PoliPoll Research AI · Texas U.S. Senate

A double-digit lead.
Recorded in early October.

A public record of two separate nonprobability online studies. These findings describe recruited samples—not an election forecast or an accuracy ranking.

Download research brief (PDF)3 pages · Approved public brief

Two studies. Both in the record.

First study · Unweighted

October 10–11, 2026

1,000 completed interviews

James Talarico
53.8%
Ken Paxton
37.4%
Ted Brown
8.8%

+16.4 percentage points

Talarico’s lead over Paxton in the recruited sample.

Unweighted online sample with party-affiliation and sex quotas. Participants self-reported being registered Texas voters and likely or certain to vote.

The three-candidate ballot required a choice; there was no undecided option.

Download full poll (PDF)Original 17-page report · Fieldwork October 10–11, 2026 · Released October 12, 2026

Separate follow-up · Weighted

October 13–18, 2026

A separate 1,000 completed interviews

James Talarico
43.6%
Ken Paxton
34.1%
Undecided
11.6%

+9.4 percentage points

Talarico’s lead over Paxton, calculated from unrounded estimates. Selected results shown above.

The primary analysis included all voting-likelihood categories and adjusted for age, sex, race/ethnicity, education and geography. It was not a likely-voter forecast.

An illustrative 2024 presidential-vote-recall stress test produced a +4.2-point lead. This is sensitivity analysis—not primary weighting, a correction or a forecast.

What these findings do—and do not—say

Both studies used nonprobability online samples. No conventional probability-sample margin of error is claimed. Quotas and weighting do not remove every possible source of selection or measurement error.

Different samples, ballot choices, weighting and treatment of undecided voters limit direct comparison. We do not claim those differences caused the change, or that one study proves the other right. Neither study establishes “most accurate.”

Read the public record

The approved three-page brief includes selected context with linked sources. Original reports and methodological detail are available on request.

Read the published LinkedIn article (opens in a new tab)

Media contact

John A. Mazzola

Founding Managing Director

PoliPoll Research AI, LLC

john.mazzola@polipollresearch.com