Chapter 6 Methodology
6.4 Data Collection: Teacher Practice
6.4.6 Processing and Analysing the Data
6.4.6.3 Analysing the Interview Data
As stated above, analysis and interpretation of the interview data, began during the transcription phase, in fact, keywords and themes had already emerged from the first two conversations. However, I was also anxious to avoid shutting down openness to the emergence of alternative or contradictory issues. While the semi- structured interview was guided by the pre-interview questionnaire, the data derived from the conversations added meaning to the quantitative responses already
collected. This interpretative phase was a key part of the research because it asked “…what does this mean…? (Willig, 2014, p. 137). Although the analysis began with the data handling and management through the transcription; bringing order to the
data to summarise and look for patterns was an on-going analytical process (Bathmaker, 2010).
The diversity of research designs and qualitative interview strategies invite a
corresponding diversity in strategies for conducting qualitative data analysis; framing these are the researcher’s theoretical assumptions and representational strategies. Different approaches, for example, hermeneutics, ethnographic, and narrative methods influence how interview data is analysed (Roulston, 2014). Generally, the type of analysis and interpretation carried out depends largely on the researcher’s ontological and epistemology positions (Willig, 2014; Braun and Clarke, 2006) and involves the use of theory and theorising (Bathmaker, 2010). Each analytical approach moves from description to explanation and possible theory generation (Cohen et al., 2017) and at each turn research subjectivity is inescapable.
Research analysed deductively begins with a certain system of rules and decides if the phenomena observed obeys that rule. Working inductively moves from the specific observation of a particular phenomenon towards developing a theory. Alternatively, abduction takes its starting point from the empirical data. The analysis goes on to explore the data to uncover new knowledge, this new knowledge is then tested in different contexts (Reichertz, 2014). Whereas retroduction involves making inferences from a description of a problem or phenomena, leading to an understanding of the casual properties producing it (Sayer, 2010). Each strategy has its strength and limitations (Cohen et al., 2017; Braun and Clarke, 2006). In reality, I sense that all of these approaches to data analysis occur at some point before, during and after the research process, what is important is that as the researcher, I make explicit how and why a particular path was taken.
The analysis of the qualitative data was conducted using a combination of thematic analysis and framework analysis. This maintained an open dialogue between me and the data through the search for meaning and alternative explanations. Thematic analysis can be flexibly applied within any theoretical position and works as a
method to reflect reality and unpick the surface of reality (Braun and Clarke, 2006). Although, critique of this method suggests that it is merely the first step in analysis before the real decisions about representation and interpretation are made (Willig, 2014), thematic analysis avoids the pitfalls of content analysis, where meaning and context are lost through extensive reduction and codification of the data (Vaismoradi et al., 2013).
A theme “captures something important about the data in relation to the research and represents some level of patterned response or meaning” (Braun and Clarke, 2006, p. 82). Themes can be abstract concepts formulated through the words, expressions and images that the interviewee reveals (Ryan and Bernard, 2003). During the analysis categories and themes were identified that not only described the issues but looked behind the text to uncover the latent meaning (Vaismoradi et al., 2013). The judgement about what constituted a theme was made after repeated reading of the entire interview data set and was a continuous, iterative process based on several factors. These included, representation within the theoretical framework, relevance to the research questions, prevalence within each data item and the entire data set, and the expressions and metaphors used by the interviewees (Ryan and Bernard, 2003) .
In my research, the data collection, transcribing and coding of each data interview took place concurrently throughout the fieldwork. Immersion in the data began whilst transcribing, with the stop-start, back and forth replay of the interviews. This
generated initial ideas for the structure of the coding, themes and categories. On first reading of the early interviews frequently used words and phrases were coded as keywords and labelled as nodes in NVivo, the entire interview data set was coded with this initial set, with similar key words grouped into categories. During the remaining data collection process new codes and themes emerged, these gave unexpected insights. Subsequently, once all interviews were completed, each transcript was re-read and recoded with the additional themes to ensure that, as much as possible, all of the features had been captured from the data. The semantic nodes (Table 15) were categorised to bring together data that reflected what
teachers did in the classroom, their teaching and learning activities, whole school issues and specifics about the science curriculum.
Table 15: Data analysis: Descriptive Semantic Codes
Key nodes Key Categories
Rigour Target grades Practicals Teaching hours Time Pressures Fun Flight path Whole school
Teaching and Learning Activities Curriculum Changes KS3 KS4 KS5 Primary CPD
Grouping by key stage was included as a node to help investigate areas where teaching strategies had changed for different students. The search tool option Matrix Intersection was used to construct tables of coding across groups to facilitate this (Woolley, 2009).
The concept-driven coding stage sought out data that explored the emergent latent themes and the a priori themes derived from the theoretical framework. The latent
themes included those related to ideas around fairness, the nature of science for all and the teacher’s sense of professionalism. The a priori themes focussed on
identifying where teachers had expressed their thoughts on the reforms relative to their past experience, future projections and present challenges to their decision- making (see Table 16).
Table 16: Data analysis: Latent Codes
Actions/interactions Concept – Agency
Decision making within rules - Discretion
Skills & Experience - Professionalism Sense of Fairness - Justice
Suitability for students – Justice
Past: Iterative Future: Projective
Present: Practical-evaluative Beliefs
Self-Efficacy
The latent themes were interpreted through the theory and literature to examine the participant’s conceptualisations of the current context, this brought an
understanding of how teachers used their agency and discretion in difficult circumstances.