AP Voter Poll will survey 75,000 voters for the 2026 midterms

Initial results are set for 5 p.m. ET Nov. 3, then updated through the night

The Associated Press will survey about 75,000 voters before the Nov. 3, 2026 election, with about 25,000 additional in-person interviews at polling places in 15 states on Election Day. The survey, called the AP Voter Poll, is conducted by SSRS, a nonpartisan research firm, for The Associated Press, ABC News, CBS News, CNN, Fox News Media, NBC News and MS Now. The first set of results is scheduled for 5 p.m. ET on Nov. 3, with additional state-level and national findings to be published throughout the night. Linley Sanders reported for The Associated Press.

Launched in 2025, the AP Voter Poll is an extensive survey of verified voters that blends the methodologies of AP VoteCast, which was conducted in general elections from 2018 to 2024, with traditional exit polls. It interviews voters by phone and online in all 50 states in the days leading up to the election and collects additional data from in-person interviews at polling places in 15 states on Election Day.

The pre-election component runs from Oct. 22 through Nov. 3 in all 50 states. To assemble it, the AP Voter Poll draws a random sample of registered voters from state voter files. Those voters are contacted through the mail and invited to take the survey online, by phone or by text, and some voters who do not respond to the initial invitation are contacted by phone.

The samples draw on probability- and nonprobability-based opinion panels. Other voters are recruited through the probability-based SSRS Opinion Panel, which matches each respondent to the voter file. The poll also interviews voters recruited through the Verasight Verified Panel, where some participants are not selected randomly but are instead recruited through internet advertising or other methods. The respondents’ voter registration status is verified separately, and the sample is combined with the other pre-election data sources so the additional interviews do not overrepresent or underrepresent important demographic and ideological groups.

Many voters will have already cast their ballot by the time they take the survey. Voters who have not yet voted will be recontacted on Election Day, when the poll verifies their final vote choice and the method by which they voted.

On Election Day, in-person interviews will be conducted with voters at polling places in Alaska, Florida, Georgia, Iowa, Maine, Michigan, Minnesota, Nebraska, Nevada, New Hampshire, North Carolina, Ohio, Pennsylvania, Texas and Wisconsin, adding approximately 25,000 interviews across those states. Interviews are also being conducted at early-vote centers in Alaska in late October. Data from the pre-election and in-person surveys are blended so the combined surveys are representative of the estimated 2026 electorate in each state where an exit poll is conducted.

The result is a national survey drawing on interviews from all 50 states, along with separate state-by-state results in every state where a statewide race for U.S. senator or governor is taking place. Interviews conducted in the five states without a statewide race — Indiana, Missouri, North Dakota, Utah and Washington — will be included in the national survey but will not be reported separately.

Voters are asked a standard set of demographic questions, including their age, sex, race and ethnicity, education level and ideology. The survey also asks how voters assess the country’s current leadership and direction, how they are evaluating the candidates on their ballots and what they believe are the key issues facing the country, such as crime, the economy, foreign policy, healthcare and immigration. The questionnaire asks about voters’ choices in U.S. Senate and governor races, and this year California voters will also be asked how they voted on Proposition 40, a proposed one-time tax on billionaires in the state.

Participants in the national survey are asked whether they voted for the Democratic or Republican candidate in their local U.S. House race. That national finding, often called the “generic” House ballot, is meant to shed light on the shape of the race for Congress across the country, but it does not include results from specific congressional districts.

The survey is not a replacement for the vote count, and the topline findings — how many voters overall say they supported a specific candidate or party — will not be reported. The Associated Press says the design is intended instead to provide insight into how specific groups voted and why they voted the way they did.

The AP Decision Team will have the Voter Poll available as it analyzes races on Election Night. Races will not be called on the survey alone; the team also considers election results, vote history, registration and turnout data, demographic information and other data. In some contests where one political party has an established history of lopsided victories, and other election data — overwhelming advantages in voter registration, advance ballot returns and pre-election polling — already point to a victory by that party, the survey results may be the final data point that allows the Associated Press to declare a winner as soon as polls close.

The poll has already been used to explain past results. In the 2025 New Jersey governor’s election, AP Voter Poll results showed that Democrat Mikie Sherrill won after getting support from about half of white voters in the state and a strong showing among Black, Hispanic and Asian voters. In the 2025 New York City mayoral race, the survey found that younger voters overwhelmingly voted for Zohran Mamdani in his successful campaign, while Andrew Cuomo’s coalition was much older.

The Associated Press describes the Voter Poll as an integral part of its coverage of the 2026 election and a tool for explaining the voting coalitions that shaped the race for congressional control and decided key races for U.S. Senate and governor.