The most popular post on this blog is this one, about grounded theory and the emergence of themes. The post where I used Queen lyrics as an example of coding processes is also one of the most read. So, I thought I’d write about thematic analysis as it’s another one of the methods I use often in my research. I often find qualitative methods texts difficult to follow when they talk about the processes for these types of analysis, and find it much easier if there is an example to follow, so that’s what I’m going to do here. As writing about this whole process could get quite lengthy, I’m going to break this up into a series of posts. This is the first, which will introduce what I’m doing. So first things first, we need some data to work with.
The reason I chose this particular video to work with is because it’s relatively short, and because it is an arranged interview. Previously when I’ve tried out analysis methods to see what they are like I’ve used film scripts as practice data, but I find using actual interviews more useful as it’s not a scripted conversation, but still retains a constructed setting which is similar to what you will get in research interviews. Having natural speech is useful; scripted speech doesn’t tend to have all the “erms”, and pauses, and tripping over words or backtracking that natural speech does, and these things can be recognised by the analysis, so it’s useful to have them in the data. I tend to use a constructionist approach to my work, and so recognising that an interview isn’t a natural setting is important. Interviewees are aware that their answers are going to be analysed and considered in some detail, and this might change what they say or how they say things. Nirvana in particular are a good example of this, in interviews it seems that they are generally uncomfortable with the situation and/or the questions being asked of them, and so don’t play along. Also, I really like Nirvana, and happen to find this particular interview interesting and funny.
The first step for a qualitative analysis is often to transcribe the data. This isn’t always necessary, but I like to have transcriptions for a number of reasons. Firstly, because you really get to know the data through the transcription process. Secondly, it means you have a paper trail of exactly what you worked with, and can show where each code came from (this is also true if you use NVivo). Thirdly, I can send participants a copy of their interview transcript which they can check, clarify any points if they wish, or ask to have certain things removed (important for data protection/retaining anonymity). Finally, I prefer to read information rather than listen to it, so a written transcript is what I like to use. You’ll be spending a lot of time with your data, so knowing what formats work for you is helpful.
Transcription can be a long process, that 15ish minute interview might take me up to an hour to transcribe, and that is just to do a verbatim transcription, not including timing of pauses, marking intonation etc. For this example I’ve transcribed from 1 minute 28 seconds in the video to 3 minutes 54. I’ll use this transcript in the second post to show how I go about the coding process. I’ll still consider the rest of the interview, but for me this takes a slightly different approach. So Part 2 will appear some time next week, and will look at how I code data.
If you want to, you could analyse the same data as me over the next few weeks, and I’ll do a post where everyone’s findings can be compared. If you’re interested in doing that, leave me a comment/message/tweet!