Compelled Turnout and Democratic Turnout: Why They Are Different

Abstract

One strategy in defence of compulsory voting is based on what I call the non-instrumental value of high turnout: the idea that almost-universal participation in elections is valuable per se. This article argues that we do not have democratic reasons to value compelled turnout. First, thanks to an original analysis of the practice of voting, I identify three constitutive rules that make the physical acts of marking and casting a ballot count as proper voting. This preliminary analysis serves to illuminate the fact that the act of voting has democratic value if it is performed in a free and reason-responsive way. Second, I identify political equality and popular control as democratic values that high turnout expresses. Finally, the article rejects the non-instrumental case for compulsory voting because it cannot ensure that people vote in a reason-responsive way and, if they do not, high turnout lacks democratic value.

Introduction

As the history of modern democracy has been dominated by struggles for universal suffrage, the right to vote is its distinctive feature (Ceva and Ottonelli, 2021) and the most widely exercised form of participation by citizens (Verba et al., 1995). Yet, voter turnout has been steeply declining for decades, making electoral participation an unequal business in which primarily age, education and income determine who shows up at the voting booth (Cancela and Geys, 2016; Kostelka and Blais, 2021; Smets and Van Ham, 2013). This decrease has been less dramatic in countries where voting is mandatory, which has led various scholars in the past 20 years to advocate for compulsory voting (henceforth CV) as an effective remedy for the plague of declining turnout (Hill, 2002b, 2006, 2014; Lijphart, 1997; Birch, 2009; Chapman, 2019; Elliott, 2017; Engelen, 2007; Lacroix, 2007; Umbers, 2020).

Among numerous defences of CV, one strand revolves around high turnout as a democratic good. Intuitively, if democracy requires that all the people rule, there seems to be something inherently democratic about having high-turnout elections. Famously, Arend Lijphart (1997) asserts that ‘the democratic goal should be not just universal suffrage but universal or near-universal turnout’ (2). Similarly, Lisa Hill (2014: 166) claims that ‘when turnout is high and socially even, not only are elections more procedurally legitimate, but the governments they deliver are also more able to be “of the people, by the people, for the people”’. In her recent article, Emilee Chapman (2019: 103) claims that ‘when characterized by nearly universal participation’, elections ‘command the attention of the general public and manifest the equal political authority of all citizens’.

The democratic value of high turnout can be fleshed out either in instrumental or non-instrumental terms. Instrumentally, high turnout is valuable for what it supports, such as more egalitarian political outcomes (Bechtel et al., 2016; Birch, 2009; Fowler, 2013; Lijphart, 1997; cf. Miller and Dassonneville, 2016) and more responsive politicians (Chapman, 2019; Engelen, 2007; Hill, 2002b). By contrast, a non-instrumental view values high turnout as a democratic good regardless of its impact on how people vote and how elected officeholders respond to their votes. These two cases are usually endorsed in conjunction, but they are different. While advocates often fail to sharply distinguish between them, that high turnout is relevant regardless of what people vote for is a claim that some explicitly endorse (Engelen, 2007: 28; Hill, 2002a: 93), as is the idea that high turnout enhances democratic legitimacy by expressing citizens’ equal and collective authority (Chapman, 2019: 103–104). ¹

This article focusses on the non-instrumental case for CV and aims to show that, unless further qualified, it fails. Critics have already objected that high turnout is not a public good (Lever, 2009a: 70) and that it is not democratically desirable (Saunders, 2012: 307), but I do not deny that high turnout may be democratically valuable. I only contest the assertion that CV is the appropriate way to achieve the kind of high turnout required by the democratic ideal. In other words, I argue that if we consider the democratic principles invoked to defend the non-instrumental value of high turnout, we realise that high turnout is good for democracy only under certain conditions. CV must thus satisfy these conditions if it is to achieve the right kind of high turnout.

The article expands the recent literature on the ethics of voting in three ways.

To my knowledge, it offers the first analysis of voting as a social practice defined by three constitutive rules. ² Voting is a practice that takes place when (1) a group must make a collective decision from a slate of options, (2) each member expresses their choice (3) in a way that influences said decision. While this conceptual analysis may appear trivial, it helps us rule out what may be considered regular instances of voting that nevertheless fail to meet these criteria.

Second, this article develops a normative account of democratic voting that defines two conditions for citizens’ act of voting to have democratic value. Voting is not always a democratic practice; voters may not be equal or free to vote as they like. I turn to a fundamental understanding of democracy as collective self-rule (Lafont, 2019) to draw out two principles that voting, as all other democratic institutions, ought to fulfil: political equality and popular sovereignty. These principles not only establish a democratic institutional framework but also identify how citizens should act within that framework. As a democratic practice, voting must be performed in a certain way: freely and reason-responsively.

