Six Things People Believe About the Midterms, Checked Against the Data

On Tuesday, Nov. 3, 2026, voters elect all 435 members of the House ^[1], 35 senators ^[3][19][20] and 36 governors ^[22]. Six beliefs about midterm elections come up every cycle, from the president’s party always losing seats to mail ballots favoring one side. We checked each one against official election returns, turnout counts and published research, and graded each as worded. One holds up: midterm turnout is always “low” when that means below presidential years. Four are partly right, and one is not supported by the evidence. This is a check of the record, not a forecast.

Every figure below describes elections that have already happened, or maps and rules in place as of Oct. 26, 2026. We use no polls and no race ratings, and nothing here says who will win on Nov. 3. Each claim ends with the evidence that would change its verdict.

The president’s party

From 1946 to 2022, the president’s party lost House seats in 18 of 20 midterms ^[1]. Across all 20, the median change was a loss of 22 seats and the mean a loss of about 26 ^[1]. Counting only the 18 losses, the median was 27 seats and the mean about 29 ^[1]. The smallest loss was 4 seats, in 1986, and the largest was 64, in 2010 ^[1].

Large losses came under both parties. Democratic presidents lost 56 seats in 1946, 54 in 1994 and 64 in 2010; Republican presidents lost 50 in 1958, 48 in 1974 and 42 in 2018 ^[1]. The two exceptions also split by party. In 1998, with Bill Clinton, a Democrat, in the White House, his party gained 4 House seats. In 2002, under George W. Bush, a Republican, his party gained 8 ^[1]. Before the years in this chart, Franklin Roosevelt’s Democrats gained 9 seats in 1934 ^[3].

1946-561950-281954-181958-501962-61966-471970-121974-481978-141982-261986-41990-81994-5419984200282006-302010-642014-132018-422022-9-80-60-40-20020

Data table

Midterm year	Seats gained or lost by the president's party
1946	-56
1950	-28
1954	-18
1958	-50
1962	-6
1966	-47
1970	-12
1974	-48
1978	-14
1982	-26
1986	-4
1990	-8
1994	-54
1998	4
2002	8
2006	-30
2010	-64
2014	-13
2018	-42
2022	-9

Source: U.S. House of Representatives, Office of the Historian (History, Art and Archives). Party Divisions of the House of Representatives, 1789 to Present (HTML table), 79th Congress onward; midterm elections only · retrieved Oct. 20, 2026 · 20 rows · as of 2022 · annual. Download CSV

The Senate pattern is weaker. The president’s party gained Senate seats or broke even in 1962, 1970, 1982, 1998, 2002 and 2018 ^[2][3], and in 2022 by the American Presidency Project’s count ^[3]: 7 of the 20 midterms from 1946 to 2022. The Senate Historical Office’s table, which lists independents separately, shows Democrats one seat lower after 2022 ^[2].

The strongest case for the belief is the record itself: 18 losses in 20 midterms, under presidents of both parties ^[1]. Three lines of research explain it in different ways. Edward Tufte’s 1975 study of eight midterms from 1938 to 1970 linked the president’s party’s share of the national House vote to the president’s popularity and to growth in real disposable income ^[4]. James Campbell found in 1985 that “the coattails/surge-and-decline theories offer somewhat more accurate predictions of seat losses” ^[5]. Joseph Bafumi, Robert Erikson and Christopher Wlezien found that early polls in a midterm year “project a normal vote result,” and that preferences then “almost always move toward the out party” ^[6].

The strongest case against it starts with the word “always.” The president’s party gained House seats in 1934 ^[3], 1998 and 2002 ^[1]. The size of the change also ranges too widely, from a gain of 8 seats to a loss of 64 ^[1], for an average to say much about any one year. Tufte’s model implies that a popular president and rising incomes shrink the loss ^[4], while Campbell concluded that “the economy does not have the impact suggested by previous research” ^[5], so even the causes of the pattern are disputed.

The president’s party in 2026 is the Republican Party. The record above covers 20 past elections and is not a forecast for Nov. 3. It reads the same whichever party holds the White House: the large losses and the two exceptions came under presidents of both parties.

What would change the verdict: Nothing that happens in 2026 changes the record from 1946 to 2022, which is what this verdict grades. A gain for the president's party on Nov. 3 would be the third since 1946.

Turnout and party

Pew Research Center wrote in 2025: “For the past several decades, high turnout often benefited Democratic candidates.” It added: “This pattern has been weakening since 1960 but was still evident in the 2020 election,” and “The 2024 election was different” ^[9]. Pew’s studies of validated voters, people it confirmed had voted, show the change.

• 68% to 29%House vote of people who skipped 2016 but voted in 2018, Democrats to Republicans ^[11]
• 71% vs. 67%Turnout in 2022 of Trump's and Biden's 2020 voters ^[10]
• 44% to 40%2024 nonvoters' preference, Trump to Harris ^[9]

In 2018, Hillary Clinton’s 2016 voters turned out at 78% and Donald Trump’s at 74% ^[11]. In 2022 the edge ran the other way, and Pew found that Republican gains came mostly from “partisan differences in turnout” rather than from voters switching sides ^[10]. In 2024, 89% of Trump’s 2020 voters voted again, against 85% of Joe Biden’s ^[9]. People who sat out 2020 had preferred Biden 46% to 35% ^[9].

