---

This article is available here to read fully.

We thank the authors of this publication for allowing us to reproduce their work.

Please note, that some images and text may not display correctly in this version, and there may be some formatting issues. We apologise for any inconvenience this may cause.

Is there is an issue, please get in contact with us at info@vote16.co.uk.

---

Tak Wing Chan Matthew Clayton

University of Oxford University of Warwick

This article is an examination of the issue of whether the age of electoral majority should be lowered to sixteen.We consider and reject several arguments raised by both sides of the voting age debate.The key issue, we claim, is the political maturity of young people. Drawing on empirical data collected in nationally representative surveys, we argue that the weight of such evidence suggests that young people are, to a signi cant degree, politically less mature than older people, and that the voting age should not be lowered to sixteen.

Introduction

The questions of whether there should be a minimum voting age and, if so, at what age it should be set, are signi cant political issues, because having the vote is widely recognised as one of the most important legal rights within a democracy. In the absence of some compelling argument,the exclusion of a particular section of the population from the franchise is standardly taken to be a serious violation of political equality.And yet, until recently, the exclusion of teenagers from the vote has received comparatively little attention in public debate (Schrag, 2004).

Nevertheless,the question of the voting age is now a subject of political debate. In 2003—4, the UK Electoral Commission conducted an extensive public con- sultation on the question of the minimum age of voting and candidacy in UK elections (Electoral Commission, 2003; 2004). In its nal report,Age of Electoral Majority, which offers a thorough review of arguments for and against lowering the age at which political rights are enjoyed, the Commission recommends that while the age of candidacy should be lowered from 21 to 18, the voting age should stay at 18 (Electoral Commission, 2004). The Commission considers several arguments that have been advanced in favour of lowering the voting age to sixteen.Among these arguments are:(1) those that demand consistency in the treatment of voting and other social and economic rights; (2) the related argu- ment that sixteen-year-olds are suf ciently mature to vote; (3) the appeal to public demand or public opinion; and (4) the argument that positive effects on political participation will follow from lowering the voting age.In each case,the Commission marshals arguments and evidence to show that the case for lowering the voting age is not conclusively established.

*' 2006The Authors. Journal compilation ' 2006 Political Studies Association

SHOULD THE VOTING AGE BE LOWERED? 547**

In this article, we begin by addressing some normative matters that attend the voting age debate — considerations that the Electoral Commission believe to be relevant to the issue, as well as certain other arguments that have been offered. We challenge certain arguments that the Commission and others deploy in defence of retaining the status quo.Thereafter, we offer arguments that support the justice of an age-based allocation of voting rights, which rest on the impor- tance of political maturity for democracy.These normative considerations moti- vate the need for an examination of certain empirical questions concerning the relationship between age and political maturity.Thus we turn to an examina- tion of available survey data that are relevant to the question of the age of electoral majority. On the basis of these considerations we defend the conclu- sion that the voting age should not be lowered to sixteen, as some have demanded.

Normative Issues Public Opinion

It might be thought that one of the rst questions that arise with respect to any proposed change to the voting age is whether the public demands such a change. Accepting this thought, the Electoral Commission asked the market research company, ICM, to survey public opinion on the issue. ICM found that a clear majority is in favour of retaining eighteen as the minimum voting age (Electoral Commission, 2004, pp. 29—42). But why does it matter what the public thinks about this issue? Its view might be thought to be relevant in two ways. First, democratic decisions, it might be argued, should be shaped by the will of the people and, therefore, however good the arguments are for a particular legal reform,it always counts against the reform if the majority or a plurality rejects it. Perhaps this is not always a decisive reason, but it is a consideration that must always be weighed in the balance.

This appeal to majoritarian choice rests on an impoverished conception of democracy. In judging the democratic character of political decisions, Ronald Dworkin proposes a distinction between choice-sensitive and choice- insensitive issues. In any democratic society there will be a range of decisions which ought to turn on, or at least be in uenced by, the preferences of the community,determined by the will of the majority or a plurality,or according to some other social choice rule.However,many issues are choice insensitive,in the sense that their resolution may proceed without reference to what the public wants. He cites capital punishment and racial discrimination in employment as examples of choice-insensitive issues (Dworkin, 2000, pp. 204—5).

Questions about the franchise are choice insensitive. Even if the overwhelming majority are appalled by the prospect of sixteen-year-olds having the vote, this cannot in itself be even apro tantoreason against lowering the voting age.The

democratic conception is one in which every member of the political community is viewed as having equal status and in which political institutions and practices embody that principle.No doubt,majoritarian procedures might be an essential part of democracy so understood,but they must operate under certain conditions: that the rules for electing representatives are free and fair,that discrimination and oppression are prohibited, and so on. Signi cantly, electoral matters concerning the size and shape of the franchise are among the most important conditions of the legitimation of majoritarian procedures,and so cannot legitimately be deter- mined by the will of the majority.It follows that the appeal to majoritarian choice must be rejected.1

Let us turn to the second way in which it might be thought the voting age debate should be informed by public opinion.Here,it is said that in the absence of any precise formula by which to measure political awareness,individual responsibility and other attributes that are constitutive of political maturity,we must rely on the public s view of whether sixteen-year-olds have suf cient maturity to have the right to vote (Electoral Commission, 2004, p. 25). This way of regarding the public s importance is different from the appeal to majoritarian choice,because it is consistent with the choice-independent view that the franchise ought to be enjoyed by citizens who are able and willing to use it effectively.Consulting public opinion,on this interpretation,might be thought to be the way of gaining the best judgement about political maturity understood in this way.But,if this is the right kind of explanation of the relevance of public opinion to the issue at hand, the factual premise of the argument remains in need of support.We need a defence of the belief in the public s authority on matters of political competence.

Political Participation: How Should the Consequences for Electoral Turnout be Assessed?

The Electoral Commission s inquiry into the voting age was motivated in part by its concern with declining participation rates in UK elections and, in particular, the low participation of young people in politics.While it does not believe this to be a decisive reason for changing the voting age, if participation rates were to increase as a consequence,this would count in favour of lowering it.Similarly,the Commission asserts that if sixteen and seventeen-year-olds were to exercise voting rights extended to them at a low rate,then this is a reason — though,again, not necessarily a conclusive reason — not to lower the voting age (Electoral Commission, 2004, p. 21).

Suppose we take electoral turnout to be indicative of political participation.The Electoral Commission suggests that we ought to look at the overall turnout,and claims that the short-term effect of enfranchising sixteen-year-olds would be a decline in turnout,because the available evidence suggests that the young in the UK are less likely to vote than older people (Electoral Commission,2004,p.47). In addition, the Commission nds arguments supporting the view that turnout would increase in the long term to be inconclusive.

Figure 1: Quality and Quantity: A and B

A B

These claims concerning the detrimental effect on turnout in the short term rest on an implausible view of how to gauge whether turnout has improved or worsened.The Electoral Commission assumes that we should consider whether the overall turnout of the electorate rises or falls.Certainly,that headline gure is relevant in day-to-day political analysis when we are interested in changes to the level of participation of a given electorate.But,we should not accept that measure when we are evaluating the effects of a change in the size of the electorate.

