Proportional Non-Voter Sortition: Legislative Inclusion for Non-Voting Citizens

Abstract

Democratic electoral systems characteristically take a Count-and-Report Approach to enfranchised citizens’ failure to turn out in political elections: The number of non-votes is counted and publicly reported, but non-votes are given no effect in the allocation of political offices. We morally assess this model in the context of legislative elections, and argue that it is pro tanto less desirable than another, counterfactual model which we label Proportional Non-Voter Sortition. This model provides that citizens’ non-votes shall be legislatively represented by legislators who are selected by lot from the non-voting part of the demos, and whose share of seats is proportional to the aggregate rate of non-voting in the legislative election. Our assessment of the two models specifically turns on the claim that a democracy’s legislature should reflect the full diversity of political views that exists within its demos (legislative inclusivity), and it offers two reasons to expect that Proportional Non-Voter Sortition serves legislative inclusivity better than the Count-and-Report Approach: 1) Proportional Non-Voter Sortition uniquely incentivizes political parties to integrate non-voter views into their political agendas and mobilisation activities; and 2) Proportional Non-Voter Sortition additionally enables the legislative articulation and advocacy of non-voter views by placing people in the legislature who are descriptively representative of non-voters. Through our discussion of different models for the post-electoral management of non-votes, we contribute to a body of normative democratic theorizing about citizens’ non-voting, but raise a distinct question that so far has remained unexplored.

Voting-eligible citizens of democratic states regularly fail to cast a ballot in political elections. Parliamentary elections in democracies rarely feature turnout rates over 70%, and even in countries renowned for high electoral participation, e.g., Sweden and Norway, the number of non-voters usually exceeds one million. (cf. IDEA 2023) Yet there is little discussion of how democratic electoral systems should absorb citizens’ non-votes—of how electoral systems should manage non-voting in the aftermath of elections.

In democratic states, electoral systems usually absorb non-votes in representative elections through a mechanism that one might call the Count-and-Report Approach: Non-votes are scrupulously counted and publicly reported as a part of the electoral result, either in the form of an abstention rate or (indirectly) as a turnout rate. However, the non-votes have no formal effect on the allocation of political office. ¹

Is such a Count-and-Report Approach a good model for absorbing non-votes? Might there be any alternatives that are more attractive? This question is important, not least since rates of non-voting are significant and indeed rising in numerous democracies around the world (Blais 2007; Kostelka and Blais 2021). Yet contemporary democratic theory has surprisingly little to say on it. Democratic theorists typically see non-voting as a democratic problem (e.g., Lijphart 1997) and intensely debate the justifiability of compulsory voting policies (e.g., Hill 2002; 2015; Lever 2010; Schäfer 2011; Brennan and Hill 2014; Elliott 2017; Booth Chapman 2019; Volacu 2020; Umbers 2020; Saunders 2016; 2018; Destri 2023) and other institutional arrangements for increasing turnout, including a lowered voting age (e.g., Zeglovitz 2013; Peto 2018; Umbers 2018) and the creation of financial (dis)incentives for (non-)voting (Saunders 2009; 2021). However, the separate question of how electoral systems (in a broad sense) should absorb—or post-electorally manage—non-votes has not been explicitly asked, let alone analyzed.

This article aims to fill this lacuna within democratic theory, and it does so by introducing and normatively defending a novel, counterfactual institutional model for the specific context of legislative elections. The model is labelled Proportional Non-Voter Sortition, and it distinctly gives non-votes a determinate formal effect: Proportional Non-voter Sortition translates non-votes into the legislature by including into it a share of legislators who are selected by lot from among the non-voting citizens, in a number that is proportional to the aggregate rate of non-voting. And by our (non-exhaustive) normative argument, Proportional Non-Voter Sortition is pro tanto preferable to the standard Count-and-Report Approach since it is more likely than the Count-and-Report Approach to further the important moral principle of legislative inclusivity—i.e., the claim that a democratic legislature should include and reflect the full diversity of political views that exists with the democratic citizenry. ²

Our defence of Proportional Non-Voter Sortition proceeds as follows. First, (1) we explain the model of Proportional Non-Voter Sortition in some more detail. After that (2) we explain and justify the principle of legislative inclusivity, showing that it rests solidly on three values dear to committed democrats: political equality, popular control over government, and epistemic governmental capacity. Then we go on to develop two separate arguments for why Proportional Non-Voter Sortition is more likely than the Count-and-Report Approach to further the principle of legislative inclusivity. The first argument (3) revolves around the behaviour of political parties under the respective models, whereas the second argument (4) homes in on the allotted legislators that Proportional Non-Voter Sortition can generate. In the penultimate step (5) we consider and rebut several potential objections to Proportional Non-Voter Sortition and our defence of it. (6) We conclude with a summary of the argument.

Before starting in earnest though, it is worth flagging two limits in our argument and making one clarification. First, our argument for Proportional Non-Voter Sortition partly invokes specific interpretations of the ideals of political equality and popular control. These interpretations are influential and widely shared in the philosophical literature, though not completely unrivalled. Still, we do not defend these conceptions as such. So, our overall argument is conditional on readers accepting some conventional conceptions of equality and control. To the extent that readers affirm other conceptions of these two values, our argument does not give them reason to support Proportional Non-Voter Sortition.

Second, our defense of Proportional Non-Voter Sortition centrally turns on the ideal of legislative inclusivity. That ideal can potentially be promoted in other ways, too, notably through a policy of legally compulsory voting. ³ But since compulsory voting is a philosophically contentious policy that, as already noted, is the object of an extensive scholarly controversy, we cannot within the scope of this article discuss adequately whether compulsory voting is an attractive competitor to Proportional Non-Voter Sortition. Our analysis accordingly brackets the policy option of compulsory voting, assumes a stable policy context of voluntary voting, and remains agnostic on the qualities of compulsory voting with respect to legislative inclusivity. And this in turn limits the argument’s scope: our defense of Proportional Non-Voter Sortition applies only to democracies that have a prior, independent commitment to voluntary voting. But although so limited, the argument remains significant as in many democracies voluntary voting is a longstanding and presumably cherished practice, and other democracies, e.g., the Netherlands, have explicitly reversed their previous compulsory voting policies in favour of voluntary voting (cf. Birch 2009).

Finally, our argument does not look to short-circuit democratic deliberation about Proportional Non-Voter Sortition. To say that Proportional Non-Voter Sortition is pro tanto desirable, on grounds that should speak powerfully to committed democrats, is not to claim that democracies should implement Proportional Non-Voter Sortition without regard to public opinion. That might be the implication if one claims that democratic values strictly require Proportional Non-Voter Sortition. However, ours is not so strong a claim, and the article therefore should not be taken as an attempt to sidestep democratic deliberation. Rather, the following argument aims to constructively open and nourish public deliberation about how democracies might properly absorb non-votes. And if it succeeds at spurring sustained, public consideration of that question, we shall indeed be satisfied that an important part of our aim has been achieved—even if our specific policy proposal is not ultimately retained.

