Comparing Party Systems: A Multidimensional Approach

The comparative study of political parties once consisted largely of attempts to differentiate among one-, two-, and multi-party systems. Eventually, however, the extreme oversimplification of this classification scheme came to be recognized, and more elaborate schemes which contained categories like "modified one-party system" and "two-and-a-half party system" began to appear. Unfortunately, these new classifications proved to be extremely arbitrary in the criteria that were used to classify party systems and highly capricious in the way that criteria were applied.' Dissatisfied with the classificatory approach, comparativists then began to develop numerical indices which they hoped would facilitate systematic analysis. Over the last fifteen years, literally dozens of these indices have made their way into the research literature.2 The purpose of each such index is to assign a single numerical value to a party system for use in comparing intersystem differences at a point in time, intrasystem differences over time, or a combination of both. In sharp contrast to both the traditional classificatory approach and the more recent index-construction approach, we begin with the assumption that party systems are inherently multidimensional. While the indices which have flourished in recent years are useful for certain purposes, they--like the earlier classification schemes-greatly understate the complexity and diversity of the party systems that one finds worldwide. The main virtue of existing indices is that they focus attention on a single aspect of the operation of party systems and treat that aspect in an "objective," replicable fashion. However, this is also their primary defect, for in focusing attention on only one aspect of party systems they forfeit the ability to capture other important aspects as well. In our view, a multidimensional approach, in which several different aspects of party systems simultaneously serve as the bases of comparison, is needed to sensitize comparativists to the considerable differences that sometimes characterize systems which seem very similar when viewed unidimensionally, as well as to the unexpected similarities that sometimes crop up between systems which are generally thought to be very different. This paper begins with a discussion of four attributes that we consider central to the analysis of party systems. While these attributes have been utilized in previous research on party systems, each is normally analyzed in isolation from the others. We, on the other hand, discuss a procedure that facilitates comparison based upon the simultaneous consideration of all four attributes. Utilizing this procedure, we employ data from forty-six nations to develop a multidimensional representation of party systems. We then discuss the potential theoretical importance of the information gained from this multidimensional representation.