Identifying National Types: A Cluster Analysis of Politics, Economics, and Conflict

Abstract

This article is founded on the assumption that cluster analysis can be used to complement regressionbased techniques to obtain further improvement in systematic understanding of the nexus of politics, economics, and conflict. It assumes such variables form part of a yet to be understood, non-linear, timedependent interactive system. Cluster analysis is used to classify entities into groups and aims toward explanations based on characteristics cutting across the objects in which they are embedded; thus, the analysis seeks a more compelling account of the complex linkages between and among economic, political, and conflict-related variables. Cross-sectional data for 1967, 1974, 1981, 1988, and 1995 from the Dataset on National Attributes is used in the cluster analysis. The data analysis identifies clusters of states based on a range of characteristics. As expected within a time-dependent system, there is evidence of consistent clustering of countries within and across years, along with evidence of change. Several clusters, such as the advanced states, are very stable and indicate patterns that should be explored further with regression analysis.