• 1 Introduction

  Research on political behaviour gives enormous attention to the question of what compels citizens to turn out to vote in democratic elections (Blais, 2006; Blais & Carty, 1990; Blais & Dobrzynska, 1998; Crepaz, 1990; Frank & Martínez i Coma, 2023; Jackman, 1987; Jackman & Miller, 1995; Kostelka & Blais, 2021; Radcliff, 1992; Radcliff & Davis, 2000). This focus is largely grounded in normative concerns. Since elections are so instrumental for democracy by giving citizens a key instrument to influence policies and to find representation, it is of vital importance that turnout is high (Lijphart, 1997; Powell, 1982). Sufficiently high turnout is the condition for elections to produce governments that are representative of the people (Dahl, 1971; Pitkin, 1967).
  Two of the most consistent determinants of voter turnout are polarisation and compulsory voting regulations (Birch, 2009; Kostelka et al., 2022; Singh, 2021). Polarisation drives voter turnout in two ways; either because ideological polarisation increases the utility of voting (Dalton, 2008; Downs, 1957; Ellger, 2023; Stokes, 1963) or because affective polarisation strengthens the emotional value that is attached to the outcome of the election (Harteveld & Wagner, 2022; Phillips, 2024; Serani, 2022; Ward & Tavits, 2019). Compulsory voting, meanwhile, drives turnout because it requires citizens by law to vote in an election (Birch, 2009; Dassonneville et al., 2023; Kostelka et al., 2022). Yet, we know little about how these two interact. That is, studies that examine the role of polarisation in stimulating turnout often exclude compulsory voting countries (Ellger, 2023; Wessels & Schmitt, 2008). This is problematic, because the context of compulsory voting can influence political behaviour (Dassonneville et al., 2019; Singh, 2023), and it is often suggested to stimulate a civic duty to vote (Chapman, 2019; Feitosa et al., 2020; Quintelier et al., 2011). The impact of polarisation on the motivation to vote among such an electorate might therefore be seriously suppressed.
  Yet, testing the polarisation-turnout mechanism in a context of compulsory voting is challenging, as both the polarised and the non-polarised voters are required to vote. Making a meaningful distinction between their turnout behaviours is therefore complicated. In this article, we aim to overcome this issue. Relying on the Belgian case, we employ a survey question that asks respondents to what extent they would still be willing to cast a vote in a hypothetical future scenario where compulsory voting is lifted. Accordingly, we examine whether polarisation – both ideological and affective – has a positive impact on this voluntary voting intention. This allows us to detect whether polarisation is also a driving force of turnout in a compulsory voting system and offers some short-term insights about a potential ‘polarisation participation gap’ in case compulsory voting would actually be replaced by voluntary voting. We test our argument relying on data from Belgian national election studies of the past three decades (1991-2019), and we analyse the two regions of Flanders and Wallonia separately.
  We find that turnout attitudes are similar across both linguistic regions and remain remarkably stable across time. Both ideological polarisation and affective polarisation are indeed positively associated with the willingness to vote, but ideological polarisation is not consistently statistically significant. Affective polarisation, on the other hand, significantly contributes to turnout in almost all elections under study. In what follows, we first review the literature on polarisation and compulsory voting, from which we derive our set of hypotheses. We then describe our case, data and methods and present the results. We conclude with some implications of our findings.

• 2 Polarisation and the Turnout Calculus

  Election and party scholars have studied the topic of political polarisation over several decades (Abramowitz & Saunders, 2008; Downs, 1957; Hetherington, 2009; Powell, 1982; Schattschneider, 1960; Stokes, 1963). Generally speaking, we can identify two main conceptualisations of political polarisation: ideological polarisation and affective polarisation. In this section, we outline these two conceptualisations, their differences and how they are theorised to stimulate voter turnout. Finally, we discuss how compulsory voting might alter these mechanisms.

  2.1 Ideological Polarisation

  Ideological polarisation revolves around ideological divides, usually at the party level (Sartori, 1976). It is concerned with the extent to which political parties occupy different positions in the ideological space, which can be captured on the simplified left-right continuum. This ideological positioning offers the main framework for parties to compete electorally and to attract voters who are ideologically close. In essence, the concept of ideological polarisation refers to one of the core tasks of political parties: to channel societal divisions into clear policy platforms that are distinguishable and to create room for political competition during elections, which provides voters the opportunity to find adequate ideological representation (Powell, 1982; Schattschneider, 1960). Accordingly, ideological polarisation is low when parties position themselves ideologically close to each other, with little discernible differences between them. In contrast, when parties are more dispersed alongside the ideological divisions of a political system – for example, the left-right continuum – ideological polarisation increases.
  From the start, research on ideological polarisation has studied its connection with voter turnout. For example, the spatial theory of voting argues that voters behave rationally during an election and engage in a cost-benefit analysis to find a party that is ideologically nearest (Stokes, 1963). With higher ideological polarisation among the parties, the choice subset of parties increases for voters. This increases the utility of voting, as the maximum distance between a voter and the party that is ideologically closest reduces. In turn, the higher utility of voting should drive more voters to the ballot box (Downs, 1957). Conversely, if the political offer is low, chances are higher that the distance between voters and parties increases, which in turn fuels the likelihood of abstention. Another way in which low ideological polarisation can contribute to vote abstention is when there are virtually no differences between the political parties. In this case, a voter might be closely aligned to several parties in the system, but sees no utility in voting since the different parties barely offer diverging views, leading to indifference about the vote.
  Importantly, the utility of voting – and thus the decision whether to turn out – depends on how parties position themselves before an election. The spatial polarisation of parties offers voters increased options for representation, which increases with the emergence of new parties that fill a gap in the ideological spectrum (Tavits, 2006). For instance, emerging populist radical right parties have appealed to voter groups that were poorly represented by existing (mainstream) political parties (Kriesi, 2014). Still, existing parties can also engage in this process by addressing new voter groups and taking more radical positions (Spoon & Klüver, 2019).
  Proximity to a party is only one part of how ideological polarisation stimulates turnout. Additionally, when voters perceive parties as ideologically distant to them, the utility of keeping them out of power should be high. Again, this is dependent on the full dispersion of ideological polarisation in the system (Dalton, 2008). If a voter is ideologically distant to all the parties in the system, and the differences between these parties are minor, the ideological distance will not drive the utility of voting. It will only do so if there are also parties that are ideologically close to the voter and the ideological dispersion in the party system is high. This is also why the original way of measuring ideological polarisation, namely through counting the number of political parties, has been insufficient to determine its effects on turnout (Blais & Carty, 1990; Blais & Dobrzynska, 1998; Crepaz, 1990; Jackman & Miller, 1995; Powell, 1982; Radcliff & Davis, 2000). Recent studies that capture ideological polarisation through the dispersion of parties on the left-right continuum indeed consistently find that it increases turnout across the globe (Béjar et al., 2020; Dalton, 2008; Ellger, 2023; Hobolt & Hoerner, 2020; Moral, 2017).

