Case Studies and Theory Development in the Social Sciences - Alexander L. George [29]
A second example is Randall Schweller’s study of how democracies behave with regard to preventive war. Like Peterson, Schweller incorporates both systemic and domestic variables, looking at how domestic structures affect state decisions on preventive wars during ongoing power shifts. Schweller concludes that only nondemocratic states wage preventive wars against rising opponents, and that democracies seek accommodation with rising democracies and form counterbalancing alliances against rising nondemocratic challengers.116
Both these studies define useful subtypes of democracies, but not every subtype is useful or progressive. Researchers might allow their subjective biases to intrude, leading them to define away anomalies through the creation of subtypes. As Miriam Elman argues, for example, “defining democracy as a regime in an independent state that ensures full civil and economic liberties; voting rights for virtually all the adult population; and peaceful transfers of power between competing political groups makes it fairly easy to exclude numerous cases of warring democracies.” 117 The creation of a new subtype is warranted if it helps explain not only the aspects of a case that led to the creation of this subtype, but also other unexplained dimensions of the case or of other cases. The assertion that “new” or “transitional” democracies are more war-prone and should be treated differently from other cases that might fit the democratic peace, for example, may warrant the creation of a new subtype. It posits testable correlations and causal mechanisms and suggests dynamics that should make states in transitions from as well as into democracy more war-prone.118 More questionable is the exclusion from assertions on the “democratic peace” of civil wars, like the U.S. Civil War.119 Also debatable is the exclusion from some data sets of conflicts that fall somewhat below the arbitrary figure of 1,000 battle deaths, such as the conflict between Finland and Britain during World War II.120
While case study methods are particularly amenable to creating subtypes and differentiating variables, they have no monopoly on such innovations. Studies using statistical methods have addressed the behavior of “democratizing” states and have examined the behavior of states that have democratic institutions but not democratic norms.121 Also, once case studies identify potentially useful subtypes, if a sufficient number of cases in the subtype exists statistical tests can assess whether these subtypes are indeed correlated with the specified outcome. In this way, case studies can often help develop sharper concepts, subtypes, or measurement procedures that can then be incorporated into statistical studies, though this can require considerable effort in recoding the cases in existing statistical datasets.
Examples of Case Study Research Design in the Interdemocratic Peace Literature
The democratic peace literature provides some of the best examples of how to implement case studies. These examples illustrate the important point that there is no single “case study research design.” Rather, different case study research designs use varying combinations of within-case analysis, cross-case comparisons, induction, and deduction for different theory-building purposes.122 An excellent example of a case study research design using both within-case analysis and cross-case comparisons is Paths to Peace, edited by Miriam Elman. Elman carefully defines the class of cases to be studied—international crises between democratic, mixed, and nondemocratic dyads—while acknowledging that this class of cases cannot adequately test the assertion that democracies