The “Hostage Voter” Problem in the United States: Negative Partisanship, Strategic Voting, and Two-Party Entrapment

Executive assessment

The phenomenon described as a “hostage voter” is not a standard term in political science, but the proposed mechanism can be decomposed into several well-established phenomena: negative partisanship, affective polarization, strategic voting, electoral coordination under plurality rules, spoiler avoidance, and negative or “against” voting. The strongest conclusion from the literature is that the component mechanisms are real, empirically measurable, and common enough to matter in U.S. elections. What is much less firmly established is the strongest system-level version of the hypothesis: that the contemporary two-party system itself causes negative partisanship, which then recursively suppresses alternatives and intensifies negative partisanship in a self-sustaining loop. Evidence for that complete causal feedback cycle is suggestive rather than conclusive. ^{source ref}

The most important conceptual point is that three things that are often conflated must be separated:

Political scientists have increasingly shown that positive attachment and negative aversion are empirically distinguishable. Alexa Bankert's Political Behavior study, for example, measures positive and negative partisan identity separately and finds that they are distinct constructs with different consequences for participation, bipartisanship, and vote choice. Comparative work by Medeiros and Noël similarly treats negative party identification as something more than simply the inverse of positive party identification. ^{source ref}

That distinction matters because a Democrat who passionately dislikes Republicans but genuinely prefers Democrats to every alternative is not, in the strict strategic-voting sense, a hostage voter. Nor is a Republican who likes Republicans and strongly dislikes Democrats. The hypothesized condition is more demanding: the voter would rather support some alternative, believes that alternative cannot realistically prevail, and therefore chooses a major party because the perceived cost of the other major party winning is especially large.

There is good evidence that many Americans do vote substantially against rather than for. Pew found that in 2016, 33% of Trump supporters and 32% of Clinton supporters said opposition to the other candidate was a main factor in their likely vote. By 2020, opposition to Trump was the most commonly cited reason among Biden supporters. At the same time, the pattern was asymmetrical that year: Trump supporters more often cited Trump's performance, issues, or partisan considerations affirmatively, whereas Biden's coalition contained especially strong anti-Trump motivation. ^{source ref}

There is also powerful evidence that voters perceive major-party defeat as an unusually serious loss. Immediately before the 2020 election, 90% of Biden supporters said a Trump victory would result in lasting harm to the country, while 89% of Trump supporters said the same about a Biden victory. Those figures are almost a textbook expression of the proposed “fear of the opposing coalition” mechanism. ^{source ref}

At the same time, dissatisfaction with both choices is not hypothetical. In Pew's 2023 survey, 28% of Americans viewed both major parties unfavorably, the highest share in three decades of that polling series, while 25% said neither party represented them well. Sixty-three percent were dissatisfied with the presidential candidates who had emerged at that point. ^{source ref} In the Biden–Trump phase of the 2024 election, roughly a quarter of Americans had unfavorable views of both candidates. ^{source ref}

Those facts make the “politically trapped” feeling empirically plausible. But they do not establish that the parties are merely holding voters through fear. U.S. voters also perceive large substantive differences between the major coalitions. In 2020, 86% of registered voters told Pew that Trump and Biden differed on the issues, dramatically higher than in earlier decades. ^{source ref} Negative partisanship can therefore reflect both an institutional strategic constraint and sincere judgments about consequential policy, social, or democratic differences. The two interpretations are not mutually exclusive.

My overall assessment is therefore:

Negative partisanship provides a credible and important mechanism connecting a plurality-dominated two-party system to the subjective experience of political entrapment. It can reinforce two-party voting even among people who are dissatisfied with both parties. But the evidence supports this most strongly as a mechanism affecting a significant subset of voters, not as a complete explanation of American two-party stability or contemporary polarization.

The two-party system makes defensive voting rational in circumstances where third alternatives are perceived as nonviable; negative partisanship raises the perceived cost of defecting from one's major-party coalition. Those mechanisms reinforce each other. Yet affirmative party preferences, real policy differences, partisan identities, candidate evaluations, elite polarization, and social identities remain independently important. ^{source ref}

What the “hostage voter” mechanism actually requires

Negative partisanship is not simply weak positive partisanship

Traditional studies of party identification treated partisanship largely as a durable psychological attachment to a party. Newer work asks whether citizens may also possess an independent identity against another party. Medeiros and Noël's comparative analysis of four Anglo-American democracies found that negative party identification contributes separately to explaining electoral choice, alongside positive identification. ^{source ref}

Abramowitz and Webster's influential work on the United States argues that growing negative partisanship has been associated with greater party loyalty, nationalization of elections, and straight-ticket behavior. Their central point is directly relevant here: citizens can express intensely partisan electoral behavior even when their attachment is substantially driven by hostility toward the opposing party. ^{source ref}

