In times where populism and discontent are on the rise, the dynamics of referendums may be different. Especially if voters have little experience with referendums they may be more likely to “vote with the boot” (Franklin et al. 1995, p. 113). This special issue therefore asks: How do referendums function in times of discontent? It will zoom in on the Dutch 2016 referendum and compares it to the Brexit referendum, both important examples of referendums in times of discontent (Qvortrup 2017; Topaloff 2017).

In this introduction, I first provide an overview of the most relevant theoretical factors used to explain referendum voting. Afterwards, I discuss one important lacuna in the field: its relatively limited geographical focus on countries that have a lot of referendum experience and highlight why this is an important factor to consider. Then I provide background into the Dutch referendum legislation and the 2016 referendum campaign. I finish by presenting an overview of the main findings of the papers in the special issue.

Theoretical background: discontent, partisan cues and referendum campaigns

Referendum research trying to explain why referendum voters vote what they vote mainly focuses on three inter-related topics: the role of political discontent, cues and referendum-specific factors.

Regarding discontent, scholars have noted that referendums are often of a second-order nature (cf. Reif and Schmitt 1980). As a result, they may function as a vote on the popularity of the government giving voters the opportunity to vote against their governments (Franklin et al. 1995). More recently, this strand of literature has been linked to the rise of populism. These scholars view the referendum as an instrument that enables those “who feel unrepresented” and it are populist parties that harvest this resentment to cast a vote against the elites (Topaloff 2017, p. 130).

Regarding cues, many scholars stress the importance of partisan cues, or party endorsements (Lupia 1994; Donovan and Bowler 1998; Colombo and Kriesi 2017). The argumentation is simple. As Van der Meer and Van der Brug state in this special issue: one way for voters to make up their mind is “to rely on trusted sources. If one is a strong supporter of party A, and the leadership of this party speaks out in favor of voting ‘yes’, one may decide to follow this advice.” During the actual referendum campaign, voters who rely on party cues simply look for the position of their preferred party.

Regarding referendum-specific factors, two elements stand out: the issue at stake (Svensson 2002; Hobolt 2009) and the referendum campaign (Farrell and Schmitt-Beck 2003; Bernard 2012). In its most generic form, it is found that voters can base their vote on attitudes related to the topic. For instance, research has found that voters in EU-referendums rely on their attitudes towards the EU (Schuck and De Vreese 2008). Furthermore, studies find that referendum voters “do consider the issue at stake” and “make use of the information provided by parties and the campaign environment” (Hobolt 2009, p. 249). Indeed, during the campaign certain arguments about the referendum topic or certain alleged/real consequences of a ‘Yes’/‘No’-vote become salient and are used by voters when deciding their vote (LeDuc 2003).

This special issue adds to the existing literature in two ways. First, while research has considered the three aforementioned factors, our knowledge of how they are inter-related and what the explanatory power of each of them is, is still in its infancy. Second, most of the available literature is based on just a few fairly extreme ‘usual suspects’ countries, which may limit the applicability of the findings.

The importance of going beyond the usual suspects

The use of referendums is increasing. What is more, this increase is not simply a matter of more Swiss referendums occurring. It is to a large extent due to a broader set of countries that occasionally hold referendums (Qvortrup 2014). Yet while the number of countries that holds referendums is expanding, the academic literature is still mainly focusing on the ‘usual suspects’. Indeed, the bulk of the research focuses on the few countries that regularly hold referendums, mainly Switzerland, the US (at the State level) and Ireland (Altman 2011, p. 61).Footnote 1

This focus on the extreme referendum countries matters when one wishes to study how and why citizens vote in referendums. Indeed, zooming in on these countries that hold regular referendums may bias the findings. As mentioned earlier, contemporary research on voting decisions in referendums generally focuses on three areas and in each of the three areas biases may be present when only looking at the extreme referendum countries.

Regarding discontent, it could be that in such countries people who feel unrepresented become fed up with the referendum process as a whole and as a result stop voting (e.g. referendum voting fatigue). Consequently, the number of dissatisfied citizens who actually vote could be lower. One could expect that the role on discontent on referendum turnout is overestimated; while its impact on the vote-choice (‘Yes’/’No’) is underestimated.

Regarding cues, it seems likely that when voters have a lot of referendum experience, they may rely more heavily on cues by (and attitudes towards) politicians (Wavreille and Reimink 2014, p. 15). Hence, the role of partisan cues may be overestimated.

