Skip to the pageelection.run

Neither wing won the wave

Progressives and centrists each say the midterms prove them right. The seats they ran in prove something else.

Illustration of Representatives Alexandria Ocasio-Cortez and Hakeem Jeffries as black and white paper cutouts in front of the U.S. Capitol, which is printed in red 0s and 1s under a sky of blue hexagons.
Illustration: election.run

With the midterms two weeks in the rearview mirror, Democrats have a new intramural fight on their hands. Progressives and moderates are battling over which way the party heads into 2028.

On one side, left-wing voices are saying that progressives overperformed. On the other side, centrist Democrats say the opposite. So who’s right?

The claims

What each side is saying

Both camps are pointing to the same results and using the same words. “We are the majority makers in this Congress that helped flip these seats,” Randy Villegas, a progressive who leads in CA-22, said on a call with Progressive Caucus leaders. “The majority makers are in the center,” Jim Kessler of the centrist think tank Third Way told Newsweek.

Justice Democrats say progressives brought the energy and platform that turned the tide. A CNN analysis found Progressive Caucus members ran a few points behind the New Democrats and Blue Dogs, measured against Harris.

Let’s find out. Who overperformed, and who overperformed once we account for more than the raw numbers?

The test

Putting the claims to the numbers

  1. Progressives ran a little behind the wave. Compared to the overall 14.6-point blue wave swing towards the Democrats, the 45 Progressive Caucus members who aren’t also New Democrats averaged 0.57 points behind the wave.
  2. Centrists ran a little ahead of it. Meanwhile, the 75 members from the two centrist blocs, New Democrats and Blue Dogs, ran 1.91 points ahead of the wave.
  3. A 2.48-point gap. This is the gap centrists point to. On raw numbers, centrists averaged 1.91 points ahead of the wave and Progressive Caucus members 0.57 behind, a gap of 2.48 points. Not a huge difference, but enough to have tipped the last three presidential elections.
  4. But they ran in very different seats. The average Progressive Caucus seat went for Kamala Harris by 32.88 points in 2024. The average centrist seat went for her by 13.45.
  5. The bluest seats gained the least. In the bluest seats, with less room to grow, Democrats tend to land below the wave whatever their politics. The line shows where they usually land.
  6. Now measure each member against their seat. Here they are again, 2.48 points apart on average. But what happens to the gap when we compare each member with Democrats in similar seats?
  7. From 2.48 points to 0.28. The line now marks the usual, where a typical Democrat would land in the same seat and race. Measured from there, the gap nearly disappears to only a 0.28-point difference between the camps.
−10+10+20+30The blue wave14.6-point swingBehind the waveAhead of the waveThe usualfor the seat and raceBehind the usualAhead of the usualProgressive Caucus only, 45New Democrats and Blue Dogs, 75Average member 0.57 behindAverage memberAverage member 1.91 aheadAverage memberPoints ahead of the waveThe wave−10+10+20+30Even+20+40+60SwingHarris 2024 marginDeep blueAverage centrist seatAverage Progressive Caucus seatWhat Democrats usually score in a seat this blueTypically 2.32 aheadTypically 0.27 behind0.57 behind1.91 ahead2.48points −10+10+20+30The blue wave14.6-point swingBehind the waveAhead of the waveThe usualfor the seat and raceBehind the usualAhead of the usualProgressive Caucus only, 45New Democrats and Blue Dogs, 75Average member 0.57 behindAverage memberAverage member 1.91 aheadAverage memberPoints ahead of the waveThe wave−10+10+20+30Even+20+40+60Swing seatsHarris 2024 margin in the seatDeep blue seatsAverage centrist seat, Harris +13.45Average Progressive Caucus seat, Harris +32.88What Democrats usually score in a seat this blueTypically 2.32 aheadTypically 0.27 behind0.57 behind1.91 ahead2.48points

Each of the 45 Progressive Caucus members who are not New Democrats and the 75 centrists, New Democrats outside the Progressive Caucus and Blue Dogs, is placed by how far they ran ahead of the wave. On average, Progressive Caucus members ran 0.57 behind the wave and centrists 1.91 ahead, 2.48 points apart. Their average seats went for Harris by 32.88 and 13.45 points in 2024, and Democrats in deep blue seats tend to land below the wave. In a seat like the average centrist's, Democrats usually land 2.32 ahead of the wave, and in one like the average Progressive Caucus member's 0.27 behind the wave. Measured against what is usual for each member's seat and race, the averages end up 0.28 points apart, with a range from 2.08 behind to 1.52 ahead.

