Misunderstanding No. 3 about Case Studies
Prof. Bent Flyvbjerg
Oxford University. IT University of Copenhagen. Villum Kann Rasmussen Professor and Chair. Advisor to 10 Downing Street. Bestselling author in 22 languages. Award-winning scholar, speaker, advisor.
Misunderstanding no. 3. The case study is most useful for generating hypotheses, while other methods are more suitable for hypotheses testing and theory building
Correction no. 3: The case study is useful for both generating and testing of hypotheses but is not limited to these research activities alone.
See sources and read more here: bit.ly/1bHZ1Mx (you may have to copy and paste link).
The third misunderstanding about the case study is that the case method is claimed to be most useful for generating hypotheses in the first steps of a total research process, while hypothesis-testing and theory-building is best carried out by other methods later in the process, as stipulated by the Penguin definition of the case study. This misunderstanding derives from the previous misunderstanding that one cannot generalize on the basis of individual cases. And since this misunderstanding has been revised as above, we can now correct the third misunderstanding as follows:
The case study is useful for both generating and testing of hypotheses but is not limited to these research activities alone.
Eckstein--contravening the conventional wisdom in this area--goes so far as to argue that case studies are better for testing hypotheses than for producing them. Case studies, Eckstein (1975, 80) asserts, “are valuable at all stages of the theory-building process, but most valuable at that stage of theory-building where least value is generally attached to them: the stage at which candidate theories are tested.” George and Bennett (2005, 6-9) later confirmed and expanded Eckstein's position, when they found that case studies are especially well-suited for theory development because they tackle the following tasks in the research process better than other methods:
- Process tracing that links causes and outcomes.
- Detailed exploration of hypothesized causal mechanisms.
- Development and testing of historical explanations.
- Understanding the sensitivity of concepts to context.
- Formation of new hypotheses and new questions to study, sparked by deviant cases.
Even rational choice theorists have begun to use case study methods to test their theories and hypotheses, which, if anything, should help deflate the decades-old antagonism between quants and qualts over case study research (Bates et al. 1998, Flyvbjerg 2006).
Testing of hypotheses relates directly to the question of “generalizability”, and this in turn relates to the question of case selection. Here generalizability of case studies can be increased by the strategic selection of cases (on the selection of cases, further see Ragin 1992; Rosch 1978). When the objective is to achieve the greatest possible amount of information on a given problem or phenomenon, a representative case or a random sample may not be the most appropriate strategy. This is because the typical or average case is often not the richest in information. Atypical or extreme cases often reveal more information because they activate more actors and more basic mechanisms in the situation studied. In addition, from both an understanding-oriented and an action-oriented perspective, it is often more important to clarify the deeper causes behind a given problem and its consequences than to describe the symptoms of the problem and how frequently they occur. Random samples emphasizing representativeness will seldom be able to produce this kind of insight; it is more appropriate to select some few cases chosen for their validity.
Table A summarizes various forms of sampling. The extreme, or deviant, case can be well-suited for getting a point across in an especially dramatic way, which often occurs for well-known case studies such as Freud’s “Wolf-Man” and Foucault’s “Panopticon.” The deviant case is also particularly well suited for theory development, because it helps researches understand the limits of existing theories and to develop the new concepts, variables, and theories that will be able to account for what was previously considered outliers.
In contrast, a critical case can be defined as having strategic importance in relation to the general problem. The above-mentioned strategic selection of lead and feather for the test of whether different objects fall with equal velocity is an example of critical case selection. This particular selection of materials provided the possibility to formulate a type of generalization that is characteristic of critical cases, a generalization of the sort, “If it is valid for this case, it is valid for all (or many) cases.” In its negative form, the generalization would be, “If it is not valid for this case, then it is not valid for any (or only few) cases.”
How does one identify critical cases? This question is more difficult to answer than the question of what constitutes a critical case. Locating a critical case requires experience, and no universal methodological principles exist by which one can with certainty identify a critical case. The only general advice that can be given is that when looking for critical cases, it is a good idea to look for either “most likely” or “least likely” cases, that is, cases which are likely to either clearly confirm or irrefutably falsify propositions and hypotheses. A model example of a “least likely” case is Robert Michels’s (1962) classical study of oligarchy in organizations. By choosing a horizontally structured grassroots organization with strong democratic ideals--that is, a type of organization with an especially low probability of being oligarchic--Michels could test the universality of the oligarchy thesis; that is, “If this organization is oligarchic, so are most others.” A corresponding model example of a “most likely” case is W. F. Whyte’s (1943) study of a Boston slum neighborhood, which according to existing theory should have exhibited social disorganization, but in fact showed quite the opposite (see also the articles on Whyte’s study in Journal of Contemporary Ethnography, vol. 21, no. 1, 1992).
