CHAPTER 3: RESEARCH DESIGN
3.7 Data Analysis
In line with accepted practice (Lee, 1999), taped interviews were transcribed verbatim and checked against the original tapes for transcription accuracy. The transcription process enhanced data accessibility and facilitated analysis. Each interviewee was asked the questions, shown in the interview schedule in Appendices 1 & 2. All interviewees were asked the same questions in the same order, with the obvious exception that the specialist commentators were not asked about the make-up of their current boards, their remuneration, nor about their reasons for becoming a director or involved in governance. Occasionally, in my role as interviewer, I asked a probing, or follow-up, question to elucidate a response (Polkinghorne, 2005). However, this was rarely needed as the participants were generally articulate and happy to talk about their experiences and views. The resulting quite structured and similarly shaped interview transcripts made a systematic ‘coding’ analysis relatively straightforward.
Written transcripts of the interviews were examined line by line, to identify key words and common themes in responses that were then grouped into categories (Holton, 2007). Essentially, as each response was read, the exact word or phrase was recorded against the code number for that participant. Where descriptions were provided, they were summarised into a key word or phrase. For example, a description of what was done prior to the meeting in relation to reading papers and understanding issues raised for discussion at the board meeting, was categorised as ‘preparation’. Once the key words or phrases were logged from each response, the whole set was read, counted and put into a table. This process enabled patterns to emerge, such as use of the same words. For example, one person may have said: “ask key questions” and another may have said: “ask important questions”. Both responses were grouped into the one category “Ask key questions”. As all responses were logged, there were cases where categories had only one or two entries. This detail was provided to ensure a
comprehensive set of perceptions was recorded and to prevent arbitrary generalisations (Morrow, 2005). From these groupings/categories, responses were tallied by gender and participant grouping (i.e. Crown director, private director or specialist commentator) to provide further detailed analysis. This also provided a frequency count to ascertain the extent to which respondents held similar views - essentially, as articulated in the research question, the perspectives of Crown Company directors and specialist commentators on each of these issues. This form of analysis was used for all of the interview questions.
While the interview data itself is not quantitative, categories were counted for clearer presentation. This “by hand” analysis largely mimicked the identification, coding and counting process that occurs with qualitative analysis support packages, such as the HyperResearch data analysis package. I chose not to use HyperResearch as I believe it is best used for grounded theory approaches (Glaser, 1992) which allow categories to surface from interviewees’ own words. In this case, the initial coding was conducted in close relation to key question areas which in turn had been derived from theory and previous research in the area. Once initial categories were established, the whole transcript was searched for further comments relative to each emerging theme in the recognition that participants in the course of the interview were likely to revert to, and elaborate on, earlier comments or in the process of answering another question recall other relevant additional information in relation to earlier questions. A ‘by hand’ analysis was also deemed more appropriate in this project as I wished to always review each response in its full interview context. Further, the characteristics of each interviewee were important in the analysis. Such relevant characteristics were, for instance, gender, experience in private sector governance as well as public sector governance, particular focus or position of specialist commentator. Investigation without the support of software was therefore regarded as helping produce a more nuanced analysis.
Questions had also been developed from the literature and were in turn linked to key themes identified in the literature as relevant to director selection, as explained above and in Section 3.4. Each answer was analysed across the participants and common categories were identified. The coding process was informed largely by the use of key
following chapter (Chapter 4), descriptions of the categories and how they were formed are explained and further clarified, supported by quotations from the interviews. Although this is qualitative data, a frequency count was undertaken, allowing for a quantitative approach within this process. The directors’ comments were separated from those of the specialist commentators. As the project highlights aspects of board diversity, particularly gender diversity, the responses and comments of male and female directors have also been coded and cumulated separately to see if they show any differences.
Whilst this quantification of the qualitative interview offers one way into the data, Brewerton and Millward (2001) point out the importance of qualitative content analysis that places the “emphasis on meaning rather than on quantification” (p. 152). Consistent with this approach, has been the careful examination of the context within which comments were made and the use of illustrative quotations in presenting the data. Further, the interviews also contained the opportunity for the participants to tell an illustrative story from which themes emerged.