Proportional Non-Voter Sortition

Proportional Non-Voter Sortition as we define and defend it is a legal-constitutional policy model that plays a role in defining how general election results translate into a specific distribution of parliamentary seats. The model provides that the legislature shall comprise elected as well as appointed members and that the share of the appointed members shall be proportional to the aggregate rate of non-voting in the last general election. The policy further provides that the appointed legislators are to be selected by lot exclusively from among the voting-eligible citizens who have failed to vote in the election. The elected and the allotted legislators are to have identical parliamentary rights and duties, e.g., equal rights to vote and deliberate in the chamber; same rights to join committees and constitute parliamentary groups; equally long mandates; equal levels of renumeration, and so on. And Proportional Non-Voter Sortition finally includes an opt-out clause that allows allotted individuals to decline the opportunity to serve as legislator, in which case the selection mechanism continues to operate until it returns a non-voter who is prepared to serve.

The expected consequences of this particular policy model are the main topic of the sections developing the party-based argument and the argument from descriptive representation. But we can already note the model’s most immediate and general effect: Proportional Non-Voter Sortition yields a legislative assembly that includes allotted members who are drawn from the non-voting population and whose number expands and contracts over time in response to changing rates of non-voting. But unlike other policy proposals that involve sortition for (a kind of) legislative office (e.g., Zakaras 2010; Guerrero 2014; van Damme and Verret-Hamelin 2017; Gastil and Wright 2018; Landemore 2020; Abizadeh 2020; Stone and Malkopoulu 2022), ours does not intend to replace electoral democracy. Proportional Non-Voter Sortition on our account is institutionally nested within a constitutional framework of regular democratic elections and merely serves to second the electoral system in the event that (some) enfranchised citizens do not participate in the elections. As such, Proportional Non-Voter Sortition clearly is a rival to the Count-and-Report Approach that records and publicizes the number of non-votes. Yet Proportional Non-Voter Sortition does not challenge or seek to displace elections per se as the primary mode of legislator selection. This much is also clear from the fact that if participation rates in an election approach 100%, then Proportional Non-Voter Sortition will not draft any non-voter legislators into the parliamentary assembly. ⁴

Equally important, the practical feasibility of Proportional Non-Voter Sortition does not depend on the pre-existence of any particular electoral system. Though intuitively most compatible with a proportional electoral system, Proportional Non-Voter Sortition can also be practiced in combination with other systems, notably single-member district election systems. To do so, one just leaves unaltered the number of elected legislators, but adds to the legislature a new, additional tiers of allotted legislators that shrinks and grows in proportion to the aggregate, national rate of abstention.

Finally, notice that as we defend it, Proportional Non-Voter Sortition is a policy model articulated at a medium level of specification. The model represents an institutionally distinct electoral arrangement; yet the practical implementation of it in a particular context necessitates further specifications. For instance, one must specify whether the share of allotted legislators should be proportional to the rate of non-voting in the voting-eligible population (i.e., number of non-votes divided by number of names on the voter register) or to the rate of non-voting in the voting-age population (i.e., number of non-votes divided by number of enfranchised citizens). Similarly, one must specify the proportionality constant (k) that mathematically mediates between the rate of abstention and the share of allotted legislators in the assembly. In other words, one must decide whether an increase by ten percentage points in the rate of non-voting should entail a ten-percentage-point increase also in the share of allotted legislators (k=1), or whether the conversion rate should be smaller or larger—e.g., k=0.5 which would turn a 10% non-voting rate into a 5% share of allotted legislators. To implement Proportional Non-Voter Sortition, one must also decide whether blank and spoilt ballots should count as non-votes; ⁵ one must develop an exact legal formulation of the policy, and one must give an account of how it fits into the wider electoral law. However, the proposal advanced here does not tackle such tasks of detailed specification. Rather, we aim to offer a model that has brightly defined core features but is still sufficiently general to be relevant for a variety of different contexts.

This concludes our argument’s descriptive part. Next, we turn to the argument’s normative leg, which begins with an explanation and justification of the ideal or principle that is at the normative argument’s center—i.e., the principle of legislative inclusivity.

Legislative Inclusivity

As we define it, the principle of legislative inclusivity is a regulative ⁶ moral ideal saying that a democracy’s law-making assembly should reflect all political views that exist within the democracy’s demos. The legislature should accommodate the articulation and advocacy of all the diverse views that enfranchised citizens have about the proper ordering of public affairs and the law especially. Thus defined, legislative inclusivity is, of course, rarely perfectly realisable. An assembly of a few hundred members cannot normally represent every shade of political opinion that exists within a citizenry of millions. But a legislative assembly can do better or worse at reflecting the diversity of citizens’ political views, and to the extent that a legislature enables the advocacy of a broad variety of citizen views, we might say that it is approximately inclusive.

In affirming the principle of legislative inclusivity, we do not claim any particular originality. Rather, we expect the principle to be quite familiar. In fact, it is precisely what John Stuart Mill advocated in Considerations on Representative Government, where a key passage reads: “Parliament has an office … to be … an arena in which’ every opinion in the country ‘can produce itself in full light and challenge discussion; where every person in the country may count upon finding somebody who speaks his mind” (Mill 2008, 104). Still, it is worth appending two clarifications.

First, legislative inclusivity is no exhaustive account of good democratic representation. On our definition of the principle, legislative inclusivity is satisfied as soon as every significant segment of political opinion is expressed and defended by one, and no more than one, member of the legislature. But this presumably is not fully satisfactory from the point of view of democratic values such as political equality and popular control over government. These values presumably imply further desiderata of good representation, most notably a desideratum of proportionality. A full, democratic account of representation will demand that in addition to legislative inclusivity, there be a measure of proportionality between the extent to which views are defended in the legislature and the extent to which they are popular among citizens. Still, this article leaves to the side the proportionality desideratum and the broader idea of good, democratic representation. The proportionality desideratum can be interpreted in more or less demanding ways. On a relatively undemanding interpretation, the proportionality desideratum just means that the distribution of opinions within the citizenry and the legislature should track each other loosely: the most popular view among citizens should have the strongest representation within the legislature, the second-most popular view should be represented by the second-most numerous group of legislators, and so on. On a more demanding interpretation, though, the proportionality desideratum means direct proportionality—if 30% of citizens hold a particular view, then it must also be defended by 30% of all legislators. The appropriate way of interpreting the proportionality desideratum is furthermore a matter of extensive scholarly debate. ⁷ So, developing a theoretically robust account of the proportionality desideratum, and good democratic representation more broadly, is quite beyond the scope of this article. And for that reason, its analysis looks exclusively at the relatively ecumenical principle of legislative inclusivity, all while recognising that it is not all there is to good democratic representation.

