Political trust and the role of uncertainty and vulnerability

Persistent cross-national differences in political trust have traditionally been explained through cultural determinism, rooted in the classic culturalist model (Lerner Reference Lerner1958; Almond and Verba Reference Almond and Verba1963; Easton Reference Easton1965; Uslaner Reference Uslaner2002). The rise of rational-culturalist perspectives has shifted the focus toward the endogeneity and volatility of political attitudes (Citrin Reference Citrin1974; Lane Reference Lane1992; Mishler and Rose Reference Mishler and Rose2001; Valgarðsson, Jennings, Stoker et al. Reference Valgarðsson, Jennings, Stoker, Bunting, Devine, McKay and Klassen2025), emphasizing short-term variations in political trust based on government performance, processes, and probity (Citrin and Stoker Reference Citrin and Stoker2018). This perspective aligns with the evaluative model of political trust, which conceptualizes trust as inherently relational: ‘A trusts B to do X’ (Hardin Reference Hardin2002: 56). Following this theoretical understanding, recent academic debates emphasize performance evaluations when it comes to explaining trends in political trust in contemporary democracies (see, for instance, Seyd Reference Seyd2024; Van der Meer and Van Erkel Reference Van der Meer and Van Erkel2024). However, while these studies explain within-country fluctuations and even some cross-country variations, they fail to account for enduring cross-national disparities (Devine and Valgarðsson Reference Devine and Valgarðsson2024; Devine Reference Devine2025) and, more importantly for our argument, they overlook the persistent differences in the winner-loser gap between new and old democracies already identified in Anderson and colleagues’ (Reference Anderson, Blais, Bowler, Donovan and Listhaug2005) classic work.

These shortcomings stem in part from an overly rationalistic conception of political trust. Political trust is not purely rationally based on how citizens evaluate the object of trust on its own merits (Harteveld, van der Meer, and De Vries Reference Harteveld, van der Meer and De Vries2013). Conceptual discussions on political trust have elaborated on related but distinct forms of trust as rooted in valent expectations of the political system (Hardin Reference Hardin2002, Reference Hardin2004; Bertsou Reference Bertsou2019), emphasizing the role of citizens’ perceptions of competence, fairness, and alignment with shared values in shaping these expectations. However, the literature of political trust has paid comparably little attention to the mechanisms by which assessments and perceptions are processed in generating an overall trust judgement (Seyd Reference Seyd2015: 73; Theiss-Morse, Barton, and Wagner Reference Theiss-Morse, Barton and Wagner2015: 167).

Evaluations depend on expectations (Miller Reference Miller1974: 952; Hibbing and Theiss-Morse Reference Hibbing and Theiss-Morse1995: 55), and the latter are not indifferent to partisan preferences and experiences, constituting an essential part of citizens’ bounded rationality (Jones Reference Jones2001). A less stringent rational approach should focus on ‘the cognitive basis of political trust: trust as an evaluation of politics against – conceptually idiosyncratic – ideals or benchmarks’ (van der Meer and Hakhverdian Reference Van der Meer and Hakhverdian2017: 83). The idiosyncratic benchmark of political trust involves vulnerability and uncertainty about what to expect from the political system. Vulnerability arises because trust exists in a vertical relationship between citizens and political institutions, where different agents wield power through access and influence over public decision-making (Hardin Reference Hardin2002). Uncertainty stems from the expectation that institutions should reduce unpredictability and regulate the actions of those in power (Six and Verhoest Reference Six, Verhoest, Six and Verhoest2017; Verhoest, Maggetti, Guaschino et al. Reference Verhoest, Maggetti, Guaschino and Wynen2025). After all, political trust is based on the translation of trustworthiness beliefs about the competence, benevolence, and integrity of political institutions and actors into citizens’ confidence in their future actions (Mayer, Davis, and Schoorman Reference Mayer, Davis and Schoorman1995; Rousseau, Sitkin, Burt et al. Reference Rousseau, Sitkin, Burt and Camerer1998; Six and Latusek Reference Six and Latusek2023). This way, when citizens exhibit high levels of political trust, they expect decision-makers to deliver outcomes that advance their individual or collective interests. However, these expectations could vary, shaped by past performance (Hetherington and Huser Reference Hetherington and Husser2012; Norris Reference Norris2022) and attitudes toward political actors (Mondak, Hayes, and Canache Reference Mondak, Hayes, Canache, Zmerli and van der Meer2017). Notably, expecting benefits and not expecting harm may function differently.

