Referendums as extended arms of the government: evidence from an illiberal regime

ABSTRACT

Public support for referendums has been analysed extensively in established democracies, but the topic is investigated less in illiberal settings. This article aims to explain why citizens in an illiberal country (Hungary) support referendums. We argue that in such political regimes, referendums may serve as extended arms of the illiberal government rather than as providing people with direct access to decision-making. Our study uses individual-level data from a survey conducted in May 2022 on a national representative sample. The evidence indicates that those citizens in favour of referendums are characterized by anti-democratic attitudes and support for the initiator.

KEYWORDS:

• Referendum
• illiberalism
• initiator
• democracy
• attitudes
• Hungary

Introduction

Opinions about referendums vary greatly. Some scholars consider referendums to be an important democratic mechanism that is useful for the political system and societal development because it provides ordinary citizens with a way to express their views and engage directly in the decision-making process. Several salient decisions in society have been reached through referendums, such as the adoption of new constitutions, the independence status of many territories, or joining (or withdrawing from) international organisations. These attitudes and decisions have led to important levels of support for referendums and a high demand for them in the past two decades (Bowler and Donovan 2019; Gherghina and Geissel 2020; Mendelsohn and Parkin 2001; Schuck and de Vreese 2015). Others see referendums as problematic because they give voice to incompetent and uninformed citizens, they can be manipulated by initiators or authorities in the implementation phase, and they may lend themselves to the tyranny of the majority (Altman 2018; Bellamy 2018; Hollander 2019). Extensive literature illustrates that in democratic settings, attitudes about referendums can be the result of expectations about direct democracy, attitudes towards representative democracy, initiators’ behaviour, or voting outcomes. Public support for referendums is normally higher among politically dissatisfied and politically engaged or sophisticated citizens (Anderson and Goodyear-Grant 2010; Bowler and Donovan 2000; Craig, Kreppel, and Kane 2001; Donovan and Karp 2006; Schuck and de Vreese 2015; Walker 2003).

However, we know little about what drives citizens’ support for referendums in illiberal settings. There is some literature about how referendums are initiated and how the process works (Esen and Gümüşçü 2017; Podolian 2015; van Eeden 2019) in such contexts, but to date, no systematic study of the attitudes developed by citizens towards referendums has been done. To address this gap in the literature, the present article aims to explain why Hungarian citizens support referendums. We focus on Hungary for two reasons: it is an illiberal political system that used to be a democracy for more than a decade (Enyedi 2016a; Pállinger 2019), and its three most recent referendums, organised in 2008, 2016 and 2022, have been initiated by the same political party (Fidesz-MPP). Hungary’s experience with representative democracy is relevant because it allows the formation of attitudes since the start of the illiberal regime to be isolated. One of the referendums was organised when the country was still a democracy: in 2008, Fidesz-MPP (at the time just called Fidesz) was in opposition and initiated a plan to overthrow the government by referendum. In 2016, the same party was in government and initiated a referendum to gain popular legitimacy for its decision to oppose the EU quotas for refugees. More recently in 2022, the party called for a referendum on four questions related to the promotion of sex reassignment treatments for minors to legitimize the amendment of certain laws for the protection of children. The different topics of the three referendums and the time interval between them strengthen the empirical observations: the existence of similar effects means that they are long-lasting rather than contextual or topic-specific.

In illiberal regimes, we expect that the variables with high explanatory power in democracies will be less important than anti-democratic attitudes and support for the illiberal initiator of the referendum. Our central argument is that citizens who live in illiberal regimes may consider referendums as an extended arm of the government rather than as providing unmediated access to decision-making. Our analysis uses individual-level data from a survey conducted in Hungary in May 2022 on a nationally representative sample of 1,337 respondents. We use ordinal logistic regression and build separate statistical models for each of the three referendums.

This article contributes to two different areas of literature. First, it adds nuance to the understanding of direct democracy as complementing representative democracy. The classic meaning of complementarity is that referendums provide people with the opportunity to participate directly in the political decision-making process, with the aim of improving the overall quality of democracy (Morel and Qvortrup 2017). Complementarity in Hungary has a different meaning, whereby the use of referendums has moved in focus from empowering citizens to empowering the government party. Put bluntly, referendums are tools in the hands of the government party which it uses to promote its policy agenda. This approach reflects that the instrumentalization of referendums, studied from the perspective of initiators (Pállinger 2019; van Eeden 2019), is echoed among the people. We also highlight a new avenue for the intersection of direct democracy and representative government. Previous studies speak about legislative interference in the aftermath of direct-democratic processes in which politicians have the tools to enhance or obstruct the implementation of popular decisions (Ferraiolo 2023; Gerber et al. 2001). We illustrate that in Hungary, the referendum is considered to be an extension of the government’s activities, and thus the role of politicians is prominent throughout the process – from initiation to implementation – rather than only in the aftermath. Second, this article illustrates that citizens in illiberal settings are instrumental, and that they support referendums not so much because of their intrinsic values, but more because referendums increase the chance that the people will get what they want (Werner 2020). They accept the illiberal policies proposed through referendums when these come from the party they support (Gidengil and Stolle 2022).

The second section of this article reviews the literature on citizens’ perspectives about referendums. Next, we formulate several testable hypotheses. The fourth section presents the research design of this article with emphasis on the case selection, data, and variable measurement. The fifth section provides a schematic presentation of the results and their interpretation in light of the political realities in Hungary. The conclusions summarise the key findings and discuss the main implications for the broader field of study of referendums and citizens’ perceptions.

