Electoral integrity matters: how electoral process conditions the relationship between political losing and political trust

Electoral integrity is increasingly becoming the topic of both the public debate and political-science research. Typically defined as “agreed upon international conventions and universal standards about elections reflecting global norms applying to all countries worldwide throughout the electoral cycle, including during the pre-electoral period, the campaign, on polling day, and its aftermath,” (Norris 2014, p. 21) electoral integrity lies at the core of democratic procedures. Deficits in electoral integrity can take many forms, such as gerrymandering, media bias, or voter intimidation (cf. Birch 2011; Lehoucq 2003; Norris et al. 2013). While such deficits are ubiquitous in autocracies—where a lack of electoral integrity is a defining criterion—we find them also in democracies. For example, authorities in the US are repeatedly confronted with complaints about voter suppression laws targeting minority voters (Hajnal et al. 2017; Wang 2012), and German Bundestag elections display signs of irregularities in vote counting (Breunig and Goerres 2011). A recent series of books and other contributions has extensively discussed what causes such deficits (Birch 2007; Lehoucq 2003; Norris 2015), how to detect and measure them (Birch 2011; Bishop and Hoeffler 2016; Hyde and Marinov 2012; Kelley and Kolev 2010; Norris et al. 2013, 2014b; van Ham 2014), how to promote electoral integrity (Birch 2011; Kelley 2012; Norris 2017; Norris et al. 2014a), and why electoral integrity matters (Norris 2014). With respect to the latter question, prior research has studied how violations of electoral integrity can deter voter turnout (Bratton 2008; Martínez i Coma and Trinh 2017; Simpser 2012), lead to post-election violence and even civil war (Norris 2014), and damage the legitimacy of political regimes (Fortin-Rittberger et al. 2017; McAllister and White 2014; Norris 2014). This paper adds a new perspective to this literature by asking how electoral integrity affects the way election results translate into citizen attitudes towards the political system.

Traditionally, research connecting election results with citizens’ attitudes focusses on a direct link between voting for the winning or losing camp and citizen support. Most immediately, those who have voted for the parties or candidates that ended up in government (i.e. those on the winning side) should ceteris paribus be more satisfied with this incumbent government than those who have voted for opposition parties or candidates (i.e. those on the losing side) since this government is more likely to implement policies favored by those who voted for it (Anderson and Tverdova 2001; Lambert et al. 1986). In addition to this utilitarian perspective, which aims primarily at the link between election results and incumbent support, Anderson et al. (2005) argue that winning or losing in an election may also affect citizens’ satisfaction with democracy and their trust in the political system. For one, winning or losing may trigger positive or negative emotions, which are then conveyed onto not only the incumbent government but also onto the political system itself. Second, assuming a need for cognitive consistency (cf. Festinger 1957), Anderson et al. (2005) argue that electoral losers may develop more negative attitudes towards the political system to avoid psychological discomfort and cognitive dissonance. Lending support to these arguments, prior research has repeatedly found the status as election loser (winner) to decrease (increase) support for the political regime, especially citizens’ satisfaction with democracy and their trust in various political institutions (Anderson et al. 2005; Anderson and Guillory 1997; Anderson and LoTempio 2002; Banducci and Karp 2003; Bernauer and Vatter 2012; Blais and Gelineau 2007; Craig et al. 2006; Curini et al. 2012; Dahlberg and Linde 2017; Jou 2009; Moehler 2009; Rich 2015; Singh et al. 2011).

This contribution seeks to expand on this research by bringing in the electoral process. It does so by first introducing a causal mechanism that links political losing to political trust via evaluations of the electoral process: citizens who voted for the losing camp are more likely to view the electoral process as unfair than citizens who voted for the winning camp, leading to political distrust. It further suggests that the effects of political losing on political trust are contingent on the actual quality of the electoral process: in conditions where electoral integrity is high and elections are conducted in a truly free and fair manner, even those who voted for the losing camp have little reason to suspect foul play. Political losing should therefore barely affect perceptions of electoral fairness. Yet whenever there are actual deficits in electoral integrity, political losers have much more reason to doubt the integrity of the electoral process than those who are content with the outcome of the election.

