SAFE acronym
3.11 Data analysis
Miles and Huberman (1994) estimate that approximately two to five times as much time is needed to process and order qualitative data, as the time that was needed to collect it. They explain that the best defence against data ‘overload’ is to have a conceptual framework and research questions in order for the analysis to be a selective process. It is also recommended that the researcher remains alert to the purposes of the research and the:
‘‘…conceptual lenses you are training on it’’ (Miles and Huberman,
1994, p.56).
However, the researcher must also remain receptive to unexpected information.
In this study, my ‘conceptual lenses’ were made very explicit by the process of formulating an Initial Programme Specification based upon the literature review, as part of a Realistic Evaluation approach. Indeed, a template analysis of the data was initially considered, so specific was the evidence I was seeking. However, as Miles and Huberman (1994) highlight, I wished to remain open to new and unexpected contexts, mechanisms and outcomes that the data presented, rather than to impose on the data a set of already identified categories (although I did have the ‘templates’ of contexts, mechanisms and outcomes in mind). I therefore decided to use thematic analysis to offer a framework and a guide to my analysis of the data, with the
‘conceptual lenses’ of the Initial Programme Specification kept in mind and
contributing to a ‘theoretical sensitivity’ to the meanings within the data (Glaser and Strauss, 1967). I therefore looked for themes that were contexts, mechanisms, and outcomes, and coded these within the data.
This thematic analysis can be described as ‘theoretical’ thematic analysis because it is driven by my theoretical interest in the research area, rather than being an
‘inductive’ or ‘bottom up’ analysis which is data-driven (Braun and Clarke, 2006). This kind of thematic analysis is more ‘analyst-driven’ and in my case involved coding for specific research questions which mapped onto a theoretical approach.
I decided that for the purposes of thematic analysis I needed all of the data to be in written form in order to be coded, but I did not feel that formal transcription of the audio data was needed. I therefore transcribed the data in an informal way which was fit for purpose, and did not use any transcription conventions because discourse analysis of the data was not needed. In the two interviews where there had not been consent to audio-record, I used my written records of the interviews and coded these directly. The observation data was in the form of ‘field notes’ and the field notes were coded directly, rather than being written up. I did not feel this was necessary, and also found that the way in which I had written my field notes also offered me clues regarding the context in which I had written them, and the order in which I had noticed aspects of the environment. Examples of coded data are included in Appendix I (page 214).
Miles and Huberman (1994) explain that coding involves differentiating and combining the data, and making reflections on it by assigning labels to units of meaning in the data. Miles and Huberman (1994) recommend condensation and analysis after each wave of data collection, so that a process of iterative reflection is ongoing throughout the data collection process. This process of iterative reflection is very much in line with the Realistic Evaluation approach. I took this advice, and after each phase of the research, I began to code interesting aspects of the data. For example; the initial set of contexts, mechanisms and outcomes that I identified after a first attempt at coding Day 1 of observations is included in Appendix G (page 205).
As my analysis progressed, and I became more knowledgeable about the school and the stakeholders, the codes were further developed, more interpretation was
involved, and patterns in the data began to emerge and to be noticed and recorded. It was at this stage that the codes began to represent ‘themes’. As Braun and Clarke (2006) explain:
“A theme captures something important about the data in relation to the research question, and represents some kind of patterned response or meaning within the data set” (p.10).
However, Braun and Clarke (2006) suggest that the greater the number of instances of the code within the data set do not necessarily indicate that the theme is more crucial, and that therefore ‘‘researcher judgement’’ is needed to decide upon themes (p.82). This can be driven by the particular analytic question that the researcher begins with, or in Miles and Huberman’s (1994) words, the ‘conceptual lenses’ the researcher is wearing.
I decided to give an indication of the source of the themes, and the prevalence of the themes in my reporting of the data, in line with the practice of Humphrey et al (2009). However, this was not provided in order to give a quantitative indicator of the
relevance of each theme, but rather to increase the transparency of my analytical procedure, and to improve the validity and credibility of the findings through demonstrating that triangulation of evidence had occurred (as Yin, 2009,
(2006), that the occurrence of a theme does not necessarily demonstrate its importance.
The process of thematic analysis that I followed is described by Braun and Clarke (2006) in Table 10 below.
Table 10: Phases of thematic analysis. Taken from Braun and Clarke, 2006; p. 87.
Phase Description of the process
1. Familiarising
yourself with your data:
Transcribing data (if necessary), reading and re-reading the data, noting down initial ideas.
2. Generating initial codes:
Coding interesting features of the data in a systematic fashion across the whole data set, collating data relevant to each code.
3. Searching for themes:
Collating codes into potential themes, gathering all data relevant to each potential theme.
4. Reviewing themes: Checking if the themes work in relation to the coded extracts (Level 1) and the entire data set (Level 2), generating a thematic ‘map’ of the analysis.
5. Defining and naming themes:
Ongoing analysis to refine the specifics of each theme, and the overall story the analysis tells, generating clear definitions and names for each theme.
6. Producing the report:
The final opportunity for analysis. Selection of vivid, compelling extract examples, final analysis of selected extracts, relating back of the analysis to the research question and literature, producing a scholarly report of the analysis.
Braun and Clarke (2006) explain that thematic analysis involves a;
“…constant moving back and forward between the data set, the coded extracts of data that you are analysing, and the analysis of data that you are producing” (p.15).
This is very relevant to my analysis, which involved an ongoing state of to-ing and fro-ing from the themes to the data, as the data set increased, and the depth of analysis progressed. It is also emphasised that there is a recursive process involved within these 6 phases of thematic analysis, and there is movement between phases. This was certainly the case within my research which took place over a 6 month period, and therefore involved a constant re-familiarisation with the data, reviewing and re-defining of the themes. There was also a re-generation of codes following the realist interviews, which added new information and led to a redefinition of some of the themes. An example of the feedback given in a realist interview is given in Appendix H (page 211).