Hansard

I can assure the committee that, by the standards of many university lecture theatres, this is a palatial building.The remit that I was given for appearing before the committee was to present the report that I wrote for the McIntosh commission, entitled, "An Electoral System for Scottish Local Government: Modelling Some Alternatives", and to expand and develop thereon. In interpreting that remit, I have given myself three tasks. First, I shall give a summary of what I said in the report to McIntosh and pick out the crucial conceptual points and the issues that this committee, the Kerley commission and eventually the Scottish Executive ought to consider in deciding on an alternative electoral system to the existing system for Scottish local government elections. Secondly—and this is not something that I was asked to do by McIntosh—I shall present some arguments on the problems with the existing system. They are also, however, problems that are not widely articulated or understood, even by advocates of electoral change. I shall try to open up the debate on why we are thinking about alternatives in the first place. Thirdly, I shall give members some insight into research on the Scottish public's attitudes to alternative electoral systems in the light of their experience of electing members of the Scottish Parliament on 6 May. The Centre for Research into Elections and Social Trends and the constitution unit at University College London were given a grant by the Economic and Social Research Council to investigate attitudes towards and understanding of the electoral system in both Scotland and Wales in the light of the first devolved elections. That research is just beginning, but I want to give the committee one or two headlines about it, because neither McIntosh nor anybody else has much evidence so far about what the public think.My first task is therefore to give some indication of what I was trying to say to McIntosh—not that McIntosh always listened to what I said, but that is another story. My first point challenges the way in which people conventionally think about electoral systems. Once one has decided that one wants an alternative electoral system, one should go to a bespoke tailor for it rather than buying something off the peg from Marks and Sparks. We should look for the electoral system that is best tailored for Scottish local government, given the objectives that one wants to set both for local elections and for local government. Electoral systems do not come in neat packages called single transferable vote, additional member system or alternative vote-plus. They come with a variety of options and ought to be tailor made. I will say more about that later. The second principal point that I made to McIntosh follows from my first point. In the classic debate about electoral reform, we ask what people think about STV or AMS and so on, but the response is that it depends on what one is trying to achieve. The first intellectual task for anybody who is thinking about desirable electoral systems, including evaluation of the merits of the existing system, is to decide what the objectives of local elections and of local authorities are. The merits or otherwise of electoral systems vary according to what one is trying to achieve. The criteria for evaluating the system therefore vary, too.The third principal message in my report is that, although one must go to a tailor, if one wants a suit, one cannot give the tailor a specification just for trousers. In other words, there are important trade-offs in devising electoral systems. One cannot have one's cake, eat it and regurgitate it for breakfast. At the end of the day, one must make choices across desirables; one may find that one's objectives point in somewhat contradictory directions. I will draw the committee's attention to two important trade-offs. As I said, STV, AMS and AV-plus do not come in simple packages. Their proportionality depends crucially on the way in which they are implemented. I could find highly disproportional versions of STV; and I could find highly proportional versions of STV. I could find highly disproportional versions of AMS; and I could find highly proportional versions of AMS. I could certainly find highly disproportional versions of AV-plus—indeed, the Jenkins commission justifies its version of AV-plus precisely on the ground that it is not a proportional system. The implementation and the detail matter. That is why the debate should be framed not around systems, but around the things that affect whether the systems will or will not produce proportionality and around the other criteria that we need to bear in mind.You may now be saying to yourself, "Oh dear—this is going to be one of those terrible academic lectures, in which I am told that the world is terribly complicated and difficult, and after which I will end up with no clearer idea of how to make a decision." Although I have said that you should not look at the world through one lens, my report offers you an alternative and relatively simple lens—but a lens that, in my view, clarifies the choices in devising electoral systems, and clarifies the things that influence the way in which those systems operate.I will outline four choices. The first is the most boring and the most technical. The report goes into some of the detail; I do not propose to go into it now, but you are welcome to ask about it. It is the so-called allocative rule. At the end of the day, no electoral system—PR or whatever—can produce perfect proportionality. If, for example, we have a