Empirically, this contribution relies on a unique database that combines survey data on political trust and perceptions of electoral fairness from three cross-national survey projects (Asian Barometer Survey, European Social Survey, Latinobarómetro) with macro-level data on electoral integrity (Varieties-of-Democracy Project) through ex-post harmonization. Covering 45 democracies across three continents (Europe, Asia, and Latin America), it uses multi-level structural equation modeling to test its conjectures on a comprehensive scale. It finds that political losing indeed decreases political trust and that it does so not directly but indirectly via perceptions of electoral fairness and satisfaction with the incumbent government. Confirming its key proposition, the empirical analysis shows that political losing has a weaker effect on political trust in countries where elections are found to be conducted in a truly free and fair manner.

Political trust, defined as citizens’ confidence that the political system, its institutions, or actors will “do what is right even in the absence of constant scrutiny” (Miller and Listhaug 1990, p. 358), is a core concept in political-culture research and considered as essential for the smooth functioning and stability of democracy. Not only is it often referred to as a key component or even the epitome of regime legitimacy (Hetherington 1998; Hutchison and Johnson 2011; Newton 2009), but it also has important and desirable behavioural consequences. Among others, high political trust is associated with a higher willingness to pay taxes (Letki 2006; Marien and Hooghe 2011; Scholz and Lubell 1998), higher compliance with the law in general (Abdelzadeh et al. 2015; Dalton 2004; Tyler 2011), and fewer demands for institutional reform (Dalton 2004; Dalton et al. 2001). Political trust makes political institutions work effectively and reduces the need for control and supervision. Investigating how losing in an election, electoral integrity, and political trust relate to each other can thus help us assess under which conditions election results have the most profound consequences for the functioning of democracy.

While prior research typically expects losing an election to have a direct effect on political trust (Anderson and LoTempio 2002; Banducci and Karp 2003; Craig et al. 2006; Jou 2009), I propose two causal mechanisms mediating this effect: one via support for the incumbent government and one via perceptions of electoral fairness. Following David Easton’s (1965, 1975) conceptualization of political support, we can distinguish between citizen attitudes towards the political regime and towards the political authorities (as well as towards the political community). First, the most immediate reaction to losing an election should be at the level of support for the incumbent authorities: citizens that did not vote for the incumbent government are unlikely to be happy with this government or satisfied with its performance, both for utilitarian and for psychological reasons (Anderson and Tverdova 2001; Lambert et al. 1986). Those less satisfied with the incumbent government may then generalize these attitudes towards the political regime itself—after all, it is incumbents that shape much of how the political system is perceived (Lambert et al. 1986; Maier 2011)—, resulting in lower levels of political trust.

However, losing may also affect political trust through another venue: by shaping citizens’ perceptions of the fairness of the electoral process. Those who have voted for an unsuccessful party or candidate have more reason—both for psychological and for utilitarian reasons—to suspect foul play and to scrutinize the election result. For election winners, on the other hand, having attained the desired outcome is likely to lead to them paying little attention to allegations of electoral irregularities, irrespective of whether these are truthful or not. Recurring to the negative emotional response to losing, Cantú and García-Ponce (2015) conjecture that election losers develop more critical views about the integrity of the electoral process in order to compensate for the gap between their party or candidate preferences and the actual outcome of the election. Testing their hypothesis on Mexico, they find that election losers indeed perceive the electoral process to have been less fair than those who find themselves on the winning side. Kernell and Mullinix (2019) similarly argue that what they call “partisan motivated reasoning” distorts information processing and interpretation among both winners and losers, resulting in losers viewing the electoral process as having been less fair than winners. Conducting a survey experiment in the US, they confirm their proposition in finding that losers are more likely to believe that votes have been miscounted than winners (as well as nonpartisans). Earlier research on the US (Alvarez et al. 2008; Craig et al. 2006), Africa (Moehler 2009), and Latin America (Maldonado and Seligson 2014a, b) substantiates the general idea of citizens viewing the electoral process more negatively when the election has not brought their desired outcome.

