We checked The NYT’s data. Moderates still win.

Democrats win more often when they run candidates who match their districts, not their Bluesky feeds.

Sen. Joe Manchin receiving the wisdom of moderation from above (Photo by Andrew Harnik/Getty Images)

Recently, The New York Times published an editorial laying out a data-backed case for the Democratic Party recapturing the political center of America. By analyzing hundreds of federal election results across the nation in 2024 and comparing the performance of moderate lawmakers to nonmoderate ones, the Times concluded that moderate lawmakers enjoyed a clear and obvious electoral advantage.

The uproar was instantaneous.

Progressive pundits, data analysts, and academics alike hammered the editorial as soon as it was released, taking issue with multiple design, framing, and methodological choices that the editorial board made.

G. Elliott Morris of Strength In Numbers and Adam Bonica of Stanford University argued that the board drastically overstated the impact of moderation by not controlling for incumbency and fundraising. Since more incumbent lawmakers were well-funded moderates, and incumbents tend to overperform the presidential margins, the argument went, a comparison that doesn’t account for these factors would naturally inflate the value of moderation.

Second, Morris and Bonica argued the groupings used by the Editorial Board were far too coarse. Candidates were split into two buckets: “moderates” and “nonmoderates,” based on whether they got the support of certain outside groups like the Blue Dog PAC. Everyone who didn’t qualify as a moderate was classified as nonmoderate.¹

Morris and Bonica took issue with this, arguing that the comparison was unfair; many of the lawmakers who the Times classified as nonmoderates were actually just phantom candidates who had no money and no presence. The pair argued that these so-called phantom candidates failed irrespective of their ideology.

I think that PAC and endorsement classifications, like those done by the newspaper, are a pretty good way of getting at a lawmaker’s ideological presentation (Blue Dogs are pretty moderate, while Justice Democrats are quite progressive). But I also share some of the methodological concerns raised by Morris and Bonica, and I think it’s worth further analysis. Thankfully, The New York Times was kind enough to give us its full underlying dataset, and so I tried to replicate² its study in a way that corrected for these things.

The first change was to use the Wins Above Replacement scores from my work at Split Ticket, which grades candidates on how their electoral performance compared to an estimate of how the race should have gone, based on fundamentals like seat partisanship, incumbency, demographics, and money. As a result, the WAR model controls for factors that Bonica and Morris criticized the Times for omitting.

The second change was to compare candidates who moderate PACs endorsed or donated to with candidates who received the support of progressive and far-right PACs. This accounts for a key (and correct) criticism that Bonica and Morris made, which is that the original analysis from the Times relied on a contaminated sample — it is, in fact, cleaner and more robust to compare well-funded moderates to well-funded progressive (or far-right) candidates.

Yet after doing all this, my results are very similar to those from the Times and validate its findings. Moderates enjoyed a smaller but very clear advantage over their more ideological counterparts. This also aligns with my previous studies.

Moderate Democratic candidates enjoyed a nearly four-point advantage in vote margin over progressive candidates. And at the extremes, this gets magnified: The average WAR of the 20 Blue Dogs-endorsed candidates was +4.5, which was nearly nine points better than the average WAR of the nine progressive candidates endorsed by the Justice Democrats.³

This is also true for the Republicans, but less so. It is still possible to find credible, overperforming Republican moderates like Brian Fitzpatrick and Mike Lawler, but this is getting increasingly hard to do as the president cements his hold on the Republican Party and makes alignment with MAGA a borderline necessity to win a primary.

Do the Blue Dogs-endorsed candidates do better than Justice Democrats’ because they’re more moderate or because of other factors, like television presence and charisma? I leave it to the reader to decide if correlation is causation in this case. But what I will say is that for one reason or another, moderates consistently do better.

Ideology obviously isn’t everything in this world, and there are strong progressives (Chris Deluzio) and weak moderates (Jim Costa). But in general, the data says that the Times is correct: Moderates tend to do better.

Who cares?

Democrats just had blowout victories in the New Jersey and Virginia gubernatorial races. The political environment has arguably never been worse for Republicans since the 2018 blue wave. So why are these data-heavy debates about the value of moderation even relevant?

First, part of the reason that moderates perform better is that they are more accurately representing the views of their constituents. Far-right and far-left candidates are genuinely out of step with many voters, even in deep-blue districts, and that’s a problem for anyone who cares about democracy.

Second, I’m worried about the authoritarian overreach of the Trump administration, and ensuring the Senate is held by the Democrats is extremely important to curb the president’s increasing lawlessness.

In order to flip the Senate, Democrats are going to need more than just another blue wave. They’re going to need credible candidates who can win over a lot of swing voters — and the evidence we have suggests that those voters are relatively moderate.

Let’s assume, for the sake of argument, that Democrats flip North Carolina and Maine, bringing them to 49 seats.

What are seats number 50 and 51?

Adjusting for incumbency, these are the types of overperformances Democrats are going to need to flip the edges of the battleground:

In other words, Democrats are going to need significant candidate effects in at least two states in order to win the Senate, even if they get a blue wave, win North Carolina and Maine, and hold all their existing seats. This is a tall ask.

It is made easier by the fact that former Senator (and proven overperformer) Sherrod Brown is running for Senate in Ohio. But how will the party win in Iowa, Florida, Texas, or Alaska? If the data provides any suggestions, moderates have been significantly more likely than not to get those four- or five-point overperformances. It is, after all, an important part of how the party has managed to hold on in states like Nevada, Arizona, and Georgia.

