Hostname: page-component-77c89778f8-5wvtr Total loading time: 0 Render date: 2024-07-24T16:44:18.395Z Has data issue: false hasContentIssue false

Election integrity across Europe: who thinks elections are held fairly and why?

Published online by Cambridge University Press:  15 March 2024

Andreas C. Goldberg*
Department of Sociology and Political Science, Norwegian University of Science and Technology (NTNU), Trondheim, Norway
Carolina Plescia
Department of Government, University of Vienna, Vienna, Austria
Corresponding author: Andreas C. Goldberg; Email:
Rights & Permissions [Opens in a new window]


If elections are to perform their legitimizing role, they should not only be objectively free, fair and non-fraudulent, but should also be perceived by the public as such. This paper investigates who perceives elections to be fair and why by contrasting two main logics: one based on the idea that perceptions of election integrity arise from external cues voters get from their environment and a second logic claiming that perceptions are internally created based on attitudes and beliefs. We use original survey data collected in ten countries around the European Elections 2019. We find that perceptions of election fairness are unrelated to country levels of integrity but mainly relate to voters’ status as winners/losers of the elections, attachment to the institutions they elect and populist attitudes. We also find beliefs on fake news influence to weakly mediate the relation between populist attitudes and perceptions of election fairness.

Research Article
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (, which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
© The Author(s), 2024. Published by Cambridge University Press on behalf of European Consortium for Political Research


It is key for representative democracy not only that elections are objectively free, fair and non-fraudulent but also that they are perceived by the public as such. In fact, it is unlikely that the public will consent to an election outcome when this is perceived to be the output of a fraudulent process, which may ultimately lead to violence and instability (e.g., Daxecker, Reference Daxecker2012; Fjelde and Höglund, Reference Fjelde and Höglund2016). The events in the United States (U.S.) Capitol in early 2021 underline how perceptions of electoral (un)fairness present a risk to the peaceful transfer of power.

Questions and problems around election integrity are not simply reserved to consolidating or transitional democracies but do indeed increasingly characterize established democracies as well (Birch, Reference Birch2011; Alvarez et al., Reference Alvarez2012; Norris, Reference Norris2014, Reference Norris2015). For instance, in the 2021 federal and state elections in Berlin, Germany, several problems such as wrong/missing ballot papers or long waiting times occurred on Election Day Footnote 1 leading to a court decision to repeat the election in several hundred constituencies, which eventually happened in February 2023. In the U.S., as well as elsewhere, election officials have started enacting some of the most significant changes to the conduct of elections with the overarching goal of increasing public confidence in the electoral process (e.g., Claassen et al., Reference Claassen2013). In the case of the European Parliament (EP) elections in 2019, European authorities had instituted an electoral package of competent national authorities responsible for monitoring and enforcing rules against fake news and election interference by foreign actors. Footnote 2 This represented an unprecedented cooperation among all European Union (EU) member states at election time and pointed to potential threats that may endanger electoral integrity of the EP elections. Potential integrity problems in EP elections that were discussed around the 2019 elections include administrative problems (e.g., remote voting from abroad, multiple voting by citizens residing in another EU country or delays/complications in national registration procedures), but also more recent problems related to electoral disinformation and fake news in online political content (Garnett and James, Reference Garnett and James2020) as well as interference from state and non-state (foreign) actors (e.g., cyberattacks).

Relying on comparative data, in this paper, we examine what shapes citizens’ general perceptions of election fairness by focusing on whether perceptions of election integrity are related primarily to external cues voters get from their environment, including national-level election integrity (such as those happening in Germany during the 2021 Berlin election), or whether they rather depend on pre-existing attachments to elected institutions and internally created factors such as populist attitudes and perceptions of fake news influence (such as was the case in the U.S. in 2020). To test our arguments, we rely on original survey data collected in the context of the European Parliament elections 2019. This dataset covers ten countries in Europe (Czech Republic, Denmark, France, Germany, Greece, Hungary, Netherlands, Poland, Spain and Sweden). The key variables for our study, measuring perceptions of electoral integrity of the EP elections, were asked a little more than a month after the elections, that is, people had time to learn about potential irregularities in the conduct of the elections.

We find that perceptions of election fairness are unrelated to national levels of integrity but derive first and foremost from voters’ status as winners/losers of the elections, attachment to the institutions they elect via those elections and populist attitudes (both targeting politics and the media). The latter effect of populist attitudes on perceptions of election fairness is mediated by beliefs on fake news influence, albeit the mediated effect is rather weak.

Our study contributes to the existing literature in three ways. First, we shed light on the case of the European elections which so far has not received scholarly attention in the context of integrity perceptions. Focusing on the European elections – the second largest democratic elections in the world – provides us with a unique opportunity to study integrity in a variety of national contexts during the same election, while at the same time examining country-related factors that may influence perceptions of electoral integrity. Second, we examine both external and internal factors in one joint model. While our framework includes common determinants of electoral integrity, we expand it with aspects related to more recent phenomena. Third, and as one such recent aspect, we examine the influence of so-called fake news, which is an ever-increasing phenomenon that threatens democracy (Freedom House, 2019) as it offers one of the easiest tools to manipulate voting during the campaign (e.g., Gunther et al., Reference Gunther2019; Tenove, Reference Tenove2020; Zimmermann and Kohring, Reference Zimmermann and Kohring2020; Chambers, Reference Chambers2021). So far, extant studies focus mainly on the identification of related integrity problems and ways how to combat them (e.g., Garnett and James, Reference Garnett and James2020; Judge and Korhani, Reference Judge and Korhani2020). An exception is the study by Mauk and Grömping (Reference Mauk and Grömping2023), which finds that disinformation campaigns are linked to less accurate and more polarised perceptions of election fairness. However, and complementing these findings, we know little about the extent of the problem of fake news influence on elections as perceived by citizens, even less about the factors underlying these perceptions and neither about its connection to overall election fairness, which is precisely one of this paper’s main contributions.

Theoretical framework

Elections are complex processes, and irregularities can occur at the pre-election stage in terms of legal framework, registration and campaigning, on Election Day in terms of the actual voting procedure, and during the post-election stage when votes are counted. Hence, election integrity can be compromised at basically any of these three stages and can take many forms (Elklit and Reynolds, Reference Elklit and Reynolds2005). Therefore, election integrity is a particularly complex concept to define (see also Martinez i Coma and Van Ham, Reference Martinez i Coma and Van Ham2015). While a single, universal definition of electoral integrity does not exist, it can be defined as ‘any election that is based on the democratic principles of universal suffrage and political equality as reflected in international standards and agreements, and is professional, impartial, and transparent in its preparation and administration throughout the electoral cycle’ (Kofi Annan Foundation, 2012). Country-level election integrity as assessed for example by country experts is possibly not enough to guarantee the public confidence in electoral processes. Put it differently, citizens need to be themselves convinced and perceive that elections have integrity to have confidence in the election process. The crucial question is thus when do people perceive elections to be free and fair?

Two main lines of argumentation are used to explain citizens’ assessments. According to a first line of argumentation, perceptions of election integrity arise from ‘external’ cues voters get from either their environment – either before the election (e.g., during voting registration), on election day (e.g., long queues at the polling station) or after the election (e.g., being on the losing side) (Anderson and Tverdova, Reference Anderson and Tverdova2001; Kerr, Reference Kerr2017) – or from media, parties or interpersonal communications with friends, neighbours, co-workers, and family (Kerr and Lührmann, Reference Kerr and Lührmann2017; Karp et al., Reference Karp2018).