local council with 25 seats, which implies one seat for 4 per cent of the vote, you have to decide whether a party that ends up with 2 per cent is entitled to a seat. You have to decide whether a party that gets 6 per cent of the vote is entitled to one seat or two. The answer to that question is not obvious, and there are different rules for providing the answer. As you probably know, in the Scottish parliamentary election the d'Hondt rule was used. It was also used to provide the United Kingdom's European parliamentary delegation. It is not the only rule, but as it happens—depending on the context—it tends to be relatively favourable to the larger parties and relatively unfavourable to the smaller parties. For example, if you reran the UK European parliamentary elections using a different rule—the Sainte Lague rule—the Conservatives would have five fewer seats, and Labour four fewer, than they do now. The rule you use can make quite a difference under certain circumstances.The second choice—and the choice that is much more important in determining the proportionality of the system than whether the system is STV or AMS, or whether you use Sainte Lague or whatever—is ward size. Opinion on that is now clear and virtually unanimous. A system will be more proportional the bigger the wards or constituencies that you use. If you have a 25-seat council, and you elect all 25 members in a single district, you can indeed ensure that you get one councillor for each 4 per cent or so of the vote. If on the other hand, the same council is divided into five separate wards with five councillors each, you will realise quite rapidly that you now need about 20 per cent of the vote in that ward to get elected. A party that always gets 10, 11, 12 or 13 per cent of the vote in each of the five wards may well end up with nothing at all; whereas if the council had been one district of 25 seats, it would have got two or three seats. Ward size is crucial to the proportionality of a system.The third choice, however—and the choice that can to some extent make it possible to achieve both small wards and proportionality—is to have a number of tiers. The additional member system that was used for the Scottish parliamentary election is one variety of a multi-tiered system. It is a system in which some MSPs were elected at what I would call a lower tier—a single-member ward of very small size and potentially very disproportional; in practice in Scotland such wards are highly disproportional. The effect of that is ameliorated by having a higher tier where the seats are allocated in such a way as to attempt to reduce the disproportionality generated at the lower tier. There are other methods—the additional member system is simply one way. The outcome also depends on whether there is one tier or, as in the Scottish case, eight different tiers. That is the other way in which systems can vary; to some extent, you can begin to have smaller wards and proportionality.Those first three choices—the allocative rule, the ward size and the number of tiers—in combination and interaction, primarily determine the proportionality of an electoral system. Whether it is d'Hondt's or Sainte Lague's allocative rule makes much more difference if there are large numbers of small wards. However, if there are large numbers of small wards, and a substantial upper tier—or more than one tier—you can still end up with a system that is proportional. The interaction between those choices primarily determines the proportionality of a system.The fourth choice—the system of candidate choice—does not have a great deal to do with proportionality, but as the debate over the European Parliamentary Elections Act 1999 showed, it can generate an awful lot of heat and perhaps a degree of light. At one end of the spectrum there are closed party list systems. In effect, voters can vote for parties, but they cannot indicate a preference for individual candidates. At the other end of the spectrum there are open list systems in which voters are required to indicate a preference for individual candidates. Under systems such as the single transferable vote, they can indicate preferences on more than one list. Voters could vote for a Liberal Democrat No 1, a Conservative No 2 and an SNP No 3. Open list systems simply give voters more than one vote and they can give each of those votes to a different candidate from a different party. The crucial element is that, under an open system, voters are able to express a preference for an individual independently of their party choice.There are also flexible systems in which voters are permitted, but not required, to express a preference for individual candidates. If they do not express a preference, it is assumed that they like the rank order that the parties themselves put up. Whether the actions of voters will overturn the rank order will depend first on the number of voters who vote against that order and, secondly, on the rules that are used to determine the impact of individual voters' choices. Those rules vary tremendously from one country to another.The main point, apart from the detail, is the principle of choice. There is a choice between systems that are, at the one end, completely closed and at the other, highly open, with a variety in between.Those are, essentially, the four important choices to make—the allocative rule, the ward size, the number of tiers and the candidate choice system. Once you have made those decisions, you might discover that you have something called STV or AMS or whatever.There are a couple of wrinkles, with