Finally, I apply this conceptual and normative analysis to the non-instrumental case for CV and conclude that CV fails because high turnout possesses democratic value only if it satisfies the conditions of freedom and reason-responsiveness. Compelling people to attend elections is insufficient to this end and raises more concerns.

The article unfolds as follows. The next section investigates the three constitutive rules of voting, while the third explains when voting has democratic value. The fourth section is divided into two parts: one reconstructs the non-instrumental argument offered by Chapman (2019), which focusses on the same democratic principles of political equality and popular control I address in the previous section, and the other argues that compulsory turnout is not the kind that we value in virtue of these principles.

What Counts as Voting

Political science teaches us that voting is a procedure for making collective decisions. Since this seems obvious, scientists tend to focus on all the irrationalities and paradoxes attached to voting, while theorists often prefer to elaborate their views of democracy around the notion of deliberation, as opposed to voting (Mackie, 2018).

The act of voting, though, is only seemingly simple. In fact, voting is both a decision-making procedure and a social practice employed in a variety of circumstances (Ceva and Ottonelli, 2021: 5). Among friends, voting can be used to pick a restaurant for dinner. Companies generally have board members vote on high-stakes decisions, and juries reach verdicts by voting. In politics, voting is meant to settle matters of common concern, from important policy issues decided by referendums to the election of officeholders, not to mention laws voted on by parliamentary representatives. While I will sometimes refer to private interactions like the restaurant case, what follows will mostly track voting in democratic elections.

As a social practice, voting attributes a particular normative role to individuals engaging in it and involves a set of rules that define it (Ceva and Ottonelli, 2021; Rawls, 1955). Following John Searle (1995), we can distinguish between regulative rules, which regulate activities that can be defined independently of the rules, and constitutive rules, which not only regulate, but constitute given activities. These activities would not exist without these rules. Most such rules take the following ‘count-as’ form: ‘X counts as Y in context C’. Think of the game of volleyball: the rules governing what players may or may not do are constitutive insofar as there is no volleyball outside those rules. If after the other team’s serve, I catch and hold the ball, I am not bad at playing volleyball; I am not playing volleyball at all. Similarly, outside the constitutive rules that define what counts as voting, there is no voting at all (Ceva and Ottonelli, 2021: 4).

Like other social practices, voting is ontologically subjective and epistemologically objective: it does not exist independently of our mind and intentions but is independent from what each individual thinks. One of Searle’s famous examples is money: ontologically it is just a piece of paper, but it counts as a means of exchange because humans view it as such. Furthermore, if I stop looking at money that way, it is not deprived of its social meaning. Similarly, the act of marking a ballot counts as voting because us human beings view it as such, and my personal denial of the practice does not deprive it of its social meaning. In other words, as a social practice, voting has its own logic and cannot be reduced to what any individual think of it.

Looking at voting through these lenses means that this investigation is neither descriptive nor explanatory. I do not explore the motivation with which people perform this act; nor am I offering a normative theory of why or how people should vote. ³ This analysis is partly conceptual and partly normative. Conceptually, I aim to offer a rational reconstruction of the logic of voting as a social practice and to illustrate what I take to be its three constitutive rules. Normatively, I suggest that we generally attach democratic value to the act of voting only if at least two conditions are met. Thus, I am not describing how most people vote or how they should vote, but rather how we ought to see voting to make sense of it as a democratic practice. The fact that not all citizens in contemporary democracies vote in the way I depict, therefore, does not invalidate the claims that I defend here.

Let us start with the first rule: more than one option and more than one voter are necessary for voting to take place. Voting is a collective activity: it is not something one individual can do alone (Birch, 2018: 11). Voting also requires that a group of people pick one among several available options. If I decide on my own which restaurant to pick for dinner, I am not voting for it; nor are my friends voting for a restaurant if only one place is available. The context in which people resort to this specific practice is that in which a group of people must make a choice. To the extent that available options are only illusory, the act of voting amounts to a farce, which undemocratic countries sometimes deploy in search of popular legitimation or international respectability. Clearly, this is not the case in democratic systems, where at least two options must not only be present, but also have some chances of electoral success (Przeworski, 1991).

