Democrats have convinced themselves that 2026 is finally the year Texas turns blue. Polling shows James Talarico (D-Texas) with a slight edge over Ken Paxton (R-Texas) in the race for Texas' open Senate seat, and that's apparently all it takes for the party to sink another cycle's worth of money, time, and energy into a state that hasn't elected a Democrat statewide since 1994 and voted for Trump in 2024 by a 14-point margin.
I've said before that I'd love nothing more than to watch Democrats burn through their donor money chasing Talarico in Texas, because he doesn't have a chance. Yet, like clockwork, Democrats are convinced, "This is the year.”
That confidence runs straight into a body of research from pollster Patrick Ruffini, and it isn't kind to Democrats. Ruffini's analysis shows that Senate polling in red states has consistently overstated Democratic support during the summer months over the last four election cycles, and that the same pattern is showing up again in 2026.
Right now, Democrats look competitive practically everywhere Republicans should be safe. A Fox News poll has Sherrod Brown (D-Ohio) leading his Ohio Senate race by eight points. Other surveys show Talarico ahead in Texas and Mary Peltola (D-Alaska) leading in Alaska. On paper, it looks like a red-state revolt is brewing.
It's happened before. In 2020, a Quinnipiac poll conducted July 30 through Aug. 3 showed a tied South Carolina Senate race between Lindsey Graham (R-S.C.) and Jaime Harrison, kicking off a summer of neck-and-neck surveys that had Democrats dreaming of an upset. Graham won by ten points.
Ohio tells the same story twice. In 2022, Tim Ryan (D-Ohio) led in most summer polling before losing to JD Vance by six points. In 2024, Brown led in all but two pre-October polls before losing to Bernie Moreno. Now he's supposedly up eight points again, and Democrats are acting like this time is different.
Recommended: The Democrats' 'Big Tent' With Socialists Is Doomed. Here's Why.
Ruffini didn't rely on hunches to make his case. He pulled 3,312 polls from the last four election cycles from the old FiveThirtyEight archive and found that the Democrat candidate's margin was overstated in every single cycle. The average polling error came in at D+3.7, with 73% of the polls overstating the Democrat relative to the eventual result.
That error peaks at 5.9 points too Democratic in late July and doesn't start shrinking until late August. By Election Day, the average error settles at D+2.6, still tilted toward Democrats but nowhere near what today's polls are showing.
On top of that, Ruffini found a sharp split by cycle type. In the Trump era, Senate polls in presidential election years have remained biased toward Democrats through Election Day. Polls in midterm years, like this one, were heavily biased in the summer but wound up accurate by the time voters actually showed up. Either way, the numbers Democrats are pointing to right now aren't where things end up.
I suspect this phenomenon isn’t limited to red states either. Six years ago, Sen. Susan Collins (R-Maine) won reelection by 8.6 points even though not a single poll ever showed her ahead. Her opponent, Sara Gideon, led in most surveys by four or five points, and at times by as much as twelve. Collins won by nearly nine. That was the same year that Joe Biden carried Maine by nine points.
Anything can happen between now and November. But Democrats have no rational reason to feel confident about Texas, Ohio, Alaska, or even Maine, frankly. Every cycle brings the same summer mirage, and every cycle it evaporates by fall. Why would anyone think 2026 would be the exception?