The second clarification to add is that legislative inclusivity is not a principle of descriptive representation. The principle does not itself require that the distribution of particular external characteristics (e.g., gender, ethnicity, socio-economic status) among legislators track the distribution of those characteristics within the wider society. Legislative inclusivity is instead concerned with political ideas, and insofar as it requires a certain correspondence, it is the correspondence between the political ideas that exist within society and the ideas that get articulated and advocated within the legislature that matters. As a practical matter, descriptive representation may well be instrumental to realising legislative inclusivity—in the following we in fact develop an argument along those very lines. But at the conceptual level, we insist on keeping distinct the principle of legislative inclusivity itself and the means for achieving it such as descriptive representation.

With these clarifications in place, we now turn to the justification of legislative inclusivity: why, if at all, is this a valid, or attractive, moral principle? Our short reply is that the principle is valid because it connects to three more fundamental values that committed democrats characteristically endorse: political equality, popular control over government, and epistemic government reliability.

Consider first the value of political equality. While the correct interpretation of political equality is the subject of ongoing scholarly dispute, a number of leading theorists converge on the general understanding that political equality requires that in the context of decision-making about the ordering of public affairs, all citizens have the opportunity to have their political views considered on their merits. Wilson (2019, 49), for example, explains political equality as the idea that every citizen has the “entitlement” to “obligate others to attend to their own judgments, and to give weight to those judgements, in the course of their deliberations about what we ought to do.” Similarly, Christiano describes political equality as the principle that “each and every citizen ought to have a say in the choice of collective features of society” (1996, 70; cf. Christiano 2008; Beitz 1989). ⁸ And an approximately inclusive legislature can in turn significantly help to further the very objective that each citizen have the opportunity to have their political views considered on their merits. Because an approximately inclusive legislature enables a wide range of citizen views to feed into the parliamentary deliberations and voting cycles that shape the content of the law, a society is all else equal more politically egalitarian the more its legislature is inclusive. And so, legislative inclusivity is an attractive moral principle on politically egalitarian grounds. Political equality maybe does not strictly entail or imply the principle of legislative inclusivity—we remain agnostic on that point—but all else equal, political equality certainly favors a more inclusive legislature over one that only reflects a restricted set of citizen views.

Legislative inclusivity is attractive also on the grounds of popular or democratic control. This second value is closely related to the idea of collective self-government and mandates that a state’s policy stand in a particular relationship to the plural political judgments of all its citizens. Popular control over government asks that state policy be broadly responsive to, or track, the considered political judgments of its citizens, and that it should do so over time, across changes in the views of citizens. Popular control is in other words the claim that there should be an “ongoing alignment between processes of political opinion- and will-formation in which citizens participate and the policies to which they are subject” (Lafont 2019, 23). Popular control is the idea that citizens should jointly “impose an equally acceptable direction on government” (Pettit 2012, 179). And an approximately inclusive legislature can again aid to further this particular idea. An approximately inclusive legislature does not necessarily guarantee an alignment of the required kind. But since it enables a broad variety of citizen views to influence the legislative processes that shape the content of legislation and public policy more broadly, an approximately inclusive legislature certainly increases the likelihood that laws and policies continually track the views of citizens in all their plurality. Hence, legislative inclusivity is a valid, desirable principle also from the perspective of popular control. And the same goes, too, for our final value of epistemic government capacity.

The notion of epistemic government capacity refers to a state’s reliability at selecting public policies that are substantially correct or reasonable, its robust tendency to select policy options that are effective at producing the intended outcomes and conform to the substantial requirements of justice. In turn, an important pre-condition for this reliability is that the state’s decision-making process mobilizes diverse interpretations of the problems at hand and varied approaches to solving them. Indeed, in the context of collective decision-making, there are strong reasons to believe that the propensity to select for substantially correct options correlates tightly with the cognitive diversity of the decision makers, more so than with their individual competence (Hong and Page 2012). So, for a state to be epistemically reliable, it is important that its law and policy-making processes engage an array of different perspectives and ideas. ⁹ This diversity may indeed be especially important at the level of the legislature, as it is the institution that substantively tackles problems of law and policy and its membership is sufficiently restricted to enable the deliberative exchange of considerations that is central to unlocking the full epistemic gains of diversity (cf. Landemore 2012). An approximately inclusive legislature, for its part, helps to secure precisely this precondition of perspectival diversity within the law-making assembly. And so, the principle of legislative inclusivity should be attractive also from within a concern for epistemic government capacity.

Legislative inclusivity is a valid regulative principle, then, that rests squarely on the values of political equality and popular control as conventionally interpreted, and on the value of epistemic governmental capacity—values that characteristically matter a great deal for democrats. And by implication, there is an important pro tanto democratic case for any policy that furthers this ideal. But it still remains open how the Count-and-Report Approach and Proportional Non-Voter Sortition perform in this regard. So, we now turn to this issue, developing two separate, probabilistic arguments to the effect that Proportional Non-Voter Sortition is more likely than the Count-and-Report Approach to promote legislative inclusivity. We begin with an argument that homes in on the behavior of political parties.

The Party-Based Argument

As a growing body of normative scholarship recognises, political parties are essential components of representative democracies (Rosenblum 2008; Muirhead 2014; White and Ypi 2016; Bonotti 2017; Wolkenstein 2020) that perform crucially important activities. When at their best, political parties channel citizens’ views into their legislative programmes and broader political agenda. Parties continuously collect political demands, perspectives, and judgements from their members, sympathisers and potential voters, organising and integrating them into distinct visions of the common good and corresponding coherent policy programmes that they seek to implement (Wolkenstein 2020, 27). But not only that; political parties also play a role in the formation of the political views and judgements that citizens hold in the first place. Instead of merely picking up and channelling pre-existing citizen views into political action, political parties continuously work to discursively relate their political action to particular political values, interests, and conceptions of the common good that ideally appeal to all citizens, or at least to the parties’ members and prospective voters. Additionally, parties continuously strive to diffuse and publicise their justificatory narratives across society (White and Ypi 2016, Bonotti 2017), thus “building political constituencies whose members feel a sense of common purpose” (White and Ypi 2010, 812) and shaping what citizens think is politically desirable and expedient, good and bad public policy.

So, political parties are effectively “bilingual” entities (Muirhead and Rosenblum 2006) that operate as a two-way channel of communication and expression (Sartori 2005, 24-25), which has led scholars to interpret political parties both as “agents of popular sovereignty” (Wolkenstein 2018) and a “source of … political justification” (White and Ypi 2016, 58). But this is not all: If political parties continuously collect and integrate citizen views into their legislative programmes and political agendas, they aid to maintain correspondence between the views that are expressed in the legislature and the views that are present within the citizenry. Equally, if parties continuously expend efforts to ideologically mobilize and win citizens’ support for their various policies, that also helps to maintain congruence between the views expressed in the legislature and the views that circulate within the citizenry. In a word, political parties can be powerful agents of legislative inclusivity. However, the likelihood that parties thoroughly unfold this potential is importantly affected by the wider institutional context, and under a Count-and-Report Approach to non-voting, parties are rather unlikely to function as strong vectors of legislative inclusivity. To understand why, notice first a basic point about non-voters.

