共時比較、通時比較、因果推論(I)

概要

論文の詳細を見る
The main topic of this paper is a much-debated problem in comparative research methodology, namely, is it possible to draw longitudinal inferences from cross-sectional comparisons? The author attempts to answer this question with particular reference to S.M.Lipset's economic development hypothesis on democratization. The hypothesis that "wealth fosters democracy", which was put forth in the late 1950s, has survived increasingly statistical tests drawing mainly on cross-sectional data. After reviewing the outlines of this hypothesis, the author explains how it is characterized by a tendency to infer longitudinal diachronic generalizations from cross-sectional, synchronic ones. It is then considered whether the "developmental equivalence" assumption underlying this inference is legitimate or not. The answer is in negative. In short, synchronic comparisons cannot substitute for diachronic ones. But even if historical informations and diachronic propositions are needed for avoiding falling into the "cross-sectional fallacy", can we assert the epistemological superiority of diachronic comparisons over synchronic ones? Considering the problem of multi-collinearity in temporal series of information, we cannot assert so. In general, for the causal interpretation of longitudinal change of a phenomenon, synchronic comparisons are as essential as diachronic ones to the process of verifying hypotheses and generalizations.
尚美学園大学の論文
2001-10-31