If Scotland votes ‘No’ on 18 September, what happens next? And might voters’ answers to that question influence how they vote in the referendum? Here, Rob Johns from the University of Essex analyses a British Election Study Internet Panel survey experiment which examined reactions to various post-‘No’ scenarios. It looks as if referendum voters will pay little heed to which party might win the 2015 UK general election, but rather more to whether that winner will agree to further devolution.
The Independence Referendum on 18 September 2014 looks like a straightforward choice between a major constitutional change and the status quo. Recent debate has focused on various implications of that change – for EU membership, for public spending, for the currency in which that money will be spent, and so on.Voters may still be unsure about what an independent Scotland would look like but this cannot be because the campaign has ignoredthese questions.
Less attention has been paid to the implications of a No vote. This makes sense, of course: there is bound to be more uncertainty about change than about the status quo. But there are good reasons for considering what happens if the ‘Better Together’ campaign wins out. For one thing, that still remains the most likely outcome. Even if the polls have narrowed recently, they still point unanimously to a No. And, taking constitutional referendums in general, support for change tends to ebb as the crunch approaches. The second reason is that status quos are themselves changeable. Scottish politics will not be set in aspic following a No vote. As the referendum approaches, voters’ thoughts are likely to turn to what will happen if the polls are right, and the conclusions they reach may influence the way they vote on 18 September.
Here, I take a look at what British Election Study Internet Panel (BESIP) wave 1 data tell us about two particular ‘known unknowns’:
- Which party will win the UK general election of May 2015?
Support for independence barely quivered after the 2010 general election, and so Conservative government in London is clearly not enough in itself to deliver a Yes vote. Nonetheless, the experience of Tory rule for which few Scots voted has been a powerful driver of support for constitutional change, and those who fear the same experience after 2015 might be more inclined towards Yes.
- How ready will that new government be to devolve more powers to the Scottish Parliament?
Throughout the referendum campaign (and long beforehand), there has been strong support for further devolution. An important factor in referendum decisions, then, is whether voters believe the UK parties’ pledges of ‘more powers’. Given the lack of harmony and clarity among those parties when it comes to proposals in this area, voters could be forgiven for both confusion and scepticism. The question is whether such scepticism pushes them towards a Yes vote.
So how do we find out whether these two issues really will influence referendum choice? One common polling method is to ask a hypothetical question. ICM recently asked voters how they would vote in the referendum if, at the time, they thought the Conservatives would win the 2015 general election. They then asked the same about Labour, and compared the results. A similar approach is used in the famous ‘£500 question’, asking people whether they would support independence in the event that it would make them so much better or worse off. Questions like this are better at generating headlines than real opinions, though. First, they are rather clumsy and leading. By including a reason in the question itself, pollsters make it hard to ignore and thus overstate its importance relative to the many other influences on voting decisions. The second problem lies in the hypothetical nature of the questions. Someone may think “well, I don’t believe for one minute that the Tories will win but, if I did, then yes, I might vote for independence”. We want to know whether people believe the premise, not just how they would act on it.
The alternative approach used here is known as a survey experiment. In this subtler approach, respondents are randomly chosen to read different arguments or pieces of information, and then we observe whether and how this influences their answers. So, in Wave 1 of the BESIP, a random 20% of the approximately 5,000 Scottish respondents read the following:
Referendum polling has consistently suggested a No vote and so attention has turned to what might happen afterwards. According to political expert Dr Mark Pickup from the University of Oxford, Labour look likely to win the most seats in 2015. The key question is how Labour will respond to a No vote. Experienced journalist Andrew Whitaker reports that Labour would then agree to transfer substantially more powers to Scotland – the so-called “devo-max” option.
Another 20% read a parallel story but with the likely 2015 winner changed from Labour to Conservative. Then there were two parallel groups which were told respectively that Labour and the Conservatives would refuse to transfer more powers to Scotland.(In each case, the experts cited were changed so that the information remained true.) That leaves the final 20% of the sample making up the control group that is needed in any experiment.
