Case Studies and Theory Development in the Social Sciences - Alexander L. George [67]
One example of a theory that failed an easy test case comes from Arend Lijphart’s study of the Netherlands, which cast doubt on David Truman’s theory of “cross-cutting cleavages.”244 Truman had argued that mutually reinforcing social cleavages, such as coterminous class and religious cleavages, would lead to contentious politics, while cross-cutting cleavages would lead to cooperative social relations. In the Netherlands, however, Lijphart found a case with essentially no cross-cutting cleavages but a stable and cooperative democratic political culture. This cast doubt on Truman’s theory not just for the Netherlands, but more generally.
Cases usually fall somewhere in between being most and least likely for particular theories, and so pose tests of an intermediate degree of difficulty. Short of finding toughest or easiest test cases, researchers should be careful to specify, for each alternative hypothesis, where the case at hand lies on the spectrum from most to least likely for that theory, and when the theory predicts outcomes that complement or contradict other theories’ predictions.
For example, Graham Allison’s study of the Cuban Missile Crisis, Essence of Decision, is in some respects a strong test case for the rational actor model, a moderate test of the organizational process model, and a strong test of the bureaucratic politics model.245 However, it is not the strongest possible test of any model and just how strong it is depends on which of Allison’s research questions is under consideration.
Let us consider the first two of Allison’s three research questions as examples. On the question of “Why did the Soviet Union place missiles in Cuba?” rational actor considerations should have been strong given the clear strategic stakes. Organizational processes should not have been very strong because the Soviet Union was taking the initiative and had time to adapt its procedures. Bureaucratic politics should have been of moderate importance given the stakes involved for Soviet military budgets and missions. On the question of “Why did Kennedy react as he did?” rational actor considerations were constrained by the incomplete information and short time period, but strengthened by the president’s direct involvement. On the other hand, the nature of the crisis favored U.S. decision-making that approximates the rational actor model. Organizational processes were a moderate constraint—the president’s personal involvement could and did modify procedures, but the short time available limited possible adaptations. Bureaucratic politics should have been constrained by the president’s role and the overriding importance of national concerns (rather than parochial institutional concerns). One could add details on what makes each question a most- or least- likely case for each of the models, but the general point is that many contextual factors must be taken into account and that they rarely all point in the same direction on the high likelihood of one theory and the low likelihood of others.
It is important to note that a case in which one variable is at an extreme value is not necessarily a definitive test. Rather, if the variables of competing explanations make the same prediction and are not at extreme values, this may represent an easy test that provides only weak evidence for the importance of the extreme variable. Such easy tests are not very probative, and if they are incorrectly used to infer strong support for a theory, they may constitute a problem of selection bias. Such a case may be more useful for the heuristic purpose of identifying the outsized causal mechanisms related to the extreme variable.
Conclusion
Generalizing the results of case studies is not a simple function of the number or diversity of cases studied. A researcher may study diverse cases that prove to have no common patterns, so that only unique historical explanations