one of which you are probably familiar. The other is one with which, in the Anglo-Saxon debate on electoral principles, people tend to be less familiar. It has, however, been quite important in debates on French electoral systems. If we use the example of a 25-seat council, the first wrinkle is that a party that gets around 3 per cent of the vote is likely to do well enough to get a seat, depending on the allocative rule. You may wish to specify that you want a system that is highly proportional but not necessarily one that makes it too easy for very small parties to get elected. Therefore, you would want to set a threshold that is higher than the natural minimum that would otherwise occur.That is precisely what will be done in the elections to the greater London authority, next year, when a de jure threshold of 5 per cent has been imposed instead of what would otherwise be the de facto threshold of around 3 or 4 per cent. It was not an issue in the debate on framing the Scottish parliamentary electoral system, because the de facto threshold was already about 5 to 6 per cent. However, the threshold can be raised artificially in a system that is otherwise highly proportional by setting a limit.The other issue—into which I shall not go too deeply—is that it is possible, under party list systems, for parties to declare a linking of their lists. Let us imagine that the red and blue parties decide to put up separate lists, but that both parties recognise that they have a fairly substantial common interest. They do not hate each other. They would like to be able to compete with each other, but they are prepared, to some extent, to form an electoral alliance. If those parties are allowed formally to link lists, under the rules for the allocation of seats, the red and blue votes are added together in the first instance: the lists are combined. If I vote for the red list, that counts as a vote for both the red and blue parties combined. In the allocation of seats, the red and blue lists are regarded as a single list, and the total number of seats to be won by red and blue is a function of their combined vote.Subsequently, the number of seats that have been won by red and blue parties combined are divided between those two separate lists, according to the individual shares of the votes that they have received. You may wonder what the hell the point of that is. First, the advantage of linking party lists is that, under most proportional representation systems, the parties are likely to increase their probability of winning seats at all, and large parties stand a chance of faring a little better. If red and blue parties regard themselves as parties that might form a coalition, but they are not sure whether they will win 50 per cent of the seats, by combining their lists, they might ensure that they win the extra one or two seats that would give them that 50 per cent. In other words, linking lists is a mechanism for encouraging parties to declare in advance their coalition preferences. Secondly, very small parties that, individually, may not pass the threshold, may, collectively, succeed in doing so. The mechanism can act both at the top and at the bottom of the scale. It would be imposed legally, entirely separately of the other four choices that I outlined.I have indicated that the first three choices are ways of creating proportionality. I now want to integrate my foregoing comments and describe the way in which a disproportional system can be created. By inference, the converse—how to create a proportional system—should be made clear. I shall then illustrate, taking a simple example from our reporter, McIntosh, how such things make a difference within systems that are called STV and AMS. It should become clear why the choice is not simply between those two systems.As I have indicated, if it is desirable to choose an allocative rule that creates disproportionality, the d'Hondt rule would be used—resulting in the system that is used for the Scottish and European parliamentary elections. To create further disproportionality, a local authority must be divided into many small wards—the smaller the wards, the less likely it is that the system will be proportional, however wonderful the allocative rule is. The third rule is to have multiple tiers. If there is a small number of geographically divided top-up seats, the system will be more disproportional. The smaller the number of top-up seats, and the more geographically divided those top-up seats are, the less proportional the system is likely to be. It is not widely appreciated that in no fewer than five of the eight regions in the elections to the Scottish Parliament, the disproportionality that was created by the first-past-the-post tier was greater than the seven top-up seats were capable of redressing. Even in a region with a 9:7 ratio, we still generated quite significant disproportionality.The fourth way of creating disproportionality—although we can argue about whether this really is disproportionality—is to have systems in which second preferences make a difference to the allocation of seats. Any system, such as the alternative vote system or the single transferable vote system, that allows voters to express second preferences encourages parties to declare in advance potential coalition alliances and, by so doing, to encourage their voters. If the red and blue parties think that they are going to be in coalition after the elections and want to ensure that they have a majority, the red party will tell its people to give their second and lower preferences