Scottish Affairs CommitteeWritten evidence submitted by John Curtice, Professor of Politics University of Strathclyde

The Merits of Holding a Single Question Versus a Multi-Option Referendum on Scotland’s Constitutional Future

1. On both sides of the debate, there is a professed wish that the outcome of the forthcoming referendum on Scotland’s constitutional future should be widely accepted by politicians and the public. The UK Government has stated that the referendum on Scotland’s constitutional future should be “legal, fair and decisive”. The Scottish Government has indicated that the conduct of the referendum needs to be “beyond reproach”. In any event, any referendum that lacked these qualities would be unlikely to provide a stable basis for the future government of Scotland.

2. However, there is considerable disagreement about the form that such a referendum should take so that that objective is achieved. The UK government takes the view that to be “decisive” the referendum should only ask a “single straightforward question” on whether the Scotland should remain part of the UK or become an independent country, arguing that “on this one issue” such a vote would “end the uncertainty”. In contrast, in the belief that “there is support across Scotland for increased responsibilities for the Scottish Parliament short of independence”, the Scottish Government has stated that “it is willing to include a question about further devolution on the lines of “devolution max” if there is sufficient support for such a move”.

3. There are different ways in which a referendum might be considered “decisive”. One is that it is widely accepted as providing an authoritative decision on the particular question put before voters. The other is that it is widely regarded as providing an authoritative resolution of the issue that the referendum addresses. The referendum on the alternative vote held in May 2011 might be considered a referendum of the first type. It is widely accepted as having resolved the question of whether that particular system should be used to elect the House of Commons without necessarily having ended the debate about the respective merits of using a proportional rather than a majoritarian system. In contrast, the referendum in 1975 on Britain’s membership of the then Common Market might be considered to have resolved for a considerable period of time at least the debate about whether Britain should be a member of that organisation.

4. A referendum is less likely to resolve an issue if it fails to encompass (and specify fairly) what are widely regarded as all of the key policy alternatives. Because of the wide variety of electoral systems in existence, it is always quite likely that any referendum that asks voters to choose between just two alternative systems will fail to bring a debate about electoral systems to a conclusion.

5. However, the decisiveness of a referendum may be thought to depend not only on the character of the options put before voters but also on whether the outcome unambiguously identifies which is the most popular option. Achieving that objective is often regarded as less straightforward in the case of a referendum in which more than two possible alternative responses appears on the ballot paper. This is despite the fact that it is rarely regarded as unproblematic to hold an election in which voters are asked to choose between more than two candidates, as is nowadays the usual position in elections to the House of Commons.

6. This distinction between an election and a referendum arises in people’s minds because it is widely felt that to be decisive a referendum has to demonstrate clear popular support for whichever proposition is declared the winner. Thus it is widely presumed that the winner should secure the expressed support of a majority of those voting. This requirement is also sometimes reinforced by a stipulation that the outcome of a referendum is only deemed to be valid if one of the propositions secures the assent of a given proportion of the total voting population. In contrast, in the case of elections it is not always considered necessary for the winner to secure the expressed support of a majority, an outlook that was endorsed in last year’s referendum in which the public opted to keep single member plurality rather than switch to the alternative vote.

7. It thus should not come as a surprise that referendums in which voters are asked to choose between more than two alternatives have only been held relatively rarely. However, they are not unknown. Key examples include:

Newfoundland (constitutional status) 1948.

Puerto Rico (constitutional status) 1967, 1993 and 1998.

Sweden (pensions) 1957.

Sweden (nuclear power) 1980.

Finland (prohibition) 1931.

New Zealand (prohibition) 1919–87.

New Zealand (electoral system) 1992.

8. Different approaches have been taken to conducting such ballots and determining which option should be declared the winner. In Puerto Rico, Sweden and Finland, voters were invited only to cast one vote, and the winning option was simply the one that secured most votes, irrespective of whether that constituted more than half of all those voting. In practice that threshold was reached on two of the three occasions in Puerto Rico and the ballot on prohibition in Finland, but not in the 1993 referendum in Puerto Rico or in either multi-option referendum in Sweden.

9. In the case of the referendums on prohibition, which were held contemporaneously with every New Zealand general election between 1919 and 1987, voters were invited to cast either one or two votes, as they saw fit. This practice might be regarded as a form of “approval voting” under which voters are invited to indicate which options they are willing to accept rather than simply which they prefer. The winning option was required to secure 50% of the vote if any change was to be implemented. Prohibition only narrowly failed to meet that threshold in 1919, but from 1928 over half of all votes cast were for no restriction beyond the existing law.

