Introduction
Citizens do not evaluate democracy against a single shared standard. Some primarily expect free and fair electoral competition; others emphasize rights and the rule of law, social protection, direct citizen influence, or an unmediated popular will. This matters because the same democratic institutions can appear more or less satisfactory depending on the democratic standard citizens use, making it an important task to better grasp citizens’ specific understandings of democracy (UoD) and the effect these understandings have on their political attitudes and behavior (among others, see Dalton et al., Reference Dalton, Sin and Jou2007; Cho, Reference Cho2015; Ferrin and Kriesi, Reference Ferrin and Kriesi2016; Claassen et al., Reference Claassen2025).
In this research note, we contribute to this debate by asking whether these normative benchmarks of citizens’ UoD help explain satisfaction with democracy (SwD) and the forms and levels of political participation (PP) citizens choose when they see a gap between democratic ideals and democratic practice. A better understanding of the linkage between the three concepts will also help us analyze why several groups of citizens become dissatisfied with democracy, disengaged with politics, and potentially more likely to support anti-democratic actors. Taking up the seminal ideas by Lu and Chu (Reference Lu and Chu2022) and Landwehr and Steiner (Reference Landwehr and Steiner2017) as well as the desiderates identified by König et al. (Reference König, Siewert and Ackermann2022), we make two systematic contributions to the literature.
First, conceptually, we propose a theoretical framework outlining the mechanisms linking UoD, SwD, and PP. We hypothesize that UoD is associated with patterns of unconventional political participation (UPP) and that this relationship may be mediated by SwD. In this framework, SwD serves as an indicator of the perceived congruence between individuals’ democratic ideals and their evaluations of how democracy functions in practice.
Second, and empirically, we extend the test of the link between UoD, SwD, and PP – which Lu and Chu (Reference Lu and Chu2022) have analyzed for the Global South – to the European context, as more recent research indicates substantive heterogeneity in the coherence of democracy conceptions, specifically between publics in more and less developed countries (Claassen et al., Reference Claassen2025). Europe serves as a critical case that allows us to test if these theoretical mechanisms persist in established democracies with high degrees of democratic institutionalization.
Using the most recent ESS wave, which includes measures of a range of key UoD discussed in the conceptual literature (ESS Round 10, 2020), we find that citizens’ UoD are associated with their SwD, with citizens holding an electoral understanding of democracy being much more satisfied than citizens with direct, social, liberal or populist UoD. Furthermore, citizens’ UoD have an effect – both directly and mediated via SwD – on their UPP, with citizens being more satisfied with democracy being less likely to participate unconventionally. In the final part, we reflect both on the implications but also the limitations of our findings.
Theory
Democracy remains a contested multidimensional concept. Minimally, it can mean free and fair elections (Schumpeter, Reference Schumpeter1942) as well as a more substantial interpretation that includes principles of fundamental rights, accountability and transparency (Dahl, Reference Dahl1971). While the scholarly debate about what makes a democracy is still ongoing, citizens also have varied perspectives on what democracy means, which can diverge from scholarly definitions. Research on citizens’ UoD has hence sought to analyze what democracy means for citizens (for a review, see König et al., Reference König, Siewert and Ackermann2022). This research has become all the more pressing by the puzzling observation that SwD among citizens remains at very high levels while at the same time anti-democratic actors with populist and authoritarian political programs seem to gain more and more public approval (König et al., Reference König, Siewert and Ackermann2022, Lu and Chu, Reference Lu and Chu2022).
The reason for this, as Lu and Chu (Reference Lu and Chu2022) as well as Landwehr and Steiner (Reference Landwehr and Steiner2017) claim, is that citizens have divergent concepts or UoD which are connected to different expectations.Footnote 1 These divergent expectations are fulfilled to varying degrees by a country’s current democratic regime. This results in some citizens perceiving their understanding of democracy to be fulfilled by their country’s democratic regime, while this is not true for citizens with other UoD. In effect, the former citizens should report higher levels of SwD, while the latter should report lower levels. However, both groups of citizens perceive themselves as democrats, making UoD an instance “where democrats disagree,” the title of Landwehr and Steiner’s article (Reference Landwehr and Steiner2017).
