Psychology by BogdanovaThe Encyclopedia of Psychology for the Artificial Era

RESEARCH DOSSIER

Democratic Norms and Support for Undemocratic Practices

Abstract. Democratic norms are shared expectations about electoral fairness, opposition rights, civil liberties, peaceful transfers of political power, and constraints on partisan rule breaking. This fundamental research monograph examines when psychological endorsement of these principles constrains actual political choices and when partisan incentives, perceptions of threat, and elite signals encourage support for undemocratic practices. A critical narrative synthesis traces the historical study of political tolerance, develops a multidimensional account of democratic commitment, and compares conjoint choice experiments, symmetric political scenarios, elite surveys, longitudinal designs, correction interventions, and cross-national evidence. Original studies and contemporary research through 2026 reveal that general support for democracy, sanctioning a transgressive candidate, endorsing a concrete procedural violation, and supporting political violence are separate outcomes. Findings about misperception corrections are mixed, and ballot structure and institutional context can substantially change voters' responses. The analysis identifies measurement, causal inference, ethical, and cross-national limitations and establishes priorities for replication, real-world behavioral validation, and research on information mediation in the Artificial Era. The method is critical narrative synthesis; no original participant data, preregistered systematic search, or meta-analysis is claimed.

Keywords: democratic norms; support for undemocratic practices; democratic commitment; political tolerance; democratic backsliding; partisanship; affective polarization; partisan violence; electoral accountability; candidate conjoint experiments; democratic hypocrisy; out-group threat; elite cues; misperceptions; political legitimacy; political psychology; electoral choice context; democratic balancing; research methods; Artificial Era

Democratic Norms as a Psychological Research Object

Democratic norms are shared expectations about legitimate political competition and the restraints that make recurring competition possible. For this dossier, their core domains are electoral integrity, the lawful transfer of executive authority, equal political participation, opposition rights, protection of civil liberties, and acceptance of institutional constraints on officeholders. Psychological commitment is the willingness to apply these expectations when doing so has costs for one's own political side. That qualification is operationally central: approving democracy in an abstract questionnaire may coexist with approving a particular incumbent's attempt to evade an adverse court ruling. The distinction between an endorsement of a constitutional ideal and a costly choice about political power underlies the electoral-tradeoff model in Graham & Svolik (2020) and the contextual experiments in Simonovits et al. (2022).

Support for undemocratic practices is an attitudinal or behavioral willingness to excuse, authorize, or choose acts that weaken fair political contestation or the rights of opponents. Research vignettes may describe closing polling locations in an opponent's neighborhood, censoring critical media, sidestepping legislative oversight, rejecting an election outcome, or favoring candidates who endorse such practices. These examples differ in severity and legal context; they should not automatically be collapsed into one score. An authoritarian preference, a readiness to bend a particular procedural rule, tolerance of an opposing movement, and acceptance of partisan violence can be correlated without denoting an identical psychological disposition. The outcome-specific intervention evidence in Voelkel et al. (2024) makes these distinctions consequential for both theory and measurement.

The central explanatory question is when citizens' professed democratic principles constrain their decisions and when competing motives redirect them toward procedural violations. The primary units of analysis include individual voters, partisan groups, political candidates, elected legislators, and the institutional opportunities within which choices take place. Explanations therefore span social identity, perceived threat, motivated reasoning, normative expectations, instrumental competition, political socialization, and rule enforcement. This monograph is a critical narrative synthesis of published primary studies, reviews, and conceptual works. It is not an original data analysis, preregistered systematic review, or quantitative meta-analysis; its causal conclusions are limited to what identifiable designs can establish.

What Constitutes a Democratic Violation?

Classification of a political act requires a specified institutional rule and a stated mechanism of harm. Restricting the lawful opportunities of opponents to vote, changing rules to prevent credible competition, punishing independent reporting, manipulating electoral administration, or refusing a binding result threatens the conditions by which parties can alternate in office. The analysis should identify which actors have legal authority, what procedural safeguards exist, and who bears the costs. Legitimate disagreement about a policy does not by itself identify democratic erosion. Przeworski (1991) treats regulated competition and the capacity of political losers to compete again as organizing democratic features; this is useful for specifying why a violation matters beyond any one party's preferences.

Rules, informal norms, and ethical standards overlap but should be analytically separated. Some violations contravene explicit law, while others exploit discretion, weaken reciprocal restraint, or use powers selectively against political opponents without a readily adjudicated offense. Levitsky & Ziblatt (2018) foregrounds mutual toleration and institutional forbearance as historically significant informal commitments. Their invocation in a psychological study should be anchored to observable choices and actor conduct rather than inferred from respondents' party labels. Researchers should code the act, the actor, the beneficiary, and the institutional condition separately.

A useful coding system records whether the proposed conduct concerns ballot access, electoral results, civil liberties, judicial independence, legislative oversight, opposition participation, executive overreach, or political intimidation. It also records whether respondents approve the act, expect their side to use it, believe opponents have already used it, or would vote for a politician who proposes it. These responses answer different questions. Hypothetical acceptance does not establish real-world misconduct, and voting for a candidate with a rule-violating position may coexist with explicit disapproval of that position when other priorities dominate, as the design of Graham & Svolik (2020) illustrates.

Historical Foundations: Tolerance, Competition, and Democratic Breakdown

Postwar American political-tolerance research established an enduring measurement problem: public endorsement of civil liberty can coexist with resistance to granting liberties to particular disliked groups. Stouffer (1955) examined attitudes toward unpopular groups in the context of ideological anxiety and civil-liberties controversies. The historical setting matters, because the groups seen as dangerous and the anticipated costs of tolerance change. The methodological legacy is more general: an abstract statement about freedom of expression is weaker evidence of an operative norm than willingness to extend that freedom to a specific opponent.

Sullivan et al. (1982) developed the political-tolerance tradition by emphasizing how an interviewer's choice of target group can distort comparisons of tolerance. A respondent who supports free speech for one presented group may favor restrictions on whichever group they personally find most objectionable. The least-liked-group strategy therefore studies tolerance toward an individual's own salient adversary. Its conceptual contribution to democratic-norm research is the principle of symmetric application: a norm has practical force when it governs the treatment of an adversary rather than merely preferred speakers. The approach also requires careful documentation of the target population, local legal context, and survey wording.

Comparative studies of democratic breakdown shifted attention toward political actors' decisions, institutions, and the conditions of regime survival. Linz & Stepan (1978) analyzed breakdown and reequilibration in historical political contexts, while Przeworski (1991) emphasized organized, recurring competition under rules that permit political losers to return. Later work on executive-led erosion asked why elected citizens might tolerate violations by their preferred leaders. This history joins political tolerance to a political-psychological puzzle: norms can be verbally endorsed yet fail as constraints when the distribution of political power is at stake. Earlier classics supply conceptual and historical frameworks; they do not themselves establish the contemporary prevalence of any measured attitude.

From Constitutional Ideals to Costly Political Choice

A survey that asks whether democracy is important elicits a broad evaluative judgment with weak information about the respondent's likely action when policies, partisan identity, and electoral rules collide. Researchers increasingly study preferences among candidates or governing scenarios that combine attributes. This change in design is substantive, not merely technical: the respondent faces an allocation problem in which defending an institution might entail an electoral loss. Svolik (2019) explained why polarized electoral competition can turn citizens into unreliable checks on elected leaders who weaken democracy. It makes democratic protection contingent on electoral incentives and on the existence of credible alternatives.

