On the referendum day these were last minute decisions that most probably turned the tide in favour of the Leave camp. In this article Thomas Milic and Uwe Serdült draw up an analogy of the June 23rd vote between the referendum votes on the EU that took place in Switzerland over the past couple of years. They argue that would the Brits be Swiss, they would have voted for an exit as well.
Since 1970 there have been more than 50 referendum votes on Europe. However, a vote on leaving the EU has been very rare. Thus, it was not surprising to see that public opinion was ‘measured’ meticulously in the run up to the vote on 23 June 2016. Starting in September 2015 one opinion poll after the other was published – pollsters ruled supreme in the land of Her Majesty. In contrast to Switzerland, which usually only has two, there were so many pre-election polls that several news outlets came up with aggregate polls, trying to offer a more precise grasp on the voters’ intentions. In the last two weeks heading into the vote they all displayed the same pattern: the opinion was shifting to “Leave”, notwithstanding a short dip after the tragic death of MP Jo Cox. Such public opinion dynamic in the late stage of a campaign came as a surprise – not only to many British pundits, but also to the observers of the UK political life abroad. Their surprise was so great, beacuse it had been generally assumed that in the end the undecided voters would lean towards the status quo – true to the motto: “When in doubt, vote no”.
But how much of a surprise was it really? Not a tremendous one if one takes a closer look at the Swiss referendum campaigns on the issue of Europe. The Swiss are rather familiar with such highly controversial votes and they are also familiar with the predicament of undecided voters in tight referendum races. After all, the Swiss might have never won a football cup, but when it comes to national referenda they are the unchallenged world champions. We can therefore make use of the Swiss experience to draw insights on what may have happened during the last few days before the Brexit vote.
For such a comparison we took a closer look at all Swiss plebiscites on Europe over the last 25 years. However, not all of them are equally well suited for our purposes. The first vote on the bilateral treaties between Switzerland and Europe in the year 2000, for example, did not have an exit but only an entry character. On this occasion, the Swiss citizens were not deciding on a “Swexit”, but rather voted on whether to agree to bilateral treaties with the European Union or not. The two referendums on the extension of the free movement of people (one in 2005 and the other one in 2009), the cooperation with Central-Eastern Europe (2006) and the Schengen/Dublin treaty (2005) indeed had an exit character. A “No” would have put an end to the bilateral arrangements or at least seriously called them into question. However, in these votes the purely “formal status quo logic” contradicted the “substantial status quo logic”. For the typical risk averse voter this type of configuration was certainly more challenging than the Brexit referendum.
There, the logic was rather simple. Whoever was unhappy with the current state of European politics voted for the exit (Leave), whoever wanted to leave everything as it is voted Remain. A comparable situation for a Swiss referendum vote can be found for the vote on the popular initiative against mass immigration (February 9, 2014). This vote had an exit character and the voters were well aware of that, as studies have shown. Similar to the Brexit case, this was also a vote for which both camps conducted a robust and fierce campaign. Note that the initiative “against mass immigration” was accepted by 50.3% yes votes of citizens (turnout was 56.7%) and 14 ½ cantons in favour of it (out of 26). If one were to follow the same electoral logic for the UK, the Brexit vote actually would not have passed as only two (namely England and Wales) of the four British constituent units voted to leave. In Switzerland the passing of constitutional changes and major treaties not only requires a simple popular majority, but also a majority of the cantons.
How do the undecided, ambivalent and late deciders vote on European matters in Switzerland? First some theory: from election research we can deduct two contrary scenarios. One of the schools of thought postulates that vote intention increasingly moves towards the status quo the closer the polling day. The justification for such a behaviour in general is in the first place risk aversion. Risk aversion affects voting behaviour the stronger the more uncertain and uninformed the individual is. Late deciders, in turn, are oftentimes ambivalent. It is due to this ambivalence that they leave their decision until the very last minute. If this assumption holds true, we should expect opinion dynamics for European referendums to trend towards a status quo vote.
On the other hand, we know that the willingness to proclaim one’s point of view in an opinion poll depends on the general climate of opinions. When you consider yourself to be in the minority, you are more reluctant to voice your opinion (it is less socially desirable to do so). In general, one does not want to deviate from the masses. However, once you don’t feel stigmatised, because of your opinion, and you realise that there are many likeminded people around you, you express your own opinion freely and may – reinforced by opinion polls – influence others as well. The dynamic of a spiral of silence, which initially suppressed the actual opinion of a majority, can thus break down during the campaign. A prerequisite for this to happen is an intense campaign, especially from the camp taking the presumed minority position. If this assumption holds true we should expect a strengthening of a minority position in an ongoing campaign.
What can we observe in general regarding opinion formation during European referendums in Switzerland and regarding our most similar case for a comparison, namely the “mass immigration initiative”? For this purpose, we analysed the traditional post referendum VOX polls. In order to analyse opinion dynamics during the referendum campaign we used the (retrospective) question about the time point. For this survey question the respondents let us know at which point in time they made up their minds. This should not to be confounded with the formal casting of the vote, for example the time when the ballot was sent by post or dropped into a ballot box, but rather about the mental fixation on a certain decision.
We have to keep in mind that we have no knowledge whether the respondent changed his or her mind during the referendum campaign, or how many times. We only know their answers about the point in time when respondents took their final verdict. From this information we mapped the course of vote intention for each respective referendum vote of the voters who were ready to decide. In parallel, we reconstructed the difficulty to decide according to the point in time the decision was taken. This proportion corresponds to the bars in Graphs 2 and 3.
