Introduction
To survive, democracy needs at least a minimum level of support from citizens (Dahl Reference Dahl1971; Easton Reference Easton1953). Yet traditional institutions and mechanisms of liberal representative democracy are facing serious legitimacy challenges (Dalton Reference Dalton2016; Keane Reference Keane2009; Norris Reference Norris2011; Tormey Reference Tormey2015). In many countries, democratic discontent is high. When people do not feel adequately represented by politicians, they are more likely to abstain from elections or to vote for extremist parties whose allegiance to the principles of liberal representative democracy is questionable (De Mulder Reference De Mulder2024). The realization that legitimacy is under threat has fueled a vast literature on political support for democracy as a political system (Claassen Reference Claassen2020), on confidence in government and other key institutions of liberal representative democracy (Zmerli and van der Meer, Reference Zmerli and van der Meer2017), and on satisfaction with the way democracy works (SWD) (Anderson and Guillory Reference Anderson and Guillory1997; Singh and Mayne Reference Singh and Mayne2023).
Citizens’ perception that they are not adequately represented by political parties, governments, parliaments, and other key institutions of liberal representative democracy raises the question of whether alternative actors – outside the realm of formal politics – might alleviate this democratic discontent (Montanaro Reference Montanaro2012, Reference Montanaro2017). An example of such an actor is Greta Thunberg, who Time magazine named Person of the Year in 2019 (Felsentahl Reference Felsentahl2019). The Swedish climate activist has gained a massive international following and significant media attention with her school strikes and Fridays for Future movement. She has been invited to speak at the European Parliament, the UN General Assembly, and the World Economic Forum, as well as to private meetings with world leaders such as Angela Merkel. Her speech to the House of Commons in April 2019 was followed by the British Parliament declaring a ‘climate emergency’ within days. In a way, Thunberg acts as an ‘unelected representative’ (Keane Reference Keane2009). She claims to speak on behalf of future generations, and these claims gain significant resonance among young people, mass media, and political authorities. Her actions take place outside of the electoral channel of representation, yet nevertheless play an influential role in how democracy works. As a result, they might influence the legitimacy of the system.
The conceptualization of this type of unelected representation is rooted in the theoretical innovation of the constructivist turn in political representation (Disch, van de Sande, and Urbinati Reference Disch, van de Sande and Urbinati2019; Wolkenstein Reference Wolkenstein2024). Central to this constructivist account is the representative claim (Disch, van de Sande, and Urbinati Reference Disch, van de Sande and Urbinati2019; Saward Reference Saward2010). In all societies, not least in established democracies, a wide variety of political activists often claim to be representatives of various constituencies (De Wilde Reference De Wilde2020; Saward Reference Saward2010). Unelected representatives, as we understand them, are political activists without a formal political office who publicly voice a representative claim (ie a policy demand on behalf of a constituency) and gain significant resonance in the public sphere because of it. In light of the long-term trends of personalization and mediatization of politics (Esser and Strömbäck Reference Esser, Strömbäck, Esser and Strömbäck2014) that tend to empower individuals over organizations, constructivist theorists argue that individual unelected representatives may constitute an alternative channel of representation in pluralist democratic systems (Disch Reference Disch2011; Saward Reference Saward2010). Yet the implications of this alternative representation channel for indicators of democratic legitimacy, like citizens’ satisfaction with democracy, remain understudied empirically.
As explained by Merkley et al. (Reference Merkley, Cutler, Quirk and Nyblade2019): ‘… in mass democracies with a complex field of electoral and non-electoral representation (Kuyper Reference Kuyper2016) the voicing of political positions and agenda-preferences, as well as ‘representative claims’ (Saward Reference Saward2006) by many actors in the system, all contribute to a citizen’s evaluation of the degree and quality of recognition, deliberation and representation in the polity’ (850). We apply such argumentation to unelected representatives and ask: When activists behave as unelected representatives by voicing political demands on behalf of other people, does that affect citizens’ satisfaction with democracy? That is the central research question of this article. We argue that claims from unelected representatives may increase satisfaction with democracy when such individuals constitute an effective channel of representation – that is, when (a) their demands substantively represent citizens (Pitkin Reference Pitkin1967) and (b) they are included in political processes (Merkley et al. Reference Merkley, Cutler, Quirk and Nyblade2019). Furthermore, we theorize that underrepresented or marginalized individuals, in the forms of (1) being an electoral ‘loser’ and (2) having low levels of political efficacy, become more satisfied with the way democracy works when they witness activists making claims they agree with.
We investigate the research question in a cross-national survey experiment fielded in four European countries: Sweden, Germany, Italy, and Romania (N = 8196). These countries range from very affluent, high-trust societies to poorer, low-trust societies within the European context. The cross-national research design strengthens the external validity of the findings and allows us to explore contextual scope conditions for the relationship between unelected representatives’ claim-making and citizens’ satisfaction with democracy. Overall, our results are mixed. Substantive representation from unelected representatives alone is not enough to increase SWD. We find mixed support for the prediction that issue-congruent claims from unelected representatives further increase SWD when these representatives are included in the political process. Corroborated through robustness checks, we find that electoral losers become less dissatisfied with the way democracy works when they witness unelected representatives making claims that substantively represent them, irrespective of whether or not these activists are included in politics. In other words, substantive representation through unelected representatives reduces the ‘winner–loser gap’ in satisfaction with democracy, one of ‘the most robust relationships documented in political science’ (Nadeau, Daoust, and Dassonneville Reference Nadeau, Daoust and Dassonneville2023: 483). This effect is strongest in Romania, less so in Italy and Germany, and almost absent in Sweden. This might be because countries with a higher quality of government have smaller winner–loser gaps (Dahlberg and Linde Reference Dahlberg and Linde2016), providing less opportunity for unelected representatives to reduce the gap.
The key implication is that citizens look beyond their elected representatives when assessing their satisfaction with democracy. Hence, citizens’ opinions about the way democracy works may be affected by informal politics and individuals without formal political standing. In the conclusion, we reflect on how these logics may be shaped by domestic contexts.
Theoretical framework
Unelected representatives and democracy: a constructivist perspective
In the constructivist turn in representation, political theorists like Disch (Reference Disch2011) and Saward (Reference Saward2010) challenged the standard account of how we understand the concept of representation. Where the standard account sees political representation as the product of elections, the constructivist perspective reverses the relationship between representation and authorization. Instead of starting with authorization in the form of elections and then asking whether this generates adequate representation, these authors start with the practice of representation in the form of ‘representative claims’ (Disch Reference Disch2011, Reference Disch2021; Kuyper Reference Kuyper2016; Montanaro Reference Montanaro2017; Näsström Reference Näsström2015; Saward Reference Saward2006, Reference Saward2010) and conceive of authorization as a possible response to this practice. This reversal allows them to problematize the role of both elected and unelected individuals in political representation. Candidates running for election make a range of claims about what they stand for and whom they represent. Voters subsequently judge these claims and authorize their preferred candidate through the ballot. But many more individuals make representative claims, ranging from activists like Greta Thunberg (Zulianello and Ceccobelli Reference Zulianello and Ceccobelli2020) to celebrities like Ben Affleck, Angelina Jolie, and Leonardo DiCaprio (Budabin and Richey Reference Budabin and Richey2021; Street Reference Street2004). These individuals contribute as ‘unelected representatives’ (Keane Reference Keane2009) to political processes of agenda-setting, decision-making, policy formulation, and – ultimately – the authoritative allocation of resources. Hence, they may, in that capacity, influence citizens’ opinions about democracy and the legitimacy of the system.