Graham & Svolik (2020) estimated such tradeoffs using a nationally representative candidate-choice experiment and a natural experiment associated with Montana's 2017 special congressional election. Candidate attributes included positions threatening democratic principles, allowing the researchers to observe whether voters would penalize transgression despite ideological or partisan attraction. The published study found that voters' willingness to defend democratic principles in electoral choices was limited and declined with several dimensions of polarization. Its experiment identifies responses to described candidate attributes; the complementary electoral setting improves real-world relevance while retaining context-specific limitations. A statement that many people favor democracy cannot substitute for this kind of behavioral tradeoff evidence.

The resulting distinction separates principled disapproval from effective sanction. Voters may judge a candidate's procedural act wrong and nevertheless support that candidate because of policies, distrust of the rival, or a belief that the violation is exceptional. An experiment must therefore measure both the perceived acceptability of the act and the resulting electoral choice where feasible. The difference is essential for evaluating institutional resilience: citizen opposition to democratic erosion constrains politicians most directly when it is translated into punishment, public resistance, or withdrawal of cooperation. Research should model the cost at which commitment ceases to govern action rather than treating endorsement as a constant personal trait.

Analytical Map of Democratic Commitment

Democratic commitment has at least four distinguishable components: a normative belief that certain political procedures ought to be followed; a general expectation that others will follow them; a conditional willingness to uphold them when one's own side is disadvantaged; and a choice or action that enforces them. Evidence for the first component may come from survey scales, for the second from perceived-group-norm elicitation, for the third from in-party versus out-party counterfactual scenarios, and for the fourth from election-related decisions or observed civic behavior. The domains may influence one another, but no single indicator reveals all four. Simonovits et al. (2022) shows why manipulating who holds power makes conditional commitment visible.

Support for undemocratic practices likewise has layers: an individual's approval of a specified procedure, willingness to vote for someone who proposes it, beliefs about the practice's desirability to allies, and willingness to perform or facilitate it. Elite actors face additional incentives, since they can directly enact rules and may hold inaccurate beliefs about their constituents. Druckman et al. (2023) supplies an experimental example involving American state legislators and their perceptions of opposing-party voters. Treating elite perceptions as a causal input suggests a different level of explanation from a simple census of voter attitudes.

A complete research structure therefore follows the links between principles, perceptions, identities, tradeoffs, electoral sanction, political violence, and intervention. Each link is a potential mechanism with separate observable implications and scope conditions. The strongest evidence includes randomized candidate attributes, context-balanced scenarios, experimental corrections, repeated measurements, and triangulation with actual electoral events. This architecture also guards against a common interpretive error: the appearance of partisan hostility does not by itself demonstrate willingness to censor opponents, undermine elections, or endorse violence. The empirical relationship between these outcomes must be estimated rather than presumed, as Voelkel et al. (2023) and Voelkel et al. (2024) demonstrate.

Affective Polarization, Partisan Identity, and Norm Application

Affective polarization refers to evaluations in which members of one political party favor their own partisan group and view the other with greater suspicion or hostility. Iyengar et al. (2012) established a social-identity approach that treats partisan feeling as a dimension of political conflict distinct from ideological distance. Finkel et al. (2020) later described political sectarianism through othering, aversion, and moralized opposition in the American context. These constructs are relevant to democratic norms because perceptions of the out-party may alter the felt legitimacy of protecting its civil liberties, electoral access, or institutional influence. The strength of this pathway cannot be deduced from an affect thermometer alone: a respondent may dislike opponents while still supporting their rights.

Kingzette et al. (2021) offers important evidence for conditional rather than uniformly reduced support. Across American surveys, partisan affect was associated with opposition to or support for constitutional protections in ways that depended on which party controlled the presidency. In the analysis, affectively polarized Republicans were more likely to oppose protections during a Republican administration and polarized Democrats more likely to support them; the earlier Democratic presidency showed the reverse orientation. The resulting pattern supports a cue-taking and context-sensitive account. It does not support treating all partisans' affective polarization as a constant, direction-free cause of intolerance; controls, elite signals, and temporal context remain integral to interpretation.

The mechanisms require disaggregation. Hostility can make exceptions feel deserved, identity can make a politician's position especially credible, and perceived threat can heighten the anticipated cost of restraint. Yet elite endorsement of a rule violation may also reorganize preferences among supporters without a preceding increase in hostility. A valid empirical test manipulates the political beneficiary and the source of an endorsement, measures initial affect independently, and specifies the causal comparison. The distinction is indispensable when evaluating interventions: Voelkel et al. (2023) found that treatments reducing affective polarization need not improve antidemocratic attitudes. A single-factor explanation would predict more parallel change than observed.

Democratic Hypocrisy and the Party-in-Power Counterfactual

Conditional principle application is observable when a person evaluates a substantively identical breach differently depending on who benefits. The term democratic hypocrisy identifies a discrepancy between declared rule commitments and power-dependent approval, but it should be defined operationally rather than used as a moral label for an individual. A symmetric counterfactual asks whether the respondent would apply the same restriction to a favored and a disfavored officeholder, holding the procedure itself stable. This design isolates party advantage more convincingly than comparisons of different governments facing different crises.

Simonovits et al. (2022) developed hypothetical experiments that varied whether respondents' party controlled government, allowing a more direct test of the costs and benefits of tolerating erosion. The authors found that American support for norm-eroding policies was greater when the respondent's party held power, with stronger associations among expressive partisans and those who perceived greater threat from the opposition. This is evidence about power-contingent attitudes under the specified scenarios. The design cannot alone demonstrate that respondents will tolerate an identical real-life decision after years of party campaigning, litigation, or media framing.

A rigorous replication should randomize the incumbent's party, candidate name or anonymized identifier, breach content, institutional stakes, and severity. It should measure perceived fairness and expected political effects after treatment, while treating those reports as potential mediators rather than automatically established causes. Researchers must also detect demand characteristics: respondents may infer the investigator's normative expectation from extreme examples. Symmetry in wording is particularly important because examples of executive action can carry different legal meanings across political systems. The psychological finding concerns how allegiance conditions normative application; the legal classification of the underlying act remains an independent empirical question.

Threat, Anticipated Retaliation, and Defensive Rule Breaking

A voter may favor democratic procedure in the abstract while expecting that the opposing party will violate it first. Under such beliefs, a rule violation can be perceived as defensive compensation, an attempt to prevent unilateral disadvantage, or a temporary exception necessary for long-term preservation. This mechanism combines an expectation about the rival with a moral and strategic appraisal of proportionality. It predicts that exposing a respondent to trustworthy evidence of rival commitment can reduce support for preemptive violations, particularly among those who initially exaggerate rivals' antidemocratic intent.

Braley et al. (2023) investigated precisely this pathway. An observational study and two experiments examined American partisans' misperceptions of their opponents' willingness to undermine democracy. Experimental information indicating that out-partisans were more supportive of democratic norms than respondents believed increased participants' own reported commitment and reduced willingness to select candidates who violated those norms. The design provides stronger evidence for a causal effect of the correction within the studied settings than for a universal theory of democratic erosion. Treatment effects depend on who delivers the information, what was initially believed, how long the correction persists, and the surrounding competing claims.

