The University of Houston’s Hobby School of Public Affairs just dropped its February 2026 statewide survey covering the March primaries and November general elections. The political press will spend the next week telling you who’s ahead. That’s the least interesting thing in this data.
The real story is what every political reporter in Texas will ignore: the Texas electorate — Republican, Democrat, and everyone in between — is checked out of the current arrangement. Not apathetic. Not disengaged. Done settling.
The Governor Can’t Get to 50
Greg Abbott has been governor for over a decade. He has functionally universal name recognition — only 2% of likely November voters don’t know enough about him to have an opinion. He’s running against a state representative that 45% of likely voters have never heard of.
And he’s at 49%.
That’s not a commanding lead. That’s an incumbent failing to close the deal against an opponent most Texans couldn’t pick out of a lineup. His net favorability among likely general election voters is a razor-thin +4 (51% favorable, 47% unfavorable). For a three-term Republican governor in the reddest large state in the union, those are the numbers of a politician who is tolerated, not supported.
Dan Patrick is worse. The lieutenant governor is actually underwater — 41% favorable to 43% unfavorable — with 16% still not knowing enough to judge him. He leads his Democratic opponent by only 5 points, 46% to 41%. His opponent, Vikki Goodwin, is a state representative that 58% of likely voters have never heard of. Patrick has been one of the most visible politicians in Texas for a decade, and he’s losing to a name no one recognizes among women (39% to 45%) and college-educated voters (38% to 49%).
The Hobby School itself flagged the comparison worth noting. In six hypothetical U.S. Senate matchups from their first report, the median Republican advantage was just 2 points. Abbott’s 7-point lead and Patrick’s 5-point lead are running above the generic Republican baseline — meaning their incumbency and name ID are still worth something. Just not much.
The Primary Electorate Is Shopping
Here is where it gets interesting.
In the Republican attorney general primary: 25% undecided. In the comptroller primary: 29% undecided. Agriculture commissioner: 34% undecided. Railroad commissioner: 41% undecided. On the Democratic side, the gubernatorial primary has 32% undecided. The attorney general primary: 40%.
Those are not early-race numbers. This survey was conducted between January 20 and January 31 — five weeks before election day. These voters had months of campaign exposure, mailers, digital ads, and endorsement announcements. A quarter to two-fifths of primary voters looked at the full menu and couldn’t find anything worth ordering.
The standard political consultant explanation is “low information, low salience races.” That’s the explanation of people whose livelihood depends on the current system working. The simpler explanation is that the candidates on offer aren’t giving voters a reason to commit. Across both parties, across every statewide race, the pattern is identical: massive blocks of voters who are paying enough attention to show up in a likely voter screen but not motivated enough by any candidate to pick one.
That’s not an information gap. That’s an inspiration gap.
The Gender Gap Nobody’s Talking About
Buried in the cross-tabs is a pattern that repeats like a drumbeat across every single Republican primary: women are dramatically more likely than men to be undecided.
In the attorney general race, 33% of women are undecided compared to 18% of men. In the comptroller race, it’s 39% to 21%. Agriculture commissioner: 41% to 27%. Railroad commissioner: 48% to 37%.
Women make up 45% of the likely Republican primary electorate. Nearly half of them, in race after race, cannot find a candidate worth supporting. That’s not a “women are less engaged in down-ballot races” problem. That’s a candidate quality problem. The Republican primary field, top to bottom, is failing to connect with a massive share of its own voters.
No one running for any of these offices appears to have noticed.
The Myth of the Ethnic Monolith
One of the most persistent fictions in Texas politics is that the state’s future is a demographic inevitability — that as the Latino population grows, Texas will turn blue. The data in this poll says otherwise.
In the November general election matchup, Abbott holds Latino likely voters 49% to 43% over Hinojosa. Patrick holds them 45% to 42% over Goodwin. Within the Republican primary electorate itself, Latinos make up 24% of likely voters.
This is not a community that belongs to either party. These are Texans making individual calculations about which candidates and which vision of the future serves their interests. The old binary — Anglo Republicans vs. minority Democrats — is a relic. It persists in the minds of political consultants and editorial boards because it’s simple and it sells. The actual electorate has moved on.
For those of us who believe in Texas self-governance, this matters. The case for an independent Texas has never been about one party, one ethnicity, or one ideology. It’s about the recognition that Texans — all of them — are capable of governing themselves. The polling data confirms what we’ve always argued: Texan identity crosses every demographic line the political class uses to divide us.
A Note on the Numbers Themselves
Before anyone builds a campaign strategy on these figures, a word of caution.
This is a YouGov online panel survey — not a probability sample. The respondents are people who signed up to take surveys for compensation and were then statistically matched to a sampling frame based on four demographic variables: gender, age, race/ethnicity, and education. That matching makes the sample look like the target population on those dimensions. It does not address the fundamental selection bias in who volunteers to sit on an online survey panel. These are people who are more politically engaged, more online, and more opinionated than the actual voter population.
The likely voter screen — the single most important methodological decision in any pre-election poll — is described in a single vague sentence: a combination of self-reported intention and past voting history. No thresholds. No weighting details. No transparency. In a Texas March primary where actual turnout runs around 10-15% of registered voters, the likely voter model is everything. They’ve told you nothing about how it works.
The sample sizes for primary analysis — 550 per party — produce a headline margin of error of ±4.18%. But when you start slicing into cross-tabs — Latino Republican primary voters, Millennial Democrats, Independents voting in the GOP primary — the cell sizes drop into the double digits. The real margin of error on those sub-groups is likely ±8-12%, which makes most of the cross-tab comparisons the researchers highlight statistically indistinguishable from noise. They never report the subgroup margins of error.
There is no disclosed geographic weighting. Texas is enormous and politically heterogeneous. The difference between the Republican primary electorate in Tarrant County and in Jefferson County is not trivial. If the YouGov panel over-represents one region and under-represents another, every number is skewed — and you’d never know.
All of that said, the directional findings are almost certainly real. The specific horse-race numbers should be treated as approximate. The gap between Abbott and Hinojosa is probably real. Whether it’s 7 points or 4 points or 10 points, nobody can tell you from this data.
What the Political Class Won’t Say
Every election cycle, Texans are presented with the same proposition: this is the election that matters, these are the candidates who will fix it, and the only question is which team you’re on.
And every cycle, the data tells the same story. An incumbent governor who can’t break 50%. A lieutenant governor whose own state is split on whether they even like him. Primary fields full of candidates that a third to two-fifths of their own party’s voters can’t be bothered to choose between. An electorate that crosses every demographic line the consultants use to sort us into manageable categories.
The voters of Texas are not confused. They are not uninformed. They are not waiting for the right candidate to come along and make the current system work. They are telling us, in the clearest language available to them, that the current arrangement is not delivering.
The question of Texas self-governance is not competing with these candidates for voter attention. It is the answer to a question none of them will ask — and most of them are afraid to hear.
The numbers don’t lie. The only question is whether anyone is willing to read them honestly.
The Hobby School of Public Affairs survey cited in this analysis was conducted January 20–31, 2026 among 1,502 likely November general election voters (±2.53% MOE), 550 likely Republican primary voters (±4.18% MOE), and 550 likely Democratic primary voters (±4.18% MOE). The full report is available from the University of Houston.