Non-voting citizens in contemporary democracies roughly divide into three ideal-typical groups that we respectively can label as circumstantial, principled, and disconnected non-voters. Circumstantial non-voters are not in principle opposed to voting but fail to turn out for what are broadly practical and circumstantial reasons. Circumstantial non-voters remain absent from the polls because they or their kids are sick, because they work on election day, because accessing the polling station is difficult, because they think that in this particular election their vote is dispensable, and so on.

By contrast, principled non-voters remain absent from the polls for more deliberate and principled reasons. Notably, this category includes citizens who do not vote because they disapprove of the available candidates or party platforms, or because they are politically cynical, thinking that politicians are generally not dependable or impotent such that it makes no difference who is in power. This second category of non-voters is well exemplified by Jaqueline, a French non-voter who in a focus group interview explains: “The real issue is that … the politicians always put their own ego first. They do not really address the inequality, the rising cost of living, the poverty” (Llorca 2022, 14; authors’ translation; cf. Eilfort 1994, 335; Muxel 2007). Another example is Callum Clayton-Dixon, an Aboriginal Australian who motivates his refusal to vote in terms of a constitutional critique of Australia and its unjust incorporation of Aboriginals: “I’m not going there [to the polling station], I’m not going to participate. We never consented to this political order being imposed on us, it doesn’t cater for us.” (Jabour 2018).

And lastly, the disconnected non-voters are those citizens who refrain from participating in elections because they experience politics as a distant activity. Disconnected non-voters may, for example, view voting and politics as epiphenomenal, as largely irrelevant to their prospects for leading a good life. Or they might be unknowledgeable about the policy platforms of the different political parties. This last category of non-voters is represented effectively by Jo and Becca, two British non-voters. When interviewed about their motives for not voting, Jo explains that “it’s not going to affect my life directly if I don’t [vote]” (Coleman 2013, 131) and Becca admits to a lack of knowledge and understanding: “I might sound thick, but I don’t understand who is going to do what and what would be best, so I just don’t bother [to vote]” (Coleman 2013, 111).

Importantly, this three-partite distinction between circumstantial, principled, and disconnected non-voters is an ideal-typical simplification. Any given non-voter may have mixed motives such that they can fit into several categories at once. And more fine-grained analyses can allow a division of non-voters into further, more refined categories. For instance, Kevin Elliot (2017, 662, 665) reports data on American non-voters’ reported reasons for non-voting that would suggest a division of non-voters into at least five categories: a) those who are too busy for voting (18.9%); b) those for whom registering or accessing the polling place is practically difficult (11.7%); c) those who cannot vote due to illness or disability (14%); d) those who lack an interest in voting (15.7%); and e) those who disapprove of the candidates or campaign issues (12.7%). Yet our three-partite typology is broadly consistent with this more fine-grained division. The categories a), b), and c) fit into our category of circumstantial non-voters; group d) slots into our category of disconnected non-voters, and our category of principled non-voters captures group e). Other fine-grained non-voter typologies (see, e.g., Doppelt & Shearer 1999; Schäfer 2015; Gray 2015) are also broadly consistent with our three-part typology, which, at any rate, suffices for the present purpose.

Returning now from the categorisation of non-voters to the behavior of political parties, we can expect that under a Count-and-Report Approach political parties will lavish considerable attention on circumstantial non-voters. The desire to secure a maximum number of votes powerfully drives each party to electorally mobilize its own supporter base and to work for high turnout among citizens who are at risk of not turning out for purely circumstantial reasons.

But with the other non-voters, the picture is quite different. Political parties can potentially increase their electoral success by paying attention to the views of principled non-voters like Jaqueline and Clayton-Dixon and incorporating those views into their political platforms and political agendas. Parties can also potentially gain votes by trying to partisanly educate disconnected non-voters. More votes might be obtained by making non-voters like Jo aware that policy decisions deeply affect their lives and making sure that party programmes are known and understood by non-voters like Becca and the population strata most prone to non-voting—i.e., the poor, the young, and those with little formal education (cf. Smets and van Ham 2013). However, programmatic revisions and partisan education are costly and risky. Programmatic revisions may not succeed at attracting additional electoral support from principled and disconnected non-voters and can even drive away parts of a party’s originally acquired electorate. A party’s attempt to partisanly educate and mobilize disconnected non-voters can, further, lead to increased turnout, but not necessarily in favor of the party doing to the educational work. ¹⁰ So, under a Count-and-Report Approach to non-voting, the optimal strategy for political parties is to not substantially engage with principled and disconnected non-voters. A rational political party will make considerable efforts to electorally mobilise its own supporters, but remain wary of mobilizing principled and disconnected non-voters through programmatic revisions and partisan education respectively. A perfectly rational party may actually look to foster non-voting among the supporters of rival parties ¹¹ and aim to consolidate the non-voting behavior of principled and disconnected non-voters so as to ensure that their votes do not accrue to another party (negative campaigning).

So, the Count-and-Report Approach gives political parties incentives to engage only marginally with non-voters who are motivated by reasons other than purely circumstantial ones. And as a result of this incentive structure, it is unlikely that the Count-and-Report Approach leads parties to unfold their full potential as vectors of legislative inclusion. Non-voter views in particular will likely be absent from party programs and the legislative assembly, as a result of parties’ limited (but rational) uptake of principled non-voter views and parties’ limited (but rational) efforts to win hearts and minds among disconnected non-voters. ¹²

By contrast, Proportional Non-Voter Sortition changes all this. Through its distinct way of transposing non-voting into the composition of the legislative chamber, Proportional Non-Voter Sortition creates a particular, negative incentive for parties to promote electoral turnout. Under Proportional Non-Voter Sortition non-voting results in the allocation of legislative seats to randomly selected non-voters and corresponding losses for political parties. Under Proportional Non-Voter Sortition non-voting means that all parties equally lose legislative seats as well as any financial resources that attach to those seats (e.g. resources for staff for the party group). If non-voting is widespread and rises to, say, 33% or 51%—as in the most recent UK and French general elections respectively (IDEA 2023)—then it can also hazard parties’ ability to discharge their core mission—i.e., to form legislative majorities capable of sustaining governments and passing bills. Such outcomes are highly unpalatable for political parties, and Proportional Non-Voter Sortition thus provides a strong negative incentive for parties to promote electoral participation, not only among their own respective supporters but across the demos more generally. In turn, a key way for parties to do this is to eliminate obstacles that stand in the way of citizens participating in elections (e.g., non-automatic voter registration, absence of advance voting) and to partisanly educate disconnected non-voters so that they adopt crystalised political views and become regular voters—just like European mass parties of the nineteenth and twentieth century expended large efforts to politically socialise and ideologically mobilise the newly enfranchised working class and women (cf. Duverger 1954; Neumann and Barghoorn 1956; Katz 1990). An additional way to promote electoral participation is to integrate non-voter views, notably those of principled and disconnected non-voters, into the parties’ programmatic platforms. And so, Proportional Non-Voter Sortition likely drives parties to engage with non-voters beyond their own core constituencies and to partly integrate those citizens’ views into their respective platforms.

