three, and you get 218 seats — an exact toss-up for control of the House.

But when we adjust for historical bias in the ratings (the table above), the picture changes dramatically. If we calculate the mathematically implied seat total for Democrats using the empirical historical probabilities of the out-party winning seats in each rating bucket, the number rises from 218 to 228 on the three-rater average. Cook’s own record moves Cook’s total from 217 to 224.

A gain of ten seats may seem small, but it is a huge move in terms of probabilities. I wrote a program to simulate potential error in the race raters’ forecasts based on their historical misses, and then see the distribution of seat totals we might end up with given those errors, and the results are notable. (A description of this program can be found in the footnotes.)

If you assume (wrongly) that Cook et al.’s probabilities translate symmetrically to outcomes, today their forecast is 217 seats for the Democrats and a roughly 47% chance of winning control of the chamber.

But if you assign each seat to the Democrats or Republicans using the win probabilities implied by Cook’s historical errors, that seat total is 224, and the win probability is 75%.

This is still a likely severe underestimate of Democratic win probability, but it’s a big shift in the right direction given the national indicators. And the story is the same for the other race-rating organizations: Inside Elections goes from 217 seats and 49% to 230 seats and 87%. Sabato goes from 220 seats and 57% to 231 seats and 89%.

Pool all three, and the House goes from a coin flip, 218 seats and 51%, to 228 seats and 83%.

The Senate tells largely the same story, but I’m going to save those findings for another article, since the big difference between the raters and models right now is in the House.

The handicappers still know things

There is still some value to the handicappers, even if they have less alpha compared to more comprehensive models. Rating organizations (especially Cook) get access to candidates and a lot of data that isn’t made public. They have a good track record and are free to put their thumb on the scale when data looks off, a gamble that worked out for them in 2024.

Our election model at FiftyPlusOne uses race ratings as one input to our model, but we do so after processing them for various historical biases and other empirical wonkery.

Maybe that scale-thumbing will work again for the race raters in 2026. But historically, they have always underestimated the out-party in midterms — to the tune of about 30 percentage points in win probability in the House this year. That’s a heck of a bias, and not one that the quant modelers have.

Subscribe now

Related Articles

Footnotes

Details of simulation program:

Step 1. Turn each rating into a seat probability. Every one of the 435 House races gets a label from the rater. The symmetric reading maps the label straight to a number: Tossup 50, Lean 75, Likely 90, Safe 97 for the favored party. The empirical reading instead looks up how that rater’s label actually performed 45 days before the 2010 through 2022 midterms, after recoding each label as favoring the president’s party or the opposition. So a Cook Lean R seat in 2026 gets the rate at which Cook’s Lean in-party seats have been held in past midterms, which is 78 percent, meaning a 22 percent Democratic win probability. A Cook Tossup gets 67 percent Democratic, because that is how often the out-party has won Cook’s House tossups.

Step 2. Add the seats up. Summing the probabilities gives the expected Democratic seat total. That is the 217 versus 224 figure for Cook. On its own, this says nothing about the odds of a majority, because it has no error attached.

Step 3. Calculate historical rater error. For each past midterm, the program finds the single nationwide shift that, when applied to every competitive seat’s probability at once, would have actually given the result of the midterm. A year like 2010, when the president’s party did even worse than the average midterm, needs a shift toward the out-party. A year like 2022, when it did better, needs a nudge the other way. The nudge is measured on the logit scale, which is a way of shifting probabilities that keeps them between 0 and 100 and moves a 50 percent seat more than a 90 percent seat. Across 2010, 2014, 2018, and 2022, those shifts were roughly plus 0.8, plus 0.4, zero, and minus 1.5. Their spread, about 1.0, becomes the size of the simulated national error. A one-unit nudge moves a 50 percent seat to 73 percent, or to 27 percent in the other direction.

