Is child disenfranchisement justified?

ABSTRACT Children are among the few social groups that are systematically and universally disenfranchised. Although children are citizens worthy of equal moral treatment and rights, their right to vote is restricted in almost all states, and this is seen as legitimate by most democratic theories. What is particular about childhood that justifies the restriction of their right to vote? How can democratic systems legitimise the exclusion of a section of their citizenry? This article provides a critical analysis of the principles that ground child disenfranchisement, and of the allocative mechanisms proposed for restricting the right to vote. It shows that most arguments in favour of child disenfranchisement are unsatisfactory, as they are inconsistently used to target the child population (but not other groups) or are incompatible with a commitment to equality. Among the principles for child disenfranchisement, only those that appeal to political incompetence may be valid for excluding children. Regarding the allocative method of exclusion, this article shows that age thresholds cannot be justified to restrict franchise, and that, while competence tests fare better in theory, there may be issues of implementation. The article contributes by providing a clear and systematic analysis of the core arguments in favour of child disenfranchisement, and the conditions that must be met to justify such a practice.


Introduction
Children are among the few citizen groups that are systematically and universally disenfranchised. 1 While in recent decades various countries have lowered their voting age to 16, this is an exception to the rule that being a minor means not having the right to vote. 2 This article explores the main arguments used to support the systematic disenfranchisement of children. It aims to show that most principles used to justify disenfranchisement are invalid, and that neither age thresholds, nor competence tests, are fully satisfactory as methods for exclusion.
Two elements are at play in justifying disenfranchisement: a claim about the principle for disenfranchisement, and a claim about the appropriate method to select a section of the population to disenfranchise. This article looks at these two elements separately, and critically analyses how the philosophical literature attempts to justify child disenfranchisement both in terms of principle and selection method. It, thus, answers to the justifiability of child disenfranchisement by breaking it into two separate issues: 1. Principle for Disenfranchisement: What are the legitimate reasons for restricting the right to vote to a citizen? 3 2. Method of Disenfranchisement: What mechanisms are available to legitimately select which citizens can be justifiably disenfranchised and which cannot?
While it may seem that the justifiability of child disenfranchisement is a binary affair (either it is justified or it is not), the subject is much more complex. The literature appeals, sometimes implicitly, to one of three principles to justify disenfranchisement of children: the unawareness principle considers that children are not affected by not having voting rights as they are unaware of what they are missing, so their equal standing is not harmed by disenfranchise; the best interest principle argues that it does not benefit children to bear political rights and responsibilities, as it imposes unjust burdens on them; finally, the harm to democracy principle argues that child enfranchisement threatens the legitimacy of a democratic system, due to children's political incompetence. This article claims that only arguments that appeal to harm can provide legitimate reasons for disenfranchisement. However, it argues that the justifiability of the harm principle depends on how the legitimacy of democracy is understood: if democratic legitimacy relies on ensuring political equality among its citizens, then child disenfranchisement cannot be justified; if democratic legitimacy relies on ensuring quality procedures and outcomes, then disenfranchisement can be justified.
Standing on 'harm to democracy' as the only principle that can potentially justify child disenfranchisement, the article looks at the three main methods proposed to exclude from franchise: age-thresholds as proxy for political incompetence, substantive competence tests, and procedural competence tests. It argues that age-based categorisations cannot be justified, especially if they are the sole indicator used to track incompetence. As to competence tests, in principle, both substantive and procedural tests are more legitimate mechanisms for allocating franchise (as they do not discriminate children based on age); however, their justifiability is affected by whether one gives more value to the equality or quality dimensions of democracy. If democracy is valued predominantly due to its epistemic capacity to produce quality outcomes, competence tests can be justified. But if democracy is valued predominantly due to its egalitarian credentials, competence testing cannot be used as a method for allocating franchise.
The article contributes to the literature by providing a clear and systematic analysis of the core arguments in favour of child disenfranchisement, and the conditions that must be met to justify such a practice. It claims that disenfranchisement is justified only if the legitimacy of democracy is threatened by political incompetence. However, age cannot be used as an indicator for tracking political incompetence; if political incompetence is of concern, it ought to be tracked for everyone regardless of age. Restricting the right to vote based on age is patently discriminatory, and should be abolished.
Section 2 analyses three principles used to justify disenfranchisement of children (unawareness, best interests, and harm to democracy). Section 3 assesses the validity of three mechanisms for exclusion from franchise (age thresholds, substantive competence testing, and procedural competence testing). Section 4 concludes.

Principles for disenfranchisement
To assess the justifiability of child disenfranchisement, we must begin by analysing the validity of the reasons given to restrict the right to vote to children. If democratic processes aim to recognise some form of equal standing among citizens, how can the exclusion of some citizens from franchise be justified?

Unawareness and self-respect
A first argument in favour of child disenfranchisement claims that children are not democratic equals as they are not affected by (or unaware of) being excluded from the political process. One could read Thomas Christiano's (2001) argument in favour of child disenfranchisement as relying partially on this justification. 4 Christiano claims that, beyond political incompetence (see Section 2.3), children are justifiably excluded from the democratic process as 'their status is not undermined by exclusion from collective decision making about matters pertaining to justice' (Christiano, 2001, p. 207): children are unaffected by exclusion as they lack awareness of the 'facts of judgment' (Ibid.).
