Now that the primaries are over, it’s time to draw the appropriate lessons, while warning against over-interpretations, and I tried to do both at New York:

Primary season came to an end this week when voters in Delaware headed to the polls on September 15. That same day, early in-person general-election voting started in four states. So it’s a good time to take a step back and look at what the primaries told us about where the American electorate is right now. It’s always important to beware of simplistic narratives about the primaries. We don’t have a national system of nominating contests — even for the national office of the presidency — and the 50 states have widely varying rules, traditions, and party dynamics. But there are some clear trends that may carry over into November and beyond.

A lot of incumbents are leaving, but not involuntarily.

2026 has not been a great year for incumbent officeholders. But most open seats were triggered by retirements rather than primary defeats. According to a tabulation from the Pew Research Center, 74 House members (42 Republicans and 32 Democrats) chose not to stand for reelection, the most since 1992, according to a Brookings Institution study. It should be noted that 28 of these retirees weren’t hanging it up but were instead running for other offices. Still, 11 House incumbents (six Democrats and five Republicans) and two Senate incumbents (both Republicans) lost 2026 primaries for reelection. That’s no higher than the “kill rate” in the last midterm election, in 2022, when 14 House members and no senators lost primaries.

While Democrats did shift a bit to the left, the trend was overstated.

One of the most prominent national narratives about the 2026 primaries was that Democrats made a decided shift toward progressivism or even socialism. The most obvious reason for skepticism about this interpretation is that it represents a self-serving story line for both the left and the right. Republicans have decided that given Donald Trump’s terrible job-approval ratings, it’s critical to focus on the alleged extremism of the opposition party. And Democratic progressives see 2026 as a unique opportunity to challenge the ancient assumption that their candidates aren’t electable.

But ignoring the hype and looking at the results, any shift to the left was pretty modest and to a significant extent limited to noncompetitive districts. The latter were typified by three June 2026 congressional primary wins in deep-blue New York City districts by progressives backed by Mayor Zohran Mamdani, soon followed by another win for a candidate of the left in deep-blue Denver. More generally, the ideological character of Democratic primaries was mixed. According to a deep dive by the Brookings Institution on 43 Democratic House and Senate primaries in competitive jurisdictions, 34 were won by mainstream Democrats, eight by progressives, and just one (in a Lansing, Michigan, House district) by a member of the Democratic Socialists of America. Looking at these primaries from a perspective of factional wins and losses, mainstream Democrats won half their races, progressives won 17.8 percent, and DSA members won 20 percent.

Two Senate primary wins by non-DSA progressives received disproportionate attention for all sorts of reasons. Graham Platner of Maine chased moderate Governor Janet Mills right out of that state’s crucial Senate race, but then was forced to withdraw after multiple sexual-misconduct scandals broke. He was replaced on the ticket, however, by another progressive, Troy Jackson, so it remained a progressive win. And in Michigan, progressive former state health director Abdul El-Sayed narrowly defeated moderate congresswoman Haley Stevens in a race that wound up heavily focused on the two candidates’ very different attitudes toward Israel. These shifts to the left were partially offset in the midterm-primary story by the startling defeat in Wisconsin’s gubernatorial contest of DSA member Francesca Hong by the more moderate David Crowley, who had dropped out of the contest earlier.

Loyalty to Trump remained the key factor in GOP primaries.

Many eyebrows were raised over an unusual number of Trump defeats in GOP primaries this year, including gubernatorial contests in Georgia, Iowa, Minnesota, South Carolina, and Wyoming. He did better in federal elections, losing no Senate primaries and only four House primaries (out of 229). But it’s worth noting that in almost every case, the vanquishers of Trump’s endorsees swore fealty to the president and in many cases tried to out-MAGA their opponents. The one viable GOP candidate in a contested primary who shook a fist publicly at Trump, Kentucky’s Thomas Massie, lost. Meanwhile, the president’s “revenge tour” against senators who voted to convict him of impeachable offenses in 2021 took down Bill Cassidy of Louisiana, and alleged lack of fidelity to the MAGA cause led to the more surprising purge of John Cornyn of Texas.

Texas, where scandal-plagued MAGA favorite Ken Paxton is in a tough race with Democrat James Talarico, joined Michigan and possibly Maine as states where competitive Senate races could be lost due to the perceived ideological extremism of major-party nominees.

Electability was an issue in both parties.

Given the intensely competitive national electoral landscape and the general perception that this is a very high-stakes election, it’s not surprising that primaries in both parties frequently revolved around claims of superior candidate electability. As usual, moderate Democrats and Republicans argued they would have a better chance to win over swing voters. But more ideologically astringent candidates had their own electability claims, with both progressive Democrats and ultra-MAGA Republicans asserting that they would do a better job of mobilizing base or low-propensity voters in November. Progressive Democrats also contended they were more in tune with public opinion than their moderate rivals on matters ranging from aid to Israel to health care to corporate corruption. Democrats generally emphasized affordability issues in anticipation of general-election clashes with Republicans, while Republicans depicted themselves as warriors against socialism and “wokeness” as rehearsals for their own general-election campaigns.

Polling primaries is hard work.

Public-opinion experts have long warned that polling primary contests accurately is difficult due to unpredictable turnout patterns and response rates and the limited amount of money available for competent polling. And for consumers of polling data, the relatively sparse number of public polls in primary contests (especially in House races) makes the available data even less reliable. What stood out in 2026 were a couple of high-profile “misses” in Democratic primaries, like David Crowley’s upset win over Francesca Hong in Wisconsin’s chaotic gubernatorial contest and the underperformance of poll-driven expectations by Michigan Senate candidate Abdul El-Sayed. There were also a couple of truly “fake” polls that were initially reported as legitimate. A parallel development involved suspicions of campaign or candidate manipulation of the increasingly influential prediction markets. All these problems both fed and drew upon mistrust in elections and the infrastructure surrounding them, which are very likely to reach critical levels (especially among conspiracy-theory-prone MAGA Republicans) during the general election.