Case

Can Independents Vote in Primaries, and Why Does It Matter?

In most states, yes, with conditions. In about a fifth of states, no. It matters because in most House districts the primary is the election that actually decides the seat, and a small fraction of voters decide it. Here is how primaries work, what opening them does and does not change, and one way to let every voter into the election that counts.

Updated September 13, 2026 · 5 min read · An independent project; the site’s own view is marked where it appears.

The short answer

It depends on the state. By the National Conference of State Legislatures' count:¹

Eight states say no. Only registered party members may vote: Delaware, Florida, Kentucky, New Jersey, New York, Pennsylvania, Tennessee, and Wyoming.
Nine leave it to the parties, which decide each cycle whether to let unaffiliated voters in.
Eight let independents pick a party's ballot, though a registered Democrat or Republican cannot cross over.
Twenty let any voter choose either ballot on primary day.
Four skip party primaries altogether. Alaska, California, Washington, and Nebraska put every candidate on one ballot.

The list moves. Louisiana, which for decades ran a single all-candidate election in November, switched in 2026 to closed party primaries for Congress, its Supreme Court, and two state boards; the first was held May 16, 2026. Voters registered with no party may pick either major party's ballot; voters registered Green, Libertarian, or with another party may not vote in those races at all.² ³ Unite America, which advocates for open primaries, counts about 17.6 million registered independents who were locked out of a 2024 primary in the closed-primary states; treat that as an advocate's tally, but the mechanism behind it is not in dispute.⁴

Why the primary matters more than November

Because most November races are not close. In 2024, only 37 of the 435 House races were decided by fewer than five points, and 304 were decided by more than fifteen; the average margin was 27 points, and twelve states did not have a single competitive House race.⁵ ⁶ For 2026, the Cook Political Report rates 21 seats as toss-ups.⁸ In a district like that, the general election ratifies a choice already made. The real contest was the primary of whichever party is safe there, and turnout in primaries is low: 21.3 percent of eligible voters in 2022, by the Bipartisan Policy Center's count.⁷ Unite America's estimate, using its own definition of a seat decided in the primary, is that 380 of 435 House seats in 2024 were settled by about 18 million primary voters, 7 percent of voting-age Americans.⁴

That is the whole reason the rules matter. A member of Congress from a safe district answers to the people who can vote in the primary and do. In a closed-primary state, that is a fraction of one party's registered members. Everyone else, including every independent in the district, votes in November on a decision that was made in the spring. That is true whichever party holds the seat.

The systems, and where each one breaks

States have tried several ways to open the deciding election. California once let every voter pick any candidate for each office from a single list; the Supreme Court struck that "blanket primary" down in 2000, 7 to 2, as a violation of the parties' right of association.⁹ Washington and then California answered with a top-two primary: all candidates on one ballot, no party nomination at all, the two highest finishers advancing regardless of party. The Court upheld that design in 2008, also 7 to 2.¹⁰

Top-two has a known failure. In California's 31st District in June 2012, four Democrats split 48.5 percent of the vote among them while two Republicans took 26.7 and 24.8 percent, so both November slots went to Republicans in a district Democrats had the votes to win; the general election was Republican against Republican.¹¹ Once that risk exists, parties clear their fields early and quietly, which moves the real decision even further from voters. Alaska tried top-four with ranked-choice in the general election. In its 2022 special House election, Nick Begich was eliminated in the first round with 28.5 percent, and Mary Peltola beat Sarah Palin 51.5 to 48.5 on the transfers; the ballots themselves showed Begich would have beaten either opponent head-to-head.¹² ¹³ Such outcomes are rare in the ranked-choice record, but that one happened in a federal race. Approval voting, in which voters mark every candidate they find acceptable, has been used for St. Louis city primaries since 2021, with the top two advancing; Fargo used it from 2020 until North Dakota banned both approval and ranked-choice voting in 2025.¹⁴ ¹⁵