To see why,consider an analogous case concerning population offered by Derek Par t (1986), who asks, how many people should there be? . Suppose, in some range, there is a trade-off between quality of life and population size.We might have more people with a lower quality of life, or fewer people whose lives are better.The choice is illustrated in Figure 1 (Par t, 1986, p. 145), in which the width of the blocks indicates the size of the population and the height indicates the quality of life people enjoy.Which would be better: A or B ?This question is similar to our question about the franchise and turnout.Should we have a larger electorate including sixteen and seventeen-year-olds in which turnout is predict- ably lower overall, or a higher turnout from a smaller electorate?

In his discussion of the ethics of population, Par t (1986, p. 151) asks us to consider A and A+ (see Figure 2). In A there is a population with a quality of life that,let us suppose,is high.InA+ the population ofA remains with the same

quality of life. However, there is an extra group of people who have lives well worth living, but which are not as good as those of the rst group. Is A+ worse than A? True, the overall average quality of life in A+ is lower than in A. Nevertheless,inA+ the rst group are no worse off than they would be inA,and since the additional group in A+ have lives well worth living,it is implausible to

suggest that it would be better if they had never existed. Why , Par t asks rhetorically, are they such a blot on the Universe? (Par t,1986,p.152).For these and other reasons,we might reasonably conclude thatA+ is no worse thanA,even though the overall average is lower in A+.

Figure 2: Quality and Quantity: A and A+

A A+

Par t s discussion tells against a preoccupation with changes to the overall average turnout from an extension of the franchise. If we enfranchise a younger section of the population, who then exhibit an adequate level of turnout, there is no reason to think that this will reduce the higher level of participation of older voters, and it may have positive effects for the electoral process, or for the integration of citizens in the political process.We conclude that the argument to keep the voting age at eighteen arising from a concern for average electoral turnout should be rejected.2

In any event,it is not obvious that considerations of turnout should in uence the age at which electoral majority should be set. The concern with turnout is motivated by,among other things,the thought that the government s mandate to rule is dependent upon the voice of the people as expressed in free and fair elections.If few electors register a vote,it is said,the government s mandate is to that extent weakened.But,this concern cannot be a legitimate reason to change the franchise in order to produce a better turnout,or to retain the existing size of an electorate so as not to worsen average turnout.If turnout is a problem,it needs to be addressed by politicians and political parties changing their behaviour and relationship to the electorate, not by changing the electorate.

It might be objected that the concern over turnout in the case of sixteen and seventeen-year-olds has a different character to the case of the older electorate.In the former case,low turnout may be explained by a lack of interest in politics as such; in the latter case, it indicates a disaffection with our particular political system or party system (Curtice,2004).The argument might be advanced that the

lack of interest explanation is consistent with exclusion from the electorate; it does not weaken a government s mandate if a group that displays no interest in the question of how we ought to live together in society is denied the vote.The mandate is weakened only if those who withhold their votes do so because their political institutions and politicians are failing to deliver an acceptable vision of collective life.

Yet, this objection can be rebutted with the observation that some sixteen and seventeen-year-olds do express considerable interest in political matters.True,on the assumption that many would not exercise their votes due to lack of interest in politics,a low turnout among this age group might not be deemed problematic in the same way as a low turnout of older age groups would.Nevertheless,if it is likely that those sixteen and seventeen-year-olds who did vote would do so competently, there would be little cost in enfranchising this age group. Any lowering of the overall turnout of the larger electorate might be discounted to the extent that the lack of interest among young voters applies, while the participa- tion of (let us suppose) competent sixteen and seventeen-year-olds would enhance our democracy.

Ultimately,level of turnout is not the issue.What matters is quality of turnout.If it is likely that those sixteen-year-olds who would use their vote would be competent voters,then a low turnout among this group would not,other things being equal,be a cause for concern.The worry about extending the franchise is not that sixteen and seventeen-year-olds would not use their vote, but that too many of them would vote and do so incompetently, in a way that would be detrimental to our democracy.

Electoral Maturity and Equality

The Electoral Commission rightly regards maturity as the fundamental issue in determining the appropriate age of electoral majority.Given its importance,it is disappointing that only a few pages of the report are devoted to its discussion.We should, it is argued, enfranchise age groups with suf cient social awareness and responsibility . These are de ned in terms of one s ability and willingness to consider the effects of decisions on society at large as well as oneself.Little more guidance on the maturity question is offered, apart from the Commission s acknowledgement of the dif culty of measuring maturity speci ed in this way, and its insistence that, in the absence of clear indicators, we must rely on the public s view of when maturity is suf ciently developed (Electoral Commission, 2004, pp. 24—5).

As observed above, no argument is offered as to why the public are experts on the issue of maturity, and very little sense is given of the different kinds of evidence that might be brought to bear on the issue. Later, we offer a pre- liminary discussion of some empirical data that might be considered in deter- mining an appropriate age of electoral majority. Before that, however, we must consider three important challenges to the idea of setting eighteen as the minimum voting age on the basis of considerations of maturity.The rst invites us to abandon age as a qualifying condition and replace it with a competence test.The second appeals to the insigni cant differences that might lie between sixteen and eighteen-year-olds to argue for the injustice of excluding the

former. The third rests on analogies between enfranchising sixteen and seventeen-year-olds and other emancipatory causes, or the ideal of consistency in the enjoyment of different rights.

The Abandonment of Age as a Qualifying Condition

Some argue that if the enjoyment of voting rights ought to vary with political maturity,then society should exclude individuals from the franchise on the basis of competence rather than age (Harris, 1982).An age-based franchise, it is said, arbitrarily discriminates against young people who possess the capacities, moti- vation and understanding that are relevant to the act of voting to a higher degree than some older people do.

We should reject this argument.3 It is a mistake to assume that the discriminations we make in law or policy should always be guided by what isfundamentally important. Suppose that age is not a fundamental consideration in judging qualifying conditions for the vote.Nevertheless,age might be a valuableproxyfor what is fundamental.The distribution of capacities that we decide are fundamen- tal might be correlated with age, albeit imperfectly. Consequently, age-based discrimination might be an effective way of tracking those capacities that are fundamentally important.This is merely an instance of the rationality of employ- ing action-guiding rules.As in other contexts, we seek, among other things, an effective way of discriminating between competent and incompetent individuals. In selecting for university places,for example,admissions tutors aim,other things being equal,to select those who will perform well in the course for which they have applied.Quali cations already achieved clearly do not matter fundamentally in such a selection process, but they might serve as a useful, albeit imperfect, indicator of academic potential.Similar claims might be made with respect to age as a proxy for identifying competent voters.

Consider now a related issue of justice.Suppose that age is an effective proxy in discriminating between the competent and incompetent. Some claim that age- based discrimination is, nevertheless, unjust.When we are allocating rights and duties to individuals, they claim, like cases should be treated alike. It is unjust to deprive a sixteen-year-old of the vote when an older individual enjoys that privilege without possessing any more political competence. Call this the anti- ageist principle. The anti-ageist principle condemns the use of any rule that excludes any competent sixteen-year-olds from the vote. But it overlooks our concern to exclude incompetents.Yet,we have good reasons of justice to prevent the incompetent from voting, since their votes might impact negatively, not merely on themselves,but also on the legal rights and duties that apply to others. Following the anti-ageist principle might inhibit our pursuit of justice,all things considered.