Bankert subsequently developed a multi-item approach that measures positive and negative partisan identity independently. Her analysis of the 2016 U.S. election finds that the positive and negative dimensions are related but not interchangeable and that they have different political effects. That is important evidence against treating “votes Democrat” or “votes Republican” as automatically equivalent to affirmative support for the chosen party. ^{source ref}

There is nevertheless a live measurement dispute. Yphtach Lelkes argues that scholars have used “negative partisanship” for conceptually different things, including simple disdain for the out-party and a more identity-like conception of defining oneself in opposition to another party. Those constructs need not behave identically, and evidence that negative attitudes have become intense does not by itself prove that negative identity has overtaken positive identity. ^{source ref}

That caution is important for the hostage-voter hypothesis. Someone can hate Party B while also enthusiastically identifying with Party A. That voter is negatively partisan, but not obviously trapped.

Affective polarization is broader than strategic voting

Affective polarization describes the growing gap in citizens' feelings toward the political in-group and out-group. The U.S. literature documents increasingly negative evaluations of opposing partisans and treats partisan social identity as an important source of the phenomenon. The major review by Iyengar, Lelkes, Levendusky, Malhotra, and Westwood synthesizes a literature in which Americans increasingly distrust or dislike the opposing political camp; earlier experimental and survey work by Iyengar, Sood, Lelkes, and Westwood supplies much of the empirical foundation. ^{source ref}

Pew's recent public-opinion findings are consistent with this literature: Republicans and Democrats increasingly describe the other party's adherents as closed-minded, dishonest, immoral, or unintelligent. ^{source ref}

But affective polarization still does not prove strategic constraint. Consider three hypothetical voters:

Voter One: “I really like Party A, and I really dislike Party B.”

This is conventional positive and negative partisanship.

Voter Two: “I am lukewarm about Party A, but Party B is unacceptable, so A is clearly my first choice.”

This is principally negative motivation, but may still be a sincere vote if A genuinely ranks first among the available alternatives.

Voter Three: “I prefer Party C to Party A. But C cannot win, and I regard B winning as disastrous, so I vote A.”

Only the third cleanly instantiates the proposed hostage mechanism.

The distinction is crucial because negative voting concerns the direction and intensity of motivation, whereas strategic voting concerns the relationship between a person's sincere preference and the ballot they actually cast.

Strategic voting supplies the missing link

In the strategic-voting literature, a voter is strategic when information about other candidates' chances changes which candidate the voter supports. Alvarez and Nagler's work is especially useful methodologically: rather than merely asking voters whether they voted strategically, they model the electoral context and the standings of parties. They find evidence of strategic behavior but also show that estimates based on voters' self-reported motivations can exaggerate its prevalence. ^{source ref}

This is the political-science form of “lesser-of-two-evils,” “defensive,” or “spoiler-fear” voting. Those informal labels describe overlapping intuitions rather than neatly separated academic literatures. The common mechanism is:

[ \text{preferred alternative is weak} \quad+\quad \text{opponent is very costly} \quad\Rightarrow\quad \text{support strongest viable blocker}. ]

The strength of the effect depends not merely on whether the voter likes A more than B but on preference intensity. Suppose a voter ranks the choices:

[ C > A \gg B. ]

If C has very little perceived probability of winning and A is the only realistic way to prevent B, the large utility loss attached to a B victory can dominate the smaller benefit of expressing a sincere preference for C. This is the “asymmetric preference intensity” component of the proposed mechanism. Empirical research usually captures this through separate candidate/party affect ratings, negative identities, “for versus against” questions, and perceived viability rather than through a literature bearing that exact label. ^{source ref}

Electoral coordination turns an individual calculation into a party-system equilibrium

Plurality electoral systems generate incentives for voters, candidates, donors, activists, and political elites to coordinate around viable contenders. The familiar logic associated with Duverger's law contains both a mechanical component—votes for dispersed smaller parties translate poorly into seats—and a psychological/strategic component—people anticipating that fact have incentives to desert weak alternatives. Modern strategic-voting research formalizes precisely this kind of viability calculation. ^{source ref}

The hostage-voter hypothesis should therefore be understood as a special case of electoral coordination in which the incentive to coordinate is amplified by negative partisanship.

A compact operational definition would be:

A hostage voter is a voter whose sincere first preference is outside the major-party option they support, who regards that preferred option as electorally nonviable, and whose vote for the major party is substantially caused by an intense desire to prevent the opposing major party from winning.

That definition makes the phenomenon testable rather than metaphorical.

Evidence that the mechanism operates in the United States

“Voting against” is plainly real

One of the clearest pieces of evidence comes from surveys that explicitly ask voters why they support a candidate.