More negative perceptions of the electoral process should then subsequently decrease trust in the political system because (perceived) violations of electoral integrity violate the rules of the democratic game and therefore are bound to disappoint citizens’ normative expectations of the political system (Norris 2014). In contrast, the perception that the electoral process was conducted in a free and fair manner should boost citizen confidence in the political system by increasing its perceived accountability and responsiveness, giving citizens the impression that they can effectively shape policy outcomes (Fortin-Rittberger et al. 2017; Hooghe and Stiers 2016). Prior research confirms this assumption. Alemika (2007) shows that trust in political institutions is higher if citizens perceive elections to be free and fair in Africa; Rose and Mishler (2009) discover similar effects in Russia; and McAllister and White (2014) find perceptions of electoral fairness to increase satisfaction with democracy in Belarus, Russia, and Ukraine. More globally, Norris (2014) demonstrates that citizen trust in elected institutions as well as satisfaction with democracy decrease if citizens perceive electoral malpractice to be more widespread. Overall, then, political losing should indirectly affect political trust.¹

However, this negative effect of political losing may itself be contingent on the nature of the election. In contexts where elections are generally conducted in a free and fair manner (i.e. in high-quality liberal democracies), even those who voted for the losing camp should still evaluate the electoral process positively, weakening the linkage between political losing and political trust. In contrast, in political regimes where the electoral process is flawed to begin with, being on the side of the electoral losers should further skepticism towards the electoral process and, in conjunction, greatly reduce political trust. Esaiasson (2011, p. 103) argues that losers will “react well” to their defeat under four conditions: first, they must perceive to gain utility from a peaceful solution of conflict through electoral processes; second, they must feel like being involved in the decision-making process; third, they must consider the system of government itself as legitimate; and fourth, losers must have the impression that their own preferred party or candidate “has a fair chance to win the next time around.”

At least the latter three conditions are all dependent on the integrity of the electoral process, suggesting a moderating effect of electoral integrity on the link between political losing and political trust. This idea is supported by Maldonado and Seligson (2014a, b), who, analyzing up to 23 democracies in Latin America, observe that the gap in political trust between winners and losers is indeed smaller in countries with a higher level of electoral integrity. Contrastingly, Fortin-Rittberger et al. (2017) find that election winners only express more satisfaction with democracy than election losers when elections were conducted in an impartial way. At the same time, Esaiasson (2011) as well as Anderson et al. (2005) find the winner-loser gap in political support to be smaller in older democracies than in newly established ones. As older democracies ceteris paribus have fewer problems with the electoral process, these findings may also indicate a moderating effect of electoral integrity on the link between political losing and political trust. We can therefore hypothesize that the negative effect political losing has on political trust, in particular the indirect effect that is mediated through perceptions of electoral fairness (H₂), is mitigated by electoral integrity.

In contrast, we would not expect the indirect effect that is mediated through satisfaction with the incumbent government (H₁) to be contingent on electoral integrity. The link between losing and government satisfaction stems primarily from voters’ expectations that a government they did (not) vote for is more (less) likely to implement the policies they prefer and there is little reason to expect electoral integrity to affect this relationship.

To test these conjectures on a broad geographical scale, this study combines micro-level data from three different cross-national survey projects: the Asian Barometer Survey (2010–2012; Asian Barometer 2013), the European Social Survey (2012–2013; European Social Survey 2016), and the Latinobarómetro (2012–2013; Corporacion Latinobarómetro 2014) with macro-level data from the Varieties-of-Democracy Project (v9; Coppedge et al. 2019b). Together, these data cover 45 democracies spread across Europe, Latin America, and Eastern Asia (for a list of countries, see Table 2).²

All survey data underwent a process of ex-post harmonization in order to prepare them for use in the comparative analysis. In general, linear transformation was applied so that the variables ranged from 0 to 1 and so that high values always meant agreement with the respective concept. The formula used for this transformation wasor, in case the direction of the scale had to be reversed,

Table 1 provides an illustration of this linear transformation process. It shows how values on the source variables (in this case, for measuring trust in institutions) translate into values on the target variable.

