3. Methodology
3.7 Analysis
I analysed my data by adapting Braun and Clarke’s (2006) approach to thematic analysis. In summary, this approach involves familiarisation with data, coding of data, identification of themes, reviewing themes, defining and naming themes, and producing a report. This process does not proceed linearly but is recursive, moving “back and forth as needed, throughout the phases” (Braun & Clarke, 2006, p. 86). There is some debate over whether thematic analysis should be regarded as an analytic method in its own right or whether it is merely a process common to a range of analytic methods, such as grounded theory (cf. Willig, 2013). Following Braun and Clarke (2006), I regard thematic analysis as method in its own right as long as the worldview informing the analysis is made explicit. Indeed, flexibility is one of the advantages of thematic analysis since it is not wedded to any particular worldview (Braun & Clarke, 2006). In this section I describe the worldview underpinning my research before providing a more detailed account of how I analysed my focus group and interview data.
My research is informed by a critical realist worldview. Critical realism has its roots in Roy Bhaskar’s philosophy of science and social science (e.g., Bhaskar, 1989). A critical realist perspective asserts that unlike the natural world, the social world is “not independent of human minds,” but it is nevertheless “independent of any particular human mind” (Gorski, 2013, p. 666). In other words, critical realism maintains that social reality exists independently of what a particular individual thinks about it while recognising that all “description[s] of that reality [are] mediated through the filters of language, meaning-making and social context” (C. Oliver, 2012, p. 374). Furthermore, critical realism proposes that social reality has both discursive and extra-discursive dimensions. Sims-Schouten, Riley, and Willig (2007) characterise the extra-discursive dimensions of social reality as embodied (e.g., physical health), material (e.g., availability of resources), and institutional (e.g., government policies), each of which can influence and constrain an individual’s discursive constructions. Critical realism is my preferred stance because, as Sims-Schouten et al. (2007) argue, it provides “an alternative both to naïve versions of realism and to totalizing versions of relativism” (p. 103). A critical realist stance acknowledges that accounts of reality are provisional, fallible, and contextually situated, but it does not accept the relativist assertion that all accounts of reality are equally valid (C. Oliver, 2012). Critical realism is also appropriate for my research because it is compatible with mixed methods research (Sayer, 2000) and with the application of
Saldana, 2014). Although I do not use mixed methods in this study, I outline directions for further research on the assumption that my results may be validated through additional research, including quantitative research.
The concept of a theme is crucial to thematic analysis. Adapting Braun and Clarke’s (2006, pp. 82–83) definition, I define a theme as a conceptually distinct form of response with respect to a research question. A theme may be represented by one or more participants. In relation to my research questions, a theme is therefore (1) a distinctive reason for assigning importance to scientific realism in fiction, (2) a distinctive way of evaluating unrealistic science in fiction aesthetically, and (3) a distinctive function of discourse about scientific realism in fiction.
I was primarily interested in participants’ self-reported responses in relation to each of my research questions. I share a basic assumption with the uses and gratifications approach to audience research, which is to say that audiences “are sufficiently self-aware to able to report their interests and motives in particular cases” (Katz, Blumler, & Gurevitch, 1973–1974, p. 511). This assumption is reflected in my focus group and interview guides (Appendix A) where my questions were often direct and transparent in purpose (e.g., explicitly asking why realism is personally important or unimportant). Having said this, I recognise that there is more to participant responses than what is explicitly self-reported, and where appropriate, I comment on the implicit meanings and strategies that I see at work in the focus groups and interviews.
Given that my third research question relates to the functions of discourse about scientific realism, it is worth clarifying that I performed a thematic analysis of self- reported functions of discourse about scientific realism rather than a discourse analysis of discourse about scientific realism. In other words, my thematic analysis should not be confused with discursive analysis as introduced by Potter and Wetherell (1987), Foucauldian discourse analysis (e.g., Parker, 1992), or other forms of discourse analysis. My approach resembles Potter and Wetherell’s approach to the extent I am interested in how “talk fulfils many functions and has varying effects” (p. 168). However, rather than analysing language use in participants’ discourse about scientific realism, I asked participants to self-report why they participated in discourse about scientific realism. I then organised their discursive motivations and purposes into distinctive themes.