In 2016, Pew found nearly symmetrical negative motivation: 33% of Trump supporters and 32% of Clinton supporters said opposition to the other candidate was a main factor in their likely vote. In 2020 the balance changed: opposition to Trump became the most frequently cited reason for Biden support, whereas Trump supporters more frequently supplied affirmative evaluations of Trump's leadership, policy positions, and performance. ^{source ref}

This demonstrates the requested distinction directly. Voting for a candidate is observably not equivalent to saying, “I strongly support this candidate.” Substantial numbers of respondents voluntarily characterize their own choice in negative terms. ^{source ref}

Research explicitly described as “negative voting” likewise treats voting primarily against an opponent as a distinct behavioral phenomenon. Garzia and Ferreira da Silva subsequently developed a broader theoretical framework relating negativity to political behavior and affective polarization, emphasizing that opposition-oriented voting deserves analysis independently from affirmative candidate support. ^{source ref}

The magnitude varies sharply across elections and candidates, however. That matters. Negative voting is not a stable personality trait distributed identically across parties; candidate-specific dislike, incumbency, issues, and political events can all change it. The sharp decline in the share of voters disliking both presidential choices when the Democratic nominee changed from Biden to Harris in 2024 is one piece of evidence that “double dissatisfaction” is partly candidate-contingent rather than simply an immutable consequence of the two-party structure. ^{source ref}

Perceived stakes are extraordinarily high

For defensive voting to become powerful, voters must see losing as substantially worse than merely failing to elect their favorite.

The 2020 Pew result is unusually revealing: 90% of Biden supporters believed a Trump victory would do lasting harm to the country, and 89% of Trump supporters believed the analogous proposition about Biden. ^{source ref}

That is stronger evidence for the proposed mechanism than ordinary “candidate favorability” measures. It establishes that the overwhelming majority of committed voters on both sides perceived the alternative outcome as more than an ordinary disappointment. Under those beliefs, casting a ballot for a low-probability third choice involves a psychologically and instrumentally larger perceived risk.

This is exactly where negative partisanship can interact with spoiler fears. If the subjective loss from an opposing-party victory rises, the threshold at which a voter is willing to abandon a preferred minor party falls.

Put schematically:

[ \text{higher perceived damage from B} \rightarrow \text{higher cost of “wasting” vote on C} \rightarrow \text{greater willingness to coordinate on A}. ]

This equation is a theoretical inference from strategic-voting theory combined with the observed intensity of anti-opponent attitudes; it is not itself a causal estimate from one U.S. experiment. ^{source ref}

Negative partisanship appears to increase major-party loyalty

Abramowitz and Webster connect the rise of negative partisanship to greater partisan loyalty and straight-ticket voting in U.S. elections. That finding is highly consistent with the idea that hostility toward the opposing coalition can keep voters inside their own electoral coalition even when enthusiasm for their own side is weaker. ^{source ref}

Bankert's work adds evidence that negative partisan identity has consequences independently of positive identity. Her later study of self-identified independents in the 2020 presidential election is especially relevant because “independents” are a population in which affirmative partisan attachment is, by definition, less straightforward. Negative partisanship among those voters still predicted political engagement, emotional reaction, turnout-related behavior, and attitudes toward bipartisanship. ^{source ref}

Thus, one part of the hostage-voter proposition receives strong support:

It is possible for opposition to the other party to sustain reliable major-party electoral behavior without equally strong affirmative identification with the party receiving the vote.

That is not merely semantic. Positive and negative identification can be separated statistically, and both predict behavior. ^{source ref}

Dissatisfaction with both parties coexists with continued two-party dominance

Pew's 2023 data provide unusually direct evidence for the apparent paradox. At that time, 28% of Americans held unfavorable views of both parties, 25% felt neither represented them well, and 63% expressed dissatisfaction with the emerging presidential choices. ^{source ref}

Pew's 2022 study had already found simultaneous growth in partisan hostility and signs of frustration with the two-party system, with frustration particularly visible among partisan “leaners” rather than strong identifiers. ^{source ref}

And in early 2024, about one quarter of voters disliked both Biden and Trump. “Double-negative” voters were especially common among people who leaned toward a major party rather than identifying firmly with it. ^{source ref}

These figures establish that there is a nontrivial population for whom the simple equation “major-party vote = favorable view of that party/candidate” is false.

They do not tell us how many of those voters have a sincerely preferred third party. Some may simply regard one major party as less bad than the other without having any attractive alternative. Others may prefer staying home. Others may have a genuinely preferred minor party but vote strategically. The headline “both parties unpopular” is therefore an upper bound, not an estimate of hostage voters.

How important is strategic third-party desertion?

