Over the last decades, concurrently with the increasing number of referendums held around the world since the 1970s (Qvortrup 2014), the issue of the determinants of referendum outcomes has attracted a growing number of scientific analyses.Several authors writing on referendums have argued that second-order factorsi.e. factors beyond the issue-voting -exert a significant influence on voting. Political partisanship is commonly regarded as dominant among the second-order factors.However, as we intend to show, political partisanship meets with serious constraints, from a split in the party to the voters' dissatisfaction with the latter; and further aspects, such as the turnout, usually disregarded as determinants of the referendum outcome, could, on the contrary, play a role in it.
Political partisanship as a referendum determinantPioneering studies of the early 1970s on the determinants of political choices focused on attitudes towards the European integration (Inglehart 1971), which started to be subject to referendums in 1972: these studies found political party preferences, as well as the voters' value priorities -especially post-acquisitive values -to predict support for supranational integration. Later studies added stronger evidence in favour of the relevance of partisanship. Marradi (1976) found that partisanship was the main determinant of the attitudes towards the divorce bill that was the issue-voting of the Italian 1974 referendum. Pierce et al. (1983), analysing referendums held in Britain and Norway, found proportions of "Yes" from 70% to 93% among the supporters of "Yes" parties, and from 3% to 34% among supporters of "No" parties. Franklin et al. (1995) found a significant association between the preferred party position and voting at a Scottish referendum.In 2000, Hug and Sciarini, analysing as many as 14 referendums on the EU integration held in nine West European countries, showed that the percentage of voters endorsing their party line was rather stable across the referendums, and ranged between 60% and 87%. Leduc's analysis ( 2002) of referendums held in several world's countries confirmed that pre-existing partisan attitudes are among the strongest clues as to the referendum voting choice. In turn, Hobolt ( 2006), analysing two Danish referendums, showed that party recommendations had a significant impact on voting, also when controlling for the voters' education and their satisfaction over the government's results. And party affiliation was also found to matter in the vote intention as to the 2016 Brexit referendum (Hobolt 2016;Clarke et al. 2016).However, the partisanship-referendum-voting link is more intricate than it sounds. First of all, referendums are characterised by an issue-voting: usually issues, rather than parties, are at stake, and this can make partisanship relatively less influential than in political elections, as to the vote (Laycock 2013). Others argued, however, that when issue-voting and partisanship collide in a referendum, voters tend to favour partisanship (Selb et al. ...