Still, while one would expect that citizens’ perceptions align somewhat with their experiences or cues, in the U.S. Bowler et al. (Reference Bowler2015) show that perceptions bear very little relation to what actually happens during the elections. In fact, a second line of argumentation claims that perceptions are ‘internally’ created, either affective and deeply irrational based on the general propensity to believe in conspiratorial theories (Edelson et al., Reference Edelson2017) or based on populist attitudes, reflecting beliefs that legitimate sovereignty lies with the people rather than the elites (Norris et al., Reference Norris2020).

It might be acknowledged that the distinction between external and internal cues may not always be sharp. For example, parties’ rhetoric as an external cue may activate voters’ internalised attitudes. Still, the distinction between external and internal cues is important because it allows us to better disentangle the sources of why certain people may differ in their integrity perceptions, that is, being more or less susceptible to situational cues and/or because of their internal norms and attitudes.

Building on this existing research, which often focuses on only one or a few aspects that may affect perceptions of election integrity, we contribute to the literature by testing a theoretical framework which integrates the aforementioned theories into a joint model to study the influence of external cues and internally created factors with novel data from the European elections. In addition, to accommodate more recent developments that may be related to electoral integrity, we equally aim to expand the selection of potential determinants. More precisely, among the internally created factors, we additionally consider perceptions of fake news influence as a potential more recent correlate of electoral integrity. We argue that (political) actors’ increasing (mis)use of political messaging via (social) media – and devoid of professional journalistic responsibilities – damages the information environment during elections (Crilley and Gillespie, Reference Crilley and Gillespie2019). In detail and following the definition of ‘fake news’ as a two-dimensional phenomenon by Egelhofer and Lecheler (Reference Egelhofer and Lecheler2019), both deliberately created pseudojournalisic disinformation (fake news genre) and actors who politically instrumentalise the term to delegitimize news media (fake news label) may be responsible for citizens’ perceptions of such a ‘fake news’ influence.

Hypotheses: perceptions of election fairness

Our first set of hypotheses derive from the first line of argumentation arguing that perceptions of election integrity arise from external cues. We start with a key premise derived from cue theory. This theory routinely used to study public opinion towards European integration (e.g., Hooghe and Marks, Reference Hooghe and Marks2005) posits direct attention to the cognitive short-cuts European citizens use to make sense of European politics. The basic argument is that cues from the national context (in addition to those coming directly from the EU level) help frame citizen views on Europe (Franklin et al., Reference Franklin1994; Anderson, Reference Anderson1998; Sánchez-Cuenca, Reference Sánchez-Cuenca2000; Rohrschneider, Reference Rohrschneider2002). Since our concern in this paper is with citizens’ views of EP election fairness, it is natural to begin with the assumption that – in line with cue theory – these views respond, in some way, to the level of fairness of national-level elections. At the contextual level, we hypothesize that in those countries where national elections are normally held freely and fairly such fairness evaluations are used by voters as a relatively simple and costless information cue applied to the EU level – not least because EP elections are, administratively speaking, a collection of parallel organized national elections. Put it differently, we hypothesize a spill-over effect from the national to the European level (Hobolt, Reference Hobolt2012), so that individuals living in countries with higher national election integrity are more likely to perceive a higher election fairness at the European election level (Hypothesis 1).

Another important variable relates to the results of the EP elections. Democratic theory teaches us that elections increase satisfaction with democracy among citizens who vote compared to those who abstain, and especially those who ‘win’ the election (e.g., Anderson and Guillory, Reference Anderson and Guillory1997). This is true also at the EP election level (see Plescia et al., Reference Plescia2021). The ‘winner–loser gap’ literature expects a boost in satisfaction with democracy after the election for at least three reasons: because winners expect to get more utility from the system than losers (for example in terms of implementation of public policies), because winning can generate positive emotions, and third, winners’ drive for cognitive consistency may motivate them to adjust legitimacy beliefs about the system that supports their political views (Anderson et al., Reference Anderson2005). Following Cantú and García-Ponce (Reference Cantú and García-Ponce2015) and Kernell and Mullinix (Reference Kernell and Mullinix2019), we hypothesize a similar effect when it comes to fairness. Voting for a party that wins the EP elections – considering both the national or the European level – can increase perceived fairness for at least two of the just listed reasons, namely because winning generates positive emotions, and because winners adjust legitimacy beliefs about the system that supports their political views (see also Mauk, Reference Mauk2022). Our second hypothesis thus states that citizens who voted for the winners of the EP elections are more likely to perceive a higher election fairness at the European election level (Hypothesis 2).

Moving to the second line of argumentation, namely internally created factors, mass perception of election integrity at the EP elections may be related to long-term determinants of attachment to or evaluations of the EU. There is no agreed way in the existing literature on how to measure attitudes towards the EU, but Boomgaarden et al. (Reference Boomgaarden2011) show that citizens’ attitudes towards the EU are multidimensional and, among others, include two components that may be relevant for integrity perceptions: first, an identity component which represents a symbolic attachment to the history and symbols of the EU, and second, an affective component representing positive or negative feelings towards what the EU is and represents. Lower attachment or negative feelings towards European institutions may lead people to perceive the EP elections as less responsive to their preferences. In other words, broader representational appraisals about the EU reflect the publics’ confidence in the capacity of the EU to properly articulate mass preferences and also conduct elections (see also Rohrschneider, Reference Rohrschneider2002). As such, we hypothesize that attitudes towards the EU correlate with how people perceive fairness at the European election level so that individuals with stronger European attachment and positive feelings towards Europe are more likely to perceive a higher election fairness at the European election level (Hypothesis 3).

In keeping with the line of argumentation suggesting that perceptions of fairness bear very little relation to what actually happens during the elections, we focus on populist attitudes, specifically on populist attitudes linked to two different (democratic) actors – namely the political elite and the news media (Fawzi, Reference Fawzi2019; Fawzi and Krämer, Reference Fawzi and Krämer2021). Populist attitudes towards the elite focus on politicians and their presumably lost touch with ordinary citizens (Norris et al., Reference Norris2020) and we label this dimension ‘political populism’. Populist attitudes towards or rather against the media is defined as ‘a specific form of populism that emphasizes elite hostility toward the media, stemming from the populist ideology’s anti-elitist dimension’ (Macaraig and Hameleers, Reference Macaraig and Hameleers2022: 2082). This ‘anti-media populism’ (Krämer, Reference Krämer2018) is based on the portrayal of the news media as self-serving, corrupt, detached from the people and (mis)using their power to indoctrinate people against their interests (Fawzi and Krämer, Reference Fawzi and Krämer2021). We argue that both these populist attitudes can relate to lower perceived levels of election fairness, and this via two routes: First, directly. Fuelled by populist rhetoric by political elites (e.g., De Vreese, Reference De Vreese2017; Engesser et al., Reference Engesser2017), people predisposed to (political) populism more easily believe respective populist claims resulting in dissatisfaction with democratic performance and mistrust of (electoral) authorities, which ultimately affects how these people judge how well elections work (Norris et al., Reference Norris2020). Relatedly, people with stronger anti-media populist attitudes are more susceptible to alternative information sources, which are said to produce popular political mythologies and are often associated with domestic or foreign actors’ strategies to undermine institutional legitimacy and to destabilize elections (Bennett and Livingston, Reference Lance and Livingston2018). As a result, even in objectively fair elections, the deep-seated distrust towards the established system, including both the political elites and news media, means that citizens with populist attitudes are less likely to believe in the fairness of elections due to electoral fraud or other flaws (Edelson et al., Reference Edelson2017; Partheymueller et al., Reference Partheymueller2022). Hence, we expect that people with stronger populist attitudes are more likely to perceive a lower level of election fairness at the European election level (Hypothesis 4).