Perceived threat is therefore both a plausible mediator and an object of strategic communication. When a leader describes routine electoral rivalry as existential, supporters may infer that procedural protections for opponents are dangerous. But threat may also follow genuine risks, and analysts should not infer falsity from intense concern alone. Sound research measures the accuracy of the belief, its emotional consequences, and the selected remedy separately. McCoy et al. (2018) explains how severe relational polarization aligns multiple disagreements into encompassing opposing camps; this framework supplies a comparative account of how political rivalry can become socially totalizing.

Elite Cues, Institutional Authority, and Norm Enforcement

Democratic rule maintenance is a coordination problem involving voters, officeholders, media institutions, courts, and other elites. A respondent may have limited independent information about whether a technical procedural change is legitimate and rely on trusted political figures to interpret it. Elite cues can make obscure institutional conduct salient, provide a justification, or signal that enforcement is politically costly. A credible experiment therefore varies both the proposed violation and the partisan affiliation or institutional authority of its endorser, separating information acquisition from identity-consistent compliance.

Political elites have their own beliefs about the electorate. In an experiment with 534 American state legislators, Druckman et al. (2023) found that legislators substantially overestimated hostility and antidemocratic attitudes among opposite-party voters, despite having comparatively lower reported levels of some such attitudes themselves. The study reported that correcting certain misperceptions could reduce legislators' support for undemocratic practices. The population matters: state legislators are elected institutional actors rather than a random sample of citizens, and treatment effects on survey responses are not automatically effects on votes in legislatures, judicial appointments, or actual rule changes.

Norm enforcement depends on political incentives as well as individual virtue. If a party benefits from a transgression and voters tolerate the resulting tradeoff, other officeholders may anticipate little electoral punishment. Conversely, visible penalties imposed by co-partisans can establish that constitutional restraint has practical force. Graham & Svolik (2020) demonstrates why the public's credibility as a check needs analysis in the candidate-choice setting. Measuring the political cost of a rule violation requires attention to media visibility, feasible opposition alternatives, electorate composition, office type, and the legal powers available to the actor.

Moralization, Justification, and Perceived Legitimacy

An anti-democratic act can be reframed as a morally authorized response to perceived injustice, danger, or prior misconduct. The underlying psychology involves selective attribution of motives, unequal thresholds of evidence for in-group and out-group actions, and beliefs about whether an outcome is procedurally deserved. Such possibilities require direct measurement; they should not be assumed of any political tradition or social category. The field should distinguish moral approval of an end from moral approval of the means and from expected effectiveness. An individual may sincerely favor a policy goal yet reject disenfranchisement used to achieve it.

Research on rule-breaking judgments must preserve the difference between legal legitimacy, social legitimacy, partisan usefulness, and subjective moral endorsement. An executive action that increases efficiency may remain politically controversial but procedurally authorized; another action might unlawfully restrict opposition participation while being justified as urgent. Coding all intense disagreement as anti-democratic would destroy construct validity. Levitsky & Ziblatt (2018) situates reciprocity and restraint in historical institutional analysis, while Sullivan et al. (1982) shows the importance of asking about rights for personally disfavored groups. Together they help define democratic commitment as a principle governing adversarial relationships.

Competing accounts can generate similar responses. A respondent may approve censorship because they distrust a particular outlet's accuracy, because they desire partisan advantage, because they expect social harm, or because an authority describes censorship as legally necessary. Discriminating among mechanisms requires factorial manipulation of the target, stated rationale, actor identity, and independent evidence. Explanatory claims should report what the study actually randomized. Where a design only observes self-reported ideology and support for restrictions, alternative explanations such as information environments, institutional trust, and risk perception remain open.

Measuring Democratic Norms: Construct Validity and Scale Design

Measurement begins by specifying the domain to which a norm applies. Electoral fairness includes equal opportunity to participate and accurate counting; opposition rights include expression, organizing, and lawful candidacy; institutional constraint concerns compliance with binding legal limits. Each domain should be represented by multiple items rather than a single abstract question about democracy. Researchers must ask whether scores form one stable dimension, several correlated dimensions, or patterns contingent on partisan beneficiary. Factor analyses are informative only when item content, response options, and sampling frames are explicit. Repeated evidence from Simonovits et al. (2022) and Kingzette et al. (2021) makes the identity of the beneficiary a candidate measurement facet, not a nuisance variable to be ignored.

Norm items are susceptible to acquiescence, social desirability, asymmetry of political examples, differential interpretations of institutional legitimacy, and changing news contexts. Balanced scenarios can hold the act constant while randomizing which party proposes or benefits from it. An item that appears facially neutral can nonetheless evoke a specific historical event or disliked leader in one population. Surveys should therefore include cognitive interviews, pretesting, comprehension checks, and analysis of item functioning across affiliations, ages, education levels, and countries. Sullivan et al. (1982) remains methodologically relevant because its approach exposed how the selected target of intolerance could change the inferred level of tolerance.

Reliability is necessary but insufficient. A high internal consistency statistic can arise from repeated partisan cues rather than coherent commitment to democratic procedure. Convergent validity requires links to defensible external indicators, and discriminant validity requires separation from general government dissatisfaction, mere out-party dislike, and approval of political violence. Predictive validity is difficult because violations are uncommon, context dependent, and often enacted by elites rather than voters. Instruments should report their item universe, scoring, missing data, interval between observations, and validity evidence rather than present an index as a transparent measurement of a timeless trait.

Candidate-Choice Conjoint Experiments and Electoral Tradeoffs

A candidate-choice conjoint task presents profiles varying on party affiliation, policy positions, competence, personal characteristics, and conduct relevant to democratic rules. Random assignment permits estimation of how one attribute affects the probability of selection within the distribution of profiles presented. Graham & Svolik (2020) used this logic to study Americans' willingness to penalize candidates for violations of democratic principles. The research question concerns the marginal electoral consequence of the violation when other features compete for attention. Because respondents choose among alternatives, the design is often more behaviorally relevant than an isolated approval statement, while still remaining a survey experiment.

Interpretation requires care about estimands. An average marginal component effect is not the proportion of citizens who sincerely prefer democracy; it is a change in choice probability averaged over specified randomized profiles. If the candidate description includes a salient policy that respondents value intensely, a democratic-violation attribute may receive a smaller electoral penalty without proving that respondents approve of the violation. Researchers should compare within-party and cross-party choices, inspect whether a nonviolating candidate is feasible, and distinguish abstention from selection of an otherwise undesirable opponent. Profile realism and temporal relevance affect external validity.

Candidate experiments often face hypothetical bias and simplified information. Respondents cannot observe the full institutional consequences of a violation, and real candidates are discussed over months under highly partisan media conditions. Actual elections introduce other sources of ambiguity: candidate quality, turnout, ballot access, and strategic voting can produce aggregate results that do not map neatly onto an individual's democratic principles. A strong research program triangulates experiments, natural experiments, panel surveys, and where feasible documented behavioral choices, while treating disagreements among estimates as opportunities to refine the explanatory model. Carey et al. (2022) offers a related candidate-tradeoff perspective among voters and partisan donors.

Survey Experiments, Counterfactual Symmetry, and Time

Scenario experiments can isolate the psychological cost of a rule violation by varying whether the actor is an ally or an opponent, whether an election is close, and whether the stated rationale invokes security or fairness. Simonovits et al. (2022) used this strategy to reduce confounding between political context and party identity. Because participants are randomly assigned contrasting vignettes, the difference between conditions has a clearer causal interpretation than a comparison of two contemporaneous real governments. However, a vignette with no durable institutional consequence measures response to the imagined situation, not the subsequent conduct of a voter faced with real uncertainty.

