Midterm Election Exit Poll Statistics 2026 | Voter Poll by SSRS, Networks, Methodology & Key Facts

Midterm election exit poll statistics for 2026 are not available yet: The Voter Poll by SSRS, which replaced the Edison and National Election Pool exit poll, expects about 100,000 interviews, a national sample above 75,000 with a margin of error of plus or minus 0.6 points, and state results that post as polls close from 6 PM ET on November 3, 2026 until 1 AM ET on November 4. In-person exit polls run in 15 states, and every result is re-weighted to the actual vote count overnight.

Midterm Election Exit Poll in US 2026 – Introduction

Midterm election exit poll statistics do not exist until voters finish casting ballots. Election Day is November 3, 2026, and the polls close in waves between 6 PM ET and 1 AM ET. Exit polls then reveal who voted, how they voted and which issues drove their choices. This year the poll has a new name and a new design, with about 100,000 interviews planned across all 50 states.

The exit poll no longer relies only on interviews at polling places. Millions of Americans voted before Election Day, so the poll now blends in-person interviews with online and phone surveys of verified voters. That change affects how the numbers should be read, and how fast they arrive. Here are the midterm election exit poll statistics that are known today: the sample sizes, margins of error, closing times and release schedule. This report will be updated as results come in.

Interesting Facts about Midterm Election Exit Poll in US 2026

VOTER POLL MARGIN OF ERROR BY GEOGRAPHY

National sample (76,130) | ███ 0.6
Texas (5,000)            | ██████████ 1.9
California (4,400)       | ██████████ 2.0
Utah (340)               | ████████████████████████████████████ 7.1
North Dakota (250)       | ██████████████████████████████████████████ 8.3

Source: SSRS, The Voter Poll Methods Statement, October 27, 2026; Associated Press; 270toWin poll closing times. Results are not yet released.

The Voter Poll by SSRS replaces two older systems. For years the National Election Pool used exit polls run by Edison Research, and the Associated Press and Fox News used AP VoteCast. In 2025 SSRS bought Edison Research and merged the two approaches. Seven news organizations now share one poll with about 100,000 interviews, a national sample above 75,000, and a margin of error of 0.6 points nationally.

Precision depends on geography. Texas has 5,000 respondents and a margin of error of 1.9 points. Utah has 340 and North Dakota 250, with errors of 7.1 and 8.3 points. In-person interviews happen in 15 states only. The first polls close at 6 PM ET and the last at 1 AM ET, so complete results cannot exist by 11 PM.

The Voter Poll by SSRS Sample Size and Survey Modes 2026

VOTER POLL INTERVIEWS BY SOURCE

Probability-based sources | ███████████████ 30,189
Verasight verified panel  | ███████████████████████ 45,941
Election Day exit polls   | ███████████ 22,100

Source: SSRS, The Voter Poll Methods Statement, October 27, 2026, Table 1. The exit poll total and the combined total are our calculation from state targets.

The pre-election survey covers 76,130 registered voters, split between 30,189 from probability-based sources and 45,941 from a verified online panel. The probability sources include voter-file samples invited by postcard, text, email and phone, plus the SSRS Opinion Panel. Every pre-election respondent is matched to the L2 voter file. About 22,100 more interviews come from in-person exit polls, which brings the total to about 98,200, close to the stated 100,000.

Only registered voters are in the sample, and the poll does not sample the District of Columbia because it would not affect the national balance of power. Respondents can answer in English or Spanish. Pre-election respondents who had not yet voted are recontacted from 5 PM local time on Monday, November 2 until near poll closing on Tuesday, to confirm whether and how they voted. Their final answer replaces their earlier one.

Exit Poll States and Election Day Interview Sites 2026

ELECTION DAY EXIT POLL LOCATIONS BY STATE GROUP (each █ = 10 locations)

13 states at 30 each | ███████████████████████████████████████ 390
Michigan             | ███ 32
Alaska (Election Day)| ██ 20

Source: SSRS, The Voter Poll Methods Statement, October 27, 2026. The 442 location total and 22,100 interview estimate are our calculation.