This approach represents common practice and is considered reasonably valid when question wordings and response categories are similar (Veenhoven 1993). Ex-post harmonization enables us to extend the cross-national coverage of survey data, which is often limited to specific groups of countries (e.g., regionally), and therefore increases variation and facilitates cross-national comparison (Dubrow and Tomescu-Dubrow 2016). Common challenges to ex-post harmonization include the quality of the source data and documentation and the comparability of the data (Granda et al. 2010; Tomescu-Dubrow and Slomczynski 2016). The following paragraphs describe in more detail the measures used and discuss specific challenges that arose during the harmonization process of the data used in this analysis.

The dependent variable political trust is modeled latently as citizens’ confidence in four key regime institutions: parliament, legal system, police, and political parties (for question wordings, see Table A-1). Modeling the dependent variable latently in a structural equation model (instead of, for example, using a summative index or factor scores) allows to account for the different weight of each of the four indicators as well as for measurement error (Meuleman 2019). Institutional confidence is a commonly used measure of political trust in both democratic and autocratic contexts (e.g., Chen 2017; Dalton 2004; Hooghe, Dassonneville, and Marien 2015; Moehler 2009; Pietsch and Clark 2015). By combining confidence in four different institutions, the measure captures institutions of the legislative, executive, and judiciary as well as political and implementing institutions. While question wordings are rather similar across all three surveys (cf. Table A-1), the number of response categories varies considerably: Both the Asian Barometer Survey (ABS) and the Latinobarómetro (LB) use 4 response categories ranging from “a great deal”/“a lot” of trust to “none at all,” whereas the European Social Survey (ESS) offers respondents an 11-point response scale. Yet, the extremes are labeled similarly to the ones in the ABS and LB: “no trust at all” and “complete trust.” It thus seems justified to harmonize them using linear transformation (see Table 1 for an illustration of how the original variables translate into the target variable). Multi-group confirmatory factor analysis showing metric measurement invariance to be present across surveys further validates this practice: it suggests that respondents interpret questions and response scales in similar ways across survey projects (for a more detailed discussion of measurement invariance, see Online Appendix/Table A-3). To facilitate interpretation, political trust is transformed onto a scale from 0 (no trust at all) to 100 (complete trust).

For measuring the status as election loser, the analysis makes use of the retrospective vote choice questions. In accordance with the mainstream of the literature (e.g., Anderson et al. 2005; Blais et al. 2017; Singh 2014), it codes a respondent as an election “winner” whenever they voted for a party that joined the government subsequent to the election in question (regardless of the number of vote shares or cabinet seats won by that party) and as an election “loser” whenever they voted for a party that did not become part of the government.³ The resulting binary variable takes the value “1” for election losers and the value “0” for election winners. All countries were checked with regard to whether a national election (parliamentary or presidential) was held during the respective survey’s fieldwork period. This was the case in the Netherlands, which held national parliamentary elections on September 12, 2012, which coincided with the ESS fieldwork period starting in August 2012. As government was not formed until November 5, winners and losers are almost impossible to identify for this period, and the analysis thus excludes the Netherlands. The same is true for Israel, where the ESS was fielded between September 2012 and February 2013 and which held national parliamentary elections on January 22, 2013, taking until March 14, 2013, to form a government. No national elections were held in any of the other European nor the Asian or Latin American countries during survey fieldwork periods.