  2.2 Affective Polarisation

  Affective polarisation, which focuses on the level of political polarisation between voters rather than parties, has recently received much scholarly attention (Iyengar et al., 2019; Wagner, 2024). Affective polarisation is rooted in social identity theory, and it posits that people are strongly influenced by group identities that offer cues to categorise the world around them (Tajfel et al., 1971). Most evidently, group categorisations lead people to positively assess other people from the same group, while out-group members are treated with bias, discrimination or outright hostility (Tajfel, 1970). By applying this theory to political behaviour, Iyengar et al. (2012) argue that one of the core political identities functions through partisanship and, as such, offers a salient group membership for voters. Voters thus tend to categorise other people according to the party for which they vote: they familiarise with voters of the same party but show more negative attitudes towards people of other parties (Garzia et al., 2023; Harteveld, 2021; Iyengar & Westwood, 2015; Reiljan, 2020; Wagner, 2021). The extent of affective polarisation among individuals is therefore reflected by the extent to which affect towards the in- and out-party diverges: increased positive affect for the in-group party (voters) and/or increased negative affect towards out-groups of parties and voters.
  Overall, affective polarisation is studied to apprehend a range of negative consequences on political (Kingzette et al., 2021; Torcal & Carty, 2022) or social (Huber & Malhotra, 2017; Lee, 2022; Martherus et al., 2021) behaviour, yet it has also been suggested to foster political participation (Ahn & Mutz, 2023). However, the mechanism between affective polarisation and turnout is different than for ideological polarisation. While ideological polarisation should trigger a rational calculus of whether or not to participate, the effect of affective polarisation is rather rooted in emotions. In this regard, both constitutive aspects of affective polarisation matter: positive in-group emotions and/or negative out-group emotions. On the one hand, positive feelings towards the political in-group entail that voters are positively attached to people who share their political opinion. These likeminded people offer them a social identity that is often expressed through partisanship (Huddy et al., 2015), or an issue that is strongly politicised (Hobolt et al., 2021). The larger the sympathy of voters for their own side, the higher the likelihood that they want it to perform well – which activates their political participation. Indeed, voters who are strongly affectively polarised tend to mix their social identity with their political identity (Ward & Tavitz, 2019). As a result, the success of their party also becomes a voter’s personal success, and accomplishments – as well as failures – are taken at a personal level. As elections are the key moment of competition between parties in electoral democracies, the best way to increase the chances to experience success and to avoid failure is thus turning out to vote and casting a vote for the own party.
  On the other hand, voters tend to dislike political opponents and the parties that represent these ideas. In fact, these out-groups might even be viewed as a threat, especially when dislike for them is high. During elections, voters who strongly dislike other parties are thus more likely to vote because they fear the success of these parties, which comes at the expense of the success of the own party. A straightforward mechanism that leads negative affect towards voting is that affectively polarised voters strongly dislike the issue positions or ideologies of other parties (Algara & Zur, 2023; van Erkel & Turkenburg, 2022). As such, voting can be an act to keep the parties that represent these unfavourable ideas out of office. This is especially important considering that many voters are not partisans; nor do they exhibit strong attachments to one of the parties. Rather than being concerned with the success of the party voted for, they are mostly concerned with the failure of other parties that are strongly disliked, also known as negative partisanship and negative voting (Bankert, 2021; Mayer & Russo, 2024; Weber, 2021). These effects have also been discerned in Belgium, in particular towards the radical right party Vlaams Belang (Boonen, 2019).
  Another mechanism taps into the role of emotions connected to winning and losing: voters who strongly dislike other political parties and their voters should be particularly happy when their party wins the elections and other parties are defeated (Janssen, 2023; Ward & Tavits, 2019). The prospect that their party could lose the election against this disliked group should motivate them to influence the election as much as they can, with the most obvious act being voting on election day.
  Recent findings support the positive association between affective polarisation and political participation. Serani (2022) finds that the propensity to vote in Spain indeed increases as affective polarisation rises, more specifically because of out-group dislikes. Similarly, Harteveld and Wagner (2022) show that affective polarisation is indeed an important driver of actual turnout in Germany, the Netherlands and Spain (controlling for partisanship and levels of ideological polarisation). Additionally, affective polarisation also positively affects other types of political participation, such as protest behaviour (Bettarelli et al., 2022) or political activism (Wagner, 2021).

  2.3 Compulsory Voting

  Existing work linking polarisation to turnout has predominantly – if not solely – focused on countries with voluntary voting systems. In fact, most cross-national analyses that studied the impact of voter polarisation on turnout simply exclude countries with compulsory voting (Ellger, 2023; Wessels & Schmitt, 2008). Accordingly, there is little evidence as to whether the turnout-polarisation mechanism also holds in a context of compulsory voting, such as in Belgium.
  Importantly, research on compulsory voting suggests that the mechanism between polarisation and turnout might not equally apply in these systems or that it may even not apply at all. Evidently, regulations that make voting compulsory are among the most robust predictors of electoral turnout (Birch, 2009; Singh, 2023). Countries that require their citizens to vote, rather unsurprisingly, witness consistently higher turnout rates than countries that do not, and these differences are substantial. Indeed, compulsory voting also answers the ‘equity dilemma’, as famously presented by Lijphart (1997): given that citizens with higher education and income are more likely to vote, policies are biased in favour of this group and tend to disadvantage citizens of lower socioeconomic backgrounds (Gilens, 2012; Peters & Ensink, 2015; Schakel, 2021). Correcting such bias is usually an argument to favour compulsory voting.
  Compulsory voting might also affect the polarisation-turnout mechanism in multiple ways, to the extent that it might not apply in Belgium. First, while compulsory voting is an established effective way of raising turnout levels, the reasons behind this relationship remain understudied. The most straightforward explanation – voters do not want to risk the legal consequences of abstention – fails to account for the fact that turnout is also exceptionally high in compulsory voting countries without sanctioning for nonvoting (Kostelka et al., 2022) or without enforcement of the legislative penalties for nonvoting (Dassonneville et al., 2023). For example, while Belgian voters should officially be fined when they abstain from voting, this law has barely been enforced in the last 20 years (Engelen, 2005; Kużelewska, 2016), with public prosecutors openly saying that they do not give priority to its enforcement (Vlaamse Overheid, 2021).
  As a consequence, factors besides legal consequences should also play a role in compulsory voting systems, as “the presence of a compulsory voting law has led many Belgians to view voting as a moral obligation” (Dassonneville et al., 2023, pp. 54-55). For example, compulsory voting is often argued to foster a civic duty to vote (Chapman, 2019; Feitosa et al., 2020; Quintelier et al., 2011). Of course, it is possible that this sense of civic duty solely rests on the legal obligation to vote, but it is also conceivable that such moral effects are part of an electorate that is used to turn out when elections take place. That is, citizens who voted in past elections are also more likely to vote in future elections (de Kadt, 2017; Denny & Doyle, 2009), most likely because the act of voting is self-reinforcing by creating an image among citizens of being a regular voter (Dinas, 2012). Furthermore, while non-voters are likely to become habitual voters during their lives, habitual voters are much less likely to become habitual abstainers (Plutzer, 2002).
  In turn, this could mean that polarisation matters less for the decision to vote. Voters who do not find ideological representation or who are not strongly emotionally involved in the election still cast a ballot because they believe it is their civic duty to do so. Indeed, in terms of ideological polarisation, citizens in compulsory voting systems are less likely to vote ideologically coherently (Dassonneville et al., 2019; Selb & Lachat, 2009). In addition, compulsory voting also changes political behaviour beyond turnout, for example, by decreasing the post-election winner-loser gap in democratic satisfaction (Singh, 2023).
  Finally, compulsory voting was introduced in many countries in order to reduce costs for political parties by not having to worry about mobilising voters (Birch, 2009). This suggests that parties under this system do need not to polarise to move voters to the ballot box and that they might focus on other aspects during the campaign. This might also decrease the importance of the mechanism that runs from polarisation to turnout. Simply put, if the electorate is not used to polarising parties, it might be driven by other factors to make the effort to vote.
  Accordingly, while polarisation usually drives turnout, we acknowledge that this mechanism might be different in a country with a history of compulsory voting such as Belgium. We therefore deem it important to test this mechanism in this context. Still, the exact change of the mechanism is difficult to predict; if anything, the previous discussion would lead us to expect null results at most. However, we formulate the hypotheses as we would expect them to apply in regular voluntary systems. In the data section, we discuss in more detail how we interpret potential null results, particularly in relation to our measure of turnout.

  H1: In compulsory voting systems, voters who perceive larger ideological differences between parties are more likely to vote.

  H2: In compulsory voting systems, voters with higher levels of affective polarisation are more likely to vote.

