It is all About Dissimilarity: Party System Characteristics and Their Proper Measurement

In this article we propose a new conceptualization of the crucial party system characteristics: disproportionality, electoral volatility, territorial heterogeneity and inter-election incongruence. We argue that these characteristics can be studied as dissimilarities between vectors of votes or seats. We present different specifications of vectors in order to address various research questions important for students of parties and party systems. Subsequently, developing the analyses of Monroe (1994) and Taagepera and Grofman (2003), we present nine measures of vectors’ dissimilarity: index of dissimilarity, Gallagher’s least squares measure and its transformations, cosine measure, Gini coefficient, Kullback-Leibler divergence (relative entropy), weighted variance and weighted standard deviation of ratios. We discuss their utility in empirical studies of main party system characteristics, using several dimensions of comparison, based on the formal postulates. We also add two new postulates concerning measure’s decomposability: horizontal (sumtype) and vertical (variance-type)