The Cornyn Voter Threatening To Cross The Aisle In May Won’t Cross It In November

Through the spring, public surveys consistently warned that between 24% and 30% of John Cornyn’s voters were telling pollsters they would cross party lines and vote for Democrat James Talarico (cis) in November if Ken Paxton became the Republican nominee. Last week, Paxton did in fact become the nominee, beating Cornyn by more than 20 points. Post-runoff polling now shows that as many as 30% of Cornyn voters claim they will vote for Talarico instead of Paxton. Texas Public Opinion Research went into the field on the night of May 27, fewer than 24 hours after the polls closed, with the first general election survey of the cycle, and reported the upper bound of the spring warning had landed exactly on schedule. The piece of cable commentary that followed wrote itself. Paxton was too damaged to win. The runoff had hollowed out the GOP coalition. Texas was finally in play. Each of these conclusions is wrong, and each is wrong for the same reason.

Disaffected Cornyn voters telling a pollster on Wednesday night that they would rather vote for a 35-year-old Austin Democrat than the man who beat their candidate by 20 points are not making a voting decision. They are venting a primary wound. The professional question for any responsible analyst is not what those voters say tonight. It is what they will say in October.

The man who has answered that question more authoritatively than anyone alive is John Ashcroft. The former Missouri senator and U.S. attorney general has watched Republican Senate primaries from inside campaigns since the Ford-Reagan era. His rule is that it takes 30 to 60 days for the wounds of a bitter primary to close, and that intra-party polling is meaningless until that window has passed. The runoff was held on May 26. Apply Ashcroft’s lower bound and the first meaningful survey of Cornyn voters can be taken on June 25. Apply his upper bound and we are in late July. The Texas Public Opinion Research poll was in the field roughly 18 hours after the runoff was called. By Ashcroft’s measure, the 30% defector figure is the bruise three minutes after the punch, not the bruise three weeks later when it has faded to nothing.

It helps, at this point, to look at who the 30% actually are. The crosstabs reveal a coalition that will not survive a campaign. Of the Cornyn voters telling surveyors they would vote for Talarico, 51% give as their reason Paxton’s past indictment or his 2023 impeachment by the Texas House. These are not Talarico voters. They are Cornyn voters who have not yet performed the cognitive reconciliation between primary anger and partisan reality. Only 10% of Talarico-leaning Cornyn voters offer any positive reason for supporting him at all; another 10% describe themselves as “reluctant” supporters who see no alternative. A further 23% of Cornyn voters are parked in “undecided” or “will not vote” categories. The combined pool of soft-Talarico, undecided, and sit-out Cornyn voters is on the order of 35 to 40% of the Cornyn coalition. Standard partisan-revert behavior, documented since Atkeson 1993, predicts that roughly two-thirds of those voters return to the Republican column within 60 to 90 days. That alone closes the current gap and then some.

This is not a partisan claim. Lonna Rae Atkeson’s seminal paper “Moving Toward Unity” in American Politics Quarterly found that supporters of losing primary candidates revert to the eventual nominee at predictable rates as the general election approaches. Jeffrey Lazarus, writing in Legislative Studies Quarterly, established that the apparent correlation between divisive primaries and weaker general-election performance is not causal at all, but a joint product of candidate quality and pre-primary expectations. Fouirnaies and Hall at Stanford in confirmed that in base-state seats with partisan leans above 7 points, the measurable divisive-primary penalty is statistically indistinguishable from zero. Texas leans R+10 to R+13. It is the cleanest possible case for the proposition that the primary will leave no footprint on the general election.

But the historical record is where this argument lives or dies. The frame is simple. The runner-up’s coalition comes home. It has always come home. The mechanism is what political scientists call partisan reversion, and it has been visible in every contested base-state primary of the modern era.

Consider Reagan against Ford in 1976. The Republican primary that year went all the way to the convention floor in Kansas City. Ford defeated Reagan by 117 delegates on the first ballot. Reagan declined Ford’s offer to chair the California general election effort and did almost no surrogate work for the ticket. Reagan supporters polled in mid-August reported in large numbers that they would stay home rather than support Ford. Polls had Jimmy Carter ahead by 33 points. By Election Day, that 33-point Carter lead had collapsed to a 2.1-point Carter win. The Reagan coalition, which had given every appearance in August of permanent estrangement, voted for Ford in November in numbers that produced one of the closest popular-vote margins of the modern era. The Paxton-Cornyn runoff was not as bitter as Reagan-Ford, and Cornyn has already conceded the race. The Texas case in 2026 begins from a substantially better position than Ford did in 1976.

