The emerging consensus among pundits is that the GOP is underperforming relative to expectations in the 2022 midterm elections. The normal account of U.S. midterm elections is that the incumbent president’s party loses seats in midterm elections. In 2022, this would mean that the Democrats should expect to lose seats in the House, the Senate, and in statewide elections. This isn’t an ironclad law—witness how the GOP did in 2002, in the midterm elections of George W. Bush’s first term—but it’s a pretty robust pattern.

An anti-incumbent midterm swing would seem to be particularly likely in a highly polarized environment with a president whose approval ratings remain very low. And yet the Democratic candidates in last night’s elections did much better than anticipated. For example, President Trump carried my own congressional district* by 11 points in 2020, but in last night’s special election to name a replacement for Tom Reed (R), the GOP candidate won by only 5 points. That’s not a swing against the Democrats, it’s the exact opposite.

Given the pundits’ consensus interpretation of these results, one might expect the GOP to infer that nominating extreme candidates like Carl Paladino is a mistake, and that to win elections they ought to run more to the center. That would be a good story of electoral accountability in which partisan competition incentivizes parties to run towards the center as they learn from the results of past elections about where the median voter’s preferences are.

But there’s a problem: President Trump. President Trump has regularly alleged that the results of the 2020 presidential election were rigged, somehow marred by fraud and biased against him. In some new research, Andrew Little, Andrew Mack, and I are studying what happens to partisan competition when one party believes that elections are fraudulent. The short answer is that under these conditions, parties cannot learn properly about what voters want, so they do not converge to the center in the way that they would if they trusted that election results were unbiased.

The implication is, the GOP may infer from the results of last night’s elections not that their candidates are unpopular, but rather instead as further evidence that the system is rigged against them, driving the parties apart rather than pulling them towards the median voter.

A full treatment will be available in a working paper soon to come, and that version will state our results precisely. But I can offer the intuition behind our analysis here. Imagine that you are a party that suffers an electoral defeat. You might think that this means that your platform wasn’t very popular. But what if, instead, you believed that elections might be biased against you?** You would have a tough time attributing your party’s performance to your platform versus what you think is fraud. So you might respond to your defeat by concluding that your platform should remain where it is and that the system is unfair.

Your opponent, in turn, can offer a more extreme platform then they would otherwise because you have ceded some of the center ground on the basis of your belief that you are more popular than you are (which, again, emerges because you attribute your own poor showing to fraud rather than to your own unpopularity).

Some of this divergence will dissipate over time in repeated elections, but under generous conditions the partisan divergence will be strictly greater than if you didn’t believe in fraud.

There is a lot more to dig into here—watch this space for more. But the key takeaway is that our results make me skeptical that the GOP will infer from last night’s results that they need to offer more moderate candidates with more moderate platforms. It could be that they are right not to make much of last night’s results. Maybe they were a fluke, or maybe they were unusual. But this is a further corrosive implication of the widespread belief among the GOP that the 2020 presidential election was fraudulent.

NOTES

* NY-23, soon to be redistricted away.

** In fact, these results emerge even when the party has no intrinsic interest in believing in fraud. You can get these results purely from motivated reasoning from one or both parties, who find comfort in the belief that their platform is close to the median.