The constant comparison method of analysis, combines coding and analysis, re-visiting whether, and how much, cases are described well by the coding framework that is developed, to eventually reach a final theory or themes following much iteration (Boeije, 2002; Glaser, 1965; Murphy et al., 1998).
Although constant comparison is often discussed in relation to developing a theory (especially in grounded theory methods), it has relevance to other methods of qualitative analysis, though may not be explicitly labelled in that way (Boeije, 2002). Constant comparison was used within the broader Braun and Clark (2006) framework for thematic analysis used in this thesis. This iterative approach, described earlier, involves the development of a coding framework, simultaneous to the analysis, with both informing each other, along with comparing the findings between participants. Although there are some frameworks for undertaking
constant comparison in the context of a grounded theory approach, these were not specifically taken, since they were not wholly compatible with the thematic analysis framework followed (Boeije, 2002; Braun and Clark, 2006).
6.7.4 Inter-coder reliability
When analysing qualitative research, there are mixed views in particular on inter-coder reliability (Cohen & Crabtree, 2008). Some report inter-coder reliability
nature of analysis, the assumed element of subjectivity and the expertise developed by the researcher, it was felt unrealistic to expect any inter-coder
reliability tests to be high and unnecessary given the philosophical approach to the research. This falls within the critical realist approach to the whole thesis,
described in Chapters 1 and 3, and earlier in this Chapter, which assumes that the analyst of data will influence the findings through their own understanding of the world and the topic discussed (Bhaksar, 2010; Maxwell, 2010; Scott, 2007). Inter- coder reliability was, therefore, not undertaken. It was, nevertheless, felt useful for supervisors to examine the data and associated codes at different points in
analysis to enable discussion and refinement (Barbour, 2001). During the second phase of analysis, supervisors examined two transcripts to get a general sense of the codes emerging. These were discussed, which further enhanced the analysis. During the fourth phase, supervisors again examined two transcripts, which this time were coded. This enabled further discussion, deliberation, and refinement of codes and code names.
6.7.5 Negative cases
Attention to negative cases is a further area for consideration in analysis. Most commonly, this refers to the identification and discussion of cases that
deviate from the norms found in the data (Morse et al., 2002; Murphy et al., 1998). Therefore, ‘negative cases’ are not considered pejorative; more, it is a way of describing examples of the variety of responses that may not represent the majority of participants’ narratives. Within analyses, negative cases, where
qualitative results Chapter (7), and in the discussion. This represents ‘fair dealing’, where the data balances individual’s narratives with presenting results that are collectively robust (Murphy et al, 1998). Therefore, where themes or codes are only expressed by a very small number of individuals, they may not be representative of men with cancer on the whole, and may be expressed as a negative case, rather than fully integrated into the key findings.
6.7.6 Respondent validation
A further approach that can be used to increase rigour in analysis is
respondent validation (sometimes known as member checking; Long and Johnson, 2000; Mays and Pope, 1995; Murphy et al., 1998). This most commonly involves the researcher feeding back findings to participants and gathering views on the perceived accuracy of findings. Potential difficulties with member checking arise since it relies on participants’ reading the results in detail, that they will be unbiased in their assessment, and, crucially, understand and relate to the results as
representing collective findings of multiple participants, not them as an individual (Morse et al., 2002; Murphy et al., 1998). Since the final results would include the collective data from multiple participants, it was felt that member checking would not effectively increase the rigour of analysis of this work, since the collective results would not fully represent individual’s stories.
6.7.7 Triangulation
this approach suggest that it is not possible to triangulate research, or that it does not contribute to validity (Cameron, 2009; Moran-Ellis et al., 2006; Murphy et al., 1998; Sale et al., 2002). Rather than attempting triangulation, and assuming a greater validity through the research, it was instead assumed that the mixed- methods approach employed enabled a breadth and depth to findings.
6.7.8 Reflexivity
As demonstrated in Chapter 7, a large amount of data were acquired from the interviews. Reflexivity is usually defined as the influence that a researcher has on the findings, acknowledging that qualitative research has an element of
subjectivity (Jootun, 2009; Mays and Pope, 2000; Murphy et al., 1998). Here, I reflect on various factors that may have influenced the design and interpretation of the qualitative research, recognising that my own interpretation of ideas, concepts and the data will influence the results presented, in line with the critical realist approach discussed above (Bhaksar, 2010; Maxwell, 2010; Scott, 2007). Firstly, I was developing the interview questions and analysing the interview data as a Health Psychologist, who had worked in practice with oncology patients. This naturally helped shape the focus of questions, and influenced the analysis, both around the particular code names ascribed to data, and the interpretation of the data. Therefore, it is likely that someone from another discipline, and perhaps a Health Psychologist with purely research experience, would have developed questions and interpreted men’s narratives differently. Further, my training in applied psychology will have influenced how I prompted and probed for details within men’s stories. This, too, was influenced by my qualitative research training,
however, the use of reflections in particular is an approach I use in practice and was not part of my research training. Indeed, I feel this was a useful approach to help men elaborate on points. At other times, I did probe more deeply than I would or could have done had I been working with these men in applied practice. My role as a researcher both enabled me to do this, since I was there to discover, and helped facilitate men to talk about their stories, since they had entered into the conversation as willing research participants, and they saw their role to respond to questions (naturally knowing that they also were not under pressure to do so). There were times when men either did not know an answer off the top of their head, gave very short answers, or perhaps did not see the point of a question asked during the interviews. These were occasions when I asked them to
elaborate. Often, this resulted in insights that I would not have achieved otherwise, nor would I have gain such insights had I been working with these patients in my role as a Health Psychologist working in practice.
The analysis was clearly shaped by my understanding of health and illness, as part of my disciplinary training. This includes the knowledge of models of illness, stress and coping, which informed the analysis and results presented in Chapter 7. Due to part of the interview questions investigating the role of masculinity, I feel that being a woman may have assisted men to open up, since there was no need to ‘prove’ their masculinity to another man. Similarly, being younger than all the participants meant that I could have been seen as being different to them. This
asked and more detail given in answers. Men were also aware that they were speaking to a psychologist. It is possible that they did use this as an opportunity to talk in detail about things that they would not have done otherwise. Some men commented that they had not discussed some things with anyone before. As such, although I was not there to offer therapy, knowing that I was a psychologist may have increased their confidence that they would not be judged and that I may understand any difficulties that they had experienced may have also helped facilitate them opening up. Therefore, a range of factors may have influenced the development of interview questions, how I asked questions, how men responded and how I analysed the data. This is an accepted part of qualitative research and is acknowledged in the discussion as both a strength and limitation.
6.8 Summary
The semi-structured interview study aimed to elucidate from men insights into the use of support, help seeking, and the factors that affect this. Thematic analysis followed the six phases of analysis detailed by Braun and Clark (2006). Analysis included the development of a coding framework and an inductive iterative approach. The results are detailed in Chapter 7.