Influence of non-democratic experience on winner-loser gaps in trust

As it is well known, winning and losing elections can trigger a gap in political trust between those who obtain the electoral victory and are able to form the government and those who lose the elections and become part of the political opposition. This pattern seems to be consistent for different democratic settings and different democratic attitudes (Anderson, Blais, Bowler Reference Anderson, Blais, Bowler, Donovan and Listhaug2005: 42–43; see also Anderson and LoTempio Reference Anderson and LoTempio2002; Craig, Martinez, and Kane Reference Craig, Martinez and Kane2006; Moehler Reference Moehler2009; Singer Reference Singer2023; Singh, Lago, and Blais Reference Singh, Lago and Blais2011; van der Meer and Steenvoorden Reference Van der Meer and Steenvoorden2018; Martini and Quaranta Reference Martini and Quaranta2019). The presence of this winner-loser democratic attitudinal gap is often attributed to the direct impact of electoral outcomes on individuals’ perceptions of institutional fairness, responsiveness, and their own political efficacy, where victory tends to affirm these aspects and defeat tends to undermine them. Following this logic, and according to the classic work by Anderson, Blais, Bowler et al. (Reference Anderson, Blais, Bowler, Donovan and Listhaug2005: 102–109), losers in new democracies tend to exhibit significantly lower levels of political trust in their parliament than losers in older and more established democracies. This difference is often attributed to the limited electoral experience of the citizens in new democracies.

This explanation presupposes that experiences of winning and losing commence within the democratic framework. We contend this assertion for one key reason: any non-democratic regime generates winners (those controlling the institutions and influencing policymaking) and losers (those who do not have any possibility of influencing public policy decisions). Moreover, non-democratic regimes are increasingly characterized by significant institutionalization, allowing for restricted competition among parties, candidates, or factions (Gandhi and Przeworski Reference Gandhi and Przeworski2007; Levitsky and Way Reference Levitsky and Way2010). This implies that in non-democracies, the status of being a winner or a loser often has more profound consequences than in democracies. In non-democracies, losers may be excluded or repressed and face severe political, economic, and personal consequences. At the same time, benefits and rents of winners in non-democracies tend to exceed those enjoyed by government supporters, leaders, or policymakers in democracies (Brum Reference Brum2018). In democratic regimes, by contrast, governments’ actions are constrained by a set of checks and balances (the constitution and law, separation of powers, etc.) and monitored by watchdogs and the free media, which together control the benefits available to electoral winners. At the same time, losers in democratic settings are free from government coercion for having diverging preferences, ideology, or economic and social interests, as is the case in many non-democracies.

How, then, do prior non-democratic experiences shape individuals’ responses to winning or losing in democratic elections? We argue that such experiences condition the relationship between electoral status and political trust by amplifying the perceived vulnerability and uncertainty that electoral outcomes can generate. This heightened sensitivity stems from the legacy of non-democratic rule, which amplifies both the perceived stakes of electoral competition and the psychological impact of its results.

After prolonged non-democratic experiences, both winners and losers become more attuned to the broader implications of electoral outcomes in terms of security, status, and policy implementation. Former winners who prevail in the new democratic setting may interpret victory as confirmation that their previously privileged position will persist. As access to influence, resources, and status appears safeguarded under the new rules, their attachment to governing institutions is reinforced, and their trust in democratic institutions may strengthen, as they perceive the system as fair and responsive, especially when these institutions serve their policy preferences. Likewise, former losers in a non-democratic system who become electoral winners post-transition to a democracy may develop high trust in the institutions that facilitated their success.

