Page 27 - Demo

1Introduction25In the analysis of the data, before analysing the visual and textual content, for the most part, I conducted an actor analysis to acquire familiarity with the actors active in the controversy studied. When analysing the gathered data for their visual and textual meanings, I followed the advice of scholars from various disciplines to note the textual context when analysing visuals (e.g., Bleiker, 2018; Doerr & Milman, 2014; Rodriguez & Dimitrova, 2011; Rose, 2016, p. 121). Hence, in all chapters, I applied textual coding for sentiments or tone of voice (i.e., positive, negative, neutral/balanced). In Chapters 2, 3 and 4, I first coded the textual content and then the visual content. In Chapters 3 and 4, I also coded the text separately from the visuals for storylines and frames of the technology studied. Textual codes were defined according to the specific research question which I was answering. In Chapter 5, I applied a multimodal approach and coded the visualisations together with the text for image–text storylines. For Chapters 3, 4 and 5, I further developed existing code books based on inductive coding of samples of the data. For visual analysis, I inductively developed code books for each chapter.To guard the rigour and relevance of the analysis and to prevent biases, I took several measures. When using digital tools and wherever possible, I followed recommendations to mitigate bias characteristics of the tools (for example, recommendations on how to avoid getting personalised Google results, Rogers, 2019, p. 33). I discussed the code books and the results of the interpretative analyses in routine meetings with my co-authors and made sure that we agreed on them. I also organised coding-in-action sessions and sessions of reflection on the results with various groups of researchers and other participants, such as members of the Public Administration and Policy Group, the consortium members of the project this PhD dissertation is part of ‘Travelling of Framed Facts and Uncertainties’ (TOFU), the international valorisation panel we launched for this project, other scholars involved in the study of policy controversies and stakeholders specific to a technology (e.g., invitees from the Dutch food information centre, Voedingscentrum). Table 1.2 summarises the method used for each chapter.The abovementioned research protocol of first conducting a stakeholder analysis, then scraping websites and Twitter, and analysing for storylines, frames and sentiments was specified for each empirical chapter. In Chapter 2, which takes a digital methods approach, a variety of digital tools are used and significant effort is put into grounding them by linking them with the happening in the studied Efrat.indd 25 19-09-2023 09:47