Perspectives about referendums

The literature on citizens’ perspectives about referendums includes several important strands of research. One of them argues that disaffection with how democracy works might lead to more positive perceptions and increasing support for alternative ways of decision-making, such as referendums. Conversely, citizens who are satisfied with the current functioning of the democracy are expected to be more inclined to preserve the status quo. In line with this, the political disaffection thesis holds that citizens who are disappointed by the political system are more inclined to support direct democracy through referendums. Generally, citizens who feel like outsiders in society are the most disappointed. Moreover, the lack of trust in politicians, the main party in government, and populist attitudes are generating higher appeal for referendums as important instruments for citizens to express their will (Bowler and Donovan 2019).

There is a positive relationship between politically dissatisfied citizens and support for direct democracy. Thus, referendums are more appealing to citizens who are less politically interested, less educated, and supporters of protest parties (i.e. far-right parties) (Dalton, Bürklin, and Drummond 2001). The citizens who are disconnected from the political system, sceptical of the established elites, and who yearn for an alternative political process are more supportive of referendums when compared to citizens who are satisfied with how representative democracy works (Schuck and de Vreese 2011; 2015). Similarly, the problem-based approach advances that referendums are considered appropriate instruments for solving some of the political systems’ issues. Citizens show increased support for referendums when they are dissatisfied with the government’s ability to listen to their needs. Support for referendums is weaker among citizens disappointed with the unresponsiveness of the government in countries with more referendums experience and stronger in countries with fewer referendums experience (Werner, Marien, and Felicetti 2020).

Further evidence from countries that experienced referendums in 2014 and 2016 (Switzerland, the Netherlands, the UK, and Hungary) shows that these are perceived in more positive terms among the “losers of modernization”, more specifically among politically dissatisfied citizens, disadvantaged by their income, and have anti-immigration views, in contrast to the “winners of modernization”, namely, the citizens who are politically satisfied, advantaged by their socio-economical status, and are open to immigration citizens who are less likely to support referendums. In line with this, beyond their democratic purpose, referendums can also become a threat by allowing illiberal views to advance in political life, mostly originating from politically marginalised citizens who might feel anxious about immigration due to their limited resources or skills. Moreover, the lack of trust in the parliament, political parties, and politicians consolidates citizens’ support for the use of referendums. Findings indicate that ruling parties feel less threatened or undermined by the use of referendums in countries where these are used more frequently and are more likely to be endorsed by citizens who support the governing party, than by supporters of the opposition parties (i.e. Hungary). One explanation is the influential and populist character of the incumbent government in Hungary (Rojon and Rijken 2021). Moreover, in countries where referendums are used, citizens have higher levels of subjective regime legitimacy due to their opportunity for direct engagement in decision-making (Gherghina 2017).

The second line of argument holds that citizens are more likely to support referendums if they support the initiator. Based on the partisan cue reasoning and similar to other types of plebiscites, several studies indicated that citizens are influenced in their perspectives on the topic of the referendum by their preferred party position (Aimer and Miller 2002; Gherghina and Silagadze 2019; Hobolt 2006; 2007; Nemčok, Spáč, and Voda 2019). There is an ongoing debate over the main important determinants for voters’ support for a referendum, namely party or policy. In line with the previous literature, citizens are more likely to use party cues, however, policy information is still relevant. Citizens with higher levels of political knowledge are more inclined to rely on partisan cues. They do not exclude policy information but use it instead to confirm their preferred party’s issue position along the campaign (Colombo and Kriesi 2017). In this regard, it can be expected that citizens might perceive more positively the referendums that are initiated by their preferred party or politician.

Another strand of literature argues that citizens support referendums if the prospects of winning on a certain policy are high and if they support a specific policy proposal (Werner 2020). Individuals with populist attitudes support referendums despite their unfavourable outcomes (Werner and Jacobs 2022). In democratic regimes, referendums are initiated more by non-populists, while in populist initiators are more common in authoritarian regimes and transition countries. In this context, citizens are more supportive of referendums initiated by populists, while “their image of referendum promoters is associated with a positive outcome”. Populists win more referendums than non-populists (Gherghina and Silagadze 2020, 8). These results might reflect that since populists-initiated referendums occur more often in authoritarian regimes and have a higher rate of victory, both the support for the initiator and the anti-democratic views could be a determinant factor of citizens’ continuous support for referendums.

The instrumentalization of referendums

All referendums are instrumental by nature because they adopt policies that people support. However, they might sometimes be used by some initiators for a purpose other than the content of the question. Such goals may include, among others, weakening an opponent, polarising a debate, going against a decision made by a judicial court or by the EU, or gaining visibility (Beach 2018; Daly 2022; Gherghina 2019; Morel 2001; 2007; Walker 2003). The literature refers to the policy and institutional goals for referendum instrumentalization and emphasises two types of actions: party-oriented (centripetal) or against opponents (centrifugal).

The centripetal approach leads to goals that favour the party, such as agenda-setting, popularity, legitimacy (policy goals), and party organisation (institutional goals). The centrifugal approach is expected to determine favourable outcomes at the expense of the party’s opponents, such as electoral mobilisation (policy goals), and deadlocks, bypassing institutions, and early elections (institutional goals) (Gherghina 2019). For instance, the Bulgarian Socialist Party (BSP) capitalised on the 2013 referendum initiative for the “Belene project”. The referendum was instrumentalized mainly for the purpose of electoral mobilisation against the governing party and had several beneficial outcomes for the party, such as consolidated internal discipline, organisational unity and support for the party leadership, gaining power against the governing party, and a secured position in power after the following parliamentary elections (Stoychev and Tomova 2019). In addition, the presence of party-cartelized settings with low party polarisation favours the manipulation of referendums by populist parties to overrule, destabilise, or bypass debates in or decisions by representative political institutions (van Eeden 2019).