López-Guerra (2012) follows a similar line of argument, while focusing on the individual's interest in self-respect. He claims that enfranchisement has a distributive and a relational value. Unequal distribution of voting rights affects individuals' interests because it, first, provides unequal opportunities to affect decision making; and, second, because they harm the self-respect of those who are not granted equal political rights (López-Guerra, 2012, p. 126). While all citizens would be unjustly treated due to the distributive inequality of disenfranchisement, not everyone is unjustly treated based on voting's relational value. Children under 10-years-old, he argues, even if harmed in distributive terms, are not misrecognised by disenfranchisement as their selfrespect is not affected by not being allowed to vote (López-Guerra, 2012, p. 127).
Christiano and López-Guerra thus claim that individuals whose self-respect (or lack of awareness of the facts of judgment) is not affected by disenfranchisement may be justly disenfranchised. This misses an important element of the problem of respect: unawareness of being disrespected does not preclude disrespect. Both authors assume that the harm caused by lack of inclusion is conditioned by the subject's awareness of her mistreatment and the harm it causes. This is far from true. Relying on an individual's own awareness of their disrespect for it to be considered an injustice has problematic implications, particularly for the protection of the vulnerable and dependent. Take, as an example, laws that protect children from sexualisation through prohibitions of child pornography or sexual exploitation. These protections are especially stringent precisely because most children who are sexually exploited are unaware of the objectification and harm caused by being treated as sexual objects. It is their lack of awareness of being treated sexually what makes their sexualisation so problematic. If one appealed to unawareness as justifying disrespect in this case, the whole system that protects children (or individuals with mental disabilities) from sexual exploitation would be lost, especially in the most terrifying cases, as younger children are less aware of the harm that is inflicted on them through disrespect. I believe that appealing to unaffectedness as the principle that justifies disenfranchisement is unsatisfactory, and can have grave consequences on the interpretation of other cases in which lack of awareness to harm is present.
Philip Cook's proposal for a minimum procedural competence test (2013) can be considered as a partial endorsement of the unawareness justification for disenfranchisement, while not relying (entirely) on the argument presented above. Cook argues that disenfranchisement can be justified by defining a minimal procedure through which the individual must pass in order to vote (tracking literacy and independence; Cook, 2013, pp. 450-455). The procedure limits enfranchisement only to those aware of it, while filtering out those who would be unaffected by being disenfranchised: 'As those incapable of completing these tests would be ignorant of the test, they would not suffer harm to self-esteem' (Cook, 2013, p. 454). Cook proposes a radical change in the threshold for inclusion, expanding enfranchisement to almost all the population, while still justifying disenfranchisement for those whose illiteracy or dependence makes them unaffected by (or unaware of) their exclusion.
While Cook's is one of the most consistent and fair proposals in the literature, I still wish to raise an issue with his justification for disenfranchising those who do not pass a procedural test: namely, its irrelevance. If the procedural test aims at excluding those, and only those, who are entirely unaware of their exclusion, it seems that the test is unnecessary as it filters out individuals who would most probably not be a part of the procedure anyways. What would change if they were enfranchised, and simply did not vote? Within Cook's own understanding of the competences required for enfranchisement, recognising the right to vote to those unaware and unaffected by disenfranchisement (who would be a very small fraction of the population) would not have an impact on the democratic outcome whatsoever (Hinze, 2019, p. 297), while it would play an important symbolic role in recognising all citizens as equal members in society.
The question of principle, thus, cannot be answered by appealing to children's unawareness and unaffectedness of being excluded from franchise. First, assuming unawareness of harm as legitimising disrespect is highly problematic; second, the unawareness claim (if true) would only apply to a minority of the child population, making disenfranchisement irrelevant.

The best interests principle
An alternative argument claims that franchise harms children's fundamental interests (Beckman, 2009(Beckman, , 2018. The principled justification is that enfranchisement would harm children's best interests. The concept of 'best interests' grounds Beckman's claim that children's interests may be harmed by enfranchisement. He argues that children's political rights ought to be determined primarily with reference to their best interests (Beckman, 2009, p. 92). Following the Convention on the Rights of the Child (CRC; UN General Assembly (UNGA), 1989, Art. 3.1), children's best interests must be the primary concern when making any decision that may affect them. Thus, whether children ought to have voting rights should rely on an assessment of the benefits and harms that this may incur. Due to children's vulnerabilities and developmental needs, primary concern should be given to understanding the (positive and negative) impact that voting has on their fundamental interests (Beckman, 2009, p. 114;see also, Clayton, 2006, p. 183).
Relying on the 'best interests' principle, Beckman follows the literature on the 'special goods of childhood' (C.M. Macleod, 2010;Gheaus, 2015) to argue that granting political freedoms to children, with the duties this entails, threatens the protection and promotion of a core interest that children have in being children. Beckman uses the moral wrongness of child labour as an analogy: 'The reason why children have no moral right to take employment is not that children are particularly bad workers but that to work is bad for them.' (Beckman, 2018, p. 291). It is not that children are necessarily incapable of voting; rather, the claim is that certain goods specific to 'being a child' (i.e. freedom to play, freedom from political burdens, and innocence from adult affairs) would be compromised by their enfranchisement. Voting would impose responsibilities on children, limiting the realizability of the goods of childhood: children would bear the burden of making decisions over society as a whole, they would be compelled to use their free time to acquire information and knowledge on political options and on deliberation over their decisions (Beckman, 2009, p. 115; see also, Schrag, 2004, p. 373). It is, thus, claimed that it is not in children's best interest to engage with political questions and bear political responsibilities; their best interests justify their disenfranchisement.