The strategic-voting literature makes it clear that voters respond to perceived viability, including in plurality contests, and U.S. presidential research has specifically examined whether minor-party supporters defect toward major candidates when electoral stakes and competitiveness make doing so advantageous. ^{source ref}

But estimating the exact number of hostage voters is difficult for three reasons.

First, counterfactual preference is latent. Observing someone vote Democratic does not reveal whether they would have voted Green, Libertarian, independent, or something else under different viability conditions.

Second, strategic motivation is itself hard to measure. Alvarez and Nagler show that self-reported explanations can produce errors and that those errors become worse when respondents are questioned farther from Election Day. ^{source ref}

Third, positive and negative motives coexist. A voter may honestly say:

“I support A's health-care policy, dislike A's candidate, would ideally prefer C, and above all cannot tolerate B.”

There is no natural rule requiring this person to possess only one “true” motivation.

Consequently, the literature supports the existence of the hostage mechanism much more confidently than it supports a single population percentage for “hostage voters.”

Can surveys distinguish “I support A” from “I must stop B”?

Yes—but only imperfectly.

Direct questions already separate the two

The simplest method is to ask voters whether their vote is mainly:

for their chosen candidate,

or

against the opponent.

Pew's 2016 and 2020 results demonstrate that respondents can and do make this distinction. ^{source ref}

Researchers can also ask respondents separately how much they like their own party and dislike the opposing party. Bankert's positive/negative partisan identity scales were designed precisely to avoid treating one as the mirror image of the other. The finding that the scales have different political consequences is evidence that surveys can empirically distinguish affirmative and negative partisan motivation. ^{source ref}

Feeling thermometers and candidate favorability questions add another layer. A voter who gives Party A a 45/100 but Party B a 5/100 looks quite different from someone assigning A 95 and B 5 even though both cast the same A ballot. The difference is exactly what a simple vote-choice variable conceals; affective-polarization research routinely relies on separate in-party and out-party evaluations for this reason. ^{source ref}

But “for versus against” is not enough to identify the hostage mechanism

A stronger survey would need at least four independent measurements:

Sincere first preference. Which candidate or party would the voter select if every option had an equal realistic probability of winning?

Perceived viability. What probability does the voter assign to each candidate winning?

Negative preference intensity. How bad would the voter regard each possible winner?

Actual ballot choice. Did the respondent choose someone other than their sincere first preference?

The decisive counterfactual would be something like:

“Suppose your preferred third-party candidate had the same realistic chance of winning as the Democratic and Republican nominees. Whom would you support?”

followed by:

“Suppose voting for that third-party candidate could not change which of the two major-party candidates was more likely to win. Whom would you support?”

A voter who switches from A to C under those counterfactuals is much closer to the precise hostage-voter construct than somebody who merely says “I'm voting against B.”

The strongest designs would manipulate viability

Observational surveys face a basic identification problem: voters who believe third parties are unviable may differ systematically from voters willing to support them.

A stronger design would randomly vary credible information about candidate viability, polling, or electoral competitiveness and observe whether voters shift from their preferred minor candidate toward the major-party “defender.” Such a study could directly estimate the strategic component while separately measuring negative affect.

Panel surveys are also useful. Researchers could measure first preferences months before Election Day and watch whether supporters desert weaker candidates as viability information becomes clearer. This would be preferable to retrospectively asking after the election why a respondent voted as they did.

The reason for this methodological caution is empirical, not merely theoretical: Alvarez and Nagler found that direct self-reports of strategic motivation can misestimate strategic voting, whereas incorporating objective electoral context produces lower estimates. ^{source ref}

“Negative partisanship” itself must be disaggregated

Lelkes's critique is particularly important here. A respondent can dislike:

1. the opposing party;
2. its politicians;
3. ordinary opposing partisans;
4. particular policies;
5. partisan conflict itself;
6. or the existence of both major parties.

Those attitudes are not interchangeable. Someone who rates both parties poorly may dislike partisan warfare rather than feel deeply attached to an anti-Republican or anti-Democratic identity. ^{source ref}

This means that a serious hostage-voter survey should not infer motivation from a single feeling thermometer.

The proposed phenomenon is therefore measurable

An ideal classification could look like this:

This classification is not a claim that existing surveys routinely publish these five categories. It is the measurement architecture implied by the existing literature on positive/negative identity and strategic voting. ^{source ref}

The empirical challenge, therefore, is not that political science is unable to distinguish “for A” from “against B.” It can. The harder problem is distinguishing negative but sincere major-party support from negative, strategically constrained support that suppresses a preferred alternative.

The feedback loop: can negative partisanship stabilize a dissatisfied two-party electorate?