to the blue party and the blue party will reciprocate. If two parties do that under systems that give voters preferences, they are likely to end up with a larger number of seats than one would expect, given their share of first-preference votes. On one famous occasion in Ireland, which uses STV, the combined first-preference votes of Fine Gael and Labour fell as compared with the previous election, but they managed between them to secure office because they persuaded their voters to transfer their second preferences.As I have indicated—and here we come to the wrinkles—high formal thresholds are another way of creating disproportionality. That may all sound wonderful in theory, but does it really make a difference? Let us take the example of Glasgow in 1995, using a simple index of disproportionality—examining how the proportion of seats won by each party varies from the proportion of votes.First, I have compared the outcomes under STV in two sets of circumstances. In the first set of circumstances, I use three-member wards; in the second, I use six-member wards—in other words, they are twice as big. I have held everything else constant—the second preferences have been held constant, there are no upper tiers and so on. What this illustrates clearly is the impact of ward size within the same system. With three-member wards, my simulation suggested that there would be disproportionality of 13 per cent, but with six members per ward there would be disproportionality of 8 per cent. Doubling the size of the wards from three to six members has a substantial impact on the disproportionality exhibited.My second comparison is between two varieties of the additional member system. Here I illustrate the potential impact of the number of top-up seats that are available. I have applied the top-up across the whole of Glasgow, rather than dividing the city up. In the first case, I assume that 75 per cent of Glasgow's councillors are directly elected and that 25 per cent are top-up members. In the second, the division is 50:50—that is the system that is used in Germany. With only a 25 per cent top up, we get disproportionality of 8 per cent, the same as with the six-member constituency under STV. However, if I choose 50:50 AMS, with the 50 per cent of top-up members allocated across the city as a whole, I get an outcome that is almost perfectly proportional. Because I am operating a system with a large proportion of top ups and not dividing those up, I get high proportionality. Once again, it is clear that under AMS the details of the system make a difference. That is why I say that the real issues are the size of wards, the nature of the tiers and, to some extent, the nature of the rule governing allocation, rather than a simple choice between STV and AMS. The details are fundamental.As I said, the second point that I made in my report to McIntosh was about the need to think about objectives before getting lost in the details of the system. You need to know what you want to achieve before you start worrying about how to go about it, as the arguments about what you want to achieve are the crucial normative arguments. You can get boring anoraks such as me to work out how to deliver your objectives—you just need to ensure that we are not telling you too many porkies.I want to move on a little from the report, as it would not be helpful for me simply to reiterate what I told McIntosh. First, let me remind the committee of what I regard as the McIntosh-Kerley objectives. I want to suggest that other objectives, implicit in the McIntosh report, are relevant to devising an electoral system in Scotland and I want to illustrate that there are conflicts among the objectives.The first criterion that McIntosh puts forward is proportionality. We have already talked about that. The second criterion is the councillor-ward link, about which I will talk when we consider the first-past-the-post system. It is clear to me from the debate that the Parliament had on 2 July that the subject inspires considerable enthusiasm. The third criterion is the fact that Scotland votes for a few independent candidates—McIntosh implies that we should not construct a system that makes it impossible for independents to be elected. The fourth criterion is that Scotland is a varied country and it might not be possible to have the same system for Glasgow as for the western isles. The fifth criterion, which has come out of the arguments with the existing local government boundary commission for Scotland, relates to the wish for ward boundaries to reflect communities rather than lines on a map.Members will be familiar with all that. It will be obvious to them that there is a conflict between the desire for councillors to represent relatively small wards—as they do if the councillor-ward link is defined in the conventional sense—and the desire for proportionality.The other factor—which might not seem to be a McIntosh-Kerley objective but which I think of as one—is the issue of appropriate numbers. If Kerley were to come to the conclusion that efficient delivery of services and effective community leadership could be delivered by 20 councillors in every council in Scotland, it would not be desirable to set that as the ideal number of councillors if it prevented proportionality from being achieved or the representation of communities from being enabled. Kerley's recommendations on the number of councillors need to take into account the other factors that the electoral system has to deliver. Other objectives are implied by McIntosh—I think that some of them