This article examines methodological issues for case studies in educational research and suggests a methodological framework for qualitative research in education. Researchers tend to separate empirical research into two types: quantitative research and qualitative research. Quantitative research is often used as a means of making causal inferences and testing hypotheses. In the social sciences, however, we can examine cases that cannot be studied using quantitative methods, such as field survey, ethnography, discourse analysis, and historical analysis. We often think that quantitative research is best suited to hypothesis testing and a nomothetic, generalized approach. Quantitative research may also require a large number of cases (N is large). On the other hand, qualitative research is suitable for hypothesis formulation and an idiographic, differential approach. In social science methodology, however, we often examine causal inferences and test hypotheses in cases with only a small number of subjects (N is small) or even one single subject (N is one). In education research in Japan, however, when researchers analyze small- or single-N cases, they almost always choose the idiographic, differential approach. There are two problems with this trend. First, it creates a divide between quantitative and qualitative research. Many researchers tend to regard quantitative and qualitative research as different methodologies. Second, in educational research in Japan, research designs for small- and single-N case studies have been significantly biased. As a result, researchers have made many idiographic monographs concerning various topics, but their findings cannot be systematically combined or used for theory construction or testing. To solve these problems, this paper advocates the importance of the logic of inference. When we analyze small-N cases, for example, we can choose between the method of agreement and the method of difference. In the method of agreement, the researcher chooses cases in which the dependent variable is not dispersed. In the method of difference, the dependent variable in the cases is dispersed. The method of difference is usually the better choice, but the method of agreement can make finding new hypotheses easier. If researchers can only observe a single case, they can analyze the causal relationship to examine an outlier case or a most likely (unlikely) case or a counterfactual case. This paper discusses two implications for educational research. First, the view that quantitative and qualitative research are different methodologies and logical forms are biased and underestimate the potential of small- and single-N case studies. Small- and single-N case studies can be analyzed using quantitative and/or qualitative research (i.e., a hypothesistesting, nomothetic, generalized approach and/or a hypothesis-formulation, idiographic, differential approach). Second, educational researchers in Japan should be more conscious of methodology and theory. Many of them have been eager to investigate the facts but tend to be cool toward examining causal relationships and constructing theory. Researchers should be as interested in theoretical contributions as they are in factfinding inquiries.
View full abstract