Case Studies and Theory Development in the Social Sciences - Alexander L. George [51]
In any case, if the data and generalizations available to the investigator do not permit him or her to choose from competing explanations, then both explanations for the case should be retained as equally plausible, and the implications of both for theory development should be considered in phase three of the study.
Transforming Descriptive Explanations Into Analytical Explanations
In addition to developing a specific explanation for each case, the researcher should consider transforming the specific explanation into the concepts and variables of the general theoretical framework specified in Task Two.190 (In Harry Eckstein’s terminology, such research is “disciplined-configurative” rather than “configurative-idiographic.”) To transform specific explanations into general theoretical terms, the researcher’s theoretical framework must be broad enough to capture the major elements of the historical context. That is, the set of independent and inter-vening variables must be adequate to capture and record the essentials of a causal account of the outcome in the case. The dividing line between what is essential and what is not is whether aspects of a causal process in a given case are expected or found to operate across the entire class of cases under consideration. For example, if some instance of organizational decision-making was decisively affected by the fact that one of the key participants in the decision process caught a cold and was unable to attend an important meeting, this would not constitute a basis for revising our theory of organizational decision-making to endogenize the susceptibility of actors to disease. It would, however, constitute a basis for a general argument about how outcomes are affected by the presence or absence of important potential participants.
Some historians will object to this procedure for transforming a rich and detailed historical explanation into a more abstract and selective one couched in theoretical concepts, arguing that unique qualities of the explanation inevitably will be lost in the process. This is undoubtedly true: some loss of information and some simplification is inherent in any effort at theory formulation or in theoretically formulated explanations. The critical question, however, is whether the loss of information and the simplification jeopardize the validity of the conclusions drawn from the cases for the theory and the utility of that theory. This question cannot be answered abstractly. The transition from a specific to a more general explanation may indeed lead a researcher to dismiss some of the causal processes at work in the case simply because they are not already captured by the general theory or because the researcher fails to recognize a variable’s general significance. To say that avoiding these errors depends on the sensitivity and judgment of the researcher, while true, is not very helpful. One slightly more specific guideline is that researchers seem more susceptible to this error when trying to discern new causal patterns than when attempting to evaluate claims about some causal patterns already hypothesized to be operating in a particular case; and second, that the more fine-tuned and concrete the description of variance, the more readily the analysis will accommodate a more differentiated description of the causal processes at work.191
To the extent that the case study method has arisen from the practice of historians, it has tended to follow certain procedures that are not really appropriate for social scientists. One feature of most historians’ work is a relative lack of concern with or discussion of methodological issues en-countered in the performance of research. We believe that case researchers should explicitly discuss the major research dilemmas the case study researcher faced in