Overview
Most people imagine democratic collapse as a dramatic rupture — a coup, a burning parliament, a dictator seizing the radio station. The scholarly record shows the opposite: modern democratic erosion is typically gradual, legal, elite-driven, and invisible until well advanced. The gap between this popular mental model and the empirical evidence is not a minor quibble — it shapes whether citizens and institutions recognize and respond to threats in time.
Brief
The dominant popular script for democratic collapse — tanks in the street, a suspended constitution, a general on the radio — is an artifact of 20th-century history that fits the current global pattern poorly. The scholarly literature, anchored in large-N datasets from V-Dem and Freedom House covering the post-Cold War period, consistently finds that the dominant mode of contemporary autocratization is incremental erosion by elected incumbents, not sudden rupture by military actors. As the V-Dem 'State of the World 2024' report (published April 2025) documents, as of 2024 there are more autocracies (91) than democracies (88) in the world for the first time in over 20 years — and the route most countries took to get there was not a coup.
The 'it can't happen here' assumption — that wealthy, long-established democracies are structurally immune — has been decisively challenged by events and by formal analysis. Political scientist Adam Przeworski (2025) noted, as cited in the International IDEA Global State of Democracy 2025 report, that wealth and a history of peaceful transitions can foster a 'false perception of democratic consolidation,' leading to reduced vigilance and weaker resistance. Research published in PNAS (January 2025) using V-Dem data found a robust statistical association between income inequality and democratic erosion, explicitly including wealthy democracies — the level of institutionalization was not clearly protective. The Economist Intelligence Unit's 2024 global democratic index recorded its lowest score ever, with 130 of 167 countries measured showing no improvement or decline.
The assumption that an informed citizenry will reliably punish leaders who erode democratic norms has also been repeatedly tested and found to be, at best, conditionally true. A November 2025 study in the British Journal of Political Science found that while a majority of American respondents were willing to electorally remove an incumbent as backsliding unfolded, this disposition was heavily conditioned by partisan and policy interests — not an unconditional democratic reflex. The broader literature identifies a well-documented 'partisan double standard' or 'democratic hypocrisy' dynamic: voters punish co-partisan candidates less harshly for democratic transgressions than out-party candidates. Citizens are not reliably the last line of defense.
The picture from Freedom House's 'Freedom in the World 2026' report is consistent: the gradual erosion of democratic institutions — manipulation of elections, undermining of judicial independence — is identified as 'extremely damaging to freedom,' while being less immediately legible than coups or armed conflict. Crucially, elections continue to occur during erosion — they just become progressively less free and fair through legislative and procedural changes that constrain challengers, suppress voters, and shift the power to certify results. The continued existence of elections is itself a mechanism of legitimation, not evidence that erosion is absent.
What the evidence does support is that institutional architecture matters more than civic will as a first-order safeguard: V-Dem data analyzed by Boese et al. (2020) shows that judicial autonomy predicts regime survival during backsliding episodes even after controlling for economic development and democratic age. Courts and legislatures function as the real tripwires — which is exactly why incumbents with autocratic intentions typically target judicial independence and legislative constraints first. Understanding this sequencing is the single most important correction the evidence offers to the popular mental model.
Myths & Realities (7)
Myth
Democracies die when the military seizes power — a coup is the real threat to watch for.
Reality
The dominant mode of democratic collapse since the 1990s is not coups but gradual erosion by elected incumbents using legal means: court-packing, electoral rule changes, media capture, and the systematic weakening of oversight institutions. Coups still occur, primarily in Sub-Saharan Africa and parts of Asia, but they are not the primary mechanism through which established democracies fail.
Evidence: Brookings (March 2025) and Freedom House both document that it is now rare for an authoritarian to come to power via coup in established democracies; instead, Freedom House data over 20 years identifies elected leaders' overreach as the primary driver of decline in Free countries. V-Dem's third-wave autocratization research (Lührmann & Lindberg) established that incumbents 'legally access power and then gradually, but substantially, undermine democratic norms without abolishing key democratic institutions' as the modal contemporary tactic.