One may object that citizens are still expected to take part in local elections that have only a single candidate running for office. ⁴ The reason, so the objection goes, is that voting authorises a person to take up their role as public official and the presence of alternatives is not required for proper voting to occur. This view of authorisation as essential to voting does not seem to apply to voting on specific decisions, be they policies in referendums or restaurant choices among friends. In these cases, the presence of an alternative is required for people to vote; yet, these are all instances of the same social practice. Moreover, even uncontested elections offer an important alternative: citizens can either vote for that candidate or cast a blank ballot. I think we should view uncontested elections as referendums on one candidate, whose democratic mandate will be different depending on the share of votes received. While the logic of voting would have the candidate elected only if they received at least a majority of votes, considerations of governance and stability put the only candidate in office regardless. Furthermore, if elections routinely presented one single candidate as a compelled choice, I suspect we would not only consider them undemocratic, but also quickly lose the point of holding a vote at all. In other words, the authorising function of voting is only made possible by the fact that citizens have a way to make their choice for or against known.

The second constitutive rule of voting concerns its expressive dimension: by voting we express a choice. When I raise my hand for a Japanese restaurant, I express my preference for that type of meal; when I vote for a candidate, I express my political views, attitudes and beliefs. The reasons behind our vote are manifold, and we might not be aware of all of them. However, in all voting contexts, people are generally considered responsible for their vote, which is usually interpreted as a sign of support for a candidate. If I sleepwalked to a polling station and cast a ballot in my sleep, would we say that I am voting? Or, if you took a ballot and put a slice of ham on it before folding and casting it in the box (I have been a scrutineer, it happened!), would we understand what you did as voting? It does not seem so.

That people sometimes vote out of anger or group loyalty does not shake the underpinning logic of voting as a practice meant to give people the opportunity to express their support for one option. If after having voted on a restaurant I complain with you that I voted for pizza only as a joke because I thought sushi would easily win, you can rightfully remind me that I had my opportunity to have my preference known and counted. A feeling of regret, for instance, is not unusual when we realise we have voted for the ‘wrong’ candidate or failed to vote at all. After all, voting is not like sortition: while the latter serves to make collective decisions in an explicitly neutral way, the former is supposed to be sensitive to citizens’ choices and allow those who take part to select what they want. ⁵

Yet, voting cannot be reduced to simple endorsement. If I were only aiming to express my views, why should I do so by casting a ballot? Many other options, especially nowadays, seem more suited to publicly expressing one’s views. Social media make it easier and less costly to express one’s political beliefs while also preserving the anonymity that voting ensures. I could post my beliefs and attitudes on both specific candidates and the democratic system at large while comfortably sitting on my sofa.

What is distinctive about voting is that I must express my beliefs ‘specifically at the ballot box’ (Brennan and Sayre-McCord, 2015: 53), because in such a way my vote affects the electoral outcome. Clearly, voting is not the only means at our disposal to influence the selection of candidates. If one is sufficiently powerful and well-connected, there are much more efficient ways to do just that. But when I vote, I express my support by having my vote counted in the tally. Casting a ballot enables me to express my support for one option (Thompson, 2002: 22), while also contributing to tilting the balance in its favour (Goldman, 1999; Mackie, 2015; Tuck, 2008). ⁶ The third constitutive rule of voting thus stems from its instrumental dimension. Something counts as a vote to the extent that such an expression of choice is counted in the final tally and contributes to determining the result. Naturally the way each vote influences political outcomes depends on the type of election (presidential vs parliamentary) and the broader electoral system (single-member or multimember districts, etc.), but the common element of all voting procedures is that their outcome should be sensitive to each vote. ⁷

One may ask what this analysis tells us of invalid ballots. These votes are sometimes the product of mere apathy but often they are expressive: they may manifest voters’ attitudes towards the political process and available options, such as dissatisfaction, frustration, protest, and rage. What they may fail to express, however, is a clear choice among available options in a way that contributes to determining the result. Naturally, this depends on the context in which voting takes place. Take again the case of uncontested elections: a blank ballot in that case clearly expresses a choice against the only candidate and may affect the robustness of their democratic mandate. In Colombia, for instance, new elections are called when the blank vote wins (Taylor and Shugart, 2017). In this context then, blank votes are not only expressive, as they can be correctly interpreted as ‘not this candidate’ votes, but they may also have an impact on the outcome. Spoiled votes, on the contrary, are hardly seen as expression of a choice and never influence the electoral outcome in one way or another. Hence, they fail to respect the constitutive rules of a voting practice. ⁸

If this analysis of voting as a social practice is sound, then an action must satisfy at least three constitutive rules to count as voting. First, it must be performed in the context of a collective choice with more than one option available. Second, it must indicate a person’s choice: the expressive dimension. Third, it must affect the electoral outcome: the instrumental dimension. When in the context of a collective choice with more than one option available, individuals give a clear indication of their choice in a way that influences the result, they are voting.