There are three problems with this line of argumentation: one lies in its interpretation of the 'best interests' principle, the second is empirical, and the third substantial. First, Beckman's interpretation of the support that the CRC's 'best interests' principle gives to his argument is flawed. 5 Part of the reason why the 'best interests' principle exists is for it to partially compensate for children's lack of political rights (Eekelaar & Tobin, 2019). The 'best interests' principle requires accounting children's own views and claims in order to decide what is best for them. This means that children's interests cannot be conceptualised unidirectionally by external actors, as Beckman presumes; within the CRC, children must be included in the deliberation over what these interests are, and how to realise them (Eekelaar, 1994; UN General Assembly (UNGA), 1989 Art. 12).
The 'best interests' principle is not a pre-emptive justification for unilaterally deciding what is best for all children (Eekelaar, 1994); it is a dynamic principle that should adapt to the particular circumstances of particular children, balancing the reasoning of experts with children's own views on what their own interests are (Skivenes, 2010). It is not mean to be used as a ground for generalisable prescriptions for the whole childhood population; but rather to accommodate what justice requires to individual circumstances. Claiming, as Beckman does, that the best interests of all children would be necessarily harmed by enfranchisement obviates both children's own views on the subject, and the variable impact that franchise can have on children in different stages of development (Archard & Skivenes, 2009).
One could avoid the above issue by not appealing to the CRC, and state that the argument relies on a paternalistic definition of children's best interests (Beckman, 2009, p. 114). It would, however, still fall prey to an empirical and a substantive critique. As to the empirical issue, Beckman unduly inflates the burdens and responsibilities that come with voting to make his point. He claims that children's interests in free time and play would be irremediably harmed due to the responsibilities of being enfranchised (Beckman, 2009, p. 115). This is not necessarily the case. If we assumed that every voting citizen has the duty to acquire all necessary information and to deliberate in a fully reasonable and rational way, then, not only would children not have any free time left, but possibly everyone else would have to spend all their time deliberating politically. No one expects the individual citizen to bear the burden of democracy to this extent (López-Guerra, 2012, p. 130). Imposing this requirement on children when the rest of the population does not comply with it would be patently unjust. An even if it were so, the fact that children are enrolled in schools can palliate the threat of this responsibility taking over their free time and play (Umbers, 2020, p. 738). One could conceive of civic education classes during school time (which is already not free nor time for play) targeted particularly to children's acquisition of political information, and as a forum for debate on the different positions and candidates on the ballot (Franklin, 2020;Mahéo & Bélager, 2020).
Even if children's interests in free time and play are not affected, a 'best interests' argument could claim that political innocence, and an interest in being shielded from the 'adult world' could be threatened by enfranchisement. As Rehfeld rhetorically asks: 'Do we want, for example, a 12-year-old to deliberate about how best to deal with female circumcision in Africa or debate whether waterboarding is a good or bad policy of interrogation?' (Rehfeld, 2011, p. 153). The claim is that children have a fundamental interest in being innocent; in being shielded from certain realities that political rights uncover (C. Macleod, 2015).
This line of argument is problematic for more substantial reasons than the former. Appealing to innocence to justify the restriction of fundamental rights begs various questions: can 'innocence' be defended as an objective and universal value that justifies such restriction? Shouldn't children themselves have a say in what constitutes their interest in innocence? Using the political discourse of 'innocence' to restrict a group's rights has bad pedigree. Jane Rutherford (1997Rutherford ( , pp. 1479Rutherford ( -1489, for example, showed how the discourse of 'protecting innocence' was used against the suffragette movement during the turn of the twentieth century. Although it could be argued that the case of children is different from that of women (children are in fact innocent, while women are not), there are still problems in applying it to this case (Jenks, 2005, Ch. 6). 6 As Joanne Lau argues, '[c]alling someone 'innocent' is often a covert form of oppression; it is a failure to take him or her seriously as a political agent.' (Lau, 2012, p. 861). Innocence implies a refusal of having to justify to those we deem as 'innocent' why we keep them in this state; while, at the same time, not providing any actual content to what 'innocence' entails, or why it is beneficial to be innocent. The concept proposes a highly moralised conception of what 'childhood' should be, assumes it as true (and objectively good), and then imposes it unilaterally. Each of these steps would require thorough justification for the argument to hold. First, one must justify the imposition of a WEIRD 7 conception of 'childhood' without accounting for the variability of who children are (Lancy, 2015, pp. 397-399). Children are a radically heterogenous group (depending on their age, where they live, and other intersections); thus, claiming that innocence from the adult political world is good for them (in general) fails to account for this plurality of the childhood experience, and how engaging with reality may benefit or harm them (Jarkovská & Lamb, 2018, p. 78). Second, one must show why political innocence is necessarily good for the child population. Many children do not have the option of 'being shielded from the adult world': they are caretakers, workers, heads of households; they are treated differently because of their gender, race or religion; they care about climate change, social justice, war. Many children live already (for better or worse) in the adult world, and disenfranchising them will not change this situation, nor make them innocent again. I do not see how disenfranchisement can be justified based on children's best interests when children are forced to live in adult worlds, or when they value being political.
An argument for disenfranchisement that relies on the 'goods of childhood', and on a paternalistic understanding of children's best interests does not stand. It does not account for children's own voice and agency in defining what their best interests are; it overestimates the harms that may be imposed on children by voting; and it assumes that interference is justified regardless of the particular condition of different children.