The anti-ageist might reply that we should test for competence and license individuals as voters, in the same way as individuals take a test to qualify for the

right to drive. However, there are various kinds of costs and dif culties in identifying competence on a case-by-case basis. First, it might be too costly to make available an accurate test to screen for political competence;for example,it is doubtful that the kinds of judgement required from voters in the political process are amenable to being tested in a relatively costless manner,as is screening for driving competence. Second, such a test might have certain undesirable consequences which we ought to avoid.Those who are denied the franchise on grounds of incompetence might suffer a loss of self-esteem that would impact detrimentally on various aspects of their lives. Such problems are avoided by an age-based rule rather than a competence-based rule. Ideally, then, we should adopt an age-based rule that sets the voting age at the point at which a suf cient proportion of citizens above that age are politically competent. So, age-based voting entitlements can be both ef cient and just.4

The Appeal to Insigni cant Differences

Suppose we retain an age-based qualifying condition for voting. At what age should we set it? Consider an argument for lowering the voting age to sixteen, which runs as follows.Eighteen-year-olds should have the vote.The competence of sixteen-year-olds is only insigni cantly lower than that of eighteen-year-olds. Therefore, sixteen-year-olds should have the vote.

Suppose we accept the normative premise that eighteen-year-olds should have the vote. Suppose we grant, in addition, that there is only an insigni cant difference in competence between sixteen and eighteen-year-olds.Still,we might resist the conclusion that sixteen-year-olds ought to be enfranchised.The argu- ment is weak, because, for all we know, we could use the same argument repeatedly until we have enfranchised six-year-olds, which would be absurd.A series of insigni cant changes can amount to a signi cant change,and we ought to identify a suitable stopping point so that we can achieve the bene ts of enfranchising those who would enhance our democracy, without jeopardising that good by continuing incrementally to extend the franchise.And, we might insist on that stopping point even when there is no signi cant extra loss of political competence by lowering the voting age further.5

Other Egalitarian Arguments

Finally,we address a set of arguments for enfranchising sixteen-year-olds that rests on analogies with other emancipatory causes, or the ideal of consistency in the enjoyment of different legal rights.

Comparisons are sometimes drawn between the political emancipation of women and slaves and lowering the voting age.Arguments in favour of votes for women offered in previous centuries,for example,are sometimes cited as relevant to the voting age debate. However, many comparisons of this kind are unconvincing,

because sex and race are permanent features of people s lives while,in the normal course of life, everyone enjoys childhood, youth and adulthood. Because they affect everyone, ageist restrictions are not obviously as wrongful as restrictions based on race or sex.

Other parallels drawn between these different debates are also problematic. One concerns the ideal of consistency in the enjoyment of apparently related rights. For instance, while it does not argue for a common age of majority at which age all legal rights enjoyed by adults become available, Votes at 16 (2003), the principal coalition campaigning for the reduction of the voting age in the UK, claims that the right to vote is akin to other rights enjoyed at sixteen, such as the right to have a full-time job, to have sex and beget children, and so on. If sixteen-year-olds are judged suf ciently mature to make freely the momentous decision to beget children, should they not also have the right to in uence less momentous decisions about political matters? But these kinds of argument often suffer from inattention to the complex reasons which guide the allocation of legal rights. For example, the argument for consistency between the rights to beget and to vote might be rebutted with the observation that granting the right to have sex to sixteen-year-olds might be an acknowledgement, not of the sexual maturity of that age group, but of the dif culty of prohibiting sex. That problem of enforceability does not translate to the exclusion of sixteen- year-olds from the franchise.

Another related set of arguments asserts that it is inconsistent to hold individuals to certain legal duties without extending the franchise to them.The most familiar version of this view is the slogan no taxation without representation , but some substitute conscription ,or other duties,for taxation .Yet,arguments of this kind ignore the countless ways in which the negative freedoms of those without the franchise — very young children or tourists, for example — may legitimately be constrained. For example, like others, adolescents are legitimately subject to the restrictions laid down by the criminal law in relation to the life and liberty of others. The legitimacy of such constraints is not diminished by the fact that adolescents do not have a voice in the process by which these laws are enacted.6

Nevertheless,at least two claims for consistency between different rights do retain force.First,there are certain cases in which the attributes suf cient to generate a valid claim for one right are also suf cient for other rights.Second,the enjoyment of a certain right supports the enjoyment of other legal rights when the latter are preconditions of the former. For example, it is widely held that the freedom to hear others views is a precondition of democratic choice. So, there is a case for consistency between the voting age and the age at which censorship laws are relaxed,such that when an individual has the vote they might be as informed as any other voter. It would be odd to allow sixteen-year-olds the vote while retaining 18 certi cates for lms.But note that neither of these valid claims for consistency supports sixteen, rather than eighteen, as the age of majority.

Empirical Considerations: How Should we Test for Maturity?

Political maturity is the pivotal issue in the debate over the voting age.To test for maturity,we need a set of criteria that are indicative of an ability and willingness to engage in democratic choice and which, ideally, are capable of measurement. For example, one indicator of willingnessis the level of interest in politics expressed by the age group in question.Theabilityto engage competently in democratic political choice is multifaceted.Voters ought to have knowledge of the political system,and understanding of the nature and signi cance of issues that are the subject of public and political debate. Ideally, they should possess political convictions that are consistent (do not directly contradict each other), hang together (succeed in connecting relevant reasons in common among different public issues), consequentially rational (appreciate the expected consequences of a choice) and are not subject to whimsical revision.With these considerations in mind, we now turn to examine relevant survey data.7

Interest in Politics

Survey data consistently show that young people are generally less interested in politics than older individuals (e.g.Park,1999;Russellet al.,2002).However,most of these surveys sample individuals who are eighteen years old or above.What about sixteen and seventeen-year-olds?To answer this question,we draw on data collected in the British Household Panel Survey (BHPS). In 1991 and 2001, BHPS respondents were asked, How interested would you say you are in politics? Would you say you are (1) very interested; (2) fairly interested; (3) not very interested; or (4) not at all interested .8

In Figure 3 we show how interest in politics varies with age.(To reduce sampling variation,we have combined neighbouring age groups.Thus,16 and 17-year-olds form our rst data point,18 and 19-year-olds form the second data point,and so on.) It is clear that in both 1991 and 2001,the proportion of respondents who said that they were very interested or fairly interested in politics increased almost linearlywithage,untilaboutage60whentheincreaselevelledoff.Levelofinterest in politics was generally (about 10 percentage points) higher in 1991 than in 2001. Such a difference could to a large extent be explained by the politics of the day. For example,the race between the main political parties was tighter in 1991 than in 2001.The PollTax riots and the end ofThatcher s premiership in 1990 might also have heightened interest in politics at that time. It seems reasonable to attribute much of the observed gap between 1991 and 2001 to aperiod effect.

