The Structuring of Mass Politics in the Smaller European Democracies: A Developmental Typology

There is a curious awkwardness about discussions of “types” in comparative politics. Aristotle and Montesquieu taught us to proceed by classifications of regimes, and the current generation of computer enthusiasts have offered us more and more powerful tools for the handling of wide ranges of attributes of political entities and for the establishment of complex multi-dimensional typologies. Yet as soon as we are confronted with a concrete table of alternative types and look over the lists of cases assigned to each cell, our first reaction is almost immediately to add further distinctions, to reject the imposition of similarities across historically distinct units. The student of politics is torn between two sets of super-ego demands: he feels an obligation to reduce the welter of empirical facts to a body of parsimoniously organized general propositions but he also feels under pressure to treat each case sui generis , as a unique configuration worthy of an effort of understanding all on its own. This is of course a dilemma common to all social sciences but is particularly difficult to handle in the study of such highly visible, amply documented macro-units as historical polities. Students of census records, elections and survey data have an enormous analytical advantage: they deal with large numbers of anonymous units and can therefore proceed with the analysis of their data with a minimum of interference from exogenous “noise”. The student of comparative politics is roughly in the position of a social scientist asked to analyze the census records or the survey responses of a set of close friends: he cannot prevent himself from bringing into his analysis of the coded data on the punched cards a wide range of uncoded “surplus” information acquired through years of acquaintance with the subjects. The standardized sample survey derives great methodological strength from its programmatic insistence on equality, anonymity and distance in the treatment of the information collected: the data are given once and for all in the protocols or on the IBM cards and there is no allowance for fuzzy interaction with the subjects outside that framework.