The national political environment is favorable for Democrats heading into November’s midterm elections, but is it favorable enough for them to retake the Senate?
In this week’s New York Times/Siena polls of six key Senate contests, Democrats are highly competitive, but for now they seem just out of range of flipping the chamber.
The polls find Democrats ahead in only two Republican-held states, less than the minimum of four they would need to win the chamber, with Roy Cooper leading by seven percentage points in North Carolina and Graham Platner leading by two in the previously reported poll of Maine. Republicans lead by two points in Alaska and Iowa and by three points in Ohio. The race in Texas is tied.
It’s important to emphasize that the results of most of these races — including the three races where Republicans lead in the poll — are within the margin of error. And at the beginning of this election cycle, few would have expected the Democrats to be this competitive in the race for the Senate. With the exception of Maine, President Trump won all of these states in 2024, and he won four of them by at least 10 points. No party has flipped two Senate seats in states that tilted toward the other party by double digits in the previous presidential election since Democrats did it in 2008, when Barack Obama won decisively.
Although Mr. Trump’s unpopularity has unequivocally put the Senate into play, the poll finds that the political environment in these reliably Republican states is still not quite as favorable to Democrats as they might need. And it may not be quite as favorable to Democrats as one might have expected, based on national polls.
When voters in Ohio, Iowa, Texas and Alaska are asked which party they support for Congress, they prefer Republicans over Democrats by at least five points, even though Mr. Trump’s approval rating is beneath 50 percent in all four states. They prefer Republican control of the Senate by a similar margin.
It’s not surprising that Republicans hold a clear lead on the generic ballot in such Republican-leaning states. Democrats lead on the generic ballot by about six points nationwide, which would not be enough to give Democrats the lead in states where Mr. Trump won by double digits.
But the Republican lead on the generic ballot in these states is perhaps a tick larger than one might have guessed based on national polls. And when it comes to the Senate races, that tick may make the difference between a true tossup or a discernible if narrow Republican edge.
One possible explanation is demographics: National polls show Democrats making outsize gains among young, nonwhite and especially Hispanic voters, who swung toward the right in the last election. But these demographic groups are generally underrepresented in these states, most of which are relatively white. If so, it’s possible that Democratic gains are inefficiently concentrated in other states. Only Texas — with its large Hispanic population — counts among the Senate battlegrounds where Democrats would be expected to post above-average gains.
For good measure — and again with the crucial exception of Texas — Republicans haven’t nominated deeply flawed candidates. More voters have a favorable than unfavorable impression of the Republican senators Dan Sullivan in Alaska and Jon Husted in Ohio, and the congresswoman Ashley Hinson in Iowa. They may not be powerhouses, but they’re candidates who would be expected to prevail in these states, even in this political environment.
Democrats are nonetheless highly competitive, as they’ve nominated the kind of top-quality candidates who could win when other Democrats might not. Former Representative Mary Peltola in Alaska, former Senator Sherrod Brown in Ohio and the state senator Josh Turek in Iowa aren’t just popular overall. They’re actually more popular than their reasonably popular Republican opponents, even in Republican states.
All three are running ahead of their party’s standing on the generic ballot. Only 25 percent to 36 percent of voters say these three Democrats are “too far to the left,” even though between 52 percent and 57 percent say the same of the Democratic Party more generally in these conservative states.
Nonetheless, all three narrowly trail in our new polls. For Democrats, this is probably the most disappointing in Ohio and Alaska, where Mr. Brown and Ms. Peltola have recently held statewide office and outran their party in 2024. But while some Republican-leaning voters may have a favorable view of each, it doesn’t necessarily follow that they would be inclined to vote for them over an acceptable incumbent senator of their own party.
Alaska quirks
There’s always good reason to be cautious about interpreting survey results, especially in tight races with four months to go until the election.
But there were a few idiosyncratic reasons to be cautious about the Alaska results. Of the six states, it’s the only one that hasn’t had its primary. The race is less developed there, and it showed in a few unusual respects.
The strangest issue is Alaska’s primary ballot, which features two Dan Sullivans — one the U.S. senator, the second a retired teacher and registered Republican who appears to be running to hurt the senator. If the second Sullivan advances to the general election in Alaska’s unusual top-four system, in which the top four primary vote-getters regardless of party advance to a general election with ranked-choice voting, Republicans fear he will confuse voters and draw support from the senator in November.
But in the poll, the other Mr. Sullivan may have had the opposite effect. The poll asked only about Ms. Peltola versus an unspecified Mr. Sullivan (without title or middle name), and a handful of Democrats may have either expressed support for him for their own amusement or confused the survey interviewer when asked about him.
One Democratic respondent, for instance, supported Democrats on all questions, including expressing a “very favorable” opinion of Ms. Peltola, except they supported Mr. Sullivan. When asked if they had a favorable view of Mr. Sullivan, they asked, “Which one?” When asked the open-ended question about why they supported Mr. Sullivan over Ms. Peltola, they said, “I don’t.” Of course, one or two respondents wouldn’t have affected the poll result by more than a point, but that could be half of Senator Sullivan’s lead in such a close poll.
A more conventional issue is undecided voters. At this early stage, a significant share of Ms. Peltola’s strongest constituencies — Alaska Natives and the state’s isolated Far North and West regions — were disproportionately likely to be undecided, with 17 percent of Alaska Natives and 20 percent of voters in the Far North and West declining to express a preference in the Senate race. As it was, Ms. Peltola, who is an Alaska Native herself, led by 51-32 among Alaska Natives and by 57-22 in the North/West region of the state.
It’s not a given that undecided voters, who represent around 7 percent of the poll overall, will flock to Ms. Peltola. Mr. Trump appeared to fare relatively well among Alaska Natives in 2024, based on precinct-level election results, and his approval ratings among both groups were surprisingly strong in the poll. It helped keep his overall approval rating in Alaska quite a bit healthier than one might have guessed based on the national polls or the results of other battlegrounds. It will be interesting to see if we find the same thing when we poll Alaska next time.
Weighting postscript (wonky)
As we wrote earlier this week, this wave of Times/Siena polls is the first using a new methodology. You can read more about it here.
How much did it make a difference? Here’s what the results would have been with the old method (the actual result is reported in parenthesis):
There’s a clear story here: The new methodology shifted Iowa and Ohio far to the right; there wasn’t much movement elsewhere. There was a similar if less pronounced pattern when back-testing this method on Times/Siena results from prior cycles: The new approach tended to move results toward the right in relatively white, working-class states where the polls have erred in recent cycles, like Iowa or Ohio, but usually did quite a bit less elsewhere.
What drove the change? The new methodology contains a lot of adjustments, from new weighting targets to a new weighting algorithm, and it’s a little hard to untangle how each individual change affected the outcome. But in this particular case, it’s clear that “synthetic past vote” — a measure derived from asking voters how they voted in the last election — is a major driver of the shift in both Iowa and Ohio.
Why did states like Alaska or Texas move by a point or two one way or the other? This might not be satisfying, but it’s hard to say and I wouldn’t read much into it. Even substantively irrelevant changes — whether to weight by metropolitan status versus density, or those 18 to 39 years old versus those 18 to 29 — can potentially yield a one-point shift. Given the magnitude of the underlying change in methodology, a difference of one or two points doesn’t amount to much.
It can be hard to remember, but almost any poll could have a result that’s a point or two different if the pollster had used a different and entirely legitimate weighting choice.