• 3 Case Selection

  We analyse the relationship between polarisation and voter turnout in Belgium, which constitutes a relevant case for two reasons. First, the party system is highly fragmented, particularly since political parties split alongside the linguistic divide in the second half of the 20th century. In fact, Belgium essentially harbours two party systems: Flemish parties compete for votes in the Dutch-speaking part of the country (Flanders), while Francophone parties represent voters in the southern French-speaking part of the country (Wallonia). Regarding polarisation, previous studies have found that Belgian voters in both systems show important variation in their levels of affective polarisation (Bettarelli et al., 2022; Westwood et al., 2018), which is to some extent explained by ideological polarisation (van Erkel & Turkenburg, 2022). In terms of ideological polarisation, both systems also present a key difference: the presence or absence of a successful and established radical right party. In Flanders, the radical right Vlaams Belang is one of the most successful parties since the 1990s, while no radical right party has established a continuous and significant presence in parliament in Wallonia thus far.
  Although both regional systems include parties within the same party families, the ability of these parties to reach governmental agreements has shrunk over the years. This is symbolised by exceptionally long coalition formations at the overarching federal level in recent years (De Winter, 2019), which contrasts sharply with the classical image of Belgium of a typical consociational democracy with elites bridging social cleavages by compromise. The most recent elections also saw a polarising trend in the political party offer, with higher seat shares both for the radical left (PTB-PVDA) and for the radical right (Vlaams Belang).
  Second, Belgium is a classical example of a compulsory voting system, with mandatory voting included in the constitution since 1893 and enforced for the first elections with universal suffrage in 1894. One of the core reasons to introduce compulsory voting was to increase the legitimacy of elections, as compulsory voting was designed to raise turnout levels and as such reflect the general will of the people more adequately (Kużelewska, 2016). Indeed, the effect of the constitutional change directly led to an enormous increase of turnout, reaching levels above 90% (Robson, 1923). On top of legitimacy arguments, compulsory voting was also instrumental to try and reduce the influence of radical parties, notably in urban areas among working-class voters. Elections were originally always held in Brussels, which forced political parties to reimburse voters for their incurred travel costs. With the introduction of compulsory voting, political parties did not need not to worry about mobilisation anymore, as voters were required to cast a ballot in their own surroundings (Kużelewska, 2016).
  The system of compulsory voting in Belgium continues to date. Officially, voters can be sanctioned with a fine if they do not participate on election day, although these have seldom been issued in the past 20 years (Dassonneville et al., 2023; Engelen, 2005). Despite the virtual absence of sanction, the system has remained highly effective: all elections in the 21st century have attracted turnout levels of around 90%. Throughout its history, the abolishment of the compulsory voting system has been an important topic of discussion in Belgian politics, with particularly liberal parties arguing that compulsory voting infringes on personal freedoms. Still, the topic has not been particularly salient in political debates, and Hooghe and Deschouwer (2011) notice that the high (constitutional) barrier to replace compulsory voting by voluntary voting at the federal level has made parties reluctant to even propose it, especially because the Socialist parties has indicated to veto such proposals.
  Nonetheless, the Flemish government abolished compulsory voting at the municipal and provincial elections in Flanders in a majority vote in July 2021. Given that the aforementioned constitutional barrier does not apply at these levels, the centre-right government (consisting of the conservative N-VA, the Christian-democratic CD&V, and the liberal Open Vld) was able to change the election rules for these lower-level elections. The compulsory voting system remains unchanged for elections at the regional, federal and European elections (as well as for the municipal and provincial elections in Wallonia and the Brussels-Capital Region). In an explanatory memorandum, the Flemish government explained that it decided to abolish compulsory voting because it is not in line with most other advanced democracies, and the non-enforcement of sanctions in case of nonvoting has made the system essentially already a voluntary one (Vlaamse Overheid, 2021). Such transformation has increased the relevance of mobilisation strategies in Flanders and, thus, the question whether polarisation could contribute to turnout.

• 4 Data and Methods

  4.1 Dataset

  To map polarisation and turnout attitudes, we study Belgian election surveys conducted over almost three decades. More specifically, we pooled data from eight election surveys (seven at the federal level and one at the regional level) between 1991 and 2019 (Table 1) (for an overview of national election studies from 1991 to 2007, see Frognier et al., 2011). These surveys were either cross-sectional post-electoral studies or part of larger panels survey fielded around the time of the particular election (in 2009, 2014 and 2019; for an overview, see Michel et al., 2023).¹x Panel surveys of 2009 (Deschouwer et al., 2009); panel of 2014 (Deschouwer et al., 2014); panel of 2019 (Walgrave et al., 2022). In this article, we only rely on post-electoral surveys. Our pooled dataset is thus composed of all election surveys that include the same question on respondents’ willingness to vote in case of voluntary voting (see the Dependent Variable section). Additionally, most studies also include questions that allow us to measure ideological polarisation, affective polarisation, or both. In total, we can study the effect of either ideological polarisation and affective polarisation on vote intention in four elections separately, and in three elections combined. The dataset also provides relevant control variables for each election year. All election surveys provide us with representative samples of the voting population in both linguistic regions and, thus, allow for studying differences between the party systems.

  Table 1 Election Surveys in the Dataset

  Year	Election	Compulsory Voting	Ideological Polarisation	Affective Polarisation
  1991	Federal	X	X
  1995	Federal	X
  1999	Federal	X	X	X
  2003	Federal	X		X
  2007	Federal	X
  2009	Regional	X
  2014	Federal	X	X	X
  2019	Federal	X	X	X

  4.2 Dependent Variable

  Election surveys generally over-represent turnout, which complicates establishing the relationship between one variable and turnout. Mapping turnout attitudes in an election with compulsory voting comes with the additional challenge of differentiating between citizens who turned out volitionally and those who only turned out because of the law. This is one of the core reasons why it has been so challenging for scholars to map the impact of polarisation on turnout in compulsory voting systems. Fortunately, the collected election studies in Belgium address this with a variable that asks respondents whether they would still vote in elections in case the system of compulsory voting would be abolished. This question is frequently used in research to map the impact of (abolishing) compulsory voting (Jackman, 1999; Mackerras & McAllister, 1999).²x Specifically, the question asks respondents: “If voting for parliament was no longer obligatory in Belgium, would you then always, generally, sometimes, or never vote or don’t you know for certain?” Respondents could choose between the answer options ‘never’, ‘sometimes’, ‘generally’, and ‘always’. We code the answer options from 1 to 4, such that higher values reflect a higher intention to vote (1 = never; 4 = always). Up to and including 2003, respondents were also offered a ‘don’t know’ option.
  Obviously, this survey question does not go without criticism, as it asks respondents about behaviour in a hypothetical scenario about the future. Still, we believe that this measure can be interpreted for two purposes in this study. First, it should map rather accurately the Belgians’ willingness to vote in the current system. Even though future behaviour cannot be predicted perfectly, we do believe that this question gives respondents a straightforward way to answer whether they wanted to have voted in the past election at all would they have had the freedom to make this decision themselves. This holds regardless of whether respondents voted because they think that penalties are enforced in case of abstention or simply because they think it is just to abide by the law. Second, we concur with Dassonneville et al. (2023) that since voting is such a regular behaviour (particularly in a compulsory voting system), the attitude towards it should predict future behaviour relatively well. As such, we think that this attitude should also reflect future turnout behaviour relatively well in case compulsory voting would be replaced by voluntary voting, but only in the short term (e.g. the first couple of elections).
  This also guides our interpretation of the absence or presence of effects of polarisation. That is, in case we do find a positive effect of ideological or affective polarisation, we can assume that the same causal mechanism is at play in voluntary voting systems. Yet, in case of an absence of effect, our interpretation is that (1) the citizens’ willingness to vote under a compulsory voting system is not determined by levels of polarisation, either ideologically or affectively and (2) if compulsory voting would be replaced by voluntary voting, the role of polarisation on turnout should be smaller, or even absent, at least in the short term. We emphasise that our research design does not allow for making inferences about the effects of lifting compulsory voting on political behaviour in the long term.
  Given that the survey question was asked in all election studies of our dataset, we can map the willingness to vote under a voluntary voting system over time. Figure 1 shows the average score towards this question as well as its distribution, with 95% confidence intervals, but excluding the respondents who indicated that they do not know.³x The percentage of voters who gave the ‘don’t know’ option is only somewhat considerable in 1991, with around 10% in both regions. In the three following election studies, the percentages are around 5 or even lower. Overall, this score is very stable, averaging just below a score of 3 (‘would generally still vote’), and there are no substantial differences between both regions. In Flanders and in Wallonia, the willingness to vote if mandatory voting would be replaced by voluntary voting is largely similar. Arguably, in the past ten years, the willingness to vote slightly increased – specifically, during the elections of 2014 and 2019.

  Development of willingness to vote if voting became voluntary
  

  Line graphs on top present the mean score of willingness to vote if voting became voluntary (1-4 scale). The bars present the proportion of respondents for each answer category.