Longitudinal studies add information about persistence and change. Respondents can be surveyed before and after political events or treatment, permitting analysis of within-person trajectories and lagged outcomes. Such data do not guarantee identification of causal effects, since exposure to media, changing officeholders, and unobserved experience may influence both attitudes and predictors. Randomized interventions embedded in panel designs improve identification of immediate effects and reveal decay or persistence when follow-up is obtained. Mernyk et al. (2022) provides an example of combining cross-sectional, experimental, and follow-up evidence for perceived out-party support for violence.

The standard for time-dependent inference should include preregistration of primary outcomes, analysis of attrition, measurement invariance across waves, and interpretation of repeated exposure to the survey itself. Panels may condition participants to familiar terminology or increase attention to issues they would rarely consider otherwise. Researchers must not interpret a post-election change as the effect of an election without evaluating coinciding news, elite statements, and mobilization. Cross-time symmetry is often as important as cross-party symmetry, especially when incumbency changes the meaning of political restraint.

Political Violence: Distinct Outcome and Measurement Debate

Support for partisan violence refers to approval of physical coercion or aggression directed toward political opponents or political ends. It differs conceptually and legally from support for nonviolent but undemocratic procedures, such as suppressing votes or ignoring court orders. Researchers must distinguish approving violence in principle, believing it justified under specified conditions, reporting a personal willingness to engage in violence, and actually participating in violent acts. Each moves progressively farther from ordinary political attitude items and entails different predictors, base rates, and ethical measurement requirements. Treating an affirmative survey response as evidence of imminent violent conduct is a serious inferential error.

Westwood et al. (2022) investigated whether survey estimates of support for American political violence were inflated by ambiguous question wording and inattentive or expressive responses. The article argued that improved measurement substantially reduced estimates of approval, while retaining concern about genuine violent minorities. Mernyk et al. (2022), in separate preregistered research, found large overestimates of rival partisans' willingness to endorse violence and reported reductions after informational correction. These findings address different estimands: the measurement study evaluates how valid an apparent prevalence estimate is, whereas the correction experiment evaluates responses to a treatment. Both require attention to their questionnaire wording and sampling procedures.

A rigorous violence instrument should define whether the act is offensive, retaliatory, self-defensive, lawful, or directed at property or people. It should separate judgments about an actual event from generalized approval of violence; describe response scales with meaningful anchors; and consider whether socially undesirable answers are concealed or provocatively exaggerated. Voelkel et al. (2024) explicitly measured partisan animosity, support for undemocratic practices, and support for partisan violence as separate outcomes. The incomplete correspondence of intervention effects across them demonstrates why a study about political violence cannot automatically be used to estimate support for electoral rule breaking.

Sampling, Cross-National Comparability, and Causal Limits

A finding from a quota-matched American partisan sample addresses a different population from one about politicians in a multiparty parliamentary system. Democratic institutions vary in who administers elections, how courts constrain executives, how parties coordinate, and what counts as an exceptional emergency power. These differences alter respondents' interpretation of apparently identical survey items. Transporting a result requires the investigator to identify whether the same psychological construct, equivalent institutional act, and comparable cost structure exist in the new setting. McCoy et al. (2018) shows why severe polarization should be treated as a relational and political process with multiple historical forms.

Probability recruitment, high-quality nonprobability panels, convenience crowdsourcing, and elite surveys each have strengths and limitations. Descriptions must specify selection, exclusions, weighting, treatment assignment, and attrition rather than rely on a broad claim of national representativeness. Voelkel et al. (2024) used a large quota-matched sample of American Democrats and Republicans; that scale provides useful precision for its experimental comparisons but does not establish prevalence in every national population. Druckman et al. (2023) sampled state legislators, an informative but structurally distinct decision-making group. Findings about groups differ also in their baseline opportunities to act on undemocratic preferences.

Causal inference must follow the exact source of exogenous variation. Randomized information can establish whether that information changed a measured attitude in the study conditions; it cannot by itself establish how the attitude emerged over a lifetime. Candidate attribute randomization identifies the causal effect of a hypothetical description, while observational comparisons of strong and weak partisanship remain vulnerable to confounding and reverse causality. Well-founded scientific synthesis records this evidence hierarchy and explores whether reported effects are robust to item definitions, subgroup choices, alternate model specifications, and preregistered versus exploratory analyses.

The Graham–Svolik Electoral Evidence and Its Interpretation

The study by Graham & Svolik (2020) is important because its outcome is electorally consequential choice rather than only general democratic approval. Respondents evaluated candidates with features that could appeal to ideological interests while some candidates endorsed practices inconsistent with democratic principles. The study examined how the electoral penalty for transgressions changed as partisanship, policy extremism, and platform divergence varied. Its overarching result was that democratic principles often received limited priority in voters' selections. The natural-experiment component associated with the 2017 Montana congressional special election provided a setting in which attitudes toward a real candidate could be investigated alongside the controlled experimental evidence.

This work supports a tradeoff interpretation, but the design cannot reveal a respondent's full private reasoning without supplemental measurement. A citizen who chooses a rule-breaking co-partisan might believe the accusation exaggerated, judge a particular action constitutionally permissible, or regard the rival's policies as unacceptable. Some accounts are consistent with instrumental preference; others entail contested perceptions of facts or institutions. The electoral result alone cannot distinguish them. A careful report therefore identifies candidate attributes, framing, available alternatives, measurement dates, and sampling strategy before inferring motivational mechanisms. The central causal result concerns described positions and choice behavior under experimental assignment.

The political implication is conditional and testable. If voters penalize violations weakly when party stakes are high, politicians may expect to survive electoral accountability despite procedural transgressions. If cross-cutting considerations or strong shared commitments increase the penalty, the electoral check may be more credible. Those predictions depend on the electoral system and the opportunity to replace an incumbent. Svolik (2019) supplies the broader comparative model connecting partisan polarization to democratic vulnerability, while Carey et al. (2022) studies tradeoffs in candidate support among voters and partisan donors. The combined literature argues for observing the costs of applying democratic principles rather than relying exclusively on the stated popularity of democracy.

The Kingzette and Simonovits Findings: Contextual Enforcement

The contrast between evaluations of in-party and out-party power provides a more precise test of asymmetric norm enforcement than a comparison of raw partisan means. Kingzette et al. (2021) found that the relation between affective polarization and support for constitutional protections depended on the presidency's partisan affiliation. The temporal comparison included contexts with different parties in national executive office, exposing a pattern consistent with elite cue-taking. Its principal contribution is evidence that norm judgments are politically situated. Because the study draws on survey data rather than a complete manipulation of real governing context, the mechanism is best understood as supported by the observed pattern and modeling rather than as uniquely established.

Simonovits et al. (2022) addressed the context problem experimentally by constructing hypothetical situations that distinguish whether the participant's own party holds power. The study found greater willingness to accept democratic erosion when a respondent's party benefited, with stronger effects under expressive partisanship and perceived out-group threat. This adds experimental leverage to the claim that political position conditions adherence to principles. It also challenges simple theories in which a fixed preference for authoritarianism explains every violation. An otherwise norm-supportive citizen may become permissive under a particular distribution of political benefit, while another remains consistently restrictive.