Harm to democracy, and political incompetence
If the act of voting affects the legitimacy that the democratic process is meant to secure, then it might be justified to withhold franchise from those who can harm the system. Most (if not all) democratic theorists agree, at least in principle, that harm to the legitimacy of the democratic system can justify disenfranchisement. As voting gives political power to the individual, it allows voters to wield coercive authority over others (Brennan, 2011). Certain voters can vote 'wrongly', and if collectively the majority vote 'wrongly', this would threaten the legitimacy of democracy.
The 'wrongness' of voting can be interpreted in two ways: as generating 'bad' outcomes, or as being construed in the 'wrong' way (Estlund, 2008, pp. 160-167). While some democratic theorists directly mention bad outcomes as a potential problem with the democratic system (i.e. Caplan, 2006;Estlund, 2008;Goodin and Lau 2011), most agree that the threat to the legitimacy of the system lies in citizens voting through corrupt cognitive processes (i.e. ignorance, bias, irrationality, unreasonableness). The assumption is that, if the individual's cognitive process of deliberation is flawed, the system is corrupt, and the results are illegitimate (Brennan, 2016). 8 In short, voting wrongly can either lead to bad political outcomes that harm others, or it makes the system lack legitimate authority to rule.
Some form of harm can justify limiting the political power of individuals. If a democratic system cannot protect its citizens from the worst public bads, or if it systematically elects the worst possible candidates and the worst possible policies, what is the use of having a democratic system (besides its symbolic value) (see, Brennan, 2016: Ch.1)? Harm to democracy, thus, seems to be a justified principle for disenfranchisement, and one can see how children may fit into this picture: if it can be shown that children vote 'wrongly', their disenfranchisement might be justified.
A just cause, however, is insufficient to disenfranchise a particular collective. Justifying disenfranchisement requires showing that the morally relevant traits track the outcome one wants to avoid. In this case, political incompetence is considered the most relevant trait that traces an individual's proneness to cause harm to democracy through voting. 9 Choices made through the democratic process have an ample impact on everyone in society, and, so the argument goes, those who are allowed to vote should be competent enough to do so.
Individuals who are less competent in doing X are more likely to cause harm to others while doing X than individuals who are more competent (Christiano, 2001, p. 197). Less competent voters are more likely to harm through their votes than more competent voters, thus, a distribution of political power that tracks incompetence is justified to ensure that democratic decisions are made only by those competent to do so (Brennan, 2011, p. 709). Political powers and privileges are conditioned and limited by the harm they may inflict on others. If one does not have the competence required to ensure that one's actions do not cause relevant harms, it is legitimate to have one's actions restricted. In the case of children, the capacities required for being politically competent come with maturity, thus, it may be legitimate to disenfranchise the youngest population as they are those with highest probability to cause harm through their low political competence.
What does it mean to be 'politically competent'? A politically competent person is characterised, not by the content of their political choices, but by the process through which they make these political choices. As Christiano argues, political competence 'is a disposition of a person. It does not entail always having the right answer. One may be morally competent but make mistakes or have the wrong moral views . . . or act irresponsibly.' (Christiano, 2001, p. 200). It is not necessarily about choosing the 'correct' option on a ballot; it is about choosing through the 'correct' cognitive process. It entails reaching a certain level of rationality and reasonableness, which allows one to form and revise one's conception of justice and the good, and to understand and abide by fair terms of cooperation (Fowler, 2014, pp. 99-100). It entails a minimal ability to understand what the democratic process is, and one's responsibility in being part of this process (Archard, 2004, p. 101;Chan & Clayton, 2006, pp. 538-542).
The acceptance of political incompetence as justificatory of disenfranchisement assumes the epistemic reliability of the democratic system: democracy has value because it can produce correct political outcomes more reliably than alternative political systems. This assumption is based on Condorcet's 'jury theorem': if a voter's has fifty-one percent or higher probability of choosing correctly among a set of options, then the system can be expected to produce correct outcomes reliably (Goodin & Lau, 2011). For this to be achieved, however, it must be ensured that the average voter's probability of making a correct choice is better than a fifty-fifty chance (see, Estlund, 2008, Ch. 8, esp. 156-158); this requires ensuring that the majority of the electorate is not politically incompetent (with fifty-fifty or less probability of making a correct choice). Children, it is argued, due to their low political competence, would lower the quality of democratic choices to be worse than fifty-fifty, thus, must be excluded, as the epistemic reliability of the democratic system requires it (Chan & Clayton, 2006).
One can respond in one of two ways to this claim: first, by showing that the quality of the democratic process would not be harmed by including the politically incompetent; or second, by appeal to the egalitarian value of democracy, claiming that the outcome is secondary if in tension with the principle that all should be recognised as equals in the democratic process.
Many defenders of the enfranchisement of the 'incompetent' have argued, based on Condorcet's 'jury theorem,' that the inclusion of children and other 'incompetent' collectives would not have a negative impact on the quality of the democratic process as the probability of the average voter choosing incorrectly is inexistent (Goodin & Lau, 2011;Munn, 2018). If the quality of the process is of concern, then individual political competence is irrelevant; only the collective competence of the whole electorate (understood as average competence) can allow us to assess whether the outcome would be reliable (Hinze, 2019). Children's votes would not reduce sufficiently the average individual competence to affect the outcome; children's votes would be, at worst, random, thus simply increasing the number of votes without affecting the outcome (Goodin & Lau, 2011, p. 160), or, due to the influence of competent voters (such as their parents, teachers, etc.), their political choices would reflect those of informed voters, thus, actually improving the overall quality of the outcome (Olsson, 2008).