Regarding referendum-specific factors, when voters have a lot of experience with referendums, they know where to look for information, know which sources to trust (and which not to), and may have developed information shortcuts to deal with information overload or shortages. Conversely, voters that have little experience with referendums do not have such shortcuts and may be more susceptible to campaign information. As a result, when analysing how citizens came to their vote decision, one can expect an overestimation of the impact of generic attitudes about the broader process at stake (e.g. European integration). At the same time one can expect an underestimation of attitudes and evaluations of the specific topic of a given referendum and an underestimation of the impact of the campaign.

In sum, systematic research on these three topics in the context of a country that does not hold referendums frequently is therefore of importance. The Dutch 2016 referendum on the EU–Ukraine association agreement offers a unique opportunity to do so. The comparative perspective is offered by looking at the 2016 Brexit referendum held in the UK, adds another country that does not hold referendums frequently.

Background: new Dutch referendum legislation allowing bottom-up referendums

Even though the Netherlands saw it first national referendum in 2005; the 2016 referendum was the first one based on new referendum legislation that came into effect on the 1st of July 2015. Under the referendum law, citizens can force the government to hold a referendum about a law or treaty that the parliament just approved.Footnote 2 To trigger a referendum, citizens have 4 weeks to collect 10,000 signatures. When these are deemed valid by the electoral management body (‘Kiesraad’), they then have 6 weeks to collect 300,000 new signatures. A referendum is then organized. When a majority of those who voted is against the law or Treaty and when a turnout threshold of 30% of the electorate is met, the law or treaty has to be reconsidered by the parliament. Either it introduces a bill to repeal the law or one to allow the law to come into effect. The result is, thus, non-binding.Footnote 3

The Dutch 2016 referendum campaign

The background to the Brexit campaign is discussed in the paper by Fischer and Renwick.

Eurosceptic organizations started to look for suitable legislation to have a referendum on as soon as the new referendum law came into effect. The Association Treaty between the EU and Ukraine was the first opportunity to test the waters. This treaty would intensify the cooperation between the EU and Ukraine and was mainly geared at trade (roughly 66% of the articles in the treaty) and economic cooperation (roughly 25% of the articles in the treaty). The remaining articles primarily dealt with democracy, financial cooperation, justice, peace and security (Hendriks et al. 2017, p. 25).

On the 10th of July 2015, the Eurosceptic organization Burgercomité-EU and the shock-blog Geenstijl joined forces and announced that they would start a campaign to collect signatures to trigger a referendum about the treaty. A second organization, Forum voor Democratie joined them and helped by ample media attention, a clever online campaign and a large group of volunteers they succeeded to collect enough signatures (Hendriks et al. 2017, pp. 21–24).

The referendum was held on 6 April 2016. The specific question that the voter needed to answer the following: Are you in favour or against the law that approves the Association Treaty between the European Union and Ukraine? During the referendum campaign several issues emerged and the ‘Yes’- and ‘No’-camp clearly had different strategies. The ‘Yes’-camp was torn: if they campaigned heavily, they risked increasing the turnout and thereby helping the ‘No’-camp get passed the turnout threshold. However, if they did not campaign the threshold might still be passed, while the percentage of No-voters would be higher. In the end they chose to wage a modest campaign focussing mainly on the economic benefits of the treaty. Additionally, they stressed that the treaty helped to bolster Ukrainian democracy and would help to fight corruption. Lastly, they decided to avoid inciting fear and wage a positive campaign (Hendriks et al. 2017, pp. 37–38).

The ‘No’-camp mostly stressed different issues. They were open about the fact that for them this referendum was not about the treaty (Heck 2016), but rather about the legitimacy of the EU. They were also helped by an early interview with the Ukrainian president, Porosjenko, who openly said that he hoped to be able to join the EU after his first or second term as president. This was not explicitly mentioned in the treaty, and many people in the ‘Yes’-camp denied that the treaty was a first step towards EU-accession. Another important argument by the No-campaign was that the treaty would ‘provoke’ Putin (Hendriks et al. 2017, p. 40). Lastly, corruption was also used as an argument by the No-camp: Ukraine was so corrupt that it would be impossible to build a healthy economic relationship with the country. This issue rose to prominence right before the referendum, when the Panama-papers revealed that President Porosjenko had been trying to evade taxes (Hendriks et al. 2017, p. 38).