Step 4. Simulate 20,000 elections. Each simulated election first draws one national shift from a bell curve with that spread and applies it to every Likely, Lean,n and Tossup seat in the same direction. Safe seats are left alone, because Safe-rated seats have not flipped in the record, and shocking 360 of them at once would have widened the uncertainty interval of our simulations beyond the empirical historical range. Then each seat is resolved individually by a coin flip weighted to its adjusted probability. Counting the Democratic wins gives one seat total, and repeating 20,000 times gives us a bunch of simulations we can use to calculate chamber win probabilities (and show that histogram).

Two notes:

Why this moves the probability so much. The two readings differ by only 7 seats in the mean, but the symmetric mean sits right on the 218 line, so half the simulations fall below it. The empirical mean is 6 seats above the line, and with a spread of about 12 seats, 75% of the simulations clear the majority threshold.

I checked historical calibration. For each past midterm and rater, the program compares the actual Democratic seat count after Election Day with what the translated ratings expected. The typical historical miss is 13 seats. The simulation’s implied standard deviation is 12. So the uncertainty intervals from this model are as wide as the raters’ own records say they should be.

This video is free to watch, but it was not free to produce. If you value this podcast and want to help keep it going, please consider becoming a premium subscriber to Strength In Numbers. Paid subscriptions keep the show ad-free, fund the newsletter, and ensure we have the time it takes to do original data-driven analysis of politics and elections. Paying subscribers also gain the ability to send in questions during our livestreams, so you can directly shape our conversations on the podcast.

Upgrade to support Strength In Numbers

Share

A couple of big takeaways from the conversation:

• Polling bias is not predictable from cycle to cycle. In 2022, many election forecasters ended up making worse predictions than they otherwise woudl have by adding extra support for Republicans to “unskew” the polls.

• Weighting by recalled 2024 vote could be quietly inflating support for Republicans. Pollsters increasingly weight samples to match the results of the last election, using respondents’ self-reported vote. But people can misremember or change their reported vote — if disaffected Trump voters are the ones now declining to admit they voted for him, the remaining self-identified “Trump voters” skew more MAGA than the actual 2024 coalition. Elliott’s simulations suggest this could be suppressing the Democrats’ generic-ballot margin by up to a point.

• Named-candidate polling shows Democrats running ahead of the generic ballot. A YouGov/Economist survey using actual candidate names in each district showed Democrats up 11 points among likely voters, versus 8 points on CBS’s standard generic-ballot question over the same period (also fielded by YouGov) — a gap directionally consistent with but larger than the historical average overperformance for high-quality candidates.

• The 50+1 forecast now gives Democrats a 97% chance of winning the House, up sharply from where it stood in prior weeks, and a 65% chance in the Senate, where polls have moved toward Democrats in most competitive states.

Thanks to everyone who joined us for the weekly livestream on Substack. We’ll be back next week!

As a reminder, paid subscribers to Strength In Numbers get to participate in our live Q&A! If you missed our video livestream, you can watch it by clicking play on the web version of this post at gelliottmorris.com.

You can also subscribe to us on your favorite podcast app to listen on your own time. And if you do listen via one of those apps, please drop us a five-star rating and review if you feel we’ve earned it — it really helps people discover the show!

Subscribe now

Share

You can read the transcript of our conversation by clicking the headline of this article to take you to the web version of the podcast, then clicking the button just below the byline that looks either like a piece of paper or is labeled “Transcript,” like so:

If you’re a reader of Strength in Numbers and haven’t yet subscribed to David’s newsletter, head to the-downballot.com.

Subscribe to The Downballot

And if you’re coming from David’s site, subscribe here to get the numbers behind the news.

Subscribe to Strength In Numbers

This piece is possible because paying subscribers to Strength In Numbers have supported our original, independent polling project that has gathered data every month since May 2025. If you want to get this kind of data-driven reporting in your inbox every week — plus get first access to the full poll and crosstabs each month — a paid subscription is the best way to support it.

Upgrade to support Strength In Numbers

Last month, the Republican pollster Patrick Ruffini made the case that polls today are overstating support for Democrats. Across four cycles, he showed, error toward Democrats in Senate polls peaks at about six points in late July and doesn’t fully recede by election day. Polls, he said, always overshoot Democrats in the summer.

