CHAPTER 1: ORIENTATION AND STATEMENT OF THE PROBLEM
4.4 RESEARCH METHODOLOGY
4.4.8 Data verification
4.4.8.2 Transferability
Morgan (2007:72) refers to the distinction between knowledge that is either specific and context-dependent or universal and generalized. Morgan does not believe it is possible for research results to be either so unique that they have no implications whatsoever for other actors in other settings or so generalized that they apply in every possible historical and cultural setting. From a pragmatic approach, an important question is the extent to which one can take the things that were learned with one type of method in one specific setting and make the most appropriate use of that knowledge in other circumstances. Once again, this involves a process of working back and forth, in this case between specific results and their more general implications. Morgan (2007:72) borrowed the idea of transferability of research results from Lincoln and Guba, who treated the question of whether the things learned in one context can be applied in another as an ‘empirical’ issue. The classic example is assessing whether the results from one particular programme evaluation has implications for the use of
similar programmes in other contexts. This advocacy of transferability thus arises from a solidly pragmatic focus on what people can do with the knowledge they produce and not on abstract arguments about the possibility or impossibility of generalizability (Mertens, 2005:427).
Within this single case study the researcher was therefore not primarily interested in generalizations. All observations were defined by the specific contexts and case in which they occurred. For that reason the researcher did not claim that knowledge gained from one context will necessarily have relevance for other contexts (Babbie & Mouton, 2001:277). Replication of this study will therefore not yield the same results, but this does not discredit the results of this particular study. The more important question for the researcher was rather whether the results are consistent with the data collected. Guba and Lincoln (1985:288) were the first to conceptualize reliability in qualitative research as ‘dependability’ or ‘consistency’. That is, rather than insisting that others get the same results as the original researcher, reliability lies in others’ concurring that given the data collected, the results make sense, they are consistent and dependable (Mertens, 2005: 351). With transferability the burden of proof for generalizability lies with the reader. It is the reader, not the researcher, who determines what can apply to his or her context (Merriam & Associates, 2002:179). The following strategies for transferability were employed during this study:
Triangulation
Triangulation is the strategy whereby the researcher used multiple methods and sought out different types of data sources that could provide insights about the same events or relationships (De Vos, 2002:341; Frank & Barzilai, 2004:46) in order to support the strength of the interpretations and conclusions reached (Mertens, 2005:426). By combining methods in the same study observers could thus partially overcome the deficiencies that flow from one method (Babbie & Mouton, 2001:275,358; Greene & McClintock, 1991:19; Merriam & Associates, 2002:26; Terre Blanche & Durheim, 1999: 215). The use of multiple methods or triangulation furthermore reflected an attempt to secure an in-depth understanding of the phenomenon in question. Triangulation was not a tool or strategy of validation, but rather an alternative to validation (Denzin & Lincoln, 2000:5).
Member checks
A second common strategy for ensuring transferability is member checks which entail going back to the source of information and checking both the data and the interpretation. The aim was to assess the intentionality of participants, to correct for obvious errors, and to provide additional volunteer information (Babbie & Mouton, 2001:277). Here the participants were asked to comment on the researcher’s interpretation of the data. That is, the researcher took the tentative findings back to some of the participants and asked whether their interpretation ‘rang true’. This usually occurred during informal conversations with colleagues after data interpretation prior to compiling the two evaluation reports (Merriam & Associates, 2002:26).
Peer reviews
Peer reviews were yet another strategy employed by the researcher. Here the researcher asked a colleague to scan some of the raw data and assess whether the findings were plausible based on the data (Merriam & Associates, 2002:26). Although colleagues did not necessarily have to agree with the claims made by the researcher, it was important that they should be willing to concede that the researcher’s conclusions were plausible, at least from the perspective of the researcher herself (Weber, 2004:viii). Furthermore the researcher reviewed her perceptions, insights and analysis with an expert in the field of PBL who was outside of the context of the study but who had a general understanding of the nature of the study (Babbie & Mouton, 2001:277).
Thick descriptions
Providing rich, thick descriptions involved providing sufficiently detailed descriptions of data in context and reporting them, with detail and precision, to allow judgements about transferability to be made by the reader (Babbie & Mouton, 2001:277). That is, enough description and information were provided so that readers would be able to determine how closely their situations matched the single case study and thus whether findings could be transferred to another setting (Merriam & Associates, 2002:28; Mertens, 2005:427).
Audit trial
An audit trail describes in detail how data were collected, how categories were derived, and how decisions were made throughout the inquiry (Merriam & Associates, 2002:27-28). An audit trial thus indicates the degree to which the findings are the product of the focus of the inquiry and not of the biases of the researcher (Babbie & Mouton, 2001:278). The audit trail
was dependent upon the researcher keeping a research journal throughout the conduct of the study. What went into this journal were the reflections, questions, and decisions on the problems, issues, ideas encountered in collecting data as well as a running record of the researcher’s interaction with the data whilst engaged in analysis and interpretation (Merriam & Associates, 2002:27-28). The aim of the audit trail was to allow independent readers to authenticate the findings of the study by following the trail of the researcher. While the researcher did not expect others to replicate her accounts, the best she could do was to explain how she arrived at the results.