Introduction
In recent years, democratic innovations, particularly deliberative mini-publics (DMPs), have received growing scholarly attention. DMPs are defined as ‘carefully designed forums where a representative subset of the wider population (is selected and) come together to engage in open, inclusive, informed and consequential discussions on one or more issues’ (Curato et al. Reference Curato2021, 3). Since 2000, the POLITICIZE project has identified 159 deliberative citizens' assemblies at the national and regional levels in Europe, most of them after 2015 (Paulis et al. Reference Paulis2020).Footnote 1 In a report released in 2020, the OECD referred to the growing interest in these institutions in Europe as a ‘deliberative wave’ (OCDE 2020). This abundance of experience has often been interpreted through the prism of ‘the deliberative turn’ by political theorists (Goodin Reference Goodin2008).
In parallel to their growing use, scholars have focused on evaluating the level of public support for DMPs in shaping policy decisions. The main conclusion of these studies is that citizens who are politically dissatisfied and/or are unhappy with representative politics are those most likely to support a greater role in these democratic innovations (Bedock and Pilet Reference Bedock and Pilet2020, Reference Bedock and Pilet2021; Goldberg and Bächtiger Reference Goldberg and Bächtiger2022; Jacquet, Niessen, and Reuchamps Reference Jacquet, Niessen and Reuchamps2022; Walsh and Elkink Reference Walsh and Elkink2021). However, the sources of political dissatisfaction that trigger this greater support for mini-publics are less clear.
In this article, we claim that this dissatisfaction is rooted in a position of losers of representative democracy, which we could describe as being in a situation of ‘persistent minority’ within the democratic system (Abizadeh Reference Abizadeh2021; Christiano Reference Christiano1994). It is well established that being on the losers' side of representative democracy is associated with a lower degree of political satisfaction (Anderson et al. Reference Anderson2005) and leads to being more open to various forms of institutional change that challenge the purely representative logic (Bowler and Donovan Reference Bowler and Donovan2019; Bowler, Donovan, and Karp Reference Bowler, Donovan and Karp2007; Ceka and Magalhaes Reference Ceka and Magalhães2020; Smith, Tolbert, and Keller Reference Smith, Tolbert and Keller2010). We extend these arguments to examine support for DMPs as a key to understanding the link between political dissatisfaction and backing this type of democratic innovation. In this study, we explore various situations in which citizens may feel as though they are on the losing side in representative democracy, who would be inclined to support a greater role for mini-publics. We argue that it is the position of being a ‘loser of representative democracy’ which explains the recurring findings that citizens who are more politically dissatisfied are most favourable to mini-publics (Goldberg and Bächtiger Reference Goldberg and Bächtiger2022).
Specifically, we investigate three different situations in which a citizen might feel that they are on the losing side of representative democracy. Regarding support for mini-publics, we first look at the effect of voting for a party that is not in government. We then build on a classical conceptualization of political representation (Pitkin Reference Pitkin1967) and look at its descriptive and substantive facets. We examine descriptive representation by testing whether citizens belonging to socio-demographic groups that are under-represented in parliament are more supportive of mini-publics. Then, we look at substantive representation by examining whether citizens whose political views are less accurately represented in parliament (and in government) are more likely to be supportive of mini-publics. Our study tests the three situations separately to examine how they are linked to support for mini-publics.
Building on survey data from over 15,000 citizens across fifteen Western European countries, our findings confirm a link between being a ‘loser’ of representative democracy and supporting the use of DMPs to replace elected politicians. Citizens who are descriptively and substantively underrepresented and/or citizens who vote for opposition parties are more favourable to the idea of moving away from the classical representative model of democracy and are thus more likely to support DMPs. The three ‘losing’ situations we examine influence support for DMPs, although the magnitude of effects varies to some extent. Poor substantive representation, in particular, seems to drive citizens to support giving mini-publics a greater role in replacing elected politicians.
This article contributes to several fields of political science. It first engages, logically, with the growing community of scholars studying this form of democratic innovation, but its scope extends well beyond that. It also connects to the broader literature on various reforms that have expanded the scope of citizen participation in policy making. As explained by Manin (Reference Manin1997), the history of representative democracy involves the slow, difficult, and lengthy inclusion of the masses by political elites who are often reluctant to relinquish part of their power to ordinary citizens. In this respect, our study can relate to earlier works on support and opposition to electoral reforms (Bowler and Donovan Reference Bowler and Donovan2007; Przeworski Reference Przeworski2009) and support for instruments of direct democracy (Bowler and Donovan Reference Bowler and Donovan2019; Schuck and de Vreese Reference Schuck and de Vreese2015). The study also contributes to broader debates on process preferences (Hibbing et al. Reference Hibbing2023; Pilet et al. Reference Pilet2023). Authors have examined preferences for greater citizen involvement and other political actors (experts, strong leaders, political outsiders such as business people, or religious leaders) within this research field. One of this literature's key findings is that support for alternatives to representative democracy is rooted in political dissatisfaction (Bertsou and Caramani Reference Bertsou and Caramani2022; Bertsou and Pastorella Reference Bertsou and Pastorella2017; Gherghina and Geissel Reference Gherghina and Geissel2020; Webb Reference Webb2013). By examining the mechanisms linking political dissatisfaction and support for mini-publics, this article could pave the way for similar approaches to study other models of government. Finally, this article contributes to the broader debate on the consequences of inequalities in representation and on losers' consent (Anderson et al. Reference Anderson2005; Hansen, Klemmensen, and Serritzlew Reference Hansen, Klemmensen and Serritzlew2019; Schäfer Reference Schäfer2012; Singh, Karakoç, and Blais Reference Singh, Karakoç and Blais2012). This literature has mostly focused on how this might affect citizens' attitudes toward policy making within the representative model and support for representative actors and institutions. But, with this article, we shift the focus by considering how it could affect support for alternatives to this model of government.