The longer record is mixed too. Thomas Hansford and Brad Gomez used Election Day rainfall to separate the effect of turnout from everything else that moves votes. Studying county results in presidential elections from 1948 to 2000, outside the South, they found that a 4-point swing in turnout moved the national Democratic share by “just under one percentage point” ^[7]. They concluded that “Democratic candidates, on average, are helped by higher turnout,” with the size of the effect depending on local partisanship and on which party held the White House ^[7]. Daron Shaw and John Petrocik summarized their 2020 book, The Turnout Myth, in National Affairs, a journal published with the American Enterprise Institute, an advocacy group ^[8]. They wrote that “seventy years of survey data and election returns suggest that turnout has no systematic partisan consequences” ^[8].

People who hold the belief can point to the best causal estimate for 1948 to 2000, which points the Democrats’ way ^[7], to Pew seeing the pattern as late as 2020 ^[9], and to the new voters of 2018, who broke strongly for Democrats ^[11]. Against it, the effect was small and conditional even in that study ^[7], Shaw and Petrocik found little or no systematic effect over 70 years ^[8], and in the two most recent national elections the turnout edge went to Republicans ^[9][10].

One reading fits all of these findings: higher turnout helps whichever party the occasional voters lean toward in a given year, and that has changed as the parties’ voters have changed. Our post on The Party Realignment: How Democrats and Republicans Changed Over Time traces those shifts.

What would change the verdict: A causal study covering the 2016 to 2024 elections, or validated-voter studies of 2026. Pew's study of 2022 came out in July 2023.

Mail voting

In 2024, about 30% of all ballots were cast by mail, according to the U.S. Election Assistance Commission, and most voters cast their ballots in person, before or on Election Day ^[16]. The parties’ voters used mail very differently: 44% of Democratic voters and 26% of Republican voters said they voted by mail ^[15].

• About 2 pointsRise in turnout where counties moved to universal vote-by-mail, 1996 to 2018 ^[12]
• No effectChange in either party's share of turnout or of the vote, same study ^[12]
• 44% vs. 26%Democratic and Republican voters who voted by mail in 2024 ^[15]

Whether a rule that makes mail voting easier helps one party is a separate question, and three studies tested it by comparing similar voters under different rules. Daniel Thompson, Jennifer Wu, Jesse Yoder and Andrew Hall studied counties in California, Utah and Washington that moved to universal vote-by-mail at different times from 1996 to 2018 ^[12]. Turnout rose about 2 points, and neither party’s share of turnout or of the vote changed ^[12]. Michael Barber and John Holbein used 40 million voter records from Washington and Utah and found turnout rose 1.8 to 2.9 points ^[13]. Their estimate of the change in the Democratic vote share was 0.7 points, with a 95% range from -0.7 to 2 points, so it cannot be told apart from zero ^[13].

For 2020, Yoder and seven coauthors used the age-65 cutoff for absentee voting in Texas and Indiana. They found that 65-year-olds voted at nearly the same rate as 64-year-olds, with no meaningful partisan effect on turnout ^[14].

The case for the belief rests on who uses mail. Democrats now vote by mail far more than Republicans ^[15], so rules that widen or narrow mail voting fall unevenly on each party’s voters. The two earlier studies also cover states that moved every voter to mail, and Thompson and his coauthors say their results hold “during normal times” ^[12]. The case against rests on the causal studies: three different designs, one of them set in 2020, found no partisan effect of mail-voting policy ^[12][13][14]. The 2020 study suggests why. Voters just old enough for the easier absentee rules turned out at about the same rate as those just too young ^[14], so the rule changed how they voted more than whether they voted.

Questions about the security of mail ballots and voter eligibility are a separate debate, covered in SAVE Act: Costs, Burdens, and Benefits Compared.

What would change the verdict: A causal study of a mail-voting change after 2020 that found a partisan effect.

Competitive seats

A House race is one line on the ballot. On Nov. 3, 2026, voters in 36 states elect a governor ^[21][22]. Thirty-five Senate seats are up: the 33 regular seats ^[19] plus special elections in Florida and Ohio ^[3][20]. Eighty-eight state legislative chambers have seats on the ballot ^[22], and 145 statewide measures are on the November ballot, out of 164 certified for 2026 in 40 states ^[23].

• 36Governor races ^[22]
• 35U.S. Senate seats, 33 regular ^[19] and 2 special ^[20]
• 88State legislative chambers with seats on the ballot ^[22]
• 145Statewide ballot measures in November ^[23]

The first half of the belief has a basis. We counted every House race in 2024 from the Clerk of the House’s official returns ^[18]. Of the 435 races, 69 (about 1 in 6) were decided by fewer than 10 points, and 37 by fewer than 5 ^[18]. The other 366 were won by 10 points or more, or had no opponent on the ballot; 259 were won by 20 points or more against a listed opponent, and 22 had none ^[18]. That is one election; we did not count earlier years.