Second, indirectly. So-called fake news during the electoral campaign may influence elections by manipulating voters in their decision-making process, e.g., via receiving distorted information (cf. Zimmermann and Kohring, Reference Zimmermann and Kohring2020). We argue that people with higher levels of populist attitudes believe more strongly in such an influence. Importantly, whether there really is much fake news spread during the campaign or whether such fake news indeed influences citizens’ voting is not at the core of our argument, but rather that people have the perception of such an influence on elections. People with stronger populist attitudes may be more likely to perceive this influence for two related reasons.

The first reason relates to the dimension of the fake news label, that is, the political instrumentalization of the term by (political) actors to delegitimize news media (Egelhofer and Lecheler, Reference Egelhofer and Lecheler2019). Chambers (Reference Chambers2021) calls such a strategy ‘fake fake news’, that is, actors attack real facts or fact-based journalism by accusing them as fake news. Donald Trump’s accusations of CNN being ‘fake news’ during his 2020 presidential campaign is a prime example for this strategy (see also Bennett and Livingston, Reference Lance and Livingston2018; Fawzi, Reference Fawzi2019). We argue that citizens with populist attitudes are more likely exposed and/or more receptive to messages that falsely accuse fact-based news as ‘fake news’ following their lower trust in and hostile attitudes towards traditional news media (e.g., Mitchell et al., Reference Mitchell2018; Schulz et al., Reference Schulz2020) and their subsequent higher consumption of non-traditional social media sources (e.g., Newman et al., Reference Newman2019). In this context, non-traditional media, especially in the digital environment, are said to connect ‘communities of belief’, e.g., people with similarly strong populist attitudes, rather than the general public with more diverse opinions (Waisbord, Reference Waisbord2018). Being exposed to such a biased (media) environment may not only confirm populist citizens’ perceptions that the news media are not an objective information source (Fawzi, Reference Fawzi2019), but also already sow the seeds of doubt about the subsequent election outcome and the role of the (‘fake’) media/news in it. The latter would be in line with the more general conclusion regarding journalism by Krämer (Reference Krämer2018: 458) that ‘(f)or anti-media populists, manipulation is clearly at work’.

The second reason works via the fake news genre, that is, the presence of deliberately created disinformation or at least having the (mis)perception of such a presence. Footnote 3 The latter may be the larger problem as the majority of peoples’ news diets consists of fact-based news, at least when looking at the aggregate information ecosystem (Allen et al., Reference Allen2020). As populist supporters are portrayed as misinformed and ‘receptive to these, supposedly inaccurate, portrayals of reality’ (Van Kessel et al., Reference Van Kessel2021: 589), they more easily accept simplified explanations for complex problems without taking the effort to find out the truth about related information (e.g., Aalberg et al., Reference Aalberg2017). In turn, they may simply struggle to distinguish factual information from fake news, which may result in the (erroneous) impression of much fake news present in the media that potentially deceives voters (see also Hameleers et al., Reference Hameleers2022).

To sum up, we argue that populist citizens both being told about the biased or ‘fake news’ media, i.e., a higher exposure to external claims about fake news influence, as well as citizens’ self-perceived higher occurrence of fake news during the electoral process may relate to their perception of fake news influence on voters. This perceived influence of fake news may then negatively influence the generally perceived election fairness (cf. Zimmermann and Kohring, Reference Zimmermann and Kohring2020). As a result, the argumentative chain follows a mediation logic in which people with stronger populist attitudes (IV) have a higher likelihood of perceiving fake news to be an influential factor during the electoral process (MV), which in turn (negatively) prompts the perception of overall election fairness (DV) (Hypothesis 5).

Data and methods

To test our hypotheses, we use originally collected survey data in the context of the European Parliamentary Elections 2019 (Goldberg et al., Reference Goldberg2021). This dataset covers ten countries in Europe (Czech Republic, Denmark, France, Germany, Greece, Hungary, Netherlands, Poland, Spain and Sweden). The surveys were conducted by the company Kantar using Computer Assisted Web Interviewing (CAWI). Light quotas (on age, gender, region and education) were enforced in sampling from these databases to ensure representative samples according to these variables (checked against information from the National Statistics Bureaus or Governmental sources). The data collection followed a panel logic with three to seven waves per country (with the final three waves, including two post-election waves, conducted in parallel across all countries). We do not take advantage of the panel structure other than by retrieving our variables from different times before or after the EP elections took place. The key variables for our study, measuring perceptions of electoral integrity of the EP elections and related fake news influence, were asked in the final (post-election) wave running from July 1–12, 2019. The timely distance of this survey wave of more than a month since the elections took place (May 23–26) has the advantage that potential integrity problems that were discovered in the aftermath of the elections could have been part of citizens’ evaluations as well, instead of mainly relying on information during the campaign. Almost all other explanatory and control variables were asked in previous waves, which helps to reduce issues of reverse causality, that is, that integrity perceptions drive any of the included explanatory variables (see online Appendix B for more details per variable). The exception is one of our populist attitudes (anti-media populism), which was equally measured in the final wave. The numbers of respondents in the final wave per country are: NCZ = 733, NDE = 518, NDK = 563, NES = 557, NFR = 776, NGR = 494, NHU = 588, NNL = 1067, NPL = 857, NSE = 497. Footnote 4


Starting with our dependent variable, it is measured by asking respondents the extent to which they think ‘The European Parliament elections were held in a fair way’ on a scale from 1 (not at all) to 7 (very much). We treat this scale as continuous (recoded into a 0 to 6 scale) to run ordinary least squares models. Footnote 5 While we acknowledge the fact that election integrity is a multidimensional concept, the measure we have available in our survey is meant to capture an overall perception of election fairness and is in line with similar studies on the subject (see Mauk, Reference Mauk2022 for example). Figure 1 displays the distribution of the variable across all ten countries.

Figure 1. Perceptions of election fairness across ten European countries.

To test our first hypothesis, we need a measure of national-level election integrity that is ideally exogenous to voters’ perceptions; to this end, we rely on the data collected by the perceptions of Electoral Integrity Project (PEI) dataset, to be precise, release 7.0 (v2) which covers the period 2012–2018 (Norris and Grömping, Reference Norris and Grömping2019). In particular, we make use of the PEI index designed to provide an overall summary evaluation of expert perceptions that an election meets international standards and global norms (measured on a 0–100 point scale). It is generated at the expert level. Footnote 6 We rely on the country-level values which are based on the average PEI indices per country for all national elections between 2012 and 2018 (usually two elections, with exceptions of one election in Denmark and five elections in the Czech Republic). We consider the PEI index as a relatively long-term indicator of national-level election integrity, without assuming that people know all the details of a single previous national election but only that they have a sense of the general election integrity in their own country. Figure A1 in the online Appendix displays the respective values for our ten countries in comparison to other European states, indicating sufficient variation within our countries for this indicator. Note that a parallel indicator for EP elections does not yet exist. Footnote 7