The crux of the experimental approach is that, because respondents are allocated information at random, the different groups begin the experiment with roughly the same level of support for independence. So any noticeable gaps that open up between the groups can be safely attributed to the different information that they read.
Following the experiment, respondents were asked to place themselves on a scale from 1 (“Scotland should definitely stay in the union”) to 7 (“Scotland should definitely become independent”). If we separate three groups of people based on that scale (1-2 = No, 3-5 = Don’t know, 6-7 = Yes), the percentages roughly match those in the opinion polls at the time: 49% for No, 35% for Yes and 16% undecided. The first question for us here, though, is whether the percentages depend on who is forecast to win the 2015 general election. And the first graph tells us that the answer is: absolutely not. Look first at the left-hand half of the graph, which is based on everyone who completed the survey. Those told that Labour would win split in more or less exactly the same way as those told to expect a Conservative victory.
The right-hand side of the graph shows the results only for those who said, earlier in the survey, that they didn’t know how they would vote in the referendum. These ‘floating voters’ are, of course, those over whom the rival campaigns are fighting the hardest. Here there is a glimpse of the expected pattern – that the prospect of a Conservative victory can push people towards independence – but the effect is small and not statistically significant. Not surprisingly, the large majority of this group remains in the middle ‘undecided’ reaches of the seven-point scale. (And, unless the referendum turnout is to surpass even the most optimistic expectations, we might reasonably expect plenty of these “don’t knows” not to vote anyway.)
So we turn to Q2. Maybe referendum voters care less about who will win in 2015 than how that government will respond to the demands for further devolution.A first step is to check whether the journalists’ claims in the experiment were believed. After all, many voters may well have made up their minds about the parties’ readiness to devolve more powers. The graph below suggests that this is often the case. Regardless of which information they read, clear majorities thought that further devolution but short of devo max was the likeliest option. However, the experiment clearly did influence at least some views. Among those told that the winning party would refuse to devolve powers, one in three believed as much and picked the status quo as the likeliest outcome. This proportion was closer to one in five among those told that the parties would be more forthcoming. (That said, not all were concerned that this would go as far as the devo max mentioned in the experiment.) One non-barking dog in these results is that it continues to matter little which party is expected to be in power. There was certainly no wholesale rejection of any claim that the Conservatives would devolve powers.
Since at least some people were persuaded by the message that a No vote would not result in further devolution, were they in turn persuaded to vote Yes? This time, the answer is: maybe a few of them. The next graph shows that, among those told that the parties would refuse a substantial transfer of powers, support for independence was at 36%. This compares to 32-33% among those told to expect devo max.
This is obviously quite a small difference –no wider, incidentally, among those still unsure how to vote – and it leaves the usual clear advantage for the No side. Plainly, expectations about devolution will not decide this referendum. But then no single thing will. In referendums, like elections, few things matter very much but lots of things matter a bit. This BESIP experiment suggests that the prospects for devolution following a No vote can be added to the list of factors likely to swing at least some referendum votes.Which way those votes swing will depend, of course, on the credibility – and perhaps therefore the concreteness – of the UK parties’ promises of ‘more powers’ over the next few months.
Meanwhile, there is no sign that the polling data for May 2015 will have much impact on polling day in September 2014. On one hand, this is heartening. It would seem short-sighted for voters to make a historic constitutional judgement based on which set of ‘here today, gone tomorrow’ politicians look set to win a particular election at Westminster. Less reassuringly, the fact that Scottish voters react in the same way to the prospects of Labour and Conservative victories may be another symptom of the complaint that “the parties are all the same”. Recent studies of Scottish voters highlight a weakening of Labour’s reputation – and thus their advantage over the Conservatives – as a party seen to stand up for Scottish interests. And, if the referendum campaign is ultimately a battle about how best to serve those interests (rather than simply a quest for anti-Tory government wherever it can be found), the partisan stripe of the winner at Westminster has become less relevant.