Case Studies and Theory Development in the Social Sciences - Alexander L. George [108]
Process-tracing is fundamentally different from methods based on covariance or comparisons across cases. In using theories to develop explanations of cases through process-tracing, all the intervening steps in a case must be as predicted by a hypothesis (as emphasized later in this chapter), or else that hypothesis must be amended—perhaps trivially or perhaps fundamentally—to explain the case. It is not sufficient that a hypothesis be consistent with a statistically significant number of intervening steps.
Process-tracing complements other research methods. While process-tracing can contribute to theory development and theory testing in ways that statistical analysis cannot (or can only with great difficulty), the two methods are not competitive. The two methods provide different and complementary bases for causal inference, and we need to develop ways to employ both in well-designed research programs on important, complex problems.413
Nor is process-tracing incompatible with rational choice approaches. Process-tracing is a research method; rational choice models are theories. Many proponents of the rational choice approach agree that its efficacy must be judged in part by empirical testing of decision-making processes; process-tracing provides the opportunity to do so. In fact, scholars are using process-tracing within a general rational choice framework to construct detailed historical case studies (or analytic narratives).414 Elements of a rational choice approach have been used, together with other theories, in developing rounded, more comprehensive explanations of complex events.415 Similarly, case study methods can be used to test and refine theoretical insights built from deductive frameworks developed in game theory.416
However, even when rational choice theory or other formal models predict outcomes with a fairly high degree of accuracy, they do not constitute acceptable causal explanations unless they demonstrate (to the extent the evidence allows) that their posited or implied causal mechanisms were in fact operative in the predicted cases. Adequate causal explanations require empirically substantiated assertions about both the causal effects of independent variables and causal mechanisms or the observed processes that lead to outcomes.
Since process-tracing shares some of the basic features of historical explanation, we discuss the logic of historical explanation and indicate its similarities and differences with various types and uses of process-tracing. 417 Process-tracing takes several different forms, not all of which are seen in historical studies; and process-tracing also has quite a few uses, several of which are not usually encountered in historical studies. These differences stem from process-tracing’s emphasis on theory development and theory testing.
Process-tracing can sometimes be used for theory testing and is frequently valuable in theory development. Many theories available thus far on problems of interest in international relations, comparative politics, and U.S. politics are probabilistic statements that do not specify the causal process that leads from the independent variables associated with the theory to variance in the outcomes.418 Such theories cannot generate predictions or hypotheses about what should be observed regarding this process.419 For example, the first generation of studies on the democratic peace thesis were correlational studies that seem to indicate that democratic states do not fight each other or seldom do so. While a number of ideas were put forward as possible explanations for this phenomenon, they were not well enough specified to permit use of detailed process-tracing of individual cases to assess whether there is evidence of the