Referendums which are initiated based on plebiscitary motivations seek to enhance the power of the initiator (Rahat 2009). Government-controlled referendums, which are also pro-hegemonic (i.e. in favour of the government authority) are not genuinely more democratic because they merely give the impression of popular power but are instead orchestrated to benefit the government (Morel 2001). Some referendums favour elite manipulation because more power is given to the executive and the “risks are particularly acute in the absence of codification and where the executive is strong relative to the legislature” (Qvortrup and Trueblood 2022, 13). When initiating a referendum, politicians have access to a set of resources that makes the instrumentalization possible. These resources include extensive information, campaign resources, and control over the referendum’s details, such as the topic, the wording of the question, and the timing of the vote (Gherghina 2019; Morel 2007; Walker 2003). For example, referendums can be scheduled to take place on the same day as general elections in order for elites to maximise their votes.

For this reason, government-initiated referendums may undermine democracy since the key goal is not to allow people to express their preference, but to achieve the initiators’ goals (Morel 2007). In France, the president may use a referendum to reinforce their power or to divide the opposition, particularly in situations where the president is in the minority (Daly 2022; Morel 2001; 2007). Poland is another case of the strong instrumental use of referendums by presidents and political parties for their personal political goals. For instance, the 1996 and 2015 referendums were instrumentalized by presidents Wałęsa and Komorowski respectively to secure their own second term in office. Political parties make use of referendums for several reasons: to challenge the president’s authority, to increase their electoral support, and to promote their foreign policy (Hartliński 2019).

The absence of formal and clear constitutional provisions about referendums enhances the likelihood of referendum instrumentalization, as was the case in the UK’s Brexit referendum (Daly 2022). Sovereignty referendums can also be instrumentalized for political motives, on the part of the secessionist, or on the part of the actors countering these movements (Mendez and Germann 2018; Sanjaume-Calvet 2021). Sometimes secession referendums are not entirely neutral tools that serve the people’s will but could also be instruments that serve the secessionists to create leverage over the central power of political elites (Cortés Rivera 2023).

Referendums are used even in non-democratic regimes mainly because “Dictators use them to increase their power. By contrast, in democratic countries they are […] used to constrain the powers of the elected representatives” (Qvortrup 2014, 250). In non-democratic regimes, elites use referendums to further legitimize their rule or to intimidate the population. The referendums endorsed by 99% of votes are more common in countries dealing with profound ethnic division and in tyrannical regimes: “the puzzling propensity of dictators to submit themselves, or their policies, to the ‘vote’ becomes a rational way of organizing the regime’s support and disorganizing its potential opponents” (Qvortrup, O’Leary, and Wintrobe 2020, 15).

Anti-democratic attitudes and support for initiators

This section argues that people who live in an illiberal setting can perceive the positive benefits of referendums when they have anti-democratic attitudes and when they support the initiators. The debate bridges two distinct bodies of literature. First, there is the idea that direct democracy can replace representative democracy. Some existing literature reports that direct democracy has become more appealing to citizens who are striving to find a solution to what they perceive to be the shortcomings of representative democracy (Bowler and Donovan 2019; Donovan and Karp 2006; Gherghina and Geissel 2019; Rojon and Rijken 2021). There are two broad perspectives about the role of direct democracy (referendums): it can complement, or replace, representative democracy. These two views are not mutually exclusive since some citizens prefer both types of democracy. Citizens’ power in the decision-making process is enhanced, especially through referendums, but the elite’s role is not replaced or bypassed given their responsibility and ability to organise (e.g. the wording, timing), and interpret the results of the referendums (Gherghina and Pilet 2021; Lupia and Johnston 2001). The checks and balances of representative democracy could prevent referendum manipulation by specific initiators (Durán-Martínez 2012). On the other hand, referendums are considered a replacement for representative democracy due to the latter’s failure to account for citizens’ preferences. Here, two main lines of argument explain the enhanced support for referendums among citizens who reject representative democracy.

The political dissatisfaction argument holds that citizens are more supportive of direct decision-making if they feel discontented with the political system under the current framework of representative democracy. Citizens who are dissatisfied with the institutions of representative democracy, particularly the parliament, as the best way to control the legislative process can have a stronger incentive to endorse direct forms of decision-making (Gherghina and Geissel 2019; 2020). People might consider that the law-making process is hindered by obstacles in the legislative arena. Likewise, citizens might be dissatisfied with the government’s performance due to slow policy implementation or what they see as mistaken prioritisation of issues on the agenda (Gherghina and Geissel 2020). If citizens feel misheard or excluded by political elites then they might consider referendums to be essential in addressing the democratic deficit (Bowler and Donovan 2019). Therefore, in both cases (disaffection with the parliament and government) referendums might represent viable solutions for citizens’ troubles with democratic institutions. For instance, citizens can use referendums to intervene directly in decision-making with no interference from political elites, thus bypassing the complicated process of political debates.

One reason why citizens develop anti-democratic attitudes is because they have lost their confidence in self-interested politicians. The growing decline in political trust is mainly caused by politicians’ low levels of integrity and honesty and is associated with more support for direct democracy (Kildea, Brown, and Deem 2021). When legislators’ decisions are driven by their personal interests, referendums represent a good alternative for citizens to counter-balance the government’s power by making it more responsive to people’s interests (Bowler, Donovan, and Karp 2007). The growing feelings of popular disaffection are amplified by “out of touch” legislators and unresponsive governments that fail to represent or address citizens’ demands (Gherghina and Geissel 2020; Kildea, Brown, and Deem 2021). For this reason, referendums can represent effective instruments to re-engage citizens based on their various positive effects on democracy, such as increased confidence, civic engagement, turnout, political knowledge, legitimacy of the policy-making process, and power to check elected officials (Bessen 2020; Bowler, Donovan, and Karp 2007; Rose and Weßels 2021; Setala and Schiller 2009). Citizens who consider politicians unreliable deciders of the people’s best interests might prefer to have a direct say in policies, unmediated by political elites. According to these arguments, we expect that:

H1: Citizens with anti-democratic attitudes support referendums.