For our second hypothesis, we need to identify election winners, which is far less straightforward at the EU level compared to national elections. In this regard, we follow the existing literature (Anderson and Guillory, Reference Anderson and Guillory1997) and consider winners as those voting for the largest party, that is, parties having received the largest vote share. However, following specific works on EP elections (Plescia et al., Reference Plescia2021) we consider respondents as winners at either the national or the European level. Footnote 8 The winners at the European level are the voters of the conservative European People’s Party (EPP), namely the party that obtained the highest percentage of the votes at the EU level. To be precise, as voters voted for national parties at the elections and the survey question provided the respective national parties (or party alliances) listed on the national ballot sheets, we recoded the national party choice into belonging to the EPP group. The winners at the national level are those voting for the party that at the national level received the largest vote share: these are the voters of ANO in Czech Republic, of Venstre in Denmark, of Rassemblement National (RN) in France, of CDU/CSU in Germany, of New Democracy (ND) in Greece, of Fidesz-KDNP in Hungary, of the PvdA in the Netherlands, of PiS in Poland, of PSOE in Spain and of Socialdemokraterna (S) in Sweden. In addition to winners (coded as 2), our nominal variable further distinguishes between losers (1), i.e., voters having voted for any other party, and self-reported non-voters (0). Footnote 9

For the third hypothesis, we need a measure of attitudes towards the EU, which we capture by two variables, namely EU identity and negative affect (developed by Boomgaarden et al., Reference Boomgaarden2011 and validated over time and cross-nationally by De Vreese et al., Reference De Vreese2019). EU identity is a mean index of three items: ‘Being a citizen of the EU means a lot to me’, ‘I share a common tradition, culture, and history with other Europeans’ and ‘The European flag means a lot to me’ (Cronbach’s Alpha = 0.82). Negative affect is a mean index of three items: ‘I feel threatened by the European Union’, ‘I am angry about the EU’ and ‘I am disgusted with the European Union’ (Cronbach’s Alpha = 0.87). Agreement with all these items is measured using a scale from 1 ‘Fully disagree’ to 7 ‘Fully agree’ with a labelled middle point 4 ‘Neither agree nor disagree’ (recoded into a − 3 to +3 scale).

To capture peoples’ populist attitudes in Hypothesis 4 (and 5), we use two variables measuring political and anti-media populism, respectively: First, we rely on common items from the literature to measure political populism. In detail, we created an index out of three items ‘The ordinary people instead of politicians should make our most important policy decisions’, ‘Politicians in the government are corrupt’ and ‘Politicians make decisions that harm the interests of the ordinary people’ (Cronbach’s Alpha = 0.81). Second, to measure anti-media populism we rely on the measure of perceived disinformation in the news media by Hameleers et al. (Reference Hameleers2022), which was designed to capture beliefs regarding the honesty of the media and the media representing the ‘enemy of the people’, that is, using the typical populist distinction between the people and the elites (news media in this case). The variable is captured again by creating an index out of three items: ‘The news media are an enemy of the ordinary people’, ‘The news media are deliberately lying to the people’ and ‘The news media only serve their own interest’ (Cronbach’s Alpha = 0.89). Both populist attitudes are measured with agreement scales ranging from 1 ‘Fully disagree’ to 7 ‘Fully agree’ with a labelled middle point 4 ‘Neither agree nor disagree’ (recoded into a − 3 to +3 scale). The two populist attitudes are correlated with r = 0.52, which confirms previous studies that political and anti-media populism are related but should not be treated as a single factor (Fawzi and Krämer, Reference Fawzi and Krämer2021; Hameleers et al., Reference Hameleers2022).

For our fifth and final hypothesis, our mediator is measured asking respondents the extent to which they think ‘Voters were influenced by (so-called) “fake news” in the EP 19 elections’ on a scale from 1 (not at all) to 7 (very much). We treat this scale as continuous (recoded into a 0 to 6 scale). To match the direction of our dependent variable (higher values representing less integrity problems), we reversed the scale of the fake news variable, that is, higher values representing no perceived fake news influence. Figure 2 shows the answer patterns for each of the ten countries (for mean values across countries see Table A1 in the Appendix). The correlation between our mediating and dependent variable in pooled data is r = 0.13, while the respective correlations with the two populist variables out of the mediation model are r = −0.21 (political) and r = −0.31 (anti-media).

Figure 2. Perceptions of no fake news influence across ten European countries.

As our research design is mainly cross-sectional (notwithstanding the use of variables measured at different panel waves), it is particularly important to include a broad array of control variables to account for potential confounders. Several factors have been used in previous research to explain why people perceive certain elections to be fair or not. Starting with socio-demographics, it is important that we control for age, given that younger voters may experience socialization processes that leave them to be less deferential to elections and democratic institutions (Denemark et al., Reference Denemark2012). We also control for sex: although women are not a demographic minority, they do constitute a minority in terms of political representation, which may lead them to view elections as less fair. Moving to non-sociodemographic factors, we control for media exposure (both general across various types of media and specifically for the election campaign as well as interpersonal discussion online and offline) to account, among other things, for the possibility that people who frequently use social media are more likely to be exposed to stories that feature suspicion about electoral malpractice and about political scandals, which may cause them to view elections as unfair. This, and related phenomena, are commonly discussed in terms of a so-called ‘toxic’ media and communication environment (e.g., Crilley and Gillespie, Reference Crilley and Gillespie2019 or Marquart et al., Reference Marquart2020). Further, and especially for our mediation variable of perceived ‘fake news’ influence, it is crucial to control for general media exposure as a potentially more general anti-media sentiment of the respondents, that is, people who distrust mainstream media may avoid them and rather consume social or alternative media (cf. Zimmermann and Kohring, Reference Zimmermann and Kohring2020). Following previous studies on perceived election integrity in the U.S. we control for (objective) political knowledge (sum score of six knowledge questions) and ideological extremism based on the left-right scale (Norris et al., Reference Norris2020). Appendix B provides full information on the measurement of all used variables and Appendix A on the summary statistics and their bivariate correlations.

Statistical modelling

In order to test all direct effects of our first four hypotheses (H1-H4) we run multivariate OLS models with clustered standard errors by countries. We first run a simple setup including only the variables of interest and a second setup adding all mentioned control variables. Due to the nested data logic, that is, respondents are nested in ten countries, an alternative setup is to run a multilevel regression model. Running an empty multilevel model results in an intraclass correlation (ICC) of 0.033, which means that only a bit more than 3% of the variance in respondents’ perceptions of election fairness is explained at the country level. While this low ICC value does not speak for the need to model the data in a multilevel logic, we still run a multilevel model with random intercepts as robustness check. As further robustness checks, we treat our dependent variable as ordinal and run ordinal logistic regressions. To exclude country-specific effects driving our pooled results, we rerun our main model by excluding one country at a time (resulting in ten models including nine countries each). We also run separate models for all ten countries to detect potential country-specific differences in the coefficients.

As robustness checks for specific hypotheses, and starting with H1, we test our main model with five alternative country-level integrity measures. Next to the two sub dimensions of the PEI data focusing on procedural fairness (proceduresi) and fair media coverage (mediai) more specifically, we use three external measures for press freedom (World Press Freedom Index (0–100 scale) from Reporters sans Frontières, 2019), the level of corruption (perceived level of public sector corruption (0–100 scale) according to Transparency International, 2019) and the number of EP elections a country participated in (to represent the country’s experience with and confidence in this type of supranational elections). For H2, and notwithstanding the mentioned difficulties of calculating alternative winner/loser measures for EP elections, we run additional models with the difference in the national vote share between the 2014 and 2019 EP elections, i.e., the gains or losses in vote share, as an alternative winner/loser measure.

