Thickening Thin Concepts and Theories: Combining Large N and Small in Comparative Politics

Abstract

In recent controversies about comparative politics, it often seems as though there are only two approaches: the systematic logical deduction of universalistic theory and the more traditional case studies and small-N comparisons leading more inductively to middle range theory.1 The purpose of this article is to situate large-N quantitative analysis in this controversy. Quantitative analysis has its weaknesses, but they could be counterbalanced by some real strengths in small-N analysis. And quantitative analysis has certain methodological advantages that help compensate for some of the weaknesses of small-N analysis. On the one hand, small-N analysis tends to develop (complex or multidimensional) concepts and theories that are well-suited for description and for making inferences about simple causation on a small scale or in a few cases, but thick concepts and theories are unwieldy in generalizing or rigorously testing complex hypotheses. On the other hand, quantitative analysis is justifiably criticized for its (reductionist or simplistic) concepts and theories, but it is the best method available for testing generalizations, especially generalizations about complex causal relationships. Quantitative analysis has hardly begun to exploit its full potential in assimilating complex concepts and testing complex theories, largely due to data limitations. In order to realize its potential, scholars need to answer two key questions that arise at the intersection of small-N and quantitative analysis. Can thick concepts be translated into the thin format of quantitative data? And can the nuanced, conditional, complex, and contextualized hypotheses of small-N analysis be translated into quantitative models? I argue that the answer to both questions is yes in principle, but that in order to make these approaches complementary in practice we must collect different data and more data and do it more systematically and rigorously.