To gauge the mediating variable perceptions of electoral fairness, I employ an item asking respondents how free and fair they thought the last national election to be. This item was not worded identically across the three surveys (cf. Table A-1). Whereas the ABS and ESS both ask rather generally about how free and fair respondents think the last election was, the LB queries more specifically about the fairness of the election “regarding the opportunities of the candidates and parties to campaign.” While the latter refers to only a subcomponent of electoral integrity, we may still consider respondents’ replies to this question to reflect their overall assessment of the freedom and fairness of the electoral process. Prior research has shown that survey respondents often use rather general assessments as heuristics when being confronted with survey questions on specialized topics (Tourangeau et al. 2000). This seems especially likely in our case since the LB contains no other questions on different aspects of electoral integrity, so respondents have little incentive to disentangle their perceptions of different aspects of the electoral process. The second mediator, satisfaction with the incumbent government, is measured by items asking how satisfied respondents are with the current government or its performance. Again, the ABS and ESS ask very similar questions, querying how satisfied respondents are with their country’s current government, whereas the LB asks whether respondents approved or disapproved of the performance of the current government. Given that government performance is likely the main evaluation criterion for satisfaction with government, we can assume that either question taps into the same latent construct.

For measuring the moderating variable electoral integrity, the analysis relies on macro-level data from the Varieties-of-Democracy (V-Dem) Project (version 9; Coppedge et al. 2019b). As a primary and summative measure, it uses V-Dem’s Clean Elections Index. The Clean Elections Index gauges the extent to which elections are free from “registration fraud, systematic irregularities, government intimidation of the opposition, vote buying, and election violence” (Coppedge et al. 2019a, p. 44) and thereby captures the most central components of electoral integrity. It is composed of indicators for electoral management body (EMB) autonomy, EMB capacity, election voter registry, election vote buying, other voting irregularities, election government intimidation, other electoral violence, and election freedom and fairness.

Based on this measure, we can identify countries and regions with deficits in electoral integrity. As Fig. 1 illustrates, V-Dem rates levels of electoral integrity to be universally high across Western Europe. However, deficits in electoral integrity become apparent already for some of the Eastern European democracies (for example, minor deficits in Lithuania and Hungary, and rather major deficits in Albania and Ukraine). The picture looks much bleaker for Latin America, where only Brazil, Chile, Costa Rica, and Uruguay reach top marks in electoral integrity, while countries like Guatemala, El Salvador, or Colombia exhibit what can be considered substantial deficits. In East Asia, already the region with the smallest share of democracies, only a single country—Japan—registers virtually no deficits. The magnitude of deficits varies across the remaining East Asian democracies, with the Philippines showing particularly low levels of electoral integrity.

Combining the system-level V-Dem data with the individual-level survey data requires matching respondents (in the survey data) to country-years (in the V-Dem data). As none of the surveys used for this analysis is an election study, fieldwork periods are not aligned with election cycles. Yet, as levels of electoral integrity can only be measured in election years, for the Clean Elections Index, V-Dem repeats the value for the most recent election for all years up until the year in which the next election takes place. This means that we can mostly use the year in which the respective survey was fielded to match these system-level data to the individual-level survey data. Exceptions had to be made for those cases in which elections⁴ took place in the same year in which the survey was fielded. As outlined above, whenever elections were held during survey fieldwork periods, these had to be excluded from the analysis due to difficulties in identifying election winners and losers. Whenever these elections were held after the survey fieldwork period, the matching year had to be adjusted: Using the survey year would have implied matching electoral-integrity data for the upcoming election, so instead the year preceding the survey fieldwork period was used for matching the macro data. Table 2 gives an overview of survey fieldwork periods, election years/dates, and matching years used.