What the research says, honestly

Reformers often promise that open primaries produce more moderate officials. The evidence for that is mixed at best. The most cited study, covering state legislators across every primary system, found that "the openness of a primary election has little, if any, effect on the extremism of the politicians it produces."¹⁶ A later look at California's and Washington's top-two found "a modest and somewhat inconsistent effect on representation."¹⁷ One 2020 study did find top-two primaries associated with less extreme legislators.¹⁸ The participation effect is on firmer ground: in the 2022 primaries, states with fully open, top-two, or top-four systems averaged 24.5 percent turnout, against 21.5 percent for semi-open and 20.7 percent for closed states.⁷ This site's case for opening primaries rests on who gets to vote in the deciding election, not on a promise about who wins it.

What could change

Where Common Sense Again comes in

One open primary ballot for every voter

Every voter should be able to vote in the election that actually decides who represents them. In most districts that election is the primary, and in too many states most voters cannot enter it. The common-sense fix is one open ballot for everyone.

This site proposes that every qualified candidate appear on one primary ballot open to every eligible voter regardless of party registration; that voters mark every candidate they would accept; and that the two with the most approvals advance to a head-to-head general election that produces a majority winner. The permanent commitment is four tests, not the mechanism: every eligible voter can take part in the election that decides the seat; supporting one acceptable candidate can never hurt another you would also accept; the final winner holds real majority support head-to-head; and the ballot is understandable in line, countable on existing equipment, and auditable by hand. A pick-one top-two ballot fails the second test, which is what happened in California's 31st District. Ranked choice is not the recommendation because of what the Alaska ballots showed.²⁰

The honest limits: St. Louis is the only American jurisdiction using an approval primary, so the record at federal scale is nonexistent. Approval changes strategic voting rather than eliminating it; a campaign can tell its supporters to approve only one name. And critics in St. Louis argue approval can dilute a cohesive minority bloc, an objection this site treats as serious and not yet fully answered. If a better method meets the four tests more completely, it should be used instead.

Read the repair and its evidence →

Go further

A closed primary in a safe district is where this site's argument starts: the election that decides the seat is the one most voters never enter, and the winner answers to the few who did. The three-minute case connects that to the maps, the money, and the rules that keep fights alive instead of settling them.

Read the three-minute case →

The repair in full → the four tests, the approval ballot, and why ranked choice is not the pick.
What is gerrymandering? → why so many districts are safe in the first place.
The scorecard → measure 1 counts states with both an open primary and formula-drawn maps; today that is zero.

Share this caseEmail Post on X

Sources

Primary records first. Where a source has a known lean or a limit, the note says so. Records marked with an S-number also appear in the site’s source register.

