Observed sources and real demographic data seed a modeled response. The output is synthetic—not observed polling, a forecast, or a statement of what the public believes.
Read the method →Published polls and forecaster ratings first, each dated and linked to its source. Then Pheme's synthetic public answering the same questions, set beside the polls so the gap shows. Pheme does not forecast seats or call races, and its modeled answers are a simulation, not a poll.
Election day is Tuesday, Nov. 3, 2026.
Observed snapshot as of Oct. 8, 2026
This snapshot is older than 10 days. Polls and ratings move — open the linked sources for current numbers before relying on any figure below.
53 R · 47 D
35 seats are on the ballot. Democrats (the 47 include two independents who caucus with them) need a net gain of four seats for a majority, because the vice president breaks a 50–50 tie.
Source: 270toWin / Inside Elections ↗ · Oct. 1, 2026
All 435 seats
218 seats make a majority. Republicans hold a narrow majority going in; the exact count moves with vacancies and special elections, so check the Press Gallery tally rather than a number copied here.
Source: House Press Gallery party breakdown ↗ · Oct. 8, 2026
Redrawn mid-decade
Several states redrew their House maps before 2026, including Texas and California, where voters approved Proposition 50 in November 2025. Comparing a district to its 2024 result can mislead where the lines moved.
Source: Ballotpedia — redistricting ahead of 2026 ↗ · Oct. 8, 2026
The generic ballot asks which party's candidate you would back in your own district. It does not name candidates, and it is a national number: it cannot say which seats change hands.
| Published by | Democrat | Republican | Margin | Dates |
|---|---|---|---|---|
| Silver Bulletin average ↗Polling average; Democrats crossed 50% in it for the first time this cycle. | — | — | D+8.9 | Oct. 8, 2026 |
| FiftyPlusOne average ↗Polling average weighted by pollster quality and recency. | 51.2% | 43.4% | D+7.8 | Oct. 8, 2026 |
| Pew Research Center ↗One survey of 3,917 registered voters; 15% not sure, 5% another candidate. | 44% | 36% | D+8.0 | Sept. 28 – Oct. 4, 2026 |
Observed published polling; an average blends many surveys, a single survey has its own error
A midterm is usually read as a verdict on the sitting president. These are the published measures of how much this one is about Donald Trump.
Most voters say the Iran war is driving gas prices up, and most say the administration's policies are pushing grocery prices up.
Source: CBS News / YouGov Battleground Tracker ↗ · Oct. 2, 2026
Observed survey and average figures as published; Pheme has not adjusted them
Share of registered voters calling each issue “very important” to their vote. Each issue is rated on its own, so the shares do not add to 100.
Source: Pew Research Center ↗ · Oct. 4, 2026
A rating is a forecaster's judgment, not a poll.
| State | Was | Now |
|---|---|---|
| Alaska | Tilt R | Toss-up |
| Michigan | Toss-up | Tilt D |
| Texas | Lean R | Tilt R |
| Georgia | Tilt D | Lean D |
| North Carolina | Tilt D | Lean D |
| Minnesota | Likely D | Lean D |
| Nebraska | Likely R | Lean R |
Source: Inside Elections rating changes, via 270toWin ↗ · Oct. 1, 2026
Every day Pheme puts these questions to its synthetic population: Census-weighted personas, reasoned by a language model. Nobody is surveyed. The Gap column is modeled minus observed. A large gap is the model missing, and it is shown on purpose.
“If the 2026 election for the U.S. House were held today, would you vote for the Democratic or the Republican candidate in your district?”
Modeled Loading the latest daily run…
| Answer | Modeled | Observed | Gap |
|---|---|---|---|
| Democrat | — | 44% | — |
| Republican | — | 36% | — |
| Undecided | — | — | — |
Bias-corrected reading: shown once a run carries one.
Observed column: Pew, registered voters, Sept. 28 – Oct. 4. Pew also offered “another candidate” (5%) and left 15% not sure, so compare the Democratic-minus-Republican margin rather than each party's level.
“Do you approve or disapprove of the way Donald Trump is handling his job as president?”
Modeled Loading the latest daily run…
| Answer | Modeled | Observed | Gap |
|---|---|---|---|
| Approve | — | 30% | — |
| Disapprove | — | — | — |
| Unsure | — | — | — |
Observed column: Pew, Sept. 28 – Oct. 4. Pew reports approval only in this release; the modeled Disapprove and Unsure have no published counterpart here.
“Will your vote for Congress this November be a vote to support Donald Trump, a vote to oppose him, or is he not a factor?”
Modeled Loading the latest daily run…
| Answer | Modeled | Observed | Gap |
|---|---|---|---|
| Support Trump | — | 19% | — |
| Oppose Trump | — | 44% | — |
| Not a factor | — | — | — |
Observed column: Pew, Sept. 28 – Oct. 4. The remaining Pew respondents said Trump is not a factor or gave no answer; that residual is not published as one number.
“Which issue matters most to your vote for Congress this year?”
Modeled Loading the latest daily run…
| Answer | Modeled |
|---|---|
| Cost of living | — |
| Gas and energy prices | — |
| Health care | — |
| Immigration | — |
| Democracy | — |
| Something else | — |
No directly comparable published question: Pew asks how important each issue is, not which one matters most.
Every tracked question, over time → · Where the model misses →
These open Ask with the question filled in; nothing runs until you press Run. What comes back is a modeled reaction to your hypothetical, not a measurement.
Your input a hypothetical question; the answer is modeled