Jason Corley

In the end, Quantus Insights delivered. @QuantusInsights

Tomorrow morning, @QuantusInsights will release our final Texas poll of the 2026 midterm election cycle, featuring the U.S. Senate and gubernatorial races. Just days before Election Day, all eyes are on Texas. 🇺🇸

Interesting. We're not the lone ranger over here.

The combination of a large, habitual Republican turnout pool and an unusually young, non-white and relatively infrequent swing electorate is precisely what makes this election difficult to project from historical turnout alone. Lucky for us, we can also poll it.

Glad to see someone else considering this. The theory goes like this. There’s persuasive evidence that Texas Democrats heavily front-loaded their vote, while the second week has seen habitual Republican voters entering the electorate. The Democratic EV advantage may also be overstated by the 2026 primary expanding the pool of voters classified as Democrats. Because their previous absence from Democratic primaries doesn’t mean they weren’t already part of the Democratic general election coalition. The real question is whether we’re seeing existing Democrats newly identified, differential turnout, crossover voting, or genuine partisan realignment. Because those are very different things. Maybe more on this tomorrow.

The substantial Democratic early voting advantage has the look of heavy front loading. The following week suggests Republicans are beginning to draw down their considerable remaining pool of voters. The question now is whether that acceleration continues through Friday and what in the world Election Day will look like.

Reviewing Texas registrants, with each voter’s 2026 ballot status through Oct. 27, turnout history, and modeled party. The early electorate is R+2 on modeled party. The full 2022 electorate was R+12; 2024 was R+9. White voters are 55% of early voters, down from 62% in 2022. Democrats have banked more of their expected vote. Lean Democrats have reached 83% of their 2022 turnout, Strong Democrats 73%. Strong Republicans are at 62%, Lean Republicans 57%. Under 30 voters have already matched their entire 2022 turnout. But the pool still available is R+17 to R+20 on modeled party and 52 to 62% White depending on who turns out. Republicans have a substantial reserve, and the later days of early voting have run more Republican than the first. Project the remainder using either the 2022 or 2024 turnout pattern and the final electorate lands at R+8 to R+10. That holds whether total turnout is 8.7 million or 10.6 million. The shape of this electorate is largely set. Scored on its 2024 presidential vote, that electorate backed Trump by 12 to 14 points. Composition alone moves the margin a point or two toward Democrats, not ten. So how does that square with public polls showing Talarico ahead by about 2 and Abbott by 3 or 4? Only through switching. If 2024 Trump voters hold at 89% for Paxton, the going consensus, this electorate produces Paxton +3. A Talarico lead needs 13 to 15 points of net movement among 2024 Trump voters, or a final electorate several points bluer than its 2024 vote suggests. Abbott leading in the same polls tells you the movement is specific to Paxton: same voters, different candidate, five points apart. Early vote trackers tell us who is showing up. They cannot tell us how those voters are choosing. That is what we have been measuring this week, among early voters and among the voters still to come.

Took a Texas Senate poll about an hour ago. Finally emerged from my office after working on our own Texas poll, only to get polled on the same race myself. Too much irony, so I participated.

Thank you for taking our Texas poll and participating in the process! We’re still actively sending and collecting.

We’ll see what we find for our last look at Texas. More to come.

What if some of these races really are this close? What if pollsters are independently refining their likely-voter models and arriving at similar conclusions? We saw similarly competitive battleground polling in 2024, and polling generally performed well, despite accusations of herding then too. Unlike 2020, there were no Wisconsin D+16 polls. Did that pollster earn higher ratings for departing from the consensus? Respectfully, I find it more problematic to characterize a poll that missed by double digits very late in the election as “brave.” An egregious outlier isn’t more useful than a poll near the consensus, particularly if that consensus proves correct. And agreement alone isn’t evidence of herding. Pollsters should be focused on accurately measuring the likely electorate, not manufacturing disagreement to demonstrate independence.

It’s Jackson +2 according to Marist’s latest release. -Jackson 49% (+2) -Collins 47% Their last poll, fielded during the first few days of October, showed Jackson +4.

This is good stuff. The thing worth considering now is electorate composition and vote choice. 2018 sort lays that interaction out for us. Democratic turnout gains shifted the partisan composition of the electorate by 2 to 3 points in several key states. If 2026 produces a similar shift, alongside the stronger Democratic preferences we’re seeing in some polling and early voting, the effects could compound. Obviously that matters considerably in important Senate races OH, TX, NC, AK, IA, etc. Right now these races are extremely competitive and already showing a sizable swing from 2024.

We found unaffiliated voters in this sample recall voting Trump +5 in 2024 and back Cooper 55 to 36 today. Two years apart, a 24 point swing, among largely the same cohort that voted in 2022 and 2024. And the polling consensus agrees unaffiliateds are breaking for Cooper somewhere between 15 and 20+ points. Cooper is polarizing, 39% very favorable against 40% very unfavorable, but his net favorability is +9. He’s a known quantity who has already won this state twice on ballots Trump carried. So it's not a case of unaffiliateds discovering him, they’re plainly backing him. Republicans still have roughly 920k of the state’s most reliable voters unspent. If they turn out hard on Election Day, the electorate moves several points toward the GOP, Whatley outperforms the polls, and everyone points and laughs. If they don’t, a Democrat wins a North Carolina Senate race for the first time since 2008, by about the same margin Kay Hagan got there with. 8.5 pts.

Remaining Texas Republican primary history advantage: 🔴 570k Have yet to cast a ballot. That’s considerable when the 2026 early electorate has shifted 6-7 points Democratic from 2024.

That’s an awful lot of certainty about how people are going to behave in an election that hasn’t happened yet. Uncertainty deserves a place in the equation, especially when the premise depends on voters behaving differently than they have in the past. You can make a case for it, but you can hardly demand that it happen. He may well get the outcome he expects, but I suspect it will owe less to Republicans suddenly behaving unlike Republicans than to differential turnout peculiar to this election.

When we released our @QuantusInsights Texas Senate poll on October 13, we found Talarico and Paxton tied at 47% (Paxton: 47.2% to Talarico 46.7%) compared to the preceding seven polls that averaged Talarico +3. Six polls since have averaged Talarico +1.5, while the three most recent average a dead heat. Polling steadily converged toward our original finding.

According to Fox News poll: “Views of Talarico are net positive, albeit by just 1 point (49% favorable, 48% unfavorable), while Paxton’s ratings are underwater by 17 (41% favorable, 58% unfavorable). Among "double haters," voters who view both candidates negatively, Paxton leads by 50 points.” A 50-point advantage among 5% of voters contributes approximately 2.5 points to Paxton's statewide margin. There may be a ceiling on how much Talarico can benefit from Paxton's personal liabilities. The voters who dislike Paxton but remain unwilling to vote for Talarico because of his own liabilities, are precisely the voters he needs to convert.

That’s all you can do and all you should do. We’ve shown this Hispanic vote between +11 to as high as +30 if I recall. All the noise comes from those who have very little idea what they’re even criticizing. Your team had a great 2024 for a reason.

Put simply, smoothing methods suggest that Talarico's earlier 3–4-point advantage has disappeared in the most recent polling. LOESS models place the latest margin tied, and perhaps Paxton +0.5. -Early advantage: Siena and Trafalgar showed Talarico ahead by 3–4 points. -Mid-October variation: Cook showed Talarico +5, but Quantus found a tie and InsiderAdvantage found Paxton +1. -Latest reading: Fox News found a 49–49 tie, and pulled the fitted trend toward an even race.