From codes to theory-building
By Pat Bazeley, PhD Use the code MSPACEQ323 for a 20% discount on Dr. Bazeley's books on qualitative and mixed methods data analysis.
Have you ever been stuck in a morass of codes or themes?
We’ve tended to become very good at coding or at identifying themes in qualitative data – sometimes too good, leading to the coding equivalent of the immobilising ‘I’ve got a thousand pages of transcript’ problem. If, along the way, you noted what each code’s job was, where it might take you in terms of your overall purpose and, critically, what its mates were, that will help enormously. Nevertheless, most of us don’t manage to do that consistently, and how to move forward from this point remains an issue for many students.
See a recent review.
As I analysed my own practice, I realised that the way I worked with coded qualitative data was similar, as a process, to what I’d done in an earlier life with survey data – data that also present you with assorted pieces of information that need to be wound together into a coordinated picture of ‘what is going on here?’ This experience was crystallised into a three-step strategic process through my work with students who had become caught in a coding trap as they learned to use qualitative software. The result is an approach that will have you climbing up and out of the mire. From a bundle of codes, initially working them one at a time, you will gradually move toward a comprehensive picture, a coherent story, or a soundly-argued theory, as told by your data.
Description is a good non-threatening place to take a first step. Select a code, mind-map it, write about it. Everything you know. Review the data you have coded there – what does it tell you? How distinct is it? What patterns or puzzles do you see as you write?
Now take the data for the code you are working with, and compare what has been said, done, or recorded by different demographic subgroups or at different times or in different places. With qualitative data these comparisons don’t stop with simple counts; sometimes the shape or content of the data within each comparative group reveals variations that give new understanding of the dimensions within that code, regardless of counts. Write about this, but you’re on a roll now, so don’t stop there.
In the third step, pursue your interrogation of the code with questions prompted by those comparisons: What is going on here? Why are these groups expressing this differently? I wonder if it has something to do with …? More mind-mapping. And, just as I did with my survey data, start exploring relationships between what is said or found here and what is happening elsewhere that might help to explain those variations. Now, you are not only gaining more insight into the code you started with, you are beginning to relate it to others – to build a web of connections across your data.
Some of the writing and drawing you do at each stage will contribute directly to your report, while the work you do at the third step will help you begin to sequence and structure your report as you ask: ‘What did I need to know about in order to understand this?’ It will also be what the reader needs to know to understand your conclusions and how they were developed.
Principles in Practice
Let me briefly illustrate these principles by creating, as I write, an example from a current project. In the context of evaluating the benefit for older women of regularly attending an exercise-focused ‘Wellness Centre’, I am also considering what wellbeing means for older women and how that relates to health. Drawing on interview data, I find the code for independence, for example, is best defined (described) as a capacity to care for oneself, to be in control of one’s life; it is clearly affected by mobility and, historically, by the experience of widowhood. Comparing what was said across values for a range of variables, but primarily health status, reveals some difference between those women who are used to being independent about how they conduct(ed) their lives, some of whom have now have either lost it or fear losing it, and those who really only gained a sense of independence much later in life (e.g., after being widowed, and/or when they joined the Wellness Centre). The former group record either high or low levels of health status, affected by their current level of independence. The latter group, more likely to be of middling health status, see threats to their independence as something to be managed rather than feared, and even though they relish their newfound independence, they are quite prepared to seek and receive help when needed.
Reading the qualitative text for the comparative categories thus suggests there can be a historically-based attitudinal difference, not evident on initial reading of the coded text as a whole, that points to further links that warrant exploration in relation to independence – and to maintaining health more generally – when an older woman is faced with problems of mobility or a health crisis. How do these attitudinal differences arise, and how do they come into play? Codes to explore include adaptability-resilience in particular, but also the sense of achievement and identity that comes from successful independent action; giving and receiving help; and the experience of and response to widowhood.
I referred to this earlier as working a code. It is an essentially practical working-through of possibilities that has you looking at your data in more detail and depth. Two further strategies lift you above the mechanics of this process. One is a reminder to write about what you are seeing as you work. Apart from keeping a useful record, writing prompts clear thinking and fresh ideas as you sift through what you have and how to say it. Then, after a such a period of immersion in working through a code or set of codes, you deserve a break! Take a long walk, or sleep on it! Waking up with a flash of insight circulating around in my brain is what gets me out of bed in the morning. Today, it was thinking about the possibility of ‘facing decline’ as a simple but useful integrating concept for this data – with dimensions that vary (such as independence, adaptability, acceptance of help) in response to prior and current experiences, with varying outcomes for health and wellbeing. Like all flashy ideas, it will need careful assessment against the data
More Methodspace Posts about Qualitative Data Analysis
Learn how inductive and deductive styles of reasoning are used to interpret qualitative research findings.