1.PRIMARYNational Conference of State Legislatures, “State Primary Election Types,” updated May 19, 2026. Closed: Delaware, Florida, Kentucky, New Jersey, New York, Pennsylvania, Tennessee, Wyoming. Partially closed (party discretion on unaffiliated voters): Connecticut, Idaho, Kansas, Maryland, Oklahoma, Oregon, South Dakota, Utah, West Virginia. Open to unaffiliated voters only: Arizona, Colorado, Maine, Massachusetts, New Hampshire, New Mexico, North Carolina, Rhode Island. Partially open: Illinois, Indiana, Iowa, Nevada, Ohio. Open: Alabama, Arkansas, Georgia, Hawaii, Michigan, Minnesota, Mississippi, Missouri, Montana, North Dakota, South Carolina, Texas, Vermont, Virginia, Wisconsin. Multi-party: Alaska (top-four), California and Washington (top-two), Nebraska (nonpartisan legislature), Louisiana (listed as all-comers). ncsl.org →
NCSL's table still lists Louisiana under its old all-candidate system; the state's own election office is authoritative on the 2026 change (source 2).
2.PRIMARYLouisiana Secretary of State, “Closed Party Primary Elections”: starting in May 2026, U.S. House, U.S. Senate, the Louisiana Supreme Court, the Public Service Commission, and the Board of Elementary and Secondary Education fall under closed party primaries; voters registered “No Party” may choose the Democratic or Republican ballot or sit those races out; voters registered Green, Libertarian, or Other Party may not participate in closed-party primary races. sos.la.gov →
3.REPORTINGLouisiana Illuminator, “Party primary primer: What Louisiana voters need to know for the May 16 election,” April 20, 2026: candidates of the same party compete May 16, party runoffs June 27, and the winners advance to the November 3 general election; the change applies only to U.S. Senate, U.S. House, the state Supreme Court, the Public Service Commission, and BESE. lailluminator.com →
4.RESEARCHUnite America, “The Primary Problem 2024”: in the 2024 primaries, 380 of 435 U.S. House seats (87 percent) were effectively decided by 18.1 million voters, 7 percent of 260 million voting-age Americans; the organization counts a seat as decided in the primary when the general election is not competitive; it also reports 17,592,955 independent voters locked out of closed primaries in 2024. uniteamerica.org →
Unite America advocates for nonpartisan primaries; its “decided in the primary” figure depends on its own competitiveness threshold. Cited with that label, alongside the neutral margin data in sources 5 and 8.
5.RESEARCHBallotpedia, “Election results, 2024: Congressional margin of victory analysis”: across 435 U.S. House elections the average margin was 27.3 percentage points, the narrowest since Ballotpedia began the series in 2012; by margin, 37 House races were decided by under 5 points, 74 by 5 to 15 points, and 304 by more than 15 points (Democratic and Republican winners combined). ballotpedia.org →
6.RESEARCHFairVote, “Dubious Democracy 2024”: 84 percent of 2024 House elections were uncontested or won by more than 10 points and 66 percent (285 races) by more than 20; only 37 of 435 were decided by fewer than 5 points; twelve states (Hawaii, Idaho, Kentucky, Louisiana, Mississippi, North Dakota, Oklahoma, South Dakota, Utah, Vermont, West Virginia, Wyoming) had no competitive House election. fairvote.org →
FairVote advocates ranked-choice voting. Its margin counts match Ballotpedia's where they overlap.
7.RESEARCHBipartisan Policy Center, “2022 Primary Turnout: Trends and Lessons for Boosting Participation,” March 2023 (read from the report PDF): turnout of all eligible voters in the 2022 primaries was 21.3 percent; states with fully open, top-two, or top-four primaries averaged 24.5 percent, semi-open states 21.5 percent, and closed states 20.7 percent; over the past four midterm cycles, fully open states averaged 21.9 percent against 18.5 percent for closed states. bipartisanpolicy.org →
A descriptive comparison, not a causal estimate; the report itself notes that contested top-of-ticket races move turnout more than primary type does.
8.RESEARCHCook Political Report, 2026 House race ratings as of September 11, 2026, as compiled by 270toWin: 21 toss-ups of 435; 365 seats rated safe for one party. 270towin.com → S34
Analyst ratings that change through the cycle.
9.PRIMARYCalifornia Democratic Party v. Jones, 530 U.S. 567 (2000), 7–2, opinion by Justice Scalia: California's blanket primary, under which every voter could choose any candidate for each office regardless of party, violated the parties' First Amendment right of association. law.cornell.edu →