Theory and Hypotheses: Support for Mini-Publics and Losers of Representative Democracy
Political Dissatisfaction and Support for Deliberative Mini-Publics
A new strand of literature on public support for mini-publics has proliferated in recent years, enriching the existing works on support for various process preferences such as referenda (Bowler and Donovan Reference Bowler and Donovan2019; Schuck and de Vreese Reference Schuck and de Vreese2015; Smith, Tolbert, and Keller Reference Smith, Tolbert and Keller2010; Werner Reference Werner2020; Werner, Marien, and Felicetti Reference Werner, Marien and Felicetti2020), electoral reforms (Bowler and Donovan Reference Bowler and Donovan2007), or technocracy (Beiser-McGrath et al. Reference Beiser-McGrath2022; Bertsou Reference Bertsou2022; Bertsou and Caramani Reference Bertsou and Caramani2022; Bertsou and Pastorella Reference Bertsou and Pastorella2017). Studies have examined what kind of mini-publics (consultative, binding, with few or many participants …) were more popular among citizens (Bedock and Pilet Reference Bedock and Pilet2020; Christensen Reference Christensen2020; Goldberg Reference Goldberg2021; Goldberg and Bächtiger Reference Goldberg and Bächtiger2022; Pow Reference Pow2021; Rojon, Rijken, and Klandermans Reference Rojon, Rijken and Klandermans2019), as well as how mini-publics affect the perceived legitimacy of policy decisions (Boulianne Reference Boulianne2018; Christensen Reference Christensen2020; Jacobs and Kaufmann Reference Jacobs and Kaufmann2021). This article dialogues with another strand of research that assesses which citizens are more supportive of DMPs.
Authors have identified several factors that influence citizens' support for mini-publics. Some have stressed the role of political attitudes such as political efficacy, political interest, social trust, or left-right self-placement (Christensen and von Schoultz Reference Christensen and von Schoultz2019; García-Espín and Ganuza Reference García-Espín and Ganuza2017; Gherghina and Geissel Reference Gherghina and Geissel2020; Jacquet, Niessen, and Reuchamps Reference Jacquet, Niessen and Reuchamps2022; Rojon and Pilet Reference Rojon and Pilet2021). Others underline the effect of socio-demographic factors such as age, gender, and education (Coffé and Michels Reference Coffé and Michels2014; del Río, Navarro, and Font Reference del Río, Navarro and Font2016; Már and Gastil Reference Már and Gastil2023; Vandamme et al. Reference Vandamme2018). These explanations also apply to support for models of government that broadly give citizens a greater role (Bengtsson and Christensen Reference Bengtsson and Christensen2016; Font, Wojcieszak, and Navarro Reference Font, Wojcieszak and Navarro2015; Gherghina and Geissel Reference Gherghina and Geissel2020; Webb Reference Webb2013) or for other instruments of citizen participation such as referendums (Bowler and Donovan Reference Bowler and Donovan2019; Bowler, Donovan, and Karp Reference Bowler, Donovan and Karp2007; Schuck and de Vreese Reference Schuck and de Vreese2015). However, the influence of those factors on support for DMPs and greater citizen participation has not been systematically confirmed. Their robustness varies across studies and countries.
The single factor for which results are most consistent is political dissatisfaction as support for DMPs is greater among politically dissatisfied citizens. This correlation has been observed in a variety of studies. Bowler, Donovan, and Karp (Reference Bowler, Donovan and Karp2007) talk about ‘enraged citizens’ to refer to the association between dissatisfaction and support for reforms that give citizens a greater role in policy making (see also Bengtsson and Mattila Reference Bengtsson and Mattila2009; Gherghina and Geissel Reference Gherghina and Geissel2020; Jacquet, Niessen, and Reuchamps Reference Jacquet, Niessen and Reuchamps2022; Webb Reference Webb2013). Political dissatisfaction is also related to support for DMPs. Pilet et al. (Reference Pilet2022) recently established that dissatisfaction with the ruling political elites is the strongest driver of support for DMPs across fifteen European democracies. Similarly, Goldberg and Bächtiger (Reference Goldberg and Bächtiger2022) underlined that support for DMPs is not widespread among German citizens, especially when giving such assemblies more than a consultative role. Yet, support becomes more pronounced when they look specifically at the attitudes of politically dissatisfied citizens.
Losers’ Consent and Support for Deliberative Mini-Publics
Even though the evidence about the impact of political dissatisfaction on support for DMPs is significant, the roots of political dissatisfaction that lead citizens to believe that citizen assemblies could improve the situation remain unclear. This study analyses the sources of political dissatisfaction that may trigger support for DMPs. It tries to understand why a low degree of satisfaction with democracy is associated with support for DMPs.
