Detecting Causal Chains in Small- n Data

Abstract

The first part of this article shows that qualitative comparative analysis (QCA)—also in its most recent form as in Ragin (2008)—does not correctly analyze data generated by causal chains. The incorrect modeling of data originating from chains essentially stems from QCA’s reliance on Quine-McCluskey optimization to eliminate redundancies from sufficient and necessary conditions. Baumgartner (2009a, 2009b) has introduced a Boolean methodology, termed coincidence analysis (CNA), which is related to QCA, yet, contrary to the latter, does not eliminate redundancies by means of Quine-McCluskey optimization. The second part of the article applies CNA to chain-generated data. It turns out that CNA successfully detects causal chains in small- $n$ data.