Exemplary Case Design

In case study research, although a selected case occasionally is treated as a unique historical event (the emergence of the United States as a 20th-century superpower), more commonly it is treated as an instance of a class of comparable events (e.g., the emergence of the United States, Russia, and China as 20th-century superpowers). Sometimes (e.g., as part of a program evaluation) the results of a case study are expected to generalize to a set of comparable cases that exhaust a particular class (e.g., a specifiable group of schizophrenics in a particular hospital). At other times, the results are expected to generalize to a set of comparable cases within a broad or even theoretically infinite class (e.g., schizophrenics in general). Regardless of whether the relationships observed are expected to generalize to a finite or to a theoretically infinite class, the relations between the particular case selected for study and the class to which generalization might be expected should be—but often are not—articulated.

A frequent assumption is that the selected case is an exchangeable member of the class as a whole. That is, members of the relevant class are considered sufficiently homogeneous to warrant arbitrary, perhaps even random, selection of any member of the class for case study. When such uniformity can be substantiated, it is plausible that (a) each instance of the class possesses the features that define class membership; (b) the features that define class membership constitute—or are within—the boundary conditions for the occurrence of relationships observed in the selected case; and (c) the relationships observed in the selected case will, therefore, generalize to the other instances of the class. In social science research, however, members of a class repeatedly—and perhaps generally—are neither homogeneous nor exchangeable, and, to that extent, attempts to generalize from a single case become problematic.

One way to address this issue is by selecting for study a case that exemplifies the relevant class. When an exemplar has been appropriately identified, the relationships observed in that particular case may generalize to other cases to the extent that they, like the exemplar, possess the features that define class membership. This claim requires clarification of what it means to exemplify class membership, that is, what it means to say that members of the relevant class possess the features that define the relevant class. So, if an “exemplar” is “a typical example” of a class, how is such “typicality” to be understood?