In particular, we believe, like Kriesi (Reference Kriesi2020, 246), that democratic dissatisfaction is rooted in deficits of political representation and that a key driver of political dissatisfaction is being a loser of representative democracy. This idea also relates to the concept of a ‘permanent (or persistent) minority’ as coined by political theorists (Abizadeh Reference Abizadeh2021; Christiano Reference Christiano1994): in a democracy, there are individuals who never can influence political decisions due to their social and demographic characteristics, ideological preferences, or political choices. We develop a similar argument and examine whether citizens in a position of permanent minority are indeed more likely to be dissatisfied with representative policies and supportive of DMPs. Interestingly, political theory has also shown that a lottery could be a solution to the issue of persistent minority (Saunders Reference Saunders2010). We rely on three dimensions of political representation: political preferences (through voting choice), descriptive representation, and policy congruence.
Our first hypothesis builds on the dimension of voting choice and government composition and the literature on loser's consent. Research has shown that, especially in western democracies, those who lost the election are more politically dissatisfied (among many other studies, see Anderson et al. Reference Anderson2005; Blais and Gélineau Reference Blais and Gélineau2007; Farrer and Zingher Reference Farrer and Zingher2019). In particular, voting for a party that does not enter government significantly affects political dissatisfaction (Singh, Karakoç, and Blais Reference Singh, Karakoç and Blais2012). As Anderson and colleagues explain, ‘losing seems to be one of the first steps in the direction of (institutional) change and reform’ (Anderson et al. Reference Anderson2005, 181). Losers of elections are more prone to support change in various institutions. For example, supporters of parties most often in the opposition are more likely to support electoral reforms (Vowles et al. Reference Vowles and Vowles2002; Wenzel, Bowler, and Lanoue Reference Wenzel, Bowler and Lanoue2000). Other authors have demonstrated that election losers are more inclined to support reforms that increase citizens' direct participation in policy decisions (Bowler and Donovan Reference Bowler and Donovan2019; Bowler, Donovan, and Karp Reference Bowler, Donovan and Karp2007; Smith, Tolbert, and Keller Reference Smith, Tolbert and Keller2010). Consequently, we expect that the same holds true for DMPs.
H1a: Voters of parties in the opposition are more likely to support giving deliberative mini-publics a greater role.
However, a distinction should be made between voters of parties who are in opposition but might have been in power before and supporters of parties who remain permanently out of power. The latter may feel more alienated by representative democracy than the former (Bedock and Panel Reference Bedock and Panel2017), who could hope to return to government in the near future. Therefore, we expect that supporters of parties that never come to power are more likely to support a radical change involving the replacement of elected politicians by a mini-public.
H1b: Voters of parties that are constantly in opposition are more likely to support giving deliberative mini-publics a greater role.
Descriptive Representation and Support for Deliberative Mini-Publics
The notion of ‘losers' of representative democracy should not be limited to having voted for a party that remains in opposition. It also relates to citizens' ability to be well-represented within representative institutions. Here, we build upon Pitkin's work on the concept of representation and examine two facets of representation: descriptive and substantive (Pitkin Reference Pitkin1967). Descriptive representation refers to the politics of presence and the idea that citizens may feel better represented by elected politicians with similar sociodemographic characteristics. From this perspective, the assembly should mirror society (Pitkin Reference Pitkin1967, 61). However, in reality, this is rarely the case. Elected politicians are predominantly male, well-educated, economically well-off, and from the majority ethnic group. Citizens with different profiles – women, those with a low level of formal education, economically disadvantaged individuals, and ethnic minorities – are often not descriptively well-represented in most parliaments (Bovens and Wille Reference Bovens and Wille2017; Giger, Rosset, and Bernauer Reference Giger, Rosset and Bernauer2012; Wängnerud Reference Wängnerud2009). This lack of descriptive representation affects their satisfaction with representative democracy (Arnesen and Peters Reference Arnesen and Peters2018).
Inadequate descriptive representation could be another facet of being a ‘loser’ of representative democracy, potentially leading to greater support for DMPs. Since citizen assemblies are generally composed through sortition to reflect the general population, the hypothesis regarding descriptive representation and support for DMPs appears plausible (Curato et al. Reference Curato2021). A few studies have recently begun to explore this question and seem to indicate that citizens from descriptively underrepresented groups support democratic innovations (Gherghina, Mokre, and Miscoiu Reference Gherghina, Mokre and Miscoiu2021; Talukder and Pilet Reference Talukder and Pilet2021; van der Does and Kantorowicz Reference van der Does and Kantorowicz2022). However, these findings are based on single-country studies, while the presence of women, citizens with lower levels of formal education, or economically worse-off citizens may differ across countries. Our data, derived from a survey across fifteen countries, allows us to test the impact of gender, education, and (subjective) income on support for DMPs. However, some recent studies have critically examined the characterization of underrepresented groups based solely on single aspects such as gender, income, or education. They argue for the importance of acknowledging the fluid nature of political identities (Martínez-Palacios Reference Martínez-Palacios2017; Wojciechowska Reference Wojciechowska2019) and emphasize the need to consider intersectional theories when addressing political representation (Severs, Celis, and Erzeel Reference Severs, Celis and Erzeel2016).
By focusing exclusively on these three characteristics, the concerns of citizens from smaller groups (for example, natives or unregistered individuals) who may not feel represented are often neglected. Although citizens from highly disempowered and smaller groups might not feel entirely descriptively represented within small-scale DMPs due to limited representation opportunities, they may still feel partly represented by selected citizens who share at least one characteristic (for example, gender, income, or formal education).