10. The expectation that whatever option is implemented should secure at least half the vote can be achieved by holding more than one ballot. In the event that no one option secures 50% of the vote, a second ballot is held in which the least popular option(s) is (are) eliminated and voters invited to choose between the two most popular options in the first vote. This was the approach used in Newfoundland in 1948, as a result of which the most popular option on the first ballot (self-government) was overtaken on the second ballot by the second most popular (becoming a province of Canada).

11. An alternative mechanism by which the criterion that the winner should have the expressed support of at least half of those voting may be met is through the use of an alternative vote ballot. This path was not pursued in any of the referendums cited in para. 7, but there would seem no reason in principle why it could not be.

12. The New Zealand referendum on the electoral system combined a two-ballot approach with the use of a “gateway” question to determine which option should be set against the status quo in the second ballot. In the first ballot voters were invited first of all to state whether they wanted to retain the current system, and if it were not to be retained, which of four possible alternatives they would prefer. As an overwhelming majority voted against the status quo (single member plurality) what proved to be the most popular alternative (additional member system) was included on a second ballot together with the status quo, and eventually secured a narrow majority. The second ballot was held even though well over half of voters had backed one of the alternative systems and more voters had voted for a second alternative (STV) than backed the status quo. The procedure might thus be regarded as one that gave the status quo a privileged position as well as ensuring that the eventual winner had the expressed support of a majority of those who voted.

13. Apart from the widespread expectation that the eventual winner of a referendum should secure the expressed support of a majority, a further criterion that it is often argued any winner of a multi-option ballot should fulfil is that it represents the Condorcet winner. The Condorcet winner is the option that is preferred by most voters in all of the possible pairwise comparisons of the options on the ballot paper. In other words, if there are three options on the ballot paper, A is the Condorcet winner if more people prefer A to B and more prefer A to C.

14. The attraction of identifying the Condorcet winner is that it is the option with the best chance of generating a consensus, and thus perhaps the option most likely to resolve an issue. However, there is no guarantee that a Condorcet winner exists –in the case of electoral systems one could well imagine that most voters might prefer the additional member system (AMS) to single member plurality (because it is more proportional), the Single Transferable Vote (STV) to AMS (because it gives voters more choice of candidates), but single member plurality to STV (because it retains single member constituencies).

15. There is then clearly a risk that a multi-option referendum could fail to ensure that the winner has the expressed support of 50% of those voting, and especially so if a simple plurality rule were used. Meanwhile, there is no guarantee that a plurality, two-ballot, or alternative vote system will necessarily pick as the winner the Condorcet winner (if it exists), with the risk being greatest with plurality rule. At the same time, however, there is good reason to believe from polling evidence that in the case of Scotland’s constitutional future a potential Condorcet winner does exist—and that a referendum on that subject that only invited voters to choose between independence and the status quo would result in the exclusion of that potential Condorcet winner from the ballot paper. If so, then such a referendum would seem unlikely to provide any kind of conclusion to the constitutional debate and in that sense at least prove decisive.

16. Surveys and polls that have invited voters to choose which one of independence, devolution max/plus and the status quo they most prefer have consistently found that none commands the support of half of voters. The following table summarises the findings:

	Independence	Devo Max/Plus	Status Quo
ScotCen/SSA 2010	28	32	27
ScotCen/SSA 2011	43	29	21
TNS/BMRB (11/12)	28	33	29
TNS BMRB (1/12)	26	30	32
ICM (1/12)	26	26	37
Panelbase (2/12)	31	26	28

Thus if a multi-option referendum is to be held in which the winner is seen to secure the expressed support of at least 50% of voters, it is highly unlikely that a simple plurality ballot will prove suitable

17. A Condorcet winner usually exists when voters’ preferences are arranged along a single dimension, such as from left to right. It appears to be the case that most voters’ preferences in respect of Scotland’s constitutional future do have this character, and that consequently a Condorcet winner probably does exist in the form of remaining in the UK but with the Scottish Parliament having significantly enhanced powers.

18. These features of public opinion are discernible from those polls that have asked voters separately whether they would vote for or against independence and whether they would vote for or against devolution max/plus. Most such polls find that a majority would vote in favour of devolution max/plus while a majority would vote against independence. The following table illustrates three recent examples; in each case those saying they did not know how they would vote have been excluded from the calculations.

