Post Context

The new playbook: a measurable partisan shift in conceding elections

Contested elections are as old as the republic. What's new is measurable: a specific pattern of explicit, evidence-free election denial has emerged almost entirely on one side of the aisle, at a rate with no real historical precedent. Bright Line Watch's study of concessions in federal races found that one in five losing Republican candidates in 2020 explicitly refused to concede — versus 2% of losing Democrats that year, and zero the next cycle. That gap is the trend.

The old version: rare, close, and eventually resolved

For most of American history, contesting a result was the exception — reserved for races so close that doubt was reasonable, and it ended with the loser stepping aside once the process ran its course.

1876 — Hayes v. Tilden. The closest call to a true constitutional crisis over a presidential result. Returns from Florida, Louisiana, South Carolina, and Oregon were disputed, with both parties submitting competing slates of electors. Congress created a 15-member Electoral Commission to resolve it, and the dispute wasn't settled until an informal bargain in early 1877 — months after Election Day, and only days before inauguration. This is the closest historical precedent for genuine institutional crisis, but it was a one-off product of Reconstruction-era chaos, not a repeatable playbook.

1960 — Nixon v. Kennedy. Nixon lost Illinois by roughly 9,000 votes and Texas by about 46,000, amid credible (if never fully proven) allegations of irregularities tied to Chicago's Democratic machine. Advisors urged him to demand recounts. Nixon declined, reportedly out of concern for the country's stability during the Cold War. He conceded the next day. Whatever his private suspicions, he treated a contest as available but not worth using.

2000 — Bush v. Gore. Gore actually retracted a concession call to Bush overnight as Florida's count tightened to within a few hundred votes out of six million cast. What followed was a genuine 36-day legal battle — recounts, court fights, and eventually a 5–4 Supreme Court decision in Bush v. Gore that stopped the recount and effectively decided the presidency. Gore conceded immediately after the ruling, saying he disagreed with the decision but would accept it "for the sake of our unity as a people and the strength of our democracy."

The through-line: each dispute involved a margin small enough — thousands of votes, sometimes hundreds — that reasonable people could disagree about the outcome. And each one ended: with a commission's ruling, a quiet concession, or a Supreme Court order the loser accepted even while disputing its reasoning.

Case	Year	Margin	Claim	Resolution
Hayes v. Tilden	1876	Disputed electors in 4 states	Competing elector slates submitted by both parties	15-member Electoral Commission and an informal bargain, months after Election Day
Nixon v. Kennedy	1960	~9,000 votes (IL), ~46,000 (TX)	Alleged Chicago machine irregularities	Nixon declined to contest; conceded the next day
Bush v. Gore	2000	~537 votes (FL)	Contested recount procedures	36-day legal battle; Gore conceded after the Supreme Court's Bush v. Gore ruling

The base rate: what the numbers say

Bright Line Watch, an academic election-monitoring project, tracked concession behavior among all defeated major-party candidates for federal office in 2020 and 2022. Their findings:

Overall, 65% of losing candidates conceded across both years and both parties.
Losing Democrats conceded 76% of the time in 2020 and 73% in 2022. Only 2% (four candidates) explicitly refused to concede in 2020, and zero did so in 2022.
Losing Republicans conceded only 45% of the time in 2020. One in five defeated Republican candidates that year — 43 people — explicitly denied the results, the highest rate for any party in any year in the dataset. That improved in 2022 — 59% conceded and 8% explicitly refused — but Republicans still trailed Democrats' 73% concede / 0% refuse rate that year.

The 2020 cycle produced a fivefold-plus partisan gap in explicit-denial rates. It also set the template: Bright Line Watch's own expert panel, surveyed ahead of the 2024 election, put the median forecast probability that Trump would refuse to concede a loss at 93%, and the probability that at least one other statewide candidate would refuse at 75%. Experts treat the tactic as a durable, recurring feature of close or high-profile races now — not a one-off reaction to a single election.

The newer pattern: bigger losses, faster claims, no exit ramp

Individual cases since 2016 show what that shift looks like in practice, in three ways.

The margin threshold has collapsed. Roy Moore (2017) and Matt Bevin (2019) lost by roughly 1.5 and 0.4 points — genuinely recount-territory margins, in keeping with the historical pattern. But Doug Mastriano lost by 14 points in 2022, and Mike Lindell lost by 11 points in Minnesota's 2026 primary. Neither number is remotely ambiguous. Claiming "anomalies" after a blowout is a qualitatively different act than requesting a recanvass after a margin under half a percent.

Unsubstantiated claims have moved from last resort to first reaction. Nixon in 1960 had actual, specific allegations tied to identifiable machine politics — and still chose not to act on them. Trump's 2020 claims, and the ones that have echoed his language since (Bevin's "irregularities," Lake's "fraud and misconduct," Lindell's "couple of anomalies"), are typically asserted within hours of a loss, before most votes are even counted, without specifics — and in most cases, never substantiated afterward. More than five dozen of Trump's 2020 lawsuits were dismissed or rejected; Lake's fraud claims lost at every level up to Arizona's Supreme Court; Kentucky's secretary of state found no evidence supporting Bevin's claims. The assertion now regularly precedes the evidence rather than following it.