Consider Obama against Clinton in 2008, which is the cleanest data we have on what happens to a 25 to 30% intra-party defector pool over a six-month timeline. The Democratic primary did not end until June 7. Polling through May and June consistently showed roughly 25% of Clinton supporters telling surveyors they would vote for John McCain or stay home. The “PUMA” movement, “Party Unity My Ass,” made national news. Pundits warned the Democratic coalition was structurally broken. Obama still won the general election by 7.2 points, with Clinton voters consolidating behind him in the high 80s by Election Day. The published 25% Clinton defector pool in May simply disappeared between Memorial Day and November. The 30% Cornyn defector pool reported this week is the same phenomenon at an earlier stage of the same curve.

Consider Trump against Cruz in 2016, the direct Texas analogue. The two men fought a primary that ended in personal feud, with Cruz refusing to endorse Trump at the Republican National Convention on July 20 and being booed off the stage. Cruz did not endorse Trump until September 23. Polling through July and August suggested Cruz holdouts among Texas Republicans could amount to a quarter of the GOP coalition. Trump carried Texas by 9 points and won the presidency. The Cruz coalition came home. It came home even though the runner-up actively withheld his endorsement for nine weeks. The Cornyn-Paxton race has already produced a faster reconciliation, with Cornyn conceding within hours and Speaker Dustin Burrows, the NRSC, and the Texas Republican Party publicly aligning behind Paxton.

Consider Cassidy against Maness in Louisiana in 2014. Conservative challenger Rob Maness took 14% in the November 4 jungle primary, accusing the eventual nominee of insufficient conservatism. The runoff against Democratic incumbent Mary Landrieu was held December 6. Total primary-to-runoff healing time, 30 days. Cassidy consolidated essentially all of the Maness voters and defeated Landrieu by 12 points. That is the lower bound of Ashcroft’s window, and it produced complete consolidation in a state with a partisan lean materially less pronounced than Texas’s. Paxton has 23 weeks, not 30 days.

There is, second, the Texas polling record. In 2018, public polls had Cruz vs. O’Rourke leads ranging from Cruz +9 to O’Rourke +2 through October. Cruz won by 2.6. In 2020, the UT-Texas Tribune poll on October 23 had Cornyn up just 8 over MJ Hegar. Cornyn won by 9.8. In 2024, the Houston Public Media UH Hobby School survey in mid-October showed Cruz ahead by only 4 over Colin Allred. Cruz won by 8.5. Most striking, the May 2026 Cornyn-Paxton runoff itself was missed by an average of 8 points by the same firms now reporting Talarico ahead by 3. The actual Paxton margin was north of 20. The Texas polling apparatus understated the Republican vote by 12 points in the runoff. Three days later, the same apparatus told us 30% of Cornyn voters had defected. One feels, at moments like this, the way Charlie Brown must have felt about the football.

The prediction market, which aggregates real capital at risk rather than the opinions of a 1,670-person sample called on a Wednesday night, currently prices Paxton at 61% to win the Texas Senate seat against Talarico at 40%. The market is reading the same crosstabs the rest of us are reading. It knows the 30% Cornyn defector pool reported this week has, in every comparable case, been reduced to single digits by Labor Day and to a partisan-baseline figure of 4 to 6% by Election Day. The market is pricing the structural baseline. The pollsters are pricing a coalition that will not exist in 60 days.

Texas has not elected a Democrat to the US Senate since Lloyd Bentsen in 1988. That is 38 years. Donald Trump carried Texas by 13.7 points in 2024. Greg Abbott won the 2022 gubernatorial race by 11. Ted Cruz won the 2024 Senate race by 8.5. The state’s structural Republican baseline is R+10 to R+13. For the 30% Cornyn defector pool to actually deliver a Talarico win, it would have to hold all the way through November while the rest of the Texas GOP coalition behaved as if nothing had changed. That is not a scenario the political-science literature can locate in any modern American election.

The honest assessment is that the Cornyn voter telling a pollster this week that he would vote for Talarico is not making a forecast about November. He is registering a grievance about May. By July, the grievance will have softened. By August, the Republican messaging machine will have introduced him to James Talarico’s actual record. By September, he will have spent four months watching Speaker Burrows, the NRSC, and the Trump White House communicate that Paxton is the standard-bearer. By October, partisan gravity will have done what partisan gravity always does. By November, the 30% defector pool will have shrunk to the partisan-baseline 4 to 6% that every modern Texas Senate race has ended with, and Paxton will win by between 5 and 10 points.

The drive-by media will spend the summer reporting Texas as a competitive race. The same drive-by media reported Beto, Hegar, and Allred as competitive. They were not. Talarico is not. The 30% number measured a wound. The race in November will measure a state.

You don’t poll a bruise. You wait for it to fade.

If you enjoy my work, please subscribe https://twitter.com/amuse/creator-subscriptions/subscribe.

Sponsored by the John Milton Freedom Foundation, a nonprofit dedicated to helping independent journalists overcome formidable challenges in today’s media landscape and bring crucial stories to you.

READ NEXT: Democrats Erupt Over Effort To Honor Trump On Currency