Unelected representatives are authorized too – albeit not through votes. Instead, their authorization comes in organizational or discursive form (Dryzek and Niemeyer Reference Dryzek and Niemeyer2008; Montanaro Reference Montanaro2018). Citizens can demonstrate their acceptance of activists as representatives through donations, joining an organization, or through engagement on social media. Exposure to claims by unelected representatives can trigger feelings of representation among citizens (Vik and de Wilde Reference Vik and de Wilde2024), which may subsequently lead to authorizing behavior on their part. Politicians and journalists can facilitate or inhibit such authorization, which in turn can have profound policy implications.
Greta Thunberg’s authorization, for example, has been amplified by politicians who have invited her to speak, journalists who have decided to feature her in news stories, and citizens who have decided to follow her and to share her messages on social media. Thunberg’s activism contributed to elevating climate change on the political agenda (Sabherwal et al. Reference Sabherwal, Ballew, van der Linden, Gustafson, Goldberg, Maibach, Kotcher, Swim, Rosenthal and Leiserowitz2021), which in turn fueled a victory for Green parties in the European Parliament elections of 2019. These victories propelled the European Commission into developing the European Green Deal, a package of legislation instrumental in transitioning Europe toward a sustainable economy (Euronews 2021).
Democracy has always been a system that contains both electoral and non-electoral channels of preference aggregation (Held Reference Held1996; Urbinati and Warren Reference Urbinati and Warren2008; Warren Reference Warren1996). We vote for politicians who elect or control the executive. At the same time, we are members of or donors to interest groups such as employers’ organizations, trade unions, churches, or more diffuse interest groups like Amnesty and Greenpeace, who influence policy through lobbying, protests, and other actions. Adopted policies are a function of the relative power that different political parties and interest groups have in our pluralist polities (Schattschneider Reference Schattschneider1960). Empowered through the rise of social media and the more general trends of personalization and mediatization of politics (Altermark, Hadjievska, and Johansson Reference Altermark, Hadjievska and Johansson2023; Bennett and Segerberg Reference Bennett and Segerberg2013; Esser and Strömbäck Reference Esser, Strömbäck, Esser and Strömbäck2014; Mazzoleni and Schulz Reference Mazzoleni and Schulz1999), activists who speak out on behalf of broader societal groups may perform a similar function to political parties and interest groups: a function of preference articulation and aggregation, and of providing a key link between citizens and the arenas and processes of formal politics and policy formulation (Disch Reference Disch2011; Kuyper Reference Kuyper2016; Tormey Reference Tormey2015). Our study treads new ground through rigorous empirical investigation of the idea that citizens see unelected representatives as legitimate political representatives, akin to those who are elected (Vik et al. Reference Vik, de Wilde, Treib and Aarøe2025). If this is so, then ways in which elected politicians affect citizens’ opinions about the system might travel to the world of the unelected. This is a key part of the study presented here.
Proposing an effective channel of unelected representation: substantive representation and political inclusion
Satisfaction with democracy ‘taps into the extent to which citizens are satisfied with the way democracy works in their country in practice at a particular point in time’ (Hobolt, Hoerner, and Rodon Reference Hobolt, Hoerner and Rodon2021: 855). It is about ‘… system support at a low level of generalization. It does not refer to democracy as a set of norms but to the functioning of democracy’ (Anderson and Guillory Reference Anderson and Guillory1997: 70). Of various indicators that gauge the legitimacy of democracy and current liberal democratic representative political systems, we argue that ‘satisfaction with democracy’ is the most likely to be influenced through representative claims made by unelected representatives. This is so because the practice of claim-making, which involves the initial speech act by would-be representatives, followed by recognition from existing political authorities, resonance in the media, and citizens’ recognition is part and parcel of the functioning of democracy. It is a dynamic practice, an exchange between representatives and represented that unfolds over time. Representative claims (Saward Reference Saward2010) contribute to deliberation through the ‘… offering and receiving of cognitively compelling reasons about a common concern’ (Warren Reference Warren2017: 47). As such, they contribute to the workings of democracy understood in a maximalist, pluralist way, where democracy does not restrict itself to free and fair elections but also includes a healthy public sphere with influence exerted through public debate (Hoerner and Hobolt Reference Hoerner and Hobolt2020). Such influence often comes in the form of agenda-setting (Kingdon Reference Kingdon1984), or raising the salience of a particular issue, a process amplified through appearances in ‘strong publics’ (Fraser Reference Fraser1990) like parliaments, which may in turn sway elections or the positions of elected officials, ultimately translating into policy change. This is the path through which Thunberg exerted influence on EU climate policy. Because representative claims contribute to the way democracy works, we expect them to affect citizens’ satisfaction with democracy. This is in line with Merkley et al. (Reference Merkley, Cutler, Quirk and Nyblade2019), who argue ‘… the more wide-ranging (in terms of issues) and more inclusive (in terms of actors) political discourse is in a polity, the more citizens will feel that their claims are being made. Because of this, they will feel adequately represented and will judge their democracy more positively’ (850).
From this reasoning, we propose that unelected representatives may influence SWD if and when such individuals constitute an effective channel of representation. That happens when (1) activists substantively represent individuals and (2) they are included in political processes. Furthermore, we theorize that underrepresented or marginalized individuals, in the forms of (1) being an electoral ‘loser’ and (2) having low levels of political efficacy, become more satisfied with the way democracy works when they witness activists they agree with. We elaborate on this below.
Substantive representation: having your voice heard
Substantive representation is essential for the functioning of a healthy democracy, as it enables representatives to leverage their positions of authority and influence to actively advocate for policies that benefit their constituents (Dahl Reference Dahl1971). Scholars of political representation emphasize the significance of substantive congruence – the principle that elected officials’ policy preferences should mirror those of the constituents they represent (Thomassen and Schmitt Reference Thomassen and Schmitt1999). This alignment, known as issue congruence, is a crucial step in the representation process (Miller and Stokes Reference Miller and Stokes1963; Powell Reference Powell2004).
Democratic theory posits that for political representation to be effective, the opinions and preferences prevalent in society must be reflected among elected officials. As Pitkin (Reference Pitkin1967) defines it, substantive representation requires actions by representatives that reflect the interests and needs of the represented, rather than merely the act of being represented. In essence, substantive representation is often equated with congruence, where politicians’ policy preferences align with those of citizens (Thomassen and Schmitt Reference Thomassen and Schmitt1999).