But,should the observed age gradients be interpreted as acohorteffect or anageing effect?As is well known,the three variables of age,period and cohort arelinearly dependent on each other, and together they present a dif cult identi cation problem.9 Much of the age gradients of Figure 3 could in fact be interpreted in terms of a combination of cohort and period effects, with no reference to any

Figure 3: Proportion of Respondents Who are Very or Fairly Interested in Politics

by Age (N1999= 9,267,N2001= 15,703)

1991

2001

20 30 40 50 60 70

age

ageing effect.This view is made up of three parts:(1) people donotbecome more or less interested in politics as they get older;(2) people from early birth cohorts happen to be more interested in politics than those from more recent cohorts;and (3) something made everyone less interested in politics in 2001 than they were in 1991.To visualise this argument, consider the following: if the line for 1991 was shifted ten years to the right, much of it would still stand clearlyabovethe observed line for 2001.The horizontal shift of the 1991 line represents the strong assumption of no ageing effect: everyone keeps the same level of interest in politics they had ten years before.The post-shift gap between the two lines then represents the period effect.

While it is very likely that there is a strong cohort component in the observed pattern,there is no need to preclude the presence of some ageing effects,at least among certain age groups.10 A hint of this can be gleaned from the fact that the observed age gradients are in factnon-linear.In both years,16 and 17-year-olds are especially unlikely to report interest in politics,in the sense that their data points fall signi cantly below the extrapolated regression lines estimated for people aged 20 to 60.If the data point of the 16 and 17-year-olds was shifted 10 years to the right, it would be found signi cantly belowthe observed line for 2001.Thus, in sharp contrast to most people in the sample, the 16 and 17-year-olds of 1991 actually became much more interested in politics when they turned 26 and 27, respectively,in 2001.This suggests to us that,at least for the youngest respondents in our sample,a signi cant ageing effect is operating as well.(A parallel argument suggests that the same is true for older respondents.)11

Direct evidence supporting our interpretation can be obtained by exploiting the panel nature of the BHPS data. For respondents interviewed in both 1991 and

Figure 4: Change in Interest in Politics between 1991 and 2001 by Age (N= 4,907)

|| | :- | || || 20 30 40 50 60 age in 1991

2001, we match their responses ten years apart.This then allows us to trace, for such individuals,how their interest in politics might have changed over time.We report in Figure 4 net gain in interest in politics for each age group.12Three points are notable here.First,taken over the entire sample,interest in politics fell by just over ve percent between 1991 and 2001 (see the lower horizontal line of Figure 4).This corresponds to the period effect discussed above. Secondly, there was a net loss in interest in politics for all but two age groups above the age of twenty in 1991; and for the two age groups with net gain, their gain is very marginal.Thirdly, in sharp contrast to the pattern for the bulk of the sample, we see a clear and substantial net gain in interest for our rst two data points, representing people aged sixteen to nineteen in 1991. In other words, our youngest respondents, who were disproportionately uninterested in politics (see Figure 3), also show the greatest gain in such interest over time.

Party Identi cation

BHPS respondents were also asked, Generally speaking do you think of yourself as a supporter of any one political party? .If the answer was no,they would then be asked, Do you think of yourself as a little closer to one political party than to the others? .On the basis of their responses to these two questions,we classi ed our respondents as either: (1) supporters of a political party, (2) non-supporters, but with af nity to a political party or (3) non-partisans.

Figure 5 shows that in both 1991 and 2001 the proportion of party supporters (category 1) increased with age in an almost linear fashion,while the share of the non-partisans (category 3) declined with age.13 The fact that there seem to be more non-partisans in 2001 than in 1991 might be a matter for concern. But again this might be due to the politics of the day. For present purposes it is

Figure 5: Party Identi cation by Age (N1991= 9,265,N2001= 16,620)

supporters

1991 2001

20 30 40 50 60 70

age non−partisans

2001 1991

20 30 40 50 60 70

age

important to note that,in both years,16 and 17-year-olds were especially unlikely to be supporters of a political party (and especially likely to report non- partisanship),in the sense that their data points lie signi cantly below (above) the extrapolated regression lines estimated for people aged 20 to 60. Following the logic of the argument set out in the previous section, we argue that although strong cohort and period effects might underlie the age gradients of Figure 5,14 sixteen and seventeen-year-olds do become more partisan politically as they get older.

Once again, direct evidence supporting our interpretation can be found by matching responses given by the same respondents in 1991 and 2001.The solid line of Figure 6 refers to a shift between being a supporter of a political party (category 1) or a non-supporter with af nity to a party (category 2) on the one

Figure 6: Change in Political Partisanship between 1991 and 2001 by Age

(N= 5,283)

|| | :- | || 20 30 40 50 60

age in 1991

hand,and being a non-partisan (category 3) on the other.15The dashed line refers to a more narrow de nition of change, that is between being a party supporter (category 1) and being a non-partisan (category 3).16 It can be seen that, using either de nition of change, a much higher proportion of our youngest respon- dents,between the ages of sixteen and nineteen,became politically partisan over the 1990s than among older respondents. This is so in spite of the fact that teenagers are especially likely to be non-partisan (see Figure 5).Together with evidence on change in political interest as shown in Figure 4,it might be said that the teenage years are a period of political awakening.

It might be argued that non-partisans are not really politically disengaged.Some of them might simply feel that none of the mainstream political parties come close to representing their views and interests.As a result,they do not support or feel close to any mainstream party.However,this does not mean that they are not involved in other forms of political activities, campaigns or organisations.

While this might be true for a small minority of the non-partisans,we think this interpretation is generally invalid. Data from the 2001 BHPS survey reveal that, compared with supporters of political parties (category 1),or those who feel close to a political party (category 2), the non-partisans are invariably less likely to be members of,let alone active members of,a wide range of organisations,including environmental groups, feminist groups or voluntary service organisations (see Table 1).17 What about the possibility that participation in non-mainstream poli- tics is especially common among sixteen and seventeen-year-old non-partisans? We believe this is rather unlikely, given that a disproportionate share of young people of this age group profess a lack of interest in politics (see the previous section).

Table 1: Percentage of Respondents Who are Members of (Panel A) or Active in

(Panel B) Organisations by Political Partisanship

(1) Supporters (2) Non-supporters (3) Non-partisans

' 2006The Authors. Journal compilation ' 2006 Political Studies Association POLITICAL STUDIES: 2006, 54(3) SHOULD THE VOTING AGE BE LOWERED?

Panel A: members Political party

Trade union Environmental group Tenants groupa Voluntary services group Other community group Feminist organisationb Any organisationc

N

Panel B: active members Political party

Trade union Environmental group Tenants groupa Voluntary services group Other community group Feminist organisationb Any organisationc

N

4.3* 0.4* 0.1 18.7* 17.1* 12.5 3.0* 3.6* 1.3 5.2* 5.0* 2.6 4.4* 4.4* 2.8 2.8* 2.7* 1.3

1.0* 0.9 0.5 59.0* 58.7* 44.3

6,155 3,609 6,714

2.0* 0.1 0.1 4.2* 3.3 2.6 1.7* 1.7* 0.6 3.1* 2.4 1.8 4.3* 3.9* 2.8 2.6* 2.5* 1.4 0.9 0.8 0.5 49.5* 50.3* 39.3 6,212 3,636 6,765

' 2006The Authors. Journal compilation ' 2006 Political Studies Association POLITICAL STUDIES: 2006, 54(3) SHOULD THE VOTING AGE BE LOWERED? 549**

Notes: *the difference between category 1 (or category 2) and category 3 is statistically signi cant at the 5% level; aincludes residents group and Neighbourhood Watch; bincludes women s group; cincludes organisations listed, plus parents/school association, religious group or church organisation, social club/working men s club, sports club, Women s Institute/Townswomen s Guild, professional organisation, pensioners organisation, scouts/guides organi- sation, other groups or organisations.

