Sara Binzer Hobolt, "How Parties Affect Vote Choice in European Integration Referendum," Party Politics, 12 (September, 2006), 623-647.

First Paragraph:
In the referendums on the European Constitution held in 2005, a sizeable majority of the French and Dutch electorates voted No, despite a broad consensus among mainstream parties in favour of the Constitutional Treaty. This is not the first time that the public has disregarded the consensus among national parties and rejected an EU treaty. The Danish rejection of the Maastricht Treaty in 1992 and the Irish No to the Nice Treaty in 2001 dealt a serious blow to the European political establishment, even though both decisions were overturned in subsequent referendums. This article examines how - and to what extent - parties can influence vote choices in referendums. To evaluate the influence of political parties in direct democracy, it is important to understand the decision-making processes of voters. While there is an extensive literature on voting behaviour in elections, there are very few general theories of how voters behave in direct democracy. Referendums present voters with a different choice than elections. No political parties or candidate names appear on the ballot, and voters must choose among alternatives that are often unfamiliar. However, this does not imply that the information and endorsements provided by political parties are irrelevant to the outcome. If voters know little about the specific ballot proposal, it is mainly the information made available to them by parties and other information providers over the course of a campaign that provides the basis for their opinion on the ballot question. Campaigns thus play a very important role in referendums, and examining how different party strategies can make a difference to the vote choice is therefore crucial if we are to understand the outcomes (De Vreese and Semetko, 2004; LeDuc, 2002). To examine how parties can influence the vote choice in referendums, the article develops a comprehensive, spatial model of how voters choose in referendums. On the basis of this theoretical model, it derives several hypotheses which are evaluated empirically in a 'controlled comparison' of the two Danish referendums on the Maastricht Treaty. This analysis thus focuses specifically on referendums on European integration, yet it could be extended to understand how political parties can influence referendum outcomes more generally.

Figures and Tables:
Figure 1. Spatial representation of voter choice in EU referendums
Table 1. Attitudes towards European integration (percentage in favour)
Table 2. Predicting the Yes-vote in the Danish Maastricht referendums
Table 3. Simulated effect on Yes-vote given changes in attitudes and partisanship
Figure 2. The impact of attitudes and partisanship on the Yes-vote.
Table 4. Reasons for voting Yes and No in the 1992 Danish referendum
Table 5. Who switched?
Table 6. Yes-vote by party preference in the two Maastricht referendums

First Paragraph of Conclusion:
Occasionally, voters answer ballot questions in referendums in ways that both shock and dismay political parties. However, this does not imply that parties are powerless in the referendum process. As this article has suggested, political parties have considerable power to influence the way in which the referendum choice is perceived by voters. The statistical analyses of voting behaviour in the two Danish Maastricht referendums have shown that the party endorsements matter, even when we control for issue-specific preferences. However, these analyses have also demonstrated that people's EU preferences are a stronger predictor of vote behaviour and that many voters 'defect' from the party line. In other words, partisan loyalties may not be sufficient to persuade voters to vote in a certain way. This evidence indicates that the influence of parties may be primarily indirect. As pivotal information providers, parties can frame the meaning of the choice that voters face in referendums. Building on a spatial model of voting behaviour, this article has suggested that parties can convince voters to vote in favour of (or against) a ballot proposal by framing the proposal as close to the ideal point of the median voter and the reversion point as more extreme. The two Danish Maastricht referendums examined in this article illustrate the importance of issue framing. The Danish case was presented as a 'critical case' because Danish voters have comparatively well-informed and stable opinions about European integration, and hence we would expect voters to be resistant to partisan cues (Franklin, 2002). The analyses suggested that the effect of party cues were more significant in the second Maastricht campaign compared with the first. A plausible interpretation of the change in outcome between the two campaigns thus suggests that the Yes-parties managed to provide a different 'framing' of the choice (aided by the Edinburgh Agreement) in the second referendum, which was closer to the 'bliss point' of a critical proportion of voters. Although further comparative work is needed to substantiate these propositions, the findings give insights into the role of parties in referendums. They have shown that campaigns matter in referendums and that parties not only have the power to ask the question that voters have to answer, but also to guide their understanding of that question during the campaign. Moreover, the evidence suggests that voters take cues from the parties they support when asked to vote on a policy issue, but that issue preferences condition the extent to which parties can persuade their own supporters to follow the party line. Hence, to understand the effect of parties on vote choice in referendums, we not only need information on the party recommendations, but also on the way in which cues are disseminated by the media, and importantly how receptive voters are to the elite cues and the media framing. Parties can influence outcomes in direct democracy, but, as in elections, the final say rests with the electorate.