The studies together motivate a research program on cross-situational consistency. Do individuals preserve their rule preferences when the beneficiary, party name, severity of act, and policy consequences are independently randomized? Does compliance persist after an election and during an emergency? Would the same individual penalize a favored leader for tampering with elections but tolerate overreach against a threatening media organization? Such questions make democratic commitment a structured set of conditional decisions. Any average index that suppresses these interactions risks misclassifying both committed proceduralists and consistently partisan rule opportunists.

The Braley Studies: Expectations of Opponents and Defensive Erosion

Braley et al. (2023) proposed that some democratic backsliding originates in inflated expectations of opponents' willingness to violate democracy. The research combined observational analysis of American partisan beliefs with two experiments providing corrective information. Its reported observational sample included 1,973 participants, while the experiments involved 2,543 and 1,848 participants. The manipulated information indicated that opposing partisans were more democratically committed than the participant had anticipated. The authors found greater stated commitment to democratic norms and reduced willingness to select candidates who broke those norms after correction.

The study advances the field by testing an epistemic mechanism rather than inferring it from general partisan hostility. The treatment's plausible pathway runs from corrected second-order beliefs, through reduced expectations of unilateral vulnerability, to a lower willingness to excuse one's own side's violation. Nonetheless, the observed changes are tied to a particular information presentation and time horizon. Respondents may update in the presence of an experimental display and then encounter contradictory elite messages outside the study. The data establish a promising causal intervention in specified conditions, not a universal solution to the incentive to violate democratic rules.

A theoretically useful implication is to distinguish an opponent's actual preference from a respondent's belief about it. The two can diverge markedly even when overall party hostility stays high. A future factorial experiment could independently vary source credibility, reported opposing-party commitment, the actor's strategic incentives, and an opportunity to impose an electoral sanction. This would clarify whether expectation correction works principally by reducing fear, by changing perceived social norms, or by making restraint strategically less costly. A single post-treatment approval scale cannot conclusively isolate these mediating paths.

The Voelkel Intervention Program: Outcome Specificity

Voelkel et al. (2023) tested whether interventions designed to decrease affective polarization necessarily improve antidemocratic attitudes. The research showed that reducing animosity between party groups can occur without corresponding improvements on the latter outcomes. This is an important falsification of an overly broad psychological model. Partisan hostility may contribute to political mistrust or social distance, yet support for a particular violation can involve distinct justifications, institutional expectations, or strategic calculations. The appropriate intervention target must therefore match the measured democratic harm rather than rely on a general assumption that friendlier partisan feelings will improve every political behavior.

Voelkel et al. (2024) expanded the intervention agenda through a large experimental megastudy of 25 treatments selected from 252 submitted proposals, using 32,059 quota-matched American Democratic and Republican participants. The preregistered analysis found that 23 treatments significantly reduced partisan animosity, while six significantly reduced support for undemocratic practices and five reduced support for partisan violence. The maximum reported shifts were approximately 10.5, 5.8, and 2.8 percentage points, respectively, across these distinct outcomes. These are outcome-specific maxima rather than guaranteed effects of a single universal intervention, and the relevant scales and comparisons must be consulted when discussing practical magnitude.

Strategies with reported benefits for support for undemocratic practices included correcting exaggerated estimates of rival partisans' beliefs and emphasizing the risks of democratic collapse. Strategies associated with reduced violence included selected belief corrections and democratic-principle messages from elites. Some treatments had broader benefits, including willingness to support undemocratic candidates, but correlations among the effects suggested that partisan animosity and several antidemocratic outcomes are not interchangeable constructs. The megastudy's size and preregistered design strengthen its internal comparisons, while limited follow-up and the American online survey environment constrain claims about long-term behavioral change. Its major theoretical contribution is the empirical separation of endpoints that political commentary frequently conflates.

Contradictory Replications and the Misperception-Correction Debate

Positive results from individual information-correction experiments must be interpreted alongside tests of robustness. Mernyk et al. (2022) reported that corrected perceptions of rivals' support for violence decreased both approval of partisan violence and self-reported willingness to engage in it across experimental studies, with a follow-up effect in one study. Its preregistered program included 4,741 participants across correlational, longitudinal, and experimental components. Druckman et al. (2023) subsequently reported beneficial correction effects among state legislators. These studies support the hypothesis that inaccurate second-order beliefs can sometimes sustain anti-democratic attitudes, but they differ in population, outcome measures, intervention framing, and follow-up structure.

Druckman (2023) tested the resilience of misperception corrections when recipients also encounter competing information. The author reported that the corrective messages did not reliably reduce support for undemocratic practices or partisan violence under those more contested informational circumstances. Dias et al. (2024) offered a further challenge using panel evidence, a meta-analytic assessment of published tests, and new preregistered experiments. The latter paper reported a panel of 9,810 partisans, a synthesis of 67 statistical tests from 12 studies in eight papers, and experiments involving 2,846 respondents; changing out-party perceptions did not consistently move the targeted antidemocratic outcomes. Its own meta-analytic results require interpretation according to heterogeneity and the studies included rather than as a definitive proof of zero effects.

The disagreement is scientific evidence about boundary conditions. Message corrections can create immediate belief updating in a controlled environment without reliably influencing a distant action, surviving hostile counter-messaging, or producing durable change at population scale. Effects may also differ for beliefs about opponents' violence, procedural transgressions, and ordinary policy positions. Research should preregister the mechanism and outcome, measure whether beliefs actually changed, compare correction-only with contested-information conditions, and estimate persistence in multiple populations. Both the positive interventions and the robustness critiques must remain in a faithful evidence synthesis; selecting only one would distort the current state of the field.

Democratic Commitment Beyond the United States

Political psychology requires comparative evidence because the same verbal endorsement of democracy can have different behavioral implications under different institutional environments. Electoral systems vary in party fragmentation, ballot design, administrative neutrality, judicial enforcement, and the social meaning of political opposition. The relevant psychological variable is therefore not simply national identification with democracy but an individual's conditional application of rules under credible local costs. McCoy et al. (2018) examines how political entrepreneurs can align social divisions into comprehensive adversarial camps. Its comparative framework helps generate hypotheses about elite polarization and reciprocal escalation, while leaving specific effect sizes to country-level primary research.

Svolik (2023) studied the rerun of Istanbul's 2019 mayoral election after the governing party's attempt to overturn its earlier loss. Combining the quasi-experimental electoral situation with survey and experimental evidence, the study distinguishes voter defection to the opposition, opposition mobilization or backlash, and withdrawal of turnout support. These mechanisms matter because a net loss in incumbent vote share can arise from different psychological choices. The empirical case demonstrates that voters sometimes impose meaningful costs on actors perceived as undermining electoral integrity, even in a highly polarized environment. Its unusual electoral sequence makes the case informative but does not make it representative of all episodes of democratic erosion.

Comparative inference should attend to the stage of regime change and the available means of sanction. A voter can punish an incumbent through elections only when electoral competition and vote counting retain enough integrity to translate preferences into officeholding. In other systems, fear of repression, restricted media, or fragmented opposition can constrain expressive and behavioral measures. The same citizen may report support for democratic rules yet regard an opposition victory as unattainable. Research must therefore combine survey psychology with credible institutional analysis. It should also examine whether questions about rights and election procedures have equivalent meanings across languages and legal traditions.

