NINE145 EDGE+ · ELECTION EXPLAINER ·
Americans vote on Tuesday, 3 November. Selected late-October surveys favour Democrats nationally, but control of Congress will be decided by individual House and Senate contests. Here is how to read the numbers without confusing a poll, a projection and a result.
- 52% versus 44%: Marist’s likely-voter poll gave Democrats an eight-point lead; Emerson’s later survey put the split at 52% versus 42%.
- 218 of 435: that is the majority threshold in a fully occupied House of Representatives.
- 53 versus 47: Republicans hold 53 Senate seats; 45 Democrats and two independents make up the Democratic-aligned side.
The midterms decide who can turn political demands into congressional action. They do not elect a president. For readers outside the United States, the simplest way to follow the contest is to keep three layers separate: public opinion, the seat arithmetic and the official count.
This pre-election edition uses the dated primary-source releases linked throughout. It does not contain live returns. Scroll through the story; tap a coloured chart item or its label to inspect the value.
READOUT 01
The national picture starts with people, not seats
Pew’s survey of registered voters put Democratic House candidates at 44% and Republicans at 36%. Another 5% preferred a different candidate and 15% were unsure. Interviews ran from 28 September to 4 October; these are an earlier snapshot, not election-eve returns.
The eight-point gap describes support across the country. It does not award eight extra seats or mean an eight-point chance of victory. The width of each ribbon’s vertical cross-section follows its published share. The ribbons travel together because this is one distribution; they do not show voters switching parties.
A useful first question is therefore: what exactly has been measured? Here the answer is the stated preference of registered voters in a national survey. The eventual electorate will consist of people who actually cast a valid ballot.
View chart data and source
Pew: registered voters; 28 September–4 October 2026. Band thickness = share. Total 100%.
| Measure | Value | Unit |
|---|---|---|
| Democratic | 44 | % |
| Republican | 36 | % |
| Other candidate | 5 | % |
| Not sure | 15 | % |
Pew Research Center — national survey, 28 September–4 October 2026
READOUT 02
Two later surveys favour Democrats, without settling the race
Marist’s 19–20 October survey recorded 52% Democratic support and 44% Republican support among likely voters. Emerson’s 24–26 October poll recorded 52% and 42%, respectively, also among likely voters. The Democratic advantages were eight and ten percentage points.
The shared direction is worth watching. The different gaps should not be read as a measured two-point swing: these were different surveys, with different samples and methods. Nor does the shared Democratic figure establish an exact national vote share.
These bars show each published party percentage on the same 0–100 scale. The unshown responses are not relabelled as undecided. Published percentages can be rounded, and a remainder alone does not tell us which answers people gave. This comparison is a small selection of primary-source surveys, not a polling average or an exhaustive election forecast.
View chart data and source
Likely voters. Marist: 19–20 October; Emerson: 24–26 October 2026. Separate surveys, not a time series. Both original sources are linked in the methods table below.
| Measure | Value | Unit |
|---|---|---|
| Marist · Democratic | 52 | % |
| Marist · Republican | 44 | % |
| Emerson · Democratic | 52 | % |
| Emerson · Republican | 42 | % |
Marist Poll — 19–20 October 2026; commissioned with NPR/PBS News
READOUT 03
For some voters, the congressional ballot is also a judgement on the president
Marist also asked how the president influenced the congressional vote. Among likely voters, 44% described their choice mainly as a vote against him, 21% mainly in his favour and 35% said he had little influence on their choice.
The radial pattern gives every percentage point one equal mark. Its rings and colours are a way to see the balance; they are not states, demographic groups or individual survey participants.
The distinction matters because a congressional election can carry several messages at once. A voter might care about the national administration, the local candidate and the household budget. This question captures one stated motivation. It cannot tell us how much each concern caused a particular choice, and it cannot replace the district-by-district count.
View chart data and source
Marist likely voters, 19–20 October 2026. One mark = one percentage point. Responses sum to 100%; positions are decorative.
| Measure | Value | Unit |
|---|---|---|
| Mainly against | 44 | % |
| Mainly in favour | 21 | % |
| Little influence | 35 | % |
Marist Poll — 19–20 October 2026; commissioned with NPR/PBS News
READOUT 04
Household economics sits near the centre of the election
In Pew’s earlier survey, 84% of registered voters rated the economy very important to their congressional vote. Democracy followed at 73%, healthcare at 68% and education at 62%. Immigration, foreign policy, energy policy and violent crime also drew majority responses.
