republicans likely to sweep

It was just two weeks before the election, and the Republican Party was surging.

For anyone who followed politics, it was impossible to escape the news. If you checked Politico, you’d be presented with articles telling you how to “[Make] sense of the GOP’s October surge.” Go over to the New Yorker and you’d see an extensive feature explaining “Why Republican Insiders Think the G.O.P. is Poised for a Blowout.” For the other side of the aisle, the New York Times would tell you that the Democrats were “Fearing a New Shellacking.” Scroll down a bit further to the Times opinion section, and you’d see an article purporting to explain “Why Republicans Are Surging.” Switch over to leading probabilistic forecasts, and you'd see exactly this happening: a once-solid Democratic position eroding rapidly, with Republicans clearly holding the momentum and on track to be favorites on Election Day.

In case you can’t tell what I’m getting at here, it’s this: every single one of these articles were released in advance of the 2022 midterms. In fact, almost all of them were released within the same few days in late October. This is the forgotten part of that race—the big part about it that was buried under all of the media’s self-congratulatory postmortems. Right at the closing stages of the election, after months of correctly covering the race as a more-or-less neutral year, nearly every single elite political commentator in the country flinched. On the forecasts and aggregates, they opened the floodgates to every poll that showed Republicans ahead. When polls found Democrats ahead, the pollsters who conducted them would use their platforms to talk their own data down. Opinion writers like David Brooks, linked above, pre-wrote entire narratives about how Democrats lost by being out-of-touch elites who went too far to the left. Even the most prestigious analysts in the country would change entire state ratings with the sole purpose of finding some way to show the GOP ahead.

In short, they used everything in their power to make Republicans appear in as strong a position as possible. They ended up being completely wrong, and they also ended up facing no consequences for it. Now, with the lights as bright they can be, they may be doing the exact same thing again to an even more extreme extent. Here’s how the stars might be aligning for a polling error that favors Trump in 2024, or: how pundits and pollsters might be courting complete disaster so they can cover their own asses.

What 2022 Doesn’t Tell us About 2024…

To begin, I want to establish something important. Unlike what many said at the time and are still saying now, something big did happen in polling and punditry in 2022. It was real, and it caused people to significantly, and, in my mind, unreasonably underestimate Democrats by Election Day that year. But while this story does provide many valuable lessons for 2024, it doesn’t prove something that many Democratic partisans would like to believe: that all polling now is essentially fake, and that any bad numbers Democrats see are a result of partisan Republican pollsters flooding the zone.

You can call that the Simon Rosenberg theory of politics, and it’s an example of something with a grain of truth to it getting pushed way, way too far. The core idea behind it is very true and very important: that, in 2022, a huge number of unreliable, Republican-leaning polling firms were added, and that they skewed some averages significantly to the right at the closing stages of the race. I will be the first one to tell you that this absolutely happened, that it still stands to be a potential problem going forward, and that it is something that aggregators and modelers should take far more seriously than they do. It’s something I’ve written about extensively. What I object to, however, is those who attempt to use the fact that this happened once as a get-out-of-jail-free card for any and all negative signs that Democrats ever see.

This is why I mention Rosenberg here, as he has made a very lucrative career of doing exactly this. A literal Democratic hack who spent decades in D.C. fighting on behalf of Clintonism against “labor unions and old-line liberals,” Rosenberg received a lot of attention for declaring that the 2022 elections wouldn’t be a red wave very early on. Since then, he has spent the past two years making it very clear that he does not, in fact, have any special insights, but instead just says that Democrats are always going to win no matter what. In this effort, he has blamed supposed “zone flooding” by right-wing pollsters for any and all problems Democrats have faced. This has included:

• The poor approval numbers Biden saw at the beginning of 2023, when he launched his campaign. Rosenberg’s only reference to the poor nonpartisan polls that Biden saw at the time was when he brought up a poll from The Washington Post that he said shouldn’t have been published on account of being too bad for Biden.
• The polling deficits Biden saw at the end of 2023 following the October 7th attacks. Again, Rosenberg also said that his readers should ignore a bad nonpartisan poll for Biden that came out around that time, this time from CNN, on account of it being too bad for Biden.
• Recent shifts towards Trump in the polls this month, which he has once again said is due to what he calls “red wave pollsters” flooding the averages.

For reference, Rosenberg currently defines “red wave pollsters” as such:

American Greatness, Daily Mail, co/efficent, Cygnal, Echelon, Emerson, Fabrizio, Fox News, Insider Advantage, McLaughlin, Mitchell Communications, Napolitan Institute, Noble Predictive, On Message, Orbital Digital, Public Opinion Strategies, Quantus, Rasmussen, Redfield & Wilton, Remington, RMG, SoCal Data, The Telegraph (sic), Trafalgar, TIPP, Victory Insights, Wall Street Journal (sic).

There are some obvious problems with this list, like the fact that many of the pollsters he has included on it—namely Fox News, the Wall Street Journal, and Emerson—are actually non-partisan pollsters, entirely different from the likes of Trafalgar or Rasmussen. He also missed Patriot Polling, which is inexcusable for anyone trying to explain the world of right-wing polling slop. But the most concerning thing for him is this: that this purported gap between these “bad” pollsters and the “good” pollsters doesn’t really exist.

Back in 2022, there was a clear gap between the national environments presented by different types of pollsters. Partisan polls from both Republican firms showed a clear national Republican advantage, while nonpartisan polls showed what was essentially a tied race. Things were even more extreme in statewide polls—in Washington, for instance, GOP-aligned polls erred by as much as 13 points compared to the final result. Notably, this gap showed up as a result of just splitting pollsters according just based on one very relevant aspect of their background: their partisan affiliation (or lack thereof). There was no cherry-picking involved, unlike with Rosenberg’s specially-selected burn book.

Things are different now. As shown by Eli McKown-Dawson in the Silver Bulletin, even removing Rosenberg’s hand-selected list of “bad” polls from the national averages only makes a marginal difference for Kamala. After testing his recommendations and cutting out every pollster from his personal burn book, McKown-Dawson found that his site’s national average only shifted to the left by a mere 0.4 points. When he took a different route and just cut out low-quality pollsters, Trump’s odds in their forecast actually improved. Although this article doesn’t close the book entirely—McKown-Dawson notably didn’t test Rosenberg’s claims by adjusting state polling averages, where Rosenberg has said that the alleged zone-flooding is at its most prevalent—it is a significant point against claims from the likes of Rosenberg that what we’re seeing now is an exact replay of what happened two years ago. The gap between partisan and nonpartisan firms from 2022, so extreme and influential back then even in the most high-volume averages, just isn’t on display now.

As for why this is the case, there are a few reasons that stand out. Part of it is that even the most notorious bad actors from two years ago are simply putting out less extreme results compared to what they put out in 2022. Trafalgar, for instance, repeatedly put out polls that year that showed Republicans either leading or in striking distance in solid Biden states, from New Hampshire to Washington to Colorado to even fucking Vermont. This year, they’ve limited themselves to core swing states, and although they have “found” Trump leads in almost every single survey they’ve conducted, it’s always only been by a few points. It’s enough for them to push their preferred narratives and get appearances on Fox, but also not extreme enough to make them stick out by as much as they did in 2022 if Democrats win. But while these changes are a big deal, they’re only half of the story, because partisan pollsters were far from the only problem we had in 2022.