Of course, this integration is unlikely to be complete or perfect. Some non-voter views presumably are too idiosyncratic or too much in tension with parties’ commitments to allow for a meaningful incorporation. For instance, die-hard anarchist views may not easily fit into any partisan agenda. But to the extent that the views of non-voters are not fundamentally incompatible with the political parties’ commitments, their fear of large-scale non-voting likely drives parties to be responsive to those views: parties are likely to build non-voter perspectives into their respective politics, thus helping to maintain congruence between the views that exist within the demos and the views that receive legislative expression.

There is another reason, too, why Proportional Non-Voter Sortition is likely to reinforce parties’ role as strong vectors of legislative inclusivity. Proportional Non-Voter Sortition not only creates negative incentives for political parties to pre-empt large-scale non-voting; it also creates certain positive incentives for parties that are nearly majoritarian during a given legislative term. Under Proportional Non-Voter Sortition, such parties may find themselves short of a number of votes that can potentially be provided by allotted non-voter legislators. Under such circumstances, the party may well seek to secure cooperation from the allotted lawmakers or desire to co-opt them into its own ranks. A major way to accomplish this in the presence of stringent and effective anti-corruption measures is in turn to offer policy concessions that can appeal to non-voters. And so, there is a reason grounded in distinctively positive incentives, too, to expect that Proportional Non-Voter Sortition drives political parties to partly articulate and defend non-voter views and preferences in the legislature.

In response to the last line of argument, a skeptic might think that if parties can co-opt the allotted legislators after they are drafted into the legislature, then parties will not be all that sensitive to the negative incentive to reduce non-voting in the first place. Parties, in this way of thinking, might accept non-voting even at a large scale and just seek to capture the allotted lawmakers once they arrive in the legislature. For that might, indeed, be less trouble than trying to engage in ideologically costly platform revisions and resource intensive ideological education.

However, this view neglects the loose-cannon nature of the allotted lawmakers. The loyalty of co-opted non-voter legislators is far from certain, and since they are not electorally indebted or accountable, they may not easily be disciplined to follow the party line. Parties may furthermore compete with each other for the allotted lawmakers. And since the allotted legislators are drawn from among the non-voters, it is not even certain that they are open to co-operate with any of the existing parties. So, exclusively relying on the post-electoral co-optation of allotted non-voter legislators is a high-risk strategy that parties are not likely to adopt. Under Proportional Non-Voter Sortition, parties presumably will be sensitive both to the positive and negative incentives to incorporate non-voter views into their programmatic platforms and legislative activities. And to the extent the Proportional Non-Voter Sortition provides such incentives, it is distinctively more likely to yield a legislatively inclusive assembly than is the Count-and-Report Approach, which, recall, gives political parties preciously few incentives to care about the views of non-voters outside of their core constituencies.

Altered party behavior is not the only reason, though, why Proportional Non-Voter Sortition is more likely to generate an approximately inclusive legislature. This expectation also rests on a second reason that relates to the randomly selected legislators themselves.

The Argument from Descriptive Representation

To the extent that Proportional Non-Voter Sortition leads to the appointment of allotted legislators, it places people in the parliamentary chamber who are relatively unconstrained and free to advocate views of their own rather than the views of others. Unlike elected legislators, the allotted ones have made no election promises, nor are they liable to electoral sanction at the issue of their term. So, when participating in the legislature’s work, when deliberating, voting, negotiating with other lawmakers, and cooperating with the established parties, the allotted legislators will likely present and advocate for their own political standpoints rather than try to purposely speak for others.

However, in their defense of seemingly individual political standpoints, allotted lawmakers are likely to advocate for views that are actually common among non-voters more generally. The reason for this is fundamentally that allotted legislators are selected from among the non-voting population specifically by lot. Unweighted random selection has the well-known property that it tends to produce a sample that has the same distribution of characteristics as the underlying population. Any characteristic that is frequent within the population is likely to resurface within a random sample from that population. So when deploying random selection for appointing the non-voter legislators, Proportional Non-Voter Sortition is likely to place persons in the legislature who as a group have political views that resemble the range of views that are affirmed by non-voters as a (heterogenous) group. If, for example, the non-voting population includes many circumstantial non-voters who in principle agree with the position of a specific political party, then the allotted lawmakers will likely include many persons who also agree with that partisan position. If sizable numbers of non-voters think that existing policies for the benefit of the poor or state-Aboriginal relations are inadequate, then the sorted non-voter legislators are likely to include people like Jaqueline and Clayton-Dixon, and so on. This reflective dynamic has as a limit, of course, in that a two- or three-digit number of allotted legislators cannot possibly reflect every shade of opinion that exists within a non-voter population of millions. But the random sampling of the non-voter legislators still makes it likely that the aggregate opinion profile of the allotted legislators resembles that of non-voters as a (heterogenous) group. And that in turn means that Proportional Non-Voter Sortition is likely to produce a tier of non-voter legislators that as a group affirm and defend an array of positions that roughly track the array of political attitudes and judgments that exist among non-voters more generally—all whilst it looks as though the allotted legislators merely defend their individual politics.

Resemblance in terms of political opinions is not the whole story, though. Non-voters and allotted non-voter lawmakers likely resemble each other in social and economic terms, too. As students of electoral turnout know well, non-voting is socially biased. Socio-economically weak citizens are strongly overrepresented among the citizens who do not vote, as are young citizens, people with lower levels of education, and members of ethnic or immigrant minority groups (cf. Smets and van Ham 2013; Fraga 2018; Dahlberg and Mörkenstam 2019). So, when Proportional Non-Voter Sortition picks allotted legislators specifically from the non-voting segment of the citizenry, it likely places people in the legislature who themselves have experiences of disadvantage and can function in a way similar to other descriptive representatives.

Advocates of descriptive gender and racial minority representation (e.g. Phillips 1995; Williams 1998; Mansbridge 1999; Young 2000) have persuasively argued that women and racial minority lawmakers can effectively further the interests and political judgements of their likes, not because “political perspectives are determined by one’s location in gender or racial hierarchies,” but rather because “the capacity to recognise and challenge these hierarchies depends on exposure to them, whether this be direct or mediated through others” (Phillips 2020, 177). And with allotted non-voter legislators it can arguably be the same. Some allotted legislators will likely have own experiences of poverty, ethnic discrimination, and classist stigma, which probably gives them a particular ability as well as inclination to recognise and call into question these forms of disadvantage and domination. That the allotted non-voter legislators do act in this way is not, of course, certain. Nor is it guaranteed that any one of the allotted legislators has experience of the disadvantages that characteristically affect non-voters. But Proportional Non-Voter Sortition makes this likely. And this in turn reinforces the probability that allotted non-voter legislators hold, articulate, and advance a set of perspectives that is roughly congruent with the views and judgements of many other non-voters—all whilst it looks as though the draftees merely advocate for their own, individual political standpoints.