To analyse the mediation effect of our fifth hypothesis, we conduct causal mediation analysis using the ‘mediation’ package (v4.5) in R (Tingley et al., Reference Tingley2014). As part of the mediation model, we present both coefficients for the first, intermediate model with fake news influence perceptions as the dependent variable, and for the second model, with perceptions of election fairness as the actual outcome of interest and fake news influence perceptions as an explanatory variable. Both indicators of political and anti-media populism serve as independent variables in the mediation model (plus all other variables from our main model as controls). To be sure about the mediation effect, one must consider the overall indirect effect via the two subsequent paths (populist attitudes → fake news influence perception → perception of election fairness). We calculate the size and statistical significance of these paths for both our populism indicators via bias-corrected bootstrap confidence-intervals (95%; 10,000 simulations). Furthermore, following the recommendation by Imai et al. (Reference Imai2010), we conduct a sensitivity analysis for the mediation effect. Given our nested data structure, we should use the specific multilevel setup for causal mediation analysis provided in the ‘mediation’ package via the ‘lme4’ package. However, at the time of writing, this setup does not allow for the calculation of the bootstrap confidence-intervals, nor for the sensitivity analysis. Hence, we present the simple setup in our main models and provide the multilevel setup as robustness check in the Appendix.

Empirical findings

Starting with a descriptive overview of our dependent variable, Figure 3 shows the distribution of the mean perception of election fairness by country (y-axis), plotted against the country-level integrity (PEI index; x-axis). Focusing on the y-axis, we observe quite some variation between countries: perception of election fairness is higher in the two Scandinavian countries Denmark and Sweden, plus Poland (mean values of around 3.9), but lower in France and the Czech Republic (mean values of around 3.2) (for exact mean values see Table A1 in the Appendix). Comparing the pattern to the country-level integrity on the x-axis shows only a weak positive relationship (r = 0.14, n.s.). Our following regression models provide a clearer answer for this relationship.

Figure 3. Perceptions of election fairness vs. country-level integrity (PEI index) across ten European countries.

Table 1 shows the results of our multivariate models. The first model includes only our main independent variables of interest – namely country level integrity, EP election winners, EU attitudes and populist attitudes. The second model further adds all other control variables (see Table A3 for complete model output). In line with Hypothesis 1, Table 1 shows a positive association between country level integrity level and perceived election integrity, but the coefficient does not reach conventional levels of statistical significance. One should, however, be cautious in interpreting this result as a sign that objective measures are unrelated to perceived election integrity due to the low number of countries in our sample. For instance, the study by Mauk and Grömping (Reference Mauk and Grömping2023) uses a larger number of countries and finds a significant positive association of de facto integrity with perceived electoral fairness. Regarding the external cues variables (Hypothesis 2), we see positive associations between being a winner and perceiving the elections as being more integral. Yet, the coefficient is significant only for being the winner at the national level and not necessarily consistent across countries (see following robustness checks). Moving to the EU attitudes and Hypothesis 3, Table 1 shows powerful associations between both EU identity and negative affect and perception of election integrity – and in the hypothesized directions. Finally, we see that both our measures of populist attitudes display significant negative coefficients in line with Hypothesis 4.Footnote 10 Since both EU and populist attitudes are measured on 7-point scales and have highly similar standard deviations (see Table A2), one can compare the respective coefficients with regards to their substantial relevance. As it becomes clear, all variables play their part for electoral fairness perceptions without one of them standing out as particularly relevant. EU identity followed by anti-media populism are somewhat more relevant, and they are also similar in relevance to the EP winner national dummy coefficient. Given the large number of variables, especially in Model 2, we checked for potential multicollinearity problems by calculating the variance inflation factors (VIF). The respective VIFs reach maximum values of around 2, that is, much smaller than the critical value of 5.

Table 1. Explaining perceptions of election fairness (OLS models)

Note: Clustered standard errors (by countries) in parentheses:*P < 0.05, **P < 0.01, ***P < 0.001.

Our conducted robustness checks confirm our main findings (see Appendix C for all model outputs). Both the alternative multilevel setup (see Table C2) and the ordinal logistic regression (see Table C3) result in highly similar coefficients. None of the five alternative country-level integrity measures of the two subdimensions of the PEI index, press freedom, the level of corruption and the number of EP elections a country participated in displays significant contextual effects, but neither produces changes for the individual-level explanatory factors (see Table C4). The alternative winner/loser measure based on the gains or losses in national vote shares does not reach statistical significance (independent of including or excluding non-voters), nor does it change the coefficients of our categorical measure of EP election winners (see Table C5). Finally, excluding one country at a time does not significantly change any of the coefficients (Figure C1). In line with that, running the full model (M2) separately for each of the ten countries (Figure C2) shows that most effects of our pooled model also show up in the separate countries, albeit not always reaching statistical significance, partly due to the smaller sample sizes.

Turning to our mediation hypothesis (Hypothesis 5), Table 2 displays the underlying regression models. The first model is the intermediate model with fake news influence perceptions as the dependent variable. The two negative coefficients for our two populist attitudes support the first part of the mediation link as people with stronger populist attitudes have a higher likelihood of perceiving fake news to be an influential factor during the electoral process (recall the reverse coding into ‘no’ fake news influence). The second model displays the complete model with fake news perceptions included as an explanatory factor for election fairness. The respective positive coefficient supports the second part of the mediation link, that is, the perception of (no) fake news influence is related to election fairness more generally.

Table 2. Explaining perceptions of election fairness (mediation model)

Note: Standard errors in parentheses; *P < 0.05, **P < 0.01, ***P < 0.001.

Turning to the overall indirect effect of the mediation model, Figures 4 and 5 display the respective results for both populist indicators. Both indirect paths (dashed line) are confirmed as statistically significant via bias-corrected bootstrap confidence-intervals. Yet, we must put the respective indirect effect sizes into perspective, as the direct effects (solid line) on election fairness perceptions are (much) larger. Figure 4 shows that the indirect effect of political populism via fake news influence accounts for roughly 4% of the overall effect. For anti-media populism in Figure 5, the indirect effect accounts for almost 9% of the overall effect. While the results thus provide support for our mediation hypothesis (Hypothesis 5), the substantial relevance is weaker than expected. Further, considering the results of the sensitivity analysis (see Figure C3 in the Appendix), the correlation between the error terms (rho) of the mediation and outcome model at which the mediation effects are 0 amount to 0.06. Notwithstanding the large number of controls included in the model, it may thus be sensitive to unobserved confounders.

Figure 4. Mediation results for political populism.

*P < 0.05, **P < 0.01, ***P < 0.001.

Note: Non-parametric bootstrap for variance estimation; confidence intervals (95%, 10'000 simulations) do not include zero for the indirect path (CI [−0.005, −0.0003]), which is thus statistically significant.

Figure 5. Mediation results for anti-media populism.

*P < 0.05, **P < 0.01, ***P < 0.001.

Note: Non-parametric bootstrap for variance estimation; confidence intervals (95%, 10'000 simulations) do not include zero for the indirect path (CI [−0.023, −0.008]), which is thus statistically significant.

Our robustness check, running the mediation analysis specifically controlling for the multilevel structure of our data, leads to the same overall findings (see Table C6 and Figures C4 and C5 in the Appendix). While not the main focus here, adjusting for the multilevel structure also confirms the non-significant coefficient of the country-level integrity variable as in our previous models, unlike this variable being significant in Table 2. The latter is most likely due to not having controlled for the nested data structure in this simple(r) mediation model setup.