Voting as a Democratic Practice

The three constitutive rules allow us to see when marking and casting a ballot count as voting. They do not tell us whether the practice has democratic value. The point and purpose of voting is to give people the opportunity to express their views in a way that affects collective decisions. This happens in various institutional settings, from oligarchies to company boards. Nothing in the concept of voting tells us how we should weigh each individual vote. The principle of ‘one person, one vote’, which comes as a recognition of citizens’ equal status as decision makers (Beckman, 2009), is generally considered essential to democratic voting but is not inherent to the practice of voting in general.

The ethics of voting typically involves ‘micro-democratic’ and ‘macro-democratic’ dimensions (Beerbohm, 2012: 17–21). While the former concerns issues of institutional design, the latter responds to an agent-relative question regarding what individuals qua citizens ought to do given a certain institutional framework. Insofar as providing a satisfying answer to macro-democratic issues does not settle all disputes concerning micro-democratic questions, it makes sense for democratic theory to investigate whether citizens have individual responsibilities. Furthermore, understanding what it means for individuals to participate in democratic procedures may provide valuable insights concerning how these procedures ought to be arranged. If political institutions ought to be assessed based on democratic principles, these same principles determine which type of citizen agency is valuable and how we should arrange our institutions so to as to foster it.

Let us start with the traditional understanding of democracy as collective self-government. As Cristina Lafont (2019: 17–33) argues, if democracy amounts to collective self-government, it is concerned with two important principles: political equality and popular control. The former principle reflects a concern with the distribution of political power and requires that all citizens are treated as equals (e.g. Christiano, 2008), while the latter rules out political alienation by ensuring that citizens’ relation to the content of political decisions is not one of blind deference or alienation (Pettit, 2012).

From the macro-democratic point of view, political equality is realised in a voting procedure insofar as the procedure treats all participants as equals, which may or may not require equal voting power. ⁹ Popular control is realised insofar as the outcome of a voting procedure is responsive to people’s choices expressed in their vote, so that the option selected is the one with the most votes. When the institutional features of a voting practice are such that citizens can view its outcomes as the product of their joint and equal authority, such a practice is democratic. ¹⁰

On the contrary, we need to ask what these democratic principles require at the level of micro-democratic theory. If a voting procedure realises political equality and popular control in its institutional design, is this sufficient for the practice to be democratically valuable regardless of how citizens act?

To answer this question, let us start with a couple of examples:

An Apparently Blessed Democracy (ABD): In this country, people always exercise their voting rights at elections and referendums. Voting procedure is egalitarian, inclusive and responsive to people’s votes. However, most citizens are under the sway of a secret criminal organisation that threatens their safety if they vote for candidates the organisation dislikes.

A Divinely Determined Democracy (DDD): In this country, people always exercise their voting rights at elections and referendums. Voting procedure is egalitarian, inclusive and responsive to people’s votes. However, since they conceive of elections as a means of ascertaining God’s will, citizens are used to voting by closing their eyes, drawing a mark on a random part of the ballot and casting it in the box.

Both cases show that a formally democratic voting procedure seems to lack true democratic value if voters cast their ballot in an unfree and/or random fashion. In the Apparently Blessed Democracy (ABD), a vote is an expression of choice and determines the result, but this choice has not been made freely because citizens must pay heed to the criminal organisation’s will. In the Divinely Determined Democracy (DDD), votes are free but random. If it is an expression of choice at all, it is a choice taken by chance, which lacks what we may call responsiveness to reasons. While in ABD, citizens’ votes are not theirs because they are subject to undue interference, in DDD, they are not theirs because they have not decided based on their values and interests but on a whim.

First, let me be clear regarding what this reason-responsiveness means. One can endorse a rich view of democracy according to which voting has epistemic value and as such it ought to be responsive to objectively good reasons (Cohen, 1986; Estlund, 2008), and accurate tracking of these reasons would be required for voting to have democratic value. This is not what I am claiming here. Less demandingly, I am looking for the minimal conditions a voting practice must fulfil for it to have democratic value according to the more ecumenical idea of democracy as collective self-rule. ¹¹

Second, both freedom and reason-responsiveness are required for popular control. If citizens must not be alienated from political decisions they are supposed to obey, then it is not enough that they are not subject to external interference as per the freedom requirement. They must also understand the voting outcome as sensitive to their reasoned choice rather than the product of chance.

Take the case of democratic voting at elections. Elections are meant to select candidates for political offices and hold current officeholders accountable (Thompson, 2002). It is hard to believe that officeholders will be responsive to citizens’ opinions and interests if the latter lack the opportunity to exercise control over them (Lafont, 2019; Pettit, 2012), and to exercise such control citizens must be able to steer the electoral process in the direction they want. They must select candidates who have their trust and share the same political objectives (Mansbridge, 2009), and this requires them to reflect on their values and interests to make the choice that is responsive to their subjective reasons to advance those values and interests.

