Times/Siena poll puts Talarico up 7 in Texas, and our estimate rises to 78%
After several new polls left our Texas forecast largely unchanged, the final Times/Siena survey moved it 5 points toward the Democrat two days before the election.
Written by the newsroom · anthropic/claude-opus-5.5 · November 1, 2026 at 9:02 PM EST
A New York Times/Siena University survey of 1,215 likely voters, fielded October 27 to November 1, has Democrat James Talarico at 50.2% and Republican Ken Paxton at 43.1% in the race for the open Texas Senate seat. That is a 7.1-point gap. The registered-voter version of the same poll, also reported with a sample of 1,215, shows a wider spread: Talarico 51.5%, Paxton 39.3%.
The poll moved our forecast. Our model now gives Talarico a 78 in 100 chance of winning, up from 73% in the run before these numbers arrived. Our projected margin rose from D+3.0 to D+3.6. Paxton's chance fell from 27% to 22%.
This is a change from what the site has reported in recent days, when new polls largely confirmed the existing estimate and left it near 73%. The Times/Siena likely-voter margin is about twice our model's projected margin, which is why it pulled the estimate up rather than leaving it in place.
The race remains within reach for both candidates in our model's estimate. Its 5th-to-95th percentile range now runs from R+4.2 to D+11.5, compared with R+4.9 to D+10.8 before. The outcomes in which Paxton wins are fewer than they were, and they are still a meaningful share of the total.
Texas is one input into our national picture. Our model gives Democrats a 78% chance of winning Senate control, which requires 51 seats; Republicans need 50, since a 50-50 tie goes to them through the vice president. Our model's average Democratic outcome is 52.4 seats. The election is November 3.
Polls this dispatch was written from
- The New York Times/Siena University · Nov 1, 2026 · n=1,215 · Texas Senate
- The New York Times/Siena University · Nov 1, 2026 · n=1,215 · Texas Senate
Every constant behind these numbers is on the methodology page.