So, in their work as legislative debaters, negotiators, and voting members of the chamber, allotted non-voter legislators are likely to form and defend an array of perspectives that resembles the range of views that exists within the non-voting population at large. The explanation for this lies in allotted legislators’ resemblance to other non-voters, and as such it is independent of how political parties operate. Allotted non-voter legislators help to further non-voter views whether or not parties adjust their agenda and ideological mobilisation efforts to cater to non-voters. Moreover, this mechanism has no counterpart under the Count-and-Report Approach to non-voting. There, non-voters typically do not enter parliament, and no other mechanism exist to make it likely that non-voter perspectives receive a hearing in the law-making assembly. Some elected representatives may share and even contingently advocate views that resemble those of some non-voters, but under the Count-and-Report Approach there is little institutional incentive to encourage a systematic articulation of non-voters views within the legislature. So, in addition to the party-based argument, there exists a separate argument based on resemblance or descriptive representation as to why Proportional Non-Voter Sortition is more likely than the Count-and-Report Approach to transport non-voter views into the legislature and to approximate the ideal of legislative inclusivity. And if this is all correct and legislative inclusion is indeed a valid ideal—as we have argued on the basis of prominent conceptions of political equality, popular control, and epistemic reliability—then it follows that Proportional Non-Voter Sortition is the pro tanto preferable policy approach to non-voting. Democrats who accept our underlying conceptions of equality, control, and epistemic reliability should reject the standard Count-and-Report Model in favor of Proportional Non-Voter Sortition.

Replies to Concerns

Having developed the positive case for Proportional Non-Voter Sortition, it is now time to consider worries and objections that may still linger. In particular, we shall address seven distinct worries that respectively deal with: legislative inclusivity under a Count-and-Report Approach, voting secrecy, the efficacy of Proportional Non-Voter Sortition, the effects of the opt-out clause within Proportional Non-Voter Sortition, alternative policy solutions for securing legislative inclusivity, the value of popular control, and the scope of legislative inclusivity.

So first, does our argument overstate the extent to which the Count-and-Report Approach is deficient regarding legislative inclusivity? Critics might press this point in two ways. They might either argue that non-voters lack meaningful political views and that legislative inclusivity therefore is not set back if “non-voter views”—a spurious notion, in this way of thinking—are not vindicated in parliament. Or critics might say that non-voter views do not substantially differ from those of voting citizens such that legislative inclusivity is satisfied by the Count-and-Report Approach and political parties’ ordinary advocacy of voter views.

The first version of this criticism seems especially powerful for it appears to be supported by public opinion research. Philip Converse (1964) famously argued on the basis of survey results that large parts of the American public lack determinate and stable views on a wide range of timely, political questions. Converse further suggested that the phenomenon of non-attitudes would be particularly pronounced for “chronic non-voters” (1964, 253). And subsequent public opinion research indeed shows that survey respondents are particularly likely to give “don’t know” or “no opinion” answers when they present features that are also associated with non-voting such as low levels of education, income, political interest, and knowledge (Althaus 2003, 66).

However, a closer inspection reveals several problems with this line of criticism. For one thing, we have seen that some non-voters remain absent from the polls because they disagree politically with the available candidates or parties, with Clayton-Dixon being a clear case in point. We have seen, too, that some non-voters fail to turn out due to merely practical reasons—sick kids, inconveniently placed polling stations, etc.—and thus may have rather clear and articulate partisan sympathies for a determinate party. So claiming that non-voters across the board lack political views surely is an overstatement: the idea seems plausible mostly in relation to the non-voters that we call disconnected non-voters.

Further, that non-voters fail to articulate determinate views in opinion surveys does not necessarily indicate a lack of political views as such. “Don’t know” responses may reflect a desire to not bear the social costs of expressing one’s true opinion (Berinsky 2004). Organizing one’s political beliefs into a specific, articulated position can furthermore take effort and time that is not necessarily available in a standard survey interview situation (Inglehart 1985). And experiments with deliberative mini-publics indeed show that when participating in formally deliberative events that provide opportunities, resources, and reason to do so, people ordinarily form determinate political views on the relevant issues if they did not have any to start with (Fishkin 2009; Farrar et al. 2010). So even if in surveys non-voters look unopinionated across an array of topics, they may still have embryonic, inchoate political views that, in a manner of speaking, are ready to crystallize and become articulable when conditions become favorable.

And lastly, from the fact that respondents lack meaningful views on pollster-defined questions, it does not necessarily follow that respondents lack political views altogether: they may still be opinionated with respect to a personal domain of interest not covered by the survey. In fact, Converse himself was open to this possibility, speculating that the American mass public is fragmented “into a plethora of narrower issue publics” (Converse 1964, 245), i.e., delimited groups that are each concerned about and opinionated on a relatively specific political question. Jon Krosnick and colleagues likewise find that the majority of Americans likely belong to at least one issue public (Krosnick 1990; Krosnick, Berent and Boninger 1994). So, even if non-voters tend to lack determinate views across the major political topics of their time, they still are likely to entertain views on niche issues within their personal domain of interest. One non-voter may have a view on state-Aboriginal relations, another is opinionated on abortion rights, a third has a position on animal protection legislation, and so on. And so long as those views are political in a broad sense—views about the proper ordering of public affairs and the law especially—they are sufficient to place a citizen within the ambit of the legislative-inclusivity principle.

Turning now to the second version of the criticism, the question is whether non-voter views really are distinguishable from those of voters. And here there is some empirical work suggesting that the political preferences of American voters and non-voters differ only marginally (Shaffer 1982; Highton and Wolfinger 2001). But as Ruth Dassonneville et al. (2021, 584) note, these studies focus on the U.S. context during limited time periods, “which severely limits generalizations.” Moreover, through a longitudinal and cross-national analysis of their own, Dassonneville and her colleagues (2021, 590) find that “in most countries voters and non-voters’ ideological preferences differ significantly” and that “the median voter tends to be situated to the right of the median citizen.” So, to the extent that general empirical knowledge is available about the political views of voters and non-voters, the two are not congruent—and thus it is a mistake to think that political parties’ advocacy of voter views suffices to satisfy legislative inclusivity.

All that said, the two criticisms we have discussed so far do usefully pinpoint that in an important sense our case for Proportional Non-Voters Sortition is empirically conditional. It turns on non-voters having (rudimentary) political views and preferences on some issues at least, and on those views and preferences being distinct from those of voters. While we have tried to show that both conditions plausibly obtain, our argument indeed weakens to the extent this is not the case, e.g., in a particular national context.

Next, does Proportional Non-Voter Sortition compromise voting secrecy? This worry is understandable in that Proportional Non-Voter Sortition requires a record compiling the personal identities of all non-voting citizens, as the basic, infrastructural record to which the random selection mechanism can be applied. Yet one should bear in mind that as conventionally formulated, voting secrecy is a principle that applies to the partisan or candidate choice that voters express on the ballot paper. The principle means that “voters are entitled to keep their votes to themselves” (Lever 2007, 376, emphasis added; cf. Brennan and Pettit 1990; Theuns 2017), not that it must be secret whether or not they attended the polls. So the concern about voting secrecy is spurious. Besides, one should note that Proportional Non-Voter Sortition is no more revelatory than existing electoral procedures. In virtually every democracy, electoral authorities keep lists or registers of the eligible citizens. As citizens turn out to vote on election day, they are ticked off the list to prevent multiple voting. So, existing electoral procedures already systematically identify non-voters.