The constructivist turn in political representation challenges the ontological foundations of this focus on substantive representation and congruence (Disch Reference Disch2021; Saward Reference Saward2010). Citizens’ policy preferences and group identity are constructed as part of the process of representative claims-making. In other words, policy preferences are endogenous to the process, rather than exogenous. While we subscribe to the constructivist turn, we see the process of reconstructed preferences and identity as a result of continued exposure to a plurality of representative claims. A single representative claim, in contrast, is unlikely to fundamentally alter citizens’ preferences or identity. Reconstruction is even less likely to happen as a result of exposure to a single claim when this claim is made by an unknown person with whom citizens have no prior constructed relationship. When that claim is made in a saturated context in which a political issue has long featured prominently in public debates, it is even less likely to reconstruct citizens’ preferences or identity. For the purposes of our single-shot experiment in which we expose citizens to one claim made by an unknown fictitious activist on a highly salient and polarized issue, we therefore treat citizen preferences as exogenous and analyze congruence as the proximity between that preference and the preference articulated in our stimulus.
Ultimately, the influence of claims made by unelected representatives can impact citizens’ satisfaction with democracy (SWD) in varying ways. Prior research has shown that a significant gap between citizens’ policy preferences and those of their preferred party correlates with lower satisfaction with democracy: ‘if what elites say and do is not what citizens want, democracy somehow falls short’ (André and Depauw Reference André and Depauw2017: 380). Wratil and Wäckerle (Reference Wratil and Wäckerle2023) shows that citizens’ legitimacy perceptions of the system are influenced by their perception of whether a majority of citizens support key policies, and that these legitimacy perceptions can be influenced even through single-shot survey experiments. Arnesen and Peters (Reference Arnesen and Peters2018) demonstrate that citizens take into account whether decision-makers have an official mandate when assessing the legitimacy of their decisions. This resonates with the call by Wolkenstein and Wratil (Reference Wolkenstein and Wratil2020) to direct the study of the effect of representative acts to those made by elected politicians, as they play the most important role in constructing relationships of political representation. But Vik et al. (Reference Vik, de Wilde, Treib and Aarøe2025) show that unelected representatives can evoke feelings of political representation among citizens too, despite not having a formal mandate. We thus anticipate that substantive representation or demand resonance – defined here as issue congruence, or the alignment between policy preferences articulated by representatives and those held by citizens – will enhance citizens’ satisfaction with democracy. This alignment increases citizens’ sense of having a voice in public discourse that reflects their preferences. By confronting respondents with a claim made by a fictitious unelected representative, we present a hard test for this idea and tread new ground, investigating empirically the influence unelected representatives might have on the democratic legitimacy of our political systems.
H1: Witnessing issue-congruent claims from unelected representatives increases satisfaction with democracy (compared to witnessing incongruent issue claims).
Political inclusion: power to your voice
However, it may be the case that voice outside of the electoral channel alone is insufficient to affect satisfaction with democracy because the non-electoral channel is more disconnected from the policymaking process than are electoral campaigns. Perhaps the influence of exposure to issue-congruent claims from unelected representatives on SWD is dependent on those making them being included in formal political processes. Prior research focusing on elected representation emphasizes that if citizens don’t feel that their preferences are translated into policies, or are at a minimum acknowledged by elected politicians, they perceive the political system as unresponsive and become less satisfied with democracy (Harteveld et al. Reference Harteveld, Kokkonen, Linde and Dahlberg2021). Hobolt, Hoerner, and Rodon Reference Hobolt, Hoerner and Rodon2021 demonstrate how the effect of intrinsic support for a candidate, understood as ideological congruence, is conditional on instrumental effect. That is, citizens’ satisfaction with democracy increases the most when they have a chance to support a candidate in an election with whom they substantively agree and who will be in a position of power to shape policies. When the candidate is considered to be powerless, the effect of substantive representation on satisfaction with democracy is not significant. In other words, it may not be sufficient to have your voice heard; that voice must be empowered.
Indeed, Flynn Reference Flynn2019 wrote ‘power without representation is blind, representations without power are empty’ (65). Wolkenstein Reference Wolkenstein2024 has further argued that we need to pay more attention to power in the constructivist turn by asking ‘to what degree can those who advance claims to represent others exercise power over others?’ and argues that power is how we really make sense of unelected representation. We have mentioned Greta Thunberg as an example of a powerful unelected representative. Her impact makes her stand out, but she is not alone. Angelina Jolie has appeared in the US Senate and British House of Lords to speak on behalf of victims of sexual violence in conflict areas (Sky News 2022), Ben Affleck has testified in Congress on behalf of the Eastern Congo (Rogers Reference Rogers2015), Seth Rogen on behalf of patients with Alzheimer’s disease (Dockterman Reference Dockterman2016), and Malala Yousafzai has spoken at the UN General Assembly to champion girls’ right to education (BBC News 2013). In contrast, Andrew Tate, championing very traditional gender roles, has been arrested (Rainsford Reference Rainsford2024). Thus, empowering unelected representatives often takes place through invitations by elected politicians to appear in formal politics arenas, amplifying their agenda-setting power. Who gets included in such a way varies greatly.
This major finding – that the combination of having a representative one agrees with and this person being in a powerful position positively affects satisfaction with democracy – forms the basis of our effective channel of representation argument. If this logic travels outside of the electoral arena, then unelected representatives making claims that citizens agree with could be particularly effective in increasing satisfaction with democracy when these representatives are included in formal political processes. Overall, inclusion in politics should increase the likelihood that preferences might be translated into policy. We also expect it to increase people’s sense of being heard and of their preferences receiving acknowledgment. Based on this link between inclusion in the political process and citizens’ experience of effective representation, we hypothesize:
H2: Witnessing issue-congruent claims from unelected representatives increases satisfaction with democracy to a greater degree when the unelected representatives are included in the political process (compared to excluded from the political process).
The moderating role of citizens’ political marginalization
One of the key factors influencing citizens’ satisfaction with democracy is whether they voted for a party currently in government. Electoral winners – those who voted for or plan to vote for a governing party – report significantly higher satisfaction with democracy than electoral losers (Anderson and Guillory Reference Anderson and Guillory1997; Ezrow and Xezonakis Reference Ezrow and Xezonakis2011; Mayne and Hakhverdian Reference Mayne and Hakhverdian2017). Citizens who do not feel represented by any party show even stronger signs of disenchantment with democracy (De Mulder Reference De Mulder2024; Nadeau, Daoust, and Dassonneville Reference Nadeau, Daoust and Dassonneville2023; Rich and Treece Reference Rich and Treece2018). This ‘winner–loser gap’ is among ‘the most robust relationships documented in political science’ (Nadeau, Daoust, and Dassonneville Reference Nadeau, Daoust and Dassonneville2023: 483). Research on elected representation suggests that when citizens feel their preferences are ignored in policymaking, they view the political system as unresponsive and become less satisfied with democracy (Harteveld et al. Reference Harteveld, Kokkonen, Linde and Dahlberg2021). Experimental studies further indicate that political inclusion of one’s party can soften the blow of election loss on satisfaction with democracy among losers (Merkley et al. Reference Merkley, Cutler, Quirk and Nyblade2019). Extending this research, we would expect that substantive representation from unelected representatives may minimize the electoral winner–loser gap:
H3a: Witnessing issue-congruent claims from unelected representatives increases satisfaction with democracy more for electoral losers compared to electoral winners.