are more important than the objectives that McIntosh explicitly set out. The first element is that, as I said, we have to think about what we want councils to do, rather than only what we want the electoral system to do. McIntosh sets out a clear objective for councils: he says that they are about providing services and representing their communities. We have to find the most effective way in which councils can do that. Do we believe that the best way is to ensure that a council is a microcosm of the political preferences of the electorate, in which case we would want a highly proportional system, or do we believe that the best way is to have majority governments in councils? The second element is that, as McIntosh has rejected the idea of elected provosts—for the time being, the Executive has rejected the idea—and proposes cabinet administrations instead, we need councils that are capable of sustaining an executive. Councils will still have to maintain an executive—if no one can form a majority administration, that may imply that a totally proportional system is inappropriate.The third element, which I think is particularly important, is that although McIntosh wants to increase proportionality—by implication, proportionality for parties—the report also says that parties should be less important in Scottish local government decision making. In particular, it calls for less use of whips. Those are wonderful words, but we need to think about how we construct a system for local government that ensures that there are incentives for that to happen. If we want a system that allows councillors to vote against a whip and that encourages councillors to think for themselves rather than follow the party line, we may not want a closed party list system, which makes it easy for a party to get rid of awkward councillors by putting them at the bottom of the list. If we want to reduce party power in local government, we need to devise an electoral system that does not reinforce party influence.Another element relates to the criticism that local government is too involved in details. That is an argument about the need for councillors to be more strategic—they need to be more concerned about the broad strategic direction of their authority and less concerned about the fine detail of which pavement currently needs mending. That may imply that some councillors should be elected to represent not small wards, but large geographical areas, which would give them an incentive to think about the district as a whole rather than in terms of individual wards. What we want councillors to do has a clear implication for the kind of electoral system that is needed. If we want councillors to be involved in discussing which pavement has not been repaired, we may want small wards. However, if we want councillors who are concerned about the broad strategic direction of services, we may not want councillors to be elected in small wards.The other two elements that we have to take into account are voter choice—whether voters should have as much choice as possible—and voter comprehension. Voters do not have to understand the relationship between seats and votes—to be honest, I do not think that many people understand that relationship under the first-past-the-post system, let alone under proportional representation. The cognitive task in which voters are required to engage in the polling booth needs to be one that the vast majority of the Scottish public can tackle straightforwardly.Those are some of the objections that need to be borne in mind. Some aspects of the McIntosh report that are not explicitly linked to the electoral system imply that there could be cross-pressures. We need to recognise the existence of trade-offs. As I have said, small wards produce low proportionality, all other things being equal. There is potential for conflict between some conceptualisations of the council-ward link and proportionality. There is no way in which we can have all councillors elected in small wards and secure proportionality. Those elements are in direct conflict with each other. It may be possible to ease the trade-off by introducing multi-tiering, through the additional member system, for example. However, the implication is that the additional member system that is used must be one in which a relatively large number of councillors are elected for fairly large districts. Even then, if some councillors are tied to their pavements, the price of achieving proportionality will be to have others tied to a relatively large area.Another factor, about which I have not said much but which is important, is complexity. You may believe in open lists and that voters should be able to choose between candidates, but if you also want large wards so that there can be a high degree of proportionality, a problem of complexity could arise. If, for example, we have open lists in wards that elect 10 or a dozen councillors, that will ask voters to make judgments about a large number of candidates. If we assume that the four major parties each put up eight or 10 candidates in the wards, that makes 40 candidates—with a few other parties, the ballot paper could have 60 names on it. There is a limit to how far voters can be expected to make meaningful judgments about a large number of candidates. The solution could be a flexible list, where voters can overturn the party ranking if they want, but do not have to. That reduces the cognitive task and enables voters that do not want to engage in that choice to ignore it. Those are two crucial trade-offs that we should be