In Graph 1 we can note that the proportion of late deciding voters for European referendums in Switzerland is about as high as in the Brexit polls: one week before the vote it is on average ten percent. In comparison to the other referendum topics in Switzerland this proportion is low. The reason is that referendums on the European question are a recurring issue in Switzerland. It is hence well known to citizens and many already have an opinion on the issue.
Graph 1: Proportion of late deciders across Swiss EU referendums
The percentage of respondents in the various surveys with a clear opinion from the outset fluctuates more. Especially noteworthy is the fact that it was relatively low for the so called «referendum vote of the century» on the European Economic Area Agreement (1992), then rose to about 60% and has lately been declining again. This figure documents the fact that the bilateral treaties with the European Union were rather undisputed ten years ago, but that they are increasingly coming into doubt. Furthermore, these values hint at an increased potential of referendum campaigns to convince people. This is due to the proportion of the voters who stated that they were sure about their opinions (and thus rather immune to any campaigning) has consistently dropped in the recent years.
Last but not least, the development of vote intention for European referendums in Switzerland is often very much in sync with opinion formation during the Brexit campaign: it was not the advocates of the EU status quo who were gaining ground towards the end but the sceptics and adversaries of the EU. The parallel with the Swiss experience is particularly strong with our prime example vote, the initiative to stop “mass immigration” (see Graph 2). During the campaign these were the proponents of the initiative to have immigration quotas who gained support, not their adversaries (meaning those in favour of bilateral arrangements). The referendums on the Schengen/Dublin treaty and the extension of free moment in 2009 show a similar pattern. For the cooperation with Central-Eastern Europe, the extension of free movement in 2005 and the bilateral treaties we can hardly recognise the pattern (see Graph 3). However, we know that compared to their counter parts the anti-European party – the Swiss People’s Party (SVP) – only invested very limited funds into these particular campaigns. But as we mentioned earlier to break through the spiral of silence effect one needs an intense referendum campaign. Moreover, the displayed graphs are a convincing record of the fact that late deciding voters are having a much harder time making up their mind than the ones deciding early on.
Graph 2: Proposed pattern for late deciders holds
Reading example for the initiative “against mass immigration”: Around 40 percent of those who were certain of their vote intentions from the outset (“from the outset”) , voted in favour of the MEI. That proportion, however, gradually grew the closer the voting day came. Besides, less than 20 percent of the early deciders (“from the outset”) reported to have had difficulties making up their minds, while roughly half of the late deciders (“1-6 days”) had a hard time to come to a voting decision.
Graph 3: Proposed pattern for late deciders does not hold
In general, we can thus state that for Swiss referendums on questions relating to EU policy the theory of the spiral of silence, or more precisely the fact that it gets broken up by noisy and intense campaigning, fits better with empirical data than the status quo model. Moreover, it is also noteworthy that voters’ intentions did not change dramatically for all the Swiss votes we checked. As we have seen, most voters have made up their minds rather early. Swiss voters are highly familiar with foreign policy matters and thus have persistent, highly stable opinions about it. A spectacular change of minds is therefore highly unlikely in the Swiss case. That might be different in the UK, simply because they have much less experience with referendum votes on Europe.
In short, based on what we know about the opinion making process in the run-up to the European referendums in Switzerland, we can say: if it were the Swiss voting on the Brexit we would expect a Leave rather than a Remain vote. We actually boldly predicted that outcome in a Swiss news outlet one day before the vote. And it is what happened.
This post represents the views of the author and not those of the BrexitVote blog, nor the LSE. An earlier version of this article appeared in German on watson. The authors would like to thank Michele McArdle and Fernando Mendez for polishing the English version of the article. Image credit.
Thomas Milic and Uwe Serdült are working as scientific collaborators in the Centre for Democracy Studies Aarau (ZDA) at the University of Zurich. The ZDA is a research center jointly managed by the University of Zurich and the University of Applied Sciences and Arts Northwestern Switzerland (FHNW).
 Mendez, Fernando; Mendez, Mario and Vasiliki Triga (2014) Referendums and the European Union: A Comparative Inquiry. Cambridge: Cambridge University Press. Link to respective entries in the c2d database on referendum results.
 Serdült, Uwe (2014) Referendums in Switzerland, in: Qvortrup, Matt (Ed.) Referendums Around the World: The Continued Growth of Direct Democracy. Basingstoke, Palgrave Macmillan, 65-121. Kriesi, Hanspeter and Trechsel, Alexander (2008) The Politics of Switzerland. Cambridge, Cambridge Universtiy Press.
 For those not familiar with the Swiss direct democracy, the following information might be helpful to understand the line of argument: In the case of an optional or mandatory referendum the no-vote blocks a reform decision adopted earlier by the parliament. Thus, voting «No» always means to maintain the legal status quo. However, in the case of those referendums mentioned above, a no-vote would have risked the bilateral status quo of Switzerland.
 Milic, Thomas (2015) For They Knew What They Did: What Swiss Voters Did (Not) Know About The Mass Immigration Initiative, Swiss Political Science Review, 21 (1): 48-62.
 Until June 2016 conducted by a university consortium of the political science departments in Bern, Geneva and Zurich. After a call for tender those referendum poll reports will from now on be taken care of by Thomas Milic from the ZDA and Georg Lutz at FORS.
 See: Herrmann, Michael (2012) Das politische Profil des Geldes: Wahl- und Abstimmungswerbung in der Schweiz. Zürich: Forschungsstelle sotomo. Milic, Thomas; Rousselot, Biance and Adrian Vatter (2014) Handbuch der Abstimmungsforschung. Zürich: Verlag Neue Zürcher Zeitung, p. 312.
 Marquis, Lionel and Pascal Sciarini (1999) Opinion Formation in Foreign Policy: The Swiss Experience, Electoral Studies, 18 (4), 453-471.