The proposed feedback loop is theoretically coherent:

[ \text{fear of opposing coalition} \rightarrow \text{strategic major-party voting} \rightarrow \text{third-party weakness} \rightarrow \text{fewer credible alternatives} \rightarrow \text{dependence on a major party} \rightarrow \text{stronger binary conflict}. ]

The evidence for the links is uneven. That distinction is essential.

Fear of the opposing coalition → strategic major-party voting

Evidence: strong for the ingredients; moderate for the causal link.

Extremely negative views of the opposing coalition are well documented. The 89%–90% “lasting harm” responses in 2020 show how severe perceived losses can become. ^{source ref} Negative partisanship is also associated with greater party loyalty, and strategic-voting research demonstrates that perceived chances of victory alter electoral behavior. ^{source ref}

Putting those findings together strongly supports the mechanism. What is less common is a study that experimentally varies negative partisanship and then shows that this specifically causes a minor-party supporter to defect to a major party.

So this arrow is well grounded but not perfectly identified.

Strategic voting → third-party weakness

Evidence: strong at the theoretical level and well established across plurality-election research.

Plurality rules reward electoral coordination. A small party can have genuine support yet become strategically unattractive if voters believe only two candidates can contend for first place. Anticipating that, some voters coordinate on their preferred candidate among the viable pair. ^{source ref}

This is what makes spoiler fear self-reinforcing. Viability is partly a belief about other voters' behavior. If enough people believe C cannot win and therefore abandon C, that belief becomes correct.

This equilibrium does not require explicit collusion by the major parties. It can emerge from decentralized rational choices by voters, candidates, donors, activists, and media organizations responding to perceived viability.

Third-party weakness → fewer viable alternatives

Evidence: strong as an equilibrium implication, but “fewer choices” needs qualification.

The United States can and does have minor parties and independent candidates. The issue is not literal ballot availability; it is credible governing viability. A voter's subjective choice set can be effectively binary even when several names appear on the ballot.

Thus the relevant variable is not:

[ N = \text{number of candidates} ]

but something closer to:

[ N_v = \text{number of candidates perceived capable of winning}. ]

When (N_v=2), a voter who intensely fears one of those two confronts a much more constraining choice than the raw ballot would suggest. Strategic-voting theory is fundamentally about this distinction between formal availability and effective viability. ^{source ref}

Fewer alternatives → dependence on one major party

Evidence: highly plausible but not directly quantified as “dependence.”

Suppose a voter prefers C, moderately prefers A to B, and regards B as catastrophic. Once C ceases to be viable, A becomes the voter's necessary electoral vehicle even if the voter dislikes A.

In repeated elections, that voter can therefore display very stable A voting without stable A enthusiasm.

This is one way to reconcile two otherwise puzzling observations: high levels of dissatisfaction with parties and candidates, alongside highly polarized and loyal electoral behavior. Pew's finding that 28% disliked both parties while negative partisanship remained extensive illustrates the coexistence; Abramowitz and Webster document the party-loyalty side. ^{source ref}

Calling this “dependence” does not mean voters literally cannot defect. They can abstain, protest-vote, participate in primaries, support independents, change parties, or organize a new party. The claim is narrower: when the immediate goal is preventing B from governing, the electoral price of abandoning A may be perceived as prohibitive.

Dependence → stronger binary conflict

Evidence: plausible but the weakest link in the proposed causal loop.

A two-coalition structure can make politics cognitively binary: every gain for “us” is easily interpreted as a loss for “them.” But contemporary affective polarization also has many possible causes beyond electoral rules, including elite polarization, nationalization of political conflict, partisan social identity, demographic and ideological sorting, and information environments. The affective-polarization literature does not reduce the phenomenon to plurality voting. ^{source ref}

There is therefore a substantial endogeneity problem:

[ \text{two-party structure} \leftrightarrow \text{negative partisanship}. ]

Negative partisanship may help maintain two-party voting, while the two-party contest may sharpen negative partisanship; yet both may also be produced by deeper social and policy conflict.

The existence of the loop is consequently more convincing than a claim that researchers have already established its total causal magnitude.

A useful evidence grading

That evidentiary gradient is, in my judgment, the most defensible reading of the research.

Plurality versus proportional and multiparty democracies

Multiparty systems generally expand the set of meaningful first choices

Proportional systems lower the electoral penalty for supporting smaller parties because seat shares can track vote shares more closely than in single-member plurality contests. The effect is especially apparent as district magnitude increases: comparative research finds that electoral institutions influence the success of smaller parties, including parties on the ideological extremes. Golder's 19-country, 165-election study, for example, found that populist extreme-right parties performed better where district magnitude was larger and more seats were distributed in upper electoral tiers. ^{source ref}

For the hostage-voter question, the important consequence is that a voter in a multiparty system is more likely to face several parties capable of obtaining representation even when only a subset can lead the government.