Relatedly, external political efficacy – the belief that one has an influence on political decision-making – and political sophistication – knowledge about politics – have an effect on satisfaction with democracy (Mayne and Hakhverdian Reference Mayne and Hakhverdian2017). As explained by Morrell Reference Morrell2003, ‘efficacy is citizens’ perception of powerfulness (or powerlessness) in the political realm’ (589). Having high political efficacy thus translates to feeling powerful, and low political efficacy to feeling powerless or marginalized (Bollwerk, Schlipphak, and Back Reference Bollwerk, Schlipphak and Back2022). The more a citizen considers herself to be included in the political process, involved, in control, and influential – that is, the opposite of feeling marginalized – the more satisfied she is with the way democracy works. Again, we expect marginalized individuals, with this sense of ‘losing out’, to be more influenced by unelected representation.
H3b: Witnessing issue-congruent claims from unelected representatives increases satisfaction with democracy more among individuals with low external political efficacy compared to individuals with high external political efficacy.
Finally, the moderating effects of being an electoral loser or having low external political efficacy on the impact of issue congruence with an activist’s claim-making (H3a and H3b) might be conditional on the activist being included in politics (H2). For electoral losers who are otherwise in the minority in electoral politics, such inclusion signals that their views still have a voice in formal institutions. As cited above, prior research shows that political inclusion of one’s party can soften the blow of election loss on satisfaction with democracy losers (Merkley et al. Reference Merkley, Cutler, Quirk and Nyblade2019). We argue that this dynamic extends to unelected representation such that unelected representatives’ issue-congruent claim-making increases satisfaction with democracy more among electoral losers when they are included in the political process.
For individuals with low political efficacy, the inclusion of the activist also signals that their perspective is being heard, as these individuals are otherwise predisposed to view political authorities and formal institutions as unresponsive (Bollwerk, Schlipphak, and Back Reference Bollwerk, Schlipphak and Back2022). Thus when such individuals see issue-congruent claims from unelected representatives who are included in the political process, the perceived responsiveness of the system increases, which in turn should increase their satisfaction with democracy. Thus, we expect the positive effect of issue-congruent claim-making by unelected representatives on satisfaction with democracy to be especially pronounced among electoral losers and individuals with low external political efficacy when these representatives are included in political processes.
H4a: Witnessing issue-congruent claims from unelected representatives increases satisfaction with democracy more among electoral losers when the unelected representatives are included in political processes (compared to excluded).
H4b: Witnessing issue-congruent claims increases satisfaction with democracy more among individuals with low external political efficacy when the unelected representatives are included in political processes (compared to excluded).
To conclude, we assume that there is less desire for an additional channel of representation among highly satisfied people who already experience effective political representation, either because their party is in power (ie electoral winners) or because they possess high political efficacy and in that way are empowered to make their voices heard. For those who are politically marginalized as either electoral losers or because they personally feel that they have low political efficacy, an effective unelected channel of representation could potentially be quite meaningful and enhance satisfaction with democracy. For them, this would not come on top of already existing effective representation; it would rather be the difference between being somewhat represented and not being represented at all.
Yet the effect of subjective marginalization goes beyond the political, at least in some situations, to include social and/or economic marginalization. Fan and Zhang (Reference Fan and Zhang2022) demonstrate that digital exclusion in the form of being disconnected from social media has a negative effect on satisfaction with democracy. Daoust and Nadeau (Reference Daoust and Nadeau2021) show that there is a different logic to satisfaction with democracy in rich countries as opposed to poorer countries. In affluent countries, citizens’ satisfaction with democracy is heavily affected by whether they are electoral winners or losers. In poorer countries, economic prospects have a stronger effect. To account for such effects, we include self-perceived cultural and economic marginalization as controls in our design. The winner–loser gap varies across democracies. Institutional quality and performance, rather than institutional type, have been found to be key in reducing the support gap between electoral winners and losers in European democracies (Dahlberg and Linde Reference Dahlberg and Linde2016). We therefore also need to account for our countries of analysis.
Methods
To examine the relationship between unelected representatives and citizens’ satisfaction with democracy, we conducted a vignette experiment in Germany, Italy, Romania, and Sweden (N = 8279). The pre-analysis plan (PAP) is available online (OSF link), and any clarifications and deviations from the PAP are explained in the manuscript and in Appendix D.
All data were collected by the polling company Kantar Germany in July 2022, and respondents were incentivized to participate according to Kantar’s policies.Footnote 1 In Appendix A3, we include an elaboration of the ethical evaluation of the design and its execution. The current preregistered study was collected in the same survey as another preregistered study with a different research design and different hypotheses. The full survey, therefore, consisted of a pre-treatment questionnaire module, a conjoint module for another study, the vignette experiment, and a post-treatment questionnaire module. A detailed description of the survey flow can be found in Appendix B3.
Kantar collected 2073 responses from Germany, 2090 from Sweden, 2097 from Italy, and 2019 from Romania, giving us a total sample of 8279 respondents. Quota sampling was employed, and the sample was set to be representative in terms of age, gender, and education. To assess whether we have sufficient power to detect small effect sizes (f = .1), we performed a compromise power analysis using G*Power 3.1 to make sure our three-way interaction would be well powered in our country subset. We set the significance level to be .05 and power to be 95% with eight groups, giving us a power of 97%. All analyses reported in the manuscript should therefore have sufficient power. In accordance with our PAP, ‘don’t know’ answers for our dependent variable led to the removal of these observations, giving us a sample of 8196 for the analysis. Appendix A2 contains participant information and the sampling procedure.
Cross-national research design and samples
The difference in SWD between electoral winners and losers varies across countries (Linde and Ekman Reference Linde and Ekman2003), indicating that we need to take into account contextual country-level factors to identify possible scope conditions of witnessing representative claims on satisfaction with democracy.
We investigate our hypotheses in a cross-national survey experiment fielded to nationally representative samples of citizens in four European countries: Sweden, Germany, Italy, and Romania. We selected this set of countries to capture contextual variation within Europe, especially regarding quality of government. Prior research has found that quality of government – and effective government in particular – is an important predictor of aggregate satisfaction with democracy (Aarts and Thomassen Reference Aarts and Thomassen2008; Dahlberg and Holmberg Reference Dahlberg and Holmberg2014; Newton Reference Newton, Dalton and Klingemann2007; Vik et al. Reference Vik, de Wilde, Treib and Aarøe2025; Wagner, Schneider, and Halla Reference Wagner, Schneider and Halla2009). Quality of government is relatively low in Romania, moderate in Italy, high in Germany, and very high in Sweden (see Appendix A1 for more information regarding case selection). The countries also differ on other potential scope conditions such as media use, degree of democratic consolidation, and economic development. This country selection allows us to explore the generalizability of the predictions across these country-level scope conditions within a European context (Vik et al. Reference Vik, de Wilde, Treib and Aarøe2025). We discuss possible interpretations of such patterns as suggestions for future research in the final section of this article.
Migration has long been a controversial issue in both national and transnational politics in Sweden, Germany, Italy, and Romania, as in most European and Western countries. Immigration within the EU is one of the few political issues in which unelected representatives actively campaign on both sides in a transnational setting. For example, British actress Lena Headey has been an ardent proponent of more welcoming refugee policies in Europe, criticizing poor treatment of refugees on Greek islands, while Austrian Identitarian leader Martin Sellner has campaigned against immigration in the United Kingdom, Austria, Greece, and many other places (Headey Reference Headey2020; Zeller and Virchow Reference Zeller and Virchow2024). Hence, migration is an issue where various manipulations will appear credible to respondents.
