Political partisanship versus turnout in Italy’s 2016 referendum

Italy’s 2016 referendum represents an opportunity to discuss the main hypotheses posited by current literature as to the determinants of referendums. The issue-voting of this constitutional referendum was too technical to arouse great passions in the voters. This provided room for second-order determinants, and the referendum campaign became a competition between Government and Opposition. In such a scenario, political partisanship is expected to play a significant role in the voter’s choice. This paper intends to test the impact of partisanship on the vote, versus that of further possible determinants: intra-party dissent, the voters’ dissatisfaction with the government, as well as the referendum turnout. Particular attention has been paid to the latter, also because the impact of turnout on referendums is understudied. Correlations and linear regression models, based on data of all the Italian provinces, revealed that, apart from partisanship, other second-order features impacted on the referendum outcome: in particular, it emerged that, where the turnout had been higher, the “No” share was lower. However, the turnout was intertwined with other determinants, first of all, partisanship. To disentangle the turnout from its correlates, a lagged instrument, namely the turnout of the previous referendum held in 2006, was introduced. The Two-Stage Least Squares model proved the instrument to be a robust predictor of the 2016 referendum vote. The result of a mediation model further supported these findings, which suggest that the turnout—as an expression of civic commitment—is not independent of the voters’ decision.