How about the efficacy of Proportional Non-Voter Sortition? Does our argument overstate the extent to which the model transfers non-voter views into the legislature? Critics might think so based on a concern about the allotted non-voter legislators. The potential worry is specifically that the allotted legislators will include disconnected non-voters like Jo who may be ineffective at presenting and advocating their views in the legislative setting. More fundamentally still, these legislators may fail to refine their inchoate, embryonic political views into more specific and actionable ones, with the result that Proportional Non-Voter Sortition maybe is less apt than our argument suggests at channelling non-voter views into the legislature.

The starting premise for this worry must indeed be granted: disconnected non-voter legislators may be lame ducks. But the already mentioned results from mini-public experiments are encouraging also here. Participants in deliberative events tend to form determinate judgments if they did not have any prior to joining the deliberative body. Deliberating citizens also tend to change their views into a more informed and coherent direction (Fishkin 2009 and Farrar et al. 2010). So, while deliberative mini-publics and legislatures are quite distinct, there may be reason for more optimism than the critic grants about disconnected non-voters’ ability to gain political knowledge and develop sophisticated policy judgements once they are allotted a legislative office. And in any case, our overall account does not depend critically on disconnected non-voter legislators being efficacious. The actions of disconnected non-voter legislators are only one of the channels by which Proportional Non-Voter Sortition transports non-voter views into the legislature. The actions of circumstantial and principled non-voters are another channel, and the incorporation of non-voter views into the programmatic platforms and legislative activities of established parties are yet another. So, even if the lame-duck worry should prove empirically accurate, it does not necessarily follow that Proportional Non-Voter Sortition is bad at transferring non-voter views into the legislature.

But will disconnected non-voters make it into the legislature in the first place? This is the core concern of a further potential criticism. In particular, the thought is here that the opt-out provision within Proportional Non-Voter Sortition enables non-voters who are disconnected, experience politics as a distant activity, or experience socio-economic disadvantage to disproportionally opt out of legislative service much like they opt out of voting and other forms of political participation. In turn, that self-selection can render the allotted legislators unrepresentative of the non-voting population at large, and critics might thus think that our specific design of Proportional Non-Voter Sortition actually has a limited capacity to further legislative inclusivity.

There are two things to say about this concern. For one, selection for legislative service during a full parliamentary term means getting access to an income, as well as a position of power and social status that otherwise remain unattainable for many people, especially for the underprivileged who are overrepresented among non-voters. So, one cannot straightforwardly presume that people will self-select for legislative service in the same socially biased ways as they do for other forms of (unremunerated) political participation. That said, the critics are fundamentally right: the opt-out clause indeed entails a risk that the allotted legislators self-select in ways that compromise the promotion of legislative inclusivity. Yet—and this is the second point—the opt-out clause seems necessary out of respect for people’s individual self-determination. People holding a legislative office must spend a large part of their time in the capital, away from their families and friends; they have to interrupt their careers and must shoulder responsibility for the common good; they must move in a context rife with ideological conflict and power struggles. This is not necessarily a life that everyone desires or is prepared to tolerate. So, as a concession to people’s individual self-determination, we think that Proportional Non-Voter Sortition must include an opt-out clause, even though it does pose a risk to the model’s capacity to deliver legislative inclusivity.

A further critical point relates to the choice of Proportional Non-Voter Sortition in the face of other policy alternatives: Is Proportional Non-Voter Sortition really the best means for pursuing legislative inclusivity? Critics might doubt this notably on the grounds that legislative inclusivity can be furthered also by an electoral system that is proportional. Proportional electoral systems are well known to strengthen the parliamentary representation of many different parties, especially as compared to majoritarian electoral systems (cf. Duverger 1954; Norris 2004, ch. 4), and to positively impact electoral turnout (Blais and Carty 1990). Moreover, a proportional electoral system may not involve the risks to governability and policy quality that can be associated with legislators who are selected by lot. And hence, critics might charge that the institutional part of our argument has selected the wrong policy tool: If legislative inclusivity is the objective, then Proportional Non-Voter Sortition is not the optimal means—proportional representation is.

This concern again invites two replies. First, while proportional representation can significantly further legislative inclusivity, it still has clear limitations. Proportionally representative electoral systems promote electoral turnout, but do not eliminate non-voting altogether. The example of Denmark illustrates this well. The Danish electoral system is strongly proportional (Gallagher 1991). Yet the Danish non-voting rate consistently ranges between ca. 15% and 20% (IDEA 2023). In other democracies with highly proportional electoral systems such as the Netherlands, Austria, and Sweden, the picture is similar. Thus, non-voting likely persists under proportional representation, and so it will only accomplish perfect legislative inclusivity if non-voters under proportional representation have views that are indistinguishable from those of voters. And while possible, this is unlikely.

Second and more importantly, the concern from proportional representation involves an important category mistake. Our aim is to consider how a democratic electoral system might appropriately absorb non-votes in the aftermath of a legislative election. The Count-and-Report Approach is one potential answer to this question, as is Proportional Non-Voter Sortition. But proportional representation is not. It is a particular way of aggregating votes, a specific model for translating cast votes into a distribution of legislative seats. And as such, proportional representation does not give any account of how the electoral system should absorb non-votes. Indeed, this is why in practice proportional representation systems virtually always treat non-votes according to the Count-And-Report Approach. So, even if proportional representation were a reliable tool for achieving full turnout and perfect legislative inclusivity, it is not a relevant policy to consider for present purposes—it does not belong to the category of policies that our analysis aims to evaluate.

This brings us to yet another potential criticism that is about popular control and argumentative incoherence. On our account, Proportional Non-Voter Sortition is desirable because it helps to further the principle of legislative inclusivity, which in turn is attractive for several reasons, including the reason that legislative inclusivity supports the value of popular control over government. Yet under Proportional Non-Voter Sortition, citizens cannot formally punish or reward the randomly allotted legislators. So upon consideration, is it maybe the case that our policy model actually undermines popular control and that our argument is incoherent?

We think not, for a combination of two reasons. First, Proportional Non-Voter Sortition is nested within a system of electoral legislative representation. The model provides that legislators are to be selected by lot only in proportion to the extent that enfranchised citizens do not vote. So, while concerns of popular control have rightly been raised in relation to models of sortition that seek to replace elections with sortition (cf. Landa and Pevnick 2021; Malleson 2018; Pettit 2012, 205), we insist that the case of Proportional Non-Voter Sortition is different. So long as non-voting rates do not go through the roof, this model entails that a large part of the legislators is going to be elected, not sorted, and citizens thus retain robust control over a significant part of the legislature. Second, Proportional Non-Voter Sortition enables citizens to vote the allotted rascals out. If citizens jointly find that the allotted non-voter legislators do not perform satisfactorily, then citizens can easily reduce their number all the way down to zero: they just need to turn out in greater numbers in the next election. So, while it is true that citizens cannot control allotted non-voter legislators in the short term, we emphasize that Proportional Non-Voter Sortition does give citizens robust control over the legislature in the medium and long run.