Examining whether citizens of these smaller groups feel particularly misrepresented would be valuable. However, conducting such research would require highly specific data and many respondents to draw meaningful conclusions.
H2: Women are more likely to support giving deliberative mini-publics a greater role.
H3: Citizens with a lower level of formal education are more likely to support giving deliberative mini-publics a greater role.
H4: Citizens with lower incomes are more likely to support giving deliberative mini-publics a greater role.
Substantive Representation and Support for Deliberative Mini-Publics
Another facet of representation underlined by Pitkin is substantive representation. This refers to the idea that citizens choose elected politicians who do not necessarily share their sociodemographic characteristics but who share their ideas and policy preferences and would, therefore, support their interests in parliament (Pitkin Reference Pitkin1967, 142). However, research has again shown that not all citizens are equally substantively well-represented (Ellis Reference Ellis2012; Giger, Rosset, and Bernauer Reference Giger, Rosset and Bernauer2012; Griffin and Newman Reference Griffin and Newman2007; Griffin, Newman, and Wolbrecht Reference Griffin, Newman and Wolbrecht2012; Reher Reference Reher2018). Indeed, some citizens hold political views only shared by a minority of MPs and are more distant from the dominant views in parliament. Some authors have demonstrated that congruence between citizen preferences and policies positively influences satisfaction with democracy (Ezrow and Xezonakis Reference Ezrow and Xezonakis2011; Ferland Reference Ferland2021). Substantively underrepresented citizens may also feel they are the losers of representative democracy and thus are more politically dissatisfied (Marié and Talukder Reference Marié and Talukder2021; Stecker and Tausendpfund Reference Stecker and Tausendpfund2016). As a result, we might anticipate that poor representation of one's political views in representative institutions could be associated with greater support for alternative institutions, such as DMPs. Research has not provided empirical evidence of a link between inadequate substantive representation and support for mini-publics. However, several studies have demonstrated that support for such democratic innovations is higher when citizens perceive that policy decisions adopted in citizen assemblies are more closely aligned with their policy preferences (Pilet et al. Reference Pilet2022; Landwehr and Harms Reference Landwehr and Harms2020; van der Does and Kantorowicz Reference van der Does and Kantorowicz2022; Werner Reference Werner2020). Consequently, we could expect a relationship between poor substantive representation and mini-public support.
H5. Citizens who see their political views less represented are more likely to support giving deliberative mini-publics a greater role.
This hypothesis can be tested for both parliament and the government as there could be difference in terms of effect between both (see Talukder Reference Talukder2023). On the one hand, one might expect that citizens whose policy preferences are badly represented in parliament are more prone to replace elected politicians with randomly selected citizens; parliament is the institution that controls the work of the government and where public debates are held. On the other hand, citizens might care more about the government's policy positions as it is the key actor in the decision making process. If this were the case, inadequate substantive representation at the government level would matter more. Both of these ideas are tested in our study. However, we should also accept the limits of our approach to this hypothesis. By looking at substantive representation in terms of congruence, we assume that voters can have clear views on their interests and how elected politicians represent those interests. Several authors have underlined the limit of such an assumption (see Dahl Reference Dahl2008). We will keep it in mind when interpreting our findings.
Data and Research Design
To test our hypotheses, we use a web-based survey, conducted between March 2 and April 3 2020, which included 15,406 respondents from fifteen Western European countries: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, the Netherlands, Norway, Portugal, Spain, Sweden, and the United Kingdom.Footnote 2 Respondents were recruited by the survey company DyNata (formerly Survey Sampling International), which employed country-specific quotas based on age, gender, education, and region – according to the latest census data. This approach ensured that each national sample was representative of the corresponding country's population in terms of these socio-demographic characteristics. The quotas were strictly enforced throughout the data collection period. The survey, which took approximately fifteen minutes to complete, included questions about political attitudes and preferences.
The primary advantage of this comparative survey is that it allows for a more robust test of our hypotheses compared to a single-country study. A cross-country study offers a range of configurations, including different types of opposition parties, variations in the proportion of female MPs, MPs with lower levels of formal education, less economically well-off groups, variations in citizens’ parliaments, and governments' policy preferences. Examining fifteen Western European countries is particularly relevant for our study. These countries are established democracies with a long tradition of representative democracy, allowing citizens to recognize their strengths and limitations. Recent studies also indicate that most of these countries have held citizen assemblies organized by public authorities at the national or regional level (Paulis et al. Reference Paulis2020). This context increases the likelihood that our respondents are familiar with this democratic innovation, thus enhancing the reliability of their survey responses regarding support for DMPs.
Our dependent variable was based on a survey question asking respondents if they supported giving a more significant role to DMPs. Respondents were first introduced to the concept of DMPs selected through sortition with a brief description: ‘People sometimes talk about the possibility of letting a group of citizens decide instead of politicians. These citizens will be selected by lot within the population and then gather and deliberate for several days to make policy decisions like politicians do in parliament.’ While this simplifies the institution, it captures its two main characteristics: (1) sortition and (2) deliberation on policy issues. Furthermore, in this web-based survey format, written instructions must be concise and straightforward to ensure readability and comprehension, maximizing the likelihood of meaningful responses. The same approach, one based on single-survey items providing basic information about the new decision-making instruments, has been used in several other recent studies on DMPs (Goldberg Reference Goldberg2021; Goldberg and Bächtiger Reference Goldberg and Bächtiger2022; Pow Reference Pow2021; Van Dijk and Lefevere Reference Van Dijk and Lefevere2023)Footnote 3 as well as in earlier studies on support for referendums (Bowler, Donovan, and Karp Reference Bowler, Donovan and Karp2007; Schuck and de Vreese Reference Schuck and de Vreese2015).