  4.3 Independent Variables

  Two different questions allow us to operationalise our two independent variables of polarisation. For ideological polarisation, we use a question asking respondents to place the different parties running for election on a left-right continuum, ranging from 0 (left) to 10 (right) (Dalton, 2008). For affective polarisation, we use a typical feeling thermometer question asking respondents how much they like each party, ranging from 0 (lowest sympathy) to 10 (highest sympathy) (Gidron et al., 2022). In 2019, the survey actually asked respondents about their sympathy towards the voters of the different parties rather than the parties themselves. This difference taps into the discussion about differences between horizontal polarisation (towards voters) and vertical (towards parties) polarisation (Harteveld, 2021; Kingzette, 2021). We acknowledge these differences, but argue that they do not affect our demonstration: both horizontal and vertical affective polarisation should positively relate to turnout.
  Both scales of ideological polarisation and affective polarisation follow the spread-of-scores calculation as proposed by Wagner (2021). While this measure was specifically designed for the calculation of affective polarisation, it also suits the calculation of ideological polarisation. As a matter of fact, the established measure of ideological polarisation as proposed by Dalton (2008) relied on a similar calculation. The spread-of-scores measure is particularly suited to measure polarisation within multiparty systems, as it acknowledges that voters can be sympathetic or ideologically close to more than one party. Theoretically, these variables can range from 0 to 5, with values of 5 reflecting the most polarised citizens. It is calculated as follows:

  

  Where p represents the particular party, i the respondent, party LR_ip a respondent’s left-right placement of a party, and v_p the vote share of the particular party. For affective polarisation, party LR is simply replaced by the sympathy score towards the party. The mean left-right placement (or mean sympathy score for affective polarisation) should also be weighted according to the party size, which is done as follows:

  

  Importantly, this calculation of ideological polarisation concerns, essentially, perceived ideological polarisation by voters rather than the actual ideological polarisation of parties. Still, the use of voters’ assessment of the ideological position of parties to calculate ideological polarisation is common in the literature (Dalton, 2008; Ellger, 2023; Hobolt & Hoerner, 2020; Lachat, 2008; Moral, 2017). In addition, as much as actual ideological polarisation matters, it mainly matters if this is picked up by voters – as such, whether they perceive ideological polarisation themselves, as this should drive the turnout decision (Enders & Armaly, 2019).
  In Figures 2 and 3, we show the average levels of ideological polarisation and affective polarisation for the years in which respective variables are included in the election study. The mean levels of ideological polarisation have been very stable over the past decades, ranging between more or less 2.2 and 2.5, which is rather average compared to other countries (Dalton, 2008). There are some noticeable, albeit small, differences between the regions: Flanders is more ideologically polarised, most likely due to the higher presence of (radical) right-wing parties such as N-VA and Vlaams Belang, whereas the only party in Wallonia that occupies a somewhat centre-right position is MR. Yet, although the level of affective polarisation is stable over the past decade, it is also slightly higher in Flanders than in Wallonia. This difference could result from the fact that affective polarisation can be predicted relatively well by ideological polarisation (van Erkel & Turkenburg, 2022), with the populist radical right voters both receiving and giving relatively high levels of dislike towards other parties and voters (Harteveld, 2021). Furthermore, the levels of affective polarisation, which barely go above a value of 2, are comparatively rather low (Garzia et al., 2023; Wagner, 2021).

  Development of Ideological Polarisation
  

  Markers depict the average level of ideological polarisation per region, including 95% confidence intervals.

  Development of Affective Polarisation
  

  Markers depict the average level of affective polarisation per region, including 95% confidence intervals.

  Finally, our models explaining willingness to vote control for several variables that are important for political behaviour and, more specifically, for individual-level voter turnout. First, we include a variable that measures respondents’ political interest, ranging from 0 to 10 (except for 1991 when the question was not included in the survey). We also control for a respondent’s left-right positioning (0 = left; 10 = right) and for political extremism by taking the square root of the squared difference between a respondent’s left-right placement and the average left-right placement of the region in each election. Finally, we control for sociodemographic variables: gender (1 = male), age, education (1 = no education; 5 = university education) and employment status (0 = unemployed; 1 = not in labour force; 2 = employed; see Appendix A1 for full variable description).

  4.4 Methods

  We employ OLS regressions for each independent election year and each region separately. The variables for ideological polarisation and affective polarisation are included separately to avoid losses of observation due to item non-response. Since the main independent variables were asked in four elections, we present eight separate regression coefficients. Three of them – 1999, 2014 and 2019 – can be compared directly, given that these elections included both variables. Additionally, we also run regression models for the election studies of 1991 and 2003. The former includes the ideological polarisation variable, while the later offers another option to study the effect of affective polarisation. For reasons of readability, we present coefficient plots with the different election years such that the impact of the independent variables can be compared over time. Full regression tables can be found in the appendix. We standardise our independent variables around the mean, such that the coefficients represent the effect of one standard deviation (except for employment status [nominal] and gender [binary]).

• 5 Results

  We start by testing how differences in ideological polarisation between citizens affect their willingness to vote (H1). Figure 4 presents the coefficients of the main ideological polarisation variable as well as control variables divided by region.⁴x Note that, for reasons of readability, we do not include the employment dummies. Generally, we find that citizens who are unemployed are less likely to vote in future non-compulsory elections than employed citizens or citizens who are not part of the labour force. Also note that we cannot control for political interest in 1991 because the variable was not included in the survey.
  With the exception of Flemish voters in 1991, the effects of ideological polarisation on turnout are in the expected positive direction: the more ideologically polarised, the more likely citizens are to vote in future elections even if mandatory voting would be abolished in Belgium. However, this effect is only significant in three of the eight regressions, which means that though we can speak of a consistent effect, it appears to be weak. When significant, we find that the effect sizes are of about 0.1, while standard deviations of ideological polarisation are around 1 as well (with some difference per election and region; see Appendix A1). Substantially, moving from the least to the most ideologically polarised citizen would thus increase the willingness to vote under non-compulsory elections by 0.5 units on the 1-4 scale.

  Effects of Ideological Polarisation on Willingness to Vote under Voluntary Voting
  

  Our results show no significant difference between Flanders and Wallonia. That is, there are election(s) where increased ideological polarisation is significantly associated with increased willingness to vote under voluntary voting in both regions. Over time, we find that the effect of ideological polarisation is most evident in the last election of the dataset, 2019, in both regions. This could be the result of the particular developments during that election, with increased voting for radical parties both on the left and on the right. For example, radical right party Vlaams Belang increased its parliamentary representation with 18 seats, and the radical left PTB-PVDA also won 12 seats, while most mainstream parties lost seats. Still, we acknowledge that our measure captures the impact of ideological polarisation between voters and not between elections. As such, the nature of the relationship – as hypothesised – should remain the same. It is, however, possible that the rise of radical parties changed the distribution of ideological polarisation. For example, voters who perceived average ideological polarisation during previous elections could be perceiving stronger polarisation during the 2019 elections, yet their attitude towards voting under voluntary voting remained unchanged. If these voters were likely future voters, it could impact the relationship. Establishing such finding would require panel data, which are unfortunately not available. Yet, we still find some evidence that points in this direction: the standard deviation of ideological polarisation is notably higher in 2019 than in all other election years (1.23 vs. 0.83), but the standard deviation of the willingness to vote in a voluntary system question is stable around 1.18.
  As the impact of ideological polarisation on voluntary voting is only marginal, we turn to the impact of affective polarisation. To recall, voters with higher levels of affective polarisation should display a higher intention to vote in future elections characterised by non-compulsory voting (H2). Similarly, we ran four separate regression models for each election year, in both regions (however, the party [or voter] sympathy used to measure the affective polarisation variable is available for 2003 but not for 1991). Figure 5 presents the effects of affective polarisation on turnout by region.