That this is our common definition of what democratic voting involves is clear from the way we speak of it. We tend to ask for reasons for why and how people have voted. We may disagree with these reasons and discuss them with friends and relatives. We tend to attribute specific motivations to other people’s votes. We tend to assess and approve or question these motives. We do not look at voting as we would look at natural events; we do not simply ask how, but why, and hold people responsible for the way they voted. All these activities that surround voting would be pointless if we did not think that democratic voting required some form of reason-responsiveness.

Citizens’ voting thus acquires democratic value insofar as it respects the two conditions of freedom and reason-responsiveness, because these conditions are necessary for voting to express citizens’ collective self-rule.

Let us consider three objections. ¹² First, electoral phenomena like donkey votes happen. A ‘donkey vote’ is a vote in which the preferences recorded simply reflect the order in which candidates’ names appear on the ballot. Preferential-voting systems, like the Australian one, make it possible to detect donkey votes because of the apparently random ranking of political preferences. Why are ballots with random rankings considered valid votes if random voting lacks democratic quality? The problem is that there is no uncontroversial way to ascertain that the apparently random ranking is not the product of a voter’s reasoned choice. Donkey votes are considered valid then precisely because they offer a preference ranking that can be considered the expression of a person’s reason-responsive choice, even if an odd one. Reason-responsiveness after all concerns subjective reasons; what matters for a choice to have democratic value is that individuals display reflection and follow the reasons they take themselves to have, not reasons that are objectively good.

Second, random choices are not always incompatible with responsiveness to reasons. For example, if a voter is genuinely indifferent between available options, due to the even weight of reasons on all sides, random voting may appear as a reason-responsive choice. My reply is twofold. First, any person who routinely makes decisions at random can hardly be seen as in control of their own fate. While Buridan’s ass is a thought-provoking philosophical paradox that shows how sometimes a random choice is better than no choice at all, I do not think it is common in real life, and even less so in political life, where parties often contest elections with complex programmes that bundle together many issues towards which citizens are hardly indifferent. Second, there is a crucial difference between individual and collective decisions. While individually a random decision may sometimes be needed, this is not the case with collective choices. If I am indifferent between two candidates, surely other fellow citizens will not be. Rather than voting randomly, then, the reason-responsive choice is to abstain or cast a blank ballot exactly because I am indifferent to all options. Take the restaurant example: if I was indifferent between sushi and pizza, I should let other friends decide, instead of raising my hand for one on a whim. Unless I must break a tie, it would be hard to justify my casual choice for pizza to those friends who voted for sushi, especially if one is allergic to gluten. Similarly, I do not find myself in need of deciding at elections because there are other people who take part in the decision. If all options are equally in line with my views, I should let other people influence the process in the way most aligned with their values and interests. ¹³

Finally, one may dispute the idea that all citizens are supposed to vote freely and reason-responsively. This would involve a fallacy of composition because even if most individuals vote randomly, their random votes will cancel each other out in the aggregate and the outcome will only track reason-responsive votes. Even if most citizens voted randomly then, this would not affect the democratic quality of elections and referendums.

Although powerful, this objection violates the principle of political equality. While this principle, when we think of citizens’ opportunities to propose new options, is quite demanding and mostly unattained in existing democracies (Lessig, 2015; Sunstein, 1994), regarding voting procedures, it clearly requires that all citizens have the same opportunity to steer the voting process towards their preferred choice among options available. From the point of view of macro-democratic theory, this means that citizens’ equal political status be reflected in their voting power. From the point of view of micro-democratic theory, this requires that all citizens can see the voting process as equally sensitive to their values and interests. But this can happen only if their vote responds to their own reasons.

As already noted, actual citizens may fail to conform to the logic of democratic voting, but the underpinning justification for having democratic voting at all is that it realises political equality and popular control, that is, it allows all citizens to see democratic outcomes as theirs. If most citizens voted randomly, I think we would not consider those citizens as equally in control of the outcome as the members of the minority: the outcome would be sensitive to all citizens’ inputs, but only some inputs would reflect citizens’ reasoned opinions concerning their interests and values. This would be particularly troubling if said minority were identifiable by relevant social and economic ascriptive characteristics, with distinctive interests not shared by most of the citizenry (Gilens, 2012; Guinier, 1995). By contrast, it is at least also because we assume that all citizens can freely and reason-responsively engage in voting practices, such as elections and referendums, that we take these to have democratic value and realise citizens’ equal authority.