Second, the support for the initiator influences the positive perception of referendums. The latter are sometimes “second-order” elections in which beliefs about “first-order” considerations (national politics) influence people’s attitudes. It “is enough that the government or the parties in power are the authors of the text submitted to the vote, or support it, for the vote to divert to their person” (Morel 2018, 159). European referendums are good examples where the government’s or president’s personal popularity counts more than the issue at stake, or where the vote on the European Treaty becomes a vote for or against the government (Hug and Sciarini 2000; Silagadze and Gherghina 2018). If this is the case, then the referendum will be perceived in line with the preference (or otherwise) for the initiator, and it will become a vote of confidence in the government (Marsh 2017). Citizens who identify with the government or the government’s programme are more likely to support its referendum proposals and less likely to have negative perceptions about the referendum (Bergman 2020). If the government is unpopular, citizens may vote against its proposals (Franklin, van der Eijk, and Marsh 1995) or refrain from participating in the referendum because non-voting in referendums that require participation thresholds to be valid is a form of action against the initiator (Gherghina et al. 2019).

The initiator of the referendum might have a stronger incentive to promote the policy at stake in order to increase its salience and mobilise the electorate. However, this does not mean that the decision to support the initiator’s proposal is blindsided or irrational, based entirely on partisan cues and not on factual information. These heuristics can guide citizens to take competent decisions despite the limited information available (Christin, Hug, and Sciarini 2002; LeDuc 2009). Instead, the support for the initiator should be seen as rational and informed precisely because the interest in, and appeal of, the initiator determines citizens to follow its stance and arguments during the referendum campaign or political debate, be more aware and informed on the issue, and finally, to take the right decision. Findings show that second-order and partisan considerations are important for less-informed as well as better-informed citizens. More significantly, second-order considerations appear to weigh more for better-informed citizens; that is, those who are influenced in their referendum vote by second-order aspects (e.g. their evaluation of the government) do not rely on political cues to simplify their decision in the absence of political interest and information. Instead, they engage in “a special kind of strategic voting”, where they vote with the political implications of their decision on the national arena in mind, a decision that is neither irrational nor uninformed (Font and Rodriguez 2009). Therefore, the supporters of the initiator might be more informed on the policy proposal put to vote and perceive referendums as beneficial to the political system because their decision will reflect their genuine preferences, interest, and knowledge on the issue:

H2: Citizens who support the initiator support referendums.

Control variables

In addition to these main effects, we control for several variables identified by previous research as potential influencers of attitudes towards referendums. First, we control for political interest as an attitude that often explains support for referendums in democratic countries (Donovan and Karp 2006; Gherghina and Geissel 2020). We could not use satisfaction with democracy, another strong predictor for support, due to its very high correlation with support for Fidesz-MPP (0.83, statistically significant at the 0.01 level). Second, we control for attitudes of support for fairness and inclusion. Citizens who value equality might see referendums as having a positive role in democracy because these may appear to be an alternative for citizens to exercise their influence over policy and challenge the influence of self-interested elites. Direct democracy facilitates political equality through agenda-setting and inclusion and can address problems of gender inequalities and marginalised segments in political participation (Saward 2001). Referendums can develop a feeling of confidence among individuals affected by inequalities that their political decisions can make a difference for themselves and to society. Inclusiveness could influence support for referendums because they enable the broad involvement of citizens in the decision-making process (Donovan and Karp 2006; Gherghina and Geissel 2020).

Third, we control for the competence and information of voters relative to referendums. If citizens consider their peers to be competent in taking decisions, then they might support referendums. Citizens are often accused of not having the ability and capacity to participate regularly in politics, either because the issues at stake are either unfamiliar or too complex, or since they lack the political expertise to make competent policy decisions (Budge 2012; Cozza, Elkins, and Hudson 2021; Lupia and Matsusaka 2004). Less-informed citizens tend to have a higher dependency on the elites’ position on the reform proposals and rely less on their own judgment and evaluation (Silagadze and Gherghina 2018). Finally, we control for age and education since they can be equated with different access to resources (information, participation). There is mixed evidence in the literature regarding their effects on attitudes related to referendums, but we also included them because of the 14-year gap between the first and the third referendums analysed in this article. Respondents had different access to these referendums according to their age.¹

Research design

Our analysis uses individual-level data from an original survey conducted online in May 2022 on a nationally representative sample of 1,337 respondents in Hungary. The sample is representative along the gender, education, area of residence, employment, and age variables. Voting with the government party was under-reported in the sample, but we use a weighing variable to bring the sample closer to the parliamentary results of 2022. The questionnaire was in Hungarian, and the average completion time was nine minutes. Three reasons make Hungary an appropriate case with which to test these hypotheses. First, the country is an illiberal setting in which the government party, Fidesz-MPP, has instrumentalized several referendums (Pállinger 2019; van Eeden 2019). Second, the three most recent referendums organised in the country had the same initiator: in 2008 Fidesz was in opposition, while in 2016 and 2022 it was in government. Third, there is high polarisation in society regarding the government party (Enyedi 2016b; Pirro and Stanley 2022). Fidesz-MPP has been in office since 2010, and while they enjoy the support of a large share of the electorate, the opposition has gained momentum in recent years. This polarisation of opinions is likely to be reflected in attitudes towards referendums.

