• No results found

2. Conducting the studies

2.1. Study I

2.1.5. Coding and analysis

I began my analysis as soon as I had written out my notes from the first observation session and interview. Even though this initial data contained my observations of only a single doctor, there were still many different consultations and situations to examine. I started by going through the data sentence by sentence, comparing incidents and coding openly without restrictions in terms of any theme. In line with the Grounded Theory approach, I constantly asked of the data “what is happening here?” and “what category does this incident indicate?” In this way, I tried to not only interpret concrete actions or words, but to also conceptualise the meaning thereof.

Where I had written in my notes that the doctor addressed me before the patient entered the room, saying, “The next patient has some psychological issues, she is a bit peculiar”, I could, for instance, code this as “doctor classifying patient”. Going through the text, I coded and conceptualised incidents and compared incidents with other incidents, looking for similarities and differences between them. Later on in the text, where I had written that the doctor smiled after the patient had left, saying “That was a consultation, like, ‘Could I have a sick note if I ever need it, just in case’!” this was also coded as “classifying”, but now as “doctor classifying situation”. Thus, both similarities and differences between incidents were noted when constantly comparing them. The properties of each category also gradually accumulated during this analysis. For instance, when it came to the concept of “categorising”, I would note how some patients were labelled negatively and others positively. It is of course possible that I sometimes interpreted a gesture, action or phrase in a way other than what was intended by the participants, even if I strived not to. However, since the analysis rested on numerous situations with different doctors and different patients, an occasional misinterpretation was unlikely to affect the end result.

After spending some time on open coding, I found that the data generated fewer and fewer new codes. Most of the text was now being coded using my established

categories, which each had various properties attached to it and thus extended to cover a large number of incidents. This development took place gradually, and after having observed ten of the participating doctors, I switched to a more selective coding process. This process focused on what I perceived to be the most important codes in the analysis. On the basis of the identified categories and the relationship between them, we then established the core category, which is the one that appeared to be central to the emerging theory and could best explain what actually happened in the field. In this first study, we initially chose “medical reconstruction” as the core category, which was intended to describe how the doctors first deconstructed the patients’ problems, redefined the pieces and reconstructed them into a specific medical complaint. However, on conferring with Barney Glaser at an advanced Grounded Theory seminar, he suggested changing the core category to

“essentialising", which reflects how the doctors focused on what seemed to be essential to their specific objective while neglecting other dimensions. This concept was not only potentially more illuminating as the core of the emerging theory, but was also more general and applicable to similar processes outside the field of medicine. Accordingly, after choosing the core category, I delimited my coding to the other categories that seemed to be related to it and were the most important in the data. This left me with 20-30 conceptual codes. I went through the data again, this time focusing only on these particular categories, conducting selective coding to enable me to add cumulative properties thereto and discover the theoretical relationship between them. Later, these sub-categories were reduced to five main categories in order to delimit the theory. These were: “breaking down”;

“concretising”; “categorising”; “existential filtering”; and “functional focus”, and were in addition to the core category of essentialising. This theoretical structure was defining for the configuration of the first manuscript, in which each of these categories was explained and clarified with empirical examples.

While this analysis slowly advanced, I continued to observe and interview new doctors. This move between data collection and analysis was very valuable when it came to conceptualising the data. In particular, it enabled me to check my formulated categories and interpretations and develop the concepts that seemed to be the most valuable. It prevented excessive speculation or theorising and ensured that the analysis and emerging theory remained close to the data. The alternation between

gathering empirical data and theoretically analysing it also generated a lot of ideas and thoughts about: the data; why the doctors and patients acted like they did; how the different concepts and parts of the analysis were related; the different aspects of theory that I had studied earlier; and similar findings elsewhere, or even links to phenomena in completely different sections of society. I tried to capture these ideas in the form of the brief, explanatory notes that Grounded Theory describes as

“memos”. Writing memos is an important part of the method, and it is emphasised that when the researcher gets an idea about the data, he or she must always interrupt other work to write a memo about it. Memo writing stimulates researchers’ creativity, encourages theoretical analysis and exposes personal conceptualisations and interpretations of the material. These memos were thus added to the data and analysed and interpreted in relation to the rest of the information we had collected.

I used the NVivo 7 analysis tool, which is computer software for qualitative data analysis, to assist with the coding and systemising of categories and memos (Gibbs, 2002). The kind of software is designed to organise and analyse unstructured data, although there is some reluctance to use it in Grounded Theory circles due to fears that it takes the focus off the collected data and the scenarios observed by occupying the researcher’s mind with computer tasks. Furthermore, the manner in which the software is structured, the opportunities it provides to make notes and links, and its preset categories force researchers to organise data in a predetermined manner that is contrary to the aim of the Grounded Theory method. While I believe that these are fair reasons to be cautious about the use of data analysis software, it is my view that the advantages outweigh the risks. Some of the concerns are related to a lack of experience in using the software. Normally, however, the more you use a computer program, the more you will be able to master it as a tool for your own purposes instead of being led by its innate possibilities. For example, while writing on the computer in Microsoft Word might have obstructed my text when I was inexperienced in using the program, now it is just as easy (or even easier) to work in this way as opposed to writing by hand. In addition, there was a large advantage in using the software to keep track of my coding. When there are hundreds of codes, categories and memos, the need to ensure that they are maintained in an accessible system is considerable. So, my primary use of the program was as a way of keeping

track of my analysis and displaying the links between related situations and connected categories.