Severity, Salience, and the Form of the Undemocratic Act

An explicit pledge to end competitive elections is cognitively and morally different from a technically framed amendment to election administration or the strategic appointment of partisan administrators. Respondents may recognize the former as threatening democracy while interpreting the latter as an ordinary partisan measure. Dahlum et al. (2025) studied reactions to norm violations in Norway and reported comparatively weak punishment for some concrete infractions, including violent rhetoric, alongside substantially stronger punishment for explicit rejection of democracy. The study also tested justifications involving pressing political goals. Its findings suggest that perceived violation severity and recognizability can matter independently of simple party attachment, although particular scenarios and their national context determine the scope of inference.

Teas (2026) examined why American respondents may tolerate norm violations that advantage their political side. Two experiments with a combined reported sample of 2,352 compared partisan advantage, perceived democratic legitimacy, and values such as access to voting versus election security. The study found rationalization of favorable violations among both major-party groups, with asymmetries that changed across the particular violations tested. This result is directly relevant to construct validity: a partisan gap observed on a mail-in-ballot item should not automatically be generalized to media censorship or refusal to transfer power. Each practice may engage different values and background factual beliefs.

The practical measurement principle is a severity-by-beneficiary matrix. Randomize relatively subtle and explicit acts, in-party and out-party beneficiaries, and neutral or emergency rationales. Measure whether the act is understood as a democratic violation before asking about support. Such comprehension measures can themselves affect later judgments, so measurement order should also be tested. Without this separation, investigators may confuse indifference to democracy with disagreement over whether the described conduct violates democratic rules. An excellent democratic-norm experiment contains a defensible institutional classification and a psychological measure of how the respondent construes that classification.

Electoral Choice Architecture and Democratic Balancing

Most candidate-choice experiments represent a single binary contest, while real voters often make multiple decisions across offices on one ballot. A voter's overall electoral choice can distribute party preference and institutional concerns across these contests. Huang et al. (2026) directly manipulated whether respondents selected candidates in single-race or multirace ballot environments and estimated the penalty for antidemocratic candidate positions. The researchers reported that the total punishment effect rose from approximately 11% in single-race settings to 17.9% in multirace settings, with related evidence drawn from the 2022 American midterm elections. These exact values belong to the published experimental comparison, not to a universal rate of democratic commitment.

The result matters for theory because it alters the choice environment while holding the general norm topic in view. Respondents might be more willing to penalize a problematic candidate when another race provides an opportunity to support their preferred party or policy, a pattern the authors call democratic balancing. This provides a concrete institutional moderator of the tradeoff documented in Graham & Svolik (2020). Individual norms have electoral force only through the decision architecture that connects them to possible votes. Psychological research on the resilience of democracy therefore benefits from formal models of substitution, compensatory choice, and the perceived consequences of divided-party representation.

The finding also suggests important tests. Do voters balance across offices when the perceived risk from one candidate is extremely severe? Does a multirace format increase information overload or enhance the visibility of procedural choices for less politically engaged citizens? Is the effect preserved under multiparty proportional representation? Are respondents' explanations consistent with deliberate balancing or do they simply respond to framing and repeated choice exposure? The 2026 experiment provides a causal estimate for ballot context in its study conditions; answering these extensions will require new cross-national designs and observed election behavior.

Winners, In-Party Favoritism, and Democratic Rationalization

Political victory can change more than the instrumental value of a procedural rule. Winning may intensify positive feelings toward the political in-group, increase perceptions of its legitimacy, and lower the threshold at which the group's actions are described as democratic. Huang (2026) used a mediator-blockage survey experimental approach to study the effects of anticipating electoral victory in the United States. The article reported that heightened in-party favoritism mediated increased approval of policies such as banning protests or disqualifying candidates, with respondents also more inclined to interpret those policies as democratic.

This finding deepens the classical tradeoff model. Some support for violation may involve a reconstructed understanding of what counts as a legitimate democratic procedure rather than a conscious admission that an anti-democratic action is worth a policy advantage. The distinction is empirically consequential: if a respondent sincerely classifies an exclusion as democratic, a message describing its instrumental cost may be ineffective, whereas an explanation of procedure and rights may be more relevant. However, mediation estimates depend on assumptions about confounding and how the mediator is manipulated, and one experiment cannot establish a universal psychological trajectory from winning to norm erosion.

Future studies should distinguish increased in-group favoritism after victory from elevated political efficacy, lowered threat, celebration, and trust in government. These mechanisms may pull in opposite directions; victory could reduce defensive fear while simultaneously increasing permissiveness toward the winning party. Measuring each process, testing mediator robustness, and following participants through actual transitions of power would clarify the balance. Simonovits et al. (2022) and Kingzette et al. (2021) provide complementary evidence that the party holding office is central to normative judgments. Together they indicate that the political environment is part of the psychological construct's expression.

Public Opposition to Erosion and the Elite–Mass Disconnect

Evidence that voters trade off principles must coexist with evidence that many voters strongly reject explicit democratic violations. Holliday et al. (2024) analyzed a repeated cross-sectional dataset of 45,095 respondents and panel data from 5,231 respondents in the United States. The authors reported that overwhelming majorities opposed democratic norm violations and that support for partisan violence was rare under their measurement approach, despite continuing partisan hostility. They also found little evidence of a general public shift toward antidemocratic views in the period examined. This result challenges broad claims that widespread voter enthusiasm for authoritarian measures is an inevitable consequence of polarization.

The tension with candidate-choice studies is partly a matter of outcome definition. A scale measuring approval of specified violations asks whether those actions are accepted; a conjoint vote can reveal a willingness to tolerate a candidate's objectionable feature while still rejecting the feature itself. Graham & Svolik (2020) and Holliday et al. (2024) can therefore be understood as probing different margins of democratic accountability. Westwood et al. (2022) adds a measurement critique of inflated estimates of political-violence support. A synthesis should resist equating willingness to make a conditional electoral tradeoff with pervasive personal endorsement of intimidation or physical aggression.

The elite–mass distinction creates another complication. Officeholders may act against democratic constraints without enthusiastic mass demand, relying on low voter attention, procedural complexity, party discipline, or limited electoral alternatives. As Holliday et al. (2024) notes, public rejection of antidemocratic attitudes does not automatically prevent elite-led erosion. Democracy depends on the actions of leaders and institutions as well as popular values. Researchers should observe organizational decisions, enforcement capacity, party nomination processes, and public responses to specific violations. The existence of principled majorities may be normatively important yet insufficient to predict the outcome of any given institutional struggle.

Evidence Quality and the Limits of Scientific Agreement

The field supports several moderately secure propositions: political allegiance can condition the application of democratic norms; candidate-choice experiments reveal tradeoffs between institutional restraints and desirable partisan or policy outcomes; misperceptions of opponents can sometimes alter support for violations; and partisan animosity, undemocratic practices, and violence exhibit different response patterns. Their evidentiary strength is not identical. Graham & Svolik (2020) offers randomized candidate attributes and an electoral case; Simonovits et al. (2022) manipulates governing-party advantage; Voelkel et al. (2024) directly compares many treatments against multiple outcomes. Each yields credible inferences within the treatment, measurement, and sample conditions specified by its authors.

Some central questions remain unsettled. Mernyk et al. (2022) and Braley et al. (2023) document beneficial effects of targeted corrections under specific conditions, whereas Druckman (2023) and Dias et al. (2024) report important constraints on robustness. Holliday et al. (2024) finds uncommon endorsement of antidemocratic acts in large public opinion datasets, while Graham & Svolik (2020) identifies limited electoral sanction under competing party and policy incentives. These findings need not be forced into one prevalence estimate because the underlying designs assess different actions and counterfactuals. A rigorous synthesis first aligns constructs and estimands, then examines disagreements that remain after alignment.

