Scotland will be voting in just eight months time on whether to maintain or dissolve the Union. There has been much speculation about the result. But remarkably little of it has investigated the lessons we can draw from past referendums in the UK and around the world. In this post, Reading politics lecturer Alan Renwick explores the evidence.
Do we know what the outcome of this year’s Scottish independence referendum will be? The polls have been remarkably stable for months, suggesting something like a 60:40 split in favour of maintaining the Union. Leading commentators, however, continue to urge caution in calling the result. Andrew Marr’s answer to the question of whether Scottish voters will opt for separation is “They micht.” The leading political scientist and BBC elections pundit Professor John Curtice says “the polls may not have moved much so far. But the important arguments and debates that will take place in the next 12 months could still make a difference.”
Such sages have a point: opinion polls record responses at the time they are taken; they do not tell us how opinion will evolve over the coming months. As the Canadian political scientist Lawrence LeDuc has shown, opinion around referendum questions is often much more volatile than opinion in elections, where many voters have stable party attachments that anchor their vote.
But we can learn more from past referendums than just that opinion changes. We can also examine how opinion changes and the circumstances that can be expected to generate one type of change rather than another. Through such comparative analysis, we can say much more about the likelihood of different outcomes in Scotland next September.
Patterns in referendums past
LeDuc’s work provides a starting point for this analysis. He examines support for the Yes option in a range of referendums from the 1980s and 1990s, comparing the final result in each case with a poll taken around one month earlier. His findings are summarized in Figure 1. As is apparent, in most cases, Yes support falls – often dramatically. By contrast, in only one case – New Zealand’s first electoral reform referendum, in 1992 – did the Yes vote substantially rise. Smaller increases occurred in five other cases: four referendums on EU accession in Austria, Finland, Sweden, and Norway in 1994 and the Quebec independence referendum of 1995.
Figure 1. Opinion change during referendum campaigns, 1980s and 1990s
Figure 2 updates this evidence for referendums taking place in long-standing democracies (countries that have been stably democratic since before the 1960s) since 2000. Here the comparison is between the final result and the average of polls conducted between six months and one month previously. The picture is very similar: in most cases, support for Yes goes down, sometimes substantially; only once – in Italy’s 2009 electoral reform referendum – did support for change show any significant rise.
Figure 2. Opinion change during referendum campaigns since 2000
In general, then, support for change ebbs as the referendum approaches. But that does not (yet) allow us to predict the same for Scotland next year, as there are exceptions. If we are going to edge towards a prediction we need to know more about the underlying causal mechanisms and then consider which mechanisms are likely to operate in Scotland.
Explaining the patterns
First, then, why does support for change generally fall? The main reason is that uncertain voters generally end up sticking with the devil they know. If you are uncertain quite what implications any given change will have, then it may be safer to hold to the familiarity of the status quo. This mechanism is accentuated if the idea of reform sounds appealing at first blush: voters may respond positively to pollsters when they have not really thought the matter through, only for doubts to develop as they begin to engage later. Examples of these mechanisms abound. The most familiar for many British voters will be the electoral reform referendum of 2011: the idea of shaking up the political system, particularly in the wake of the expenses scandal, initially appealed to many voters; but as voters engaged more, they worried about many of AV’s possible implications, and most ended up voting No. More strikingly still, Ireland’s voters opted in October 2013 to retain their Senate even though polls has long shown a majority for abolition: fear of empowering the government too far changed many minds.
In other cases, the mechanism that works against change is rather different. Where the political establishment unites behind a Yes vote, voters occasionally choose to rebel – a response that is particularly likely where the referendum is viewed as non-decisive. In Ireland’s referendum on the Lisbon treat in 2008, for example, voters knew that a No vote would only precipitate a renegotiation and second referendum, so an expression of disgruntlement carried few dangers.
A number of mechanisms can, however, dampen the shift against Yes or even generate a momentum in favour of change. One is the obverse of the mechanism just mentioned: if the establishment as a whole opposes reform and the vote appears non-decisive, a bandwagon for change can sometimes gather speed. Both conditions applied in New Zealand in 1992: voters wished to express their anger over the behaviour of the political class; a Yes vote simply meant another referendum and a second chance to decide the following year.