That changes the decision from:

[ A \text{ or } B ]

toward something like:

[ A,B,C,D,E \rightarrow \text{post-election coalition}. ]

In such an environment, a voter can more often cast a ballot for the party closest to their first preference while leaving coalition construction to elected parties.

But “more often” is important. Multiparty voting is not strategy-free.

Strategic voting persists under proportional representation

Duch, May, and Armstrong's American Political Science Review study explicitly analyzes coalition-directed voting in multiparty democracies. Voters can take anticipated coalition arrangements into account when choosing a party. ^{source ref}

A voter might prefer Party C in isolation but choose Party D because D makes a desired governing coalition more likely. Or a voter may avoid a party thought unlikely to cross a legal electoral threshold. In mixed systems, voters can also behave strategically differently across constituency and party-list ballots.

Thus PR changes the location of strategic calculation rather than abolishing it:

Plurality strategic question:

Which candidate can defeat the candidate I oppose?

Multiparty strategic question:

Which party vote best affects seats, thresholds, coalition bargaining, or which coalition governs?

The first more directly pressures voters into one of two camps. The second generally permits more differentiated party expression but introduces uncertainty about how the vote will translate into a governing coalition. ^{source ref}

Multiparty systems can soften the experience of being a permanent loser

Classic comparative evidence from Anderson and Guillory is relevant to legitimacy. They found that citizens whose preferred side lost an election were less satisfied with democracy than winners, but that losers in more consensual political systems reported higher democratic satisfaction than losers in more majoritarian systems. Winners, conversely, tended to be especially satisfied under majoritarian arrangements. ^{source ref}

That finding is compatible with one part of the hostage-voter theory: institutions that disperse governing power may reduce the all-or-nothing quality of electoral defeat. It does not prove that proportional representation reduces negative partisanship, but it demonstrates that institutional design can alter the psychological consequences of winning and losing. ^{source ref}

In a multiparty coalition system, a voter's preferred party can lose the premiership yet influence legislation through coalition bargaining. In a majoritarian system, losing can more readily mean that the opposing coalition controls executive power and legislative agenda-setting. The difference helps explain why perceived electoral stakes may differ institutionally, although cross-national political cultures and constitutional structures complicate any simple comparison. ^{source ref}

The trade-offs of multiparty politics are real

A neutral evaluation has to include the other side.

Fragmentation and coalition bargaining. Greater party-system fragmentation creates more potential coalition combinations. Government formation can therefore require bargaining among parties that campaigned separately, and voters may not know in advance exactly which coalition their ballot will facilitate. Coalition-directed voting itself is evidence that voters recognize and respond to this uncertainty. ^{source ref}

This does not mean proportional representation automatically produces unstable government. Government durability depends heavily on constitutional rules, party-system structure, thresholds, investiture procedures, and political context. The intellectually defensible claim is narrower: multiparty representation introduces a coalition-formation problem that two-party majority government often avoids.

Responsibility and accountability. When several parties jointly govern, assigning responsibility for a policy outcome can become less straightforward. Research on responsibility attribution in institutionally complex European governance finds that citizens' blame assignments depend both on the actual institutional allocation of responsibility and on political/group-serving biases. ^{source ref} The general accountability problem is therefore not imaginary: shared authority can blur who deserves reward or punishment.

Extremist entry. Lower representational barriers can allow parties that would struggle under winner-take-all rules to obtain seats. Golder's comparative analysis found that larger district magnitude and upper-tier seat allocation were associated with greater electoral success for populist extreme-right parties. ^{source ref} Whether one regards a particular smaller party as “extremist” is obviously substantive and context-dependent, but the institutional point is that easier access benefits small parties across the ideological spectrum rather than only moderate ones.

Strategic coalition maneuvering. Voters can vote not simply for their favorite party but to manipulate coalition composition. Thus post-election bargaining does not create a world of purely affirmative voting; it can substitute a more complex form of strategy for the plurality system's lesser-of-two-evils strategy. ^{source ref}

Government choice can be less direct. In a pure two-party contest, a voter frequently knows which executive coalition a ballot will support. In a fragmented parliament, the winning governing arrangement may be determined after the election through party negotiations. That can expand representational choice while reducing the voter's direct control over the exact governing coalition. Coalition-directed voting exists precisely because voters anticipate this problem. ^{source ref}

The comparative trade-off can therefore be summarized without recommending either architecture:

The relevant distinction is therefore not “strategic voting versus sincere voting.” Both systems contain strategic incentives. It is which strategic problems voters face and at what stage coalition formation occurs. ^{source ref}

Stability, dissatisfaction, and democratic legitimacy

Negative partisanship can create electoral stability without enthusiasm

This is perhaps the strongest system-level inference supported by the evidence.