After the short description of deliberative citizens' assemblies, we asked respondents to answer the following question: ‘Overall, do you think it is a good idea to let a group of randomly selected citizens make decisions instead of politicians on a scale going from 0 (very bad idea) to 10 (very good idea)?’ Most respondents understood the question as only 5 per cent of the respondents answered ‘don't know’. These respondents were removed from the dataset.
An important aspect to consider is that the survey question refers to deliberative citizen's assemblies as an alternative policy making institution. This differs from most real-life cases of citizen assemblies, which primarily serve a consultative role (Paulis et al. Reference Paulis2020; Setälä Reference Setälä2017; Setälä and Smith Reference Setälä, Smith and Bächtiger2018). Nevertheless, we chose to present a decision making version of the institution to better reveal respondents' preferences by raising the stakes. We were concerned that even those who do not particularly favour deliberative citizen's assemblies might still report some support for a consultative version of the institution because they perceive it as ‘harmless’ (see Goldberg and Bächtiger Reference Goldberg and Bächtiger2022). However, we kept the radical nature of the question in mind when interpreting the results below since previous studies suggest a higher level of public support for the consultative version than the decision making one (Bedock and Pilet Reference Bedock and Pilet2020; Rojon, Rijken, and Klandermans Reference Rojon, Rijken and Klandermans2019).
Answers to this question reveal that citizens are quite divided about DMPs. We can see that about 18 per cent of all respondents are strongly against citizens assemblies to replace elected politicians. For the other respondents, we see a normal distribution between scores of 1 to 10, with a median value of 4.32 on a 0–10 scale (standard deviation = 3.05). In the appendix, we report the same histogram by country. Strikingly, support for deliberative citizens' assemblies is relatively constant across countries. The largest difference is between Norway and Denmark (median value around 3) and Francophone Belgium (median value around 6). There may be a relationship between the level of familiarity with this tool and support for DMPs across countries. Indeed, neither Norway nor Denmark has ever implemented mini-publics, whereas French-speaking Belgium has witnessed numerous examples in recent years (Paulis et al. Reference Paulis2020; Vrydagh et al. Reference Vrydagh, Devillers, Talukder, Jacquet and Bottin2021) (Fig. 1).
Distribution of the level of support for DMPs to replace elected politicians.
Regarding our independent variables, our first set of hypotheses (H1a and H1b) posits that election losers, defined as those who voted for a party that ended up in opposition, are more supportive of mini-publics. We differentiate between those who voted for a party currently in opposition (H1a) and those who voted for parties that were consistently in opposition between 2000 and 2020 (H1b).
Our operationalization was based on respondents' vote choice and does not capture partisanship, which refers to the intensity of attachment voters might have for a specific party. This concept is particularly important in US politics, where most voters strongly or weakly identify with one of the two main parties (Keith et al. Reference Keith1992). In Europe, by contrast, partisanship is also relevant but tends to be more fluid, as party systems are more fragmented and facilitate electoral volatility (Dassonneville Reference Dassonneville2022; Huddy, Bankert, and Davies Reference Huddy, Bankert and Davies2018). Since our study covers fifteen Western European democracies, we have chosen to analyze vote choice to determine whether voters would be classified as election losers or winners. It would have been interesting to have also examined the intensity of their attachment to the party they voted for. Regrettably, such a measure of partisanship was not available in the survey.
Then, as developed in the theoretical section, those who are descriptively underrepresented in the Parliament could also be considered to be ‘losers’ of representative democracy. To test H2 to H4, we included sex (women/men), level of formal education (lower secondary education or less, higher secondary education, tertiary education), and perception of household income (living comfortably, coping, difficult, very difficult), considering that, in all countries, women, individuals with a lower level of formal education, and those who feel they are facing financial strains are underrepresented in parliament.
The last hypothesis of our theoretical framework (H5) tackles the issue of substantive (under)representation and is operationalized as opinion congruence (Ezrow and Xezonakis Reference Ezrow and Xezonakis2011; Kirkland and Banda Reference Kirkland and Banda2019; Mayne and Hakhverdian Reference Mayne and Hakhverdian2017). More specifically, we collected the ideological position of all political parties represented in parliament in all fifteen countries based on the latest (2019) Chapel Hill Expert Survey Data. To compute a score of opinion-congruence, we combined the Chapel Hill Expert Survey Data with each respondent's self-placement on the left-right axis. To operationalize the data, we first computed the mean position of parliament on the left-right axis based on the share of seats each party represented in parliament. We then computed the absolute distance between the parliament's mean position and each respondent's self-placement on the left-right axis. Similarly, we computed the mean ideological position of the government and then computed the absolute distance between the government and the ideological position of each citizen. Consequently, a perfectly congruent respondent would have a score of 0; by contrast, a score over 0 would indicate lower opinion congruence between the respondent and the parliament (either toward the left or the right).