Under 5 points375 to 10 points3210 to 20 points8520 points or more259No opponent on the ballot220100200300

Data table

Winner's margin	House seats
Under 5 points	37
5 to 10 points	32
10 to 20 points	85
20 points or more	259
No opponent on the ballot	22

Source: Clerk of the U.S. House of Representatives. Statistics of the Presidential and Congressional Election of November 5, 2024 (published March 10, 2025), 'For United States Representative', all 435 districts · retrieved Oct. 23, 2026 · 5 rows · as of Nov. 5, 2024. Download CSV

The chance that one vote decides a large election is tiny. Andrew Gelman, Nate Silver and Aaron Edlin estimated it at about 1 in 10 million in the closest states in the 2008 presidential election, and 1 in 60 million on average ^[24]. Smaller races are close more often. Casey Mulligan and Charles Hunter found that about 1 in 100,000 votes in U.S. congressional elections, and about 1 in 15,000 in state legislative elections, were cast for a candidate who tied or won by a single vote ^[25].

Read narrowly, the belief is close to right: a single vote almost never decides a House or presidential race ^[24][25], and most House seats were won by wide margins in 2024 ^[18]. Read as a reason to skip voting, it does not hold. The ballot carries more than the House race, and smaller contests are decided by small margins far more often ^[25]. Gelman and his coauthors also argue that even odds of 1 in 10 million, set against the stakes of a national election, “can provide a rational reason for voting” ^[24].

For one proposal to change how races like these are run, see Ranked-Choice Voting Explained: Why It Can Reduce Extremism.

What would change the verdict: If every race on a voter's ballot, not only the House race, were uncontested or decided by a wide margin, the second half would hold for that voter.

Redistricting

The House elected in 2024 had 220 Republicans and 215 Democrats ^[1], and 218 seats make a majority. As of Oct. 26, 2026, it had 218 Republicans, 214 Democrats, 1 independent and 2 vacancies ^[61]. Since the 2024 election, eight states have enacted or adopted new congressional maps for 2026, and a Supreme Court stay let Alabama use the congressional map its legislature enacted in 2023 ^[47].

To compare them, we count each map’s seat target: the number of districts it was drawn to move from one party’s column to the other’s, judged by 2024 presidential results by precinct (Texas, California and Florida) or as stated for the state (Ohio’s uses its redistricting commission’s index of statewide elections). It measures what the map-drawers intended, not what will happen. The per-state figures are Ballotpedia’s estimates ^[54][56][57]. By that count, maps in Texas (5), Florida (4), North Carolina, Tennessee, Louisiana and Alabama (1 each) were drawn to shift 13 seats toward Republicans, and Ohio’s up to 2 more. Maps in California (5) and Utah (1) were drawn to shift 6 toward Democrats. The net is 7 to 9 seats toward Republicans ^[54][56][57].

We checked three of the estimates against official files. On Texas’s new map, five districts Democrats won in 2024 moved at least 3 points toward Republicans in the 2024 presidential vote: three changed from Harris-won to Trump-won, and two that Trump had already won moved 3 to 6 points further his way ^[18][28][29]. That is why Trump-won districts rose by 3, from 27 to 30 of 38, while the seat target is 5 ^[28][29]. North Carolina’s 1st District, won by a Democrat in 2024 ^[18], was a narrow Biden win in 2020 on the old lines ^[70] and a Trump win in 2020 on the new ones ^[36]. A Utah court found that the map it adopted has one Democratic-leaning district ^[39], in a state whose four seats Republicans all won in 2024 ^[18]. California, Florida, Ohio, Tennessee, Louisiana and Alabama rest on Ballotpedia’s estimates alone.

Louisiana has one open question. Its legislature enacted the map above on May 29, and the act applies the new districts to the 2026 congressional elections ^[45][71]. A Sept. 5 Ballotpedia report said the state would use its 2022 map, without mentioning the new act ^[59]. The Secretary of State’s election pages we read do not name a map ^[72], so we treat the map in use on Nov. 3 as unconfirmed as of Oct. 26, 2026. The 2022 map, like the new one, had one majority-Black district where the 2024 map had two ^[57][59], so the table’s figure of 1 seat would likely hold under either.

Three of the maps, Tennessee’s, Louisiana’s and Alabama’s, were passed or cleared for use after the Supreme Court’s April 29, 2026 decision in Louisiana v. Callais ^[26]; Florida’s legislature passed its map the same day ^[41]. Writing for six justices, Justice Alito held that Section 2 of the Voting Rights Act “imposes liability only when the evidence supports a strong inference that the State intentionally drew its districts to afford minority voters less opportunity because of their race” ^[26]. Justices Kagan, Sotomayor and Jackson dissented ^[26].

Other attempts did not produce a 2026 map. The Supreme Court said on Sept. 25, 2026 that lower courts “should not order or otherwise require Missouri to use the 2025 map” ^[48]. Virginia voted on a redistricting amendment on April 21, 2026 ^[49][55]. On May 8 the Virginia Supreme Court held that the legislature’s process for proposing it violated the state constitution, which left the court’s 2021 maps in place for 2026 ^[74], and the U.S. Supreme Court denied a stay on May 15 ^[50]. Indiana’s Senate voted down a new map on Dec. 11, 2025 ^[51], and South Carolina’s effort died ^[57]. Maryland, Colorado and Georgia have no new map for 2026 ^[60].

