Case Studies and Theory Development in the Social Sciences - Alexander L. George [105]
A fourth step is possible and we strongly recommend that it be undertaken. Process-tracing of the case should be employed for several purposes: to help assess whether the consistency noted is spurious or causal; to identify any possible intervening causal process that connects the deductive theory with the case outcomes; and to provide an explanation for deviant cases that the theory failed to predict correctly. Process-tracing was used for these purposes by Aggarwal, Yoffie, and Posen, but not by Bueno de Mesquita.
AGGARWAL’S STUDY OF TRADE REGIMES
Aggarwal’s study was one of a number of studies in political economy undertaken by Ph.D. students at Stanford under the direction of Robert Keohane (with Alexander George serving as a second reader). The starting point for all these studies was the assumption that the best way to study problems of trade relations between the United States and its weaker trading partners (and also to study the development and possible transformation of international trade regimes) was to adapt structural realist theory for the specific issue-area and actors involved. (This assumption was substantially modified as students encountered the problem of developing causal inferences and explanations for outcomes of trading episodes.) The initial research design focused on the relative power advantage the United States possessed which, according to structural realist theory, should lead to outcomes favorable to the United States. When such favorable outcomes in trading episodes occurred, it might be assumed that realist theory provided an adequate explanation and could have predicted these outcomes.
However, Aggarwal realized that mere congruence of outcomes with the general predictions and expectations of structural-realist theory did not necessarily provide a reliable explanation—that it was not an adequate test of the theory. Hence, Aggarwal engaged in process-tracing of each trading episode to ascertain whether he could identify a causal process that supported the role attributed to the structural variable. He felt it necessary to proceed in this fashion since it was not possible to undertake a large-N statistical study for this purpose. In addition, to understand and explain a number of those deviant cases in which the outcomes were not as favorable to the United States as its relative power advantage would have predicted, Aggarwal undertook a detailed analysis of the dynamics of the trading interaction and engaged in process-tracing to identify how the actors’ decision-making and their strategic interaction in bargaining with each other might have led to an outcome not predicted by the theory.
Hence, Aggarwal was not satisfied to settle for the familiar fall-back position that structural realism is a probabilistic theory that does not claim to predict all cases successfully. Instead, Aggarwal attempted to explain discrepant cases and, if possible, to enrich and differentiate the theory. He referred to these cases as anomalies and argued that in the absence of a large number of cases to permit statistical analysis, “a second approach, known as ‘process-tracing,’ is an effective and potentially superior substitute. In process-tracing, the decision-making procedure in a negotiation is systematically analyzed with an eye to identifying the degree to which participants appear to respond to international systemic or other constraints.”404
As the third and fourth steps emphasize, one should not be satisfied merely with a finding of consistency. Since the data required for adequate process-tracing are often not available, the checks regarding the causal significance of consistency noted earlier should be undertaken.
Congruence and Structural-Realist Theory
Studies that use structural-realist theory to predict outcomes are in special need of supplementary process-tracing or other checks. Kenneth Waltz’s structural-realist theory is