• No results found

Chapter 3 Research Methodology

3.4 Research Design

3.4.5 Data Analysis

A recurring theme throughout this chapter was connected with the problematic nature of following case study principles. Yin (2003) set out the uncertainties faced by researchers, particularly when they are inexperienced and he warned a lack of planning in the research and case design will likely lead to a cessation of productive analysis from the data collected. As was the case with the methodological choices made at the outset of the research design, so too were choices required surrounding the strategy for the analysis of the data collected. Unlike the relative

straightforwardness of statistical methods, a qualitative approach relies heavily on the researcher’s “rigorous thinking” and “presentation of evidence” and their ability to carefully consider “alternative interpretations” (Yin, 2003, p.110). Miles and

Huberman (1994, p.50) commented the analysis should be undertaken throughout the life of the project and the researcher should leave enough scope within the

127

project for changes and variations. They strongly recommended undertaking early data analysis as this will assist in ensuring the questions set out at the outset of the study are addressed in full. The unit for analysis needs to be clearly set-out to ensure the data collection is a fruitful exercise.

Evaluating and analysing rich qualitative data in a rigorous manner is difficult and “…no single qualitative data analysis approach is widely accepted” (Neuman,

2014). The approach in this project was to use an evolving set of categories to structure the qualitative data as it was gathered. Firstly, a set of seed categories (Miles and Huberman, 1984, Fitzgerald, 1997, Wynekoop and Russo, 1997) was formulated based on the initial theoretical model and these were used to formulate the initial structured interview script. In each interview other categories and sub- categories emerged and were incorporated into interview scripts for investigation in the following interviews. So the number of categories grew as the case studies continued. ed and revealed themes and documented revisions of the theoretical model.

The presentation of data is based on illustrated narrative style, or an oral narrative told in the first person, as described by Miles and Huberman (1994) and Myers (1997) and as used in Fitzgerald (1997) and Urquhart (1998). This approach as described by (Miles and Huberman, 1994) does not resort to explicit coding but looks for “...key words, themes and sequences to find the most characteristic accounts.”

The analysis of the data followed an intuitive path and included the initial free coding of the interview transcripts, field notes and observations captured during the interviews. Several wave of coding were carried out to ensure that none of the key emergent themes have been overlooked. The researcher spent time any effort

128

reviewing each transcript repeatedly in order to be fully familiar and immersed in the data. Utilising the research questions the analysis explored the “how, what and why”

questions in order to reveal the perceptions of the individual interviewees and build a picture of the approach adopted by their individual organisations.

A recurring issue in the data collection phase of this research is attributable to the very nature of interpretivist research and as a consequence the findings, at times, appear confusing or even ambiguous, which is why the initial coding was carried out intuitively rather than by selecting specific fragments of text which might have appeared more relevant at first. Furthermore, this decision was made in view of an attempt by the researcher to ensure that no significant themes were inadvertently overlooked. Early tentative analysis was also carried out simultaneously whilst the data collection phase continued, which meant although the overarching interview questions remained static, further supplementary questions were utilised to develop individual themes.

Miles and Huberman (1994, p.51) argued early processing of the collected data is fundamental but there are many challenges in doing so in a meaningful way. They claimed the researcher must be particularly careful when writing up to ensure their personal bias does not creep in at any stage and they also advised any data which is not recorded verbatim, should marked as such, to ensure it is clear it is the researcher’s opinion rather than an actual event or statement captured. Further, they

posited it is difficult for the researcher to capture body language or tone of voice in the written word and the emphasis they place on this information within the research needs to be adequately outlined for the reader.

Miles and Huberman (1994) discussed the main concerns with the legibility of raw field notes or hand written transcription from the actual interviews and contended

129

there is much work for the researcher well in advance of any analysis taking place. Additionally, the transcription of audio files from interviews is fraught with “slippage”

as the researcher may select information which is interesting to themselves (but that might not be of any true value), in lieu of information capturing the true essence of the phenomena, which might provide a much greater richness in the analysis phase. Miles and Huberman (1994, p.52) suggested it is important to record the immediate observations of the interview and then put forward some ways in which this can be done. Use of a contact record sheet (among other techniques) allows the researcher to immediately consider some themes which may have emerged from the content of the interview. This can be particularly helpful for the researcher as it may illuminate themes meriting further investigation in later interviews. Other examples of the techniques the researcher can employ for early analysis of the data include: use of codes and coding; pattern coding; memoing; case analysis meetings; interim case summaries; vignettes; pre-structural cases and sequential analysis. This list is not exhaustive and most researchers will have personal preferences rather it provides a picture of the type of techniques available. The actual sequence of events in

analysing the data collected in this inquiry are as follows.

Table 3.3 Schedule of analysis of the data

Step no. Process Followed

Step 1 Codes are assigned to initial materials collected (interview,

documentary evidence and observations

Step 2 Reflective commentary was added to the transcripts.

Step 3 The transcripts and codes are revised and further codes were added.

Step 4 The materials are then reviewed alongside the coding and patterns of

phrases, consequences of events and langue used is reviewed to assist in identifying any relationships and early emergent themes.

130

Step 6 Some generalisations of the data are made and reviewed further to

check authenticity of the relationships.

Step 7 The generalisations were then grouped together to assist in compiling a

sensible and coherent body of information to form the thematic constructs.