But his analysis was flawed in two ways. First, it was limited to cycles after 2016, when polls have been more systematically biased than they were over the long historical term. And it assumed that bias is predictable ahead of Election Day based on the performance of polls over the last four elections. In fact, in polls going back to 1998, bias in polls varies mostly randomly from cycle to cycle, with some big reversals that make predicting bias a fool’s errand. Over that time period, if you regress a year’s polling bias on the previous four, the relationship is statistically indistinguishable from zero. In other words, the past told you nothing whatsoever about the future.

Still, it’s always a good idea to monitor the polls for signs of non-response bias toward one party. So I also ran an analysis of our individual-level Strength In Numbers/Verasight data to see if the types of people answering our surveys are more left-leaning than the electorate in 2024. They are not. I trained a statistical model to predict 2024 vote choice in public polling data among our poll respondents, and the kinds of people answering our polls are the same kinds, demographically and geographically, who were responding in early 2025.

This gives us confidence that polls today are not overestimating support for Republicans, at least in any obvious way we can detect ahead of time. There is still the chance polls are biased in some sort of way we can’t foresee — the unknown unknowns of polling error, so to speak.

But reading these pieces, you may have noticed: Everyone is asking whether or by how much the current polls are overstating support for Democrats. Almost nobody is asking whether they might be understating them. And based on the historical pattern of polling bias being unpredictable from cycle to cycle, that’s a real possibility.

In today’s Deep Dive, I show how polls might be overstating support for Republicans. One adjustment in particular could be costing Democrats up to a point of support in live phone and text polls.

The rest of this piece is paywalled for paying members of Strength In Numbers. Subscribers fund this work and make it possible for me to make a career out of independent journalism. Their support will be key to helping this business grow after the election.

Upgrade to support Strength In Numbers

The one polling methodology decision that could be skewing numbers to the right

Read more

Upgrade to support Strength In Numbers

1. If Democrats win, one reason will be great recruitment

A new YouGov poll sponsored by CBS News out today has Democrats up 8 percentage points on the U.S. House generic ballot among likely voters. That would be enough for them to win the House majority, and likely enough for the Senate if states fall the right way.

But hold on... What’s this? Another YouGov poll, sponsored by The Economist, had the Democrats up 11 points for the House on Tuesday.

Both samples are among “likely voters,” who YouGov identifies as people who say they “definitely will vote” in November. So who’s right? Are Democrats up 8 or 11? Is this just normal sampling noise across polls?

Well, the difference between these polls certainly could be due to noise. But it may be driven by differences in how YouGov asks about vote intention for House elections. See, for CBS, YouGov asked who voters would pick in the “generic congressional ballot” — a poll question that asks who voters would vote for “in the district where you live?”

But for The Economist, YouGov this week asked who voters would pick “using the names of candidates running in the respondent’s district of residence.” In the CBS poll, people don’t see the names of people running in their districts — in the Economist poll, they do. The latter approach, while rare in media polling, is common in academia; this is how the Cooperative Election Study run out of Harvard (and fielded by YouGov) has asked the question for decades. And it correlates very well with district-level election outcomes in 2022 and 2024.

In both of those years, Democratic House candidates outran their party’s presidential nominees across the board, but especially in competitive seats. I know this because when I ran my analysis of the GOP’s 2026 gerrymanders, I trained a model to forecast races in 2024 using the historical presidential results in each seat. Without adjusting for the pattern of House candidates beating the baseline across the map, this model would have badly underestimated Democratic seats.

This year, we may be looking at the same thing. Democrats have done a great job around the country recruiting candidates who are good fits for the districts they are running in, especially in competitive seats. The YouGov polls suggest this may earn them a couple points on vote margin — reasonable per the historical boosts candidates get when they run against suboptimal opponents.

In the Senate, candidates like James Talarico and Sherrod Brown also fit this pattern of good recruitment. And especially in Texas, where Talarico is running against one of the most flawed Republicans of all time — and is currently favored to win that election.