The inverse may hold true for losers in a democratic setting who have significant non-democratic experience. Those who benefited from the prior regime may interpret their electoral defeat in a democratic system not merely as a temporal setback to their policy goals but as a threat of diminished rights, reduced access to benefits, and a generalized decline of their status. This group may hence experience a backlash, increasingly questioning and distrusting the democratic institutions of representation (the parliament). Conversely, individuals who were historically marginalized may view continued electoral failure as a perpetuation of exclusion, reinforcing skepticism toward the political process.

Building on this discussion, we formulate the first hypothesis positing that strong reactions to electoral victories and defeats are learned under non-democratic regimes and continue to affect people’s political trust even after transitions to democracy:

The preceding hypothesis tests whether autocratic experience at the country level relates to the magnitude of the winner-loser gap across different national contexts. However, this aggregate-level pattern requires further examination of how the relationship operates at the individual level, given our theoretical emphasis on individuals as carriers of experiences with winning and losing under non-democratic rule. We therefore propose an additional test of the ‘non-democratic legacy’ hypothesis: we expect voters’ trust reactions to depend on their personal exposure to non-democratic rule. In other words, the longer the time an individual spent in a non-democratic country, the stronger the imprint on their perceptions of what consequences electoral outcomes may have for their lives. In exploiting within-country variation in non-democratic exposure, we build on research that connects individual democratic experience to democratic support (Brum Reference Brum2018; Besley and Persson Reference Besley and Persson2019; Acemoglu, Ajzenman, Aksoy et al. Reference Acemoglu, Ajzenman, Aksoy, Fiszbein and Molina2024). This research has found, among others, that individual democratic capital has a positive effect on democratic support even after the country’s democratic experience is accounted for (Fuchs-Schündeln and Schündeln Reference Fuchs-Schündeln and Schündeln2015). Our next hypothesis thus reads as follows:

Effects of change in democratic quality on winner-loser gaps in trust

But are these effects of autocratic experience on voters’ reactions expected to persist indefinitely, or can they be attenuated? As previously noted, the notable disparity between winners and losers in emerging democracies was initially attributed to their inexperience with democratic processes, particularly electoral victory and defeat (Anderson, Blais, Bowler et al. Reference Anderson, Blais, Bowler, Donovan and Listhaug2005: 108–109). Even the simple experience of winning after losing might increase support for democracy (Chang, Yun-Han, and Wu-Han Reference Chang, Yun-Han, Wu and Thomassen2014). Following this reasoning, some scholars have suggested that the winner-loser gap naturally diminishes as citizens gain more exposure to electoral cycles within a democratic framework. This hypothesis aligns with the broader argument that prolonged democratic experience fosters greater support for democracy and democratic values (Fuchs-Schündeln and Schündeln Reference Fuchs-Schündeln and Schündeln2015; Besley and Persson Reference Besley and Persson2019; Acemoglu, Ajzenman, Aksoy et al. Reference Acemoglu, Ajzenman, Aksoy, Fiszbein and Molina2024; Neundorf, Dahlum, Frederiksen et al. Reference Neundorf, Dahlum, Frederiksen and Ozturk2024).