In addition to these variables on losers of representative democracy, we also incorporated the variable ‘satisfaction with democracy’ (SWD) into our models. As previously discussed, the most consistent finding regarding support for mini-publics is that they are more widely supported by politically dissatisfied citizens (see Goldberg and Bächtiger Reference Goldberg and Bächtiger2022 in this journal). Our aim with this study is to demonstrate that underlying dissatisfaction with democracy is the experience of being a loser of representative democracy. This feeling leads to increased criticism of the way democracy is functioning and, subsequently, support for alternatives such as mini-publics.Footnote 4
Finally, we included two control variables that might affect citizens' support for deliberative democracy according to earlier studies: political efficacy and age. Indeed, as mentioned in the literature section, several studies have shown that citizens who feel more competent are more supportive of deliberative instruments, while older citizens are less in favour of such democratic innovations (Christensen and von Schoultz Reference Christensen and von Schoultz2019; García-Espín and Ganuza Reference García-Espín and Ganuza2017; Gherghina and Geissel Reference Gherghina and Geissel2020; Jacquet, Niessen, and Reuchamps Reference Jacquet, Niessen and Reuchamps2022; Rojon and Pilet Reference Rojon and Pilet2021).
In terms of modelling, we take advantage of the cross-country survey and perform mixed-effects regression models using restricted maximum likelihood (REML) to take into account the multilevel nature of our data (which includes fifteen clusters) and to overcome the small-n problem at level-2 units (Stegmueller Reference Stegmueller2013). We followed the procedure recommended by Elff et al. (Reference Elff2021) and used REML estimators for variance parameters and a t-distribution with appropriate degrees of freedom for statistical inferences (using Satterthwaite's method). REML facilitates the analysis of data with a hierarchical structure. It estimates the variance components of random effects that cannot be explained by fixed effects in a computationally efficient manner. This method also accounts for biases related to the non-normal distribution of error terms.Footnote 5
Results
To test our hypotheses, we ran mixed-effects regression models in which we included each independent variable separately and then ran a model with all independent variables. Table 1 below shows the results of our analyses. Model 0 integrates two key control variables according to the literature: political efficacy and satisfaction with democracy. Model 1 focuses on citizens' votes and, more specifically, on voting for an opposition party. The second and third models focus on ideological congruence with the government and parliament.Footnote 6 The fourth model focuses on descriptive underrepresentation. Finally, the models of the fifth (with ideological congruence with the government) and the sixth (with ideological congruence with the parliament) integrate all variables.
Determinants of support for the replacement of elected politicians by randomly selected citizens (Mixed-effects regressions with restricted maximum likelihood)
Standard errors in parentheses.
*** p < 0.001, ** p < 0.01, * p < 0.05.
The results corroborate several hypotheses while rejecting others. Overall, our findings are in line with our general expectation: being a loser of the representative system stimulates support for deliberative democracy even when satisfaction with democracy is controlled for. Our results apply to the various facets of being a loser of representative democracy (voting for an opposition party, low descriptive representation, and bad substantive representation).
First of all, both H1a and H1b are confirmed. Those who vote for opposition parties are more supportive of deliberative mini-publics. The effect was observed for voters of parties currently in opposition but with a history of being in power and for parties that consistently remain in opposition. The magnitude of the effect is even larger in the latter case, confirming H1b, which posits that permanent election losers are more likely to support a shift away from a representative logic.
These findings are confirmed both in Model 1 and in Models 5 and 6. Model 5 predicts that those who vote for a government party give DMPs an average rating of 4, whereas this score reaches 4.5 for respondents who vote for a party permanently in opposition. It is also striking that those individuals who abstained are not more supportive of DMPs compared to those who voted for a party in government. This suggests that those who express a vote for an opposition party should not be lumped together with non-voters when it comes to attitudes towards DMPs: ‘voice’ (vote for an opposition party) leads to a more critical stance towards the representative status quo compared to ‘exit’ (electoral abstention).
The second, third, and fourth hypotheses pertain to descriptive representation in politics, particularly in parliament. Our results support some of these hypotheses but not all. Citizens with a lower level of formal education (H3) tend to be more in favour of DMPs as a replacement for elected politicians. A similar pattern is observed among citizens who feel their household income does not allow them to live comfortably (H4). The differences recorded for these two variables are relatively similar. Model 5 predicts that respondents with a lower secondary diploma or less assign a rating of 4.5 (on a support scale ranging from 0 to 10) to DMPs, while those with a tertiary level of education and living comfortably on their current income rate them at 4.1. The model predicts that those who feel they live comfortably with their current income assign a rating of 4 to DMPs, whereas those who feel that they are really struggling financially give them a rating of 4.6.
By contrast, H2 is not confirmed. Women are slightly less supportive than men to the idea of replacing elected politicians with randomly selected citizens. As argued during the debate on gender quotas in France, this may be because ‘women are not a category’ (Bereni and Lépinard Reference Bereni and Lépinard2004): in other words, even though they are typically underrepresented in parliament, they may not see themselves as a coherent group defending shared interests. Citizens with lower incomes and lower levels of formal education tend to have more precarious living and working conditions than the rest of the population, they may be more likely to see themselves as a coherent group with interests that are underrepresented in the current political system, unlike women.