Source: BHPS (2001).

Political Knowledge

Some understanding of how key political institutions operate,and knowledge of the main political issues, policies and their implications are, no doubt, crucial indicators of political maturity.In this section,we consider whether young people are as politically informed as older people. Standard survey instruments provide some measures of factual or propositional knowledge (Frazer and Macdonald, 2003, pp. 67—8). That is, propositions such as the number of members of parliament is about 100 are read to the respondents,and they are asked whether the propositions are true or false.However,the number of correct answers given in such tests is probably not the quality that really matters.Well-informed citizens would have the practical skills to obtain and interpret relevant information from the government and other sources.They would know how to organise other

citizens with similar interests, and engage with relevant of cials or politicians in debate. Such practical or substantive knowledge is clearly more important than simple factual knowledge.But measures of the latter are all that are available to us. Furthermore,it seems reasonable to assume that people who lack factual political knowledge are also less likely to be political informed in a substantive sense.18

With the above caveat in mind, we note that the following items on factual political knowledge were included in the 1997 British Election Studies (BES):(1) Margaret Thatcher was a Conservative prime minister ; (2) The number of members of parliament is about 100 ; (3) The longest time allowed between general elections is four years ;(4) Britain s electoral system is based on propor- tional representation ;(5) MPs from different parties are on parliamentary com- mittees ; (6) Britain has separate elections for the European Parliament and the British Parliament ; and (7) No one may stand for parliament unless they pay a deposit .

However,unlike the BHPS,the BES sample individuals aged eighteen or over.So, to compare sixteen and seventeen-year-olds with older people, we need to turn to another data source which contains comparable data for young people,namely theYoung People module of the 1998 British Social Attitudes Survey (YBSA). The YBSA sampled young people aged twelve to nineteen living in private households. It also contains seven items on factual political knowledge, three of which are identical with items 2,3 and 6 of the BES,and one is substantively very similar to item 1 of the BES.19 We group YBSA and BES respondents into two-year age groups,20 and show the proportion of respondents giving correct answers to each of the four items in the panels of Figure 7.

Overall,almost all respondents,youth (the four solid dots on the left) or adult (the hollow dots),got the prime minister question right.However,for the other three items,there is a clear age gradient,in which the proportion of respondents giving correct answers increases with age,until about age 50 when the increase levels off.

It can be seen that regarding the two items on the number of MPs and the longest time between general elections, theYBSA data points for eighteen and nineteen-year-olds are signi cantly lower than the BES data points for the same age group. Such discrepancies could arise from differences in the sampling procedures or response rates of the two surveys. But this could also be an example of what is known as context effects in surveys. This problem arises when,for example,the order in which a set of questions is asked changes the way a particular item is interpreted by respondents and thus alters their response (Tourangeau,Rips and Rasinski,2000,ch.7).The implication of context effects is that comparing even identical items across surveys is fraught with dif culties. Such effects are especially problematic for attitudinal questions,but it seems that a variant of context effects might be affecting the two items as well.But,allowing for the possibility that some context effects might be at work, it would seem reasonable to conclude from Figure 7 that factual political knowledge does tend

Figure 7: Political Knowledge by Age (N= 3,770 in Each Panel)

' 2006The Authors. Journal compilation ' 2006 Political Studies Association POLITICAL STUDIES: 2006, 54(3) SHOULD THE VOTING AGE BE LOWERED?

Prime minister YBSA 1998 BES 1997

10 20 30 40 50 60 70

age

The longest time allowed between general elections is four years

BES 1997

YBSA 1998

10 20 30 40 50 60 70

age

The number of members of Parliament is about 100

BES 1997

YBSA 1998

10 20 30 40 50 60 70

age

Separate elections for the European parliament and the British Parliament

BES 1997

YBSA 1998

10 20 30 40 50 60 70

age

' 2006The Authors. Journal compilation ' 2006 Political Studies Association POLITICAL STUDIES: 2006, 54(3) SHOULD THE VOTING AGE BE LOWERED? 559**

to increase with age, and that young people are less knowledgeable than older people about political facts.This should not be surprising given the lower level of interest in politics shown by young people.

Consistency and Stability of Attitudes

It is of course desirable that voters have consistent political attitudes, otherwise they might support policies that are mutually incompatible.For example,it would be logically inconsistent for someone to support comprehensive education and demand selection at the same time, because comprehensive education is non- selective.In this section,we ask whether young people are less consistent in their attitudes.

Figure 8: Consistency of Attitude by Age (N= 16,146,N= 16,205 for Left-hand and

Right-hand Panels Respectively)

trust in government−2001 women’s employment−2001

|| | :- | || || | :- | || 20 30 40 50 60 70 20 30 40 50 60 70

age age

But, it should be noted that, apart fromlogical consistency, one can speak of other types of attitudinal consistency (Heath, 1986). For example,technical consistency concerns the means—end relationship. It may be completely consistent, logically speaking, for people who favour egalitarian income distribution to oppose pro- gressive taxation,if they believe that progressive taxation istechnicallynot the best way to achieve income redistribution. Because survey data often do not record subtleties of beliefs about means—end relations, the results presented here should be interpreted with caution. Furthermore, Anthony Heath (1986) discusses normative consistency, which is about the relations between different ends. For example, a person who supports gender equality would probably also support racial equality. But, there is no logical requirement that this should be the case. Our discussion below is mainly about normative consistency.

We consider attitudinal consistency in two substantive domains,namely (1) trust in government and (2) women s employment. For these two domains, we select relevant questions from the BHPS in 2001 (see Appendix).The principal measure we use is Cronbach s alpha.This coef cient is often used by psychometricians in questionnaire design and scale construction. In that context, the goal is to assess the internal consistencyof a battery of items; that is, to check if the items are measuring the same thing (Nunnally, 1978, pp. 229—30). Here the logic is reversed:we assume that the items are indeed measuring a one-dimensional latent construct, but we wish to test if age is related to the level of consistency of response.

In Figure 8 we show how alpha varies by age in the two domains.Given that our scales are relatively short (there are three and six items respectively), the overall

Table 2: Attitudes on Women s Employment by Age (Column Percentagesa**)**

A woman and her family would all be happier if she goes out to work All in all,

family life 16—17 (N= 614) 30—39 (N= 3522) suffers if a

woman has a

full-time job. agree neither disagree agree neither disagree

agree b 17.5 13.3 23.8 21.4 19.9 50.6 neither 20.4 34.2 18.9 14.7 40.6 15.1 disagree c 62.1 52.5 57.3 64.0 39.5 34.3

Notes: aPercentages within each column add up to 100%; bincludes strongly agree and agree; cincludes strongly disagree and disagree.

alpha values,0.60 for trust in government and 0.69 for women s employment (see the horizontal lines in the panels), are quite acceptable (Heathet al., 1994).21 In both domains,alpha increases almost linearly from 16 to about age 40 (age 30 in the case of women s employment) before it attens out and hovers about at a higher level.The increase is gentler for trust in government than for women s employment.22 Taking alpha values as measures of attitude consistency, it would appear that young people s attitudes do become more consistent as they become older.