However, as Nadeau, Daoust, and Dassonneville (2021) have demonstrated, the key determinant is not simply the passage of time but rather the quality of electoral democracy; according to them, the extent of the winner-loser gap in satisfaction with democracy is shaped by the information environment, which in turn depends on the democratic quality of the electoral process. In political systems where democratic norms have eroded, electoral losers will be more likely to be exposed to information that highlights electoral irregularities, reinforcing concerns about their future electoral prospects and diminishing their trust in the system (Levitsky and Ziblatt Reference Levitsky and Ziblatt2018). Conversely, in flawed democracies, winners are more inclined to perceive their victory as confirmation of electoral fairness, engaging in motivated reasoning that leads them to dismiss unfavorable news and reinforce their trust in institutions (Nadeau, Daoust, and Dassonneville 2021). By the same token, when elections are reasonably well executed, losers build system support rather than undermine it (Esaiasson Reference Esaiasson2011). Thus, divergent levels of uncertainty and vulnerability due to processes of democratic backsliding may contribute to an increased polarization of political trust between winners and losers. While the analysis by these preceding studies focuses on the variation in the quality of democracy regardless of its within- or between-country character, we formulate a distinctly longitudinal argument: in addition to the effects of between-country differences in democracy on the winner-loser gap, we propose that the size of the winner-loser gap is shaped by within-country changes in democracy. Specifically, as democracy improves, the winner-loser gap narrows, whereas democratic backsliding increases it. Thus, our third hypothesis is:

Data

We use survey data as well as macro-level indicators describing the political and economic context. The survey data come from two cross-national projects. The first one is the European Social Survey (ESS 2020), a biennial survey of European countries conducted since 2002. The second one is the Americas Barometer, also carried out roughly every two years in the Americas by the LAPOP at Vanderbilt University (LAPOP 2019). Both surveys are academic endeavors striving to maintain the highest quality standards and constitute reputable data sources for comparative research.

We restrict the survey data to countries that are located in Europe or Latin America, have at least five surveys, and include all the variables necessary for our analysis. Applying both criteria leaves the following 40 countries from the three regions:

• • Western Europe: Austria, Belgium, Denmark, Finland, France, Germany, Ireland, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and United Kingdom. All these, except Portugal and Spain, are countries without non-democratic experience after WWII.

• • Eastern Europe: Bulgaria, Czechia, Estonia, Hungary, Lithuania, Poland, Russia, Slovenia, Slovakia, and Ukraine;

• • Latin America: Argentina, Bolivia, Brazil, Chile, Colombia, Dominican Republic, Ecuador, El Salvador, Honduras, Jamaica, Mexico, Nicaragua, Panama, Peru, Paraguay, and Uruguay.

Our dependent variable is trust in the parliament. Political trust is ‘object-dependent’ (Hardin Reference Hardin2002; Harteveld, van der Meer, and De Vries Reference Harteveld, van der Meer and De Vries2013), meaning that overall levels vary depending on the specific institution. Cross-national research reveals a ranking: trust in political parties typically resides at the bottom, while trust in ‘state’ institutions like the police or army tends to be at the top (Denters, Gabriel, and Torcal Reference Denters, Gabriel, Torcal, Van Deth, Montero and Westholm2007). Parliaments occupy a middle ground, reflecting public evaluation of their performance and responsiveness (Linde and Ekman Reference Linde and Ekman2003; Dalton Reference Dalton2004; Torcal Reference Torcal2014).

We focus on trust in the parliament rather than trust in the government for theoretical reasons. Parliament is the central institution through which popular sovereignty is expressed and where the plurality of political preferences – including those of electoral losers – finds an institutional voice. The composition of the parliament directly reflects the distribution of winners and losers; if the experience of winning or losing shapes political trust, this effect should be most visible in attitudes toward the parliament. Trust in government, by contrast, tends to conflate evaluations of democratic representation with support for the incumbent (Miller and Listhaug Reference Miller and Listhaug1990; Rudolph and Evans Reference Rudolph and Evans2005) – winners may express high trust simply because ‘their’ people are in charge, and losers low trust for the same reason.

In LAPOP, the trust in Congress item is part of a block of questions about attitudes toward institutions and is measured on a scale from 1 (not at all) to 7 (a lot). In the ESS, trust in the parliament is the first item in a battery on trust in institutions and is measured on a scale from 0 (no trust at all) to 10 (complete trust). Due to differences in the item design, the two trust indicators are not directly comparable. We address this by specifying an ordinal response model that accounts for the differing number of response categories, as we describe in the ‘Methods’ section.