The models include several control variables. On the individual level, they control for other common determinants of political trust: economic performance evaluations (e.g., Mishler and Rose 2001), political interest (e.g., Lü 2014), and social trust (e.g., Zmerli and Newton 2008). Economic performance evaluations are measured in terms of how respondents evaluate their country’s present economic situation; a question asking respondents how interested they are in politics gauges political interest; and respondents’ assessment of whether most people can be trusted captures social trust. The analyses also include a number of sociodemographic variables: subjective social status, level of education (recoded into none, primary, secondary, tertiary), gender, and age. Table A-1 documents question wordings for all variables and Table A-2 details the descriptive statistics. All analyses are conducted as multi-level structural equation models. As this paper is interested only in individual- and cross-level effects, the empirical models include no system-level control variables other than the moderating variable electoral integrity: The multi-level nature of the models already separates the individual- from the system-level variance, thereby rendering any system-level control variables unnecessary for analyses of individual- and cross-level effects (Leyland and Groenewegen 2020, pp. 116–117).

The analysis proceeds in three steps. In a first step, multi-level structural equation models (cf. Meuleman 2019) analyze the overall effect of voting for the losing party or candidate, controlling for alternative explanations (see above; Model 1) and, additionally, for perceptions of electoral fairness and satisfaction with the incumbent government (Model 2). A second step includes the mediations of the loser effect via perceptions of electoral fairness and government satisfaction (Model 3). A third and final step tests for the conditionality of the mediation effect by incorporating the cross-level interaction with electoral integrity (Model 4). For each of these multi-level structural equation models, the measurement model of political trust is estimated dynamically. Factor loadings are always fixed to be equal across countries. The baseline measurement model, i.e. the unconstrained multi-level CFA model (Model 0a), is presented in Table A-4 in the Online Appendix. Constraining the factor loadings to be equal across levels (metric isomorphism; Model 0b in Table A-4) significantly reduces model fit and inflates the SRMR (between) beyond acceptable values. The models used in the multi-level SEM analyses therefore do not assume metric isomorphism. Table A-5 in the Online Appendix presents the resulting measurement models for each of these models 1 to 4.⁵

Turning to the substantive analysis of how political losing affects political trust, Table 3 presents the results of the analysis.⁶ It shows that, in the basic model, political losing decreases political trust (Model 1). All control variables affect political trust as we would have expected from the literature: economic performance evaluations, political interest, social trust, and socioeconomic status all increase political trust, while education tends to decrease it. Women are slightly more trusting than men. The effects of all control variables remain stable across model specifications. In contrast, the effect of political losing on political trust disappears as soon as controlling for perceptions of electoral fairness and satisfaction with the incumbent government (Model 2). The latter finding substantiates the idea that, contrary to much of the previous literature, political losing does not actually have a direct effect on political trust.

Instead, political losing appears to affect political trust indirectly. Model 3 in Table 3 includes both the mediated effect via perceptions of electoral fairness and the mediated effect via satisfaction with the incumbent government. Both of these indirect effects are negative and statistically significant. Those who have voted for a candidate or party that did not become part of the government (coalition) tend to perceive the elections to have been less free and fair than those who voted for the winning camp and, subsequently, express less trust in the political system’s institutions (Fig. 2, upper path). In addition, those who voted for the losing camp are also less satisfied with the incumbent government and this in turn, again, results in lower levels of institutional trust (Fig. 2, lower path). Summing up those two indirect effects (− 0.87 through perceptions of electoral fairness and − 4.47 through satisfaction with the incumbent government) gives us a total indirect effect of − 5.34. This means that being an election loser reduces political trust by about 5 points on a 100-point scale, or about 5 percent.⁷ The majority of this effect (around 80%) is mediated by government satisfaction, while the path via perceptions of electoral fairness accounts for around 20 percent of the total indirect effect. As both Model 3 in Table 3 and Fig. 2 demonstrate, the indirect effects completely absorb the effect of losing: there is no direct effect remaining.