On the side of the belief, 9 seats, the top of that range, is more than the whole swing in 5 of the 20 midterms from 1946 to 2022 (1962, 1986, 1990, 1998 and 2002) and equal to it in 2022 ^[1]. Maps also last. Ohio’s is set “for the rest of the decade” ^[38], and California’s runs through 2030 ^[54]. On the other side, the swing was larger than 9 seats in the other 14 midterms ^[1]. By one map-wide measure, the 2026 map leans no more Republican than 2018’s: Kyle Kondik of the University of Virginia Center for Politics puts the median district 3.4 points right of the nation, against 4.3 in 2018 ^[69]. Redrawing also ran both ways, with California and Utah moving seats toward Democrats ^[32][39], Virginia’s amendment voided ^[74], and efforts in Indiana and South Carolina failing ^[51][57].

How independent commissions draw maps, and how New York’s works, is covered in Independent Redistricting Commissions: What They Change (and What They Don’t).

What would change the verdict: The 2026 results compared with the national House vote, after the election, would show how much the new maps mattered.

Midterm turnout

Among eligible voters, turnout in every midterm from 1950 to 2022 was lower than in the presidential elections two years before and two years after, by about 16 points on average ^[62]. In 2014, 36.7% of eligible voters cast a ballot. In 2018 it was 50.1%, and in 2022, 46.1% ^[62]. The two most recent presidential elections drew 66.4% in 2020 and 64.3% in 2024, a figure that is still provisional ^[62].

Midterm yearsPresidential years

Data table

date	Midterm years	Presidential years
1948-01-01		52.2
1950-01-01	43.6
1952-01-01		62.3
1954-01-01	43.5
1956-01-01		60.2
1958-01-01	45
1960-01-01		63.8
1962-01-01	47.7
1964-01-01		62.8
1966-01-01	48.7
1968-01-01		62.5
1970-01-01	47.3
1972-01-01		56.2
1974-01-01	39.1
1976-01-01		54.8
1978-01-01	39
1980-01-01		54.2
1982-01-01	42
1984-01-01		55.2
1986-01-01	38.1
1988-01-01		52.8
1990-01-01	38.4
1992-01-01		58.1
1994-01-01	41.1
1996-01-01		51.7
1998-01-01	38.1
2000-01-01		54.2
2002-01-01	39.5
2004-01-01		60.1
2006-01-01	40.4
2008-01-01		61.6
2010-01-01	41
2012-01-01		58.6
2014-01-01	36.7
2016-01-01		60.1
2018-01-01	50.05
2020-01-01		66.38
2022-01-01	46.12
2024-01-01		64.3

Source: UF Election Lab (Michael McDonald), National VEP Turnout Rates, 1789-Present (v1.4). US_VEP_Turnout_Rates_v1.4.csv, columns TURNOUT_RATE_PRES and TURNOUT_RATE_MIDTERM · retrieved Oct. 20, 2026 · 39 rows · as of 2024 · annual. Download CSV

The Census Bureau’s survey tells the same story. In its series of reported voting, which starts in 1964 and counts all adults of voting age, citizens or not, the share who said they voted was lower in every midterm from 1966 to 2022 than in the presidential elections before and after it ^[75]. For citizens, its numbers run higher than the Election Lab’s. “Fifty-three percent of the citizen voting-age population voted in 2018,” the Bureau reported, a “historic 11 percentage point increase” and the highest midterm turnout in four decades ^[63]. Turnout among 18- to 29-year-olds rose from 20% to 36% ^[63]. In 2022, 52.2% voted, the “second highest for a nonpresidential election year since 2000” ^[64], and below 2018’s 53.4% ^[65]. Pew described both years as “unusually high turnout levels, with rates not seen since the 1960s” ^[9]. By the Election Lab’s count, that fits 2018, the highest midterm rate in the chart, while 2022’s 46.1% was below the 1962, 1966 and 1970 midterms and second only to 2018 since then ^[62].

The belief is right that every midterm since 1950 drew fewer voters than the presidential elections on either side of it ^[62]. The smallest gap was 8.6 points, in 1950 against 1948, and the highest midterm, 2018 at 50.1%, was still below the lowest presidential year in the chart, 1996 at 51.7% ^[62]. “Low” has no fixed line in absolute terms, so we grade the comparison the belief makes, with presidential years, and it holds in every case. Midterm turnout still varies from one midterm to the next. Against other midterms, 2018 was a four-decade high, with an 11-point jump ^[63], and 2018 and 2022 were the two highest midterms after 1970 ^[62]. Why turnout rose is less clear. We found no peer-reviewed explanation, and the Pew studies above describe who voted ^[10][11], not why.

What would change the verdict: Nothing in 2026 changes the record from 1950 to 2022, which is what this verdict grades. A midterm that matched or beat the presidential election before or after it would be the first exception since 1950. For 2026, the 2024 rate to match is 64.3%, still provisional.