In each survey, we identify electoral winners and losers, defined as voters who support one of the government parties or leaders or one of the opposition parties or leaders, respectively. In the ESS, winners and losers are identified based on responses to questions about the voting behavior in the most recent general election. ESS party codes are translated into Party Facts IDs (Döring and Regel Reference Döring and Regel2019) with the Political Parties Crosswalk (Kołczyńska and Powałko Reference Kołczyńska and Powałko2022) and merged with party status in or outside of government from the V-Party version 2 dataset (Lindberg et al. Reference Lindberg, Düpont, Higashijima, Kavasoglu, Marquardt, Bernhard, D¨oring, Hicken, Laebens, Medzihorsky, Neundorf, Reuter, Ruth–Lovell, Weghorst, Wiesehomeier, Wright, Alizada, Bederke, Gastaldi, Grahn, Hindle, Ilchenko, von R¨omer, Wilson, Pemstein and Seim2022). As a consequence of these transformations, respondents are grouped into four categories: those who had voted for one of the government parties (winners), those who had voted for one of the opposition parties (losers), those who had not voted at all, and those who had not replied to the relevant survey questions.

In LAPOP, we also used responses to questions about voting in the most recent presidential election, excluding surveys where the question only asked about the first round of presidential elections in situations when a second round was carried out. Using the outcome of presidential elections to define electoral status while relying on trust in the national legislature (trust in the parliament in the ESS and trust in the Congress in LAPOP) as our measure of institutional confidence is theoretically justified for two reasons.

First, and following Anderson and LoTempio (Reference Anderson and LoTempio2002), the definition of who wins and loses an election is much more strongly associated with the presidential than the legislative outcome. We argue that this logic extends to presidential systems and to Latin American countries in particular, because presidential elections produce a clear winner-loser outcome and determine which political actors control the most visible and politically salient office in these regimes. The presidential outcome is also the only measure of electoral status available across the LAPOP surveys, which record the presidential vote choice but not the legislative vote.

Second, defining electoral status by the presidential contest while measuring trust in the national legislature is a deliberate design choice: it is precisely what allows for a clean and conservative test of our argument. Our interest lies in how electoral experiences shape citizens’ broader evaluations of representative institutions rather than their appraisal of the authorities currently in office (Citrin and Stoker Reference Citrin and Stoker2018). The legislature is the core institution of representative democracy and, housing both governing and opposition forces, is less exposed to the incumbent-driven bias that affects trust in the executive (van der Meer Reference Van der Meer2010). Because presidential losers are not necessarily legislative losers and because no electoral winner straightforwardly controls the legislature, especially in Latin America, trust in it is not mechanically tied to winning the presidency. Measuring trust in the executive, by contrast, would embed a near-mechanical association into our design, with presidential winners expressing confidence in the very office they have just won. Any winner-loser gap we observe in legislative trust therefore reflects a genuine spillover of the electoral experience into broader evaluations of representative institutions per se, offering a more demanding test than trust in the executive would provide.

Table A2 in the online Appendix A provides sample means of trust in parliament for electoral winners and losers by region, showing that (1) average trust among winners is higher than among losers in all regions, and (2) the winner-loser gap is greater in Eastern European countries compared to Western European countries. Comparing survey means between ESS and LAPOP is complicated due to the differing response scales in trust questions, which needs to be dealt with via modeling, an issue we address in the next section.

From the survey data, we also use four individual-level items as control variables: age, gender, years of schooling, and left-right self-placement. The latter addresses the possibility of confounding of winner and loser status with ideology. We recode the left-right scales into four categories: left, center, right, and non-response. To make the data from ESS and LAPOP more comparable, we restricted the age range of the respondents to between 18 and 90 years. Years of schooling are measured from 0 to ‘18 or more’ in LAPOP, so we constructed an analogous variable for the ESS. Finally, we use weights (the country weight for LAPOP and combined design and poststratification weight for the ESS) to correct for deviations from sample representativeness.

