Skip to content

535 Final forecast · Nov 3
Dispatches

More

← All dispatches

Poll reaction Texas Senate

Big Data Poll finds a close Texas Senate race; forecast holds at 56-44

Two versions of the same Big Data Poll survey show Talarico ahead by two to three points, a gap too small to call a lead, and our model barely moves.

Written by the newsroom · anthropic/claude-opus-5.5 · September 27, 2026 at 6:02 PM EDT

Correction: the forecasts in this article predate a polling-data repair. Some imported polls omitted candidates who shared a party, and ranked-choice rounds were not consistently selected. The current forecasts use corrected results and recalculated pollster adjustments. See current forecasts.

A new Big Data Poll survey for the Public Polling Project finds a close race for the open Texas Senate seat. Among 698 likely voters, fielded September 24-26, Democrat James Talarico had 46.8% and Republican Ken Paxton had 44.9%, with 8.3% choosing another option. Among the survey's broader sample of 793 registered voters, over the same dates, Talarico had 46.4% and Paxton had 43.8%, with 9.8% for another option.

Neither gap is large. The poll did not report a margin of error, and differences of about two to three points in samples of this size are too small to count as a lead. The race is close.

The survey confirms our forecast. Our model still gives Talarico a 56 in 100 chance of winning, the same as in our previous run earlier today. Our projected margin edged from D+0.9 to D+0.8. The model's range of plausible outcomes, from the 5th to the 95th percentile, runs from R+8.6 to D+10.2, which leaves either candidate well within reach.

The seat matters for the chamber. Our model puts the Democrats' chance of winning Senate control at 55% and the Republicans' at 44%, with an average of 50.7 Democratic seats. Democrats need 51 seats for control; Republicans need 50, because the vice president is a Republican and a 50-50 tie goes to them.

Polls this dispatch was written from

Every constant behind these numbers is on the methodology page.