Kernel of truth: Coups do still occur — Freedom House's 2024 report documents that Niger became the sixth Sahel country to experience a coup since 2020 — so the threat is real in certain regional contexts, just not the dominant global mechanism for established democracies.
Why believed: 20th-century democratic breakdowns in Latin America, Southern Europe, and Asia were often coup-led, and that historical template dominates civics education and popular media depictions of dictatorship.
Myth
It can't happen here — wealthy, long-established democracies are too institutionally robust to slide into autocracy.
Reality
Wealth and institutional age provide some protection but are not determinative, and may actually create a dangerous false sense of consolidation. PNAS research published January 2025, using V-Dem data from 1995–2022, found that income inequality — not wealth per se — is a robust predictor of erosion, and that state institutionalization was not clearly protective. Political scientist Adam Przeworski (2025) identified wealth combined with a history of peaceful transitions as a potential source of reduced vigilance, not guaranteed immunity.
Evidence: As of 2024, the V-Dem State of the World report counts 91 autocracies vs. 88 democracies globally — the first time autocracies outnumber democracies in over 20 years. The Economist Intelligence Unit's 2024 democracy index recorded its lowest global score ever. Multiple major indices — V-Dem, Freedom House, and the Economist Intelligence Unit — document measurable erosion in longstanding democracies including the United States.
Kernel of truth: Wealthy democracies with strong institutions do have measurably higher resilience on average — the causal link between development and democratic survival has historical support, and lower-income countries remain disproportionately represented among the most severe backsliders.
Why believed: Decades of political science consensus held that democracy, once established in a sufficiently wealthy state, was effectively irreversible — this finding has been substantially revised by post-2010 evidence but the older view persists in public discourse.
Myth
Citizens will defend democracy — public outrage and electoral punishment will stop backsliding leaders.
Reality
Citizen accountability of backsliding leaders is real but conditional, heavily mediated by partisan loyalty. Research consistently identifies a 'partisan double standard' or 'democratic hypocrisy': voters punish democratic transgressions by co-partisan candidates significantly less harshly than the same actions by out-party candidates. A 2024 Journal of Politics study (Frederiksen) found this dynamic operative even controlling for policy preferences. The Journal of Democracy (November 2024) found that democracies 'erode from the top' — elite behavior, not citizen preferences, is the primary driver.
Evidence: A November 2025 British Journal of Political Science preregistered experiment (N=4,234) found that while a majority of Americans would remove a backsliding incumbent, the response was conditioned on partisan and policy context — not reflexive. Survey data from U.S. state legislators and local officials during the 2024 primary elections (Frederiksen et al., cited at Yale ISPS March 2025) found that politicians were unwilling to pay the costs of opposing norm violations, seeing that as more costly than opposing either of the two other major candidates. European Journal of Political Research research (Jacob, 2025) found that anti-pluralist incumbents undermined institutions most severely when citizen support for democracy was weak — confirming that citizen commitment is a real but variable constraint.
Kernel of truth: Strong citizen commitment to democratic norms does raise the cost of backsliding for incumbents and makes erosion less likely, as the Jacob (2025) EJPR study confirms — citizen attitudes are not irrelevant, just not reliably sufficient.
Why believed: Democratic theory centers the sovereign will of the people, which generates an intuitive assumption that sufficiently blatant anti-democratic behavior will trigger mass electoral punishment; this assumption does not survive contact with affective polarization.
Myth
Backsliding is obvious when it happens — people will recognize it and respond.
Reality
Democratic erosion is structurally designed to be ambiguous. Incumbents pursue changes that are individually defensible — a court reform framed as efficiency, a media regulation framed as combating misinformation, an electoral rule framed as preventing fraud — while cumulatively dismantling the architecture of accountability. The Lowy Institute's analysis describes democratic erosion as 'often not obvious until the effects are well advanced.' The continued occurrence of elections is itself used to claim democratic legitimacy while those elections are being progressively constrained.