The poll runs in-person exit polls in 15 states. In 13 of them it samples 30 polling locations. Michigan uses 32 to ensure it represents the whole state, and Alaska uses 20 Election Day locations plus 5 early-vote centers, where interviewers worked 10 shifts between October 19 and 30. That gives 442 Election Day locations, by our calculation.

At each site an interviewer approaches every nth voter as they leave, with a target of at least 50 completed questionnaires. Locations are chosen with probability proportionate to their 2024 Election Day turnout. In Alaska’s early-vote centers, voters complete the survey mostly on tablets, with paper forms available. The 15 states get about 1,500 exit interviews each, 1,600 in Michigan and 1,000 in Alaska. Other states rely only on pre-election surveys.

Voter Poll State Sample Sizes and Margins of Error 2026

VOTER POLL TOTAL SAMPLE BY STATE (each █ = 250 respondents)

Texas          | ████████████████████ 5,000
California     | ██████████████████ 4,400
Florida        | ██████████████████ 4,400
Michigan       | █████████████████ 4,340
Pennsylvania   | ████████████████ 4,000
New York       | ██████████ 2,450
Utah           | █ 340
North Dakota   | █ 250

Source: SSRS, The Voter Poll Methods Statement, October 27, 2026, Tables 1 and 3. Margins are at the 95% confidence level for proportions near 50%, with a design effect of 1.8 for state samples.

Texas has the largest state sample at 5,000, followed by California and Florida at 4,400 and Michigan at 4,340. Pennsylvania, Arizona, Georgia, Iowa, Maine and several other states have 4,000. New York has 2,450. Smaller states have far fewer, down to 340 in Utah and 250 in North Dakota.

Margins of error rise as samples shrink, from ±1.9 in Texas to ±8.3 in North Dakota. SSRS uses a design effect of 1.8 for state samples and 3.0 for the national sample, which widens the margins to account for weighting. These margins apply to a result near 50%. They are smaller for lopsided results and larger for subgroups, such as young voters or a single racial group.

Voter Poll Margin of Error by Sample Size 2026

STATE MARGIN OF ERROR BY SAMPLE SIZE (plus or minus points; each █ = 0.5)

100 respondents   | ██████████████████████████ 13.1
250 respondents   | █████████████████ 8.3
500 respondents   | ████████████ 5.9
1,000 respondents | ████████ 4.2
2,000 respondents | ██████ 2.9
4,000 respondents | ████ 2.1
5,000 respondents | ████ 1.9

Source: SSRS, The Voter Poll Methods Statement, October 27, 2026, Table 4. Margins at the 95% confidence level for proportions near 50%; design effects of 1.8 (state) and 3.0 (national).

The table is a guide to subgroup results. A state subgroup of 100 voters carries a margin of ±13.1 points, and a national subgroup of 100 carries ±17.0. With 1,000 respondents the state margin drops to ±4.2 and the national margin to ±5.4. At 5,000 the state margin is ±1.9. The full national sample of 75,000 reaches ±0.6.

National margins run larger than state margins at the same size because the national design effect is 3.0, against 1.8 for states. A national result from 1,000 respondents is therefore less precise than a state result from 1,000. The margins are estimates for a hybrid sample that mixes probability and non-probability sources, and SSRS notes that no industry standard exists for that calculation.

From Exit Polls to The Voter Poll: Timeline of Changes 2003 to 2026

VOTER POLL PRE-ELECTION SAMPLE, 2025 VS 2026 (each █ = 2,000)

2025 (CA, NJ, NYC, VA)  | ██████ 12,753
2026 (all 50 states)    | ██████████████████████████████████████ 76,130

Source: SSRS, “Introducing The Voter Poll by SSRS,” October 28, 2025; SSRS 2025 Methods Statement; Wikipedia National Election Pool entry. The 6.0x growth in pre-election sample is our calculation.

The exit poll changed after voting shifted away from Election Day. Edison conducted the National Election Pool exit polls from 2003. For 2020 pollsters added early-voting locations and phone interviews, and for 2024 email and text outreach. In 2025, SSRS bought Edison and launched The Voter Poll for ABC, CBS, CNN, NBC, Fox News and the Associated Press, with MS NOW joining by 2026.