Measures of democratic experience and performance are derived from indicators from the Varieties of Democracy dataset version 13 (Coppedge et al. Reference Coppedge, Gerring, Knutsen, Lindberg, Teorell, Altman, Bernhard, Cornell, Steven Fish, Gastaldi, Gjerløw, Glynn, God, Grahn, Hicken, Kinzelbach, Krusell, Marquardt, McMann, Mechkova, Medzihorsky, Natsika, Neundorf, Paxton, Pemstein, Pernes, Ryd´en, von R¨omer, Seim, Sigman, Skaaning, Staton, Sund- str¨om, Tzelgov, Wang, Wig, Wilson and Ziblatt2023). We measure authoritarian history as the number of years between 1950 and 2000 in which the given country had the regime classified as an electoral or closed autocracy, based on the variable ‘Regimes of the world’, which is obtained from V-Dem’s Electoral Democracy Index and Liberal Component Index (Coppedge et al. Reference Coppedge, Gerring, Knutsen, Lindberg, Teorell, Altman, Bernhard, Cornell, Steven Fish, Gastaldi, Gjerløw, Glynn, God, Grahn, Hicken, Kinzelbach, Krusell, Marquardt, McMann, Mechkova, Medzihorsky, Natsika, Neundorf, Paxton, Pemstein, Pernes, Ryd´en, von R¨omer, Seim, Sigman, Skaaning, Staton, Sund- str¨om, Tzelgov, Wang, Wig, Wilson and Ziblatt2023: 287). The number of years of autocracy ranges from 0 in most countries of Western Europe to 50 years in Russia (see Figure A1 in the Appendix). In the analyses aimed at probing the role of individual non-democratic exposure, we construct an individual-level measure of the proportion of years under autocracy since the respondent’s age of 15 until the year of the survey. Importantly, this measure captures respondents’ cumulative exposure to authoritarian rule but does not identify whether they occupied positions analogous to electoral winners or losers under the authoritarian regime. In models testing the effect of countries’ over-time changes in democracy, we use a change in the liberal democracy index over the preceding five years (Coppedge et al. Reference Coppedge, Gerring, Knutsen, Lindberg, Teorell, Altman, Bernhard, Cornell, Steven Fish, Gastaldi, Gjerløw, Glynn, God, Grahn, Hicken, Kinzelbach, Krusell, Marquardt, McMann, Mechkova, Medzihorsky, Natsika, Neundorf, Paxton, Pemstein, Pernes, Ryd´en, von R¨omer, Seim, Sigman, Skaaning, Staton, Sund- str¨om, Tzelgov, Wang, Wig, Wilson and Ziblatt2023: 45).

In order to account for economic characteristics, we use Gross Domestic Product (GDP) per capita as provided in the V-Dem dataset (Fariss, Anders, Markowitz et al. Reference Fariss, Anders, Markowitz and Barnum2021; Coppedge et al. Reference Coppedge, Gerring, Knutsen, Lindberg, Teorell, Altman, Bernhard, Cornell, Steven Fish, Gastaldi, Gjerløw, Glynn, God, Grahn, Hicken, Kinzelbach, Krusell, Marquardt, McMann, Mechkova, Medzihorsky, Natsika, Neundorf, Paxton, Pemstein, Pernes, Ryd´en, von R¨omer, Seim, Sigman, Skaaning, Staton, Sund- str¨om, Tzelgov, Wang, Wig, Wilson and Ziblatt2023: 384), as well as the Gini index of disposable income inequality from the Standardized World Income Inequality Database (Solt Reference Solt2022). Basic descriptives of these variables are in Table A1 of the Appendix.