The dependent variable of this study is a positive attitude towards referendums related to their effect on democracy. The respondents were asked the following question: “What is the impact of the following referendum on democracy in our country?” about the 2008 referendum on fee abolition, about the 2016 referendum on the migrant quota, and about the 2022 referendum about LGBTQI+ in education. The available answers are measured on a five-point ordinal scale ranging between “very negative” (1) to “very positive” (5). We used this measurement instead of the more common question about how much the people would support the use of referendums in the country to lower partisan bias. In 2016 and 2022 the opposition’s campaign focused on boycotting the referendums through absenteeism to invalidate the results since Hungary needs a participation quorum for the referendum results to be binding. As such, respondents could associate the idea of using referendum with the government. Our question gauges support by looking at the potential effects on democracy, which could make voters think broadly about the utility of referendums.

The anti-democratic attitudes (H1) are measured through the following question: “Which of the following statements is closest to your opinion?”. Those respondents who chose “In some circumstances, a dictatorship is better than a democratic political system” were coded 1, and all the other respondents who chose “All political systems are the same” or “Democracy is better than any other political system” were coded 0. Voting for the initiator (H2) is a dichotomous variable that is coded 1 for all the respondents who indicated Fidesz-MPP as the political party for whom they voted in the 2022 parliamentary elections. Answers indicating any other parties were coded 0.

The first four control variables are measured as the answers provided to the question: “Please rank on a scale from 1 (not at all) to 10 (very much), how important are the following values to you in terms of democratic governance in general?”. The respondents could indicate the importance of equality, diversity, the broad involvement of citizens in the decision-making process (inclusiveness), and taking the views of minorities into account. The next two variables were measured through the following question: “Please indicate how much you agree with the following statements about the Hungarian society”. The competence of citizens corresponds to the item “Citizens are able to make just as good decisions as politicians”, and the degree of information on policy issues corresponds to the item “Citizens are generally well informed on policy issues”. The available answers were recorded on a five-point ordinal scale with values between “completely disagree” (1) and “completely agree” (5). Education is measured on a three-point ordinal scale that distinguishes between elementary, intermediate, and higher education. Age is measured on a five-point scale that includes categories ranging from 18–29 years old (1) to over 65 (5). Appendix 1 includes the descriptive statistics for all the variables included in the analysis.

All “do not know / no answer” options were removed from the analysis. The variable for H2 is about voting, and those who did not vote were excluded from the statistical analysis.² We used multivariate ordered logistic regression and ran multicollinearity tests. The results indicate no reasons for concern: the highest value of the coefficient is 0.48 (statistically significant at the 0.01 level between diversity and inclusiveness), while the VIF scores are equal to or lower than 2.18. The correlation between anti-democratic attitudes and voting for Fidesz-MPP is positive and statistically significant, but much weaker than expected in an illiberal setting (0.29).

The 2008, 2016 and 2022 referendums in Hungary

Hungary has organised nine referendums at the national level since 1990. The practice of direct democracy in Hungary can be divided into four phases: (1) referendums related to the democratic transition; (2) foreign-policy referendums – on NATO accession in 1997 initiated by the national parliament and the 2003 EU accession referendum required by the constitution; (3) the 2004 referendums on dual citizenship and on privatisation in health care triggered by two popular initiatives; and (4) three referendums initiated by Fidesz since 2008 (Pállinger 2019). An overview of the three referendums organised in 2008, 2016 and 2022 is now given, with a focus on the context and discourse promoted by the initiator.

The 2008 referendum had its origins in a non-public speech by Prime Minister Gyurcsány which was leaked to the press in the fall of 2006, six months after the legislative elections. In the speech, the prime minister admitted that his party (the socialists) had lied in the two years before the 2006 elections about the country’s economic situation to get re-elected. Following protests and riots organised by conservatives and the extreme right, Gyurcsány announced that he would ask for a vote of confidence in parliament. As leader of the main party in opposition, Fidesz, Orbán dismissed this proposal and requested negotiations on the instalment of an expert cabinet. In October 2006, the parliament expressed its confidence in the prime minister, which led to new protests. Recognising that its endeavours were futile, the opposition came up with the idea of overthrowing the government via a referendum. At the end of October 2006, Orbán announced that his party had handed seven questions to the National Electorate Office related to the standard of living, fees, and prices. Three of these questions – on abolishing co-payments, daily fees at hospitals, and college tuition fees – were officially approved in December 2007 and a referendum was scheduled for March the following year. During the referendum campaign, Fidesz used aggressive nationalist rhetoric and enriched its populist identarian agenda with socio-economically sensitive topics, which were traditionally covered by the socialists. The referendum was valid, with a turnout of 50.5%, and most voters supported each of the three proposals (82–84%). This made the outcome of the referendum legally binding, which helped Fidesz to retain momentum until the next general elections in 2010, when they gained a landslide victory.

In 2016, after successful re-election in the national legislative elections, Fidesz initiated a referendum on the topic of the EU’s “Migrant Quota”, which mainly aimed at gaining support for, and enhancing the legitimacy of, its foreign policy (Pállinger 2016; van Eeden 2019). The question of the referendum was worded to reflect Orbán’s arguments against the EU’s plans to use migrant quotas for member states. The referendum campaign took place in parallel with government actions indicating a clear stance towards the issue: the government had voted against the proposal in the European Council and filed an action before the European Court of Justice. The government drew on the right-wing reworking of post-colonial discourse that had gained momentum in several Central and East European countries. Fidesz framed a fight of the people against an establishment of imperialists in Brussels, comparing it to Nagy’s historical battle against communism and the Soviet Union in 1956. The government’s campaign website, Népszavazás 2016, called for “sending a message to Brussels”, picturing the EU as a semicolonial power, and adopting a discourse focused on alleged infringements by “Brussels bureaucrats” on “our communities, our families, our culture and everything that makes Hungarians Hungarian”. To provide a transnational dimension, Orbán repeatedly emphasised the claim that Europe’s Christian heritage was under siege by menacing Muslims. The referendum was invalid due to low turnout (44%), but the government kept insisting that a large majority backed its initiative. They used the number of voters as evidence: approximately 3.6 million people supported the government’s initiative in the referendum, more than the 2.5 million voters who had voted for Fidesz in the 2014 legislative elections.