The research question we posited in this paper is: what influences perceptions of election integrity? Our research is motivated by several recent events pointing toward a general decline of perceptions of election fairness alongside an increasingly large threat of fake news during election campaigns, following the emergence of new digital technologies. In particular, in this paper we focus on both internal and external determinants of perceptions of election fairness broadly conceived and the mediating role of perceptions of fake news influence specifically. We focus on the EP elections, not only because it is the second largest democratic exercise in the world but also because, since 2019, EU member states have precisely engaged in coordination to address increasing potential threats to electoral integrity of the EU level.

We find that perceptions of election fairness are unrelated to country-level integrity at the national level (based on expert judgements) but relate first and foremost to voters’ status as winners or losers, attachment to the institutions they elect via those elections and populist attitudes. The latter effect is mediated by perceptions of fake news influence, albeit the size of this mediation effect is smaller than anticipated.

These findings have several important implications. First, related to the topic of election integrity, elections are largely regarded as events to legitimize institutions but what we learn from our study is that their legitimizing property might not work when people distrust those institutions in the first place. Second, and relatedly, our study shows that the perceptions of election integrity are first and foremost explained at the individual level rather than stemming from more general country-level experiences with the administration of elections. For losers of the election or those more alienated towards the EU as well as those more prone to populist thinking, it seems that it matters less what officials might do to actually ensure election integrity. This is an idea that is worth exploring in future research. Another interpretation for the absence of an effect stemming from objective country-level integrity may be the different standards the respective populations might use when judging what a fair election implies, following specific national experiences. Still, descriptive evidence from the ten countries under scrutiny in this paper indicates that the large majority of the population (still) perceives EP elections as broadly free and fair with hence, no immediate danger in sight for events such as those that happened in the U.S. With regard to the U.S. case, it represents a good example of imagined electoral fraud in the absence of factual evidence.

Third, our research design, more precisely the use of variables measured in different waves of a panel survey, allows us to disentangle certain individual-level determinants of post-electoral integrity perceptions. Putting it differently, by studying perceptions about one month after the elections we give respondents the time to become familiar with discussions around potential fraud that may have taken place during both the election campaign, during election day and after the elections. This approach has rarely been used in prior research and, to our knowledge, this is the first systematic study of this sort. Relatedly, and building on our findings of a significant, albeit relatively weak, mediation effect, we invite future research to delve more deeply into what drives election integrity with a specific focus on non-direct effects stemming from mediating or moderation effects (in line with recent studies such as Mauk and Grömping, Reference Mauk and Grömping2023). As EP elections may be considered a least likely case to find strong effects, that is, people care less about EP elections than national elections (‘second-order’ argument), the small effects we found may be (much) stronger in national contexts.

Our study does not come without limitations. First, our mediator, i.e., fake news influence perceptions, was measured after the election in parallel to our dependent variable. While such a post-election measurement allows for the opportunity that the formation of attitudes is based on election reports of fake news some weeks after the election, methodologically the measurement of the mediator should ideally precede that of the dependent variable, this to ensure the correct causal order. Yet, data constraints restrict us to the used model setup, and previous research in the field equally measured the mediating and dependent variable in parallel (e.g., Mauk, Reference Mauk2022). Also, while we focused on populist attitudes as a form of internally created irrational beliefs, the alternative aspect of conspiratorial thinking may result in (even) stronger effects, particularly for the mediation model when examining election-specific conspiracy beliefs. Finally, next to the provided substantial interpretations, the absence of country-level integrity effects may be due to the small number of country cases and related low variance (see for instance the respective significant effects in Mauk and Grömping, Reference Mauk and Grömping2023, which are based on a much larger country sample).

Notwithstanding these limitations, our study provides important new insights into citizen perceptions of integrity for elections held in our new digital era. Yet, as discussed by Judge and Korhani (Reference Judge and Korhani2020), the regulation of fake news or disinformation is a thin line. On the one hand, one cannot restrict the free flow of political discourse too much as it serves as a cornerstone for public confidence in the political process. On the other hand, the impact of fake news should also not gain the upper hand, as this may equally undermine public confidence in electoral integrity.

Supplementary material

To view supplementary material for this article, please visit

Data availability statement

The data used for the article, and a detailed documentation of the data collection process, are freely available at: Goldberg, A.C., van Elsas, E.J., Marquart, F., Brosius, A., de Boer, D.C., & de Vreese, C.H. (2021). Europinions: Public Opinion Survey. GESIS Data Archive, Cologne. ZA5553 Data file Version 1.0.0,


Previous versions of this article were presented at research seminars at the University of Amsterdam (ASCoR) and NTNU in Trondheim (EVPOC group) as well as the 2021 European Political Science Association conference. We thank all participants as well as Diana Burlacu and Alessandro Nai for their helpful comments on earlier drafts of this article. We also thank the three anonymous reviewers and the editor at EPSR for their valuable feedback.

Funding statement

This research is funded by a grant from the European Research Council (ERC), grant number 647316.

Competing interests

The authors have no competing interests to declare that are relevant to the content of this article.


3 As we focus on citizens’ perceptions of fake news influence, that is, people may have different definitions of fake news, we do not apply the strict definition by Egelhofer and Lecheler (Reference Egelhofer and Lecheler2019) with fake news being low in facticity, created with the intention to deceive and presented in a journalistic format.

4 The goal of the final wave was to collect around 500 respondents per country. Once these numbers had been reached across countries (after 12 days), the data collection was simultaneously stopped in all countries. Therefore, we refrain from reporting retention rates from earlier waves, which would be misleading given the forced end of the fieldwork.

5 Please note that using other modelling strategies such as ordered logit leads to similar findings (see online Appendix C).

6 Note that we make use of the PEI index, imputed and thus observable for all experts and states.

7 The qualitative assessments by OSCE/ODIHR of the 2004 and 2009 EP elections ( or Election-Watch.EU ( have emphasized the diversity in how the elections are run across member states and have recommended further harmonization.

8 While we do not possess a measure of acceptance of the election result, it is quite plausible to think that election losers not only perceive election integrity to be lower compared to winners, but they are also less likely to accept the election result itself (see Lago and Martinez i Coma, Reference Lago and Martinez i Coma2017).

9 Recent studies show that voters might feel as election winners under additional circumstances than voting for the largest party (Stiers et al., Reference Stiers2018). For example, Plescia (Reference Plescia2019) shows that doing better than in a previous election or doing better than expected (as predicted by opinion polls) is related to feeling as winners of the elections. Plescia et al. (Reference Plescia2021) show that in the context of EP elections these additional operationalizations do not lead to different results. Yet, considering the previous election to operationalize winners at the EP elections is far more problematic than in contexts of national elections because parties may change group affiliation at the European level, parties may run as part of larger national alliances, and voters tend to have a hard time remembering the previous election results due to the lower saliency of EP elections. Nonetheless we have run additional models to check whether our findings are robust to alternative measures of winning/losing, see following results section.

10 Following the “sore loser” argument, we additionally tested the interaction of populist attitudes with the EP election winner variables. As displayed in Table C1 in the Appendix (models 1 and 2), the negative coefficients for populist attitudes become smaller for EP winners (due to positive interaction terms), albeit not all interaction terms reach statistical significance. In contrast, testing an additional cross-level interaction of populist attitudes with the country-level integrity does not show any significant patterns (same Table model 3).