To sum up, voting is a social practice takes place (1) when a group must make a collective decision from a slate of options and (2) each member expresses their choice (3) in a way that influences said decision. From a top-down, institutional perspective, voting is democratic when the procedure is designed in such a way as to realise political equality and popular control. From a bottom-up, agential perspective, voting has democratic value when participants engage in this practice freely and reason-responsively. High turnout has democratic value non-instrumentally if and only if democratic voting, not just any voting, takes place.

Democratic and Compulsory Voting

The Non-Instrumental Case

The non-instrumental case for CV is a good example of how micro-democratic theory influences macro-democratic theory: because having all citizens vote is a good thing, we should arrange our institutions in such a way that voting is mandatory. Because the democratic good of high-turnout elections is explained in non-instrumental terms, this argument absolves CV advocates of the burden of explaining the good consequences of high turnout. Chapman (2019) is the first to explicitly defend the expressive value of high-turnout elections based on two non-instrumental reasons. ¹⁴

The first reason invokes the value of political equality by asserting that high turnout publicly demonstrates ‘the political involvement and formal political equality of all citizens’ (Chapman, 2019: 104). A similar argument is endorsed by Engelen (2007: 34) who claims that ‘compulsory voting sends a powerful signal to the people that each and every vote matters’ (see also Lijphart, 1997: 2).

Chapman’s (2019: 104) second non-instrumental reason appeals to popular control: elections are ‘distinctively valuable mechanisms of democratic control when they predictably involve the entire citizenry in the sanctioning process’ because they provide ‘an unambiguous reminder to public officials that they are accountable to all citizens, not just the most vocal and active’ (Chapman, 2019: 104).

Chapman’s (2019: 103) justification of CV is thus particularly compelling, not only because it is clearly non-instrumental, but also because it invokes the same democratic principles I have drawn from Lafont’s conception of democracy as collective self-rule. Accordingly, when all citizens participate in elections they can be seen as equal authors of the law.

The non-instrumental case thus comprises three premises and one conclusion:

(P1) High-turnout elections are non-instrumentally valuable for democracy.

(P2) They are valuable because they express citizens’ political equality and popular control.

(P3) CV is the best means to ensure high-turnout elections.

. . . CV is justified on democratic grounds.

The last premise concerning the efficacy of compulsion at raising turnout (P3) is beyond dispute: evidence shows that CV is the most effective means to ensure high turnout, at least when strictly enforced (Birch, 2009; Boyle, 2022; Singh, 2015). These findings are after all unsurprising. Since CV involves compelling people to attend elections, we must expect that turnout spikes under such regimes, especially if we compare compulsion to other institutional means, such as automatic registration or elections on weekends, that only facilitate voting. The non-instrumental case takes high turnout to be democratically valuable in a way that is independent from how citizens vote, thereby seemingly avoiding objections concerning the subordination of CV to a duty to vote well (Brennan, 2016; Lever, 2009a; Lever and Volacu, 2019; Volacu, 2020).

Why the Non-Instrumental Case Fails

The argument fails because popular control and political equality only justify P2 on the further condition that citizens vote freely and reason-responsively. If voting is not democratic, high turnout lacks democratic value. P2 is thus conditional on how the act of voting is performed by citizens. The question thus becomes: can compulsion bring about democratic voting?

If citizens were compelled to cast a formal and reason-responsive vote, it could. As has been noted, though, this would violate their freedom (Lever, 2009a; 2010; Saunders, 2010, 2012). This is not because, as some supporters have claimed, its enforcement would require eliminating the secret ballot (Chapman, 2019: 102; Hill, 2006: 222; cf. Elliott, 2017: 567). Rather, even if it was possible to compel people to cast a formal and reason-responsive vote, ¹⁵ that vote would lack democratic value because it would violate the first condition of freedom.

To be sure, while ABD shows a criminal organisation in action, CV is defended on the premise that it is enforced by presumptively legitimate states. Yet, the mechanism is not so different because forcing citizens to cast a formal vote deprives them of the option to abstain, thereby altering the set of available options. If any form of uncontrolled alteration interferes with individual freedom (Pettit, 2012: 51–56), then CV violates the freedom condition.

This is not a minor worry, as recognised by CV advocates, who in fact dismiss it by claiming that the term ‘compulsory voting’ is a misnomer because what is mandatory is turnout (Birch, 2009: fn. 1, 22; Chapman, 2019: 102; Elliott, 2017: 657; Engelen, 2007: 25; Hill, 2002b: 82; Lacroix, 2007: 193; Lijphart, 1997: 2). Compulsory turnout is thus compatible with the freedom condition.