The 2022 referendum was on a different topic, but very similar to 2016 in terms of the government’s approach and the outcome. Since 2010, the government’s legislative agenda had given rise to several conflicts with the EU institutions. In 2021, The European Commission started an infringement procedure against Hungary, claiming that the Act LXXIX of 2021 on “tougher action against pedophile offenders and amending certain laws to protect children” discriminates against people based on their sexual orientation and gender identity; thus, the law also violates the EU’s internal market rules and the fundamental rights of individuals. The government argued that the EU initiated the infringement procedure as a disguise for the application of ideological-political pressure to push Hungary to withdraw the law. The government’s narrative during the campaign of the 2022 referendum followed the above-mentioned post-colonial discourse and called upon citizens to “send a message to Brussels”. Since the referendum was held at the same time as parliamentary elections, the party in government put efforts into voter mobilisation. However, the war in Ukraine that started in the middle of the electoral campaign became the key topic and the referendum played a minor role. The turnout was again too low (47%) for the referendum to be valid, but the government maintained the same discourse about the legitimacy of the proposal by pointing to the number of people who supported its stance in this referendum.

Analysis and results

The respondents’ opinions about the impact of the three referendums on Hungarian democracy are quite varied. Figure 1 presents the distribution of respondents, with the vertical axis depicting the values between 1 (very negative) and 5 (very positive). More individuals see the benefits of the 2008 referendum than those of 2016 and 2022. In 2016, many survey respondents considered the referendum as having had a negative impact on democracy in Hungary, with most responses recorded between the very negative and neutral stance. Regarding 2022, many respondents consider the referendum to have had a very negative impact on Hungarian democracy, with a small percentage of respondents seeing this referendum as having had positive effects. These differences in opinions regarding the impact of the three referendums are reflected in the value of the correlation coefficients (all non-parametric, and statistically significant at the 0.01 level): 0.36 between the 2008 and the 2016 referendums, 0.21 between the 2008 and the 2022 referendums, and 0.72 between the 2016 and the 2022 referendums. The high correlation between the two most recent referendums is not surprising given the similar rhetoric with which the government approached the two topics (see the previous section).

Figure 1. Distribution of respondents according to attitudes about referendums.

We ran six ordinal logistic regression models – two for each referendum – as follows: Model 1 includes the main effects and Model 2 includes all the control variables. All the statistical models are available in Appendix 2, which shows that the strength of the main effects does not change much when the control variables are added. The models’ goodness-of-fit is overall good. Figure 2 visually depicts the effects of Model 1. There is empirical evidence for both hypothesised relationships across the three referendums. The people with anti-democratic attitudes (H1) consider that referendums are beneficial to the quality of democracy. This effect is consistent across the three referendums, but it gets stronger as time passes. For the 2008 referendum, people with this attitude were 1.43 times more likely to have a positive attitude about referendums compared to those who appreciate representative democracy. In the 2022 referendum, the likelihood is greater than 3.5. One possible explanation for the weaker effect in 2008 might relate to the memory bias of some respondents regarding that specific past referendum, or to the fact that the younger respondents did not experience it. The effects are considerably stronger for the referendums organised in a period in which the country was characterised by illiberalism (2016 and 2022) than in the referendum organised when Hungary was still a democracy (2008).

Figure 2. Effects on support for referendums.

Voting for Fidesz-MPP (H2) finds very strong empirical support for all referendums. In 2008, those who supported the party were much more likely to consider that referendum as beneficial (OR = 2.43) compared to voters for other parties. In the 2016 and 2022 referendums, when the party was in government, this variable became deterministic (OR of roughly 17 and 21). In summary, people see the referendums as beneficial to the political system when they are used by the party in government, as an extension of its current practices, including authoritarian issues. In other words, the Hungarian public perceives the usefulness of referendums in line with their use by the current government.

The people’s anti-democratic attitudes and their support for a government party that has embarked on an illiberal path may interact. Both variables have strong or very strong effects on the support for referendums, and it is worth investigating how they interact empirically. Since interaction terms in logistic regression models cannot be directly interpreted, we trace this interactive relationship by plotting the predictive margins of voting for Fidesz-MPP conditional on anti-democratic attitudes. Figure 3 shows that the likelihood of Fidesz-MPP voters having positive attitudes towards referendums is conditioned to a very limited extent by their anti-democratic attitudes. However, the likelihood of people voting for other parties holding positive attitudes about referendums increases considerably once these people have anti-democratic attitudes. This observation matches the reality of ideological party positioning in Hungary: there are other parties that promote anti-democratic rhetoric, such as Jobbik, and their voters could also fall into this category.

Figure 3. Predictive margins of voting for Fidesz-MPP (95% CI).

