The Moulton Markey Senate primary is five days away, and early voting closes tomorrow. So the arguing’s nearly over. A University of Massachusetts Amherst and WCVB survey put Ed Markey at 60 percent and Seth Moulton at 32, with 7 percent undecided. YouGov ran it between 5 and 12 August. The sample covered 445 self-identified Democratic primary voters, with a margin of error of 5.8 points.

A second survey tells a gentler version of the same story. Suffolk University and the Boston Globe polled 500 likely primary voters and found Markey ahead by roughly 14, at about 52 to 38. The two polls disagree by fourteen points on the size of the lead. Neither disagrees about who holds it.

Markey turns 80 before the election, and he’s been in Congress since 1976. Moulton is 47, served four tours in Iraq as a Marine, and beat a sitting Democrat to win his own House seat in 2014. Because his whole case rests on age, what sits under the topline matters more than the topline.

Take the group at its own word first, since the numbers hold up under checking. Moulton founded Serve America in 2017 for the next year’s midterms. Its own account of that first cycle claims 21 winning House candidates, 18 of whom helped flip seats from Republican hands. Since then the running total has reached 38 Democrats elected, with 24 of them in districts that had been red.

This cycle it has backed 14 more candidates. Thirteen have already won their nominations. Whatever else is true about the veteran recruitment approach, the pipeline part works. Moulton built the version other groups copied.

New Politics recruited Moulton himself to run in 2014, and its executive director Emily Cherniack has described what the alumni are doing now as what servant leaders do for each other. Read charitably, that’s a network paying a debt. Read less charitably, it’s a candidate calling in favours because the ordinary tools have stopped working.

Moulton picked age as his opening, and the pollsters found that Markey turned it into an asset. Markey was the more experienced choice to 91 percent of them. Meanwhile just 9 percent named new leadership as the quality that mattered most, well behind the 24 percent who wanted a candidate matching them on issues.

Then there’s the transgender question, which several of the UMass pollsters flagged as the worse problem of the two. Moulton took a harder line on transgender athletes after the 2024 election. In this survey, 68 percent said Markey was better placed to speak for LGBTQ people. Another 67 percent said they would be less likely to back any candidate who opposed transgender rights. Moulton expanded an earlier apology during the final debate, though by then the ballots had already started moving.

Notice what has not appeared anywhere in this list. Nobody polled has called Moulton’s service a liability. Instead it simply hasn’t been the axis the race turned on, which is a different and quieter kind of problem for the argument this site usually makes, because a case that never gets tested also never gets confirmed.

Here’s the finding that should give this site pause. Tatishe Nteta directs the UMass poll. He listed the groups these voters think Markey would speak for better: women, progressives, union members, seniors, and veterans. Indeed, he flagged that last one as the shock, given Moulton’s years in the Marines.

His one clear edge in the survey was among the young, who turn out least in primaries. So the Marine is winning the group least likely to vote. Meanwhile he is losing the group whose experience he shares.

Be careful how much weight that carries. Veterans are a small slice inside a poll of 445 people. Findings on small groups are the least stable thing any survey gives you, so treat the number as a flag rather than a fact. Even so, nobody expected this one. Because the pollster reported it rather than a campaign, it is hard to wave away, and it points the same way as the Florida result eleven days ago.

State it properly first. Parts of it are correct, and one part may change how you read the whole cycle.

Nobody has ever claimed a service record helps a Democrat beat another Democrat. Our claim is narrower: service moves independents and soft Republicans in a general election. So Massachusetts tests none of that. It tests whether a moderate Marine can beat a popular left-wing incumbent in one of the bluest primaries in the country, and the answer there was probably always no.

Markey also holds edges that owe nothing to either man’s life story. After fifty years in Congress almost every voter knows his name, and more than four in five approve of his work. On top of that, most of the state’s House members have endorsed him, and he took about 73 percent of the delegates at the state convention in May. Jesse Rhodes of the UMass team argued that the party’s signal alone likely settled it for voters on the fence.

Then on 18 August the Boston Globe editorial board endorsed Moulton, and its reasoning is the most useful thing anyone has written about this race. The board had backed Markey against Joe Kennedy in 2020, so the switch carries weight. It praised Moulton’s record directly, including the Bronze Star with valor he earned commanding infantry at Najaf in 2004. But the sentence that matters is about the electorate rather than the man. The board noted that Markey’s attacks on Moulton’s progressive credentials land hard with Democratic primary voters and land much softer with independents.

Read that beside the poll and the whole race resolves. Nobody disputes what Moulton did in Iraq. The dispute is over which voters weigh it, and on 1 September the answer is Democrats only.

And there is a real caution about the polls themselves. Alexander Theodoridis, who co-directs the survey, pointed to recent shocks in Michigan and Wisconsin. In short, primaries are hard to poll well. Screening for likely voters is guesswork, and people often decide late. Moulton makes the same point in blunter words.

We accept most of this. One thing survives it, though. Moulton has spent nine years arguing that service is the quality Democratic voters undervalue and ought to reward, and this year he made that case with his own life story, in his own state, with an alumni network campaigning beside him, against an opponent whose age he had spent ten months turning into the issue. The voters who know the argument best preferred the 80-year-old who never wore a uniform. That isn’t proof the general-election theory fails. It’s evidence that the primary-side version of it, which recruitment shops keep acting on, was oversold.

Three things are worth watching. Only one of them is the topline.

Turnout comes first. Moulton’s support skews young and unenrolled, and Massachusetts lets unenrolled voters take a Democratic ballot. Suppose turnout runs high on the North Shore and in the Merrimack Valley while Boston stays quiet. Then the polls were measuring the wrong set of voters. That’s what Moulton means when he says the surveys are wrong, and it isn’t a silly claim.

The trouble is that much of this electorate has already voted. Secretary of State William Galvin’s office sent out close to 700,000 mail ballots, and returns were approaching half of that before the final week began. Galvin also said he doesn’t expect the Senate race to pull people to early voting at all, and that local contests for district attorney and state Senate are what get someone to the polls in August. A late surge can only move ballots not yet cast. Here a large share went in before the last debate, before the Globe endorsement, and before either closing argument ran.

Second, watch the margin rather than the winner. A Markey win by 25 or more would suggest service simply didn’t register as a variable at all. A single-digit result would mean something moved late that no public poll caught, and it would set Moulton up for 2032, a race Markey has already said he will not enter.

Third, watch what happens to Serve America after Tuesday. Thirteen of its 14 backed candidates this year have nominations in hand. If its founder loses his own race by 20 points on the same night, the pitch to donors gets harder to make, and the shops that copied the model will notice. Jason Crow’s PAC already lost two of its Florida candidates on 18 August.