In the end, the ‘No’-camp won convincingly (61.00 vs. 38.28%), while the turnout threshold was passed as well (though less convincingly so: 32.28%). In the aftermath of the referendum, the government tried to ease the worries of the ‘No’-voters by negotiating an addendum to the treaty. However, the ‘No’-camp was dissatisfied with this solution: according to them the only correct response would have been to abandon the treaty altogether. As the 2017 national election survey later revealed, voters were not convinced either. Roughly 64% of the respondents indicated that the government should have withdrawn its signature from the treaty.

Overview of the papers in this special issue

What determined the outcome of the referendum? The first paper in this special issue by Van der Brug, Van der Meer and Van der Pas considers all three of the aforementioned theoretical factors. It finds that discontent with national governments hardly mattered. More importantly though, by using a three-wave panel survey, the authors show that in the beginning of the campaign voters relied heavily on partisan cues and generic attitudes towards the EU. However, over the course of the campaign, voters picked up referendum-specific information and relied less strongly on generic EU-attitudes. Partisan cues remained important, but by the end of the campaign period, voters based their vote-choice “increasingly on referendum-specific considerations, even when these do not correspond to general EU-attitudes nor the positions taken by one’s preferred party” (Van der Brug et al. 2018). Specifically, the arguments that the treaty was good for the economy (‘Yes’) and that it might lead to Ukraine’s accession to the EU (‘No’) seemed to have played a strong role.

The second paper by Jacobs, Akkerman and Zaslove zooms in on discontent and the role of populist attitudes in the decision to vote (‘No’). The study finds that citizens with a higher degree of populist attitudes are more supportive of referendums. They are also less likely to think referendums are too expensive. Furthermore, these citizens are no more nor less likely to turn out to vote, but are clearly more likely to vote ‘No’ (Jacobs et al. 2018). No significant effect for trust in government or campaign exposure is found, but partisan cues of the populist parties did play a role, even after controlling for populist attitudes.

The next two studies focus on the impact of the referendum campaign. The paper by Goldberg and De Vreese (2018) uses a panel survey and offers a detailed unpacking of the different EU-related attitudes that may influence a referendum vote. The study finds that generic EU-attitudes mattered, but also found that referendum campaigns can impact EU-attitudes, which then in turn affect the vote decision. The authors also highlight that campaigns do matter: the more a person was informed, the less important the generic attitudes were for the final vote-choice and the more important issue-specific considerations were.

Kleinnijenhuis and Van Atteveld study the impact of exposure to the benefits and disadvantages of the treaty. They find that that exposure to the benefits and disadvantages mattered for turnout, but not for the final vote decision, where discontent matters more. The authors explain this finding by pointing to the fact that most media offered coverage of both the benefits and the disadvantages. Such competitive issue framing effectively may neutralize the effect (Kleinnijenhuis and Van Atteveld 2018). This result is in line with the findings by Van der Brug, Van der Meer and Van der Pas who also find that mere exposure did not play a role. Rather, media seem to have offered a menu of arguments from which the voter made a choice.

The final paper examined the role of discontent, party cues and referendum-specific factors in the case of the Brexit referendum. Fischer and Renwick (2018) find that issue-based considerations were “much more important than or at least mediated the effects of more general political attitudes (…) such as government approval and leader evaluations)”. Interestingly, expectations about the effects of a Brexit on the economy and immigration had most impact, regardless of whether they were based on accurate or misleading sources. Voters expected that a Brexit would cause no big economic harm, but they did expect that it would reduce immigration. Clearly, the ‘No’-camp was more successful in getting its arguments across during the campaign.

What does this tell us about referendums in countries that do not hold regular referendums and what is the role of discontent? All in all, the papers in this special issue suggest that the dynamics of referendums are different in countries where referendums are rare events. Several papers found that discontent in the form of trust in the EU or populist attitudes indeed played an important role (though the role of trust in the government seems to have been limited). Party cues also played a role, even independent from referendum-specific factors, but it was the impact of these referendum-specific factors and the campaign in particular that stood out the most. To reiterate the quote by Hobolt (2009, p. 249) mentioned earlier: a significant portion of the voters did seem to consider the issue at stake and made use of the information available. In short, in times of discontent, referendum campaigns may be more important than ever. With dissatisfied voters tending to vote against a proposal and voters motivated by party cues tending to vote in line with their party, it may be precisely this group of voters influenced by the campaign that ultimately makes the difference between a ‘Yes’ or a ‘No’ outcome.