Experimental design
We conducted a vignette experiment with a 2 × 2 between-subject factorial design. Respondents were randomly assigned to one of our four treatment conditions, which consisted of news articles that portray (1) an activist who is pro-immigration and is included in politics, (2) an activist who is pro-immigration and is excluded from politics, (3) an activist who is anti-immigration and is included in politics, and (4) an activist who is anti-immigration and is excluded from politics. This 2 × 2 design allows us to investigate the effect of substantive representation (or issue congruence) and political inclusion.
We focus on immigration as the political issue in the experiment because it is a salient issue across Western democracies, and one where unelected representatives actively campaign on both sides in a transnational setting, as noted in our discussion of the country selection. This allows us to realistically vary the message of the activist and increases mundane realism.
The news articles all introduce Kim Jones, a fictitious political activist, who is visiting the country of the respondents. She is presented as an internationally well-known activist arguing in favor of more liberal immigration policies on behalf of refugees (the pro-immigration stimulus) or more restrictive immigration policies on behalf of ‘Europeans’ (the anti-immigration stimulus). Thus, in all vignettes, she is presented as an unelected representative: an individual making a representative claim that gains public resonance. Because the celebrity status of activists like Bono or Angelina Jolie might put people off or otherwise interfere in their assessment of their claims and activism, we chose to present Kim Jones purely as an activist without prior celebrity status. To limit potential framing effects, both the pro- and anti-immigration positions have a threat-based frame. In the pro-immigration condition, the threat is that if we do not help refugees, our own cultural values will crumble. In the anti-immigration condition, the threat is that refugees threaten Europe’s culture. The position of the activist in the political process was manipulated as either excluded, that is, being refused the chance to talk with politicians during her visit, or as included, that is, being invited to talk with politicians during her visit. This is drawn from real-world examples such as how Greta Thunberg was invited to speak to the House of Commons in the United Kingdom but was refused the chance to speak at the Norwegian Parliament (NTB 2021; Thunberg Reference Thunberg2019). The stimuli are illustrated in Figure 1 and available in a larger size in Appendix B2.
Figure 1. Illustration of vignettes.
We performed a pre-test of the vignettes among a small convenience sample (N = 28), and overall, our manipulations worked as intended in the pre-test, as shown in Appendix B2. All vignettes were also found to elicit a similar emotional response in our pre-test, limiting potential confounding framing effects. The post-treatment questionnaire included manipulation checks of both the issue position of Kim Jones and her inclusion in the political process as indexed by her perceived influence. The manipulation checks consisted of two questions: ‘In the article about Kim Jones you just read, what was the goal of her activism?’ and ‘And what about her political influence? In the article you just read, did politicians listen to Kim Jones?’ While our pre-test indicated that our manipulations worked, the vignettes with an activist who was included had a higher proportion of respondents who failed the inclusion manipulation check in the survey (see Appendix B2.2). There could be several reasons why more people in this group failed the manipulation check. Notably, there is difficulty operationalizing the complex concept of political inclusion. We continue all analyses with intention-to-treat (ITT) effects to limit post-treatment bias (Montgomery, Nyhan, and Torres Reference Montgomery, Nyhan and Torres2018; Varaine Reference Varaine2023), but as preregistered we also run the analyses on only those who passed the manipulation check (see Appendix C).
Dependent measures
After the treatment vignettes, we measured our outcome variable by asking: ‘On the whole, how satisfied are you with the way democracy works in [COUNTRY]?’ (DV1), which is measured on an 11-point scale (1 = extremely dissatisfied, 11 = extremely satisfied).Footnote 2 This measure is taken from the European Values Study and is one of the standard formulations and operationalizations of SWD (Poses and Revilla Reference Poses and Revilla2022). In addition, we also measured respondents’ views on the role political activists play in democracy on the same 11-point scale by asking ‘And how satisfied are you with the role political activists play in [COUNTRY’s] politics?’ (DV2). The main reason behind including this second DV is to take into account that there might not be a link between an individual activist’s claims and involvement in politics on the one hand and citizens’ general satisfaction with democracy on the other hand. In other words, the idea that witnessing a single representative claim could affect such a general, aggregated expression of public opinion as SWD is very innovative and might be too far-fetched. Perhaps citizens do not allow their opinion about the way democracy works to be influenced by such single-shot events, much less fictional ones. Our second DV is therefore designed to capture a smaller aspect of satisfaction with democracy – the extent to which unelected activists are involved in it – where citizens are more likely to make a connection between the stimulus and the outcome variable. This furthermore addresses the call by Singh and Mayne (Reference Singh and Mayne2023) to ‘… develop alternatively worded questions that more precisely capture the literal and encompassing meanings of SWD’ (206).
Measures of issue congruence, electoral winner, and political efficacy
Our quasi-experimental factor of issue congruence is calculated by contrasting the issue position in the vignette to the respondent’s attitudes on immigration measured in the pre-treatment questionnaire. The respondent’s issue position was measured by asking them about their stance on immigration on a ten-point scale, where 1 meant that they ‘fully agreed’ with the statement on the left (ie ‘Immigration strengthens our society’) and 10 meant that they ‘fully agreed’ with the statement on the right (ie ‘Immigration is a threat to our society’). We computed issue congruence by contrasting the respondent’s issue position on immigration with Kim Jones’s position in the news article (−4 = maximum degree of incongruence, 4 = maximum degree of congruence). Like in previous research (Bakker, Schumacher, and Rooduijn Reference Bakker, Schumacher and Rooduijn2021; Vik et al. Reference Vik, de Wilde, Treib and Aarøe2025), we assume that the claim is more congruent for respondents with more extreme attitudes in the same direction as Kim Jones. A positive score on the issue congruence scale (
$0 \lt x \le 4$
) means that the issue position of the activist is congruent with that of the respondent (eg anti-immigration position and anti-immigration claim). Negative scores (
$ - 4 \ge x \gt 0$
) indicate incongruence between the respondent’s immigration position and the position of the activist (eg anti-immigration attitude and pro-immigration claim). A score of 0 indicates that the respondent has a neutral issue position, meaning that the policy demands in the claim are neither congruent nor incongruent with the respondent’s issue position. The recoding process is explained in detail in Figure A4 in Appendix B3. To ease interpretation while retaining the fine-grained information from the continuous measure, we rescale issue congruence to range from 0 (minimum congruence) to 1 (maximum congruence).
In terms of our moderators, for electoral winner/loser, we adopted the coding methodology outlined by Anderson et al. (Reference Anderson, Blais, Bowler, Donavon and Listhaug2007: 34), wherein an individual was classified as an electoral winner if they voted for a party that held government positions during the data collection period (July 2022Footnote 3 ). For a detailed recoding process, please refer to Figure A5 in Appendix B3. It is noteworthy that three out of the four countries experienced changes in government between elections. Consistent with Merkley et al. (Reference Merkley, Cutler, Quirk and Nyblade2019), respondents who indicated they did not vote or could not recall were excluded from the analysis. Recent work highlights that abstainers – citizens who reported they did not vote – can be used as a baseline, so as not to remove observations from the analysis (Nadeau, Daoust, and Dassonneville Reference Nadeau, Daoust and Dassonneville2023). We therefore also tested our hypotheses H3a and H4a with electoral abstainers as a baseline (note that this was not preregistered but added as a robustness check).