The problem at this point is that compulsory turnout is not sufficient to bring about democratic voting, because it is compatible with invalid and random voting. The presumed benefits of having almost all citizens cast a formal vote cannot be attained by attendance, which is the only thing legally required (Lever, 2009b: 224). If the opportunity to cast informal votes is needed to preserve freedom of abstention, it is difficult to see how this opportunity provides something more than abstention itself, as some advocates claim (Engelen, 2009: 219; Hill, 2002b: 93). This is for two reasons.

First, if invalid voting is a ‘functional equivalent of abstention’ (Power and Roberts, 1995: 803), it seems bound to have the same hermeneutical uncertainties: citizens may do so because they reject all the options, because they are contented with all or because they are marginalised. While contextual social norms may sometimes provide clues to interpret the choice a blank ballot expresses, the correct interpretation is often contentious. ¹⁶ Formal votes are difficult to read too; voters can vote for a candidate because they appreciate their programme or because they hate all other candidates. Regardless of their reasons, though, the choice they express is clear.

The kind of choice that informal ballots express, by contrast, is often ambiguous. Engelen (2009: 219) claims that CV makes it possible to ‘distinguish between purely apathetic and apolitical voters on one hand (blank) and anti-political protest voters on the other hand (none of the above)’, but this is merely speculative. We have no reason to think that the distinction between apolitical and protest voters maps on to the distinction between blank and ‘none of the above’ votes. Inferring the former from the latter is unwarranted. Conjecturing the reasons for informal voting is not different from conjecturing the reasons for lower turnout (Lever, 2010: 911). The expressive dimension of voting is thus violated when social norms regulating the interpretation of invalid voting are not in place.

Second, with few exceptions like Colombia, casting informal ballots fails to satisfy the instrumental dimension that proper voting requires, because these votes rarely influence the electoral outcome. While voting for third or small parties may be considered a waste strategically, it is still a clear expression of choice that is meant to influence the outcome, even if the chances of getting one’s candidate elected are slim. By contrast, when even a majority of invalid ballots cannot legally change the outcome, these ‘votes’ fail not only clearly to express a choice, but also to affect the outcome. To the extent that both the expressive and the instrumental dimension are violated, informal voting is not proper voting.

Compelling attendance is thus insufficient to give high turnout the democratic value that is supposed to have according to the non-instrumental case because attendance per se does not bring about proper voting. ¹⁷ Nor does it bring about reason-responsive voting, which is also needed to realise popular control. And while levels of invalid votes are recognisably higher under CV regimes (Kouba and Lysek, 2019; Martinez i Coma and Werner, 2019; Singh, 2019), the exact rate of random voting is bound to be undetectable. Even Australia, with its preferential voting system, can only detect donkey votes, which are likely to be the most egregious but not the only instances of random voting.

The non-instrumental case thus faces a dilemma. If formal voting is mandatory, the freedom condition is violated. If mere attendance is mandatory, the higher turnout may consist of invalid and random voting, thereby failing the reason-responsiveness condition. This is a problem for both popular control and political equality. Take a counterfactual case in which turnout is almost universal, but only a minority of citizens cast a formal and reason-responsive vote. In that case, a majority of invalid and random ballots would not only attest to citizens’ lack of capacity or willingness to make a choice among the options they are given, but also manifest their unequal political influence over collective decisions.

CV supporters may reply that although compulsion may not ensure that people vote freely and reason-responsively, it does not hinder democratic voting either. So long as the increase in democratic voting is higher than the amount of invalid and random votes, compulsory turnout seems effective at raising a valuable turnout. According to this restatement, compulsion is not justified because it directly brings about democratic voting, but on the idea that by maximising turnout, democratically justified turnout will be maximised also.

This reasoning raises two concerns. First, it justifies an obligation targeted at individuals based on a collective goal. It is because collectively increasing the pool of voters tout court will likely increase the pool of reason-responsive voters that the individual is compelled to attend elections. But what the individual is compelled to do is neither sufficient nor directly causally related to the goal that justifies the obligation. Compare this with another obligation to which CV is sometimes compared: taxation (Hill, 2002b: 87; Umbers, 2020: 1318). The law requires us to perform an action (paying taxes) that is sufficient and directly causing the goal that justifies the obligation. By contrast, when the law requires us to attend elections, this is not sufficient nor directly causally related to democratic voting, which is the reason we are asked to attend elections. Looking only at the aggregate dimension and ignoring the individual one is normatively problematic insofar as it amounts to treating individuals as mere means to maximising democratic turnout. Call this the deontological concern.