Finally, a last potential worry regards the scope of legislative inclusivity. Our argument for Proportional Non-Voter Sortition has all along assumed that the principle of legislative inclusivity extends so far as to apply to the political views of non-voters. But if this assumption is inaccurate, then we no longer have a case for Proportional Non-Voter Sortition. If non-voter views are not among the views that legislative inclusivity wants to see reflected in the legislature, then the principle does not particularly favor electoral arrangements that carry non-voter views into the parliamentary arena. The principle then becomes indifferent to such policies: it does not oppose them, but does not favor them either. So, are non-voter views within the scope of legislative inclusivity?

Some readers might worry that they are not, based on a reasoning from forfeiture or opt-out. More specifically, the potential idea is that non-voting is a way for citizens to remove themselves from the group to whom legislative inclusivity applies. And this in turn would suggests that while legislative inclusivity is important, it does not extend so far as to apply to the views of non-voters. Rather, the principle applies more narrowly to the views of people who exercise their franchise.

However, this reasoning is seriously vulnerable. Most importantly, it overlooks the distinctively institutional character of the legislative inclusivity principle. The notions of individual waiving, forfeiture, and opting-out are conceptually applicable in the context of principles that allocate individually exercised claims and entitlements. So, for example, it is conceptually fitting to employ the idea of waiving in relation to a principle that gives individuals a claim to bodily integrity, as when one says that a patient consenting to surgical treatment temporarily waives their claim to bodily integrity. But talk of individual forfeiture is quite amiss in the context of institutional principles that regulate the design and function of governmental institutions. Talking of individual forfeiture or opting-out in relation to the separation-of-powers principle, for example, is nonsensical. Individuals cannot, by their individual actions, opt in or out of such an institutional principle, and as we interpret legislative inclusivity, it is an institutional principle, too. On our account, legislative inclusivity aims to regulate the design of the law-making assembly, articulating a desideratum for its composition and relation to the citizenry as a whole. The principle does not say that citizens are individually entitled to have their political views represented in the legislature and as such legislative inclusivity is not amenable to individual waiving or forfeiture: as with the separation of powers, it is a category mistake to suppose that individuals could waive or opt out of that principle. ¹³

Additionally, even if it were the case that individual opting out from legislative inclusivity is possible, it is not clear that a failure to vote effects such an opt-out. Liberal democratic systems include multiple mechanisms for relaying citizens’ political views into the legislature and the process of policy-making. In addition to voting, citizens can write letters to their representatives, attend town-hall meetings, participate in demonstrations, sign petitions etc. And when people forego those non-electoral opportunities, they are not normally thought to opt out from the principle of legislative inclusivity. For example, we do not usually think that people who do not march in demonstrations opt out of legislative inclusivity altogether. And so, absent further argument it is far from clear that the failure to exercise one’s vote functions as an act of forfeiture or opt-out from the legislative inclusivity principle.

That not voting does not necessarily function as an act of forfeiture receives support also from the behaviour of non-voters themselves. Doppelt and Shearer (1999, 25) notably find that in the non-voter group they survey, nearly one-quarter has contacted their U.S. representative or senator. Some non-voters thus seem to retain a desire for political efficacy and seemingly do not interpret their own non-voting as an opt-out from the principle of legislative inclusivity. This is further reason to not take it for granted that non-voting is a way of removing oneself from the ambit of legislative inclusivity.

So, the reasoning based on forfeiture provides only spurious support for the claim that non-voter views do not fall within the remit of legislative inclusivity. But perhaps the claim can be argued differently? Maybe one can say that properly understood, legislative inclusivity does not require the legislative reflection of just any old view that citizens happen to have. Rather, it applies more narrowly to citizens’ considered political views, the judgments they hold after a discriminating reflective process that takes into account competing perspectives, empirical facts, and that keeps at bay cognitive vices such as motivated reasoning or wishful thinking. ¹⁴ And then one might add that non-voters’ views generally do not cross this bar, that their non-voting attests to the superficiality and knee-jerk character of their political reflection. This line of reasoning would offer a distinct, forfeiture-free explanation of why legislative inclusivity does not extend to cover the views of non-voters. It also resonates with some of the themes at the basis of deliberative democratic theory (cf. Sunstein 1991, Elster 1997). Yet the reasoning is not cogent.

Interpreting legislative inclusivity as a principle narrowly concerned with citizens’ considered political views is fair enough. But for the principle to remain reasonably consistent with the democratic commitment to universal enfranchisement, it must operate with a non-heroic, relaxed notion of consideration. Citizen views must count as considered at a relatively low threshold of critical scrutiny; otherwise it becomes hard to maintain that all citizens should have an opportunity for electoral participation. And this in turn creates a crux: To think that non-voter views consistently fail to clear a relatively low bar of critical reflexivity is not particularly plausible. The examples of Jaqueline or Clayton-Dixon rather suggest that if the notion of consideration is broad and relaxed, then non-voter views are in part going to count as considered. So, even if one accepts that legislative inclusivity is narrowly about citizens’ considered political judgments, that does not yield the conclusion that non-voter views consistently fall outside the scope of legislative inclusivity. This claim seems altogether untenable, whether it is argued by the means of forfeiture or by a restriction of legislative inclusivity that makes it apply to considered political views only.

Conclusion

Many enfranchised citizens of democracies do not vote in political elections. How should democratic electoral systems manage this phenomenon? How should electoral systems absorb citizens’ non-votes? With this article, we have sought to put this question on the table, and we have tried to answer the question with a specific institutional proposal tailored for the context of legislative elections—i.e., Proportional Non-Voter Sortition. In contrast to the standard Count-and-Report Approach to non-voting, this model brings it about that non-votes directly affect the distribution of parliamentary seats. The model provides that the legislature shall include a share of appointed legislators who are selected by lot from the non-voting part of the demos in a number that is proportional to the aggerate rate of non-voting. And this model is attractive, or so we have argued. Proportional Non-Voter Sortition incentivizes political parties to integrate non-voter views into their political agendas and legislative activities. It simultaneously enables the legislative articulation and advocacy of non-voter views through allotted non-voter legislators who descriptively represent non-voters. Thus, the model helps to further the important regulative ideal of legislative inclusivity, considerably more so than the standard Count-and-Report Approach, which gives political parties and legislators preciously little reason to care for the views of citizens who do not vote. Legislative inclusivity, in turn, is a valid, attractive ideal because on widely shared interpretations of the terms, it furthers the important democratic values of political equality, popular control over government, and epistemic government capacity, or so we have argued. And if that is all correct, there is an important (though conditional) pro tanto case for democracies committed to voluntary voting to radically amend the ways in which they absorb their citizens’ non-votes.