CHAPTER 2: REVIEW OF LITERATURE
3.5. Credibility, Validity and Trustworthiness
To defend their claims, research studies need to show that their design and processes are valid and dependable (Creswell & Miller, 2000; Golafshani, 2003; Guba & Lincoln, 1989; Maxwell, 2005; Patton, 1999; Silverman, 2005). Quantitative studies generally adhere to the rules and regulations of validity and reliability. The former requires that the study show how the instrument chosen to collect data measures what it claims to measure. Standardised units of measurements are therefore used in a standardised manner to objectively measure the same units across different contexts. Reliability requires that the results of the study be replicable in those different contexts (Golafshani, 2003; Silverman, 2005). The quantitative definitions and requirements of validity and reliability have been found to be inapplicable in qualitative research because in the latter the instruments of measurements tend to be the researchers themselves, who bring their subjective theoretical perspectives into the study processes, thus shaping the study processes and having an influence on the context under study. Additionally, data that tends to be measured in qualitative research is perceived as context-dependent and interpretation thus has to include the particular features of that context and how it influences the conclusions arrived at. These conclusions therefore cannot be expected to be generalised and replicable outside the context within which they were formed. Qualitative researchers instead focus on demonstrating credibility, validity and trustworthiness of their research processes and conclusions. This is done by displaying how decisions made for a qualitative study design are coherent, that the methods chosen were applied rigorously, and that the conclusions arrived at are valid and defensible. This is typically done by leaving an audit trail
for potential readers (Creswell & Miller, 2000; Golafshani, 2003; Guba & Lincoln, 1989; Maxwell, 2005; Patton, 1999; Silverman, 2005).
An audit trail is a rich, exhaustive description of a study’s design and processes, detailing how the study was formulated and conducted, right up to how the conclusions were arrived at. This detailed description allows reviewers and other readers to audit the credibility and validity of the research. This entire dissertation serves as an audit trail for such purposes. Each interview has been transcribed verbatim and one of them is attached under the appendices section (see Appendix 8) as a representative example. Another aspect of the dissertation that serves to demonstrate the credibility of this researcher is the use of researcher reflexivity. This is when a researcher illuminates the theoretical perspective within which they perceive the research problems, study design, and study results. This allows any reader of the study to contextualise the concepts used to describe the research problem, the selection of methodology, and the framework within which the results were interpreted. This dissertation makes it clear that this study’s design and processes have been formulated within an interpretive framework. During the analysis stage the researcher went back and forth between the interview transcriptions and the Review of Literature and Results and Discussion chapters in an effort to seek out any theoretical misinterpretations. This ensures that conclusions arrived at by the researcher are embedded in literature and not just assumed by the researcher (Creswell & Miller, 2000; Patton, 1999).
The validity of the data has been ensured by the use of rich exhaustive data, comparison, triangulation, and member checking (Creswell & Miller, 2000; Golafshani, 2003; Guba & Lincoln, 1989; Maxwell, 2005; Patton, 1999; Silverman, 2005). Sampling to the point of information redundancy ensured that participants’ accounts were not selected due to bias but were instead explored exhaustively. Interviews were conducted in a semi-structured manner to allow interviewees to speak exhaustively about the pre-identified issues as well as to elaborate on aspects that may not have already been identified by the interviewer. All interviews were recorded with audio tape recorders to ensure that all this data was captured. Comparison entails that researchers do not just seek participants who will confirm their study hypotheses, but also search for potential disconfirming cases, in other words participants from the same context who might provide different perspectives to those already identified. Seeking a diversity of perspectives is also knows as triangulation, which is data collection with different instruments and/or from different perspectives. This study opted to interview students from different levels of undergraduate and postgraduate studies, from different faculties, from different linguistic
and national backgrounds, and select both males and females. The participants’ differences in terms of their experiences with bilingual instruction contributed different angles of perception of the bilingual policy. Additionally, the questions in the interview schedules were worded to ask about the same particular issue in different ways, to ascertain whether participants remain consistent in their perspectives or change them according to the way a that issue is asked. Comparison also applies during the process of data analysis, by seeking themes that may potentially disconfirm those that have been identified. This was also done by the researcher. During the interviews, member checking, a form of validating the researcher’s interpretation of the data by asking the speaker to confirm or disconfirm that interpretation in real time, was employed by the researcher. In instances of misinterpretation, the interviewees had the opportunity to clarify (Creswell & Miller, 2000; Golafshani, 2003; Guba & Lincoln, 1989; Maxwell, 2005; Patton, 1999; Silverman, 2005).