Politician	Party	Race	Year	Margin	Claim	Outcome
Roy Moore	R	AL Senate	2017	~20,000 votes (1.5 pts)	Voter fraud, statistical anomalies alleged by "election integrity experts"	Refused to concede for weeks; lawsuit rejected by a judge; Jones certified
Matt Bevin	R	KY governor	2019	~5,100 votes (0.4 pts)	Unspecified "irregularities," "history of vote fraud" in the state	Forced a recanvass; conceded a week later when it didn't change the result
Stacey Abrams	D	GA governor	2018	~55,000 votes (1.4 pts)	Not fraud in the count — argued Kemp's oversight of his own election and a voter-roll purge made it illegitimate	Never fully conceded; repeated the framing through her 2022 rematch loss (by ~7.5 pts, which she did concede)
Donald Trump	R	US President	2020	~7M votes / 74 electoral votes	Rigged voting machines, illegal ballots, outside funding of election offices	Refused to concede for over two months; 60+ lawsuits dismissed or rejected; conceded the day after the Jan. 6 Capitol riot
Doug Mastriano	R	PA governor	2022	~14 points	No specific post-loss fraud claim, but campaign had called mail-ballot counting evidence "the fix is in"	Delayed ~5 days, then conceded cleanly
Kari Lake	R	AZ governor	2022	~17,000 votes (0.6 pts)	Fraud/misconduct by Maricopa County election officials	Sued repeatedly, lost at every level through the state Supreme Court; still wouldn't affirmatively say she lost as of a 2025 deposition
Kari Lake	R	AZ Senate	2024	~80,000 votes (2.4 pts)	Floated 2022-style "election interference" claims	Never issued a concession; said her "campaign was over" instead
Mike Lindell	R	MN GOP gov. primary	2026	~45,000 votes (~11 pts)	Unspecified "anomalies" found by his team	Refused to concede on election night; no lawsuit filed as of this writing

Trust in elections has become downstream of who won, not what happened. This is the part that separates a series of isolated incidents from an actual trend. Gallup's tracking shows Republican confidence that presidential votes are counted accurately at 55% in 2016 (when Trump won), collapsing to 28% in 2024 pre-election polling, then rebounding to 64% within weeks of Trump's 2024 win, according to AP-NORC polling. Democratic confidence has moved in something like the mirror pattern. Political scientists Steven Levitsky and Daniel Ziblatt, along with Nancy Bermeo, describe this as a hallmark of democratic backsliding elsewhere in the world: not a single dramatic rupture, but a habitual refusal by a share of the political class to accept unfavorable outcomes, which gradually convinces their voters that the system itself is rigged whenever their side loses. Once trust tracks outcome rather than process, an evidence-free fraud claim doesn't need to convince a court — it only needs to be asserted to a base already primed to believe it.

Why this matters

The Bright Line Watch numbers support a specific claim: within one party, in 2020, one in five losing candidates took the position that a definitive result wasn't real, against essentially no equivalent rate on the other side, and forecasters expect that tactic to keep recurring rather than fade. The shift is from isolated, close-margin disputes eventually resolved by institutions the loser ultimately accepted to a recurring script, used by a measurable fraction of one party's candidates, that treats "no concession" as an acceptable permanent state rather than a temporary one. Kari Lake still hadn't affirmatively acknowledged her 2022 loss as of a 2025 deposition — three years and multiple failed lawsuits later. That's not a dispute working its way through a process. That's the process being treated as optional.

Where this argument is contested

The counterpoints worth naming: 65% of losing candidates overall conceded across 2020 and 2022, and the "avalanche" of election-denial copycats some experts predicted for 2022 mostly didn't happen — the Brennan Center's review found most election-denying candidates who lost that year conceded anyway. Some analysts also argue the "trend" is overstated by media coverage that treats every sore-loser statement as equivalent to a coordinated movement, when in practice most of the figures discussed here (Mastriano, Bevin) conceded within days regardless of what they said on election night. Others note the behavior isn't confined to one party — Stacey Abrams's years-long refusal to fully accept her 2018 loss (by 1.4 points) shows Democrats aren't immune to delegitimizing a result they view as procedurally unfair, even without alleging a rigged count, even though Bright Line Watch's numbers found essentially no Democratic equivalent to outright denial in federal races. And some political scientists caution that public trust in elections has always been somewhat outcome-dependent — the losing side has reliably expressed more doubt than the winning side going back decades — so the post-2020 partisan trust gap may be a matter of degree rather than a genuinely new phenomenon.

Sources

Recent cases

Historical precedent

American Election Denial: 250 Years of Doubt, University of Idaho digital library project — cdil.lib.uidaho.edu

Polling and trust data

Base-rate and concession-rate data

Political science and democratic-erosion analysis