With fewer than 24 hours until Election Day, I did a deep dive into the final sweep of available results for the U.S. Senate. The New York Times conceded in publishing its last battleground polls that registered white Democrats were 30% likelier to respond than registered white Republicans, indicating an undersampling bias that, if systemic, could be dramatically skewing the data to disfavor GOP odds.
While Democrats are overwhelmingly likely to obtain control of the House, polling of the upper chamber remains within the margin of error across much of the map. Thus, I tried to deduce the state of the race from the actual data on ballots already cast. While many states do not release party registration data for early and absentee voters, looking at data from states that do and models for states that do not can give us a clearer picture of the race. And after looking at the data, I’m much less convinced that Republicans face a serious threat in states such as Kansas and South Carolina, but I’m less confident in Republican victories in Ohio and Alaska.
Let’s start with the states that do publish party registration data. In Maine, where incumbent Sen. Susan Collins (R-ME) faces socialist Troy Jackson, Democrats technically have a 24-point margin. But that’s only because independents and third-party voters comprise 30% of the electorate. The current vote is on par with Maine’s performance at this point in the 2022 midterm elections and 8.6 points bluer than at this point in 2024, when then-Vice President Kamala Harris beat Donald Trump by 7 points. But Collins is not Trump. In 2020, she won non-college graduates by 22 points and men by 19 points. The current 2026 electorate is skewing a little more college-educated thus far than the final 2020 electorate, but it’s skewing significantly more male. Further, the absentee margin today is within 1 point of what it was in 2020, when Collins ultimately defied dozens of polls that projected a double-digit defeat and instead won reelection by nearly 9 points. The data seem to indicate that Collins could easily lead the Election Day vote by the 13-point margin she needs to overcome Jackson’s likely existing lead in the early vote.
Kansas is a similar story. In Kansas, where incumbent Sen. Roger Marshall (R-KS) faces a much-hyped challenge from Democrat Adam Hamilton, registered Republicans currently lead Democrats by over 2 points. The electorate is currently 13 points redder than it was at this point in 2022, when Sen. Jerry Moran (R-KS) comfortably won reelection to the Senate. The racial and gender breakdowns of the existing electorate are pretty similar to what they were at the end of the 2020 cycle, when Marshall won by 11 points.
In Iowa, where Rep. Ashley Hinson (R-IA) is fighting to replace outgoing Sen. Joni Ernst (R-IA), registered Democrats lead by 12 points. But again, this is because 20% of submitted ballots are from registered independents or third-party voters. Iowa’s existing electorate is 13 points bluer than at this point in 2024, when Trump ultimately won the state by 13 points, but it’s 6 points redder than it was at this point in the 2022 midterm elections, when Sen. Chuck Grassley (R-IA) won by 12 points. The demographics are similar enough to those when Ernst won reelection in 2020, and Republican turnout seems decent enough that Hinson doesn’t face an insurmountable margin on Election Day, when she’ll need around a 7-point margin to eke out a win.
We then have to rely on models and nonparty demographics to ascertain the results in the battlegrounds that do not release party registration data. In South Carolina, the On Point Politics model indicates that Republicans enjoy a double-digit margin, but the real hope for Republican Darline Graham is that the electorate is currently less diverse than it was even in 2024, when Trump won the Palmetto State easily. In Minnesota, where Republican Michele Tafoya is running one of the best campaigns of the whole cycle, the early-vote demographics indicate the reverse: The current electorate is less male and more educated than it was in 2024, meaning that Tafoya requires a double-digit margin of victory on Election Day to pick up the seat.
The data in Alaska, given the Democrats’ fake Dan Sullivan and the state’s ranked choice voting, are next to useless. On Point Politics thinks the current vote is nearly 15 points bluer than in 2024, but Alaska doesn’t release any demographic information at all. Whether Sen. Dan Sullivan (R-AK) beats “Decoy Dan” and actual Democrat Mary Peltola remains a totally open question.
The data continue to be a bit of a mixed bag in Texas and Ohio. In the Lone Star State, where Democrat James Talarico has tried to distance himself from Peak Woke in his race against Republican Ken Paxton, the On Point Politics model thinks the current turnout is 14 points bluer than it was at this point in 2024, when Trump ultimately won by 14 points. But the electorate currently looks a good deal older, and Paxton apparently still leads the early vote.
In Ohio, where incumbent Sen. Jon Husted (R-OH) is trying to keep his seat — previously occupied by Vice President JD Vance — from former Democratic Sen. Sherrod Brown, the current electorate is slightly less diverse and less educated than the final 2022 electorate that wound up favoring Vance. But On Point Politics estimates that the electorate is still some 15 points bluer than it was at this point in 2024, when Trump and Vance eventually won the state by 11 points.
In Michigan, Rep. Haley Stevens (D-MI) dropped the mother of all November surprises when she withdrew her endorsement of party nominee Abdul el Sayed and instead backed Rep. Mike Rogers (R-MI). Theoretically, in a blue wave year, Michigan shouldn’t be close at all, and the models all think that el Sayed is favored. However, young voters, who formed the backbone of el Sayed’s primary victory, make up a share of the current electorate that’s half as large as their share in 2024, while seniors are nearly twice as dominant. Could Election Day voters swayed by Stevens make up the gap for Rogers?
WHAT IF THE VIBES ARE ALL WRONG?
If I had to bet (which I do not do, because that would be unethical), I’d assume that Iowa, Maine, South Carolina, and Kansas all stay comfortably enough Republican. I think that Tafoya’s performance will be close enough that, if the GOP didn’t have to incinerate a quarter-billion dollars on Paxton, she could have pulled it off. But ultimately, no cigar. I have no clue about Alaska, and neither do you. For Ohio and Texas, I think the numbers lean red, but my gut leans blue. For Michigan, I feel the reverse: I just do not believe that the median Michigander is OK with el Sayed’s radicalism, and Stevens spoke for many a liberal Democrat in finally defecting from his campaign.
But in the end, as always, these races will all come down to Election Day turnout. Clearly, however, none of them will begin with an insurmountable margin, or at least not one quite as unfavorable as the polling would indicate.