aware of in considering a system. Let me move on from McIntosh. I said that I wanted to assess the validity of the arguments for the existing system and to ask why we were considering change. Some aspects of change have not been widely articulated. There are two clear arguments in favour of the existing electoral system, which are perfectly defensible in terms of democratic theory. The first is that voters should decide who rules on the council. The argument says that the most important thing is not that every individual vote is faithfully reflected on the council, but that the voters decide who the leader of the council will be—as with the Prime Minister. Being able to determine who forms the executive is more important than determining the fine detail of the composition. By generating majorities for the largest party, first past the post ensures that voters determine the colour of the executive. The second argument is that the councillor-ward link ensures that councillors and political representatives have a close link to individual voters. My problem with the first argument is that first past the post cannot be relied on to achieve the objective that is set for it. I have no problem with the Labour party winning 74 out of 79 seats on Glasgow City Council on 49 per cent of the vote if, and only if, any other party that got 49 per cent of the vote also won 74 out of 79 seats. I would have no problem if the system were colour-blind in relation to the level of exaggeration. If the Labour party gets 49 per cent of the vote, it gets 74 seats, and if the SNP gets 49 per cent of the vote, it would also get 74 seats—that strikes me as fair, at least using the criteria of the traditional defence of first past the post. The problem is that that statement does not hold. The most interesting thing about the 1999 local elections in terms of the thinking about changing the system is not the results in Glasgow and Edinburgh—which people point to as examples of terrible disproportionality—but the fact that no less than 13 of the 32 councils in Scotland, despite the use of first past the post, ended up with no overall majority. The existing system fails to deliver—on any regular, reliable basis—single-party majorities in Scottish local government. If we want single-party majorities on a local authority level, it is not the right system—it is not delivering—so we would have to think of something else. The second problem with first past the post is that it cannot even be relied upon to identify the correct winner. Let me remind you of what happened in Dundee this year: the SNP got 36.4 per cent of the vote and Labour got 36 per cent, yet Labour got four more seats than the SNP. Edinburgh in 1992 is an even more notorious example. Labour got 29 per cent of the vote and the Conservatives got 40 per cent, but Labour ended up with seven more seats than the Conservatives. Just in case you think that it is always Labour that benefits, in East Kilbride in 1974 Labour won more votes than the SNP but ended up with two fewer seats. The system does not necessarily even get the winner right. The third element is that, as I implied, even when the system gets the winner right, it is not even-handed in its exaggeration. One can play games—some of this is in my report to McIntosh—and ask what would happen if there were uniform movement in an area so that the SNP were as far ahead of Labour as Labour is ahead of the SNP now. One discovers that, with the same lead, the SNP would not necessarily get as many seats as the Labour party does.That happens because, in first past the post, there is no regular or reliable relationship between seats and votes at council or national level. The result depends on how parties' votes are geographically distributed. I could demonstrate how some geographical distributions are advantageous in some circumstances but not in others. For example, a party whose vote is evenly spread geographically may come second in wards across the council and get very few seats. At a certain level of support, the party with even support will come first and, suddenly, everything falls into its lap—it moves from having nothing to having virtually everything.However, a party that does very well in some places but very badly in others will tend to do relatively well if it is narrowly in first place. Everything depends on geography. Do we want the results at council level in Scottish local government to be a function of electoral geography?The second main argument for first past the post relates to the councillor-ward link. In my view, that concept needs an awful lot of unpacking. What is the councillor-ward link under the single-member plurality system meant to deliver? As I understand the argument, it is meant to deliver two things. First, it is meant to deliver accountability—every councillor is accountable to a body of electors for their custodianship in the previous four years, or whatever, and their fate will be determined by the votes of—potentially—all the people in the ward, rather than by the votes of those who like that party and who may be making a choice between individual candidates of that party.The second thing—as I understand the argument—that the councillor-ward link is meant to deliver is service. Because every councillor represents an individual ward, they have a clear incentive to demonstrate that they are interested in the needs and demands of that ward and to ensure that they look after the problems and interests in their ward in terms of the requirements of community groups and individual constituents.If