In this research note, we elaborate on this mechanism in two ways. First, rather than presenting a new conceptualization of what citizens may have in mind when thinking about democracy, we focus on the conceptualization brought forward by Ferrin and Kriesi (Reference Ferrin and Kriesi2014, see also Ferrin et al., Reference Ferrin, Hernandez and Landwehr2023). While we acknowledge that there are other conceptualizations of UoD (Gherghina and Geissel, Reference Gherghina and Geissel2017, Reference Gherghina and Geissel2020; Pilet and Gherghina, Reference Pilet and Gherghina2024; Claassen et al., Reference Claassen2025), the Ferrin and Kriesi conceptualizations is not only theoretically consistent but also has the great advantage of being measured for a wide range of European countries in Round 10 of the ESS. Second, we do not only focus on the effects UoD may have on SwD, but also on whether or not citizens participate politically and how.
We start out by differentiating between five concepts of democracy people may have in mind when thinking about a democratic system (Ferrin and Kriesi, Reference Ferrin and Kriesi2014; Ferrin et al., Reference Ferrin, Hernandez and Landwehr2023).Footnote 2 Citizens prioritize an electoral model of democracy if the minimal definition of democracy is fulfilled, that is, whether a political system allows citizens to vote for more than one party in free and fair elections. Other citizens may understand democracy mostly in the form of a liberal model of democracy, ensuring that everyone is treated the same way. Furthermore, citizens may consider the model of social democracy – a regime being democratic if it helps the poor or prevents too high levels of economic inequality – to be most important. Citizens prioritizing the direct model of democracy connect democracy with the people having a direct say in politics, for example, by referendums. Finally, citizens may also share a populist understanding of democracy connected to the idea that the will of a homogenous people should not be obstructed by corrupt elites.
How are these different UoD connected to the way citizens feel about the state of democracy in their country? Following the argument developed by Landwehr and Steiner (Reference Landwehr and Steiner2017) and the “literal approach” of conceptualizing SwD (Singh and Mayne, Reference Singh and Mayne2023), citizens should evaluate their democratic regime in relation to how well it matches their core understanding of democracy. While we acknowledge potential feedback loops, we hold that it is most plausible to expect UoD, which are considered relatively stable and socialized identities (Ferrin and Kriesi, Reference Ferrin and Kriesi2016), to influence the more volatile SwD. Given that currently most democratic systems at least enable free and fair elections, we should observe the highest level of SwD among those who primarily hold an electoral understanding of democracy (H1).Footnote 3
Furthermore, we share the idea of Lu and Chu (Reference Lu and Chu2022) that citizens’ evaluations of the democratic regime of their country may also be reflected by the degree to which they participate politically. Yet, elaborating on the conceptualization by Lu and Chu (Reference Lu and Chu2022), we additionally argue that the channels of PP citizens use result – at least partly – from citizens’ SwD.
In this research note, we focus on the effect of UoD and SwD on UPP.Footnote 4 We focus specifically on UPP because it represents a more sensitive stress test for the theory. While conventional acts like voting are often habitual, protest carries higher individual costs. Therefore, mismatches between citizens’ normative UoD and the system’s actual performance should leave a more direct imprint on UPP.
We argue that citizens who prefer other core UoD over electoral democracy should be more likely to show higher levels of UPP due to two mechanisms, a direct and a mediated one. First, building on the insights of Gherghina and Geissel (Reference Gherghina and Geissel2017), citizens holding UoD that are associated with a more direct influence of citizens – direct or populist UoD – should also be more likely to report PP behavior that aims to work around the indirect process of electoral democracy and promises a direct influence on political decision-making. Hence, we would expect citizens holding primarily direct or populist UoD to be more likely to report unconventional forms of participation (H2).
Second, citizens preferring a format of democracy that is conceptually and normatively “thicker” – that is, encompasses more complex and more elaborate features – may be less likely to be satisfied with democracy, as we argued above. While the electoral understanding of democracy is a very basic one, the other conceptions may be considered more critical and more demanding. People holding such UoD hence have a higher likelihood of being disappointed by the actual implementation of democracy. We expect that this lower degree of satisfaction should also make them more likely to choose more active forms of PP that are outside the usual, more conventional channels of influence in the current representative format of democracy. The reason for this is that their lower degree of SwD is likely to be associated with lower levels of trust in the representative system and the actors representing them. Hence, these citizens are also less likely to make their voice heard only via the conventional channels of electoral-representative democracy. Instead, they may perceive the unconventional channels as ways to not only express their dissatisfaction but also to potentially change something about the existing democratic format, by increasing the visibility of their discontent with politics, policies, and polity. Formally, we therefore expect the following mediating effect of SwD on the relationship between UoD and UPP: liberal, social, direct, and populist understandings should be negatively associated with SwD, and SwD should be negatively associated with UPP (H3).