For external political efficacy, we used a short political efficacy scale (Groskurth et al. Reference Groskurth, Nießen, Rammstedt and Lechner2021), which encompasses four items. Two of these items encapsulate external political efficacy. The two items were ‘Politicians strive to keep in close touch with the people’ and ‘Politicians care about what ordinary people think’. Responses were measured on a six-point scale ranging from strongly disagree to strongly agree. Combining the two items on external political efficacy produced a reliable scale (
$\alpha $
= .79). We recoded the scale to range from 0 to 1, with higher values indicating higher external political efficacy.
Control variables
As respondents’ issue positions, external political efficacy, and vote choice are observed and not manipulated experimentally, we include control variables in all main analyses to increase internal validity. We specified controls a priori as gender, age, and education. We also specified that we would run analyses with the additional controls on self-perceived cultural marginalization (SPCM) and economic marginalization (SPEM) (Bollwerk, Schlipphak, and Back Reference Bollwerk, Schlipphak and Back2022). Cultural marginalization was measured with the items ‘The values of people like me are not respected enough in society’ and ‘The way of life of people like me is not appreciated enough in society’ (
$\alpha $
= .74). Economic marginalization was measured with the items ‘The work of people like me is not recognized enough in society’ and ‘No matter how hard we work, people like me are not appreciated’ (
$\alpha $
= .78). To simplify our analytical approach, and as we found no multicollinearity, we include all of the above-mentioned controls in the analysis reported in the main manuscript.
Results
To test our hypotheses, as preregistered, we performed ordinary least squares (OLS) regression but with country-level fixed effectsFootnote 4 to investigate how unelected representation influences citizens’ satisfaction with democracy. As preregistered, we also conducted supplementary analyses examining our second dependent variable (DV2), ‘How satisfied are you with the role political activists play in [COUNTRY’s] politics?’ To ease the interpretation of results, we rescaled all our variables so that 0 = minimum value and 1 = maximum value.Footnote 5
We ran several robustness checks. First, we ran all models without control variables and subsequently without fixed effects. In addition, we ran robust regression with M-estimation, which is a type of robust estimation that can handle heavy-tailed distributions and outliers.Footnote 6 As preregistered, we also perform all analyses exclusively on participants who passed the manipulation check. Additionally, we preregistered our intention to qualitatively assess differences among countries. Consequently, we have provided subgroup analyses for each country. All results are illustrated with regression tables in Appendix C. Any disparities between these supplementary analyses and the main results have been transparently reported in this main manuscript.
H1: Does witnessing issue-congruent claims from unelected representatives increase SWD?
To test H1, we regress SWD on issue congruence along with controls. The results are presented in Figure 2 (see Table A2 in Appendix C1 for full regression output and robustness checks). Contrary to our expectations, the results show that issue congruence does not have a significant effect on SWD (
$\beta $
= .002, SE = .01, p
$ \lt $
.862). In other words, simply witnessing an issue-congruent claim from an unelected representative does not increase SWD. This contrasts with a key finding of Merkley et al. (Reference Merkley, Cutler, Quirk and Nyblade2019), who showed that having one’s voice heard was enough to increase SWD. This null finding is relatively consistent in all additional analyses on SWD. Thus, H1 is not supported.
Figure 2. Testing H1: Coefficient plot of issue congruence (substantive representation) and control variables on SWD and SWD2. Country fixed effects. 95% confidence intervals. N
$ = $
8196.
As presented in Table A3 in Appendix C1, we do note, however, that issue congruence leads to higher values of our second dependent variable, ‘satisfaction with the role activists play in politics’ (
$\beta $
= .065, SE = .010, p
$ \lt $
.001). This finding suggests that having one’s voice heard through representative claims made by unelected representatives in the public sphere has some positive effects on how citizens assess their country’s politics, but not enough to sway their opinion about democracy more generally.
H2: Does inclusion in the political process increase the impact of issue-congruent claims from unelected representatives on SWD?
To test H2, we regress SWD on issue congruence, inclusion, and their interaction term along with control variables (see Table A4 in Appendix C2). Figure 3 illustrates the results and also illustrates the marginal effects for the inclusion and exclusion conditions separately. Consistent with H2, the results show that witnessing an activist making an issue-congruent claim increases SWD more when the unelected representative is included in the political process (compared to excluded) (
$\beta $
= .042, SE = .021, p
$ = $
.045). Specifically, when the activist expressed an issue-congruent claim and was described as meeting with national politicians (ie included in the political process), satisfaction with democracy (SWD) was, on average, 4.2% higher than when the activist was excluded. Although the positive effect is consistent in our robustness checks, it is not statistically significant in our preregistered OLS model without fixed effects (
$\beta $
= .038, SE = .021, p
$ = $
.07), nor in our fixed effect models without control variables (
$\beta $
= .038, SE = .022, p
$ = $
.07). As shown in Figure 3, the marginal effects of issue congruence on satisfaction with democracy (SWD) are not statistically significant for either condition. For the political inclusion condition, the effect is positive but non-significant (
$\beta $
= .023, p = .122), while for the political exclusion condition, the effect is negative and also non-significant (
$\beta $
= –.019, p = .197).
Figure 3. Testing H2: Predicted SWD by issue congruence and the activist’s political inclusion. Country fixed effects. 95% confidence intervals. N
$ = $
8196.
Interestingly, for respondents who passed the manipulation check, the effect is substantially larger (
$\beta $
= .077, SE = .027, p
$ \lt $
.01). In other words, respondents who correctly identified that the activist was included in politics showed a considerable increase in SWD. However, as the manipulation check was necessarily measured after the treatment, internal validity is lower in this subgroup analysis (Aronow, Baron, and Pinson Reference Aronow, Baron and Pinson2019; Montgomery, Nyhan, and Torres Reference Montgomery, Nyhan and Torres2018; Varaine Reference Varaine2023). This highlights important aspects of how citizens understand political inclusion, which we return to in the discussion.
At the individual country level, the interaction effect is also consistent in terms of direction, but not significant in Sweden, Germany, and Italy (see Appendix C2). We do not find the effect in Romania, where the analysis suggests that activists’ inclusion results in higher levels of SWD regardless of the level of congruence. This suggests that there might be country-level variance in effective channels of unelected representation and how political inclusion and substantive representation are weighed in citizens’ democratic attitudes. In terms of the second dependent variable, we do not find evidence for this interaction effect and find only a weak and insignificant effect in the other direction (
$\beta $
= −.012, SE = .020, p
$ = $
.51).
In sum, we find some support for H2 with our primary dependent measure in the main models – the fixed effects models with control variables. Yet this result is not significant in the various robustness checks, including the preregistered OLS model and the fixed effect models without control variables, and in none of the models with the second dependent measure. Thus, our evidence does not allow us to firmly conclude that citizens’ satisfaction with democracy increases when unelected representatives they agree with are included in politics, nor does it allow us to confidently argue that there is no such effect.
H3: Is there an interaction effect of political marginalization and witnessing issue-congruent claims on SWD?
