How we rate races

The scale

1. Safe DNot competitive. The Democrat would need an extraordinary event to lose.23
2. Likely DThe Democrat is a clear favorite, but the race could tighten.31
3. Lean DCompetitive, with the Democrat holding a real but beatable edge.13
4. Toss-upEither candidate could win. No meaningful advantage on the evidence available.3
5. Lean RCompetitive, with the Republican holding a real but beatable edge.7
6. Likely RThe Republican is a clear favorite, but the race could tighten.1
7. Safe RNot competitive. The Republican would need an extraordinary event to lose.15

“In play” means Toss-up, Lean D or Lean R: 23 races today. Races with only one party on the ballot are counted as Safe and labeled “No D filed” or “No R filed”. The number beside each step is how many of the 93 races carry that rating now.

What goes into a rating

The last general election

How the same district voted the last time the office was on the ballot, as a two-party share from the Harris County canvass. Where no same-office result exists (most judicial benches), we use the countywide judicial average from 2024 and label it a baseline.

The district's partisan lean

Presidential and other top-of-ticket results inside the district, precinct by precinct, which show how the seat behaves beyond one candidate.

Incumbency and open seats

Whether an incumbent is on the ballot, how they won last time, and whether the seat is open.

Money

Cash on hand and fundraising from the latest FEC, Texas Ethics Commission and county filings. Money moves a rating only when the gap is large and the race is otherwise close.

Reported polling

Independent public polls with a stated sample, field dates and population. We cite each poll in the race notes; we do not average or adjust them.

Events

Primary upsets, withdrawals, appointments, indictments and redistricting, each tied to a dated news report.

No single input decides a rating, and there is no formula that converts them into one. That is deliberate: a formula would treat a two-point 2024 margin the same whether or not the incumbent has since been indicted.

When a rating changes

Ratings move only on hard evidence: a candidate leaving the race, a credible public poll of the district or state, an indictment, or a shift in who is on the ballot. The desk reviews every race against the week’s reporting. Each change is dated, sourced and shown on the race page and the front page, with the old rating struck through.

• Aug 10Texas Governor: Safe R to Lean R. Three late-July polls all had Abbott within 3 points of Hinojosa, his tightest statewide position yet.
• Aug 7Texas Attorney General: Safe R to Lean R. A July poll found Middleton and Johnson nearly tied, 39 to 38, with 18 percent undecided.

Corrections to our own data are logged the same way and labeled as corrections. Found an error? Tell the desk.

Sources

• Election results: Harris County Clerk canvass returns, precinct level, 2012 through the 2026 primaries and runoffs.
• Campaign finance: Federal Election Commission, Texas Ethics Commission, Harris County clerk and City of Houston filings. Each figure shows its filing date.
• Districts and demographics: U.S. Census Bureau (TIGER boundaries, American Community Survey citizen voting-age population).
• Candidate photos: official government portraits or the lead image of each candidate’s Wikipedia article. Where neither exists, we show initials.

The Accountability Score

Each official’s profile carries one number, 0 to 100, built only from public records. Score = 30% fundraising strength + 30% legislative output + 20% peer standing + 20% experience. When a component has no data, it is dropped and the remaining weights renormalized: no official is penalized for a gap in our records.

It is not an ideology score, an endorsement or a prediction. Two officials with opposite politics can both score 85.