Retroduction, Congruence Analysis, and Causal Inference in Case Study Research

The value of case studies is the subject of some controversy among social scientists, particularly when they are used to test general theories. This has been a fruitful debate inasmuch as it has encouraged us to think more carefully about our methodological choices. However, I argue that the debate about case study methods is often informed by an overly narrow definition of science and - causal inferencenamely that science is reducible to testing general theory, and that causal inference is reducible to a "frequentist" framework. Instead, I show that using theory to explain an individual outcome is also a legitimate goal of science, but that achieving this goal requires different standards of inference than those used in "large-n" designs. In making this argument, I suggest we think of case studies in much the same way a medical examiner would think of an autopsy. The goal of the medical examiner is not to test a "general theory of death," but to use her general knowledge about the causes of death to explain an individual case. In order to do this for large scale events that are the subjects of our inquiries in the social sciences, I describe a technique called "congruence analysis." Congruence analysis is based on the logic of retroduction, which is not only common in autopsies, but also in natural sciences like geology. This technique requires looking for congruence between case study data and multiple theoretical implications in order to assess how much different factors contributed to an outcome we want to explain, rather than searching for "critical tests" that discriminate between rival explanations