H3 hypothesized that when an activist makes a congruent claim and the respondent is (a) an electoral loser and (b) has low external political efficacy, SWD should be higher. To test H3a, we regress SWD on issue congruence and being an electoral winner/loser, and their interaction term along with control variables on SWD. For H3b we do the same but replace the electoral winner/loser variable with external political efficacy. The results are illustrated in Figure 4, in which we also illustrate the marginal effects for the slope of the electoral winner and loser separately. All analyses with regression output are available in Appendix C3 and C4.
Figure 4. Testing H3: Predicted SWD by issue congruence moderated by (a) electoral winners and losers (N = 6302) and (b) individuals with high, medium, and low external political efficacy (N = 8196). Country fixed effects. 95% confidence intervals.
In support of H3a, we find a statistically significant moderating effect of being an electoral loser (compared to being an electoral winner) on the impact of unelected representatives’ issue-congruent claims on SWD (
$\beta $
= −.066, SE = .023, p
$ \lt $
.01). This finding is confirmed when including electoral abstainers as a baseline (
$\beta $
= −.030, SE = .012, p
$ \lt $
.05). Importantly, when looking at the marginal effects, the effect of issue congruence leads to significantly higher levels of SWD for the subgroup of electoral losers (
$\beta $
= .043, p
$ = $
.01). This finding mirrors that of Merkley et al. (Reference Merkley, Cutler, Quirk and Nyblade2019), namely that having one’s voice heard (or in our case being substantively represented) can minimize the SWD gap between electoral winners and losers. This effect is confirmed in our robustness checks and when we regress our second dependent variable in our model (
$\beta $
= −.085, SE = .022, p
$ \lt $
.001). At the individual country level, all countries demonstrate an effect in the same direction, but the effects are not significant. In conclusion, H3a is supported: Substantive representation from unelected representatives may soften the blow for electoral losers.
Contrary to H3a (see Table A10, Appendix C4), we do not find a statistically significant moderating effect of external political efficacy on the impact of unelected representatives’ issue-congruent claims on SWD (H3b), although the tendency is again in the expected direction (
$\beta $
= −.039, SE = .036, p
$ = $
.28). Additional analyses show that these results are similar across robustness checks. For the country-level subgroup analyses, the predicted moderating effect of political efficacy is only significantly present in Germany (
$\beta $
= .163, SE = .072, p
$ \lt $
.05). We see a similar pattern for our second dependent variable. In conclusion, there is not sufficient evidence in support of H3b, which posits that political efficacy is a key driver for SWD and that substantive representation from unelected representatives can help mitigate this, with some exceptions showing in Germany.
In sum, being an electoral winner and having high external political efficacy are strong predictors for SWD, which is well established in the literature. Witnessing congruent claims made by unelected activists can reduce the SWD gap between electoral losers and winners.
H4: Is there a three-way interaction of issue congruence, representative inclusion, and political marginalization on SWD?
According to H4a, unelected representatives’ issue-congruent claims should increase SWD more among electoral losers when the unelected representatives are included in the political process. To test this prediction, we add the electoral winner/loser variable to the interaction of issue congruence and political inclusion postulated in H2. The results are illustrated in Figure 5, Panel a, and all analyses with regression output are available in Table A12, Appendix C5.
Figure 5. Testing H4: predicted SWD by issue congruence and by inclusion of the activist moderated by (a) electoral winners and losers (N = 6302) and (b) individuals with high, medium, and low efficacy (N = 8196). Country fixed effects. 95% confidence intervals.
For H4a, we do not find a significant three-way interaction. The inconclusive interaction between congruence and political inclusion (H2) seems to be present for both electoral winners and losers (
$\beta $
= −.018, SE = .046, p
$ = $
.99), and this holds when respondents are asked to assess the role activists play in their country (SWD2) too (
$\beta $
= .073, SE = .045, p
$ = $
.69). To the extent that witnessing the political inclusion of activists with whom one agrees increases SWD, it does so irrespective of whether respondents are electoral winners or losers. The exception is in Romania, where this three-way interaction is significant for SWD (
$\beta $
= −.311, SE = .131, p
$ \lt $
.05). Otherwise, the null finding is confirmed in our robustness checks, subgroup analyses, and for our second dependent variable. We conclude that our analysis does not support H4a.
While the three-way interaction in H4a is not statistically significant, the marginal effect among electoral losers who witnessed a congruent claim by an activist that was politically included is strong and significant (
$\beta $
= .079, p
$ \lt $
.01), while the marginal effect is almost zero and non-significant for the excluded activist (
$\beta $
= .002, p
$ \lt $
.701). This provides some support for the prediction that the effect observed in H3a may be more pronounced among those electoral losers when the activist is included in the political process. However, this pattern should be interpreted with caution given the non-significance of the overall three-way interaction.
According to H4b, unelected representatives’ issue-congruent claims increase SWD more among individuals with low political efficacy when the unelected representatives are included in the political process. To investigate this hypothesis, we test the full three-way interaction between issue congruence, external efficacy, and representative’s inclusion by expanding the model for the test of H2. The results are illustrated in Figure 5, Panel b, and the regression output is available in Appendix C6.
The results show that the activist’s inclusion/exclusion in the political process does not significantly influence the effect of political efficacy on the impact of unelected representatives’ issue-congruent claims on SWD (
$\beta $
= .021, SE = .072, p
$ = $
.99). This is consistent across countries and regardless of whether respondents passed the manipulation check, and also for the second dependent variable, making us confident in this null finding. Thus, the results do not support H4b.
Discussion
In this paper, we investigate whether activists who behave as unelected representatives by voicing political demands on behalf of various constituencies affect citizens’ satisfaction with democracy. We propose that activists may serve as effective channels of representation if they substantively represent individuals. Furthermore, we argue that this effect of substantive representation is stronger when unelected representatives are included in the political process and among individuals who feel underrepresented or marginalized, either because they are electoral ‘losers’ or through low political efficacy. To test this, we conducted a preregistered vignette experiment in Sweden, Germany, Italy, and Romania (N = 8196). Our study is the first to empirically link research on unelected representation to studies of political system support.
The results are mixed. Given that our treatment is rather weak – a single claim made by a fictitious activist is unlikely to have major effects on citizens’ satisfaction with democracy – it is somewhat remarkable that we find any effects. Overall, we find that substantive representation from unelected representatives does not significantly increase satisfaction with democracy on average across the whole electorate. However, there are important conditional effects. Our strongest result is that electoral losers report less dissatisfaction with democracy when represented by unelected activists who share their views, regardless of these activists’ political inclusion. This is an important finding because it suggests that substantive representation from unelected representatives narrows the ‘winner–loser gap’ in satisfaction with democracy, a relationship that has been described as ‘one of the most robust in political science’ (Nadeau, Daoust, and Dassonneville Reference Nadeau, Daoust and Dassonneville2023: 483). The key implication is that citizens look beyond their elected representatives when assessing their satisfaction with democracy, especially when they lack representation in public office. Unelected representatives could provide a supplementary pathway to maintaining support for the democratic system, especially among electoral losers. This link is important as studies show a crisis in democratic support, with declining trust in representative institutions (Valgarðsson et al. Reference Valgarðsson, Jennings, Stoker, Bunting, Devine, McKay and Klassen2025) and widespread perceptions that ‘elected officials are out of touch’ (Pew Research Center 2024: 15). Against this backdrop, it is crucial to investigate whether unelected representation can affect citizens’ attitudes toward the political system. An important question for future research to investigate is whether the effects of issue-congruent claims from unelected representatives are stronger among some segments of electoral losers, while other subgroups of electoral losers might be less susceptible to representational compensation by activists.
