ttitudinal Polarization Measurement Through (Ordered) Latent Class Analysis

This paper presents a new approach to the measurement of attitudinal polarization for cross-national or repeated cross-sectional studies. The proposed approach is a two-stage one. At the first step, order-constrained Latent Class Analysis (LCA) is used to identify a categorical latent construct underlying a set of observed items. Basing on the best LCA solution, class membership is assigned for each individual in the sample. At the second step, a broad family of categorical polarization indices may be computed for that categorical latent scale in respect to any grouping variable of interest (e.g., country of living, or wave of study). The data from the 4th wave of the European Values Study are used, and polarization between survival and self-expression values in 28 European countries is measured. The resulting polarization scores are used to test a hypothesis assuming positive aggregate-level association between values polarization and support for radical right parties and ideologies.