To be more concrete, we report in Table 2 people s views on two statements regarding women s employment among 16 and 17-year-olds and among respondents who are in their 30s.The two statements are: All in all, family life suffers if a woman has a full-time job and A woman and her family would all be happier if she goes out to work . Since the rst statement speci es the condition of full-time employment, while the second does not, it is strictly speaking not logically inconsistent to agree with (or to disagree with) both statements. Having said that, we note that the second statement is more general and is therefore stronger than the rst. It would be odd to think that employ- ment in general is bad for a woman and her family, but full-time employment is unproblematic. It can be seen that among those respondents who disagree that A woman and her family would all be happier if she goes out to work , only 23.8 percent of sixteen and seventeen-year-olds think that family life will suffer . The proportion of 30 to 39-year-olds who take this combination of views is much higher — 50.6 percent. Correspondingly, we see a much higher proportion of sixteen and seventeen-year-olds disagreeing with both statements (57.3 percent compared with 34.3 percent).

Finally,we turn to the issue of attitudinal stability.Politically mature voters should be open to rational argument, new experience and fresh evidence. Thus, we

Figure 9: Stability of Attitude by Age (N= 7,535,N= 11,211 for Left-hand and

Right-hand Panels Respectively)

government (1998−2001) women’s employment (1999−2001)

|| | :- | || || | :- | || 20 30 40 50 60 70 20 30 40 50 60 70

age age

would expect some change in their attitudes over time. However, excessive attitudinal instability would suggest that their view is not based on convictions, but is rather subject to whimsical revision.Are young people less stable in their attitudes than older people? Because the questions we examined in this section were put to BHPS respondents in multiple years,we can match their answers to the same questions over time, and check the stability of their attitudes. Opera- tionally,we add up the scores of their responses to form an overall scale score in each domain (see Appendix).We then compute the Pearson correlation of the scale scores two to three years apart.

As can be seen from Figure 9,in terms of overall averages (see the horizontal lines in the panels),attitudes regarding trust in government (r= 0.50) are generally less stable than attitudes concerning women s employment (r= 0.69). However, in

both domains,the attitudes of young people between the age of sixteen and the mid-20s are noticeably less stable than those of older people.

Summary and Discussion

We argued in our discussion of the normative issues that the key issue in the voting age debate is political maturity, and we have reported various empirical results which suggest that sixteen and seventeen-year-olds are considerably less mature than older people.We showed that:(1) a disproportionate share of 16 and 17-year-olds are not interested in politics; (2) that 16 and 17-year-olds are especially likely to report political non-partisanship;(3) that they are less knowl- edgeable about political facts;and (4) their attitudes are less consistent at one point in time, and are less stable over time. However, it is very important to note that,

while older people generally lost interest in politics during the 1990s, teenagers actually became more interested in politics and more partisan over the same period.We interpret this as an ageing or life-cycle effect.The teenage years, we believe, are a period of political awakening.

What are the implications of our argument and ndings? In this concluding section, we consider three sets of issues. First, we consider a counterfactual question:how competent would sixteen and seventeen-year-olds be if they were enfranchised? Secondly, we consider an invalid inference from our argument, namely, that we should take the voteaway from 18, 19 or20-year-olds who are politically just as immature as 16 or 17-year-olds.Finally,we consider a question to which we do not yet have a satisfactory answer: what is the absolute level of political competence by which to judge any age group?

Counterfactual

Note that our data test for presentcompetence and interest in politics and, therefore, do not speak directly to the relevant counterfactual, how competent sixteen and seventeen-year-olds would be if they had the vote.We must address this issue with an open mind, for as Mill insisted in his defence of the emanci- pation of women, even if differences in competence are real, these might be the effect of subjugation which, therefore, cannot be taken as a basis for that subju- gation (Mill,1991 [1869],ch.1,para.18).In a similar vein,we must not be content to assess the present political competence of sixteen and seventeen-year-olds,but should ask how competent they would be if they were enfranchised. That counterfactual question takes into account the possibility that having the vote might have positive effects in terms of one s ability and willingness to engage in politics.

It is very dif cult,if not impossible,to address counterfactual claims satisfactorily. However,we are sceptical of the view that the competence gap would disappear if sixteen and seventeen-year-olds were given the vote.This is because recent research in neuroscience suggests that the brain,speci cally the prefrontal cortex, is still undergoing major reconstruction and development during the teenage years and perhaps for some time in young adulthood as well (Gieddet al.,1999). Since the prefrontal cortex is crucial for impulse control and abstract thinking, and is the part of the brain that enables us to weigh dilemmas,balance trade-offs and, in short, make reasonable decisions in politics and other domains of life, it would seem that the competence gap that we report in the third section has some biological bases.23 If that is indeed the case,it seems unlikely that simply enfran- chising sixteen and seventeen-year-olds would make them politically mature voters.The fact that the teenager s brain is still under development explains the behavioural dif culties of many teenagers,and as Richard Dawkins and Elisabeth Cornwell (2003) remark,it might also suggest that [t]he brain just isn t ready to vote at 16 .24

If this is the case, should sixteen and seventeen-year-olds be given the vote? Recent research on turnout suggests that voting is very much a matter of habit (see e.g. Green and Shachar, 2000; Plutzer, 2002). It would appear that voters really need to decide whether to go to the polling booth in the rst few elections at which they are eligible to vote,and then habit takes over (Franklin,2004,p.13). Now we know that sixteen and seventeen-year-olds are especially uninterested in politics, but we also know that over the next several years of their lives, they go through a remarkable period of political awakening. Do we want our young voters to develop a habit of not voting when they are not interested in politics?25

Taking theVote Away from Eighteen-year-olds?

On our criteria, the data we have indicate a competence gap between young people into their early to mid-20s and older age groups,and not just between 16 and 17-year-olds and older citizens. Should we, therefore, embrace the bolder conclusion that the franchise ought to be takenaway from certain age groups — 18, 19 and 20-year-olds, for example — who currently enjoy it? If we resist that conclusion, are we not arbitrarily discriminating against sixteen and seventeen- year-olds in refusing to apply the same conditions to their neighbouring age groups?

These inferences are invalid,partly for reasons discussed earlier where we argued for a suitable stopping point in the face of the argument for votes at sixteen from an appeal to insigni cant differences.A second reason against raising the voting age draws on a relevant distinction between extending rights to individuals who presently lack them and withdrawing rights from those who presently enjoy them.Even if 18 and 19-year-olds are no better placed than 16 and 17-year-olds to contribute to our democracy,we should be wary about thwarting their legally af rmed expectation that they may engage in political participation through the act of voting.Adiminutionof legal rights might have symbolic signi cance which could be detrimental to the self-respect and self-esteem of the individuals con- cerned.26 Thus, the empirical ndings relevant to the voting age may have a different signi cance for the young who lack voting rights compared to neigh- bouring age groups that presently enjoy them.