  Effects of Affective Polarisation on Willingness to Vote under Voluntary Voting
  

  All coefficients of affective polarisation are in the hypothesised positive direction in all elections. Voters with higher levels of affective polarisation show a stronger willingness to vote, even if it was voluntary, than voters with lower levels of affective polarisation. Furthermore, these effects are strongly significant (p < 0.01) in all elections, except that of Wallonia in 2003. As such, we find that the effect of affective polarisation on willingness to vote is much more robust than the effects of ideological polarisation. Effect sizes vary only slightly, from about 0.1 to 0.2. For instance, considering an average effect size of 0.15, the effect of moving from the least affectively polarised citizen (0) to the most affectively polarised citizen (5) is 0.75 unit on the voluntary voting question (1-4), provided that the standard deviation of affective polarisation revolves around 1.
  Consequently, the results suggest that the emotional mechanism that drives voters to vote in voluntary electoral systems plays a similar role in a system where voters are compelled to vote. Voters who are more emotionally invested in Belgian elections – for example, because they strongly like or dislike one of the parties – are more eager to vote. Apparently, the political culture of the country, where voting has become habitual through enforcement, has not altered the effect that emotions have on the willingness to vote. Given that this is the case for both regions, which show quite strong differences in the party offer, it makes us confident that these results are robust. They also indicate that if Belgium were to make voting voluntary, the first elections should witness some kind of a ‘polarisation participation gap’. Whether this holds in the long term is difficult to say through our research design, but the recently unfolding evidence on the impact of affective polarisation on turnout in voluntary systems (Ellger, 2023; Harteveld & Wagner, 2022) suggests that it would. We again find that the 2019 election year has a stronger impact on the willingness to vote than before, as we also did for ideological polarisation. While the same aforementioned caveats exist about the nature of the relationship, we do again find that the standard deviation of the affective polarisation increases visibly. Whether the nature of this relationship indeed changes more structurally because of the influence of fringe parties should be revealed by future Belgian election studies.
  Finally, we take a closer look at the control variables. Contemporary debates on compulsory voting often focus on the type of voters who would be affected by the abolishment (or introduction) of such a system, yet they remain largely theoretical (Lijphart, 1997; for an overview of the arguments in favour and against compulsory voting, see Birch, 2009). Therefore, we also assess the effects of control variables on the willingness to vote under a voluntary voting system based on three decades of data. To increase the number of observations, we run regressions for each election year with only the control variables and the same dependent variable in Appendix B.5.
  We find that political interest is the most influential variable on willingness to vote: it has a positive and strongly significant effect in all elections under study, with the largest effect size. This is in line with recent cross-country evidence (Dassonneville et al., 2023). As such, our results point towards a strong effect of political sophistication on voluntary turnout, since the effects of education are also positive and significant in each election year that we studied (Gordon & Segura, 1997; Lachat, 2008; Luskin, 1990). Accordingly, a likely turnout gap is expected between politically sophisticated and unsophisticated citizens if Belgium were to replace its compulsory voting system with a voluntary voting system. More specifically, the gap would emerge between citizens with lower and higher education, thereby confirming Lijphart’s warnings (1997).
  Furthermore, researchers have often argued that older people are more likely to vote than younger citizens (Bhatti et al., 2012), while a gender gap in voting points towards the trend of a higher likelihood of voting among men than among women (Franklin, 2004). In fact, recently Dassonneville et al. (2023) found that Belgian females and younger citizens are less likely to vote if voting is not compulsory anymore. In our dataset, we fail to find these effects. For age, we find both positive and negative coefficients in the elections under study, but most of them are not statistically significant. For gender, our coefficient represents the effect of being male and should thus be positive. Again, most of our coefficients are statistically insignificant, which is in line with more recent research arguing that the gender gap in voting is decreasing or even disappeared (Inglehart & Norris, 2003; Kostelka et al., 2019).
  Still, we must stress that our results are somewhat more nuanced: the null effects of gender are driven by the inclusion of political interest. If we exclude political interest in the regressions, we indeed find that males are more likely to turn out during voluntary elections. This appeals to the difference in political interest that we find between both genders in our dataset, with males being significantly more politically interested. At the same time, these results also suggest that if a male and a female have the same level of political interest, they should not display significantly different attitudes in turnout intention. Notably, the other variables display the similar association (size) when political interest is excluded, also in bivariate regressions.

• 6 Robustness Checks

  To test the robustness of our results, we run models which include both ideological polarisation and affective polarisation together (for the election years in which they are both available: 2003, 2014 and 2019). However, including both variables in the models leads to a substantial drop of observations, in particular compared to the models that tested the impact of affective polarisation. Including both variables is therefore most likely to impact the result of affective polarisation. Therefore, we test the models incrementally: first, we test the original models with only ideological polarisation or affective polarisation included (for elections in which both are available); second, we test the models but only with the observations for which both variables are available. Finally, we include both variables of polarisation jointly (see Appendix B.4).
  In Flanders, we do not detect major differences once both variables of polarisation are included in the models. The significance of the positive coefficient of ideological polarisation collapses in 1999 only, once affective polarisation is included in the model (p = 0.104). Additionally, the coefficient for ideological polarisation in 2014 changes direction from positive to negative once affective polarisation is included. Yet, since this coefficient was already statistically insignificant and very close to 0 in the original model, we do not see this is as a meaningful change. For affective polarisation, we see that all effects remain in the same direction and that their significance is robust.
  In Wallonia, the inclusion of affective polarisation does not meaningfully change the coefficient of ideological polarisation in any of the election years. For affective polarisation, we do detect some changes in 1999 and 2014, where the originally strongly significant positive coefficient loses its significance in the joint model. This collapsed significance already occurs in the original model with the observations of the joint model only. As such, we can assume that the collapse of significance is not due to spuriousness but rather due to loss of observations, which is indeed substantial (respectively 16% and 39%).
  Finally, we run the joint models using multiple imputations for the missing values of ideological polarisation and affective polarisation. This leaves us with the same models, but with the number of observations that reflect the number of observations after list-wise deletion of respondents with a missing value for employment, education, left-right placement, political extremism, and/or political interest. We impute m = 20 for every missing value in the variables of affective polarisation, ideological polarisation, or both and thus generate 20 possible values for each missing value in either of the variables (this is well above the conventional amount of imputations; see Rubin, 1987). We report these results in Figure 6 (full regression tables in Appendix B.6). For ideological polarisation, we again only detect the loss of significance in Flanders in 1999 (again not completely vanishing with p = 0.099). However, we do not observe the collapse of significance for affective polarisation in Wallonia in 1999 and 2014 using multiple imputation; the coefficients remain positive and strongly significant (p < 0.01). These results support our interpretation that the significance loss is largely due to a drop of observations across models. All in all, we conclude that the original models are robust, except for the result of ideological polarisation in Flanders in 1999.

  Effects of Affective Polarisation and Ideological Polarisation on Willingness to Vote under Voluntary Voting with Multiple Imputations
  

• 7 Conclusion

  Political polarisation among voters, either ideologically or affectively, has been consistently found to stimulate voter turnout (Béjar et al., 2020; Dalton, 2008; Ellger, 2023; Harteveld, 2021; Hobolt & Hoerner, 2020; Phillips, 2024; Wessels & Schmitt, 2008). Still, most of this evidence has been found in systems with voluntary voting. Given that compulsory voting has the potential to alter political behaviour and attitudes (Chapman, 2019; Feitosa et al., 2020; Quintelier et al., 2011; Singh, 2023), we studied whether the polarisation-turnout mechanism also holds in the context of compulsory voting in Belgium.
  Through an analysis of three decades of election studies in Belgium, we find that the willingness to vote – captured through a question about hypothetical future voting behaviour under voluntary voting – is mainly driven by affective polarisation rather than ideological polarisation. Belgian voters who are more affectively polarised show a higher willingness to vote in future elections, even when these would be non-compulsory. We find that these positive effects are strongly significant and robust across both linguistic regions. Ideological polarisation, the way in which parties are perceived to be ideologically distinct from each other, only plays a marginal role (consistent positive effects, which are statistically insignificant in most of the elections under study).
  Our findings make a threefold contribution. First, we contribute to the literature on the correlates of voter turnout (Frank & Martínez i Coma, 2023; Smets & Van Ham, 2013), which has extensively studied how polarisation contributes to turnout (Béjar et al., 2020; Dalton, 2008; Harteveld & Wagner, 2022; Hobolt & Hoerner, 2020; Wessels & Schmitt, 2008) but often overlooked this relationship in compulsory voting systems. In these systems, we show that affective polarisation matters. The mechanism that drives turnout through affective polarisation applies in a similar way as in non-compulsory voting systems. At the same time, the effect of ideological polarisation appears to be absent in such a context. Additional research in other countries with compulsory voting is needed to uphold these mechanisms even further. Second, we add to the literature on compulsory voting (Birch, 2009; Singh, 2021), which often remains theoretical in nature and has mainly analysed the common sociodemographic turnout gaps when compulsory voting is lifted rather than the effect of polarisation (Gallego, 2010; Lijphart, 1997; Singh, 2015; Söderlund et al., 2011). Finally, we contribute to the booming literature on the political consequences of affective polarisation in multiparty systems (Harteveld & Wagner, 2022; Torcal & Carty, 2022; Wagner, 2021; Ward & Tavits, 2019) and show that regardless of its often-argued negative consequences it actually has the potential to foster political participation.
  More broadly, our findings imply that if compulsory voting is replaced by voluntary voting, Belgium might witness a ‘polarisation participation gap’, at least in the short term. Less polarised voters are less likely to vote, or they might abstain for good. This could impact the way citizens engage in political discussions or other types of political behaviour. Furthermore, given that political parties should reengage in mobilising the electorate when voluntary voting is put in place, they might resort to polarising strategies. While ideological polarisation is arguably one of the core duties of political parties, this might be unlikely to move citizens to vote. Instead, polarising the electorate in more affective ways could be more fruitful, but, potentially, it also has severe negative consequences for the democratic system. Therefore, if parties resort to increased negative and uncivil rhetoric or actions to mobilise the electorate, lifting compulsory voting might actually harm democracies on the long term.