Evidence: A 2020 study cited in Wikipedia's democratic backsliding entry established that 'elections are improving and rights are retracting in the same time period, and in many of the same cases' — meaning electoral metrics and rights metrics can diverge, creating a false picture of health. Freedom House (2026) identifies 'manipulation of elections' and 'undermining judicial independence' — both individually legible as routine governance disputes — as the leading mechanisms of gradual erosion. The Journal of Chicago's 'A Wolf in Sheep's Clothing' paper (2025) models citizen uncertainty about whether norm violations are genuinely anti-democratic as a key mechanism enabling backsliding.
Kernel of truth: Some backsliding episodes do produce clear, visible moments of rupture — Hungary's 2010-2013 constitutional overhaul and South Korea's December 2024 martial law declaration (which was quickly reversed by the legislature) are examples where the anti-democratic intent was relatively legible.
Why believed: People expect categorical events — a declaration, a purge, an arrest of opposition leaders — rather than a slow accumulation of procedural changes that individually fall within normal political contestation.
Myth
Backsliding happens because governments fail to 'deliver' — fix the economy and democracy stabilizes.
Reality
The claim that democratic erosion is primarily a performance-legitimacy story — citizens abandon democracy when it disappoints them economically — is not well supported by the evidence. The Journal of Democracy (November 2024) explicitly challenges this 'conventional wisdom,' finding it does not hold on a global basis. Backsliding leaders like El Salvador's president Bukele have maintained approval ratings above 90% while in office — but this popularity came after, not before, erosion began, and reflected endorsement of the erosion itself rather than a demand for it driven by prior economic failure.
Evidence: The PNAS January 2025 study found income inequality — not GDP per capita, growth rates, or government service delivery — as the robust predictor of erosion, and traced the causal pathway through partisan polarization rather than direct performance dissatisfaction. The Journal of Democracy (November 2024) notes that some of the most popular backsliding leaders governed during periods of relative economic normalcy; their support reflected ideological alignment and polarization rather than desperation. V-Dem research identifies disinformation — documented as increasingly prevalent in nearly a quarter of all countries as of 2024 — as a distinct driver that operates independently of policy performance.
Kernel of truth: Economic dislocation and growing inequality do create conditions that make citizens more susceptible to populist appeals and less committed to incumbent institutions — the PNAS study's inequality finding confirms a real economic dimension, just not via the simple 'deliver better and democracy is safe' mechanism.
Why believed: The performance-legitimacy model is intuitive and politically convenient — it implies a tractable solution (better governance) and avoids harder conclusions about the structural vulnerability of democratic institutions to elite manipulation.
Myth
Democratic backsliding is a problem for fragile, young, or poor democracies — not for countries like the United States or Western Europe.
Reality
Multiple independent democracy indices document measurable erosion in longstanding, wealthy democracies. Freedom House's 'Freedom in the World' reported a decade of decline in the United States as of March 2025. The V-Dem Institute's 2026 democracy report classified the U.S. as an 'electoral democracy' for the first time in over 50 years, losing its liberal democracy status. The Economist's Democracy Index moved the United States from 'full democracy' to 'flawed democracy.' Research by Jacob Grumbach, whose State Democracy Index covers all 50 U.S. states from 2000 to 2023, documents that state governments have led the way in democratic backsliding domestically.
Evidence: Grumbach's SDI findings (cited in CITP Working Paper 027, November 2025) show that Republican control of both chambers in a state legislature is the strongest predictor of institutional decline at the state level. The EU's experience with Hungary and Poland through the 2010s — both EU member states — demonstrates that regional integration and shared prosperity do not immunize democracies against internal erosion, a dynamic documented by Rutgers political scientist R. Daniel Kelemen who argues EU membership may have enabled an 'authoritarian equilibrium' in some cases.