Your state on Nov. 3

Scroll sideways to see the whole map. The table below lists every state.

• AK54.6%Alaska: 54.6%, plus or minus 2.7 points reported turnout in 2022 (survey)
• ME63.8%Maine: 63.8%, plus or minus 3.6 points reported turnout in 2022 (survey)
• WI60.9%Wisconsin: 60.9%, plus or minus 2.9 points reported turnout in 2022 (survey)
• VT62.2%Vermont: 62.2%, plus or minus 3.4 points reported turnout in 2022 (survey)
• NH59.1%New Hampshire: 59.1%, plus or minus 3.3 points reported turnout in 2022 (survey)
• WA59.7%Washington: 59.7%, plus or minus 2.8 points reported turnout in 2022 (survey)
• ID48.3%Idaho: 48.3%, plus or minus 2.4 points reported turnout in 2022 (survey)
• MT56.5%Montana: 56.5%, plus or minus 3.6 points reported turnout in 2022 (survey)
• ND52.0%North Dakota: 52.0%, plus or minus 3.0 points reported turnout in 2022 (survey)
• MN63.7%Minnesota: 63.7%, plus or minus 4.0 points reported turnout in 2022 (survey)
• IL52.1%Illinois: 52.1%, plus or minus 2.8 points reported turnout in 2022 (survey)
• MI63.3%Michigan: 63.3%, plus or minus 2.9 points reported turnout in 2022 (survey)
• NY49.1%New York: 49.1%, plus or minus 2.5 points reported turnout in 2022 (survey)
• MA56.7%Massachusetts: 56.7%, plus or minus 2.8 points reported turnout in 2022 (survey)
• OR70.0%Oregon: 70.0%, plus or minus 3.2 points reported turnout in 2022 (survey)
• NV50.9%Nevada: 50.9%, plus or minus 3.5 points reported turnout in 2022 (survey)
• WY49.0%Wyoming: 49.0%, plus or minus 3.5 points reported turnout in 2022 (survey)
• SD53.4%South Dakota: 53.4%, plus or minus 4.0 points reported turnout in 2022 (survey)
• IA51.8%Iowa: 51.8%, plus or minus 3.7 points reported turnout in 2022 (survey)
• IN41.9%Indiana: 41.9%, plus or minus 2.8 points reported turnout in 2022 (survey)
• OH47.8%Ohio: 47.8%, plus or minus 2.7 points reported turnout in 2022 (survey)
• PA60.0%Pennsylvania: 60.0%, plus or minus 2.4 points reported turnout in 2022 (survey)
• NJ50.5%New Jersey: 50.5%, plus or minus 3.3 points reported turnout in 2022 (survey)
• CT49.6%Connecticut: 49.6%, plus or minus 4.2 points reported turnout in 2022 (survey)
• RI54.4%Rhode Island: 54.4%, plus or minus 3.7 points reported turnout in 2022 (survey)
• CA51.5%California: 51.5%, plus or minus 1.6 points reported turnout in 2022 (survey)
• UT52.8%Utah: 52.8%, plus or minus 3.2 points reported turnout in 2022 (survey)
• CO61.3%Colorado: 61.3%, plus or minus 4.1 points reported turnout in 2022 (survey)
• NE48.8%Nebraska: 48.8%, plus or minus 3.3 points reported turnout in 2022 (survey)
• MO52.9%Missouri: 52.9%, plus or minus 3.2 points reported turnout in 2022 (survey)
• KY52.3%Kentucky: 52.3%, plus or minus 6.0 points reported turnout in 2022 (survey)
• WV38.4%West Virginia: 38.4%, plus or minus 5.3 points reported turnout in 2022 (survey)
• VA53.2%Virginia: 53.2%, plus or minus 3.2 points reported turnout in 2022 (survey)
• DC62.9%District of Columbia: 62.9%, plus or minus 3.0 points reported turnout in 2022 (survey)
• MD57.9%Maryland: 57.9%, plus or minus 3.7 points reported turnout in 2022 (survey)
• DE54.2%Delaware: 54.2%, plus or minus 3.9 points reported turnout in 2022 (survey)
• AZ55.8%Arizona: 55.8%, plus or minus 3.3 points reported turnout in 2022 (survey)
• NM54.1%New Mexico: 54.1%, plus or minus 3.2 points reported turnout in 2022 (survey)
• KS59.4%Kansas: 59.4%, plus or minus 4.0 points reported turnout in 2022 (survey)
• AR43.9%Arkansas: 43.9%, plus or minus 2.8 points reported turnout in 2022 (survey)
• TN44.5%Tennessee: 44.5%, plus or minus 2.7 points reported turnout in 2022 (survey)
• NC45.7%North Carolina: 45.7%, plus or minus 2.7 points reported turnout in 2022 (survey)
• SC44.9%South Carolina: 44.9%, plus or minus 3.5 points reported turnout in 2022 (survey)
• OK47.0%Oklahoma: 47.0%, plus or minus 2.7 points reported turnout in 2022 (survey)
• LA48.2%Louisiana: 48.2%, plus or minus 3.8 points reported turnout in 2022 (survey)
• MS46.4%Mississippi: 46.4%, plus or minus 2.6 points reported turnout in 2022 (survey)
• AL45.4%Alabama: 45.4%, plus or minus 3.0 points reported turnout in 2022 (survey)
• GA56.9%Georgia: 56.9%, plus or minus 3.6 points reported turnout in 2022 (survey)
• HI50.0%Hawaii: 50.0%, plus or minus 3.7 points reported turnout in 2022 (survey)
• TX47.0%Texas: 47.0%, plus or minus 2.1 points reported turnout in 2022 (survey)
• FL49.0%Florida: 49.0%, plus or minus 2.1 points reported turnout in 2022 (survey)