The Condorcetian response is appealing as it shows that, if most voters are sufficiently apt at choosing a correct option, it is irrelevant whether the rest are competent or not. The response, however, has limitations: first, it requires average competence of the collective to be sufficiently adequate for it to withstand incompetent voters; and, second, its focus on the quality of the outcome does not answer to the claim of legitimate authority. 10 Regarding the first point, conceding to the correctness of the jury theorem, if average individual competence has a fifty-fifty chance, or worse, of making correct choices, then the system cannot withstand more incompetent voters (Brennan, 2016: Ch. 2). As the Condorcetian argument relies on the assumption that the average competence of the electorate is in fact sufficiently adept in making correct choices, its validity depends on this empirical assumption being true: if the average voter votes badly in a systematic way (due to incompetence or moral corruptness) then the system cannot withstand the inclusion of a larger, potentially incompetent electorate. Moreover, it would actually require increasing the threshold of competence (or moral righteousness), disenfranchising many more, to ensure average individual competence to be better than random.
Going further, Jason Brennan (2011) argues that whether the outcome is correct or not is irrelevant to justifiability, as political incompetence makes the procedure illegitimate. Competence is fundamental to voting, not only because it improves the chance of better outcomes, but because the legitimate authority of the process depends on it. Brennan (2011, pp. 702-705) uses the legitimacy of jury decisions as an example that proves the moral relevance of competence. He asks whether a jury that we know as ignorant of the facts of the case, or one which is incapable of understanding what the case is about and what its consequences are, can be considered as a legitimate authority (Brennan, 2011, p. 708). He claims that, regardless of what the outcome is (that is, whether the jury's decision is right or wrong systematically), the jury must be considered as illegitimate as no system of justice should rely on individuals who are incompetent making these high stakes choices. Brennan claims that the case of voting is analogous to juries; only those deemed capable of voting rationally, consciously and knowledgably should be allowed to do so; the authority and legitimacy of the system depends on this.
Brennan's legitimacy-based argument is strong. However, the validity of this claim presupposes that democracy merely has an instrumental value in ensuring quality political outcomes (Brennan, 2016, pp. 10-14). Democracy, for more proceduralist theorists, is valuable because it is a political system that recognises and reflects the equality of its members beyond its instrumentality (i.e. Christiano, 2002;Kolodny, 2014). In this sense, the right to vote 'is a symbolic affirmation of one's political status even when it seems to have little instrumental value for an individual.' (Archard, 2004, p. 98). Seeing democracy through this lens would imply that the quality of the outcome should be secondary to the egalitarian condition that democracy stands on. Legitimate democratic authority, a proceduralist would claim, does not arise from quality outcomes, but from the recognition by the process of the equality of all citizens regardless of competence (Goodin, 2007).
One can agree that the egalitarian rationale of the procedure has fundamental value, while defending the need to account for the instrumentalist challenge. Would a proceduralist abide by a democratic process so incompetent (or morally corrupt) that would undermine the principles upon which it stands? Conceding to the egalitarian procedural value of democracy, the threat to the legitimacy of the system by an excess of political incompetence or moral corruptness must be taken as a limit to what the procedure can withstand. Some minimal conditions that safeguard the legitimacy of the democratic system from widespread incompetence or corruptness must be in place to secure its stability.
What does this entail? The weight one assigns to different values of democracy determines to a great extent the scope of justifiability of disenfranchising based on political incompetence. Does democracy reflect the equality of those citizens who meet the competence for being treated as equals (as instrumentalists argue)? Or does democracy recognise and symbolise the equal political status of all citizens regardless of an individual's condition (as proceduralists maintain)? 11 For the task at hand, it suffices to say that, on the one hand, understanding democracy as mainly instrumental can justify more stringent requirements for inclusion to the franchise; on the other, valuing democracy primarily for its egalitarian procedures, while it shrinks the justified scope of disenfranchisement based on incompetence, it still must concede to ensuring that the legitimacy of the system is protected from systematic corruptness or incompetence. Whether procedural or instrumental, widespread political incompetence is a threat that no democratic system can withstand; disenfranchisement that prevents this scenario is, thus, justified.

Method of disenfranchisement
Following the previous section, only threats to the legitimacy of the democratic system through political incompetence can justify restricting franchise for children. Having determined the principle that could ground disenfranchisement, we must explore the justifiability of the mechanisms available for classifying who may be included or excluded from franchise based on incompetence.

Age as a threshold
Age is the standard metric used to track political incompetence and enforce child disenfranchisement. If you are below a certain age (usually 18-years-old) , you are deemed legally incompetent to exercise the right to vote. But why can age be used to justify disenfranchisement? Age is not morally relevant in itself; it is useful only as a proxy that tracks political (in)competence. Competence variability between a 17 and an 18-year-old is so subtle (if it even exists) that it would be arbitrary to judge every 18-year-old individual as competent and every 17-year-old as incompetent. The cognitive and emotional development that occurs with age is so fluid and variant from person to person, that using any particular point as a definite and accurate threshold seems highly questionable (Fowler, 2014, p. 100;Cook, 2013, pp. -444-445).
What justifies, then, treating individuals above and below an age threshold differently? Two arguments are given. First, the justifiability of statistical discrimination: even if age is not an accurate distinguishing trait, the statistical probability that individuals at each side of the threshold will behave as expected (underaged as politically incompetent, adults as competent), proves the proxy as reliable even if not precise (see, Schauer, 2018). Second, one can appeal to complete-life egalitarianism to respond to concerns that age proxies are discriminatory: a commitment to equality does not entail treating individuals equally at each moment in their life; equality only requires individuals being treated equally across their whole lifespan. If everyone's voting rights are restricted below a certain age, complete-life equality is secured.
