Getting started with coding

I'm not going to talk about what to code, or analytic strategies; there is plenty written about this (e.g., Braun and Clarke's textbook, Thematic Analysis, or Smith, Flowers and Larkin's Interpretative Phenomenological Analysis text). I want to focus on practical tips that will help you produce coded transcripts that are going to provide the strongest foundation for the later parts of your analysis.

Create more codes than you think you need

Most commonly, I see that there isn't enough coding on a transcript. There are probably multiple reasons: lack of confidence, time limitations (that looming assignment deadline), and also not having access to an example of how a noted/coded transcript should look. Perhaps the tyranny of a deadline and the need to meet assignment guidelines lead to a focus on obviously relevant data to avoid wasted work on irrelevancies. As a side note, I've also noticed that students often under-cite in their academic writing even while thinking they've cited too much. Of course, you can look at a published paper to see how often experts cite, but there are few opportunities to look at real examples of coding.

You're probably underestimating how coding is useful to your analytic process and how data can be informative even if it doesn't obviously answer your research question. I use my research question as a guide when I'm coding (I have it on a sticky on my computer screen and look at my data through the 'lens' of that question), but I try to code as much as possible to see what's going on. Sometimes my codes are questions to myself about what might be happening, and that's fine, too. If you can't decide how to code something, it's fine to pass over it. You may come back to it later (or not - sometimes pieces of data aren't relevant to our research aims). Do as much as you can initially and remember that this, like many elements of qualitative research, is cyclical, and you may well have new ideas to apply later in the process, so return to your transcripts as your analysis progresses.

You may very well find parts of your data that do relate to your research question, but not in obvious ways. You miss the opportunity for that if you don't code your data thoroughly enough.

Broad brush coding

I also often see that a transcript is coded very broadly, for example, that most of the codes are of entire paragraphs. Codes can cover a paragraph, but I wouldn't do this too often. They can also relate to a word or two, or a phrase or sentence. If you rely on coding whole paragraphs you won't get enough 'resolution' or detail in your data to build something meaningful from it. 

The analogy I like to use is imagine your interview data is a lump of clay, and coding is like cutting it into chunks. If you only cut it into a few large pieces (coding large stretches of text), you'll have limited options to put it together in different ways. Lots of small pieces give you more possibilities. You can also code the same text with different labels if you see different information or interpretations in a piece of text.

Also, just because you code data doesn't mean you have to incorporate it later on - you can be selective. So, create plenty of codes at this stage; it'll pay off later on. I suspect another concern is creating too much to work with at this stage, which can be a challenge, for sure, and managing that could be the subject of a later post.

Very top-level coding/noting

Along with 'not enough' codes or coding large blocks of text, novice researchers often jump ahead in their coding to try to make codes more conceptual, for instance, labelling with codes like 'stress' or 'coping'. When I see those codes or 'stress' and 'coping' used as individual themes in an analysis, I often point out that stress and coping was the subject of my WHOLE PhD thesis; they're not small elements! Going to these kinds of conceptual codes is jumping way ahead in the analytic process and misses creating information that will make for a better IPA and RTA, which is all about the detail and the nuance. I think this is sometimes an attempt to be more interpretative or psychological, and perhaps that these sound tidier and more formal than the often quite scrappy initial notes we make. It's understandable because we rarely get to see others' coding, and you might not have opportunities to work examples with someone more knowledgeable.

Your initial coding should focus on you thinking deeply (and labelling) what you see in your data. I never worry about trying to reuse codes at this stage (so I'm not trying to find repeated code names). I'm literally labelling as much of the text as I can with what's going on or other perspectives on my data.

What helps?

✅ Practical guidance! Code more than you think, and keep your labels straightforward and clear

✅ Find opportunities to practice coding with a more expert other - this is a key part of our IPA and RTA workshops because it's so helpful to practice with others and a guide

✅  Seek out opportunities for 1:1 feedback on a portion of your coded transcript from your instructor or supervisor (or book a 1:1 to review this). 

From my experience, it's possible to learn a lot from working with a very small amount of data, and collaboration can be a highly effective way to inspire your independent work.

References

Braun, V., & Clarke, V. (2022). Thematic Analysis: A practical guide. SAGE.

Smith, J. A., Flowers, P., & Larkin, M. (2022). Interpretative Phenomenological Analysis: Theory, Method and Research (2nd ed.). SAGE.