Evidence appraisal also requires scrutiny of publication selection, outcome multiplicity, scale interpretation, geographic coverage, and intervention follow-up. The largest sample does not automatically have the strongest causal leverage, and the cleanest randomization may study an unrealistic scenario. Reports should document whether numerical effects are absolute percentage-point differences, relative percentages, standardized effects, or model estimates. They should avoid combining these quantities without conversion and uncertainty analysis. Moreover, a causal effect on a self-report in a brief online study should not be converted into a claim that electoral law, protest behavior, or government conduct has changed.

Research Ethics, Interpretation, and Responsible Political Science

Research involving democratic attitudes must preserve the legitimacy of disagreement over ordinary policies while examining violations of the procedures governing that disagreement. Operational definitions should not equate membership in a party, political ideology, religious belief, nationality, or dissenting constitutional interpretation with support for violence or a willingness to destroy democratic competition. An experiment can investigate the impartial application of election rules without attempting to shape participants' party preferences. The investigator's responsibility is to identify observable procedures, accurately characterize empirical relationships, and disclose how choices of examples and coding affect outcomes.

Experimental work must avoid creating risks through inflammatory stimulus material, misleading statements about election procedures, or unsupported claims that a population is predisposed toward violence. Deception-based designs need ethical review where required and careful debriefing. Sensitive survey responses should receive appropriate privacy protection, especially in political environments where expression carries legal or social risk. Researchers should publish precise stimuli, preregistered outcomes when applicable, anonymized data where ethical, and replicable code while safeguarding subjects. Cross-cultural translation requires local subject expertise and attention to political vulnerability.

Scientific neutrality is compatible with explicit classification of an undemocratic act according to a defensible procedural criterion. The study should disclose that criterion, maintain symmetrical evaluation across parties, and examine the uncertainty inherent in real-world law and constitutional practice. The same standard applies when an intervention is assessed: its value is determined by the predeclared outcome and its measured effect, rather than by whether it assists one electoral coalition. An academically useful monograph is an instrument for understanding psychological mechanisms and evidence, not a manual for manipulating a population's political choices.

Artificial Era: Information Mediation and Democratic Norm Recognition

Digital platforms and generative artificial systems increasingly mediate how people encounter claims about election integrity, institutional authority, and the intentions of political opponents. Within the Artificial Era, the relevant psychological problem is how human judgments of democratic legitimacy change when information is selected, summarized, ranked, or conversationally presented by artificial agents. The mechanisms investigated in this monograph—social identity, threat expectations, perceived consensus, and authority cues—provide empirically grounded hypotheses for studying these mediated environments. Their application to a particular artificial system requires independent testing; results from a human political interview or party-message experiment cannot be assumed to transfer directly to an automated interface.

Research should distinguish exposure to a claim, credibility assigned to its source, comprehension of the underlying procedure, correction of an inaccurate belief, and eventual support for an institutional act. Each is an experimentally separable stage. The mixed results of misperception corrections in Braley et al. (2023), Druckman (2023), and Dias et al. (2024) caution against assuming that presenting more accurate information will necessarily stabilize norms in environments with competing signals. A useful program would randomize source identification, evidence transparency, conversational personalization, timing, and the availability of corroboration, while measuring both belief accuracy and downstream judgments.

The Aisentica horizon of From Homo to Artificial supplies an interdisciplinary context for new forms of information mediation and institutional interaction. The empirical claims of this dossier remain grounded in psychology and political science. Concepts such as Artificial Agency and Human–Artificial Relations can motivate future research on accountability for generated political explanations, user trust, and errors in attributed authority. Any stronger claim about artificial cognition, political autonomy, or a novel nonbiological source of legitimacy would require operational definitions and independent evidence specific to those systems.

Priority Research Questions and a Cumulative Methodological Agenda

A cumulative science of democratic commitment should prioritize direct comparisons of the same respondent's abstract principles, approval of specific violations, candidate choice, and subsequent behavior. Such designs can estimate the share of apparent discrepancy attributable to measurement rather than motivational inconsistency. The democratic balancing finding in Huang et al. (2026) indicates that the available electoral choices are an experimentally consequential context, while Teas (2026) shows how the specific violation changes observed asymmetries. Replication protocols should vary ballot structure, legal severity, party advantage, political information, and country simultaneously where feasible. Cross-national measurement invariance and institutional specificity must be preconditions for pooling.

A second priority is to identify interventions that survive real informational competition. The contrast between Voelkel et al. (2024) and Dias et al. (2024) suggests testing single-message corrections against repeated, conflicting, identity-congruent, and independently verified messages. Outcomes should be measured immediately, after weeks or months, and where feasible through documented decisions. Large preregistered multi-site studies could compare the effect of norm reminders, factual clarifications, credible co-partisan restraint, procedural explanations, and reductions in perceived threat without combining them into an unexplained treatment package. The design should distinguish improvements in civility from changes in support for democratic restrictions and from actual political behavior.

Further priorities include distinguishing normative conviction from beliefs about legality; measuring political elites and citizens in connected samples; studying people who reject party identification; testing how severe and subtle violations are recognized; and modeling the transition from approval to action under institutional constraints. Survey research should be supplemented by field experiments and careful analyses of natural institutional episodes. Researchers also need better understanding of when fear reflects an exaggerated stereotype and when it responds to independently verified threats. Such distinctions will improve causal explanation while preventing inaccurate reassurance or unjustified alarm.

Conclusion

Democratic norms are psychologically meaningful when they organize judgments and choices under partisan disadvantage, uncertainty, and real institutional consequences. Research reveals both principled commitments and conditional exceptions. Political identities, perceived threats, expectations about opponents, elite communications, and the design of electoral choices can each influence the willingness to sanction procedural violations. The classic civil-liberties literature introduced the problem of tolerance toward disliked groups; contemporary experiments investigate the corresponding problem of democratic restraint when one's political side may lose power. The resulting scientific object is a set of related but distinguishable psychological decisions rather than a single generalized attitude.

The evidence has established important contrasts. Graham & Svolik (2020) documents electoral tradeoffs; Simonovits et al. (2022) identifies power-conditioned democratic hypocrisy; Braley et al. (2023) reports that correcting exaggerated fears can reduce certain violations; Voelkel et al. (2024) demonstrates that reducing partisan animosity and reducing antidemocratic attitudes require partly different interventions. At the same time, Holliday et al. (2024) finds broad public rejection of explicit violations, Westwood et al. (2022) emphasizes sensitivity of violence-prevalence estimates to measurement, and Dias et al. (2024) highlights limits to information correction. Huang et al. (2026) further shows that ballot structure can change the estimated punishment of undemocratic candidates. The synthesis is therefore a model of conditional commitment, institutional opportunity, and outcome specificity.

The most productive future direction combines validated norm measures, credible behavioral choices, symmetric counterfactuals, elite and citizen samples, cross-national institutional analysis, and replication under competing information. A robust account will specify what people believe democracy requires, which actions they classify as violations, what they expect opponents to do, and what costs they will bear to uphold the rules. It will trace when institutional settings translate moral and procedural commitments into effective constraints on officeholders. This framework permits precise analysis of democratic resilience and erosion while preserving the empirical distinctions on which reliable political psychology depends.