Absolute vs. Relative Competence

The data presented in the third section offer both comparative information about the relative competence of different age groups and non-comparative informa- tion concerning the absolute level to which young people are able and willing to engage in democratic choice.Our preceding normative discussion suggested that the absolute level of political competence is the key variable in determining whether the voting age should be reduced.What matters is whether young people are suf ciently competent taken as a group.Yet we have not supplied, still less

defended,a particular view of what the appropriate level of competence must be. Young voters and would-be voters are apparently less competent than older voters, but are they competent enough? This is where further normative and empirical work is essential.We believe the available data indicate that the sixteen and seventeen-year-old age group lacks suf cient political maturity to be enfran- chised.Nevertheless,we recognise the need for a more ne-grained set of criteria that would enable us fully to justify this conclusion.

Conclusion

Our aim in this article has been to contribute to the voting age debate by drawing on relevant considerations from both normative and empirical sides of the discipline.From a normative point of view we rebutted simple arguments which assert that the debate should be resolved by appeal to public opinion or the ideal of consistency between different legal rights.Nevertheless,we defended the claim that the setting of a minimum voting age might be just and ef cient, and thereafter we itemised a set of criteria to test for political maturity.Motivated by these normative considerations,our examination of relevant BHPS and other data indicates that there is an age gradient with respect to political maturity. On the basis of our discussion of the survey evidence,we claim that there is aprima facie case against lowering the voting age.Yet,we are conscious that further normative and empirical work is necessary for our argument to be demonstrated conclusively.

Appendix: BHPS Questions Used in Constructing the Two Attitudes Scales

**Government (Questions included in 1992, 1994, 1996, 1998 and 2001)27

• On the whole,what governments do in Britain/UK re ects the wishes of the people (koppola).
• Ordinary people don t really have a chance to in uence what governments do (koppolb).
• Governments can be trusted to place the needs of the nation above the interests of their own party (koppold).

Women s Employment (Questions included in 1991, 1993, 1995, 1997, 1999 and 2001)

• A pre-school child is likely to suffer if his or her mother works (kopfama).
• All in all, family life suffers if a woman has a full-time job (kopfamb).
• A woman and her family would all be happier if she goes out to work (kopfamc).
• Both the husband and wife should contribute to the household income (kopfamd).
• Having a full-time job is the best way for a woman to be an independent person (kopfame).
• A husband s job is to earn money;a wife s job is to look after the home and family (kopfamf).

(Accepted: 27 September 2005) About the Authors

Tak Wing Chan, New College, Holywell Street, Oxford OX1 3BN, UK; email: tw.chan@new.ox.ac.uk*

Matthew Clayton,Department of Politics and International Studies,University of Warwick,Coventry CV4 7AL,UK; email:m.g.clayton@warwick.ac.uk*

Notes

We thank David Baker, Kathy Behrendt, Philip Cowley, John Cunliffe, David Denver, Justin Fisher, Steve Fisher, Anthony Heath, Randy Metcalfe, David Stevens, Katherine Watson,Andrew Williams, three anonymous referees, the editor of the journal and a seminar audience at Oxford for useful suggestions.

1 For more sophisticated treatment of this issue, see Dworkin (1996, p. 15 ff.).

2 A related but separate issue is the long-term effect of giving sixteen-year-olds the vote on their own turnout rate

in the future.We return to this issue later.

3 In this and the following two sections,we draw on the longer discussion in Clayton (2006,ch.5).The publisher s

permission for this adaptation is gratefully acknowledged.

4 For related and further objections to direct competence tests, see Archard (2003, p. 25).

5 See Quinn (1993) for a general discussion of problems associated with insigni cant differences and of various ways

in which we might determine an appropriate stopping point in cases of this kind.

6 See Raz (1986) for discussion of the normal justi cation of political authority . For further criticism of

consistency arguments for lowering the voting age, including those that rest on the slogan of no taxation without representation , see Cowley and Denver (2004).

7 We only consider the variation of political maturity by age.Although political maturity might also vary by gender

or other social-demographic characteristics, because selective enfranchisement of any age group is not under consideration, we will not undertake multivariate analysis in this article.

8 The same question was also asked in 1992 through to 1996.

9 The identi cation problem arises because the value of any two of the three variables jointly determines the value

of the third. For example, if a 20-year-old woman was interviewed in 2001, we know that she was born in 1981; and if a man born in 1960 was interviewed when he was 31 years old, we know that he was a respondent of the 1991 survey. In other words, the three variables actually give us only two independent pieces of information.As a result,we cannot unambiguously distinguish the effects of ageing,cohort and period.For a more detailed discussion of the ageing, period and cohort effects in the context of British politics see, for example, Tilley (2002).

10 There is quite strong evidence for cohort effects in political attitudes over the life course (e.g. Alwinet al., 1991)

and in turnout (e.g. Franklin, 2004).We shall return to this issue of ageing vs. cohort effects later.

11 Although 2001 was an election year,and 1991 was widely expected to be an election year,the pattern reported in

Figures 3 and 5 (see below) holds for 1994,a year in the middle of the UK electoral cycle.Details are available from the authors on request.

12 Net gain in interest in politics is calculated as follows:

Ng −N l × 100,

N

where Ng is the number of respondents for a particular age group who reported no interest in politics in 1991 but became interested in politics in 2001;Nl is the number of respondents who reported interest in politics in 1991, but not in 2001; and N is the total number of respondents of that particular age group.

13 The lines for category 2, i.e. non-supporters with af nity to a political party, are much atter, especially for 2001.

Details are available from the authors on request.

14 If the 1991 line for supporters of political parties were shifted ten years to the right, it would very nearly overlap

with the observed 2001 line.Thus,for supporters,if one were to take the very strong cohort effect assumption,there is no need to postulate any period effect.

15 That is, the value of the solid data points are calculated as follows:

N 3→1 +N 3→2 −N 1→ 3 −N 2→ 3 × 100,

N

where N3→ 1,for example,is the number of respondents who were non-partisans in 1991 but supporters of a political party in 2001,N2→3 is the number of respondents who were non-supporters but with af nity to a political party in 1991 and non-partisans in 2001, andN is the total number of respondents of that age group.

16 That is, the value of the data points on the dashed line is calculated as follows:

N 3→1 −N1→3 × 100.

N

17 A similar pattern is observed for the 1991 BHPS data. Details are available from the authors on request.

18 We thank an anonymous reviewer who referred us to the work of Sarah Butt. Butt (2004) uses almost the same

instrument as ours and shows that voters who are more knowledgeable of political facts are better in identifying the policy positions of political parties.They are, therefore, better in matching their own issue preferences with party platforms. More knowledgeable voters are also more aware of the state of the economy and the government s economic performance. Furthermore, although both high and low-knowledge voters blame the government for changes in the standard of living, for more knowledgeable voters,evaluations of the national economy are less likely to be a simple re ection of their own experience (Butt,2004,p.13).In other words,our measure of factual political knowledge does seem to have wider validity as a measure of political maturity.

19 TheYBSA item that is similar to item 1 of BES reads: Tony Blair is the present prime minister .The threeYBSA

items that are different from BES are: Great Britain is a member of the European Union , Northern Ireland is part of the United Kingdom and Women are not allowed to sit in the House of Lords .