10.PRIMARYWashington State Grange v. Washington State Republican Party, 552 U.S. 442 (2008), 7–2, opinion by Justice Thomas: Washington's Initiative 872 top-two primary, in which candidates state a party preference and no party nominates, is facially constitutional. law.cornell.edu →
11.PRIMARYCalifornia Secretary of State, Statement of Vote, June 5, 2012 Presidential Primary Election, U.S. Representative District 31: Gary G. Miller (R) 16,708 (26.7%), Bob Dutton (R) 15,557 (24.8%), Pete Aguilar (D) 14,181 (22.6%), Justin Kim (D) 8,487 (13.5%), Renea Wickman (D) 4,188 (6.7%), Rita Ramirez-Dean (D) 3,546 (5.7%). Statement of Vote, November 6, 2012 General Election, District 31: Miller 88,964 (55.2%), Dutton 72,255 (44.8%). elections.cdn.sos.ca.gov →
12.PRIMARYAlaska Division of Elections, 2022 Special General Election, U.S. Representative, ranked-choice detailed report: Round 1, Peltola 75,799 (40.19%), Palin 58,973 (31.27%), Begich 53,810 (28.53%); Begich eliminated, with 27,053 of his ballots transferring to Palin and 15,467 to Peltola; Round 2, Peltola 91,266 (51.48%), Palin 86,026 (48.52%). elections.alaska.gov →
13.RESEARCHAdam Graham-Squire and David McCune, “Ranked Choice Voting and Condorcet Failure in the Alaska 2022 Special Election” (arXiv 2303.00108): “Nick Begich was eliminated in the first round despite being more broadly acceptable to the electorate than either of the other two candidates. More specifically, Begich was the ‘Condorcet winner’ of this election.” Their companion analysis of the same ballots (arXiv 2209.04764) documents the monotonicity and no-show paradoxes in that race. arxiv.org →
Computed from the cast-vote record. The authors' broader work finds such failures rare across American ranked-choice elections; this Case does not claim otherwise.
14.PRIMARYCity of St. Louis, Ordinance 71410 (Proposition D), approved by voters November 3, 2020: open, nonpartisan primaries for mayor, comptroller, president of the Board of Aldermen, and aldermen in which voters may choose all the candidates they wish, with the top two advancing to a runoff general election; first used in the March 2021 primary. stlouis-mo.gov →
15.PRIMARYNorth Dakota Legislative Assembly, House Bill 1297 (69th Assembly, 2025): “relating to prohibiting approval voting and ranked-choice voting in elections,” amending home-rule powers of counties and cities; signed by the governor April 18, 2025. ndlegis.gov →
16.RESEARCHEric McGhee, Seth Masket, Boris Shor, Steven Rogers, and Nolan McCarty, “A Primary Cause of Partisanship? Nomination Systems and Legislator Ideology,” American Journal of Political Science 58, no. 2 (April 2014): 337–351: “We find that the openness of a primary election has little, if any, effect on the extremism of the politicians it produces.” onlinelibrary.wiley.com →
Peer-reviewed; state legislators, all primary systems.
17.RESEARCHEric McGhee and Boris Shor, “Has the Top Two Primary Elected More Moderates?,” Perspectives on Politics (2017): the top-two primary in California and Washington had “a modest and somewhat inconsistent effect on representation,” with stronger evidence of Democratic moderation in California that the authors attribute partly to concurrent redistricting and term-limit changes. cambridge.org →
18.RESEARCHChristian R. Grose, “Reducing Legislative Polarization: Top-Two and Open Primaries Are Associated with More Moderate Legislators,” Journal of Political Institutions and Political Economy (2020), as summarized by USC, May 14, 2020: top-two primaries reduce legislator extremity by almost double the reduction associated with open primaries, over 2003–2018. today.usc.edu →
The minority position in this literature; cited so the reader sees both sides.
19.RESEARCHBallotpedia News, “Five states reject ranked-choice voting measures, Alaska retains system after recount,” December 10, 2024: Arizona Proposition 140, Colorado Proposition 131, Idaho Proposition 1, Nevada Question 3, and Oregon Measure 117 rejected; Missouri Amendment 7 (prohibiting ranked-choice voting) approved 68.4 to 31.6; Alaska's repeal measure defeated by 743 votes after a recount; District of Columbia Initiative 83 approved 72.9 to 27.1. news.ballotpedia.org →
20.PRIMARYCommon Sense Again, Repair 01, “One open primary ballot for every voter”: the four tests, the approval-ballot mechanism, its stated limits (St. Louis as the only U.S. jurisdiction; strategic voting changed rather than eliminated; the minority-bloc objection treated as unanswered), and the reasons ranked choice is not the recommendation. →

Published September 13, 2026. Last substantive update September 13, 2026. Load-bearing facts last re-verified against primary records September 13, 2026. This page is updated in place when the subject changes; it is not republished as a new article.