Aalberg, Toril et al., eds. Populist Political Communication in Europe. New York: Routledge, 2017.Google Scholar
Allen, Jennifer et al.Evaluating the Fake News Problem at the Scale of the Information Ecosystem.” Science Advances 6.14 (2020): eaay3539.CrossRefGoogle ScholarPubMed
Alvarez, R. Michael, et al., eds. Confirming Elections: Creating Confidence and Integrity Through Election Auditing. New York: Palgrave Macmillan, 2012.Google Scholar
Anderson, Christopher J.When in Doubt, Use Proxies: Attitudes Toward Domestic Politics and Support for European Integration.” Comparative Political Studies 31.5 (1998): 569601.CrossRefGoogle Scholar
Anderson, Christopher J., and Guillory, Christine A.. “Political Institutions and Satisfaction with Democracy: A Cross-National Analysis of Consensus and Majoritarian Systems.” American Political Science Review 91.1 (1997): 6681.CrossRefGoogle Scholar
Anderson, Christopher J., and Tverdova, Yuliya V.. “Winners, Losers, and Attitudes About Government in Contemporary Democracies.” International Political Science Review 22.4 (2001): 321338.CrossRefGoogle Scholar
Anderson, Christopher J. et al. Losers’ Consent: Elections and Democratic Legitimacy, New York: Oxford University Press, 2005.CrossRefGoogle Scholar
Lance, Bennett, W., and Livingston, Steven. “The Disinformation Order: Disruptive Communication and the Decline of Democratic Institutions.” European Journal of Communication 33.2 (2018): 122–39.Google Scholar
Birch, Sarah. Electoral Malpractice. Oxford: Oxford University Press, 2011.CrossRefGoogle Scholar
Boomgaarden, Hajo G. et al.Mapping EU Attitudes: Conceptual and Empirical Dimensions of Euroscepticism and EU Support. European Union Politics 12.2 (2011): 241–66.CrossRefGoogle Scholar
Bowler, Shaun et al.Election Administration and Perceptions of Fair Elections.” Electoral Studies 38 (2015): 19.CrossRefGoogle Scholar
Cantú, Francisco, and García-Ponce, Omar. “Partisan Losers’ Effects: Perceptions of Electoral Integrity in Mexico.” Electoral Studies 39 (2015): 114.CrossRefGoogle Scholar
Chambers, Simone. “Truth, Deliberative Democracy, and the Virtues of Accuracy: Is Fake News Destroying the Public Sphere? Political Studies 69.1 (2021): 147–63.CrossRefGoogle Scholar
Claassen, Ryan L. et al.Voter Confidence and the Election-Day Voting Experience.” Political Behavior 35.2 (2013): 215–35.CrossRefGoogle Scholar
Crilley, Rhys, and Gillespie, Marie. “What To Do About Social Media? Politics, Populism and Journalism.” Journalism 20.1 (2019): 173–6.CrossRefGoogle Scholar
Daxecker, Ursula E.The Cost of Exposing Cheating: International Election Monitoring, Fraud, and Post-Election Violence in Africa.” Journal of Peace Research 49.4 (2012): 503–16.CrossRefGoogle Scholar
Denemark, David et al.Gender and Leader Effects in the 2010 Australian Election.” Australian Journal of Political Science 47.4 (2012): 563–78.CrossRefGoogle Scholar
De Vreese, Claes H. Political journalism in a populist age (Policy Paper), 2017. <> (last accessed 2 Feb. 2023).+(last+accessed+2+Feb.+2023).>Google Scholar
De Vreese, Claes H. et al.One Size Fits All? Testing the Dimensional Structure of EU Attitudes in 21 Countries.” International Journal of Public Opinion Research 31.2 (2019): 195219.CrossRefGoogle Scholar
Edelson, Jack et al.The Effect of Conspiratorial Thinking and Motivated Reasoning on Belief in Election Fraud.” Political Research Quarterly 70.4 (2017): 933–46.CrossRefGoogle Scholar
Egelhofer, Jana L., and Lecheler, Sophie. “Fake News as a Two-Dimensional Phenomenon: A Framework and Research Agenda.” Annals of the International Communication Association 43.2 (2019): 97116.CrossRefGoogle Scholar
Elklit, Jørgen, and Reynolds, Andrew. “A Framework for the Systematic Study of Election Quality.” Democratization 12.2 (2005): 147–62.CrossRefGoogle Scholar
Engesser, Sven et al.Populism and Social Media: How Politicians Spread a Fragmented Ideology.” Information, Communication & Society 20.8 (2017): 1109–26.CrossRefGoogle Scholar
Fawzi, Nayla. “Untrustworthy News and the Media as “Enemy of the People?” How a Populist Worldview Shapes Recipients’ Attitudes Toward the Media.” The International Journal of Press/Politics 24.2 (2019): 146–64.CrossRefGoogle Scholar
Fawzi, Nayla, and Krämer, Benjamin. “The Media as Part of a Detached Elite? Exploring Antimedia Populism among Citizens and Its Relation to Political Populism.” International Journal of Communication 15 (2021): 32923314.Google Scholar
Fjelde, Hanne, and Höglund, Kristine. “Electoral Institutions and Electoral Violence in Sub-Saharan Africa. British Journal of Political Science 46.2 (2016): 297320.CrossRefGoogle Scholar
Franklin, Mark et al.Uncorking the Bottle: Popular Opposition to European Unification in the Wake of Maastricht.” Journal of Common Market Studies 32(1994): 455.CrossRefGoogle Scholar
Freedom House. Freedom on the Net 2019 Key Finding: Politicians and Hyperpartisans Use Digital Means to Manipulate Elections, 2019. (last accessed 7 Feb. 2022).Google Scholar
Garnett, Holly Ann, and James, Toby S.. “Cyber Elections in the Digital Age: Threats and Opportunities of Technology for Electoral Integrity.” Election Law Journal: Rules, Politics, and Policy 19.2 (2020): 111–26.CrossRefGoogle Scholar
Goldberg, Andreas C. et al.Europinions: Public Opinion Survey.” GESIS Data Archive, Cologne. ZA5553 Data file Version 1.0.0, 2021. Google Scholar
Gunther, Richard et al.‘Fake News’ and the Defection of 2012 Obama Voters in the 2016 Presidential Election.” Electoral Studies 61 (2019): 102030.CrossRefGoogle Scholar
Hameleers, Michael et al.Mistake or Manipulation? Conceptualizing Perceived Mis-and Disinformation among News Consumers in 10 European Countries.” Communication Research 49.7 (2022): 919–41.CrossRefGoogle Scholar
Hobolt, Sara B.Citizen Satisfaction with Democracy in the European Union.” JCMS: Journal of Common Market Studies 50 (2012): 88105.Google Scholar
Hooghe, Liesbet, and Marks, Gary. “Calculation, Community and Cues: Public Opinion on European Integration.” European Union Politics 6.4 (2005): 419–43.CrossRefGoogle Scholar
Imai, Kosuke et al.A General Approach to Causal Mediation Analysis.” Psychological Methods 15.4 (2010): 309.CrossRefGoogle ScholarPubMed
Judge, Elizabeth F., and Korhani, Amir M.. “Disinformation, Digital Information Equality, and Electoral Integrity.” Election Law Journal: Rules, Politics, and Policy 19.2 (2020): 240–61.CrossRefGoogle Scholar
Karp, Jeffrey A. et al.Dial ‘F’for Fraud: Explaining Citizens Suspicions about Elections.” Electoral Studies 53 (2018): 11–9.CrossRefGoogle Scholar