If Democrats do win the House and Senate this year, it might be because they again have a sizable recruitment advantage over the Republicans.

Share

2. What Strength In Numbers published last week

Last Sunday I wrote four short essays about how this newsletter works and how I actually make a living as an independent journalist and pollster. Includes some thoughts on largely quitting algorithms and social media, which continues to be one of the best decisions I ever made.

Then on Wednesday I wrote about why you should think of all polls as models now. The old probability-vs-opt-in distinction is largely meaningless for accuracy. Response rates have collapsed so far that every pollster is modeling now, whether they admit it or not. What a pollster does to their data matters more than how they got it:

And on Wednesday night, in my coverage of the Republican Party’s midterm “convention,” I discussed why the strategic logic behind “Trumpapalooza” is so misguided.

David Nir and I discussed a little more about that in our podcast on Thursday, and also talked through a guide on how to know when polls are worth paying attention to. See above: Not all polls are created equal.

And Friday’s free Chart of the Week explains that low-turnout voters aren’t MAGA diehards. They’re anti-status-quo, and went for Trump in 2024 because of prices — which leaves many of them opposed to Trump today.

If you’re a frequent reader of Strength In Numbers, I’m confident you will get a lot of value out of a paid subscription to the site. You’ll get the Tuesday Deep Dive every week, access to a private community for paying members, and first access to our monthly Strength In Numbers/Verasight poll. More than that though, you’ll be supporting original, data-driven journalism about American politics.

Upgrade to support Strength In Numbers

3. Even more numbers!

• The latest Quinnipiac University poll has Democrats up 49-38 for control of the House. That’s an 11-point lead, which nudged our FiftyPlusOne average up to about D+7 late last week. What’s interesting is that Republicans are losing ground over time mostly to the “don’t know” bucket; Since QPac’s July poll, Democrats picked up a point, Republicans shed three, and “no opinion” climbed to 13%. Voters Favor Dems Over GOP By 11 Points For Control Of U.S. House

• It’s really not hard to understand why Democrats are doing so well. In QPac’s poll, Trump is at a 32%-65% approval rating on the economy, 29-65 on Iran, 28-62 on Canada trade, and 28-61 on Ukraine, with voters against him 33% to 59% overall. He’s even underwater on immigration by 18 points. Quinnipiac crosstabs, September 10

• I have really been enjoying Scott Clement’s new newsletter on polls at WaPo. This week he has a lot of great tips about how to read polls. Seven simple tips to understand midterm election polls

• Cook and TelevisaUnivision find that the GOP’s 2024 gains with Texas Hispanic voters have evaporated. Their new poll has Talarico up 63-30 in the Senate race with Latinos, 60-33 for governor, and Trump’s approval with Hispanics down to 33%. Exclusive new poll of Hispanic Texas voters shows collapse of GOP’s 2024 gains

• More Senate polls! CNN/SSRS has Maine at 48-45 and Michigan at 47-44 (conducted August 31 to September 6), in states where roughly two-thirds of voters disapprove of Trump. CNN polls: Maine and Michigan Senate races

• Data for Progress has Trump 16 points underwater on his handling of corruption, but doesn’t trust either party to fix the issue, rating Democrats at -6, and Republicans at -12. Voters Don’t Trust Trump — or Either Party — to Address Corruption

• New reporting by NPR finds Trump administration officials want to pull race questions out of the census and leave millions of people out of the 2030 count. Trump officials propose census changes. Quoting here: “The administration’s proposal, which was released online Wednesday, calls for an unprecedented exclusion of non-U.S. citizens without a green card from a key set of 2030 census results that the 14th Amendment says must include the “whole number of persons in each state.”“

• Consumer sentiment is down a second straight month at Michigan, to 47.8 from 51.7, with Joanne Hsu pointing at fuel prices and trade. New Surveys of Consumers

  

• And this new paper shows primary voters are badly unlike the electorates they’re picking candidates for. Good discussion of how to “make primaries more democratic.” A good read, even if you don’t agree that that should be the long-term goal for primaries. The Problem With Primary Voters

Share

And that’s it for this week! Thanks for reading. Strength In Numbers will be back in your inbox on Tuesday.