Each ring has its own 100% track. A longer coloured fraction means a larger share rated that issue very important. Compare the fraction filled and the printed percentage, rather than the physical length of different-sized circles.
These responses overlap: someone can care deeply about both prices and healthcare. Adding them together would be meaningless. The figures also do not say which party voters trust on each issue. They tell us what was on respondents’ minds, leaving the separate questions of policy preference, party confidence and actual voting behaviour open.
View chart data and source
Pew registered voters, 28 September–4 October 2026. Very important to congressional vote. Independent responses; do not add to 100%.
| Measure | Value | Unit |
|---|---|---|
| Economy | 84 | % |
| Democracy | 73 | % |
| Healthcare | 68 | % |
| Education | 62 | % |
| Immigration | 59 | % |
| Foreign policy | 57 | % |
| Energy | 54 | % |
| Violent crime | 52 | % |
Pew Research Center — issues survey, 28 September–4 October 2026
READOUT 05
Voting plans help explain the count, not the winner
Gallup’s October survey found 43% of registered voters had already voted or intended to vote before Election Day. The share was 54% among Democrats, 38% among Republicans and 37% among independents.
That is a survey of behaviour and intentions, not an administrative tally of accepted ballots. It does not establish who won early voting or what percentage of each party’s supporters ultimately voted. Someone can change a plan; someone who describes a party identity can choose another party’s candidate.
It is still useful context for election night. Different kinds of ballots can be reported at different stages. An early batch and a later batch need not resemble each other. A changing lead should prompt questions about what has been counted and what remains, rather than an assumption that the underlying electorate has suddenly changed.
View chart data and source
Gallup fieldwork 1–19 October 2026; registered voters. Already voted or plan to vote before Election Day. Separate party groups.
| Measure | Value | Unit |
|---|---|---|
| Democrats | 54 | % |
| Republicans | 38 | % |
| Independents | 37 | % |
READOUT 06
The House is won one district at a time
All 435 voting seats in the House are contested in the midterms. With full membership, 218 seats constitute a majority. The gold marks show that mathematical threshold; they are not a prediction for either party. Each mark is one seat, arranged in a simple grid rather than a geographical map.
A national poll combines voters from very different districts. Running up a large lead in one district still wins only that district’s seat. A smaller lead spread across several competitive districts can produce more seats. Candidate strength, local concerns and the boundaries of each constituency therefore matter alongside the national climate.
This is why a favourable national survey is informative but incomplete. To judge control of the House, readers need district outcomes and the accumulating seat total. A poll lead and a chamber majority measure different things. The 218 benchmark describes a full House, rather than every procedural vote under every possible vacancy or attendance situation.
View chart data and source
US House: 435 voting members. NINE145 arithmetic: floor(435/2)+1 = 218. Anonymous seats, not assigned districts or results.
| Measure | Value | Unit |
|---|---|---|
| Majority threshold | 218 | seats |
| Remaining seats | 217 | seats |
US House of Representatives — membership and responsibilities
READOUT 07
The Senate starts at 53 Republican seats
The Senate’s published party division is 53 Republicans, 45 Democrats and two independents. The two independents belong to the Democratic caucus, putting that side’s starting alignment at 47. This graphic shows the existing chamber, not the outcome of tomorrow’s vote and not the number of seats being contested.
Only 35 Senate seats are being contested: 33 regular elections and two special elections, in Florida and Ohio. Republicans are defending 22 of those seats and Democrats 13, according to the European Parliamentary Research Service’s October briefing. The election map is therefore different from the chamber’s full membership.
Starting from that balance, a net gain of three seats for the Democratic-aligned side would produce 50–50; a net gain of four would produce 51–49. A net gain subtracts losses from gains. Winning four Republican-held seats while losing one Democratic-aligned seat would therefore be a net gain of three.
At 50–50, the vice-president’s constitutional power to break Senate ties matters. At 51, a side has a numerical majority without that tie-break. These are arithmetic scenarios holding caucus alignment constant. They do not assign win probabilities, and they do not suggest that all 100 senators face election at once.
View chart data and source
US Senate party division: 53 Republican, 45 Democratic, 2 independent. Caucus alignment and vice-presidential role linked below. Baseline, not forecast.
| Measure | Value | Unit |
|---|---|---|
| Democratic | 45 | seats |
| Republican | 53 | seats |
| Independent | 2 | seats |
What each survey actually measured
“Registered voters” means people registered to vote. “Likely voters” is a pollster’s assessment of who is expected to participate. Neither description is interchangeable with the final electorate. A survey of all adults includes people who may not be registered or eligible.