Kernell, Georgia, and Mullinix, Kevin J.. “Winners, Losers, and Perceptions of Vote (mis) Counting.” International Journal of Public Opinion Research 31.1 (2019): 124.CrossRefGoogle Scholar
Kerr, Nicholas. “Manipulating Elections across the World.” Election Law Journal 15.1 (2017): 109–14.CrossRefGoogle Scholar
Kerr, Nicholas, and Lührmann, Anna. “Public Trust in Manipulated Elections: The Role of Election Administration and Media Freedom.” Electoral Studies 50 (2017): 5067.CrossRefGoogle Scholar
Kofi Annan Foundation. Deepening Democracy: A Strategy for Improving the Integrity of Elections Worldwide. International Institute for Democracy and Electoral Assistance, 2012.Google Scholar
Krämer, Benjamin. “Populism, Media, and the Form of Society.” Communication Theory 28.4 (2018): 444–65.CrossRefGoogle Scholar
Lago, Ignacio, and Martinez i Coma, Ferran. “Challenge or Consent? Understanding Losers’ Reactions in Mass Elections.” Government and Opposition 52.3 (2017): 412–36.CrossRefGoogle Scholar
Macaraig, Ayee, and Hameleers, Michael. #DefendPressFreedom: Paradigm Repair, Role Perceptions and Filipino Journalists’ Counterstrategies to Anti-Media Populism and Delegitimizing Threats.” Journalism Studies 23.16 (2022): 2078–96.CrossRefGoogle Scholar
Marquart, Franziska et al.‘This Time I’m (not) Voting’: A Comprehensive Overview of Campaign Factors Influencing Turnout at European Parliament Elections.” European Union Politics 21.4 (2020): 680705.CrossRefGoogle Scholar
Martinez i Coma, Ferran, and Van Ham, Carolien. “Can Experts Judge Elections? Testing the Validity of Expert Judgments for Measuring Election Integrity.” European Journal of Political Research 54.2 (2015): 305–25.CrossRefGoogle Scholar
Mauk, Marlene. “Electoral Integrity Matters: How Electoral Process Conditions the Relationship Between Political Losing and Political Trust.” Quality & Quantity 56 (2022): 1709–28.CrossRefGoogle Scholar
Mauk, Marlene, and Grömping, Max. “Online Disinformation Predicts Inaccurate Beliefs About Election Fairness Among Both Winners and Losers. Comparative Political Studies (2023): 00104140231193008.CrossRefGoogle Scholar
Mitchell, Amy et al. In Western Europe, Public Attitudes Toward News Media More Divided by Populist Views Than Left-Right Ideology. Pew Research Center, 2018. (last accessed 16 Feb. 2023)Google Scholar
Newman, Nic et al. Reuters Institute Digital News Report 2019. 2019.−06/DNR_2019_FINAL_0.pdf (last accessed 15 Feb. 2023)Google Scholar
Norris, Pippa. Why Electoral Integrity Matters. New York: Cambridge UP, 2014.CrossRefGoogle Scholar
Norris, Pippa. Why Elections Fail. New York: Cambridge UP, 2015.CrossRefGoogle Scholar
Norris, Pippa et al.The Paranoid Style of American Elections: Explaining Perceptions of Electoral Integrity in an Age of Populism.” Journal of Elections, Public Opinion and Parties 30.1 (2020): 105–25.CrossRefGoogle Scholar
Norris, Pippa, and Grömping, Max. “PEI Country-Level Data (PEI_7.0) v2 09–05–2019.xlsx.” Perceptions of Electoral Integrity, (PEI−7.0), 2019., Harvard Dataverse, V2.CrossRefGoogle Scholar
Partheymueller, Julia et al.Participation in the Administration of Elections and Perceptions of Electoral Integrity.” Electoral Studies 77 (2022): 102474.CrossRefGoogle Scholar
Plescia, Carolina. “On the Subjectivity of the Experience of Victory: Who Are the Election Winners? Political Psychology 40.4 (2019): 797814.CrossRefGoogle Scholar
Plescia, Carolina et al.Do European Elections Enhance Satisfaction with European Union Democracy? European Union Politics 22.1 (2021): 94113.CrossRefGoogle Scholar
Reporters sans Frontières. 2019 World Press Freedom Index. 2019. (last accessed 7 Feb. 2022).Google Scholar
Rohrschneider, Robert. “The Democracy Deficit and Mass Support for an EU-Wide Government.” American Journal of Political Science 46.2 (2002): 463–75.CrossRefGoogle Scholar
Sánchez-Cuenca, Ignacio. “The Political Basis of Support for European Integration.” European Union Politics 1.2 (2000): 147–71.CrossRefGoogle Scholar
Schulz, Anne et al.We Are the People and You are Fake News: A Social Identity Approach to Populist Citizens’ False Consensus and Hostile Media Perceptions.” Communication Research 47.2 (2020): 201–26.CrossRefGoogle Scholar
Stiers, Dieter et al.What Makes People Believe That Their Party Won the Election? Electoral Studies 55 (2018): 21–9.CrossRefGoogle Scholar
Tenove, Chris. “Protecting Democracy from Disinformation: normative Threats and Policy Responses.” The International Journal of Press/Politics 25.3 (2020): 517–37.CrossRefGoogle Scholar
Tingley, Dustin et al.mediation: R Package for Causal Mediation Analysis.” Journal of Statistical Software 59.5 (2014): 138.CrossRefGoogle Scholar
Transparency International. 2019. (last accessed 7 Feb. 2022).Google Scholar
Van Kessel, Stijn et al.Informed, Uninformed or Misinformed? A Cross-National Analysis of Populist Party Supporters across European Democracies.” West European Politics 44.3 (2021): 585610.CrossRefGoogle Scholar
Waisbord, Silvio. “Truth Is What Happens to News: On Journalism, Fake News, and Post-Truth.” Journalism Studies 19.13 (2018): 1866–78.CrossRefGoogle Scholar
Zimmermann, Fabian, and Kohring, Matthias. “Mistrust, Disinforming News, and Vote Choice: A Panel Survey on the Origins and Consequences of Believing Disinformation in the 2017 German Parliamentary Election.” Political Communication 37.2 (2020): 215–37.CrossRefGoogle Scholar
Figure 0

Figure 1. Perceptions of election fairness across ten European countries.

Figure 1

Figure 2. Perceptions of no fake news influence across ten European countries.

Figure 2

Figure 3. Perceptions of election fairness vs. country-level integrity (PEI index) across ten European countries.

Figure 3

Table 1. Explaining perceptions of election fairness (OLS models)

Figure 4

Table 2. Explaining perceptions of election fairness (mediation model)

Figure 5

Figure 4. Mediation results for political populism.*P < 0.05, **P < 0.01, ***P < 0.001.Note: Non-parametric bootstrap for variance estimation; confidence intervals (95%, 10'000 simulations) do not include zero for the indirect path (CI [−0.005, −0.0003]), which is thus statistically significant.

Figure 6

Figure 5. Mediation results for anti-media populism.*P < 0.05, **P < 0.01, ***P < 0.001.Note: Non-parametric bootstrap for variance estimation; confidence intervals (95%, 10'000 simulations) do not include zero for the indirect path (CI [−0.023, −0.008]), which is thus statistically significant.

Supplementary material: File

Goldberg and Plescia supplementary material

Goldberg and Plescia supplementary material
Download Goldberg and Plescia supplementary material(File)
File 217.9 KB