Imagine two large coalitions with voters distributed as follows:

[ \begin{array}{c|c|c} & \text{likes own coalition} & \text{dislikes own coalition}\ \hline \text{strongly fears opponent} & \text{loyal} & \text{still potentially loyal}\ \text{does not fear opponent} & \text{loyal} & \text{high defection risk} \end{array} ]

If fear of the opponent is sufficiently widespread, even the upper-right group—people dissatisfied with their own coalition—can remain electorally reliable.

This produces an important distinction:

[ \boxed{\text{electoral stability} \neq \text{affirmative legitimacy}} ]

A party can have a stable vote share because many voters positively approve of it, because many fear the alternative, or because both are true. Ordinary election returns cannot distinguish those possibilities.

Abramowitz and Webster's findings about negative partisanship and party loyalty, combined with Pew's simultaneous evidence of strong party hostility and widespread dissatisfaction with both parties, demonstrate that this combination is empirically possible in the United States. ^{source ref}

That is probably the most compelling part of the “hostage” analogy.

But participation under threat is not necessarily illegitimate

The word “hostage” can become analytically misleading if it implies that negative voting is inherently irrational or coerced.

Suppose a voter believes:

• Party B would adopt policies the voter regards as gravely harmful;
• Party A would prevent those policies;
• Party C is ideologically ideal but has effectively no chance of governing.

Voting A is then an ordinary consequentialist choice. The voter is constrained by electoral structure, but the motivation may be substantively rational.

The evidence that voters perceive large policy differences supports this counterargument. In 2020, 86% of registered voters said the two presidential candidates differed on important issues. ^{source ref}

Accordingly, negative partisanship cannot simply be interpreted as evidence of institutional coercion. Some negative partisanship may arise because the parties increasingly represent genuinely different ideological, policy, cultural, or governing projects. A voter can rationally care more about stopping an outcome they regard as disastrous than about expressing a perfect first preference.

In this interpretation, the two-party system does not manufacture the underlying conflict; it aggregates existing conflicts into two broad coalitions.

The strongest counterargument is that the causal arrow may run from real polarization to negative partisanship

The United States has undergone party sorting and nationalization: local and cross-cutting political identities have increasingly been organized around the national party conflict. Abramowitz and Webster explicitly connect negative partisanship with nationalized elections. ^{source ref}

If Democrats and Republicans increasingly disagree about genuinely consequential policies and social questions, greater negative partisanship may partly be an accurate psychological reflection of greater substantive distance.

This alternative story is:

[ \text{real coalition differences} \rightarrow \text{high perceived stakes} \rightarrow \text{negative partisanship} \rightarrow \text{stable two-party voting}. ]

The hostage-loop story instead emphasizes:

[ \text{two-party electoral constraints} \rightarrow \text{binary strategic dependence} \rightarrow \text{high perceived stakes} \rightarrow \text{negative partisanship}. ]

The evidence supports elements of both, and present research does not cleanly identify what fraction of contemporary negative partisanship belongs to each path.

Negative partisanship may stabilize voting while weakening satisfaction

The public-opinion evidence makes this combination plausible. Pew's 2023 survey reported only 16% saying they trusted the federal government always or most of the time, alongside 28% unfavorable toward both parties and 25% saying neither party represented them well. ^{source ref}

Those indicators imply dissatisfaction with political institutions and representation, but they should not automatically be interpreted as rejection of democracy itself. “I dislike both parties,” “I distrust the federal government,” “I am dissatisfied with how democracy works,” and “I reject democratic government” are analytically different attitudes.

Thus a system can simultaneously exhibit:

[ \text{high turnout / stable party voting} ]

and

[ \text{low trust / low enthusiasm / low representational satisfaction}. ]

There is no contradiction once defensive voting is recognized.

Whether affective polarization itself erodes democratic legitimacy is contested

Research gives reasons for concern but also reasons not to overclaim.

Graham and Svolik's American Political Science Review study used a nationally representative candidate-choice experiment and a natural experiment to study voters' willingness to trade democratic principles against partisan interests. Their work concludes that polarization can reduce voters' willingness to punish politicians from their own side for undemocratic behavior, highlighting a genuine tension between partisan stakes and democratic accountability. ^{source ref}

A broader review of interventions against partisan animosity notes that rising partisan animosity is associated with weaker support for democracy and greater support for political violence. ^{source ref}

But Broockman, Kalla, and Westwood provide an important causal counterweight. Across five experiments with more than 12,000 participants, they substantially altered affective polarization but found little evidence that those changes cascaded into many of the hypothesized effects on democratic norms or political accountability. Their conclusion is that researchers should be cautious about assuming that simply reducing partisan dislike will automatically strengthen democracy. ^{source ref}

So the evidence does not warrant a simple claim:

[ \text{negative partisanship} \Rightarrow \text{democratic delegitimation}. ]

A safer formulation is:

Negative partisanship can sustain electoral mobilization and party loyalty while coexisting with low trust and dissatisfaction. Strong partisan stakes can sometimes weaken democratic accountability, but the causal contribution of affective dislike itself to broader democratic erosion remains contested. ^{source ref}

That distinction between behavioral stability and normative legitimacy is essential.