Hover over or focus a state to read its 2022 reported turnout. Select a state for what is on its ballot.District of Columbia: 62.9%, plus or minus 3.0 points reported turnout in 2022 (survey)

Reported turnout, 2022 midterm (survey), percent of citizens 18 and older

38%48%51%54%60%70%

• Step 1 of 5, lightest: 38% to under 48%
• Step 2 of 5, shade: 48% to under 51%
• Step 3 of 5, shade: 51% to under 54%
• Step 4 of 5, shade: 54% to under 60%
• Step 5 of 5, strongest: 60% to 70%

United States: 52.2% ±0.5. Margins of error run from 1.6 to 6.0 points, so many neighboring states are not statistically different; the shading is not a ranking.

Every state and D.C., A to Z. Turnout is what people reported in a Census survey; ± is its margin of error in percentage points (90 percent confidence). The table scrolls sideways; the state's name stays at the left.

What is on each state's ballot on Nov. 3, 2026, reported turnout (survey) and election offices, A to Z

State	U.S. House seats (all on the ballot)	U.S. Senate race	Governor's race	House map changed since 2024	Reported turnout, 2022 midterm (survey)	Reported turnout, 2024 (survey)	How to vote	Notes
Alabama	7	Yes, one seat	Yes	Yes	45.4% ±3.0	58.7% ±4.5	Alabama election office	The Legislature's 2023 map is in use, in place of the court-ordered map used in 2024, after the U.S. Supreme Court stayed a district court injunction on June 2, 2026.
Alaska	1	Yes, one seat	Yes	No	54.6% ±2.7	62.1% ±4.1	Alaska election office
Arizona	9	No	Yes	No	55.8% ±3.3	61.7% ±3.2	Arizona election office
Arkansas	4	Yes, one seat	Yes	No	43.9% ±2.8	52.8% ±2.7	Arkansas election office
California	52	No	Yes	Yes	51.5% ±1.6	64.7% ±1.6	California election office	New map enacted Nov. 4, 2025 by voters, who approved Proposition 50, a legislative constitutional amendment; it applies through 2030.
Colorado	8	Yes, one seat	Yes	No	61.3% ±4.1	70.1% ±5.0	Colorado election office
Connecticut	5	No	Yes	No	49.6% ±4.2	66.3% ±3.4	Connecticut election office
Delaware	1	Yes, one seat	No	No	54.2% ±3.9	69.8% ±3.7	Delaware election office
District of Columbia	None; D.C. elects a nonvoting delegate	No; D.C. has no senators	No; D.C. has a mayor, not a governor	Not applicable; no House districts	62.9% ±3.0	79.5% ±3.3	District of Columbia election office	D.C. is not apportioned House seats and has no senators; it elects a nonvoting delegate to the House on Nov. 3, 2026, and a mayor, not a governor.
Florida	28	Yes, one seat (a special election)	Yes	Yes	49.0% ±2.1	60.3% ±1.9	Florida election office	Special election for the Class III seat (term ends Jan. 3, 2029) held by an appointed senator. New map (HB 1-D, ch. 2026-229) enacted May 4, 2026 by the Legislature in a special session, approved by the governor. Challenges are pending in state court.
Georgia	14	Yes, one seat	Yes	No	56.9% ±3.6	64.4% ±2.6	Georgia election office
Hawaii	2	No	Yes	No	50.0% ±3.7	58.5% ±3.4	Hawaii election office
Idaho	2	Yes, one seat	Yes	No	48.3% ±2.4	65.2% ±2.2	Idaho election office
Illinois	17	Yes, one seat	Yes	No	52.1% ±2.8	65.9% ±3.0	Illinois election office
Indiana	9	No	No	No	41.9% ±2.8	60.7% ±3.4	Indiana election office
Iowa	4	Yes, one seat	Yes	No	51.8% ±3.7	71.8% ±4.4	Iowa election office
Kansas	4	Yes, one seat	Yes	No	59.4% ±4.0	70.7% ±5.6	Kansas election office
Kentucky	6	Yes, one seat	No	No	52.3% ±6.0	67.7% ±3.6	Kentucky election office
Louisiana	6	Yes, one seat	No	Yes	48.2% ±3.8	58.0% ±3.6	Louisiana election office	The map used in 2024 (SB8) was struck down by the U.S. Supreme Court in Louisiana v. Callais on April 29, 2026. Which map applies on Nov. 3, 2026, the one enacted May 29, 2026 by the Legislature (SB 121, Act 2 of 2026) or the 2022 map, was not confirmed in an official source as of Oct. 19, 2026.
Maine	2	Yes, one seat	Yes	No	63.8% ±3.6	67.1% ±3.5	Maine election office
Maryland	8	No	Yes	No	57.9% ±3.7	71.1% ±3.6	Maryland election office
Massachusetts	9	Yes, one seat	Yes	No	56.7% ±2.8	68.9% ±3.1	Massachusetts election office