The first run was small. In 2025, SSRS surveyed 12,753 voters in California, New Jersey, New York City and Virginia, with about 2,100 exit poll interviews each in New York City and Virginia. The 2026 pre-election sample of 76,130 is about 6.0 times larger, by our calculation, and it covers every state. That scale is why national margins now run near 0.6 points.

Voter Poll Weighting and Statistical Adjustments by State 2026

STATES BY WEIGHTING TIER (each █ = 1 state)

Basic weighting (self-report and voter file)   | ███████████ 11
Expanded weighting                             | ██████████████████████ 22
Most detailed weighting, plus MRP calibration  | █████████████████ 17

Source: SSRS, The Voter Poll Methods Statement, October 27, 2026, Table 2. Tier counts are our calculation from the published state groups.

SSRS weights each state sample to match registered voters, using the L2 voter file, Census-based models and modeled political traits. Eleven smaller states get basic weights built from age, gender, race, education and recalled 2024 vote, plus voter-file variables such as party registration and 2022 turnout. Twenty-two states add interactions and variables such as 2026 primary turnout and a likely-voter score.

The 17 states with the largest samples get the most detailed treatment. Their registered-voter weight is also calibrated to an estimate of the 2026 Senate or governor vote by county-based region, using multilevel regression and poststratification, known as MRP. Texas adds Hispanic-by-age and Hispanic-by-region weights, and Michigan adds a variable for Muslim or Middle Eastern heritage. After votes are counted, all results are weighted again to match the actual state totals.

Election Night Poll Closing Times and Exit Poll Release Schedule 2026

POLL CLOSING WAVES, NOVEMBER 3-4, 2026 (hours after 6 PM ET; each █ = 1 hour)

6 PM ET  first polls (parts of IN, KY)  | 0
7 PM ET  Georgia, Virginia, others      | █ 1
8 PM ET  Florida panhandle, most states | ██ 2
9 PM ET  New York, parts of TX, MI, KS  | ███ 3
12 AM ET Alaska (most), Hawaii          | ██████ 6
1 AM ET  Aleutian Islands, Alaska       | ███████ 7

Source: 270toWin, 2026 Poll Closing Times; Outlook India and Rio Times reporting on poll closings. Times are Eastern and adjusted for time zones; local times differ.

Exit poll results post state by state as polls close, so the schedule above is also the release schedule. The first wave begins at 6 PM ET, when parts of Indiana and Kentucky close. Most major states follow between 7 PM and 9 PM ET. Many state results typically appear in the 7 PM to 11 PM window, but not all of them. Alaska’s polls do not fully close until 1 AM ET on November 4.

Early numbers change. SSRS says the final step weights results to match each state’s vote count, once votes are counted. Early exit poll tables are therefore preliminary and shift through the night as the vote totals arrive. Close races can take days to settle, and Alaska and California can take a week or more, according to SSRS’s Joe Lenski. Readers should treat the first exit poll readings as estimates.

Early Vote Benchmarks Behind 2026 Exit Poll Weights

EARLY BALLOTS CAST BEFORE ELECTION DAY

2022 midterm total      | ████████████████████████████████████ 45.9
2026, midday Oct 31     | █████████████████████████████████████████ 51.3

Source: University of Florida Election Lab, via Scripps News and Newsweek, October 30 to 31, 2026. Party shares reflect registration, not vote choice. The percentage change is our calculation.

The Voter Poll must reflect a very large early electorate. By midday October 31, voters had cast 51.3 million early ballots, 11.8% more than the 45.9 million early votes of 2022. More than 28 million were cast in person. In the final weighting step SSRS adjusts the share of mail, early in-person and Election Day voters to match the estimated composition of each state’s electorate.

Party registration among early voters is a context figure, not a prediction. In states that report it, 43.2% of early ballots came from registered Democrats, 31.7% from Republicans and 25.1% from unaffiliated voters. Registration is not vote choice, and the exit poll is the first source to show how those groups actually voted. Counts will keep changing until Election Day closes.

Voter Poll Limits and Margin of Error Bands by State 2026

NUMBER OF STATES BY VOTER POLL MARGIN OF ERROR (each █ = 1 state)

±2.1 or smaller | ████████████████ 16
±2.4 to ±3.9    | █████ 5
±4.2 to ±4.6    | ██████████████████ 18
±5.9 or larger  | ███████████ 11

Source: SSRS, The Voter Poll Methods Statement, October 27, 2026, Table 3. State counts per band are our calculation. The District of Columbia is not sampled.