Finally, we observe interesting findings regarding substantive representation (H5). The lack of ideological congruence between a citizen and parliament and government leads to greater support for citizens selected by lot to replace elected politicians. However, the size of the effect is more pronounced when it comes to the lack of congruence with the government. Model 6 predicts that respondents who are most congruent with the parliament give a rating of 4.1 to the idea of replacing politicians with citizens selected by lot, whereas this figure rises to 4.6 for citizens who have the highest level of ideological incongruence with parliament. For the government, Model 5 predicts that respondents whose position on the left-right scale is perfectly congruent with that of the government give a rate of 4 to DMPs, whereas this score reaches 4.7 for respondents whose position on the left-right scale is the furthest away from the government. These results suggest that, when it comes to substantive representation, ideological congruence with the government matters more than parliament when citizens evaluate the relevance of a radical reform of representative democracy that questions the role of elected politicians (Fig. 2).
Predicted support for randomly selected citizens replacing elected politicians according to the level of ideological incongruence with the government.
Finally, the goal of this study was not just to analyze whether being a loser of representative democracy would explain support for mini-publics to replace elected politicians but also to examine whether these configurations of being a loser of representative democracy could explain part of the recurrent finding in earlier studies that SWD was associated with support for mini-publics. This expectation can be discussed when comparing Model 0 with the full models (Models 5 and 6). Model 0 confirms the strong association between lower SWD and support for mini-publics. However, we can observe in Models 5 and 6 that when the variables capturing the configurations of being a loser of representative democracy are also included, the magnitude of the effect of SWD diminishes significantly, even if it remains significant. Its effect size is reduced by almost 2.5 times compared to Model 0, which only includes satisfaction with democracy (and control variables). For each unit increase in satisfaction with democracy, support for the replacement of elected politicians by randomly selected citizens increases by 0.04, which is relatively modest. In other words, when we control for being a loser of representative democracy, the effect of political dissatisfaction on support for DMPs does not disappear but becomes much less pronounced. In that respect, our findings directly complement recent studies (Goldberg and Bächtiger Reference Goldberg and Bächtiger2022; Pilet et al. Reference Pilet2022). It is largely because they are badly represented that the politically dissatisfied are more open to moving away from the institutional status quo and support the creation of truly powerful citizens' assemblies.
The robustness of these findings (and their interpretation) are further corroborated by additional path analyses that we ran using structural equation modelling (see appendix). With this technique, as with multivariate regressions, we can compare the effect of satisfaction with democracy on support for mini-publics with and without the variables capturing being a loser of representative democracy while also controlling for the associations between those variables and the measure of satisfaction with democracy. These path analyses confirm that the link between SWD and support for mini-publics remains significant when adding the variables of being a loser of representative democracy, but the size of the effect is much reduced. Moreover, the path analysis confirms that there is a direct effect of being a loser of representative democracy but also significant indirect effects of vote and income satisfaction through satisfaction with democracy. This is due to the association between being a loser of representative democracy and being less satisfied with the way democracy is working in your country. The analysis shows that being a loser is not fully equivalent to having low satisfaction with democracy. Rather, we believe that being poorly represented may make some voters dissatisfied with democracy, further increasing their openness to alternatives to representative democracy like DMPs.
Conclusion
Our analyses suggest that citizens who find themselves on the losing side of representative democracy are more supportive of mini-publics. Voting for an opposition party (particularly one that never forms a government), being poorly represented in descriptive terms, and even more so in substantive terms within representative institutions leads to greater support for a reform that would replace elected politicians with citizens selected – substantially. Our findings complement and provide context for earlier studies on the increased support for mini-publics among the politically dissatisfied (see, for instance, Goldberg and Bächtiger Reference Goldberg and Bächtiger2022). We demonstrate that the sources of this link might, at least in part, be related to the experience of being a loser of representative democracy for some citizens. This implies that support for DMPs does not necessarily indicate enthusiasm for the intrinsic virtues of democratic innovations such as deliberation. It also means that the roots of the link between dissatisfaction and support for mini-publics are largely related to the quality of representation. What losers of representative democracy express when they favour DMPs is the hope that this instrument may provide a greater voice and influence to citizens disadvantaged by the current institutional status quo.
These findings contribute to both academic and political debates. Academically, they add to existing research on mini-publics and can inform other fields of study. First, they complement studies on losers' consent, which show a link between losing and lower support for the functioning of (representative) democracy (Anderson et al. Reference Anderson2005). Our findings take this well-established observation a step further by demonstrating that an increased dissatisfaction with democracy could push losers to advocate for radical reforms (such as replacing elected politicians with randomly selected citizens). In this sense, they also confirm studies on political inequalities that have shown that advantaged citizens defend the institutional status quo, while disadvantaged ones are open to institutional change (Ceka and Magalhaes Reference Ceka and Magalhães2020). As political inequalities grow, the institutional stability of representative democracies could be at risk.
In this regard, our findings also contribute to the broader academic debate on process preferences. Studies on public support for alternatives to representative democracy (such as technocracy or more authoritarian regimes) have also emphasized the link with political dissatisfaction (Bertsou and Pastorella Reference Bertsou and Pastorella2017) without considering the sources of this dissatisfaction. It would be intriguing to explore whether support for other alternatives is also linked to the position of being a ‘loser’ of representative democracy. One could argue, for example, that support for technocracy may be more related to policy outputs and substantive representation (Bertsou and Caramani Reference Bertsou and Caramani2022) and less to descriptive representation than support for mini-publics.
