Reflections on case studies, modelling and theory building

Abstract

This paper provides a discussion on how case study research fits into the bigger picture of theory building in the social sciences. In particular, it discusses how case study research, as opposed to more classical quantitative (often statistical) approaches contribute to the progression of collective learning in science. As such, for arguments sake, in this paper these approaches (i.e. qualitative vs. quantitative) are put into a competitive relationship. It is acknowledged that such competitive relationship does not exist in theory, as they are often considered to be complementary, but in practice qualitative research is often more difficult to publish; and tends to be more difficult to defend. Therefore the authors believe there is a need to better recognise the role of case study research in the scientific process; and in light of recent development in participatory modelling to collectively re-evaluate how case study research contributes to the progression of scientific learning? The deductive statistical approach in social science, as it was defined in a classic exchange of articles and ideas by mathematician Bernoulli and Galton in the 18 th century is to ascertain a posteriori through knowledge based on experience, in situations when we cannot determine a “priori” (independent of experience) by statistical analysis of a number of similar instances. This assumes events in the future will follow a similar pattern as in the past, and relies on several important assumptions which may inherently limit applicability. Firstly, the output of “priori” calculations is invariably correlations of relationship similarities rather than identified as causative factors. Secondly, the assumption of “similar conditions” is problematic because by definition this requires causal surmise and cannot be ascertained purely through statistical calculations. Consequently, social science research based on statistics alone is incomplete and requires causal explanations, suggesting the need for theoretical formulations. Whilst there are modern statistical techniques to alleviate some of these issues, such as capture-recapture, jack-knife and bootstrap; this shows that social science based on statistics alone is incomplete but requires causal explanations which in turn require the formulation of theory. Conversely, inductive case study research has been criticized largely for being unable to result in laudable generalisations. The authors suggest case studies are suitable for developing human understanding of issues, and producing context dependent knowledge necessary for the formulation of theory. Specifically, case study research can be used for abductive reasoning whereby hypothetical explanations, A, of observations are found that are sufficient but do not necessarily explain observations B. Traditional scientific method would dictate that such a theory is tested using empirical falsification, but this is not always possible due to difficulties of observation. It is clear that the research process requires both (inductive) case study research as well as (deductive) statistical approaches. Furthermore, there is a need for a trained mind to synthesize the results and develop and choose the “best” theory / model to explain observations (abductive reasoning). Participatory modelling with its mix of qualitative and quantitative methods places it within two major paradigms: generalisation via theory and statistics vs. context specific knowledge via case study. The question is whether participatory modelling can be the bridge between two main paradigms, by allowing for the development of general theories through the immersion of oneself in the specific. Limitations to possible applications remain, as participatory modelling is usually warranted and feasible only in certain cases, and is ethically dependant on circumstance. The authors argue that participatory modelling has not been developed with these purposes in mind, but serves merely as a thought provoking exemplar that is hopefully paving the way for the next generation of research methods. If this is to be the case, participatory modelling requires a strong theoretical foundation considering its place in epistemology and scientific method.