Kernel of truth: The most severe and rapid collapses into closed autocracy remain concentrated in lower-income countries and those with shorter democratic histories — wealthy democracies have more institutional redundancy and typically erode more slowly.
Why believed: The canonical cases of dramatic backsliding taught in political science — Venezuela, Hungary, Turkey — are all middle-income or developing-world cases, reinforcing an unconscious association between backsliding and a certain kind of country.
Myth
Once backsliding begins, it always ends in full autocracy — there is no recovery.
Reality
Democratic erosion does not follow a deterministic one-way path to closed autocracy. V-Dem's 2024 State of the World data identified 19 countries in ongoing episodes of democratization alongside the 45 in autocratization. Countries including Austria, Chile, Nepal, and South Africa appeared on V-Dem's watchlist in prior years but stabilized by 2024. Brazil's experience — where former president Bolsonaro was barred from standing for office for eight years by the election court for efforts to undermine electoral integrity — illustrates that institutional counterweights can halt or reverse erosion.
Evidence: The IGCC Working Paper (December 2024, Kaufman) finds that just under a third of democracies with a newly elected anti-pluralist government experienced backsliding in the following year — meaning two-thirds did not. V-Dem data analyzed by Boese et al. (2020) shows judicial autonomy predicts democratic survival during backsliding episodes. Freedom House (July 2023) notes that independent courts remain 'centrally important' to reversing erosion, citing Brazil's electoral court and India's Supreme Court as examples of institutional checks functioning under pressure.
Kernel of truth: Backsliding that progresses beyond certain institutional thresholds — particularly when judicial independence and electoral administration are captured — does become self-reinforcing and very difficult to reverse; Hungary after 2013 is the canonical case of erosion reaching a tipping point.
Why believed: The most visible and documented cases of backsliding (Venezuela, Hungary, Turkey) all progressed to deep erosion or near-autocracy, creating survivorship bias toward the most severe outcomes in public awareness.
Facts & Figures (6)
The claims behind this analysis, each with its verification status — including what is contested, unverified, or could not be established.
As of 2024, V-Dem counts 91 autocracies and 88 democracies globally — more autocracies than democracies for the first time in over 20 years (V-Dem State of the World 2024 report, published April 2025).
Establishes that the global balance has already tipped, undercutting any assumption that democracy is the default or stable equilibrium.
✓ GROUNDED
The Economist Intelligence Unit's 2024 global democratic index recorded its lowest score ever, with 130 of 167 countries measured showing no improvement or decline (Lowy Institute, citing EIU 2024).
Quantifies the breadth of the current erosion wave across the full country distribution, not just fragile states.
✓ GROUNDED
Research published in PNAS (January 2025), using V-Dem data for 1995–2022, found a robust statistical association between income inequality and democratic erosion; the level of state institutionalization was not clearly protective.
Directly challenges the claim that wealthy, institutionalized democracies are structurally immune to erosion.
✓ GROUNDED
A November 2025 British Journal of Political Science preregistered experiment (N=4,234) found a majority of American respondents were willing to remove a backsliding incumbent electorally — but only against the backdrop of partisan and policy interests, not as a reflexive democratic commitment.
Shows citizen accountability of backsliding leaders is real but conditional, not automatic — partisan loyalty mediates it.
✓ GROUNDED
V-Dem data analyzed by Boese et al. (2020), cited in a December 2024 IGCC Working Paper, shows that judicial autonomy predicts regime survival during backsliding episodes even after controlling for economic development, democratic age, and other factors.
Identifies institutional architecture — specifically independent courts — as the strongest structural safeguard, not citizen mobilization alone.
✓ GROUNDED
Freedom House's 'Freedom in the World' data over the last 20 years shows that the primary drivers of decline in Free electoral democracies have been overreach by elected leaders and authoritarian-legacy conditions — not military coups (Freedom House, cited July 2023).
Directly grounds the 'coup myth' correction with the most widely used global democracy index's own causal accounting.
✓ GROUNDED