Lastly, our findings could connect to other studies on the expansion of citizen participation. Future work could compare the dynamics of support for direct democracy mechanisms with what we observe about support for mini-publics. Some of these studies have also highlighted the link between being a loser of representative democracy and support for referenda (Bowler, Donovan, and Karp Reference Bowler, Donovan and Karp2007), but other research has noted differences within the public regarding support for deliberative and direct democracy instruments (Rojon, Rijken, and Klandermans Reference Rojon, Rijken and Klandermans2019). Investigations differentiating various dimensions of ‘losing’ in representative democracy (in political, descriptive, and substantive terms) could help make sense of these differences.
Beyond academic debates, our findings are also directly relevant to contemporary discussions on the use of mini-publics in democracies. Many supporters of DMPs emphasize the intrinsic values of these instruments, which could appeal to most people: deliberation, inclusiveness, and cognitive diversity (Vandamme et al. Reference Vandamme2018). Our findings reveal that, in addition to these virtues, support for mini-publics is also highly instrumental. It is based on a negative evaluation of the current representative model of democracy and is rooted in the feeling of being badly represented in the existing system. From this viewpoint, losers of representative democracy would only support DMPs in the long run if these instruments demonstrate their ability to effectively reduce the feeling of being a loser within the political system. This notion connects to the problem-based approach to democracy proposed by Warren (Reference Warren2017). Citizens are pragmatic in their support for democratic institutions. They support reforms, provided they perceive that the new institutions can address the problems they have identified within the current institutional architecture (see also Werner, Marien, and Felicetti Reference Werner, Marien and Felicetti2020).
Building on this argument, we may argue that the losers of representative democracy would only be satisfied by introducing DMPs if the new instrument could solve the causes of their political dissatisfaction. This suggests that genuine access to power should be granted to citizens who are not adequately represented politically, descriptively, and substantively. These citizens expect to address the widely demonstrated notion that most political systems, including liberal democracies, are more responsive to the wealthy and better-educated (Bartels Reference Bartels2018; Peters Reference Peters2018; Peters and Ensink Reference Peters and Ensink2015). They also anticipate that supporters of opposition parties will have a greater say in political decisions.
However, research on institutional change and support for mini-publics has indicated that not all citizens are open to changes heading in that direction. Citizens who are currently well-represented (through the party they vote for and elected politicians with whom they share sociodemographic characteristics and political views) within parliament and government tend to be more cautious about DMPs, particularly when these mini-publics do not collaborate with elected politicians but punctually or permanently replace them. This has also been confirmed by Goldberg and Bächtiger (Reference Goldberg and Bächtiger2022), who found that most German citizens are unwilling to give more than a mere consultative role to DMPs because they remain satisfied with how their representative democracy works.
Beyond citizens, we also know that political representatives are quite ambiguous – if not sometimes openly hostile – when it comes to mini-publics. With a few exceptions, they only support giving them a consultative role limited in certain areas. They refuse any arrangement putting the current political order of representative democracy into question (Jacquet, Niessen, and Reuchamps Reference Jacquet, Niessen and Reuchamps2022; Koskimaa, Rapeli, and Himmelroos Reference Koskimaa, Rapeli and Himmelroos2023; Rangoni, Bedock, and Talukder Reference Rangoni, Bedock and Talukder2023). In other words, their preferences are incompatible with the idea of granting losers of representative democracy fairer representation and a direct say in political decisions. In that sense, we can only concur with Goldberg and Bächtiger (2022, 7) when they stated: ‘recreating feelings of “ownership” over the democratic process via deliberative citizens’ forums might turn out to be a rockier road than many advocates of DCFs [Deliberative citizen forums] have imagined’. If losers of representative democracy – who hope to regain power and better representation through mini-publics – must face the hostility of elected representatives and citizens who feel adequately represented, the ability of mini-publics, or any reform for that matter, to address the roots of their political dissatisfaction seems seriously compromised.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0007123423000479
Data availability statement
Replication data for this article can be found in Harvard Dataverse at: https://doi.org/10.7910/DVN/PEX9AR.
Acknowledgements
The authors would like to thank all of those who have contributed to this study. In particular, we would like to thank Damien Bol and Andre Blais, who coordinated the survey, including the questions used in our analyses. We would also like to thank the members of the POLITICIZE research team, Davide Vittori, Emilien Paulis, and Sebastien Rojon, for their remarks and suggestions on our work on public support for sortition. Earlier versions of the paper were presented at the panel on ‘public support for democratic innovations’ at the 2022 ECPR General Conference in Innsbruck and at the 2022 Annual Conference of the Netherlands Institute of Governance (NIG) held at Tilburg University. We would like to thank all participants for their useful comments and suggestions. Finally, we would like to thank the reviewers and the editors of the British Journal of Political Science. Their detailed comments have greatly improved the quality of the article.
Author contributions
All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication.
Financial support
The study has received financial support from the European Research Council (ERC Consolidator Grant) under the European Union's Horizon 2020 research and innovation programme (grant agreement no. 772695) for the CURE OR CURSE?/POLITICIZE project.
Competing interests
The authors declare no conflicts of interest related to this article.
Ethical standards
The data used in this study is part of a broader survey (EPIS survey) coordinated by Prof. Damien Bol (King's College London) and Prof. André Blais (University of Montreal). The full survey received ethical clearance from King's College London's Research Ethics Office prior to data collection (registration confirmation reference number: MRA-19/20-14511, date 26th Sept. 2019).