Research design: data and modeling
To analyze the hypotheses we developed, we make use of Wave 10 of the European Social Survey (ESS, 2020) including the measures needed to test our hypotheses. This survey does not only cover 31 countries with 59,865 respondents in total, but it is also considered to be one of the most methodologically rigorous cross-country survey enterprises.
In addition, and regarding our most substantive variable of interest – citizens’ core or prioritized understanding of democracy –, the ESS measures not only the degree to which citizens share several conceptions of democracy, but it additionally asks citizens to indicate which conception or dimension they consider to be most important for democracy. In the previous Wave 6 (ESS, 2012), these dimensions comprised the electoral dimension, the liberal dimension, the social dimension, the direct dimension, the inclusiveness dimension, and the representative dimension (Ferrin and Kriesi, Reference Ferrin and Kriesi2014). In Wave 10, the team coordinating the section on UoD proposes to think about five models (or: dimensions) of democracy, the liberal, the social, the direct, the populist, and the multilevel model of democracy (Ferrin et al., Reference Ferrin, Hernandez and Landwehr2023). In the latter version, the team therefore subsumes electoral democracy under the concept of liberal democracy. This leads to the somewhat surprising decision to measure the liberal dimension with two questions in the item asking citizens what they consider most important for democracy, while only using one statement for the social, the direct, and the populist model (and excluding the multilevel model). As such a procedure artificially inflates the number of respondents choosing a liberal understanding of democracy to be most important, we return to the original conceptualization and treat the statement on free and fair national elections as a measurement capturing the electoral understanding of democracy (Ferrin and Kriesi, Reference Ferrin and Kriesi2014).
Methodologically, asking respondents to choose the single most important feature of democracy, rather than endorsing or rating multiple items, offers distinct analytical advantages. This forced-choice format captures respondents’ prioritization among competing democratic principles and avoids acquiescence bias, which tends to inflate agreement with all listed items regardless of true preference.
Table 1 indicates our revised measure of our central variable, respondent’s prioritized understanding of democracy. Respondents are shown all of the five statements (in a randomized order) included in the table and are then asked to indicate which one of the statements they find most important for democracy. Based on their answers, we create a nominal variable indicating which of the conceptions connected to the statements is prioritized by each respondent when thinking about democracy. In the remainder, we call this variable citizens’ prioritized understanding of democracy.
Measuring prioritized understandings of democracy
Table 1. Long description
A table categorizing understandings of democracy by statement. The table has two columns and five rows. The first column is labeled ’Understanding of democracy’ and the second column is labeled ’Measured by statement’. The rows are as follows: Row 1: Electoral, That national elections are free and fair. Row 2: Liberal, That the courts treat everyone the same. Row 3: Social, That the government protects all citizens against poverty. Row 4: Direct, That citizens have the final say on the most important political issues by voting in referendums. Row 5: Populist, That the views of ordinary people prevail over the views of the political elite.
Figure 1 illustrates the descriptive distribution of respondents on this variable. One immediately notices that most respondents consider the electoral dimension of democracy to be their prioritized understanding. About the same number of respondents connect the core understanding of democracy to the direct, the social or the liberal model, while considerably fewer respondents connect democracy primarily to the populist model.
Distribution of prioritized understandings of democracy.
Our second substantive variable measures respondents’ satisfaction with the way democracy works in their country, on an 11-point scale ranging from 0 (extremely dissatisfied) to 10 (extremely satisfied).
To operationalize UPP, we draw on an ESS question asking respondents whether they have engaged in any of the following four forms of UPP during the past twelve months: signing petitions, attending demonstrations, boycotting products, and participating in online debates. The UPP score was calculated by summing the binary values within the index and dividing it by 4, resulting in a score ranging from 0 to 1.