Read the interview dates before the publication date. A poll released shortly before voting may describe opinions collected earlier. Also check whether the question names local candidates or asks about a generic party choice. The national party-choice figures here are generic congressional measures, not a combined list of candidate-level results.
| Producer | Fieldwork | Relevant population | Method note |
|---|---|---|---|
| Pew Research Center | 28 September–4 October 2026 | Registered voters | 3,917 registered voters within a survey of 5,062 adults. |
| Marist Poll | 19–20 October 2026 | Likely voters | Published margin of error: ±3.8 percentage points. NPR/PBS News/Marist partnership. |
| Emerson College Polling | 24–26 October 2026 | 1,000 likely voters | Published credibility interval: ±3 points. |
| Gallup | 1–19 October 2026 | Registered voters for the early-voting measure | Self-reported voting and plans, not counted returns. |
Why an eight-point lead is not an eight-point certainty
A poll estimates opinion from a sample. The reported margin of error or credibility interval is one expression of uncertainty, not a guarantee that the result must fall within a neat band. Question wording, participation, weighting and assumptions about who will vote can affect a survey too.
The uncertainty around a gap between two candidates is not simply the same as the uncertainty around one candidate’s share. Nor does a ten-point lead mean a 90% chance of winning. A probability forecast would require a separate model and its assumptions; this article does not manufacture one from the poll margins.
Several surveys pointing in the same direction provide more context than a single headline. But averaging incompatible populations or treating every poll as equally representative can create false precision. The comparison here keeps the original producers and populations visible.
What a change in congressional control could mean
The House and Senate both participate in making federal law. For ordinary legislation to advance to the president, the chambers must agree on the same text. A president can sign legislation or veto it. A congressional election therefore changes the people and coalitions involved in that process, even though it does not replace the president.
Control also affects who organises a chamber and its committees. The Senate has additional responsibilities, including considering presidential nominations. A party’s seat majority matters, but it does not mean every member agrees or that every proposal passes automatically. Procedural rules and the choices of individual legislators remain important.
For international readers, the practical question is which policies can secure enough support. Trade, public spending, security commitments and energy policy can involve different institutions and different legal routes. A change in the seat count is a reason to reassess those routes, not proof that a particular policy will change overnight.
The same discipline applies to markets. An election result, a legislative decision and a market reaction are separate events. This article does not turn polling percentages into forecasts for stocks, currencies or oil.
How to follow election night without getting misled
- Start with the jurisdiction. Use state and local election authorities for official returns. The United States does not have one single nationwide counting office publishing every local result.
- Separate counted votes from seats. A candidate can lead the running tally before enough ballots have been processed to settle the contest. A seat total should identify its basis and status.
- Ask what remains. The order in which ballots and areas report can change the running margin. Do not assume that the first reported batch represents the entire electorate.
- Distinguish a projection from certification. Election-night totals are generally unofficial. Canvassing and certification are subsequent official steps, with timetables that vary by state.
- Compare like with like. A poll of likely voters, a tally of early ballots and a final certified result answer different questions. One cannot be silently substituted for another.
Counting that continues beyond election night is part of the process. Its existence alone does not establish wrongdoing. If a total changes, the useful information is the election authority’s explanation, the ballots involved and the applicable procedure.
Quick answers
When are the US midterm elections?
Election Day is Tuesday, 3 November 2026. Early-voting and postal-voting arrangements vary by state. Readers voting in the United States should check their own election authority for polling hours and applicable deadlines.
Do Americans elect a president in these midterms?
No. Midterms occur between presidential elections. They include congressional contests and other state and local races.
Can the national House poll tell us who controls the Senate?
No. The Senate is decided through its own state contests. A national House party-preference question does not distribute Senate seats.
What does “percentage points” mean?
It is the difference between percentages. A share of 52% compared with 44% is an eight-percentage-point gap. Calling it an 8% increase would describe a different calculation.
Are the moving ribbons a prediction?
No. Their numerical proportions reproduce a published survey. Movement helps reveal the chart; it adds no respondents, votes, seats or probability.
Sources and calculation notes
Reporting and visual analysis: NINE145. Primary sources checked. Fieldwork dates stay attached to each survey.
NINE145 calculated the published party gaps by subtraction and the full-House majority as one more than half of 435, rounded down. Senate scenarios use the stated party balance and unchanged caucus alignment. No polling percentage has been converted into a predicted seat count. The graphic data are available in the accompanying tables.