• Data availability

  The replication material can be found at OSF via the following link: https://osf.io/8uqys/?view_only=7ff818b7f54e42a788995bea4ed82df6

References

• Abramowitz, A. I., & Saunders, K. L. (2008). Is Polarization a Myth? The Journal of Politics, 70(2), 542-555. https://doi.org/10.1017/S0022381608080493.

• Ahn, C., & Mutz, D. C. (2023). The Effects of Polarized Evaluations on Political Participation: Does Hating the Other Side Motivate Voters? Public Opinion Quarterly, 87(2), 243-266. https://doi.org/10.1093/poq/nfad012.

• Algara, C., & Zur, R. (2023). The Downsian Roots of Affective Polarization. Electoral Studies, 82, 102581. https://doi.org/10.1016/j.electstud.2023.102581.

• Bankert, A. (2021). Negative and Positive Partisanship in the 2016 U.S. Presidential Elections. Political Behavior, 43(4), 1467-1485. https://doi.org/10.1007/s11109-020-09599-1.

• Béjar, S., Moraes, J. A., & López-Cariboni, S. (2020). Elite Polarization and Voting Turnout in Latin America, 1993–2010. Journal of Elections, Public Opinion and Parties, 30(1), 1–21. https://doi.org/10.1080/17457289.2018.1545775.

• Bettarelli, L., Close, C., & Van Haute, E. (2022). Is Protest Only Negative? Examining the Effect of Emotions and Affective Polarization on Protest Behaviour. Politics and Governance, 10(4). https://doi.org/10.17645/pag.v10i4.5665.

• Bhatti, Y., Hansen, K. M., & Wass, H. (2012). The Relationship between Age and Turnout: A Roller-Coaster Ride. Electoral Studies, 31(3), 588-593. https://doi.org/10.1016/j.electstud.2012.05.007.

• Birch, S. (2009). Full Participation: A Comparative Study of Compulsory Voting. Manchester University Press.

• Blais, A. (2006). What Affects Voter Turnout? Annual Review of Political Science, 9(1), 111-125. https://doi.org/10.1146/annurev.polisci.9.070204.105121.

• Blais, A., & Carty, R. K. (1990). Does Proportional Representation Foster Voter Turnout? European Journal of Political Research, 18(2), 167-181. https://doi.org/10.1111/j.1475-6765.1990.tb00227.x.

• Blais, A., & Dobrzynska, A. (1998). Turnout in Electoral Democracies. European Journal of Political Research, 33(2), 239-261. https://doi.org/10.1111/1475-6765.00382.

• Boonen, J. (2019). Learning Who Not to Vote for: The Role of Parental Socialization in the Development of Negative Partisanship. Electoral Studies, 59, 109-119. https://doi.org/10.1016/j.electstud.2019.04.001.

• Chapman, E. B. (2019). The Distinctive Value of Elections and the Case for Compulsory Voting. American Journal of Political Science, 63(1), 101-112. https://doi.org/10.1111/ajps.12393.

• Crepaz, M. M. L. (1990). The Impact of Party Polarization and Postmaterialism on Voter Turnout. European Journal of Political Research, 18(2), 183-205. https://doi.org/10.1111/j.1475-6765.1990.tb00228.x.

• Dahl, R. A. (1971). Polyarchy: Participation and Opposition. Yale University Press.

• Dalton, R. J. (2008). The Quantity and the Quality of Party Systems: Party System Polarization, Its Measurement, and Its Consequences. Comparative Political Studies, 41(7), 899-920. https://doi.org/10.1177/0010414008315860

• Dassonneville, R., Barbosa, T., Blais, A., McAllister, I., & Turgeon, M. (2023). Citizens under Compulsory Voting: A Three-Country Study (1st ed.). Cambridge University Press. https://doi.org/10.1017/9781009071116.

• Dassonneville, R., Feitosa, F., Hooghe, M., Lau, R. R., & Stiers, D. (2019). Compulsory Voting Rules, Reluctant Voters and Ideological Proximity Voting. Political Behavior, 41(1), 209-230. https://doi.org/10.1007/s11109-018-9448-6.

• de Kadt, D. (2017). Voting Then, Voting Now: The Long-Term Consequences of Participation in South Africa’s First Democratic Election. The Journal of Politics, 79(2), 670-687. https://doi.org/10.1086/690065.

• Denny, K., & Doyle, O. (2009). Does Voting History Matter? Analysing Persistence in Turnout. American Journal of Political Science, 53(1), 17-35. https://doi.org/10.1111/j.1540-5907.2008.00355.x.

• Deschouwer, K., Delwit, P., Hooghe, M., Lefevere, J., Rihoux, B., & Walgrave, S. (2014). PartiRep II Longitudinal Electoral Survey. Data Archiving and Networked Services. https://doi.org/10.17026/dans-xwm-mq42.

• Deschouwer, K., Delwit, P., Hooghe, M., Lefevere, J., & Walgrave, S. (2009). PartiRep I Longitudinal Electoral Survey. Data Archiving and Networked Services. https://doi.org/10.17026/dans-2c7-pyzw.

• De Winter, L. (2019). Government Coalitions as a Reflection of National Politics: The Complex Case of Belgium. In M. Evans (Ed.), Coalition Government as a Reflection of a Nation’s Politics and Society (pp. 64-85). Routledge.

• Dinas, E. (2012). The Formation of Voting Habits. Journal of Elections, Public Opinion and Parties, 22(4), 431-456. https://doi.org/10.1080/17457289.2012.718280.

• Downs, A. (1957). An Economic Theory of Democracy. HarperCollins.

• Ellger, F. (2023). The Mobilizing Effect of Party System Polarization. Evidence from Europe. Comparative Political Studies, 57(8), 00104140231194059. https://doi.org/10.1177/00104140231194059.

• Enders, A. M., & Armaly, M. T. (2019). The Differential Effects of Actual and Perceived Polarization. Political Behavior, 41(3), 815-839. https://doi.org/10.1007/s11109-018-9476-2.

• Engelen, B. (2005). Een dam tegen het leeglopen van de democratie: Pleidooi voor het behoud van de opkomstplicht. Ethiek en Maatschappij, 8(2), 49-63.

• Feitosa, F., Blais, A., & Dassonneville, R. (2020). Does Compulsory Voting Foster Civic Duty to Vote? Election Law Journal: Rules, Politics, and Policy, 19(1), 19-44. https://doi.org/10.1089/elj.2018.0539.

• Frank, R. W., & Martínez i Coma, F. (2023). Correlates of Voter Turnout. Political Behavior, 45(2), 607-633. https://doi.org/10.1007/s11109-021-09720-y.

• Franklin, M. N. (2004). Voter Turnout and the Dynamics of Electoral Competition in Established Democracies since 1945. Cambridge University Press.

