Spatial Voting Models under Imperfect Measurement of Ideological Preferences: Evidence from Scandinavia
A report by Kirill Zhirkov at the regular LCSR seminar
On February, 6th Kirill Zhirkov (LCSR, research fellow) delivered a report on “Spatial Voting Models under Imperfect Measurement of Ideological Preferences: Evidence from Scandinavia” at the regular LCSR seminar.
Kirill told about his new project on electoral behavior in Scandinavian countries. He pointed out that there is gap in research of dimensionality in European politics. Earlier, economic factors were considered as the major determinants of the outcomes of electoral competition. However, many non-economic issues became politically significant in Europe during the last decades, such as migration, European integration, ecology etc. Thus, the study of voting in multiparty systems can be of a great interest for political sociologists, because it allows for testing a bunch of hypotheses considering so-called niche parties which exploit one social issue to achieve local electoral success.
The main goal of the current project is to test a model of spatial voting introduced by Antony Downs, an American economist, in his famous work “An Economic Theory of Democracy”. The key concept of Downs's theory is the assumption that voters support parties with policy positions that are the closest to the voters’ own views. Kirill extended the standard Downsian framework by assuming that voters assess party positions on several dimensions when deciding for which party to vote. In particular, he uses not only classic right-left scale but also position towards migration.
Kirill uses mixed logistic regression to proceed data analysis. This is an econometric technique developed for the analysis of multiple discrete choices. This method is an advanced form of multinomial logistic regression which has several important advantages. Firstly, it allows regression coefficients to vary across respondents. Secondly, it weakens the IIA assumption, crucial for standard logit regression. Consequently, researcher is able to obtain more reliable estimations of relative utilities of various alternatives between which voters choose.
Kirill uses the data of European Social Survey project on three Scandinavian countries (Denmark, Norway and Sweden). Such choice was determined by similarity of political systems in these countries and also by their political stability. It allowed combining the data from several ESS waves (2002-2010) in order to increase sample size (which was necessary because discrete choice models require large samples to obtain accurate estimations). The results of the last wave of Chapel Hill expert survey (2010) were used to estimate the positions of parties in these countries within a 2-dimensional space.
Party preferences of the respondents were used as the dependent variable. Preliminary analysis showed that the distance between personal attitudes and party positions on economic issues is a more important factor for predicting election results. Migration issue, as expected, is more important for radical right parties. However, the most interesting result according to Kirill’s own opinion, is the positive relation between probability of voting for Conservative Party (H) in Norway and Conservative People’s Party (DKF) in Denmark and the distance between the voter and the party. This result contradicts the model of spatial voting. Kirill noted that there is no a clear theoretical explanation for this counterintuitive finding. Nevertheless, he suggested several post hoc explanations.
First of all, parties to which this paradox is observed usually represent radical views on the issue of migration; less radical voters can also support such parties due to closeness of positions of these parties on other dimensions to their own preferences. Secondly, such result can be an evidence of measurement error related to difference of criteria that experts and respondents use to evaluate party positions. Decision on what explanation fits the best requires additional empirical evidence. In further research Kirill plans to focus on analysis of National Electoral Studies, as well as intends to verify expert estimations of party positions using the data from Comparative Manifesto Project.
by Olesya Volchenko