Second, the conditional effect of anti-democratic attitudes becomes stronger over time for non-Fidesz-MPP voters: in the 2016 and 2022 referendums, the existence of anti-democratic attitudes led to a considerably higher increase than in 2008, especially among voters for other parties. The rejection of representative democracy boosts belief in the benefits of referendums among voters, possibly as a replacement mechanism for decision-making in an increasingly polarised society. This observation is in line with earlier evidence about populist polarisation in Hungary (Enyedi 2016b). If a party system becomes dominated by parties that pursue populist strategies, then both the relations among parties and the relationship of the party system with its environment will be affected. As part of that process, Hungarian political parties have adapted the use of the direct-democratic instruments to the system’s logic. Direct democracy in Hungary is used and controlled by the political elites mainly as a tool with which to mobilise their supporters, in order to reproduce their relative strength (Pállinger 2019). In brief, Fidesz-MPP uses referendums as an extension of its “us vs them” approach to politics, which polarises society and leads people to take stances towards referendums.

These adaptation strategies are reflected in popular attitudes towards the benefits of referendums, shown here in Figures 2 and 3. The following lines embed these observations in the Hungarian context and explore why people with anti-democratic attitudes and supporters of the initiator consider referendums to have positive effects on the quality of democracy in the country. Data from the 8th wave of the European Social Survey conducted in 2015–2016 indicate that Hungarians do not participate in non-conventional / non-institutionalized forms of participation but favour voting. The country’s voter turnout is higher than in most other Central and East European countries. Since 2010, Hungarian elections have been quite divisive, and until 2022 the opposition was fragmented. In this context, as part of a broader strategy to unify its core electorate and to legitimize an illiberal political system, Fidesz-MPP resorted to referendums. In doing so, it focused on single issues that did not require citizens to analyse the complex political processes around the issue, and which allowed the initiator to steer public attention away from salient issues. This is an old concern regarding referendums (Lupia and Matsusaka 2004); for example, the 2022 referendum was initiated by the government party against the backdrop of the increasing visibility of opposition parties in fall 2021. Several opposition parties joined forces and organised common primary elections, aiming to select a candidate who would provide solid competition to the incumbent prime-minister Orbán in the 2022 elections. The idea of a referendum shifted attention from the actions of opposition parties and brought the government party back into the spotlight. At the same time, it chose the single issue of LGBTQI+ that had not been a salient problem in Hungarian society until then. With this approach, Fidesz-MPP created a diversion away from more important problems, such as the EU sanctions faced by the country because of its government’s actions.

The strong effect of support for the initiator in Hungary confirms two theoretical lines of enquiry. On the one hand, referendums involve a relative asymmetry between political institutions and citizens, in which the institutions have the upper hand. Voters generally follow the recommendations of the government, and referendums often inculcate values espoused by the regime (Gherghina 2019; Qvortrup 2018). The three referendums in Hungary reflect this: over time, Fidesz-MPP voters appear to associate the referendums more closely with the practices of the party. These practices included several illiberal decisions, such as the adoption of a new Fundamental Law that was not voted on in a referendum, and other plebiscitary practices meant to augment their visibility such as the National Consultations (Batory and Svensson 2019). Their supporters thus see referendums as an appendix of the party, including their actions beyond representative democracy. This attitude has become more prominent since the party entered government.

On the other hand, recent research explains that Fidesz-MPP uses legalism to kill liberalism (Pirro and Stanley 2022). This is reflected in its use of two of the three referendums discussed here – the 2016 referendum intended to bend and break broader regulations discussed at the European level, which were far from the discriminatory measures adopted by the government against migrants in Hungary (Pirro and Stanley 2022). Equally importantly, the referendum could undermine the EU’s architecture of laws and norms that provides a set of safeguards against democratic breakdown in principle (Lührmann and Lindberg 2019). The 2022 referendum was used to forge illiberal legislation that continued a long series of discriminatory acts ranging from the criminalisation of homelessness or poverty to reforms in higher education, including banning gender studies. The party displayed aversion to genderist ideology and to same-sex partnerships, reflected in its decisions to formulate marriage in terms of heterosexual couples with the commitment to have children. All these were taken one step further when Fidesz-MPP initiated a fresh referendum in the spring of 2022 to bring together some of these concerns in a heterogenous set of general questions that targeted education, sexual minority rights, and the protection of children.

Among the controls, the only variable with a consistent and statistically significant effect on positive attitudes towards referendums in Hungary is the perception that people are well-informed about policy issues. This partially contradicts previous observations according to which information and referendums do not go together (Anderson and Goodyear-Grant 2010). In this study, politically disinterested people only express more support for referendums organised by Fidesz-MP in government (2016 and 2022). In the same referendums, less educated people support the referendums more than those who are highly educated. Younger people are more likely to consider that the 2008 referendum is beneficial to the quality of democracy in the country, but in the other two referendums, age no longer makes a difference. It is somewhat counter-intuitive that respondents’ opinions about equality or inclusiveness have no effect given the sensitive nature of the referendum topics, especially in 2016 and 2022.

Conclusions

This article aimed to explain why Hungarian citizens support referendums. This first study conducted on people’s support for direct democracy in an illiberal setting indicates that the people consider referendums to be potential replacements for representative democracy, and associate them with the initiator. The positive attitudes expressed towards referendums are influenced strongly by support for the initiator. The results are consistent across the three referendums analysed here, but are noticeably stronger for the referendums organised when the country had an illiberal political system.