Michigan	13	Yes, one seat	Yes	No	63.3% ±2.9	72.3% ±2.9	Michigan election office
Minnesota	8	Yes, one seat	Yes	No	63.7% ±4.0	75.9% ±2.9	Minnesota election office
Mississippi	4	Yes, one seat	No	No	46.4% ±2.6	68.9% ±2.8	Mississippi election office
Missouri	8	No	No	No	52.9% ±3.2	69.5% ±3.3	Missouri election office	A map enacted by the Legislature in 2025 is not in use for 2026: on Sept. 25, 2026 the U.S. Supreme Court said lower courts should not require it, so the 2022 map applies. Voters decide a referendum on the 2025 map on Nov. 3, 2026.
Montana	2	Yes, one seat	No	No	56.5% ±3.6	69.2% ±2.5	Montana election office
Nebraska	3	Yes, one seat	Yes	No	48.8% ±3.3	67.4% ±4.0	Nebraska election office
Nevada	4	No	Yes	No	50.9% ±3.5	66.9% ±2.9	Nevada election office
New Hampshire	2	Yes, one seat	Yes	No	59.1% ±3.3	71.9% ±3.4	New Hampshire election office
New Jersey	12	Yes, one seat	No	No	50.5% ±3.3	72.5% ±2.8	New Jersey election office
New Mexico	3	Yes, one seat	Yes	No	54.1% ±3.2	66.0% ±3.7	New Mexico election office
New York	26	No	Yes	No	49.1% ±2.5	59.2% ±2.6	New York election office
North Carolina	14	Yes, one seat	No	Yes	45.7% ±2.7	64.6% ±2.7	North Carolina election office	New map (S.L. 2025-95) enacted Oct. 22, 2025 by the General Assembly.
North Dakota	1	No	No	No	52.0% ±3.0	71.2% ±2.7	North Dakota election office
Ohio	15	Yes, one seat (a special election)	Yes	Yes	47.8% ±2.7	67.4% ±2.5	Ohio election office	Special election for the Class III seat (term ends Jan. 3, 2029) held by an appointed senator. New map adopted Oct. 31, 2025 by the Ohio Redistricting Commission, by a unanimous vote.
Oklahoma	5	Yes, one seat	Yes	No	47.0% ±2.7	59.7% ±3.1	Oklahoma election office	An appointed senator holds this Class II seat; whether a separate special election for the rest of the current term (to Jan. 3, 2027) is also on the ballot is not confirmed from a state source.
Oregon	6	Yes, one seat	Yes	No	70.0% ±3.2	75.3% ±2.8	Oregon election office
Pennsylvania	17	No	Yes	No	60.0% ±2.4	69.7% ±2.3	Pennsylvania election office
Rhode Island	2	Yes, one seat	Yes	No	54.4% ±3.7	70.0% ±4.0	Rhode Island election office
South Carolina	7	Yes, one seat	Yes	No	44.9% ±3.5	62.7% ±3.2	South Carolina election office	An appointed senator holds this Class II seat; whether a separate special election for the rest of the current term (to Jan. 3, 2027) is also on the ballot is not confirmed from a state source.
South Dakota	1	Yes, one seat	Yes	No	53.4% ±4.0	58.9% ±3.8	South Dakota election office
Tennessee	9	Yes, one seat	Yes	Yes	44.5% ±2.7	64.8% ±2.6	Tennessee election office	New map (HB 7003, Public Chapter 3) passed May 7, 2026 by the General Assembly in an extraordinary session.
Texas	38	Yes, one seat	Yes	Yes	47.0% ±2.1	57.9% ±2.3	Texas election office	New map (PLANC2333) enacted Aug. 29, 2025 by the Legislature (HB 4, signed by the governor). The U.S. Supreme Court stayed a lower-court injunction against it on Dec. 4, 2025.
Utah	4	No	No	Yes	52.8% ±3.2	66.9% ±3.1	Utah election office	New map adopted Nov. 10, 2025 by the Third District Court, which barred the Legislature's map; it governs at least the 2026 elections.
Vermont	1	No	Yes	No	62.2% ±3.4	71.7% ±3.1	Vermont election office
Virginia	11	Yes, one seat	No	No	53.2% ±3.2	72.9% ±2.4	Virginia election office	A map tied to a constitutional amendment that voters approved April 21, 2026 is not in use for 2026: the Supreme Court of Virginia voided the amendment on May 8, 2026.
Washington	10	No	No	No	59.7% ±2.8	70.0% ±4.0	Washington election office
West Virginia	2	Yes, one seat	No	No	38.4% ±5.3	59.5% ±5.5	West Virginia election office
Wisconsin	8	No	Yes	No	60.9% ±2.9	72.2% ±3.4	Wisconsin election office
Wyoming	1	Yes, one seat	Yes	No	49.0% ±3.5	63.8% ±2.4	Wyoming election office

Data table

The table above is this figure's data table: every state and D.C., with the ballot facts and both turnout figures. Download the CSV for the ballot file. United States, reported turnout (survey): 52.2% ±0.5 in 2022, 65.3% ±0.5 in 2024.

