Method of Analysis

that stand out as significant instead of applying a predefined theoretical framework. This means that we take an inductive approach to determine what characterises organ donation as a social practice. While we attempt to approach the data with an open mind, we acknowledge that the interview guide and the responses we collect influence how we as researchers perceive and code our data (Kvale, 1996; Rubin & Rubin, 1995 in Saldaña, 2009). Thus, our interaction with the data and the analysis to some degree reflect the theories, concepts and language that formed this thesis in the first place (Merriam, 1998, in Saldaña, 2009). This includes our preliminary research and as well as what we ourselves have seen in the media and experienced elsewhere.

3.7.1. Data Coding and Processing

When you go to the theatre, you do not expect to sit through several hours of rehearsal and final preparations for the show. You arrive just before show starts, maybe a bit earlier to enjoy a glass of wine, sit back and enjoy what others have worked hard to create. Keeping in line with this analogy, we would however like to invite you backstage to see how we have coded and processed our data.

The data processing in this thesis starts already when transcribing the interviews, as we note any ideas for codes in brackets in the transcript itself. We do this to remember the idea later and for the other researcher to see it when reading the transcript. In case of long and complex ideas, we write an analytical memo; a tool that is used throughout the development of this thesis to capture ideas and thoughts. These ideas and thoughts can be on the topic, the interviews, the coding, the combining of codes as well as themes and patterns.

In the coding process, we use what Denzin (1978) calls investigator triangulation in order to capture as many different aspects of the data as possible (in Berg & Lune, 2013). This means that we code the data separately and combine our codes continuously in order to merge identical codes, discuss those that differ and be inspired by the ones that vary. We code the data as part of the interpretation process and due to the amount of data we use a coding software, Nvivo, to keep track of and organise the concepts and categories that emerge. First and foremost, we code for the attributes gender, age, occupation, religious beliefs, type of consent, type of registration and years of being an organ donor. These attributes are displayed in table 2 in this methodology chapter.

As a consequence of the grounded theory methodology, our first cycle coding includes a wide variety of coding methods, which we describe using Saldaña's (2009) terminology. Our first cycle

coding includes initial coding that captures our first impressions of the data. Examples of these codes are assumptions, deliberate ignorance and organs as things. Our first cycle coding also consists of structural coding for topics such as changes in consent, trust and values that we have asked about in the interviews. We also code for topics such as presumed consent and financial incentives as these are topics that the participants often have strong opinions about. Topics that we had not inquired about are also coded, which is called descriptive coding in Saldaña's (2009) terminology. These descriptive codes include consideration for the family, me and my body and voluntariness.

Emotions and values are extremely central in organ donation: “one can't separate emotion from action; they are part of the same flow of events, one leading into the other” (Corbin & Strauss, 2008, p. 7). As a consequence, we employ both emotion coding and value coding. By coding for emotions, we obtain insight into the participants’ world views, which is part of phenomenology.

The emotion codes include blame, gratitude and good feelings. As already mentioned, we code for values as part of our structural coding as this is something we directly ask about however in the value coding, we go more into detail and develop sub-codes such as helping others, motivation and reciprocity. Given the large variety of coding and consequently the large number of codes, we also conduct simultaneous coding to acknowledge and encompass the fact that some sections of data are relevant for more than one code.

Thanks to our thorough first cycle coding, we had little preparation prior to the second cycle coding. We had already merged similar codes, renamed codes with more appropriate names and generally discussed the utility of the codes. We had to some extent also done meta-coding by establishing patterns of themes, so there was little pattern coding left to do. During the first cycle coding and our work with the codes, categories had appeared to us and all of these were noted down in analytical memos that we would return to later. Our second cycle coding therefore mainly consisted of coding for patterns that had not already revealed themselves to us. This was done on a large whiteboard as we deemed it would grant us the best overview of our existing codes, categories, themes and patterns. At the end of this second cycle coding, we transferred what was on the whiteboard to a mind map using mindmup.com, which allows for real time collaborative work. This process increased our understanding of the categories and how they relate, which facilitated structuring and restructuring of the analysis prior to and during the writing of the analysis.