20 Thus, the data points refer to 12 and 13-year-old, 14 and 15-year-old, and so on.

21 Alpha ranges from zero to one,with high alpha values indicating high levels of consistency.It should be noted that

the magnitude of the coef cient depends on both the average correlation among the items (i.e.internal consistency) and the number of items in the battery.

22 A sharper increase in alpha regarding trust in government among younger respondents can be found if we use BHPS

data from other years.But in those cases the alpha values are also more erratic in general.Details are available from the authors on request.

23 For a non-technical introduction to recent ndings on neuroscience and teenage development,see Strauch (2003).

24 One possible way to address the counterfactual is to examine what happened when voting age was lowered from

21 to 18 in 1969.At this point,we are not sure if appropriate survey data are available from the 1960s.In any case, given the space constraint, we will have to leave this question for a future paper.

25 We argue earlier that the question of the voting age should not turn on its effect on the overall turnout rate. But

notice that here we are not considering the overallturnout rate, but the future turnout rate of sixteen and seventeen-year-olds themselves. Notice, in addition, that the importance in terms of habit formation of voting in the rst few elections in which one has a right to vote tells against the proposal to give sixteen and seventeen- year-olds half a vote.While this proposal seems to lessen the maturity problem,it is no better than offering the full franchise in so far as habit formation is concerned. If anything, because their votes would count for less, it would be unsurprising if sixteen and seventeen-year-olds with half-votes were even less likely to vote.

26 It is clear that in certain cases there can be positive symbolic signi cance in a diminution of rights.Historically,the

elimination of the right of young children to work symbolised an enhancement of their status and the concern to which they were entitled.Nevertheless,in the voting case under discussion,the relevant analogy seems to be with the withdrawal of rights related to the self-determination of criminals or the demented. In these cases, the withdrawal of rights often symbolises a loss of status, and is associated with a loss of self-esteem.

27 BHPS variable names are shown in parentheses.The answer categories of items marked by a are reversed before

calculation of the alpha coef cient.

References

Alwin,D.F.,Cohen,R.L. and Newcomb,T.M. (1991) PoliticalAttitudes Over the Life Span:The BenningtonWomen

after FiftyYears. MadisonWI: University of Wisconsin Press.

Archard, D. (2003)Children, Family and the State. Aldershot: Ashgate.

Butt,S. (2004) Political Knowledge and Routes to Party Choice in the British General Election of 2001 ,British

Elections and Parties Review, 14, 3—17.

Clayton, M. (2006) Justice and Legitimacy in Upbringing. Oxford: Oxford University Press. Cowley, P. and Denver, D. (2004) Votes at 16?The Case Against ,Representation, 41 (1), 57—62.

Curtice,J. (2004) What s the Point of Giving 16-Year-Olds theVote? ,The Independent,5 May 2004.Available

from: http://comment.independent.co.uk/commentators/article58950.ece [accessed 19 July 2004].

Dawkins,R. and Cornwell,R.E. (2003) Dodgy Frontal Lobes,Y dig?The Brain Isn t Ready toVote at 16 ,The

Guardian, 13 December 2003. Available from: http://education.guardian.co.uk/higher/comment/story/ 0,,1107333,00.html [accessed 17 December 2003].

Dworkin, R. (1996) Freedom s Law:The Moral Reading of the American Constitution. Cambridge MA: Harvard

University Press.

Dworkin,R. (2000) SovereignVirtue:TheTheory and Practice of Equality.Cambridge MA:Harvard University Press.

Electoral Commission (2003)How Old is Old Enough?The MinimumAge ofVoting and Candidacy in UK Elections.

Consultation paper. London: Electoral Commission.

Electoral Commission (2004)Age of Electoral Majority. London: Electoral Commission.

Franklin,M.N. (2004) VoterTurnout and the Dynamics of Electoral Competition in Established Democracies since 1945.

Cambridge: Cambridge University Press.

Frazer, E. and Macdonald, K. (2003) Sex Differences in Political Knowledge in Britain ,Political Studies, 51,

67—83.

Giedd, J. N., Blumenthal, J., Jeffries, N. O., Castellanos, F., Liu, H., Zijdenbos, A., Paus, T., Evans, A. C. and

Rapoport, J. L. (1999) Brain Development during Childhood and Adolescence: A Longitudinal MRI Study ,Nature Neuroscience, 2 (10), 861—3.

Green,D.P. and Shachar,D. (2000) Habit Formation and Political Behaviour:Evidence of Consuetude inVoter

Turnout , British Journal of Political Science, 30 (4), 561—73.

Harris,J. (1982) The Political Status of Children ,in K. Graham (ed.),Contemporary Political Philosophy:Radical

Studies. Cambridge: Cambridge University Press, pp. 35—55.

Heath, A. (1986) Do People Have Consistent Attitudes? , in R. Jowell, S. Witherspoon and L. Brook (eds),

British Social Attitudes:The 1986 Report. Aldershot: Gower, pp. 1—15.

Heath,A.,Evans,G. and Martin,J. (1994) The Measurement of Core Beliefs andValues:The Development of

Balanced Socialist/Laissez Faire and Libertarian/Authoritarian Scales ,British Journal of Political Science, 24 (1), 115—32.

Mill, J. S. (1991 [1869]) The Subjection of Women , in J. Gray (ed.),On Liberty and Other Essays. Oxford:

Oxford University Press, pp. 469—582.

Nunnally, J. C. (1978) PsychometricTheory, second edition. NewYork: McGraw-Hill.

Par t, D. (1986) Overpopulation and the Quality of Life , in P. Singer (ed.), Applied Ethics. Oxford: Oxford

University Press, pp. 145—64.

Park, A. (1999) Young People and Political Apathy , in R. Jowell, J. Curtice, A. Park and K. Thomson (eds),

British Social Attitudes:The 16th Report. Aldershot: Ashgate, pp. 23—44.

Plutzer, E. (2002) Becoming a HabitualVoter: Inertia, Resources, and Growth inYoung Adulthood ,American

Political Science Review, 96 (1), 41—56.

Quinn, W. (1993) The Puzzle of the Self-torturer , in Morality and Action, Cambridge: Cambridge University

Press, pp. 198—209.

Raz, J. (1986) The Morality of Freedom. Oxford: Clarendon Press.

Russell,A.,Fieldhouse,E.,Purdam,K. and Kalra,V. (2002)Voter Engagement andYoung People.Research report.

London: Electoral Commission.

Schrag,F. (2004) Children and Democracy:Theory and Policy ,Politics,Philosophy and Economics,3 (3),365—79. Strauch, B. (2003)Why areThey soWeird?What s Really Going on in aTeenager s Brain. London: Bloomsbury.

Tilley, J. (2002) Political Generations and Partisanship in the UK, 1964—1997 ,Journal of the Royal Statistical

Society, Series A, 165 (1), 121—35.

Tourangeau, R., Rips, L. J. and Rasinski, K. (2000)The Psychology of Survey Response. Cambridge: Cambridge

University Press.

Votes at 16 (2003) Submission to the Electoral Commission s review of voting age How Old is Old Enough . ' 2006The Authors. Journal compilation ' 2006 Political Studies Association POLITICAL STUDIES: 2006, 54(3)