References

  • Braley, A., Lenz, G. S., Adjodah, D., Rahnama, H., & Pentland, A. (2023). Why voters who value democracy participate in democratic backsliding. Nature Human Behaviour, 7, 1282–1293. https://doi.org/10.1038/s41562-023-01594-w
  • Carey, J. M., Clayton, K., Helmke, G., Nyhan, B., Sanders, M., & Stokes, S. (2022). Who will defend democracy? Evaluating tradeoffs in candidate support among partisan donors and voters. Journal of Elections, Public Opinion and Parties, 32(1), 230–245. https://doi.org/10.1080/17457289.2020.1790577
  • Dahlum, S., Langsæther, P. E., & Wig, T. (2025). New evidence on citizens’ reactions to democratic norm violations in an advanced democracy. Electoral Studies, 97, Article 102971. https://doi.org/10.1016/j.electstud.2025.102971
  • Dias, N. C., Aarslew, L. F., Skaaning Frederiksen, K. V., Lelkes, Y., Pradella, L., & Westwood, S. J. (2024). Correcting misperceptions of partisan opponents is not effective at treating democratic ills. PNAS Nexus, 3(8), pgae304. https://doi.org/10.1093/pnasnexus/pgae304
  • Druckman, J. N. (2023). Correcting misperceptions of the other political party does not robustly reduce support for undemocratic practices or partisan violence. Proceedings of the National Academy of Sciences, 120(37), e2308938120. https://doi.org/10.1073/pnas.2308938120
  • Druckman, J. N., Kang, S., Chu, J., Stagnaro, M. N., Voelkel, J. G., Mernyk, J. S., Pink, S. L., Redekopp, C., Rand, D. G., & Willer, R. (2023). Correcting misperceptions of out-partisans decreases American legislators’ support for undemocratic practices. Proceedings of the National Academy of Sciences, 120(23), e2301836120. https://doi.org/10.1073/pnas.2301836120
  • Finkel, E. J., Bail, C. A., Cikara, M., Ditto, P. H., Iyengar, S., Klar, S., Mason, L., McGrath, M. C., Nyhan, B., Rand, D. G., Skitka, L. J., Tucker, J. A., Van Bavel, J. J., Wang, C. S., & Druckman, J. N. (2020). Political sectarianism in America. Science, 370(6516), 533–536. https://doi.org/10.1126/science.abe1715
  • Graham, M. H., & Svolik, M. W. (2020). Democracy in America? Partisanship, polarization, and the robustness of support for democracy in the United States. American Political Science Review, 114(2), 392–409. https://doi.org/10.1017/S0003055420000052
  • Holliday, D. E., Iyengar, S., Lelkes, Y., & Westwood, S. J. (2024). Uncommon and nonpartisan: Antidemocratic attitudes in the American public. Proceedings of the National Academy of Sciences, 121(13), e2313013121. https://doi.org/10.1073/pnas.2313013121
  • Huang, K., Lovett, M., & Helmke, G. (2026). The electoral choice context and support for democratic norms. Proceedings of the National Academy of Sciences, 123(19), e2521380123. https://doi.org/10.1073/pnas.2521380123
  • Huang, Y.-S. (2026). Love blinds? Winners, in-party favoritism, and support for violations of democratic norms. British Journal of Political Science, 56, e35. https://doi.org/10.1017/S0007123426101574
  • Iyengar, S., Sood, G., & Lelkes, Y. (2012). Affect, not ideology: A social identity perspective on polarization. Public Opinion Quarterly, 76(3), 405–431. https://doi.org/10.1093/poq/nfs038
  • Kingzette, J., Druckman, J. N., Klar, S., Krupnikov, Y., Levendusky, M., & Ryan, J. B. (2021). How affective polarization undermines support for democratic norms. Public Opinion Quarterly, 85(2), 663–677. https://doi.org/10.1093/poq/nfab029
  • Levitsky, S., & Ziblatt, D. (2018). How democracies die. Crown. https://levitsky.scholars.harvard.edu/publications/how-democracies-die
  • Linz, J. J., & Stepan, A. (Eds.). (1978). The breakdown of democratic regimes: Crisis, breakdown and reequilibration: An introduction. Johns Hopkins University Press. https://doi.org/10.56021/9780801820090
  • McCoy, J., Rahman, T., & Somer, M. (2018). Polarization and the global crisis of democracy: Common patterns, dynamics, and pernicious consequences for democratic polities. American Behavioral Scientist, 62(1), 16–42. https://doi.org/10.1177/0002764218759576
  • Mernyk, J. S., Pink, S. L., Druckman, J. N., & Willer, R. (2022). Correcting inaccurate metaperceptions reduces Americans’ support for partisan violence. Proceedings of the National Academy of Sciences, 119(16), e2116851119. https://doi.org/10.1073/pnas.2116851119
  • Przeworski, A. (1991). Democracy and the market: Political and economic reforms in Eastern Europe and Latin America. Cambridge University Press. https://doi.org/10.1017/CBO9781139172493
  • Simonovits, G., McCoy, J., & Littvay, L. (2022). Democratic hypocrisy and out-group threat: Explaining citizen support for democratic erosion. The Journal of Politics, 84(3), 1806–1811. https://doi.org/10.1086/719009
  • Stouffer, S. A. (1955). Communism, conformity, and civil liberties: A cross-section of the nation speaks its mind. Doubleday. https://books.google.com/books?id=FwseAAAAMAAJ
  • Sullivan, J. L., Piereson, J., & Marcus, G. E. (1982). Political tolerance and American democracy. University of Chicago Press. https://press.uchicago.edu/ucp/books/book/chicago/P/bo3775122.html
  • Svolik, M. W. (2019). Polarization versus democracy. Journal of Democracy, 30(3), 20–32. https://www.journalofdemocracy.org/articles/polarization-versus-democracy/
  • Svolik, M. W. (2023). Voting against autocracy. World Politics, 75(4), 647–691. https://doi.org/10.1353/wp.2023.a908772
  • Teas, P. E. (2026). Partisan or principled? Explaining political differences in attitudes about violations of democratic norms. Personality and Social Psychology Bulletin, 52(10), 3496–3511. https://doi.org/10.1177/01461672251372671
  • Voelkel, J. G., Chu, J., Stagnaro, M. N., Mernyk, J. S., Redekopp, C., Pink, S. L., Druckman, J. N., Rand, D. G., & Willer, R. (2023). Interventions reducing affective polarization do not necessarily improve anti-democratic attitudes. Nature Human Behaviour, 7, 55–64. https://doi.org/10.1038/s41562-022-01466-9
  • Voelkel, J. G., Stagnaro, M. N., Chu, J. Y., Pink, S. L., Mernyk, J. S., Redekopp, C., Ghezae, I., Cashman, M., Adjodah, D., Allen, L. G., Allis, L. V., Baleria, G., Ballantyne, N., Van Bavel, J. J., Blunden, H., Braley, A., Bryan, C. J., Celniker, J. B., Cikara, M., ... Willer, R. (2024). Megastudy testing 25 treatments to reduce antidemocratic attitudes and partisan animosity. Science, 386(6719), eadh4764. https://doi.org/10.1126/science.adh4764
  • Westwood, S. J., Grimmer, J., Tyler, M., & Nall, C. (2022). Current research overstates American support for political violence. Proceedings of the National Academy of Sciences, 119(12), e2116870119. https://doi.org/10.1073/pnas.2116870119