More commonly, however, support for Yes grows because supporters of the change successfully convince voters that the status quo is not an option. Political scientist Sara Binzer Hobolt refers to the situation that ensues following a No vote as the referendum’s “reversion point”. Generally, the reversion point is the status quo: if voters opt against change, then the pre-existing situation continues. But sometimes, whatever the formal options on the ballot paper might be, the reality is (or can be claimed to be) that the pre-existing situation is not sustainable. In a number of European countries, for example, voters have been asked to vote a second time on an EU treaty that they have previously rejected and have been warned that a second No vote could jeopardize their country’s position in the Union. Things would not just carry on as before (voters have been told) in the event that the proposal was rejected. Similarly, in the four EU accession referendums of 1994, Yes campaigners argued that, in a globalizing post-Cold War world, isolation was an increasing danger and the old way of doing things was no longer sustainable.
Finally among mechanisms that may limit any drift from Yes, sometimes opinion is just very settled well in advance of the vote and the number of floating votes is therefore low. Scotland in 1997 provides a good example. It was, famously, the “settled will” of the Scottish people that they should obtain a degree of self-government, and the polls barely changed throughout the referendum campaign.
Drawing lessons for Scotland 2014
So what are the implications of all this comparative analysis for the Scottish referendum of 2014? We can start with the final point, about the degree to which opinion is already settled. It is common to point out that many voters remain undecided. Yet a comparison with the “settled will” referendum of 1997 casts doubt on that: in the polls taken over the summer of that year, an average of 15 per cent of respondents said they did not know how they would vote on the devolution question; in recent polls, the average “don’t know” rate has been virtually the same. That suggests we might not expect a huge level of change. It could be that the polls are not telling the whole story here: that the positions of many of those who do state a position are softer than they were in 1997. Nevertheless, the evidence we have does not point to unusually high levels of uncertainty in Scotland at the moment compared with other referendums.
The main contest in the referendum campaign will be the battle over uncertainty and fear. No campaigns always gain support primarily by raising doubts about what the reformed situation will really be like. That is why the Better Together campaign has been raising many doubts around the independence option: about whether Scotland could stay within the EU, whether it could keep the pound, and so on. Yes campaigns can do something of the same by shifting the referendum’s reversion point away from the status quo. The SNP argues, for example, that Scotland’s position within the EU can be secured only by a Yes vote: that if Scotland stays in the UK, it may be dragged out of the EU by English voters in a referendum in 2017. There can be little question, however, that the uncertainties of independence outweigh the uncertainties of maintaining the Union. Other things being equal, we should expect this to generate a drift in the direction of No in the run up to the referendum.
But other things, of course, may not be entirely equal. While No campaigns win largely through negative campaigning, there can be dangers in taking the negativity too far. In the case of 2014, if unionist politicians seem to be doing Scotland down, they may provoke a counteraction. That will be especially so if it looks as though the Westminster establishment is united in disrespect for Scotland and its capacity to govern itself. Anti-establishment feelings run high in almost all democracies, but a visceral loathing of “posh boy” politicians such as David Cameron, George Osborne, and, for many, Nick Clegg is particularly intense in Scotland. The No campaign will therefore need to be careful to avoid any perception of sneering and to balance negativity with a positive vision of Scotland’s future. Scope for an anti-establishment protest vote is limited by the fact that this is a referendum the outcome of which really matters. But Better Together nevertheless need to be careful in how they balance their campaign.
So do we know the referendum result already?
Overall, all the evidence points to a comfortable majority against Scottish independence when the referendum comes next September. There is already a sizeable No majority in the polls, opinion generally shifts towards No as referendums approach, and few of the conditions that might counteract that trend apply in Scotland. Still, all of this analysis presupposes a competent campaign on the part of the advocates of continued union, and there are particular dangers that the No campaign will need to avoid if they are to secure the prize that seems ready for them to take. Better Together should win easily next year, but only if they do not become complacent.
 Lawrence LeDuc, “Referendums and Elections: How Do Campaigns Differ?”, in David M Farrell and Rüdiger Schmitt-Beck, Do Political Campaigns Matter? Campaign Effects in Elections and Referendums (London: Routledge, 2001), pp. 145–62, at p. 157.