Examining Voter-Flight from Mainstream Parties using VAA Data from Different National Elections Contexts

Electoral results from numerous national contexts throughout the period following the global financial crisis of 2008 indicate heightened electoral volatility in continuation and, perhaps, intensification of a trend already in place by the late 1980s. Whether a function of some increased capacity of voters to engage in politics independently of their social groupings, a dealignment between the later and political parties or the emergence of new and more salient cleavages, such as the "cultural" dimension, with few exceptions incumbents and mainstream parties traditionally contesting power have been ceding ground to "challenger parties" and new political formations. The spread and popularity of Voting Advice Applications across Europe affords a rare opportunity to examine voter-flight from mainstream political parties, as although not without their unique problems the robustness of the datasets produced by VAAs enables in-depth analysis of "small areas", such as the electoral niche of individuals who switch away from mainstream parties between subsequent elections. This study proposes to examine this phenomenon of voters changing their party preference between the previous and upcoming election from a mainstream party to a challenger one across five different national election contexts: Poland, 2015; Greece, 2015 (January election); Spain, 2015; Germany, 2017 and UK, 2017. As a first step, using self-reported vote in the previous election and vote intention, we employ standard logistic regression models to compare voters who elected to switch their vote to those who did not, as to individual level characteristics, e.g. demographics, political orientation. Subsequently, we use conditional logit models (McFadden discrete choice paradigm) to examine this behaviour as a function what the different parties contesting the election offered the electorate in terms of policy preferences grouped into themes (as to market liberalisation, immigration etc.). As a robustness check, wherever possible, we provide equivalent analyses using data from national election surveys (e.g. the British Election Study) for individual-level data and party positions estimates from the Chapel Hill Expert Survey.