Sixteen states have margins of ±2.1 points or smaller, including the states with the largest samples. Another 5 fall between ±2.4 and ±3.9. Eighteen states sit between ±4.2 and ±4.6, and 11 have margins of ±5.9 or more, including Utah at ±7.1 and North Dakota at ±8.3. These counts total 50 by our calculation. In-person exit polls take place only in 15 states.

Several limits apply. Margins of error are estimates, because the sample mixes probability and non-probability sources. Subgroup results have wider ranges than the toplines. The exit poll also describes the people who voted, and it should not be confused with a pre-election poll that estimates where a race stands. Results will not predict winners, which the vote count decides. A state with 500 respondents has a margin of ±5.9 points on any single percentage, so small differences between groups there should not be read as real.

Frequently Asked Questions About Midterm Election Exit Poll in US 2026

When will the 2026 midterm exit poll results be released?

Results post state by state after polls close. The first polls close at 6 PM ET on November 3, 2026, and the last close at 1 AM ET on November 4.

Who conducts the 2026 midterm exit poll?

SSRS conducts The Voter Poll for ABC, the Associated Press, CBS, CNN, Fox News Media, MS NOW and NBC.

What is The Voter Poll by SSRS in 2026?

It is a survey of voters that combines in-person exit polls, online and phone surveys, and mail and text invitations. It replaced the Edison exit poll and AP VoteCast.

How many people are interviewed for the 2026 exit poll?

About 100,000 voters, including a national sample of 76,130 before Election Day and about 22,100 in-person exit poll interviews.

What is the margin of error of the 2026 exit poll?

The national margin is ±0.6 points. State margins range from ±1.9 in Texas to ±8.3 in North Dakota.

Which states have in-person exit polls in 2026?

Alaska, Florida, Georgia, Iowa, Maine, Michigan, Minnesota, Nebraska, Nevada, New Hampshire, North Carolina, Ohio, Pennsylvania, Texas and Wisconsin.

Are exit polls available before polls close in 2026?

No. Exit poll results are not published until polls close in each state, which is why no 2026 results exist yet.

Do exit polls include early and mail voters in 2026?

Yes. The Voter Poll surveys verified registered voters online and by phone, including those who voted early or by mail, then combines them with Election Day interviews.

Why do exit poll numbers change during election night in 2026?

SSRS weights results to match each state’s final vote count, so early exit poll tables shift as votes are counted.

Is Washington, DC included in the 2026 exit poll?

No. SSRS did not sample the District of Columbia because it would not affect the national balance of power.

How accurate are state exit polls in 2026?

Accuracy depends on sample size. Texas has 5,000 respondents and ±1.9 points, while states with 500 have ±5.9 points. Subgroup margins are larger.

How many early votes were cast before the 2026 midterms?

More than 51.3 million early ballots had been cast by midday October 31, 2026, above the 45.9 million early votes counted in 2022.

What is MRP weighting in The Voter Poll for 2026?

For the 17 states with the largest samples, SSRS calibrates the registered-voter weight to estimated Senate or governor vote choice by region, using multilevel regression and poststratification.

How does The Voter Poll verify voters in 2026?

Pre-election respondents are matched to the L2 voter file and deduplicated. Verasight panelists are also phone-verified when they join the panel.

Can exit polls predict who wins in 2026?

No. The Voter Poll describes the people who voted and why. Vote counts decide winners, and exit poll results are re-weighted to match those counts.

What is the difference between an exit poll and a pre-election poll in 2026?

Pre-election polls estimate what likely voters may do. The Voter Poll describes people who actually voted and what influenced their decisions.

How many states have in-person exit polls in 2026?

Fifteen states have them. The other 35 sampled states rely on pre-election surveys of verified registered voters, and the District of Columbia is not sampled.

Disclaimer: This research report is compiled from publicly available sources. While reasonable efforts have been made to ensure accuracy, no representation or warranty, express or implied, is given as to the completeness or reliability of the information. We accept no liability for any errors, omissions, losses, or damages of any kind arising from the use of this report.