Case Studies and Theory Development in the Social Sciences - Alexander L. George [49]
Yet a satisfactory integration of the five tasks usually cannot be accomplished on the first try. A good design does not come easily. Considerable iteration and respecification of the various tasks may be necessary before a satisfactory research design is achieved. The researcher may need to gain familiarity with the phenomenon in question by undertaking a preliminary examination of a variety of cases before finalizing aspects of the design.
Despite the researcher’s best efforts, the formulation of the design is likely to remain imperfect—and this may not be apparent until the investigator is well into phase two or even phase three of the study. If these defects are sufficiently serious, the researcher should consider halting further work and redesigning the study, even if this means that some of the case studies will have to be redone. In drawing conclusions from the study, the researcher (or others who evaluate it) may be able to gain some useful lessons for a better design of a new study of the problem.183
Chapter 5
Phase Two: Carrying Out the Case Studies
The fifth task in a research design—the formulation of general questions to ask of each of the cases to be studied in phase two—allows the researcher to analyze each case in a way that will provide “answers” to the general questions.184 These answers—the product of phase two—then constitute the data for the third phase of research, in which the investigator will use case findings to illuminate the research objectives of the study.
Usually one’s first step in studying a case with which one is not already intimately familiar is to gather the most easily accessible academic literature and interview data on the case and its context. This preliminary step of immersing oneself in the case, known as “soaking and poking,” often leads to the construction of a chronological narrative that helps both the researcher and subsequent readers understand the basic outlines of the case.185
After a period of “soaking and poking,” the researcher turns to the task of case study analysis, establishing the values of independent and dependent variables in a case through standard procedures of historical inquiry. (If appropriate, the researcher may be able to quantify and scale variables in some fashion.) The researcher should always articulate the criteria employed for “scoring” the variables so as to provide a basis for inter-coder reliability.
Next, the researcher develops explanations for the outcome of each case. This is a matter of detective work and historical analysis rather than a matter of applying an orthodox quasi-experimental design.186 Social scientists performing case studies will need to familiarize themselves with the craft of the historian’s trade—learning, for the context in which the case is embedded, the special difficulties presented by various kinds of evidence that may be available; using multiple weak inferences rather than single strong inferences to buttress conclusions; developing procedures for searching through large masses of data when the objectives of the search are not easily summarized by a few simple search rules.187
This chapter provides advice on these topics. The first three sections focus on the provisional nature of case explanations, and the challenges involved in weighing explanations offered by other researchers who have analyzed a given case, and the task of transforming a descriptive explanation for a case into an explanation that adequately reflects the researcher’s theoretical framework. We then turn to issues that researchers encounter when working with a variety of primary and secondary materials. Notable issues with secondary sources include the biases of their authors, and a tendency to overestimate the rationality