This was my first time at CHIIR, and it was a really enjoyable experience; lovely people, great community spirit and the sessions were full of information and discussion. I started with the Tutorial on Sunday 10th March “Coding qualitative data: you asked them, now what to do with what they said” led by Dr Rebekah Willson (University of Strathclyde). There is a pleasure in being taught by a good teacher who enjoys their subject, even if the subject is not one of direct interest. As it happens, the subject for this tutorial was right on topic for me, right now, so a double pleasure. A really good session, which Dr Willson described as a “whirlwind tour”, but in fact gave space for us to work in pairs on an exercise, discuss and feedback. I’ve come away from that tutorial feeling more confident that I can code up the qualitative data I have collected so far in my studies.
We covered a step by step approach to coding qualitative data, bearing in mind the “paradigm shift in thinking” as one moves from quantitative to qualitative methods: we’re dealing with the human and that is messy, challenging, based on experiences and beliefs, and allows a broader, holistic understanding, albeit one that is constructionist, with the researcher involved in the research, giving multiple meanings, multiple interpretations. We are there, we are part of the process, so we have to think about the role we have and what we are doing. The result of qualitative data collection is richer data that is more difficult to interpret. We are asking “Why did they do/say that?” There are several approaches to coding, and so it is important to choose one and stick with it. There are challenges of qualitative research being in itself a learning process – it is messy, it is fun, and doing it shows you how to do it. It is normal to be confused and overwhelmed. That’s a helpful thought. Dr Willson chose to show us one route through, with a series of iterating steps, providing a robust and rigorous approach to analysing qualitative data. She reminded us that a negative/opposing result can often be the most useful and interesting thing to explore – why is that case different? It is about following where the data leads, and moving from the concrete to the abstract. Looking for similarities, grouping and classifying. She talked about the process feeling uncomfortable, which I find to be true – like wandering in a fog and occasionally glimpsing the light!
When we gather data for a qualitative study, we usually have a vast volume of material – for a example, transcribing an interview can give you 1000’s of words of material. Furthermore, when you ask open questions, the answers are unpredictable and often richer than you’d anticipated. This fits with what’s happening for me. Instead of asking “what is your job title?” and “what is your education?” in a recent survey, because of a limit on the number of questions – I combined the two into “Tell me a bit about yourself” and received back long essays that told me such a variety of things, and sparked so many questions that I had not thought to ask, around ideas that I now see are interesting to explore… Dr Willson said we must pay attention to anything that is potentially interesting, code it up and then refine our ideas – grouping, splitting up, asking new questions of the data, all the time moving from a broad view of the data to a deeper focus. Also, be rigorous and trustworthy – sharing how we code the data, what steps we took, taking an iterative approach, triangulating across data sources, including negative examples, making our codebook available, making our inclusion/exclusions available. The researcher must be trustworthy, and if more than one persons is coding – this is a good thing to check for consistency of interpretation, provided that there is inter-coder reliability; we need clear codes, clear reasons for using the codes, clear inclusion and exclusion criteria. This means we’ve moved from the initial coding exercise to a focused coding stage, using a code book. The coders code separately and then compare results.
Dr Willson described several methodologies for qualitative analysis, and explained that the choice of methology is affected by the research questions. The methodology she showed us in detail, and which we practised in the exercises is Thematic Analysis. She talked about two levels of engaging with the data: the SEMANTIC level where we look for and code things that are expliciit in the data, and the LATENT level where we look ideas and assumptions implicit in the data. We need to decide ahead of time which we do. In thinking about these levels, we start to realsie that what people say and what they do can be different – so field notes about behaviour become part of the data. As well as text, we might collect and analyse video, audio, images and so on. The steps in thematic analysis are:
- familiarise – read the text several times and take notes. Do it line by line!
- generate initial codes, get to know the data – again line by line.
- start to look for patterns in the codes, perhaps ways they group
- make themes of one or more codes – overarching ideas that cut across the codes.
- review the themes against the data… do they make sense?
- and do it again…
Defining and naming the themes provides the analytic power – think about what the thme can contribute. Themes can have subthemes, so there can be a hierarchy of themes, subthemes, categories, and codes. The code book has the full description of these, and each code and theme has a single word or short phrase descriptive name. Relate the codes and themes back to the research questions. As this process is worked through, the research questions might change – because we realise the data is pointing us in a new direction. We need durign research to constantly revist our questions, out data, our themes and codes t ensure we are following the data, asking the right questions, revisiting, enlarging and clarifying, all the time. Whether we start from a deductive approach (where we predefine the codes to support our idea and the questions we want to explore) or an inductive approach (where we explore the data, come up with codes and build to themes and questions) or move between the two – always we need to keep revisiting the data. Follow up, change the questions, revisit ideas, identify what is different, look for variations…
Later in the week, the conference dinner was at the Science Museum, and while there I noticed a mural/display that said “We are all scientists; we all observe, find reasons, look for relationships, categorise and make models” Unfortunately my photo of it is too blurry to share… but it summarised the tutorial and the week for me. Thank you, Dr Willson for a brilliant tutorial!