Using age as the threshold for disenfranchisement relies on the existence of a relevant correlation between age and the development of the competences deemed necessary for voting (Archard, 2004, p. 89). Even if unprecise tracking political incompetence, it is satisfactory as it is easy to implement, transparent, and sufficiently reliable if one considers the difference in competence between individuals of a very young age and middle-aged adults (Hinze, 2019, p. 292). Defenders of an age threshold, such as Archard (2004, Ch. 6), argue that even if there is contention regarding the accuracy of age at the boundaries of the threshold (the difference between a 17 and an 18-yearold are insignificant), the progressive acquisition of competences, and clear difference in competence of the cases at the extremes proves the justifiability of the threshold. The validity of speed limits while driving works as an analogy: we need speed limits because of the risk of accidents when speeding; despite that risk does not vary if driving at 99 or 100 km/h, the difference between driving at 100 or 30 km/h proves the justifiability of a threshold, even if not precise (Archard, 2004, pp. 86-89). Same with voting: the obvious incompetence of a one-year-old justifies the establishment of a threshold even if not fully accurate at the boundaries; 'What matters is the overall balance of probabilities. We need only to be confident that the competence is most probably not possessed by those in one age group and most probably possessed by those in the other.' (Archard, 2004, p. 89).
Age thresholds do not count as instances of wrongful discrimination because they rely on statistical correlation to justify its differential treatment (Clayton, 2006, p. 186). Opposed to discrimination based on gender or race, which are triggered by biases or prejudices against a collective, discrimination based on age uses statistical correlation of age and political competence to determine its threshold (Chan & Clayton, 2006, p. 539). The method is not unjust because it accurately tracks the probability of incompetence collectively, even if it does not individually. Results of a statistical study by Chan and Clayton (2006) exploring political competence of individuals at different ages shows a correlation between willingness to participate in political activities, ability and knowledge on politics and democracy and age. Thus, they argue, current age limitations on voting are justified. 12 The argument for statistical discrimination has problems, particularly in how it applies to age-based disenfranchisement: first, statistical data, even if accurate, can be self-fulfilling and spurious; second, even if not discriminatory in itself, it may be discriminatory if applied exclusively to the under-18 age group.
Statistical discrimination is particularly problematic as it may suffer from circularity and self-fulfilment by relying on information taken from a system that discriminates in order to assess whether the discrimination is justified (Lister, 2007;Lau, 2012, p. 863). An important reason why minors are statistically less competent and interested in politics is not because of their inherent lack of capacities, but rather due to the absence of incentives provided for them to develop these capacities (see, Peto, 2018). Imagine if Chan and Clayton's (2006) survey of political competence had been used to assess the political interest and knowledge of men and women prior to female enfranchisement in the early-nineteenth century. Most surely the results would show that the female population had a radically lower interest in and capacity to understand and address political questions than the male population. But this is not an objective assessment; it does not prove women's counterfactual political incompetence; it seems self-fulfilling as it merely shows that, under conditions in which it was close to impossible for women to acquire political competence, women did not possess said competences.
Proving the current political incompetence of children under restrictive conditions is insufficient to justify their disenfranchisement. These types of surveys show that we need to improve the conditions that would allow children to become politically competent (Peto, 2018, p. 282). If individuals have the potential to become competent, they should have ensured the conditions needed to develop this ability (Brando, 2020). Empirical research on political competence shows that an outstanding difference exists between the competence of enfranchised and disenfranchised minors; comparison between political competence of adolescents in Austria (where 16-year-olds are enfranchised) and the UK (who are disenfranchised) shows that enfranchisement makes a difference in how competent minors are (i.e. Mahéo & Bélager, 2020;Wagner et al., 2012). What this means is that without the existence of a counterfactual scenario in which we can assess political competence without restrictive background conditions, the claim of statistical reliability is spurious, and, thus, questionable as justification for disenfranchisement (Mahéo & Bélager, 2020, p. 5, 18;Peto, 2018, pp. 290-292).
A second problem with the current use of statistical discrimination is its inconsistent application to different age-groups (or to other indicators relevant to political competence). Even if statistical discrimination were justified, the fact that it is used exclusively to restrict the rights of one collective can be unjust (Hinze, 2019, p. 295). Assuming that age-based statistical discrimination is valid, Joanne Lau (2012) has argued that it can only be implemented if it applies symmetrically to the elder population at the other end of the age spectrum as well. She claims that if age is relevant, it should be used as an indicator of incompetence at all stages of life, not only childhood (Lau, 2012, pp. 864-865). The only justified way to use age as a proxy for incompetence is by tracking its correlation across the whole lifespan and disenfranchise age groups according to their statistical incompetence (Fowler, 2014 proposes a similar system). This revision of the statistical discrimination argument, while not fully successful addressing the issue of using age as an indicator, at least overcomes one of the issues that affect disenfranchisement based on age: the fact it only applies to the youngest age-groups. 13 Following complete-lives egalitarian theory (Daniels, 1988), some defenders of age thresholds for disenfranchisement Fowler, 2014;Weale, 1999) argue that differential treatment based on age is justified, opposed to discrimination based on gender or race, because age discrimination affects everyone equally. '[D]iscrimination on the grounds of age is fundamentally different from discrimination because of gender or race. Since everybody ages, policies that appear to be unfair to elder or younger groups might in fact be justifiable when looked at over the whole lives of the individuals concerned.' (Fowler, 2014, p. 104ff). We all pass through the same stages of life in which our voting rights are restricted, and this restriction lasts the same for everyone; it does not violate equality and may be considered as justified allocative criteria for differential treatment. If what matters is whether individuals are treated equally across their complete lives, the restriction of certain rights (such as voting) based on age does not violate the ideal of equality, and is, therefore, justified (Clayton, 2006, pp. 188-189).