The findings have implications for the study of referendums which reach beyond the case-study investigated in this article. The two arguments tested in the case of Hungary may be extended to other illiberal cases. It is plausible to assume that the public and illiberal political forces may follow suit especially now that de-democratisation is on the rise in the region and around the world. At the theoretical level, our results add nuance to the debate about whether direct democracy is replacing or complementing representative democracy. We show that the two are not mutually exclusive, and that they can coexist in a complex understanding of referendums, especially when adjusting the understanding of complementarity. More precisely, the strong effects of anti-democratic attitudes indicate that some people see referendums as an alternative to representation. This perspective is not bound to those who support the initiator, as was reflected in the analysis including the marginal effects. The very strong effects of partisan support for the initiator show that referendums are seen as complementary to the government’s actions within an imperfect representative system, which the government is seeking to reform. In brief, in an illiberal setting, the classic understanding of referendums as complementing representative democracy is transformed. Referendums are powerful tools in the hands of the government, which uses them to expand its power and to provide an alternative to representation.

An empirical implication of our analysis is the possibility of a direct comparison between the traditional explanation for support for referendums in democracies and in illiberal settings respectively. While in democracies support for referendums is driven by a dissatisfaction with democracy, in an illiberal setting this determinant takes the radical form of anti-democratic attitudes. In democracies, people’s political engagement is a relevant determinant for their support of referendums, as opposed to a more passive approach in the illiberal setting where closeness to the initiator rather than an interest in politics predicts support for referendums. Both findings indicate that in the latter case, referendums are associated with illiberalism and people may be willing to accept them as long as they are initiated by their preferred political party. Furthermore, our findings identify a new avenue for the intersection of direct and representative democracy. The classic interference of representative politics in direct democracy is ex-post, after decisions have been reached by citizens. The evidence from Hungary indicates the existence of a different model in which the political interference starts with the initiation. In other words, referendums are called to augment the legitimacy and power of a representative institution rather than to serve the altruistic purposes imagined in relation to direct democracy elsewhere. As such, the instrumentalization signalled by previous research persists. Finally, we provide evidence about how referendums are associated with the initiator, as we find that this association becomes much stronger once the initiator reaches government office. Referendums are tools in the hands of the government party to promote its policy agenda. This view is strongly supported by those people with anti-democratic attitudes who support the government’s actions.

The use of referendums can be attractive to illiberal actors in their attempts to preserve the majoritarian legitimacy in their political systems, while pushing their opponents to the fringes. The evidence in this article indicates that this can be achieved when the public’s understanding of referendums is closely linked to an illiberal initiator. Further research could explore the causal mechanisms outlined in this article. This can be done quantitatively by using survey experiments in which the respondents to be presented with several alternatives in terms of initiator, type of policy or type of referendum. This would allow understanding of whether the public reacts differently to referendums initiated by the opposition – since the 2016 and 2022 referendums were initiated by the government – or to referendums done on salient topics such as health or political system. Qualitative evidence with interviews may also be used to explain the link between support for the initiator and the perceived benefits of referendums. Finally, a comparison between national and local levels may produce important results, since the Fidesz-MPP dominance at the national level in Hungary is successfully contested in several localities. Such a comparison would indicate if the attitudes towards referendums differ according to the initiator, or if the general features stand.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Sergiu Gherghina

Sergiu Gherghina is an Associate Professor in Comparative Politics at the Department of Politics, University of Glasgow. His research interests lie in party politics, legislative and voting behavior, democratization, and the use of direct democracy.

Raluca Farcas

Raluca Farcas is a PhD Student at the Department of International Studies and Contemporary Politics, Babes-Bolyai University Cluj. Her research interests are political leaders, political participation and citizens' attitudes towards politics.

Daniel Oross

Daniel Oross is Senior Research Fellow at the Eötvös Loránd Research Network, Centre for Social Sciences, Hungarian Academy of Sciences, Centre of Excellence. His areas of expertise are youth policy, political participation, democratic innovations.

Notes

1 We also controlled for several other variables such as trust in politicians, opinions on the frequency of referendums, gender, and place of residence – but none of these were found to have strong and statistically significant results. We did not report them in the models to keep them parsimonious.

2 For robustness checks, we ran tests in which the non-voters were included and the results are similar to those reported in the article. We excluded the non-voters because we cannot know if they support Fidesz-MPP – the absenteeism could have been caused by other reasons than those related to political apathy or attitudes (e.g. being on holidays abroad at the time of election).

Appendices

Appendix 1. The descriptive statistics for the variables used in the analysis

Table

Variables	Average	Std. dev.	Min.	Max.	N
Support for referendum 2008	2.88	1.38	1	5	955
Support for referendum 2016	2.71	1.43	1	5	1017
Support for referendum 2022	2.21	1.57	1	5	1071
Anti-democracy	0.12	0.33	0	1	1324
Vote for Fidesz-MPP	0.46	0.42	0	1	961
Political interest	3.92	1.18	1	5	1214
Equality	7.53	3.76	1	10	1270
Inclusiveness	7.24	3.61	1	10	1176
Competent citizens	2.96	1.25	1	5	1097
Informed on policy	1.88	0.95	1	5	1135
Education	1.99	0.76	1	3	1337
Age	3.42	1.40	1	5	1337

Appendix 2. The ordinal logistic regressions for effects of referendums on democracy

Table

Variables	2008		2016		2022
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
Anti-democracy	1.43*	1.39	2.96**	2.89**	3.53**	3.63**
Vote for Fidesz-MPP	2.43**	2.64**	17.40**	16.66**	28.86**	25.31**
Political interest		0.97		0.77**		0.79**
Equality		0.99		0.99		0.99*
Inclusiveness		1.00		1.00		1.00
Competent citizens		1.06		0.93		0.90
Informed on policy		1.19*		1.40**		1.67**
Education		0.97		0.75**		0.76**
Age		0.83**		0.95		0.99
N	759	707	804	744	845	776
Pseudo R^2	0.02	0.03	0.15	0.18	0.22	0.26
Log likelihood	−1184.06	−1086.78	−1093.94	−973.60	−872.43	−757.27