Model 4 in Table 3 finally includes the interaction term between losing the election and electoral integrity. This random-slope model allows the effect of political losing to vary between countries.⁸ As the theoretical argument suggested the level of electoral integrity to affect primarily the first part of the mediated effect, i.e. the effect political losing has on perceptions of electoral fairness, this is the interaction effect reported in Model 4 (i₁). The results lend support to hypothesis 3: the effect of losing becomes smaller with increasing levels of electoral integrity. For ease of interpretation, Fig. 3 plots the conditional effect of voting for the losing party or candidate on perceptions of electoral fairness. Like Model 4 in Table 3 it clearly shows the decrease in effect strength for higher levels of electoral integrity: the effect is about three times as large in countries where electoral integrity is lacking than in those where electoral integrity is high (as measured by V-Dem).⁹ Consequently, and as the full moderated mediation effect i₁*b₁ in Model 4 (Table 3) evidences, the effect of political losing on political trust is also contingent on the level of electoral integrity. Apparently, losing the election becomes less important for how citizens view the quality of the electoral process and, ultimately, how much they trust their democratic institutions, in conditions where this electoral process is generally free and fair. In contrast, losing the election makes people considerably more distrustful in conditions where the electoral process does actually exhibit at least some flaws.

As expected, this is not the case when looking at the effect political losing has on satisfaction with the incumbent government: regardless of whether electoral integrity is high or low, having voted for the losing camp always has an equally negative effect on how satisfied citizens are with the incumbent government (cf. Table A-7 and Figure A-1 in the Online Appendix).

This contribution set out to explore how electoral integrity shapes how election results are translated into citizen attitudes towards the political system. Specifically, it asked how the effects of political losing on political trust varied between countries with high levels of electoral integrity and countries with low levels of electoral integrity. Building on the idea that seeing their own preferred party or candidate lose the election may, one, result in lower satisfaction with the incumbent government and, two, make citizens view the electoral process as less free and fair, it suggested that political losing decreases political trust indirectly through government satisfaction and perceptions of electoral fairness. It further argued that citizens are less likely to perceive the electoral process as flawed in conditions of high electoral integrity, i.e. when the electoral process is in fact free and fair. Adding to the literature on political losing and political trust, it posited that this should mitigate the negative effect political losing has on perceptions of electoral fairness and, subsequently, on political trust, but not on satisfaction with the incumbent government.

Based on a dataset of ex-post harmonized survey data from three cross-national survey projects (Asian Barometer Survey, European Social Survey, Latinobarómetro) and macro-level data from the Varieties-of-Democracy Project that covers 45 contemporary democracies in Europe, East Asia, and Latin America, it demonstrated that citizens who voted for the losing party or candidate indeed express less trust in the political system’s institutions. Yet, this link between election results and political trust appears to be entirely mediated through, on the one hand, satisfaction with the incumbent government and, on the other hand, perceptions of electoral fairness. Those who voted for the losing camp are both less satisfied with the incumbent government and view the electoral process as having been less free and fair than those who voted for the winning camp. Reduced satisfaction with the incumbent government and more negative views of the electoral process both in turn result in lower levels of political trust. Political losing thus seems to affect political trust only indirectly. With regard to the question how electoral integrity conditions this relationship between election results and political trust, this contribution’s core finding is that the gap in how losers and winners perceive the electoral process becomes smaller when electoral integrity increases. Apparently, political losing primarily fosters doubts about the fairness of the electoral process in conditions when there is at least some reason to suspect violations of electoral integrity. As long as elections are conducted in a relatively free and fair manner (as assessed by V-Dem), in contrast, whether citizens win or lose in the election is hardly relevant for how they view the fairness of the electoral process.

These findings have important implications for democracies like Albania, Guatemala, or Mongolia, which all exhibit noticeable deficits in electoral integrity. Unlike democracies with high electoral integrity like Chile, Portugal, and Japan, they cannot build on the famous “losers’ consent” (Anderson et al. 2005) with regard to citizens’ view of the electoral process. While losing an election leads to citizens being less satisfied with the incumbent government and, at least to some extent, subsequently the political institutions ruled by this government, in any democracy, it is only the former democracies—those where electoral integrity is challenged—where losing an election additionally reduces institutional trust by severely diminishing confidence in the electoral process itself. Safeguarding elections from manipulation attempts and irregularities can thus help shield democracies from election-related fluctuations in citizen support.