The Census Bureau on its survey: “Voting estimates from the CPS and other sample surveys have historically differed from those based on administrative records, such as the official reports from each state disseminated collectively by the Clerk of the U.S. House of Representatives and the Federal Elections Commission. In general, sample surveys like the CPS tend to yield higher voting rates than official results. Potential explanations for these differences include question misreporting, problems with memory or knowledge of others' behavior, and methodological issues related to question wording, method of survey administration, and survey nonresponse bias.” (U.S. Census Bureau, Voting and Registration in the Election of November 2022, P20-586 (April 2024), p. 2)

Source: U.S. Census Bureau (2020 apportionment); U.S. Senate (senate.gov); USA.gov; our list of governor races, checked against the National Governors Association. Census Table 1, Apportionment Population and Number of Representatives by State: 2020 Census; senate.gov Class II, Class III and Appointed Senators pages; USA.gov state election office directory · retrieved Oct. 20, 2026 · 51 rows · as of Oct. 20, 2026 · one-time report. Reported turnout: U.S. Census Bureau, Current Population Survey, November 2022 Voting and Registration Supplement, Table 4a; U.S. Census Bureau, Current Population Survey, November 2024 Voting and Registration Supplement, Table 4a. Download CSV

Each state’s House seats come from the 2020 census apportionment ^[66]. Turnout in the panel is what people told the Census Bureau’s survey, for 2022 ^[67] and for 2024 ^[76], not official vote counts.

To check your registration, find your polling place or learn how to vote by mail where you live, start at vote.gov or the U.S. Election Assistance Commission’s Register and Vote in Your State page. Rules and deadlines differ by state, so confirm them with your state’s election office; USA.gov lists every one ^[68].

Two states need a note. In Louisiana, Nov. 3 is the open primary for the U.S. House and Dec. 12, 2026 is the open general election ^[72]. A candidate who wins a majority on Nov. 3 is elected; otherwise the race goes to a second round on Dec. 12 ^[58][59]. Missouri votes under its 2022 congressional map ^[48]. Its voters also decide Proposition A, a referendum on the map the legislature passed in 2025: a yes vote approves that map, and a no vote keeps the districts enacted in 2022 ^[73].

How we checked these claims

We grade each belief as worded, literally, and give it one of three verdicts: Supported, Mixed or Not supported. Each is defined in How we rate a belief on our methods page. A belief that says or implies “always” is tested against every case it covers, so a single exception keeps it from Supported. A belief made of a premise and a conclusion is graded as a whole, so a true premise with an unsupported conclusion is Mixed. A verdict is about the evidence, not about any party, candidate or person who holds or repeats the belief.

House seat changes are our arithmetic from the Office of the Historian’s party divisions, which count each Congress “as of the initial election results” ^[1]: seats after each midterm minus seats before. Across all 20 midterms from 1946 to 2022, gains included, the mean change was a loss of 25.9 seats and the median a loss of 22; the 18 losses alone have a mean of 29.4 seats and a median of 27. The American Presidency Project’s figures, which we use only as a check, differ from these by 1 or 2 seats in 1958, 1962, 1978, 1986, 2010 and 2018 and give the same median ^[3]. Turnout is the University of Florida Election Lab’s rate for the voting-eligible population ^[62]; the chart starts in 1948, the first year its data carry an open license. The Census Bureau’s figures ^[63][64][65] come from a survey of citizens and run higher, so we do not put the two in one chart. The Bureau’s series from 1964 ^[75], which counts all adults of voting age, is used only to check that each midterm trailed the presidential years around it, not in any chart. House margins are our count of all 435 races in 2024 from the Clerk of the House’s official statistics ^[18]: the winner’s votes minus the runner-up’s, as a share of all votes for candidates in the race. Eleven races fall within 0.3 points of the 5-, 10- or 20-point line, so another way of measuring could move a few seats between groups. The MIT Election Data and Science Lab publishes House returns back to 1976 ^[17]; we did not use them here.

Each state’s seat target is Ballotpedia’s estimate ^{[52][53][54][55][56][57]}. We checked Texas’s, North Carolina’s and Utah’s against official district results and a court order ^{[18][28][29][36][39][70]}; the other six rest on Ballotpedia alone. We added them: 5 + 4 + 1 + 1 + 1 + 1 = 13 toward Republicans, plus 0 to 2 in Ohio, where Ballotpedia says the map “could make two districts more competitive for Republicans” ^[54], and 5 + 1 = 6 toward Democrats, for a net of 7 to 9. Ballotpedia’s own running totals count Ohio as 2 ^[54].

We left out polls, race ratings and forecasts, because this post checks beliefs against what has already happened, and Nov. 3 will settle the rest. We also left out presidential approval figures: the approval series we found may not be republished, and none of the six claims needs one.