---

I should note that most of the people on the other side of the moderation debate have no issues whatsoever in accepting the value of moderation when it comes to the Republican Party. For instance, Morris — who is perhaps the loudest voice in favor of the “moderation is overrated” theory — has absolutely no problem with the idea that extremist nominees like Blake Masters and Herschel Walker cost Republican candidates dearly in swing seats. But for some reason, he is unable to imagine this happening to Democrats and refuses to believe that progressive Democrats could pay a serious ideological penalty.

This misconception actually drives a lot of Morris’ own WAR modeling, in which he claimed that it is misleading to use Kamala Harris’ performance in an evaluation of overperformance. Morris claimed that the correct baseline is to evaluate nominees against the types of candidates who tend to come from the seat’s voters in general, because Democratic voters in purple seats “would never elect a candidate as progressive as Harris.”

This is how Morris ends up labeling an extremely unpopular candidate like Iowa Republican Rep. Mariannette Miller-Meeks as above-replacement, despite her winning a Trump +8.6 seat by just 0.2 points as an incumbent. And if you look through Morris’ WAR model, you’ll find more inexplicable results like this.

For instance, Democrat Pramila Jayapal, who won a Harris +78 seat by 68 points as an incumbent, is somehow rated as above-replacement, whereas Democrat Adam Gray, who defeated incumbent Republican John Duarte in a Trump +6 seat, is graded as below-replacement.

Notwithstanding the fact that this is an approach deeply fraught with analytical flaws⁴, it’s also just wrong. Politics never works only in one direction, and the same thing that happens to Republicans can easily happen to Democrats. There is simply no reason to believe that swing-seat Democrats are invulnerable to picking more progressive candidates like Harris, especially as the rage toward the “establishment” begins to consume its base.

In the Texas Senate race, for instance, a progressive candidate who called Gov. Greg Abbott —who uses a wheelchair — “Governor Hot Wheels” is now leading several primary polls. In the Maine Senate race, a populist, progressive candidate with a Nazi tattoo has significant support from several Democratic leaders and is competitive in polls against incumbent Gov. Janet Mills.

And in 2022, in a purple suburban Oregon seat, Democrats ousted long-tenured moderate incumbent Kurt Schrader in favor of progressive attorney Jamie McLeod-Skinner, only to see her post a big enough underperformance to lose a tight election. In other words, Democratic primary voters are not immune to swooning over progressive nominees.

But there is still a point I agree with Morris and Bonica on: Ultimately, candidates from the progressive wing of the party did not lose most of the seats that slipped away in recent cycles. While the losing moderates often overperformed (and significantly so), the moderation shown by many of those losing candidates is often insufficient, especially in a world where ticket-splitting has plummeted.

In today’s world, it is extremely unrealistic to expect downballot Democrats to win a seat that their presidential nominee simultaneously loses by double digits. Picking credible, mainstream moderates can help you win an R+5 seat, but it is a very tall ask for them to win R+10 ones.

And if the party has become far too liberal or toxic to compete presidentially in places like Ohio, then it should be unsurprising that several moderate Democratic candidates still lose those races, despite overperforming.

1. “Moderates” were defined as those who received support from Bridge the Gap PAC, Welcome PAC, Blue Dog PAC, No Labels Problem Solvers PAC, New Democrat Coalition PAC, Republican Governance Group PAC, or Republican Main Street Partnership PAC and received no money from progressive or right-wing groups. Everyone else was coded as a “nonmoderate.”

2. The methodological details I used are as follows: Progressives are classified as those who received donations from Progressive Caucus PAC or were endorsed by either Sunrise or Justice Democrats. Hard-right Republican candidates are defined as those who received donations from House Freedom Fund or Ultra MAGA PAC. In keeping with the criteria used by The New York Times, moderates are defined as candidates who received donations from Bridge the Gap PAC, Welcome PAC, Blue Dog PAC, No Labels Problem Solvers PAC, New Democrat Coalition PAC, Republican Governance Group PAC, or Republican Main Street Partnership PAC. Those who received donations or endorsements from any of the listed progressive or hard-right groups were excluded from the “moderate” label. Split Ticket’s Wins Above Replacement scores were used to track overperformance.

3. You may notice that Justice Democrats have only endorsed a few candidates, but the underperformance of these groups is so sharp, and the combined N is sufficiently large, that I am confident in its statistical significance. Seven of the nine Justice Democrats-endorsed candidates had negative WAR scores; The only ones with above-replacement scores were Rashida Tlaib and Raul Grijalva. Of the 10 worst Democratic underperformers in the 2024 House elections, four (Summer Lee, Ro Khanna, Ilhan Omar, and Pramila Jayapal) were Justice Democrats.

4. If you want to compare a moderate lawmaker to a hypothetical moderate replacement in order to measure their quality, that’s a defensible choice by itself (though not one that I would advise). But you absolutely cannot then use those scores to argue that moderation is irrelevant. Because that boils down to the circular logic of “most moderates would overperform in this district, ergo moderation is just vibes.”

In general, I think the analyses from these folks around the value of moderation contain a number of numerical and logical flaws. For instance, Morris compared Rebecca Cooke to Matt Cartwright, but never made note of the fact that Cooke was running against an incumbent, while Cartwright ran as an incumbent. When you account for this, Cooke’s 2.7% loss in a Trump +7.5 seat is actually significantly more impressive than Cartwright’s 1.6% loss in a Trump +8.6 seat — it’s why we have Cooke’s WAR at D+7.3, while Cartwright’s is just D+2.9.

These are basic analytical mistakes being made here. There is, once again, an inability to imagine that the model may be buggy, and that purported mathematical sophistication may just yield findings that most practitioners would quickly identify as incomprehensible in the face of the data.