In its execution, the thematic analysis of qualitative data requires the iterative sifting through and coding of the data, and the finding of correlative thoughts, ideas and/or patterns therein in order to think more deeply about the same, recognise emerging themes and, in doing so, create meaning with which to fulfil the research aims.
Initially, I carried out this sifting and coding using the qualitative data analysis software NVivo (QSR International, 2012). However, in bottom-up coding, the act
116 of drawing ideas from a body of data, a text, coding those ideas and setting them down within any sort of software system with a view to grouping (or splitting) these codes into themes, in effect creates an entirely new body of data, a new text.
As the exercise progressed I came to feel that not only was this new text entirely too different in nature from the original transcription text but, as it expanded with the addition of each code/node (adding new free nodes, attaching branches to tree nodes, or creating new case nodes, for example), this new text became further and further estranged from the original text.
In trying to shape my own meaning from the original text, I felt I was destroying it; I was tearing it apart; I was not generating a fresh shape of text, an adaptation or rearrangement into some form of sympathetic sister text that better revealed the original's patterns and themes; I felt I was tearing bits of the text away in order to create something else, something entirely different. Not only that, I realised that upon intending to look to this new and different text to find meaning from the original, I must also intend to abandon that original.
Although any such new texts, formed from words, phrases, sentences or parts of sentences from an original text set side by side one another or set in some form of hierarchy, might render a 'thematic'-text or 'hierarchy of themes'-text wholly adequate for research founded upon other theoretical frameworks, it was this very same fragmentation and reorganisation, this tearing of words and phrases from their delicate context within something already fully formed and 'perfect', in order to assemble some sort of alternative artificial construct, that I felt was becoming more and more anathema to the fabric and the essence of the reader response framework within which the research was embedded.
Reader response theory argues that a text is called into being by the reader; and the reader, calling into being this unique and irreplicable entity, creates meaning from this transaction.
In order to satisfy the research aims, it was my intention to call the transcription text into being and thereupon create some sort of meaning. I found, however, that through the process of NVivo coding, I was calling into being – as an author – this entirely new text and was transacting with this new text as a way of making meaning from the original.
117 This new constructed text, although representative of something in itself (a
collection of ideas; a progressive/ranked organisation of patterns), felt like an artefact; it was of the original, but not the original. It was a by-product of the original and it represented my re/constructed role as 'author/reader' instead of just 'reader'.
As both author and reader of the new text, I began to question whether, when looking to that new text for the making of meaning, I had a right to claim the making of meaning from the original text.
Respecting the Text
Although I have to, of course, become 'author' in that I have to write down my findings and formed opinions – in this very paragraph, for instance – in order to relay my research to others, when looking to fulfil the objectives and aims of the research, I felt that I had to listen more directly to what the children had to say; I needed to make meaning from what the children actually said within the context in which they said it. That is, I felt I had to stay true to the children as being the authors of their texts and me as being, only, the reader.
These three issues: the creation of an artefact, my re/constructed role as author of that artefact, and the abandonment of the original text, I came to feel presented a phenomenon that could not claim any position within the way of thinking or way of being that is reader response theory – and especially reader response theory within the further context of this particular piece of research: that is, this research does not use the 'lens' of reader response theory to think about how the children feel about science and scientists and what the children have to say about the same; rather, this research is founded upon reader response theory and regards the looking upon of any text as a unique transaction in the temporal creation of the text itself and in the temporal creation of meaning. The creation of any new text is an entirely different entity in and of itself and, thinking about those fundamental convictions of reader response theory, such an artefact cannot represent the text from which was compiled. Therefore, I decided to put NVivo to one side and begin again.
The research methodology and design had incorporated Braun and Clark's "foundational" (2006, p. 78) method of thematic analysis and their inductive approach toward the identification or discovery of themes. They suggested that
118 although themes and patterns could be identified at a semantic surface level, it was upon looking beneath these surface patterns, that one might come to see the themes that shape the semantic content (p. 84).
Their inductive approach advocated looking at the data from a bottom up point of view, in as much as one might let the data speak for itself as, in iterative
transactions with the data and coding of the same, themes might gradually be revealed. This is as opposed to a top down approach wherein a researcher might approach the data with a theory or question in mind and actively seek out data that appealed to the theory or that 'answered' the question.
Having eschewed NVivo's (bottom up) artefact, however, I decided to do both, that is, both an inductive bottom up and a deductive top down coding and analysis of the data – which is entirely in keeping with the flexibility that thematic analysis, as a method of thematic analysis in itself, allows.