Structuring Empirical Insights
The qualitative data that I gathered through field observation and interviewing had to be processed—sorted, stored, carefully transcribed, analyzed, interpreted. This section describes how I handled the gathered data in the process of my analysis that enabled me to interpret observation and interview data as empirical insights. On the one hand, my data analysis needed to reflect my philosophical interests; on the other hand, I felt that I needed to do justice to the depth and abundance of the empirical data that I was able to gather. Therefore, I have chosen to code large parts of my data, that is, some of the field notes that I took during group meetings and shadowing phases, and all interview transcripts.
Coding is a method of analysis that treats empirical data as text and involves an analysis process in the course of which the data text is segmented and reorganized. Being a term adapted from grounded theory, ‘codes’ are “[...] conceptual labels placed on discrete happenings, events, and other instances of phenomena" (Strauss & Corbin, 1990, p. 61; see also Alexa & Zuell, 2000, p. 306). To code means to index text passages with descriptive or analytic categories that relate, if vaguely at first, to the research interest. Indexing text passages allows the researcher to familiarize him or herself with small data fragments. By breaking up a body of text into manageable segments, describing those segments, unfolding the information they convey, comparing and relating them, the researcher reorganizes the data with regard to the research focus and is able to interpret them as a response to the research questions (cf. Coffey & Atkinson, 1996).
The formulation of codes can be guided by the intention to describe data from the bottom-up, that is, with the conceptual resources inherent in them, or by the intention to relate them to theory, that is, to the repertoire of abstract concepts offered by the discourse in which a piece of research is to be embedded. An example of a strongly theory-guided code in my analysis would be “testimony." Labeling different text passages with this code, however, did not help me substantially in organizing my data material, because the label “testimony" did not correspond very well with the data that I had. I realized that the acts of informational exchange that I was looking for were embedded in socio-epistemic relationships that were better described as “dependence relations." Therefore, I started to code for “dependence" instead. This led me to distinguish two different forms of epistemic dependence (see Chap. 7). An example of a code that was, at least in the beginning, rather unconnected to my theoretical framework would be “frustration." While shadowing a PhD student, frustration was a recurrent theme in his conversations with me. Clearly, this was related to the high pressure to succeed, his anxiety about failure and his understanding that he had been given a high-risk project as his dissertation topic. His frustration, however, was also related to his work conditions. He perceived his research group to be only a “little interactive," a perception in sharp contrast to my observation that led me to reflect on my biases as an observer and to reconsider the interplay between deliberation and delegation (see Chaps. 5 and 6).
For my coding process, I did not formulate a template of codes prior to the analyzing stage. When I had transcribed the data gained from observing or interviewing (which forced me to go through my audio records at a painstakingly slow pace), I began applying ad hoc codes to semantic units in the text which seemed relevant to my research interest. I have used both rather descriptive codes, which did not appear to be theoretically relevant at first glance, and more interpretive, theory- inspired codes to label and organize text passages (cf. Strauss, 1987, p. 33f; King & Horrocks, 2010, p. 153). At certain stages, I put existing codes into a systematized order. This helped me to envision new, complementary codes to match the existing ones. In between coding cycles, I went back to the data in their raw, unprocessed form. Such repeated phases of unstructured immersion ensured that I did not lose touch with the original data. After I had skimmed through scribbled field notes again, and relistened to the original audio files, I wrote encompassing case descriptions and composed characterizations ofsingle interviews. In part, these descriptions form the basis of the portrayal of selected interviewees in Chaps. 4 and 5.
From coding and less structured immersion phases, I proceeded to the formulation of broader, more general categories which I call “themes." Thematic analysis is a generic technique ofanalysis in qualitative research (Attride-Stirling, 2001; Boyatzis, 1998; Charmaz, 2000). Themes ideally resonate both with the investigator’s interest and with the concerns of the people studied as expressed in the gathered data. In general, a theme “[...] captures something important about the data in relation to the research question, and represents some level of patterned response or meaning within the data set" (Braun & Clarke, 2006, p. 82). Whereas codes pertain to the micro-level of analysis, the focus on themes enables the analyst to draw a bigger picture. From iterative cycles of coding that were paralleled by my study of the literature from social epistemology and philosophy of science, two codes, “trust" and “epistemic dependence," emerged as broader themes in my analysis. These themes later became the basis for the discussions outlined in Chaps. 7, 8 and 9.
I understand the analysis that I have conducted as moderately theory- directed “editing" (Crabtree & Miller, 1992, p. 18), a form of qualitative analysis that is little formalized and has allowed me to flexibly follow various interpretive trails that connect empirical insights with philosophical concepts. Iterations of sequential, often theory-guided, coding helped me to mediate between philosophical interests and the existing abstract conceptual repertoire on the one hand, and the concrete empirical observations that I made in the fieldwork and interviews on the other hand.
-  For a critique of thematic analysis, questioning the representational status of themes, see Gomm(2004, p. 196).