Society, Politics, and Economic Development Revisited: A Disjoint Principal-Components Analysis

Abstract

This chapter presents a comparison of the economic, social, and political traits—with two different statistical techniques—of 74 developing nations during 1955–1963 to explore whether the implications concerning the modernization processes involved in economic development draw on the basis of previous analyses are confirmed or not. In an analysis presented in the chapter, a new method of analysis, that of disjoint principal-components models, was applied to the Adelman–Morris data. The method is that of pattern recognition developed for applications in the chemical and biological sciences by S. Wold. It simultaneously classifies observations into groups and derives models of variable interactions within each of the groups. The criterion used for classification is both the homogeneity of model fit within groups and the degree of difference between models among groups. The method is, therefore, theoretically appealing for those interested in classification for subsequent theory construction. The present analysis is both more economical of effort and likely to yield better results, since it performs classification and model fitting simultaneously, thereby in essence optimizing the classification with respect to model fit. The results of the two analyses compare quite closely, both with respect to country classifications and with respect to their characterizations of development processes.