Cases of the “most likely” type are especially well suited to falsification of propositions, while “least likely” cases are most appropriate to tests of verification. It should be remarked that a most likely case for one proposition is the least likely for its negation. For example, Whyte’s slum neighborhood could be seen as a least likely case for a hypothesis concerning the universality of social organization. Hence, the identification of a case as most or least likely is linked to the design of the study, as well as to the specific properties of the actual case.
A final strategy for the selection of cases is choice of the paradigmatic case. Thomas Kuhn has shown that the basic skills, or background practices, of natural scientists are organized in terms of “exemplars” the role of which can be studied by historians of science. Similarly, scholars like Clifford Geertz and Michel Foucault have often organized their research around specific cultural paradigms: a paradigm for Geertz lay for instance in the “deep play” of the Balinese cockfight, while for Foucault, European prisons and the “Panopticon” are examples. Both instances are examples of paradigmatic cases, that is, cases that highlight more general characteristics of the societies in question. Kuhn has shown that scientific paradigms cannot be expressed as rules or theories. There exists no predictive theory for how predictive theory comes about. A scientific activity is acknowledged or rejected as good science by how close it is to one or more exemplars; that is, practical prototypes of good scientific work. A paradigmatic case of how scientists do science is precisely such a prototype. It operates as a reference point and may function as a focus for the founding of schools of thought.
As with the critical case, we may ask, “How does one identify a paradigmatic case?” How does one determine whether a given case has metaphorical and prototypical value? These questions are even more difficult to answer than for the critical case, precisely because the paradigmatic case transcends any sort of rule-based criteria. No standard exists for the paradigmatic case because it sets the standard. Hubert and Stuart Dreyfus see paradigmatic cases and case studies as central to human learning. In an interview with Hubert Dreyfus (author’s files), I therefore asked what constitutes a paradigmatic case and how it can be identified. Dreyfus replied:
"Heidegger says, you recognize a paradigm case because it shines, but I’m afraid that is not much help. You just have to be intuitive. We all can tell what is a better or worse case--of a Cézanne painting, for instance. But I can’t think there could be any rules for deciding what makes Cézanne a paradigmatic modern painter . . . [I]t is a big problem in a democratic society where people are supposed to justify what their intuitions are. In fact, nobody really can justify what their intuition is. So you have to make up reasons, but it won’t be the real reasons."
One may agree with Dreyfus that intuition is central to identifying paradigmatic cases, but one may disagree it is a problem to have to justify one's intuitions. Ethnomethodological studies of scientific practice have demonstrated that all variety of such practice relies on taken-for-granted procedures that feel largely intuitive. However, those intuitive decisions are accountable, in the sense of being sensible to other practitioners or often explicable if not immediately sensible. That would frequently seem to be the case with the selection of paradigmatic cases. We may select such cases on the basis of taken-for-granted, intuitive procedures but are often called upon to account for that selection. That account must be sensible to other members of the scholarly communities of which we are part. This may even be argued to be a general characteristic of scholarship, scientific or otherwise, and not unique to the selection of paradigmatic social scientific case studies. For instance, it is usually insufficient to justify an application for research funds by stating that one’s intuition says that a particular research should be carried out. A research council ideally operates as society’s test of whether the researcher can account, in collectively acceptable ways, for his or her intuitive choice, even though intuition may be the real, or most important, reason why the researcher wants to execute the project.
It is not possible consistently, or even frequently, to determine in advance whether or not a given case--Geertz' cock fights in Bali, for instance--is paradigmatic. Besides the strategic choice of case, the execution of the case study will certainly play a role, as will the reactions to the study by the research community, the group studied, and, possibly, a broader public. The value of the case study will depend on the validity claims which researchers can place on their study, and the status these claims obtain in dialogue with other validity claims in the discourse to which the study is a contribution. Like other good craftsmen, all that researchers can do is use their experience and intuition to assess whether they believe a given case is interesting in a paradigmatic context, and whether they can provide collectively acceptable reasons for the choice of case.
Finally, concerning considerations of strategy in the choice of cases, it should be mentioned that the various strategies of selection are not necessarily mutually exclusive. For example, a case can be simultaneously extreme, critical, and paradigmatic. The interpretation of such a case can provide a unique wealth of information, because one obtains various perspectives and conclusions on the case according to whether it is viewed and interpreted as one or another type of case.