Cross-National Statistical Research and the Study of Comparative Politics

It is often suggested that cross-national statistical research is excessively narrow and oversimplified, that it is plagued by unique measurement problems, and that it can have few causal implications. In light of these arguments, this paper evaluates the role of such research in the comparative study of politics. The paper begins with a comparison of the experimental, comparative, and statistical modes of empirical analysis, and then considers, in turn, issues in the development of generalizations, in the identification of variables, and in the interpretation of relations between variables. The last section of the paper addresses problems associated with making causal inferences in cross-national research. This involves a comparison of cross-sectional and longitudinal designs, a discussion of measurement validity, and a consideration of the fact that empirical research seldom offers a full, direct test of a theory. It is concluded that the comparative and statistical methods share a common logic of analysis, and that the statistical method can be a powerful tool in the comparative study of politics.