My data analysis started during the transcription of my data. I used transcription as an opportunity to familiarise myself with the data. I also wrote analytic memos during
transcription, making a note of potential codes and themes and identifying questions that warranted further analysis. After transcription of my focus group discussions, I identified two main areas where I wanted to collect more data. First, I recognised that for many participants, unrealistic science was more problematic when it occurred in a narrative that was “serious” about realism. I wanted to find out more about what makes a narrative “serious” about realism. Second, I recognised that I needed more data about participant motivations for discussing scientific realism in a social context. As a result, I conducted interviews to further investigate these areas. After the interviews were complete, I familiarised myself with the data through transcription and recorded my early thoughts in memos in the same way as I had for the focus groups.
In the next stage of my analysis, I coded my data with the support of NVivo for Mac (Version 10.2.2). For me, Nvivo is a convenient way to code and re-code data. It is also a convenient way to organise and review coded data. However, beyond these conveniences, Nvivo was incidental to my analysis, and I do not expect my final analysis would differ if conducted with pencil and paper. I make a note of Nvivo here only in the interest of transparency.
As a point of terminological clarification, I use the word “code” to mean “a word or short phrase that symbolically assigns a summative, salient, or essence-capturing, and/or evocative attribute for a portion of language-based or visual-data” (Saldana, 2013, p. 3). In my thematic analysis, a theme is usually a category or synthesis of multiple codes, though as I noted above, a theme can also be represented by a single example. The word “code” is also a verb. Portions of text can be assigned to (coded at) a higher-level theme or a lower-level code.
I commenced the coding process with an initial cycle of eclectic coding (Saldana, 2013, pp. 188–193). I read the transcripts line-by-line, coding words, sentences, or paragraphs that were relevant to my research questions. Some of my initial codes were informed by my reading of the literature; other codes were identified in vivo. I also used structural coding (Saldana, 2013, pp. 84–87) to collect passages of text that related to a particular question before coding the passages more closely. After the initial cycle of coding, I organised codes together into preliminary themes and subthemes for each of my research questions. I then assessed the themes for conceptual distinctiveness and reviewed the data collected under each theme for consistency with the theme, revising
read each transcript to identify additional examples of each theme and, where necessary, re-coded existing examples in accordance with my current thematic scheme. In the final stage of my analysis, I refined the definitions of my themes and started producing a report of my results. Like Braun and Clarke (2006), I regard report writing as part of the analytic process. Indeed, during the process of writing up my results and relating my findings to the literature, I further refined my definitions of themes, sometimes introducing a new theme or consolidating existing themes at this stage. In cases where themes were significantly modified during the writing process, I re-read the transcripts again to check the adequacy of my thematic scheme and re-coded text as needed.
When presenting my results in this thesis, I provide an indication of the number of participants corresponding to a theme by using pronouns, such as several, various, and many. Since pronouns on their own can be ambiguous (Sandelowski, 2001), I provide operational definitions for these pronouns in Table 3.2. As I noted above, I regard a single conceptually distinct response as sufficient to constitute a theme. Furthermore, as I noted in Chapter 1, the open-ended nature of interviews and focus groups means that the absence of a response supporting a theme does not necessarily mean the theme is unsupported (Braun & Clarke, n.d., "Should I use numbers?"). The aim of my analysis is to understand the conceptually distinct themes relevant to my research questions rather than attempting to rigorously quantify the relative prevalence of each theme. Furthermore, the responses associated with a theme may range from a detailed example to a passing allusion. For these reasons, I do not regard precise counts of responses to be meaningful when reporting my results. Nevertheless, the number of participants associated with each theme does provide a tentative measure of each theme’s prevalence in my sample. Providing an indication of the number of participants corresponding to each theme also contributes to the transparency of my analysis, and transparency is a feature of good qualitative research (Tracy, 2010).
Table 3.2
Definitions of pronouns for describing participant numbers
Pronoun No. of participants
Several 3–5
Various 6–10
Many 11–27
Finally, it is worth noting that I did not verify the accuracy of the scientific claims made by the participants in this study. In some cases, participants discussed complex scenarios where it is difficult to make a definitive assessment of what is and what is not realistic. Furthermore, digressing into detailed explanations of scientific plausibility would distract from the focus of my thesis. I report on what participants perceive to be realistic or unrealistic and the justifications that they provide for their perceptions, but I do not attempt to establish what is actually realistic or unrealistic. Participant quotations should not be taken as a reliable source of scientific information.