Thesis Title: Subtitle

Case Studies and Generalisability

The question of generalisation in qualitative research, and the extent to which this is necessary is a matter of some debate (Payne and Williams 2005). Quantitative and qualitative approaches to research have different understandings of generalisation. The problem of empirical generalisation is most often associated with a positivist approach to research, which creates a probability sample in order to make inferences about a wider population (Tripp 1985; Williams 2000). Therefore, some qualitative sociologists contest
the relevance of generalisation at all and deny that generalisation should be the outcome of qualitative work (Denzin 1983; Marshall and Rossman 1989; Denzin and Lincoln 1995). Despite this, the question of generalisation must still be addressed.
For case studies, Adelman et al. (1976) identify three different types of possible generalisations. The first possibility for generalisation is from the instance studied to the class it represents, for example from studying one private school, a researcher may be able to make generalisations private schools more broadly. The second avenue for generalisation is to generalise from case-bound features across a multiplicity of classes that might share the same features such as demographic characteristics. Finally, Adelman et al. (1976) suggest that generalisations might also be made about the case itself. When detailing the findings of a case study, findings are generalised to theory through which statements can be made about other cases. Case studies are also generalisable through falsification (Popper 2000). Falsification is widely employed in scientific research, and holds that if one observation does not fit with the proposition, the proposition is invalid and must be revised or rejected (Popper 2000). Ruddin (2006) further argues that to assert that one’s analysis of a case is valid is to also claim a measure of generalisability. That is, that the claims made about that case are also applicable to other cases with similar properties. Findings are not inferred from case studies, but rather constructions and patterns of meaning are imposed on cases (Mitchell 2000; Flyvbjerg 2001). The choice of case studies is not without its limitations, but as demonstrated earlier this method is aligned with the ontological position from which this research starts.
Case studies are perceived as being difficult to generalise from because they are perceived to be too particular, context dependant and isolated from their broader context to be useful objects of generalisation (Ruddin 2006). However, scholars such as Flyvbjerg (2011) argue that case studies, far from being isolated objects must be grounded in relation to their environment. As advocated by Flyvbjerg (2011) this research is situated within its broader social context. This context includes anxieties around the social effects of SNS and the patterns of late modernity more broadly that shape our lived experiences. While Facebook is the object of research, I have positioned this object within broader debates about community and social relationships.
In addition to this, my approach to generalisability is also influenced by the subtle realist position in which Hammersley (1992) identifies three key elements as central in judging research: knowledge, reality, and the aim of social research. Instead of focusing on the validity and ‘truth’ of a claim as a measurement of knowledge, knowledge is defined by
Hammersley (1992) as beliefs about which one can be reasonably confident. This means assessing a claim based on its plausibility, credibility and compatibility with the observed phenomena. For Hammersley the plausibility and credibility of claims made in social research are assessed in two ways. Firstly, Hammersley (1992) proposes that we must consider whether claims are sufficiently plausible given our existing knowledge of the social world. However, he acknowledges that it is the researcher in the production of their research who decides what claims correspond to the phenomena being examined (Hammersley 1992). Social reality is never presented ‘as it is’ in qualitative research, but rather it is shaped by the researcher’s particular point of view. Thus reality, as it is represented in qualitative accounts is mediated twice; once by the participants’ accounts and second by the researcher and the research.
Claims that are novel, or outside of our existing knowledge about the world require extra evidence from the researcher so the reader can judge these claims as plausible and credible. Compatibility rests on the whether the account made by the researcher can reasonably claim to depict social reality. Abandoning a focus on validity as a measure of knowledge requires acknowledging, “there are phenomena independent of our claims about them, which those claims may represent more or less accurately” (Hammersley 1992: 51). In the context of this research, it means that I have sought to provide additional evidence for claims that are novel or outside of existing knowledge regarding Facebook or SNS more generally. In additional to this Hammersley (1992) acknowledges that there are certain elements of common sense to the process of judging plausibility and credibility in which the reader, with the help of the information provided by the researcher makes his or her assessment based on their knowledge and position in relation to the phenomenon.
Therefore, use of multiple methods to create this case is not to make claims to validity, but rather to provide the reader with enough information to confidently support the new claims made in this research.