we accept accountability and service as desirable, we should consider what is the best system to achieve them. It is not entirely clear to me that single-member plurality is necessarily the best means of achieving those two desirables. The first point on that is fairly obvious: many councillors have safe wards, so it could be argued that their incentive to engage in the service function is limited and that accountability for their individual actions is nil.The second point is not widely appreciated but was used by the UK Government as an argument for defending the European election, so I take the Home Office as my source. The single-member plurality system is a closed party list system—it is a closed list of one. Most voters in Scotland vote on the basis of party label rather than on the merits of individual candidates—not entirely, I accept, but it is true for the most part, especially in central Scotland.Therefore, it can be argued that we already have a party list system—with a party list of one. Few councillors are accountable for their individual merits, as opposed to the merits of the party for which they stand. That was the argument that the UK Government used—"We already have a closed list system"—to defend the use of a party list system in European elections. I suggest that closed lists will not deliver accountability for a individual councillor's actions. The third element has been particularly problematic in Scotland in recent years. Because of the constant changes to ward boundaries, it is not necessarily the case that an individual councillor is accountable to the same body of voters that he or she has represented for the past four years. The final element is that there is an inherent contradiction between the first and second arguments in favour of a first-past-the-post system. The first argument is that elections are really about enabling voters to decide who runs the council—which political party should be in control. That implies a system of strong party control, where councillors who are elected for the controlling party will defend the policy of that party, even if it is not necessarily in the interests of their ward. A system where individuals look after the interests of their wards is a system where party is weak. A similar system operates in the US Congress, where congressmen regularly defy the party whip because their voters tend to vote for the merits of individual candidates rather than for the party for which those candidates stand. In such a system, a party may not be able to rely on the backing of a majority of its councillors. I will say a little about public opinion, based on academic survey evidence collected just after the Scottish Parliament elections. Are people wise after the event? A section of the UK population experienced for the first time at least a type of proportional representation. What did they make of it? We asked voters a number of questions, such as whether they thought that we should keep to the new way of voting, because it was a fairer system, or use the old way of voting, because it was more effective at delivering strong government. In the light of experience, at least on this occasion, the Scottish public seemed to be 2:1 in favour of a PR-type system. With some foresight, we even asked them whether they agreed or disagreed that we should use the same system for local government elections. Forty-five per cent agreed, 15 per cent disagreed—members will be able to ascertain from that that a fair number either do not care or do not know. This topic is not necessarily the most exciting for the general public. The survey also contained a number of other indicators, which we put to the Scottish electorate in 1997, about whether PR should be used for both Scottish Parliament and UK Government elections. In all those indicators, there was clear movement towards PR between 1997 and 1999. The broad judgment has to be that the experience of 6 May 1999 has made the Scottish electorate on the whole somewhat more sympathetic to PR in principle than they were previously. However, we should not run away with the idea that, even with that experience, the Scottish electorate are entirely consistent in their views. One of the long-standing findings of research in this area is that the answers obtained are very much a function of the questions asked. We asked people the broad, important, intellectual questions, which I have articulated, about whether elections should be about producing a clear winner or a fair result. That relatively abstract question still produces an almost 50:50 split. There is still room for argument on both sides about the merits of the PR system. I will give my conclusions in reverse order. First, from what I have seen so far, a reasonable supposition to make is that, should the Scottish Executive accept a recommendation from the Kerley commission for a PR system, endorsed by this committee, the odds are that the public would support that change.Secondly, first past the post does not necessarily achieve the objectives that are set out for it. If you want councils to have elections that are about controlling who runs the executive rather than the council as a whole, you should go for elected provosts; first past the post is the wrong instrument.Thirdly, if you accept the arguments that there is a movement for change and that first past the post is not satisfactory, finding the best alternative implies hard choices and clear thinking. There is a better way of looking at the debate about alternatives than the one that is commonly engaged in.