PART 1: INNOVATION GOVERNANCE: THE POLITICS OF DISRUPTION
1.3 Theoretical and methodological approach
1.3.2 Methods and data
Quantitative analysis is well-suited to theory-testing of a large number of cases, but when we are dealing with uncharted territory, it is more advisable to go deep rather than wide, and provide more case-specific detail through narrative approaches. Because the research question guiding this investigation starts from a question of “how”, it increases the potential of the research to challenge existing assumptions about taken-for-granted causal relationships and relations of power (Foucault 1982). This leads to the development of more interesting insights and theory (Sandberg & Alvesson 2010). Asking “how” questions also leads one naturally into the exposition of narrative. In order to answer questions of how social life unfolds, it is necessary to investigate and interpret the “effects of social phenomena … in relation to process” (Flyvbjerg 2001, p.136). Process takes precedence over outcome. Events have to be put in proper order and the links from one to the next examined. Actors should be placed both in time and in relation to events (Abbott 2001d). Narrative is central to building understanding of complex cases through the “thick description” (Geertz 1973, p.6) of events as understood and experienced by the subjects of the study. As the study is concerned with novel and poorly understood phenomena, it is of the utmost importance that we make as few assumptions as possible and let the reality emerge out of focused attention to detailed practices and events. Only after this step has been completed do I turn to equally important questions of why, and bolster this account by drawing on qualitative computational methods to balance the experience-near immersion in the field with more
experience-distant analysis and interpretation (Geertz 1974). What this means in practice is that I begin with interviews to get a better sense of what is going on in the field, and follow up with content and network analysis of press articles to corroborate, supplement, and extend the exploration of the cases.
To focus on narrative and events as experienced by the subjects of the study seems a prudent choice when investigating complicated policymaking processes, especially in the EU: “In a polity characterized by contested institutional competencies, multiple, competing constituencies, and supranational actors vying for legitimate claims of representation, policy-making dynamics cannot be easily deduced from given political alignments or interest configurations” (Daviter 2009, p.1136). Much of the analysis is therefore informed by first-hand interpretation of primary interview data and documentary records. The data collection process commenced with desk research on the regulation of e-cigarettes and fracking in the EU. I primarily targeted the collection of documents on the EU institutions’ websites, where all policy documents are publicly available. Of specific interest were stakeholder consultations, scientific studies (both internal and external), impact assessments, expert opinions, press releases, memos, draft proposals, legislative texts, and fact sheets. Drawing on these and newspaper articles made it possible to identify a number of key interviewees who were approached with a similar set of open-ended questions. Interviews were primarily conducted in person, resorting to Skype and telephone if the interviewee was not in Brussels at the time of investigation. Rather than record the interviews, it was found more expedient to take notes in hand during the interview, allowing interviewees to speak more freely about these controversial issues. Several expressed their preference for this interview format, especially considering the sometimes sensitive contents of the discussion. In total, 51
interviews were carried out; 24 with e-cigarette policy actors and 25 with fracking actors, and a further two with people who had much experience on the European lobbying scene in general. An overview of the interviewees and their organizational affiliations can be found in Appendices 1, 2, and 3.
Interviews proceeded in the semi-structured format: a core set of questions provided a script to each interview, but the conversation was allowed to run its natural course to allow interviewees to bring up their own interpretation of events and draw attention to different details. This setup also made it easier to identify differences in how events were framed by participants. This interview style has been described as the “romantic” approach, where the purpose is to establish rapport with the interviewee in order to elicit confessional data (Alvesson 2003; Roulston 2010). The assumption is that this conversational style makes it possible for the interviewer to develop an understanding of the interviewees’ beliefs, attitudes, perspectives and opinions. Immediately following the conclusion of each interview, more detailed field memos were written to preserve the details of the interview and allow for some immediate reflection on the analytical relevance of the contents.
Interview quality was assured in three ways following recommendations in the literature (Roulston 2010, p.206): through the triangulation of data with other interviews and documents, by running multiple interviews with key interviewees in order to build the relationship and establish trust, and through the longevity of the fieldwork, which took place over 10 months in Brussels (September 2014-June 2015). The interview manual and core questions evolved as fieldwork progressed. New issues and perspectives were identified that warranted closer attention. New interviewees were identified by snowball sampling, in which key policy actors are identified by focusing
on those persons who tend to be nominated by most of their peers (Farquharson 2005). By looking at patterns in who was nominated by whom and comparing with the interview data, a distinct sense of the different camps and contending frames also emerged. Chapter 8 bolsters the account with a network analysis of frame usage in press statements, but the methodological details of the treatment will be presented within that chapter. As mentioned, the logic of employing a range of tools is that it helps boost explanatory power while producing a more complete picture of these novel and complicated cases.