	Independence		Devolution Max/Plus
	Yes	No	Yes	No
YouGov (1/12)	39	61	58	42
Ipsos MORI (1/12)	43	57	75	25
Panelbase (2/12)	47	53	66	33

In line with these findings the 2010 Scottish Social Attitudes (SSA) survey found that while 62% felt that the Scottish Parliament should make most decisions on welfare benefits and 57% said the same about taxation, just 31% reckoned Holyrood should be primarily responsible for defence and foreign affairs.

19. The contrast between the results displayed at para. 16 and those at para. 18 arises because the vast majority of those who support independence prefer devolution max/plus to the status quo. In the YouGov survey, 76% of supporters of independence were willing to vote for devolution max/plus, while in the Ipsos MORI poll, the equivalent figure was no less than 90%. (Note that don’t knows have been excluded from this calculation, which cannot be made for the Panelbase poll.) In similar vein, 94% of those who think that the Scottish Parliament should be responsible for defence and foreign affairs also believe it should have control over taxation. As a result, while the Union is preferred to independence, devolution max/plus is clearly preferred to the status quo.

20. No survey has asked people to choose specifically between devolution max/plus and independence, and thus strictly speaking we cannot be sure that devolution max/plus is currently the more popular of these two, and thus represents a Condorcet winner. However, it seems highly likely that those who prefer the status quo would prefer devolution max/plus to independence and thereby ensure that it was the more popular. In support of this expectation we find that, according to the 2010 SSA, only 8% of those who said that their first preference was the status quo believed that Holyrood should be responsible for defence and foreign affairs.

21. However, although it appears that devolution max is a potential Condorcet winner, it is evident from the table at para. 16 that not only would it not necessarily emerge as a winner in a plurality ballot, but it would also not necessarily (albeit with a lower likelihood) prove a winner under the alternative vote or a two ballot referendum—as there is no guarantee that it would emerge as one of the two most popular options amongst voters’ first preferences. Any such ballot might thus be potentially subject to procedural criticism. On the other hand, it might be felt that if devolution max/plus were not at least one of the two most popular options amongst voters’ first preferences this would mean that there was insufficient strength of support behind it to provide the basis of a resolution of the debate about Scotland’s constitutional future.

22. A number of methods have been devised for counting preferential votes (as in an alternative vote style ballot) in such a way that the Condorcet winner is identified. Typically such methods use the information contained in such ballots to ascertain which option is the winner in each pairwise comparison of the options. The apparent complexity of and current unfamiliarity with such methods may well be thought to militate against their use.

23. In practice, given that voters’ views do appear largely to lie along a single dimension, a Condorcet winner that is backed by the expressed support of at least 50% of those voting should be identified if voters were asked to vote separately for and against independence and for and against devolution max/plus. The winner would be whichever of those two options secured most support so long as it secured at least 50%. However, it would have to be accepted that such a ballot would be open to tactical manipulation; those voters whose first preference was independence and second preference devolution max/plus might vote against devolution max/plus to reduce its chances of outpolling independence. This indeed is a general problem with attempts to identify a Condorcet winner. It would also have to be accepted that it might be the case that an option that had secured the expressed support of over 50% of voters might fail to be declared the winner, on the grounds that another option had proved even more popular.

24. An alternative approach that might be thought less subject to tactical manipulation would be to adopting a variant of the procedure used to determine New Zealand’s electoral system. First of all voters would be asked whether or not they wanted Scotland to become an independent country or remain part of the UK. Second, they would be asked whether some form of devolution max/plus should be introduced if Scotland were to remain part of the UK. In this case, supporters of independence would have little reason not to indicate support for whichever unionist option they preferred. Consequently, in the event that independence was rejected it is highly likely that whichever unionist option was the more popular would secure the expressed support of at least 50% of those participating in the ballot. Note though that it would mean that independence would be declared the winner so long as it secured 50% support in response to the first question, though of course this would also be the case if the referendum were simply a vote for or against independence.

25. There is a clearly a risk that a multi-option referendum will fail to produce a result that proves indecisive because the method of voting and counting is disputed. There are though ways in which those risks can be reduced. Meanwhile, these risks need to be balanced against the risk that also apparently exists that a referendum that does not include some form of devolution max/plus will prove indecisive because, omitting a potential Condorcet winner, it fails to resolve the issue of Scotland’s constitutional future. A wise politician will wish to assess these relative risks carefully rather than dogmatically assert that one approach will indisputably prove more decisive than the other.

March 2012