Got more for next week? Email your links or add to the comments below!

Leave a comment

In that piece, I dive into the actual polling data on low-turnout midterm voters and find that the Republicans Donald Trump and his electoral strategists are trying to activate with their first-ever midterm convention this week actually disapprove of how Trump has handled the presidency, especially on prices.

Here’s the full story:

And here’s the graph I want to share with the full newsletter audience today. It shows approval of Trump overall and on various issues for people who call themselves Republicans, by those respondents’ stated likelihood of voting in November:

For some added context, here is the same chart among all voters, not just Republicans.

This chart is remarkable (as in the literal sense: you should remark about this, with your friends and peers!) for the story it tells about Trump’s own 2024 voters. First, it shows that the president’s past supporters actually get less supportive of him as they get more politically disengaged. The Republicans who say they are definitely going to vote in November say they approve of Trump at 87%, versus 13% who disapprove. But among 2024 Trump voters who say they won’t vote in November, POTUS’s approval is just 55%, vs 40% who disapprove.

As discussed in the Deep Dive article linked above, the story is much the same on voters’ priorities; low-turnout Republicans care significantly more about inflation and the economy (and less about immigration) than do high-turnout Republicans.

Substack is how I pay my bills. If you want to support independent, data-driven political journalism and get premium posts sent to your inbox up to 3x a week, become a paying subscriber to Strength In Numbers today. Members also get early access to data products, a private community on Discord for our political data nerds, and an ad-free newsletter experience.

Upgrade to support Strength In Numbers

Increasing turnout among low-engagement voters could hurt the GOP in November, not help it.

This is the type of analysis you can conduct when you do the hard work of running an independent politics poll every month. The individual-level data we have captured about Americans this year is very rich and enables us to understand voters better than the pundits — or, in this case, those Republican strategists cheering on Trump’s convention.

The second reason that chart above is remarkable is for what it tells us about low-engagement voters more broadly. See there is a conventional wisdom in Washington that, since low-engagement voters favored Trump in 2024, they must be conservatives/MAGA/whatever. The narrative I hear often from strategists and partisan consultants is that these low-turnout voters who put Trump in the White House in 2024 are durably pro-Trump and trust Republicans more than Democrats to handle big policy issues.

This is, simply, not true. And it hasn’t been for a while; as early as last summer (2025) I was writing about how non-voters had already shifted hard against Republicans because Trump was mismanaging inflation and tariffs. In February I looked again at how “normie” voters were feeling, and it was not good. Yet consultants refuse to update their beliefs about low-engagement voters.

Well, let this chart, and Trump’s palooza, force the updating of the conventional wisdom.

Many in the media are still living in early 2025, when Trump’s approval rating was 45% and the president’s victory had left journalists entranced by the right’s supposed enduring cultural dominance. Then, the popular — and wrong — view was that Trump won because voters supported his policies. Especially among low-turnout voters.

But that was wrong then, and it’s wrong now. Trump won in 2024 because inflation was high under Democrats, so voters — especially low-engagement voters — punished the Democrats. That does not make them conservative as much as it makes them anti-status-quo. Add this to the list of things I wrote about months ago here at Strength In Numbers that people are finally catching on to.

This doesn't just matter for 2026, though. In 2027-28 a lot of Democratic groups are going to spend a lot of money hiring pollsters and consultants that adopt the same flawed beliefs about low-engagement voters as Trump does. Many firms engage exclusively in left-vs-right ideological thinking, leaving no room in their mental models of voter behavior for things like status-quo bias and anti-incumbent or anti-system sentiment — components of political belief systems that are largely if not mostly responsible for Trump’s victories in 2016 and 2024.

It’s far past time for pundits to update the conventional wisdom about low-engagement voters. No, they are not MAGA!

Subscribe now

Related Articles