Table 2: Comparative Analysis of Peer Learning Environments Elements Peer
3.8 Thematic Analysis
‘Thematic analysis is a method for identifying, analyzing and reporting patterns (themes) within data’ (Braun and Clark, 2006:79), and is viewed as a foundational method for qualitative analysis. How the data is coded is an important part of the research process. Inductive analysis was proposed, with no prior coding frame in place with the thematic analysis being data driven. However Braun and Clark (2006:84) note that ‘researchers cannot free themselves of their theoretical and epistemological commitments, and data are not coded in an epistemological vacuum’.
Another issue relates to the level at which themes are identified. Two levels are noted: the semantic level or the latent or interpretative level. Semantic analysis looks at the surface meaning and no further, while analysis at the latent or interpretative level examines underlying ideas. Latent thematic analysis is used where the development of the themes involves interpretative work. Braun and Clark (2013) note that latent thematic development comes from a constructivist paradigm in line with my epistemological values.
89 3.8.1 Phases of Thematic Analysis
According to Braun and Clark (2013:121) there are six phases in thematic analysis:
1. Familiarisation with the data. 2. Coding
3. Searching for Themes 4. Reviewing Themes
5. Defining and Naming Themes 6. Writing-up
In phase one of analysis according to Braun and Clark (2006) the researcher must immerse themselves in the data, becoming extremely familiar with all aspects of the data and note that it is important to read through the entire data set before beginning coding. Taking notes from phase one is important ‘as that phase provides the bedrock for the rest of the analysis’ (Braun and Clark, 2006:87). The transcription of verbal data, while time consuming and slow, is, according to Braun and Clark (2006), a good way to become familiar with the data. Braun and Clark (2006) add that as there is ‘no one way to conduct thematic analysis, there is no one set of guidelines to follow when producing a transcript…what is important is that the transcript retains the information you need’(pg.88). Braun and Clark (2006) note that the time spent on transcription is not wasted ‘as it informs the early stages of analysis’ (pg. 88).The 20 hours of audio video transcription analysis presented a real challenge in this study. The literature (Cotton et al, 2010) states that for each hour of recording there is at least four hours of analysis. In the context of this study, I was much slower than Cotton et al (2010) proposed, perhaps because I was new to this analytical approach.
In order to familiarise myself with the data in phase one I transcribed all 10 sessions (20 hours of data, resulting in approximately 250 pages of interactions). This took 6 weeks, possibly due to the number of contributors and the speed at
90 which they spoke. I found myself continually stopping and starting to check less audible comments. Initially I transcribed all the recordings in chronological order concentrating on what was said indicating whether it was me or a student comment. This was painstakingly slow but the idea was that I would be able to read the transcripts like a book to familiarize myself with the topic content. Next, I watched the recordings with the transcripts and identified each contributor by their name if I had not done so previously. With a holistic view of the ten weeks and a general overview of both the emerging topics and patterns of participation I felt I was at the first stage of gaining some level of insight about what was said and by whom (See appendix 5).
The second part of the initial phase of analysis was just as time consuming, requiring multiple viewings of the video diaries to familiarise myself with the other modes of communication. While the primacy of language in our culture is evident, and I now knew what was said and by whom, I also understood that the other non-verbal modes of interaction also formed an important part of the representation of the data in context. According to Norris (2004) observing proxemics, the way in which individuals use their space, ‘gives insight into the kind of social interaction that is going on’ (pg.19). Open and closed posture of arms, legs torso and head and directional positioning give some indications to the level of engagement or dis-engagement of an individual. Gestures, according to Norris (2004), include pointing at someone or following a finger while head movements may have clear meaning for example nodding yes or no and gaze is associated with ‘organisation, direction and intensity of looking’ (pg. 36). These are discussed further in the findings. In order to gain a deeper understanding of the non-verbal indicators I once again re-viewed the transcripts. Initially I looked at the tutor over the ten weeks, then the students and noted head movements, eye contact, laughter and other interesting non-verbal incidences. I did this multiple times, often re-playing a certain section over and over until I was comfortable with my interpretation of it. At the end I replayed the entire ten weeks and watched it like a film in order to get a holistic sense of this situation in context.
91 Consequently phase one, including transcription of the recordings and familiarisation with the data, and multiple viewing of the diaries for the other modes of communication and interaction, took considerable time to complete prior to phase two, coding.
Phase two to five, coding, searching for themes, reviewing themes and naming themes appear linear in approach, yet, were in fact, repetitive, iterative, cyclical and time consuming. Phase two, according to Braun and Clark (2006), is about producing initial codes from the data. These are items which are interesting to the researcher in regard to the phenomenon under study. Using the inductive approach and coding every data item initially is a cycle of looking, seeing and listening to the videos while simultaneously reading and trying to interpret what is happening in context. Pink (2007) states that analysing video is not simple and ‘the ambiguity of visual images and the subjectivity of their producers…give subjective meaning to their content and form’ (pg. 117- 119). Pink (2007) notes that, in analyzing video, it is not simply evidence of conversations and actions but rather ‘images and words contextualizing each other, forming not a complete record of the research but a set of different representations and strands of it’ (pg.120). Visual methods are rarely used on their own and, according to Pink (2007), ‘should be analysed in relation to other research texts’ (pg.136). As an ethnographer, it is impossible to record a complete relationship visually: rather, a recording is a snapshot in time, making it important to reflect during analysis and write up on the contexts in which these recordings were produced in order for the reader to have a picture of the situation in context.
Phase three allows the researcher to analyse and consequently group the codes into broad themes. Reviewing themes is phase four. According to Braun and Clark (2006), the researcher refines the themes during this analysis by checking the data set to see if there is enough evidence to support each theme. Some
92 recoding may be necessary as this is a recursive process, resulting in a refined thematic map.
In phase two I divided my themes into verbal (the topics) and non-verbal (the social interactions). I analysed the emerging topics first. I once again read through the entire transcripts and noted what appeared to be the more obvious topics. Six general topics emerged these included; student behavior, teachers role, resources, school culture, the staff room and teaching as a career. I then allocated each topic a coloured index flag and re-read the transcripts sticking the index flag in the appropriate places; I did this one topic at a time. I also noted new things that I had not identified in previous readings.
In phase three and four I re-read each topic individually checking for sub-themes and to see if there was a deepening of understanding or of co-constructed meaning over the ten week period. I did this for each topic until I was satisfied that nothing new emerged. This took iterative cycles and was very time consuming but was invaluable in familiarising myself with the topics. I took the same approach with the non-verbal indicators and watched the recordings with the transcripts and noted the interactions in context. Again this took multiple viewings and was time intensive. On reflection phase two, three and four occurred simultaneously. This layered approach was invaluable for phase five and six when I started to write up the analysis.
Phase five and six according to Braun and Clark (2006) is writing; phase five of the analysis to write a detailed description of each theme and how it fits into the research story and phase six to write up the research process, creating a story that incorporates the analytical aspects and the data extracts into a cohesive story about the research study, and its place in the current literature of the subject area.
93 In using thematic analysis the researcher needs to ensure that the ‘data interpretation is consistent with the theoretical framework’ posited, that data interpretations and analytical points coincide with the data extracts, and finally that the researcher picks ‘compelling examples to demonstrate the themes’ (Braun and Clark, 2006:95).