Conclusion

The informal “hostage voter” concept captures a real political phenomenon, but only after it is stated precisely.

The literature strongly establishes that negative partisanship exists; that Americans can strongly oppose an out-party without exhibiting equally strong positive identification with their own side; that affective hostility has increased; that many voters explicitly describe their ballots as votes against an opponent; that perceived consequences of the opposing side winning can be extraordinarily severe; and that strategic voters respond to candidate viability. ^{source ref}

It also establishes something that is central to the user's proposed puzzle: substantial dissatisfaction with both major parties can coexist with highly stable two-party electoral behavior. In Pew's 2023 data, 28% disliked both parties and 25% said neither represented them well, even as the American electorate remained organized overwhelmingly around the Democratic-Republican contest. ^{source ref}

Negative partisanship helps explain how that coexistence can be sustained.

For a voter with preferences

[ C>A\gg B, ]

a plurality election in which only A and B are perceived as viable creates a powerful reason to choose A. The more intensely the voter fears B, the stronger the incentive. If many voters reason similarly, C remains weak. C's continuing weakness makes the same strategic calculation rational at the next election. In this sense, the sequence

[ \text{opponent fear} \rightarrow \text{major-party coordination} \rightarrow \text{alternative weakness} \rightarrow \text{continued coordination} ]

is a theoretically robust feedback mechanism grounded in established work on negative partisanship, strategic voting, and electoral coordination. ^{source ref}

What is not established to the same standard is the last and strongest claim:

[ \text{continued coordination} \rightarrow \text{more negative partisanship} \rightarrow \text{still stronger coordination}. ]

That endogenous psychological feedback is plausible, but contemporary affective polarization has multiple competing causes. The literature does not show that the U.S. electoral system is the dominant source of out-party hostility, nor that Americans would become substantially less negatively partisan merely by increasing the number of viable parties. ^{source ref}

Nor is every defensive major-party voter institutionally “trapped.” If the Democratic and Republican coalitions represent genuinely different policies and values, strong opposition to one may be substantive rather than artificial. The fact that overwhelming numbers of voters perceive large policy differences between major-party candidates makes this counterargument impossible to dismiss. ^{source ref}

Comparative evidence nevertheless shows why electoral structure matters. Majoritarian systems concentrate electoral competition into fewer governing alternatives and produce larger satisfaction gaps between winners and losers; more consensual systems tend to leave electoral losers more satisfied. Multiparty systems generally permit voters to support narrower first-preference parties and allow coalition construction to occur after votes are translated into seats. But they do not eliminate strategic voting: voters respond to thresholds and prospective coalitions, and greater representational openness can bring fragmentation, more complex coalition bargaining, blurred responsibility, and easier entry for fringe or extreme parties. ^{source ref}

The most defensible final judgment is therefore:

Yes. Negative partisanship provides a plausible, empirically supported mechanism connecting the U.S. two-party system to the subjective feeling of being politically trapped. A voter can dislike both major parties, prefer an alternative, yet repeatedly vote for one major party because the opposing party is perceived as substantially worse and the preferred alternative is thought unable to win. Plurality electoral coordination can make that calculation self-reinforcing by keeping alternatives weak. ^{source ref}

Its importance is substantial but should not be overstated. The evidence demonstrates negative and defensive motivation among sizable segments of the electorate, but no existing evidence justifies treating most Democratic or Republican voters as merely captive third-party voters. Positive partisan identity, policy agreement, candidate preference, and genuine ideological conflict remain powerful explanations alongside negative partisanship. ^{source ref}

At the system level, negative partisanship is better understood as a stabilizer of two-party electoral behavior than as proof of affirmative political legitimacy. It can make voters remain inside a coalition they regard as imperfect because exit raises the perceived probability of an intolerable opposing outcome. That can generate stable participation and stable party competition at the same time as low trust, dissatisfaction, and a feeling that the available choice set is inadequate. ^{source ref}

The “hostage” metaphor is therefore analytically useful only with one qualification: the restraint is an equilibrium of incentives and beliefs, not literal coercion. The voter remains free to defect, but under plurality competition and intense negative partisanship, the perceived cost of doing so can become high enough that the freedom feels politically ineffective. That gap between formal choice and effective, consequence-sensitive choice is the strongest political-science foundation for the subjective experience of being trapped.