We find mixed support for the prediction that inclusion in the political process increases the effect of unelected representatives’ issue-congruent claims. These mixed results raise some important questions about the relationship between inclusion and political power and how it affects satisfaction with democracy. As highlighted by Wolkenstein (Reference Wolkenstein2024) and other constructivist scholars, we need to ask questions about power. An important question for future research is whether inclusion alone increases the impact of activists’ claim-making on satisfaction with democracy or whether direct influence on policy decisions and/or an overall policy direction aligned with the advocacy of the activist is also required.
Furthermore, our results provide hints that political context conditions play a role in understanding unelected representation and satisfaction with democracy. We find that unelected representation can narrow the electoral winner–loser gap in the presence of substantive representation, and this effect is strongest in Romania, followed by Italy, then Germany, and finally Sweden, with only a small effect size. In the logic of our rank-ordered case selection based on varying degrees of quality of government, it may thus be the case that the impact of unelected representatives is strongest in countries with low government quality (Aarts and Thomassen Reference Aarts and Thomassen2008; Dahlberg and Holmberg Reference Dahlberg and Holmberg2014; Newton Reference Newton, Dalton and Klingemann2007). However, these results could also be explained by other contextual factors that co-vary with the quality of government. In particular, economic prosperity, which has also been shown to influence satisfaction with democracy (Daoust and Nadeau Reference Daoust and Nadeau2021), might account for these contextual differences. Further research should investigate the contextual scope conditions of our documented effects more closely, taking into account more countries, including from outside the European context.
Future studies could investigate how witnessing claims by influential activists with whom one agrees affects satisfaction with democracy in a global setting. What are the effects of involving activists in international politics as opposed to national politics – for example, by inviting them to the European Parliament or the UN General Assembly rather than the House of Commons? Does reading about a climate activist (eg Greta Thunberg) speaking at the British House of Commons, or in the UN General Assembly, affect people’s satisfaction with democracy more in places particularly affected by climate change, such as Bangladesh or Tuvalu? Taking this study from our European focus to a global level is an important avenue for future research. Studies focusing on a more domestic or less transnational setting would also be valuable, as the domestic setting might increase the strength of the effect. Citizens might perceive representation to be more directly relevant in domestic settings.
While our study represents the first cross-national investigation of the impact of unelected representation on SWD, the research design has some limitations. First, while satisfaction with democracy is the most logical indicator of political support for purposes of investigating the possible impact of claims by activists on the legitimacy of democracy, it is – like most measurements in surveys – imperfect. We have taken into account previous studies on what the most valid and reliable formulation of the question is (Poses and Revilla Reference Poses and Revilla2022) and in which countries it tends to be most reliable as a measurement (Valgarðsson and Devine Reference Valgarðsson and Devine2022) in informing our research design. We believe that the standard formulation of SWD makes sense given that ours is the first study to investigate whether there may be an effect of witnessing representative claims by activists on satisfaction with democracy. But we agree with arguments by Singh and Mayne (Reference Singh and Mayne2023) that more studies are needed using alternative formulations of the key SWD variable. We have included one such alternative formulation, specifically geared to our study and the subcomponent of democracy it entails: the involvement of activists in political processes. We expected that the likelihood of our stimuli having an effect on the former would be lower than on the latter, given that the latter relates more directly to the stimuli provided. But surprisingly, we found an effect of witnessing claims by unelected representatives on electoral losers’ satisfaction with democracy at large, not just on their satisfaction with the involvement of activists in politics.
Second, we used an experimental setup with a fictitious activist and a one-shot treatment. We used a fictitious activist for ethical reasons as it allows us to vary the message of the activist while keeping the sender constant, without assigning a fake message to a real-world activist. Still, it could be harder for people to relate to fictitious unelected representatives compared to well-known activists from real-world contexts. It is an interesting question for future research whether the celebrity status of the unelected representative shapes the impact of their representative claim-making. More broadly, what type of activist can shape citizens’ satisfaction with democracy via their claim-making? The effect of representative claim-making from well-known activists could have stronger effects, just as the effect could be stronger with repeated exposure. In short, the use of a fictitious activist and a one-shot exposure represents relatively weak treatment materials. In light of these considerations, it is interesting that we do find some effects, however small, of an activist’s representative claim-making on SWD.
Third, our operationalization of the activist’s inclusion/exclusion in the political process also has some limitations. Overall, the treatment material did not provide information about any policy decision. However, the description in the inclusion condition noted that ‘it seems incredibly likely that the goals promoted by her movement will be taken up by initiatives in parliament’, whereas the exclusion condition stated that ‘no one with political power is listening to us’. These cues signal the inclusion/exclusion of the unelected representative, but they could also cause respondents to infer that politics is pursuing the same/opposite course as the one advocated by the unelected representative. This is a major limitation of the operationalization because it makes it difficult to fully disentangle the impact of mere inclusion/exclusion relative to indirect cues of influence and cues about policy direction. An important challenge for further research is to develop experimental designs and treatments that can clearly isolate the effect of mere procedural inclusion compared to policy influence and the overall policy direction taken by politics.
Finally, our choice of the immigration issue in the research design also merits further discussion. As discussed in the methods section, we chose this issue because it is salient across Western democracies and because unelected representatives actively campaign on both sides in a transnational setting. This allows us to realistically vary the message of the activist and increases mundane realism. However, the focus on a single issue limits external validity. A question for future research to investigate is whether different characteristics of an issue shape the link between unelected representation and satisfaction with democracy.
As a final reflection, our findings raise important normative questions about democratic legitimacy. What are the democratic implications of involving activists and celebrities in policymaking? Where and how could they be involved in ways that complement rather than jeopardize democratic decision-making and accountability? What role should journalists play in covering the actions of various activists in the news? These questions remain for future research.
Our study makes the empirical point that there is reason to believe that activists can constitute effective channels of representation that affect people’s satisfaction with the way democracy works. If we want to understand citizens’ satisfaction with democracy and identify possible ways to increase it, we should not restrict our focus to the electoral channel of representation.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S147567652610098X
Data availability statement
The data underlying this study are openly available at OSF link.
Acknowledgements
This research was conducted as part of the research project ‘Unelected Representatives: The Impact on Liberal Democracy in Europe’, conducted at the Norwegian University of Science and Technology (NTNU) in Trondheim and funded by the Norwegian Research Council under grant nr. 287861. Earlier versions were presented at the Elections, Values and Political Communication (EVPOC) research group seminar at NTNU, the summer school ‘Electoral Democracy in Danger?’ at Sciences Po, and the Research on European Politics and Society (REPS) seminar at the University of Groningen. We thank all current and former colleagues for their great feedback and comments. Remaining errors are our own.
Funding statement
This research was funded by the Norwegian Research Council under grant nr. 287861.
Open access funding provided by University of Groningen.
Competing interests
The authors report no competing interests.
Ethical standards
Detailed elaboration of the ethical evaluation of the research design and its execution can be found in Appendix A3.
Open access