Appealing to complete-lives egalitarianism may be sufficient to show how allocation of rights based on age complies with equality while other forms of differential treatment do not, but it fails to provide sufficient justification as a method of disenfranchisement as it does not offer an argument in favour of any particular allocation of voting rights (Umbers, 2020, pp. 746-747). The complete-life egalitarian argument is insufficient to legitimise that the present age-based system in which only children are disenfranchised is the appropriate mechanism to abide by equality. Because it focuses exclusively on ensuring equal treatment across the life-span, many allocative mechanisms besides the current would comply with its tenant: imagine a system in which one has five years of enfranchisement then five years of disenfranchisement, or another in which only the elderly are disenfranchised, or another in which random ages are chosen to disenfranchise (López-Guerra, 2012, p. 125;Lau, 2012, p. 868). From a complete-life egalitarian perspective, there is no difference in choosing one or another; its minimal focus on the protection of equality across life does not suffice to justify specific allocative methods, nor to prove why it should only track children.
Even if age works as a simple, easily enforceable, and clear allocative mechanism for disenfranchisement, its lack of accuracy, and its reliance on dubious statistical information makes it a morally problematic method for tracking political incompetence.

Testing competence
Theorists who maintain that political incompetence justifies disenfranchisement, but doubt the validity of age proxies, have explored competencetesting as an alternative threshold for exclusion. The arguments against age proxies looked exclusively at how children are discriminated or harmed by their age-based disenfranchisement. However, the problem with using age as a proxy for political incompetence goes beyond its negative impact on the disenfranchised population; it also bestows an unfair privilege to the politically incompetent who are not tracked by age (Munn, 2018). If political incompetence is the morally relevant trait that justifies disenfranchisement, then the metric used to track political incompetence ought to track it always, not just some instances of it. Even assuming that all individuals below the age of 18 are indeed politically incompetent and, thus, justly disenfranchised, as a matter of consistency we still ought to track and disenfranchise those above the age of 18 who are politically incompetent (Lau, 2012;Munn, 2018). If political incompetence threatens democracy, then every politically incompetent citizen should be disenfranchised.
Just as it is credible to argue that some (or most) children are indeed politically incompetent, it would not be ridiculous to claim that there is a threatening section of the enfranchised population who by any account of political competence would qualify as incompetent, thus, a threat to democracy. If failure to abide by public reason and the 'facts of judgment' is an instance of political incompetence (Christiano, 2002); if ignorance of basic facts about politics, candidates and issues is an instance of political incompetence; if proneness to manipulation, bias, and an incapacity to make independent political choices is an instance of political incompetence, then, as Brennan argues, not only should many (or all) children be disenfranchised, but a majority of the adult population should be disenfranchised also (Brennan, 2016;see also, Caplan, 2006).
What these competences are, what their threshold should be, and how they should be assessed, however, are contentious issues. One can divide, for simplicity, between defenders of substantive and procedural competence tests (Christiano, 2001;Cook, 2013). Endorsers of substantive tests of competence consider that people's political competence must be evaluated through testing knowledge about the facts of the political and the electoral systems, assessing biases in reasoning, the capacity to understand the issues, and the consequences of political choices (Brennan, 2011;Munn, 2012). Substantive tests impose a high threshold to enfranchisement by ensuring that only fully reasonable and rational individuals can vote.
Procedural tests, on the other hand, filter incompetence out of the suffrage while keeping the threshold low enough for democracy's commitment to equality to not be affected. Instead of demanding 'ideal' political competence from its citizens, procedural tests filter through transparent and uncontroversial means whether a citizen is competent to vote. Cook, for example, proposes a purely procedural test in which the act of registering to vote and voting independently are themselves the procedure that determines whether an individual is politically competent (in this minimal sense; Cook, 2013, pp. 450-454). The procedural test assesses literacy through an individual's act of registering to vote, and assesses independence through an individual's act of voting privately in a booth. These are, for Cook, sufficient and minimal requirements for voting; nothing else should be expected from an individual to justify their enfranchisement.
A couple of problems have been raised with competence tests. First, the question might be raised as to whether tests should track the ideal competence for democratic politics, or if they should track the minimum requirements that could be expected from the citizenry (Archard, 2004, p. 101-102;Fowler, 2014, pp. 96-97). It is claimed that appealing to the first, ideal, understanding of what a citizen should know about politics, and how she should reason about politics, asks too much from the general population. If democracy is valued both for its instrumental reliability and for its egalitarian distribution of political power, a substantive test with a high threshold would threaten its egalitarian credentials, leading to an epistocratic system which few true democrats would be willing to endorse (see, Umbers, 2020, p. 742). If political equality is a valued tenet of democracy, imposing such high requirements for the franchise would be inconsistent with its core commitments.