Widening the groups doesn’t change the story. However you sort who’s a progressive and who’s a centrist, nothing meaningfully changes the finding that the ideological label has a near-zero effect. Put the 29 members who sit in both camps with either side, and progressives run 0.61 behind or 0.20 ahead in similar seats.

The same pattern holds for Republicans. Freedom Caucus members trail Main Street Caucus members by 1.84 points on raw numbers, but comparing similar seats puts them 0.33 ahead.

Even though caucus membership didn’t seem to matter, the races still differed wildly. So we widened the lens to 275 House races and asked how much of those differences we could explain.

The read

What the numbers can and can’t explain

  1. Why did races differ?. Some races ran far ahead of the wave and some far behind. How can our model explain these differences?
  2. Where the race was run: 44.4%. In our model, the seat’s past voting, who lives there, and how the rest of the state voted this year account for 44.4 percent. We call this the fundamentals of a seat, something that no candidate can bend.
  3. Incumbency and past performance: 14.1%. Whether an incumbent ran, and how that member did against the presidential ticket in 2024, account for 14.1 percent, the next largest share.
  4. Other measured factors: 6.4%. Fundraising, small-dollar giving, prior office, and how much of the vote is still out add 6.4 percent.
  5. Ideology: 1.4%, as much as random numbers. Our ideology measures account for 1.4 percent, about a tenth of incumbency’s share. Any measure picks up a little by chance. Random numbers in their place would also get 1.4 percent on average.
  6. “Dark matter”: 33.7%. Our model leaves 33.7 percent of the differences between races unexplained. Call it the dark matter of campaigns. It may hold candidates’ charisma, bad weather, or anything else under the sun.
Each hexagon is one percent of the differencesWhere the race was run44.4%Incumbency and past performance14.1%Other measured factors6.4%Ideologysame as random numbers1.4%Dark matter33.7% Each hexagon is one percent of the differences between racesWhere the race was run44.4%Incumbency and past performance14.1%Other measured factors6.4%Ideology1.4%same as random numbersDark matter33.7%

A grid of 100 hexagons, each one percent of the differences between races, fills block by block, largest measured block first. In our model, where the race was run, the seat's past voting, who lives there, and how the rest of the state voted this year, accounts for 44.4 percent of the differences between races. Those describe the setting of each race, not what shaped it. Incumbency and past performance account for 14.1 percent, and other measured factors, fundraising, small-dollar giving, prior office, and how much of the vote is still out, for 6.4 percent. Our ideology measures account for 1.4 percent, about the same as random numbers (1.4 percent). The model leaves 33.7 percent unexplained, the dark matter of campaigns.

Neither wing has a mandate, at least from this analysis. District differences explain most of the gap between progressive and centrist House incumbents. In similar seats, neither wing ran clearly ahead. Incumbency and past performance explain more than our ideology measures.

The take

Where the numbers stop and judgment begins

Real life is far messier. A candidate and their ideology do not come apart as cleanly as they do in our model. Past performance could reflect either ideology or candidate appeal. Same with comparisons to the 2024 ticket. Even “where the race was run” isn’t some unbreakable rule of nature. Keep that in mind whenever a model sorts behavior into neat categories, in politics or beyond.

On the ideology front, voters may read ideology off the party label, so a candidate’s own politics get layered into the party’s brand, for better or worse. That’s probably why our ideology measures look like noise. The unexplained remainder, our dark matter, could hold any mix of what our stochastic world throws at a race beyond our models, from a candidate’s brand to bad weather.

Susan Collins shows how far a brand can go. For decades Maine voters judged her apart from her party, and this year she still ran 15.6 points better than the wave, only to fall to Democrat Troy Jackson. In this midterm, even one of the strongest personal brands in Congress could not outrun the fundamentals.

Up next

The question this leaves open

So if candidate ideology isn’t moving the needle in House races, what exactly is it moving? Next week, we will tackle what this progressive versus centrist debate is really about: what voters say Democrats stand for and who is really shaping the party’s image.

How we measured

Every number here comes from one table of 2026 House races, run on counts as of Nov 15 from civicAPI, with later AP and state counts in 9 districts. The method, the code, the data, and every source are on GitHub.

Get Run Report

Every piece is free to read. Here is how to keep up.