• Frognier, A.-P., Bol, D., & Swyngedouw, M. (2011). Une démocratie multipartisane. Flandre - Wallonie: 20 ans d’analyse des comportements électoraux. In A. van Busekist (Ed.), Singulière Belgique (pp. 135-145). Fayard.

• Gallego, A. (2010). Understanding Unequal Turnout: Education and Voting in Comparative Perspective. Electoral Studies, 29(2), 239-248. https://doi.org/10.1016/j.electstud.2009.11.002.

• Garzia, D., Ferreira da Silva, F., & Maye, S. (2023). Affective Polarization in Comparative and Longitudinal Perspective. Public Opinion Quarterly, nfad004. https://doi.org/10.1093/poq/nfad004.

• Gidron, N., Sheffer, L., & Mor, G. (2022). Validating the Feeling Thermometer as a Measure of Partisan Affect in Multi-Party Systems. Electoral Studies, 80, 102542. https://doi.org/10.1016/j.electstud.2022.102542.

• Gilens, M. (2012). Affluence and Influence: Economic Inequality and Political Power in America. Princeton University Press.

• Gordon, S. B., & Segura, G. M. (1997). Cross-National Variation in the Political Sophistication of Individuals: Capability or Choice? The Journal of Politics, 59(1), 126-147. https://doi.org/10.2307/2998218.

• Harteveld, E. (2021). Fragmented Foes: Affective Polarization in the Multiparty Context of the Netherlands. Electoral Studies, 71, 102332. https://doi.org/10.1016/j.electstud.2021.102332.

• Harteveld, E., & Wagner, M. (2022). Does Affective Polarisation Increase Turnout? Evidence from Germany, The Netherlands and Spain. West European Politics, 46(4), 1-28. https://doi.org/10.1080/01402382.2022.2087395.

• Hetherington, M. J. (2009). Review Article: Putting Polarization in Perspective. British Journal of Political Science, 39(2), 413-448. https://doi.org/10.1017/S0007123408000501.

• Hobolt, S. B., & Hoerner, J. M. (2020). The Mobilising Effect of Political Choice. European Journal of Political Research, 59(2), 229-247. https://doi.org/10.1111/1475-6765.12353.

• Hobolt, S. B., Leeper, T. J., & Tilley, J. (2021). Divided by the Vote: Affective Polarization in the Wake of the Brexit Referendum. British Journal of Political Science, 51(4), 1476-1493. https://doi.org/10.1017/S0007123420000125.

• Hooghe, M., & Deschouwer, K. (2011). Veto Players and Electoral Reform in Belgium. West European Politics, 34(3), 626-643. https://doi.org/10.1080/01402382.2011.555987.

• Huber, G. A., & Malhotra, N. (2017). Political Homophily in Social Relationships: Evidence from Online Dating Behavior. The Journal of Politics, 79(1).

• Huddy, L., Mason, L., & Aarøe, L. (2015). Expressive Partisanship: Campaign Involvement, Political Emotion, and Partisan Identity. American Political Science Review, 109(1), 1-17. https://doi.org/10.1017/S0003055414000604.

• Inglehart, R., & Norris, P. (2003). Rising Tide: Gender Equality and Cultural Change around the World. Cambridge University Press.

• Iyengar, S., Lelkes, Y., Levendusky, M., Malhotra, N., & Westwood, S. J. (2019). The Origins and Consequences of Affective Polarization in the United States. Annual Review of Political Science, 22(1), 129-146. https://doi.org/10.1146/annurev-polisci-051117-073034.

• Iyengar, S., Sood, G., & Lelkes, Y. (2012). Affect, Not Ideology. Public Opinion Quarterly, 76(3), 405-431. https://doi.org/10.1093/poq/nfs038.

• Iyengar, S., & Westwood, S. J. (2015). Fear and Loathing across Party Lines: New Evidence on Group Polarization. American Journal of Political Science, 59(3), 690-707. https://doi.org/10.1111/ajps.12152.

• Jackman, R. W. (1987). Political Institutions and Voter Turnout in the Industrial Democracies. American Political Science Review, 81(2), 405-423. https://doi.org/10.2307/1961959.

• Jackman, R. W., & Miller, R. A. (1995). Voter Turnout in the Industrial Democracies during the 1980s. Comparative Political Studies, 27(4), 467-492. https://doi.org/10.1177/0010414095027004001.

• Jackman, S. (1999). Non-compulsory Voting in Australia?: What Surveys Can (and Can’t) Tell Us. Electoral Studies, 18(1), 29-48. https://doi.org/10.1016/S0261-3794(98)00040-7.

• Janssen, L. (2023). Sweet Victory, Bitter Defeat: The Amplifying Effects of Affective and Perceived Ideological Polarization on the Winner–Loser Gap in Political Support. European Journal of Political Research, 63(2), 455-477. https://doi.org/10.1111/1475-6765.12625.

• Kingzette, J. (2021). Who Do You Loathe? Feelings toward Politicians vs. Ordinary People in the Opposing Party. Journal of Experimental Political Science, 8(1), 75-84. https://doi.org/10.1017/XPS.2020.9.

• Kingzette, J., Druckman, J. N., Klar, S., Krupnikov, Y., Levendusky, M., & Ryan, J. B. (2021). How Affective Polarization Undermines Support for Democratic Norms. Public Opinion Quarterly, 85(2), 663-677. https://doi.org/10.1093/poq/nfab029.

• Kostelka, F., & Blais, A. (2021). The Generational and Institutional Sources of the Global Decline in Voter Turnout. World Politics, 73(4), 629-667. https://doi.org/10.1017/S0043887121000149.

• Kostelka, F., Blais, A., & Gidengil, E. (2019). Has the Gender Gap in Voter Turnout Really Disappeared? West European Politics, 42(3), 437-463. https://doi.org/10.1080/01402382.2018.1504486.

• Kostelka, F., Singh, S. P., & Blais, A. (2022). Is Compulsory Voting a Solution to Low and Declining Turnout? Cross-National Evidence since 1945. Political Science Research and Methods, 12(1), 1-18. https://doi.org/10.1017/psrm.2022.57.

• Kriesi, H. (2014). The Populist Challenge. West European Politics, 37(2), 361-378. https://doi.org/10.1080/01402382.2014.887879.

• Kużelewska, E. (2016). Compulsory Voting in Belgium. A Few Remarks on Mandatory Voting. Białostockie Studia Prawnicze, 20, 37-51. https://doi.org/10.15290/bsp.2016.20A.en.03.

• Lachat, R. (2008). The Impact of Party Polarization on Ideological Voting. Electoral Studies, 27(4), 687-698. https://doi.org/10.1016/j.electstud.2008.06.002.

• Lee, A. H.-Y. (2022). Social Trust in Polarized Times: How Perceptions of Political Polarization Affect Americans’ Trust in Each Other. Political Behavior, 44(3), 1533-1554. https://doi.org/10.1007/s11109-022-09787-1.

• Lijphart, A. (1997). Unequal Participation: Democracy’s Unresolved Dilemma Presidential Address, American Political Science Association, 1996. American Political Science Review, 91(1), 1-14. https://doi.org/10.2307/2952255.

• Luskin, R. C. (1990). Explaining Political Sophistication. Political Behavior, 12(4), 331-361. https://doi.org/10.1007/BF00992793

• Mackerras, M., & McAllister, I. (1999). Compulsory Voting, Party Stability and Electoral Advantage in Australia. Electoral Studies, 18(2), 217-233. https://doi.org/10.1016/S0261-3794(98)00047-X.

• Martherus, J. L., Martinez, A. G., Piff, P. K., & Theodoridis, A. G. (2021). Party Animals? Extreme Partisan Polarization and Dehumanization. Political Behavior, 43(2), 517-540. https://doi.org/10.1007/s11109-019-09559-4.

• Mayer, S. J., & Russo, L. (2024). What One Is Not: A New Scale to Measure Negative Party Identity in Multiparty Systems. Quality & Quantity, 58(3), 2887-2906. https://doi.org/10.1007/s11135-023-01793-7.

• Michel, E., Feitosa, F., Lefevere, J., Pilet, J.-B., van Erkel, P., & van Haute, E. (2023). Studying Dimensions of Representation: Introducing the Belgian RepResent Panel (2019-2021). European Political Science, 23, 199-217. https://doi.org/10.1057/s41304-023-00430-z.