Furthermore, we control for the usual sociodemographic variables, age (in years), education (in years), and gender (with 2 = female and 1 = male), as well as for the country context in which a respondent is living. We min-max transform these variables for easier interpretability.
When analyzing the impact of prioritized or core UoD on SwD, we also control for the role of what has become known as the winner–loser gap. Previous literature has shown that one of the key determinants of SwD is people’s vote choices in the last national election. Individuals who voted for a party that won the last national election are significantly more satisfied with democracy than individuals who voted for another party (e.g., see Nadeau et al., Reference Nadeau, Daoust and Dassonneville2023). We measure this by a dummy variable called “electoral winner” which receives the value 1 if a respondent reported that they voted for a party in the last national election which ended up in government and 0 otherwise. Party-level data are drawn from the WZB’s Political Parties and Electoral Programs (PPEG) database (Krause and Stelzle, Reference Krause and Stelzle2024).Footnote 5
Regarding the models, we analyze data from the ESS Wave 10 employing multilevel mixed models with random country intercepts. In doing so, we account for unobserved cross-national heterogeneity and isolate the individual-level drivers without overspecifying with a limited number of second-level units. To mitigate the dummy variable trap (Dougherty, Reference Dougherty2011), we select the electoral understanding of democracy as the reference category. This choice not only follows the conventional practice of opting for the most prevalent category as the reference group to enhance the stability of estimates, it is also motivated theoretically. While there is plenty of disagreement over how to conceptualize democracy, most scholars of democracy will agree that having free and fair elections is a key criterion of any democracy (Przeworski et al., Reference Przeworski, Alvarez, Cheibub and Limongi2000; Beckman, Reference Beckman and Hannah2021).
For testing H2 and H3, we estimated an SEM model with a maximum likelihood estimator in which we allowed different prioritized UoD to have a direct effect on UPP as well as an indirect effect, mediated by SwD.Footnote 6 We cluster observations at country level to reflect the hierarchical and nested nature of our data. This approach prioritizes a parsimonious test of the individual-level mechanism, which is the core focus of this research note, while controlling for unobserved national-level dependencies. As a robustness check, we also ran a model with bootstrapped standard errors which yielded similar results compared to the ones we report below (see online Appendix for the robustness checks).
Empirical findings
Figure 2 plots the findings of a first analysis of H1 (for the full model, see Table A1 in the Appendix). The main insight is that the different prioritized UoD have significant effects on SwD even if we control for the electoral winner–loser gap and a range of sociodemographics. The direction of the effects is in line with our expectations so that our hypothesis is confirmed. Respondents who understand democracy to be primarily about the electoral dimension display the highest levels of SwD (=H1).
Explaining satisfaction with democracy.
A look at the effect sizes suggests that respondents with a primarily populist understanding of democracy are the least satisfied with the way democracy works in their country, closely followed by respondents with a primarily direct understanding of democracy. On average, endorsing populist and direct democracy is associated with a decrease in SwD comparable to the increase associated with being an electoral winner. Proponents of primarily liberal and social UoD are also less satisfied with democracy than respondents with a primarily electoral understanding of democracy but the gap is noticeably smaller. Yet, it should be noted that the effects of primarily liberal and primarily electoral UoDs are significantly different, suggesting that it still makes sense to distinguish these two dimensions.
How do prioritized UoD affect UPP? We argued that there is not only a direct but also a mediated effect of prioritized UoD on UPP via citizens’ degree of SwD. H2 expects participants with direct and populist core understandings to be more likely to report UPP. In H3 we expected that liberal, social, direct, and populist understandings have a negative effect on SwD, and SwD should mediate citizens’ UoD by having a negative effect on UPP.
Figure 3 demonstrates that our expectations are partly met. When it comes to the effects of prioritized UoD on SwD, we find that relative to the reference category of a primarily electoral understanding of democracy, respondents with another prioritized understanding of democracy are significantly and substantively less satisfied with democracy. And when looking at the effect of SwD on unconventional forms of PP, we find the pattern expected in H3 – higher SwD decreases the likelihood of participating in unconventional ways. Conversely, lower SwD makes respondents more likely to participate unconventionally.
Explaining unconventional forms of participation.
