Case Studies and Theory Development in the Social Sciences - Alexander L. George [65]
Sagan identifies historical situations, including several aspects of the Cuban Missile Crisis, in which the theories make different predictions about the level of safety achieved and the means through which it was attained. 236
Sagan notes that his goal was to “deduce what each theory should predict about specific efforts to prevent the ultimate safety system failure—an accidental nuclear war—and then compare these predictions to the historical experiences of U.S. nuclear weapons command and control. Which theory provides better predictions of what happened and more compelling explanations of why it happened? Which theory leads to the discovery of more novel facts and new insights? Which one is therefore a better guide to understanding?”237 Sagan concludes that on the whole, the normal accidents school provides more accurate answers to these questions in the case of the Cuban Missile Crisis.
Sagan’s reasoning is as follows: given that there have been no accidental nuclear wars, one can focus on the performance of the two theories in predicting and explaining the serious—though not catastrophic—failures in the safety of nuclear weapons that have occurred. An interesting feature is Sagan’s effort to construct a tough test for the normal accidents theory in the impressive U.S. safety record with nuclear weapons, which appears to conform more closely to the optimistic predictions of high reliability theorists. That U.S. leaders attach high priority to avoiding accidental nuclear war, U.S. nuclear forces personnel are isolated from society and subject to strict military discipline, and the United States has adequate resources to spend on the safety of its nuclear weapons also favors the validity of the high reliability theory and poses a tough test for the normal accidents theory. Sagan nonetheless concludes on the basis of detailed process-tracing evidence that the lesser safety failures and near misses that did occur are comprehensible only in terms of the warnings of the normal accidents school. By arriving at this finding even in a very tough test, Sagan creates a convincing basis for generalizing beyond his cases to U.S. nuclear weapons safety as a whole.
TESTING CONTINGENT GENERALIZATIONS
To test contingent or typological generalizations, scholars must clearly specify the scope or domain of their generalizations. To what range of institutional settings, cultural contexts, time periods, geographic settings, and situational contexts do the findings apply? Here again, typological theorizing, as discussed in Chapter 11, provides a ready means for specifying the configurations of variables or the types to which generalizations apply. Tests of contingent generalizations can then consist of examining cases within the specified domain of the theory to see if their processes and outcomes are as the theory predicts. Conversely, researchers can test for cases beyond the specified scope conditions of the theory to determine if these scope conditions might be justifiably broadened.
The proper boundaries of contingent generalizations are a frequent subject of contention among theorists. An illuminating example concerns Theda Skocpol’s study of social revolutions in France, Russia, and China.238 Barbara Geddes critiques Skocpol’s analysis by arguing that in several Latin American countries, the causes of revolution that Skocpol identified were present, but no revolutions occurred, while in other countries in the region, revolutions took place even in the absence of the preconditions Skocpol noted.239 Skocpol was careful to make her theory contingent, however, clearly indicating in her introduction and conclusion that her theory is not a general theory of revolutions, but a theory of revolutions in wealthy agrarian states that had not experienced colonial domination. Skocpol in fact explicitly states that her argument does not apply to three cases