Twenty-five years ago this morning, 2,977 people were killed in New York, at the Pentagon and in a field outside Shanksville. What followed included a recruiting surge, and the people who answered it are middle-aged now. Many of them are on ballots this year. With Honor Action counted 95 women veteran candidates alone in this cycle, more than double the 2024 figure. Yet new polling says something uncomfortable about the party most of these September 11 veteran candidates chose to run in.

Change Research surveyed 2,045 registered voters nationally between 8 and 13 August, with a modelled margin of error of 2.3 points, and published the results on Tuesday. The firm belongs to the American Association for Public Opinion Research transparency initiative, so the methodology is disclosed rather than asserted.

Start with the partisan split, since it’s the number that matters here. Just 5 percent of Democrats said the Iraq war was worth fighting. Republicans came in at 32 percent. On Afghanistan the figures were 9 percent of Democrats against 36 percent of Republicans. Across all voters, 59 percent said Iraq was not worth fighting and 52 percent said the same about Afghanistan.

Neither war reads as protective either. Only 17 percent of voters said Afghanistan made the country safer, against 32 percent who said it made America less safe. Iraq scored identically at 17 percent, with 35 percent saying less safe. For scale, roughly 90 percent of Americans supported the war in Afghanistan in the weeks after the towers fell.

So a Democratic primary electorate in 2026 contains very few people who believe the defining experience of these candidates’ adult lives was worth having. That is not a claim about whether voters respect the service. It’s a claim about what they think the service was for.

Line the primary results up. The pattern is hard to miss. Maura Sullivan joined the Marines out of Northwestern in 2001 and served in Fallujah in 2005. She lost New Hampshire’s first district on Tuesday by 2.3 points, having outraised the field and led every late poll. Seth Moulton served four tours in Iraq as a Marine, founded a PAC that helped elect 38 veterans, outspent Ed Markey roughly two to one and lost by nearly 30 points. Alex Vindman raised $16.3 million in Florida and lost to Angie Nixon 56 to 44. Ryan Crosswell lost Pennsylvania’s seventh.

Every one of them is an Iraq-era veteran. Every one of them lost to a Democrat who never served. Between them they raised well over $25 million, and it bought four second places. We’ve written each of those races up separately as it happened, treating them as four different stories about name recognition, money and timing. This polling suggests they may be one story.

But there’s a fact about this cohort that a Democratic publication has to sit with. Seventy-eight veterans currently serve in the House. Fifty-nine of them are Republicans and nineteen are Democrats. The recruiting surge that followed the attacks produced a political class, and roughly three-quarters of it ended up on the other side.

So that asymmetry shapes what a service record signals. When most veterans in Congress are Republicans, a uniform in a campaign advertisement reads as coded before a voter has heard a word of policy. Republicans in the Change Research sample were six times more likely than Democrats to say Iraq was worth fighting, and the veterans they elect reflect that. So a veteran Democrat is working against an association her own party helped build.

None of which is an argument against recruiting veterans. Rather, it is an argument about what the recruitment has to overcome. And the site has probably been too quick to treat a service record as self-explaining.

The same survey contains a distinction that gets lost in the war numbers. It is a sharp one. Voters are broadly positive about counterterrorism, and broadly negative about the invasions that followed. For example, 65 percent said airport screening made the country safer. Killing Osama bin Laden scored 57 percent. Set those against 17 percent for each of the two wars and the picture stops looking like a rejection of national security work.

In short, voters are separating two things that campaigns tend to bundle. Protecting the country polls well. Occupying other countries does not. A candidate who talks about readiness, oversight and what the armed forces need is speaking to the first category, while one who leads with a war story is closer to the second.

Look at what the veteran Democrats with the best material this year have been doing, and they already know this. The carrier readiness fight, the promotion freeze, the questions about who is briefing Congress: all of it sits in the first bucket. None of it requires anyone to relitigate 2003. That is probably why it has worked better than the primaries did.

Here’s the limit, and it’s a real one. Change Research asked about the wars. It did not ask about the people who fought them, and those are separate questions. Plenty of voters hold that a war was a mistake while holding nothing against the eighteen-year-old who enlisted for it.

The best evidence on that second question actually points the other way. Gallup, working with With Honor and polling 2,132 adults in August 2025, found 55 percent of Americans saying a military background made them more likely to back a candidate, against 7 percent who said less likely. Even there, though, the partisan pattern reappears: Republicans were at 77 percent, independents at 49, and Democrats last at 40.

There are also duller explanations for this summer available. Sullivan lost to the most recognisable Democratic surname in New Hampshire. Moulton ran against a sitting senator. Vindman was a newcomer to Florida politics. None of those require a theory about Iraq. We’re offering a mechanism that fits four results, not proof that it caused them, and anyone reading this should hold it loosely until November tests it.

Two other findings deserve airing, because both complicate the way this day is usually discussed. The first is proportion. Asked what had the greatest impact on the United States over the past two decades, 41 percent of voters named Trump’s election and re-election and 34 percent named the pandemic. The attacks came third at 19 percent. Among Democrats specifically, 66 percent named Trump.

The second is stranger and worse. Two-thirds of voters said it was at least somewhat likely that people in the federal government had specific warnings and chose to ignore them. A third said it was at least somewhat likely that explosives brought down the towers, and that figure rises to 41 percent among Republicans and 40 percent among voters aged 18 to 34. Only 29 percent credited the absence of another major attack to the government and military doing their jobs. A quarter put it down to luck.

Because of that, the numbers become their own argument for having people in Congress who were there. Not because they can correct a conspiracy theory, since nobody does that in a hearing room, but because the institutional memory of what actually happened is thinning fast, and in eight years it will thin further.

General elections put a different question to a different electorate. That matters. Against a Republican, a service record does the job it has always done well, which is to close off a line of attack before it opens. And the voters most moved by it, on Gallup’s numbers, are precisely the ones a Democrat has to win over rather than turn out.

So the cohort still has its real test ahead. Cait Conley is defending a New York toss-up, and Pat Ryan another one down the Hudson. Nancy Lacore, Rebecca Bennett and JoAnna Mendoza are all on general election ballots. If the argument this site makes is right, then those races go better than the primaries did. If they don’t, then twenty-five years on we will have learned something harder about what the wars did to the politics of the people who fought them.

There is one more reason to watch closely. The generation that enlisted after the attacks is the last one with any adult memory of the day itself, and its members are now in their forties. Whatever they carry into Congress, they carry for a limited time. Meanwhile only twenty-five states require schools to teach the attacks at all, and thirteen more list them as optional. So the argument about what the wars were worth is going to be settled, eventually, by people who were not born when they started.