The 2026 primaries remade hundreds of statehouse races
A topline review of state legislative results finds 291 unique current-roster legislators who did not advance from their party primary. Their districts' 2024 presidential margins and W-NOMINATE positions show where turnover came from.
- Candidate rows
- 14,006
- States with results
- 46
- Unique incumbents out
- 291
- Losses in close 2024 districts
- 16
Turnover was concentrated, and Vermont needs its own reading
Vermont accounts for 137 unique current-roster losses in this file. Its multi-member districts and Progressive/Democratic candidate-party rows make the count different from a one-seat, two-party contest. Outside Vermont, the largest clusters are South Dakota and Wyoming with 14 each, West Virginia with 13, and Idaho with 10.
View every state count
| State | Unique losses |
|---|
A primary loss does not always follow the district's presidential lean
Positive values on the horizontal axis mean the district voted more Republican in 2024; negative values mean more Democratic. The closest contests are the useful watch list: 16 unique losses happened within five points of the 2024 presidential margin, while 116 occurred where the presidential winner matched the incumbent's party.
View the closest districts
| District | Incumbent | Party | 2024 margin | Winner |
|---|
W-NOMINATE places most matched legislators inside their party's existing spread
For 283 of the 291 unique losses, the current W-NOMINATE file supplies a first-dimension score. The chart compares that score with distance from the same chamber's party median. A lower distance means closer to the party's center in that chamber; it does not label a legislator moderate or extreme across states.
View the farthest matched positions
| District | Incumbent | Party | Score | Distance from party median |
|---|
What stands out in the topline
Turnover is not the same as a general-election forecast. A primary can remove a sitting legislator from the November ballot while leaving the seat safely in the same party's hands.
- 116 losses occurred in districts where the 2024 presidential winner favored the incumbent's party. Party voters still made a different choice than the general-election margin would suggest.
- 16 losses came in districts within five points of the 2024 presidential vote. Those are the most natural places to follow through November.
- Vermont is a structural outlier in the count because the source reports several candidates and party lines inside many districts. The unique key removes repeated candidate-party rows where the same candidate string appears more than once, while the source rows remain candidate-party contests.
- Four states have no rows here yet: Delaware's regular primary is scheduled for September 15; Mississippi and New Jersey have no regular legislative primary in 2026; New Hampshire's official results page had not published the legislative result file when this artifact was built.
How this analysis was built
The underlying result is one row per candidate, party, chamber, district, seat, and election date. It contains 14,006 rows across 7,724 primary contests in 46 states with listed results. No county or precinct rows are added to this page.
Incumbent flags come from the active legislators file for each state at api.statenavigate.com. Names are matched within a chamber and district; unresolved names stay unknown in the data and are excluded from the loss count.
Presidential context comes from State Navigate district demographic downloads. A positive margin means the Republican ticket led in 2024, a negative margin means the Democratic ticket led, and the values are percentage points.
W-NOMINATE scores come from the latest published term for each state and chamber. The first dimension is useful for comparing legislators within that term. Distance from a party median is a within-chamber comparison, not a claim that every state's score has the same absolute meaning.
The analysis deduplicates a candidate who appears on more than one party row in the same state, chamber, and district for the unique-loss headline. Advancement remains tied to the source result status; records without a verified advancement result do not become losses merely because they have a vote total.
Download the generated analysis JSON for source URLs, file hashes, coverage notes, and the complete loss list.