Second, this reasoning is premised on the idea that the number of reason-responsive voters will be equal or higher in higher-turnout elections compared with lower-turnout ones. The citizens at whom CV is directed can be divided in two categories: those who, once compelled, vote randomly and those who vote reason-responsively. ¹⁸ If at least some of the latter voters exist, the pool of reason-responsive voters is indirectly increased through compulsion.

The physical acts of arriving at the polling station, queuing, and casting a ballot are only the tip of the iceberg of a broader process, which includes, among other things, gathering diverse sources of information, reflecting on one’s values and interests and possibly deliberating with both like-minded and discordant citizens (Mackie, 2018). Because people must be aware of their subjective reasons, they need a minimum threshold of political knowledge to vote reason-responsively (Lau and Redlawsk, 2006; Lupia and McCubbins, 1998). Acquiring this knowledge is costly and time-consuming. It is thus more likely that people for whom the costs of acquiring information are not too high will take the obligation to vote as an opportunity to vote reason-responsively. By contrast, those who would gain relevant information at high cost seem less likely to vote reason-responsively.

Evidence on this is mixed. On one hand, CV seems to increase the level of political knowledge and even out its distribution, thereby helping those who are less educated become more knowledgeable (Carreras, 2016; Córdova and Rangel, 2017; Sheppard, 2015; Shineman, 2021). ¹⁹ On the other hand, the link between citizens’ policy preferences and their vote choices (i.e. proximity voting) is weakened in CV systems (Dassonneville et al., 2017; Singh, 2016) and this seems particularly true of so-called reluctant voters, that is, voters who would not vote if voting was voluntary (Dassonneville et al., 2019; Freire and Turgeon, 2020; Hooghe and Stiers, 2017; Selb and Lachat, 2009; Singh, 2019; Singh and Roy, 2018).

If CV is effective at raising voters’ political knowledge, then the presumption that getting more people to the polls will also raise the rate of reason-responsive voting seems justified. If this is not the case, however, compulsion will likely produce reason-responsive votes of those for whom acquiring information is easier and random or invalid votes of those for whom it is not. Therefore, CV supporters may find themselves in the awkward position of forcing those voters without resources to act pointlessly by casting invalid and random ballots to get to the polls those voters who have the resources to get informed and vote reason-responsively anyways. The apparent laziness or akrasia of some would then be avoided by compelling those for whom electoral participation is more difficult to do something that has no democratic value (i.e. attend elections without voting). Let us call this the unfairness concern.

Unless these two concerns are answered and CV is shown to bring about democratic voting, the conclusion does not hold. Despite some encouraging findings, the jury is still out and the non-instrumental case for CV fails to achieve its goal.

Conclusion

This article defended three main claims. First, conceptually, I offered an account of voting as a social practice defined by three constitutive rules. Second, normatively, I proposed considering voting democratically valuable if it is done freely and reason-responsively. These two conditions are required to satisfy two fundamental democratic principles: political equality and democratic control. These principles are invoked by some defenders of CV to claim that high turnout is non-instrumentally valuable. My third aim has been to argue that this non-instrumental case fails because it does not ensure the right kind of high turnout. While I agree that having almost-universal participation at elections is a good thing for democracy, I contend that it is good only conditionally. If citizens fail to vote freely and reason-responsively, high turnout is not an expression of political equality and popular control. This is not a knock-down argument against CV. It does show, however, that the non-instrumental case, as it stands, fails.

The broader importance of this academic debate cannot be overestimated. The worry that recent divisive results of elections and referendums are skewed by unequal attendance is justified. The multiple obstacles that some voters face, such as complex registration procedures (Braconnier et al., 2017) and poorly distributed polling places (Pettigrew, 2021), together with institutional mechanisms of minority-vote dilution and voter suppression (Guinier, 1995), make voting a difficult right for many to exercise. Hence, it is reasonable to highlight that the mere possession of a right to vote is not enough if the opportunities to exercise it are drastically unequal. Similarly, it is reasonable to question political institutions that fail to show equal respect for all citizens by making the exercise of this right more difficult for some (Thompson, 2002: 28).

Thus, when electoral turnout is low and unequal, there are reasons to worry, even more so if other fundamental aspects of political equality and popular control are missing. Robert Dahl, for instance, considered effective participation (the opportunity to make one’s views known to others), enlightened understanding of one’s interests and final control of the agenda as necessary criteria of democracy together with equal voting (Dahl, 1989). These other opportunities, together with the power to finance one’s preferred party and run for office, are frustrated in most liberal democracies. While CV is not meant to settle these problems, the very fact that it suggests an easy fix to the current crisis of democratic legitimacy contributes to ignoring the extent to which the promises of democracy are disappointed. ²⁰