While the above critique hints at procedural tests as better suited than substantial ones as an allocative method for franchise, a second issue affects both substantive and procedural tests: they may be biased against certain populations, incentivising abuse and discrimination (Archard, 2004, pp. 90-91). Biased tests demanding literacy were used in the past to exclude black Americans from U.S. elections (Rutherford, 1997;Cook, 2013, p. 442). Even today, voter ID laws, and voter registration procedures can be biased against social groups, affecting effective enfranchisement (Braconnier et al., 2017;Hajnal et al., 2017). Although such policies are not necessarily crafted with discriminatory intentions, they can end up being discriminatory as there is a high chance that the competences required to pass the test are more prominent in people with higher education, and higher socioeconomic status, thus excluding whole socioeconomic classes, or sections of the population (Estlund, 2008, p. 217-218;Umbers, 2020, p. 749). Brennan (2011, p. 720) has responded to these claims by arguing that this is not an injustice caused by the tests themselves, but rather that the tests reflect an already existing injustice: lack of access to education and political resources for certain sections of the population. In theory, ensuring a fair distribution of educational and political resources would compensate this potential injustice. 14 These discourses bring to the fore an important tension, and a potential inconsistency, at the root of mainstream democratic theory: democracy's attempt, first, to strike a balance between its procedural and instrumental values, and, second, to be consistent and fair in the way it does so, means that there may be important conceptual trade-offs that are not straightforward, and which tend to be omitted when discussing the case of child enfranchisement (Hinze, 2019). On one side, if the instrumental value of democracy is considered as sufficiently important to trump equality, then political competence must be taken seriously as a trait that has to be tracked and tested, to ensure the sustainment of a reliable and legitimate democratic system. If, however, the egalitarian procedures of democracy are considered its most fundamental quality, then democracy must be willing to give up a certain level of instrumental reliability, and attempt to lower the threshold of expected competence in order to enfranchise most (if not every) citizen.

Concluding remarks
This article explored the main arguments that attempt to justify the disenfranchisement of children. It considered three principles for disenfranchisement (unawareness, best interests, and harm to democracy), and argued that only arguments that appeal to harm to the democratic system based on political incompetence can justify the restriction of franchise. Based on this, it studied two methods used to select which groups count as 'harming democracy' (age thresholds, and competence tests). It claimed that age thresholds are unacceptable for disenfranchising, and considered the credentials of substantive and procedural competence tests in assessing political incompetence. Neither test is fully justifiable, and most of their validity depends on how the value of democracy is understood. Instrumentalist approaches to democracy are more amenable than proceduralists for justifying disenfranchisement based on political incompetence. If democratic equality is a primary concern, the justifiability of child disenfranchisement is weakened (even through competence testing). In any case, child disenfranchisement can only be legitimate as long as the method for exclusion is not discriminatory, and as long as it tracks and restricts the right to vote to all individuals who do not meet the competence requirements, regardless of age. Notes 1. 'Child', following the Convention on the Rights of the Child (CRC), is 'every human being below the age of 18 unless under the law applicable to the child, majority is attained earlier.' (UN General Assembly (UNGA), 1989, Art. 1). Used as synonymous of 'minor' or 'underage', as opposed to 'adult'. It includes all stages of development till majority (toddlers, children, teens and underage youths). 2. Argentina, Austria, Bosnia-Herzegovina, Brazil, Cuba, Ecuador, Nicaragua, Serbia, and Slovenia (as well as Germany and others for local elections; CIA (Central Intelligence Agency), 2020). 3. I limit the discussion to the justifiability of disenfranchising citizens by their state. 4. Some defenders of expanding the enfranchisement of children can also be included under this category (Cook, 2013; López-Guerra, 2012). 5. I thank Laura Lundy for hinting at the interpretative issues with Beckman's argument. 6. Thanks to an anonymous reviewer for encouraging me to clarify what is wrong with 'innocence'. 7. WEIRD: Western, Educated, Industrialised, Rich Democracies. 8. Many authors who endorse the expansion of franchise to children tend to rely on an outcomes-based understanding of democratic legitimacy. Through mathematical models (such as Condorcet's jury theorem) they attempt to prove that the inclusion of individuals who lack cognitive capacities to vote does not harm the outcome of the democratic process, thus, legitimate to include them (Goodin & Lau, 2011;Munn, 2018). To my knowledge, no author defends disenfranchisement exclusively based on outcomes. Thanks to an anonymous reviewer for requesting clarification. 9. Moral corruptness can be another trait that tracks harm through franchise; it seems to be the underlying principle behind the disenfranchisement of prisoners. This argument has not been used in the case of children. See debate in Beckman (2009: Ch.5). 10. Thanks to Marcus Häggrot for encouraging me to clarify how the Condorcet theorem fits into this argument. 11. Thanks to Dave Archard for encouraging me to frame it in this way. 12. Not all positions that rely on age thresholds defend the current age threshold as correct. Some, such as López-Guerra (2012) or Umbers (2020), consider current age thresholds as unjustified, arguing that the appropriate age of correlation to political competence would be between 10 and 12-years-old. This is a radical change if compared to the current system, but it still relies on the justifiability of statistical discrimination as a reason to (dis)enfranchise. 13. van Parijs (1998) also proposes a system for the disenfranchisement of the elderly. 14. Many authors show concern with competence-testing due to its practicalities (see, Rehfeld, 2011, p. 148-150;. There may be important pragmatic issues in determining the specific policy for a theory of enfranchisement, but they don't affect the (un)justifiability of disenfranchisement. As López-Guerra argues (López-Guerra, 2012, p. 134), whether tests work better than alternative mechanisms for disenfranchisement (or against unconditional universal enfranchisement) depends on the institutional practices, public culture of and capacity to deter corruption in different societies.