What VAA-generated data can tell us about the Left-Right divide: Strong in the south, weak in the north

The EUvox Voting Advice Application was launched in all twenty-eight EU member states prior to the 2014 EP elections and included twenty-one common policy issues for all member states. It therefore provides a wealth of comparable data about VAA users from across the European continent. In particular, it allows us to examine different types of voter: older as well as younger voters, university-educated as well as less well-educated voters, and politically interested as well as politically naïve voters. This paper first of all identifies latent dimensions behind user responses in each of twenty-four EU member states for which sufficient data is available. It finds that in most, but not all, EU member states, the political space is defined by two dimensions: one economic left-right dimension and one cultural dimension that embraces a variegated mix of issues including immigration, gay rights and EU integration. Using a combination of Mokken Scale Analysis and negative binomial regression on Guttman errors, I explore how user characteristics (age, education and political interest), impact on the consistency of users' responses to items that are found to load onto the same latent dimension. The results show a stark contrast between more economically stable, mainly northern European countries, on the one hand, and mainly southern European "bailout" countries on the other. In the former cases the economic left-right dimension is only strong and consistent amongst a narrow group of political literati, while the cultural dimension is the one that animates less politically interested voters. In the latter group of countries, in contrast, the economic dimension is as strong or stronger than the cultural dimension amongst the less politically adept, and even sometimes also embraces issues of EU integration (which in northern Europe tends to be subsumed under the cultural dimension). In post-communist Europe, a mixed picture emerges, possibly due to earlier legacies that determined forms of political competition during transition. The paper ends by exploring the implications of these findings for party competition. The added value of this paper is that it shows how by comparing across different user characteristics, research into VAA-generated data can overcome the "representativeness problem" that has hitherto bedevilled it.