Yet, the direct effects we expected in H2 are not corroborated by the empirical findings. We actually observe a significantly lower likelihood of citizens with a direct or populist understanding of democracy to participate in unconventional forms. These direct effects of democracy conceptions are generally stronger than the indirect effects. This suggests that while the indirect pathways through SwD do contribute to PP, their influence is weaker than the direct effects. Additionally, the negative direct effects counteract the positive indirect effects. The positive indirect effects through SwD imply that although the non-electoral prioritized UoD are associated with lower satisfaction, this lower satisfaction, in turn, increases the likelihood of UPP.
Conclusion
In this research note, we contribute to the previous literature by shedding new light on the relationship between citizens’ UoD, their SwD, and their level of UPP. Building on insights from the context of the Global South, we contribute the first systematic study on the European context. Using data from Wave 10 of the ESS, covering a plenitude of countries and an impressive number of respondents, we also demonstrate the empirical plausibility of a novel theoretical argument, which revolves around the idea that SwD can be considered a proxy for people’s perceptions of the extent to which their preferred form of democracy is implemented in their respective country.
We show that prioritized UoD have a systematic association with SwD. The more people’s primary understanding of democracy differs from the standard model of electoral democracy, the less satisfied they are with democracy. The same benchmarking logic applies to the relationship between UoD, SwD and people’s levels of UPP. Satisfaction with democracy is a substantive mediator of UoD’s varying impact on UPP. People holding a primarily electoral understanding of democracy display higher levels of SwD, and this then translates into a lower likelihood of participating in unconventional forms of PP. People with prioritized UoD that deviate from the standard electoral model, in contrast, display lower levels of SwD, and these lower levels of SwD increase their likelihood of engaging in protest activities and other forms of UPP.
Our results have implications for the discussion of how to best conceptualize prioritized UoD. While the electoral dimension has recently been subsumed under the liberal understanding of democracy (Ferrin et al., Reference Ferrin, Hernandez and Landwehr2023), our analysis demonstrates that liberal UoD, narrowly defined, have a noticeably different effect on both SwD and UPP. This suggests that it might make sense to continue treating them as separate UoDs, as originally proposed by Ferrin and Kriesi (Reference Ferrin and Kriesi2014).
Our analysis raises two questions that should be analyzed in future research: First, the theoretical mechanism presented in this article rests on the argument that SwD represents a proxy for people to assess to what extent their ideal visions of democracy are met in their country. While our results are in line with this argument, observational data such as the ones we have used in our analysis are not sufficient for a full test of the underlying mechanism. Hence, future research should apply more causally oriented methodological designs to further explore the mechanism.
Second, our analysis revealed a surprising empirical pattern: liberal, direct, social, and populist UoD (relative to an electoral understanding) all have negative effects on UPP, indicating that – beyond the effects of UoD mediated by SwD – there must be something about the UoD themselves and why respondents prioritize them that makes them less likely to participate. We do not have a good theoretical explanation for these effects, which we did not theorize. In addition, the heterogeneity found by Claassen et al. (Reference Claassen2025) for citizens’ coherence of the liberal understanding of democracy may also be an issue for the otherwise elegant and parsimonious UoD measures of the ESS. Such heterogeneity in citizens’ interpretation and normative expectations of the respective understandings may also result in such surprising effects. Hence, future research, combining observational and experimental data or survey data with more qualitative approaches should seek to shed more light on how UoD shape the way in which people do (or do not) participate politically. While our models account for country-level variance, future research could also explore how specific institutional contexts, such as degrees of polarization, might moderate the strength of these individual-level effects.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1755773926100654.
Data availability statement
The replication code can be found in the supplementary material. For the analysis, we used a merged file of ESS Round 10 that combines the data from countries with self-completion and interviewer-based survey formats. Our colleague Carsten Wegscheider kindly shared this merged file with us. While we are unable to upload this data file on a public platform for legal reasons, researchers will receive the data set by directly contacting the corresponding author.
Acknowledgements
The authors want to thank Emely Heydorn for competent research assistance as well as the participants of the INCITE-DEM workshop in Groningen in March 2025 for discussing our findings.
Financial support
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Horizon Europe Programme (project INCITE-DEM, grant agreement 101094258).
Competing interests
None.
Use of AI
The authors have not used AI at any stage of writing the paper.
Open access