• Moral, M. (2017). The Bipolar Voter: On the Effects of Actual and Perceived Party Polarization on Voter Turnout in European Multiparty Democracies. Political Behavior, 39(4), 935-965. https://doi.org/10.1007/s11109-016-9386-0.

• Peters, Y., & Ensink, S. J. (2015). Differential Responsiveness in Europe: The Effects of Preference Difference and Electoral Participation. West European Politics, 38(3), 577-600. https://doi.org/10.1080/01402382.2014.973260.

• Phillips, J. B. (2024). Affective Polarization and Habits of Political Participation. Electoral Studies, 87, 102733. https://doi.org/10.1016/j.electstud.2023.102733.

• Pitkin, H. F. (1967). The Concept of Representation. University of California Press.

• Plutzer, E. (2002). Becoming a Habitual Voter: Inertia, Resources, and Growth in Young Adulthood. American Political Science Review, 96(1), 41-56. https://doi.org/10.1017/S0003055402004227.

• Powell, G. B. (1982). Contemporary Democracies: Participation, Stability and Violence. Harvard University Press.

• Quintelier, E., Hooghe, M., & Marien, S. (2011). The Effect of Compulsory Voting on Turnout Stratification Patterns: A Cross-national Analysis. International Political Science Review, 32(4), 396-416. https://doi.org/10.1177/0192512110382016.

• Radcliff, B. (1992). The Welfare State, Turnout, and the Economy: A Comparative Analysis. American Political Science Review, 86(2), 444-454. https://doi.org/10.2307/1964232.

• Radcliff, B., & Davis, P. (2000). Labor Organization and Electoral Participation in Industrial Democracies. American Journal of Political Science, 44(1), 132. https://doi.org/10.2307/2669299.

• Reiljan, A. (2020). ‘Fear and Loathing across Party Lines’ (also) in Europe: Affective Polarisation in European Party Systems. European Journal of Political Research, 59(2), 376-396. https://doi.org/10.1111/1475-6765.12351.

• Robson, W. A. (1923). Compulsory Voting. Political Science Quarterly, 38(4), 569-577. https://doi.org/10.2307/2142479.

• Rubin, D. B. (1987). Multiple Imputation for Nonresponse in Surveys. Wiley.

• Sartori, G. (1976). Parties and Party Systems: A Framework for Analysis. Cambridge University Press.

• Schakel, W. (2021). Unequal Policy Responsiveness in the Netherlands. Socio-Economic Review, 19(1), 37-57. https://doi.org/10.1093/ser/mwz018.

• Schattschneider, E. E. (1960). The Semisovereign People: A Realist’s View of Democracy in America. Holt, Rinehart and Winston.

• Selb, P., & Lachat, R. (2009). The More, the Better? Counterfactual Evidence on the Effect of Compulsory Voting on the Consistency of Party Choice. European Journal of Political Research, 48(5), 573-597. https://doi.org/10.1111/j.1475-6765.2009.01834.x.

• Serani, D. (2022). In-Party Like, Out-Party Dislike and Propensity to Vote in Spain. South European Society and Politics, 27(1), 125-146. https://doi.org/10.1080/13608746.2022.2047541.

• Singh, S. P. (2015). Compulsory Voting and the Turnout Decision Calculus. Political Studies, 63(3), 548-568. https://doi.org/10.1111/1467-9248.12117.

• Singh, S. P. (2021). Beyond Turnout: How Compulsory Voting Shapes Citizens and Political Parties. Oxford University Press.

• Singh, S. P. (2023). Compulsory Voting Diminishes the Relationship between Winning and Satisfaction with Democracy. The Journal of Politics, 85(3), 1079-1093. https://doi.org/10.1086/723815.

• Smets, K., & Van Ham, C. (2013). The Embarrassment of Riches? A Meta-Analysis of Individual-Level Research on Voter Turnout. Electoral Studies, 32(2), 344-359. https://doi.org/10.1016/j.electstud.2012.12.006.

• Söderlund, P., Wass, H., & Blais, A. (2011). The Impact of Motivational and Contextual Factors on Turnout in First- and Second-Order Elections. Electoral Studies, 30(4), 689-699. https://doi.org/10.1016/j.electstud.2011.06.013.

• Spoon, J.-J., & Klüver, H. (2019). Party Convergence and Vote Switching: Explaining Mainstream Party Decline across Europe. European Journal of Political Research, 58(4), 1021-1042. https://doi.org/10.1111/1475-6765.12331.

• Stokes, D. E. (1963). Spatial Models of Party Competition. American Political Science Review, 57(2), 368-377. https://doi.org/10.2307/1952828.

• Tajfel, H. (1970). Experiments in Intergroup Discrimination. Scientific American, 223(5), 96-103.

• Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. (1971). Social Categorization and Intergroup Behaviour. European Journal of Social Psychology, 1(2), 149-178. https://doi.org/10.1002/ejsp.2420010202.

• Tavits, M. (2006). Party System Change: Testing a Model of New Party Entry. Party Politics, 12(1), 99-119. https://doi.org/10.1177/1354068806059346.

• Torcal, M., & Carty, E. (2022). Partisan Sentiments and Political Trust: A Longitudinal Study of Spain. South European Society and Politics, 27(1), 171-196. https://doi.org/10.1080/13608746.2022.2047555.

• van Erkel, P. F. A., & Turkenburg, E. (2022). Delving into the Divide: How Ideological Differences Fuel Out-Party Hostility in a Multi-Party Context. European Political Science Review, 14(3), 386-402. https://doi.org/10.1017/S1755773922000121.

• Vlaamse Overheid. (2021). Voorontwerp van decreet tot wijziging van diverse decreten wat betreft versterking van de lokale democratie. https://beslissingenvlaamseregering.vlaanderen.be/document-view/6087B9E6364ED90008000954.

• Wagner, M. (2021). Affective Polarization in Multiparty Systems. Electoral Studies, 69, 102199. https://doi.org/10.1016/j.electstud.2020.102199.

• Wagner, M. (2024). Affective Polarization in Europe. European Political Science Review, 16(3), 1-15. https://doi.org/10.1017/S1755773923000383.

• Walgrave, S., Celis, K., Deschouwer, K., Marien, S., Pilet, J.-B., Rihoux, B., van Haute, E., van Ingelgom, V., Baudewyns, P., van Erkel, P., Jennart, I., & Lefevere, J. (2022). RepResent Longitudinal Survey 2019—2021. Data Archiving and Networked Services. https://doi.org/10.17026/dans-xf5-djem.

• Ward, D. G., & Tavits, M. (2019). How Partisan Affect Shapes Citizens’ Perception of the Political World. Electoral Studies, 60, 102045. https://doi.org/10.1016/j.electstud.2019.04.009.

• Weber, T. (2021). Negative Voting and Party Polarization: A Classic Tragedy. Electoral Studies, 71, 102335. https://doi.org/10.1016/j.electstud.2021.102335.

• Wessels, B., & Schmitt, H. (2008). Meaningful Choices, Political Supply, and Institutional Effectiveness. Electoral Studies, 27(1), 19-30. https://doi.org/10.1016/j.electstud.2007.11.010.

• Westwood, S. J., Iyengar, S., Walgrave, S., Leonisio, R., Miller, L., & Strijbis, O. (2018). The Tie that Divides: Cross-National Evidence of the Primacy of Partyism. European Journal of Political Research, 57(2), 333-354. https://doi.org/10.1111/1475-6765.12228.

Noten

• 1 Panel surveys of 2009 (Deschouwer et al., 2009); panel of 2014 (Deschouwer et al., 2014); panel of 2019 (Walgrave et al., 2022).

• 2 Specifically, the question asks respondents: “If voting for parliament was no longer obligatory in Belgium, would you then always, generally, sometimes, or never vote or don’t you know for certain?”

• 3 The percentage of voters who gave the ‘don’t know’ option is only somewhat considerable in 1991, with around 10% in both regions. In the three following election studies, the percentages are around 5 or even lower.

• 4 Note that, for reasons of readability, we do not include the employment dummies. Generally, we find that citizens who are unemployed are less likely to vote in future non-compulsory elections than employed citizens or citizens who are not part of the labour force. Also note that we cannot control for political interest in 1991 because the variable was not included in the survey.