Themes, Codes and Buckets: What is Thematic Analysis (TA)?

There are a huge variety of ways we can analyse qualitative data, but perhaps the most commonly applied is known as Thematic Analysis. This is an approach classically described by Braun and Clarke (2006) but is straightforward and intuitive enough that most people seem to applying it in some form without reading anything about it!

Thematic Analysis, or TA is often confused or appropriated with grounded theory because it shares a common operational premise: after familiarising yourself with the data, go through and create ‘codes’ as you read through the data, to label basic things discussed in the text. These codes are a specific “single idea associated with a segment of data, and consist of pithy labels identifying what is of interest in the data” (Braun and Clarke 2009). These are ‘low-level’ usually descriptive codes, often called ‘initial codes’, for example Family, Problems, Food, Time – basic tags that start breaking down the data. A researcher applying TA will first go through the data and create these codes as a way to understand the data, but also to start thinking about higher-order or more complex issues that are being discussed.

Once this is done, the next analytic step is to start creating ‘themes’. For Braun and Clarke’s TA, themes are “an idea or concept that captures and summarises the core point of a coherent and meaningful pattern in the data” and “a common, recurring pattern across a dataset, clustered around a central organising concept” (Braun and Clarke 2009). There are fewer themes, and they are more general, and might be more conceptual and closer to answering a research question. For example ‘Family meals are sacred’ or ‘Controlling diet’. Braun and Clarke describe themes as usually a short self-explanatory sentence, where codes are usually one word or term labels. Themes when ‘fully realised’ should be a recurring concept that chimes with many participants.

If you imagine the writing up process, themes are likely to be a subheading in your report, paper or thesis, that illustrates an interesting interpretation or discovery, and you will demonstrate this with example quotes. Codes themselves are probably not interesting enough to warrant a whole section on ‘everything people said about Time’ or ‘Diet’.

Themes are usually created by putting many different codes together. It’s the creation of that ‘Theme’ level description that a lot of newcomers struggle with, due to the angst of knowing how many themes to create, what they should be, and if they are too general or vague. Although these worries can apply to the creation of codes, it’s the thematic process where the wording of themes starts to put a real stamp on the data.

This is one reason why Braun and Clarke suggest that reviewing themes is the next step after their creation (and when most codes have been assigned to at least one of them). Themes might need to be reworded, made more specific or general, or have additional themes added if not all the concepts in the data are being captured well.

But they also talk about ‘domain summaries’, crude ‘buckets’ to keep together many codes on a broad area or that seem related. This can be done as a precursor to creating themes, especially if you are ending up with a large number of codes.

More recently, Braun and Clarke have updated their terminology to prefer ‘reflexive’ thematic analysis. This recognises that researchers should not just claim that themes or codes are ‘emerging’ from the data as described grounded theory, but that it is the explicit intervention of the researcher to create them, biases and assumptions, warts and all. They also recognise that TA is not an exclusive template, and it can be combined with other ways of reading and interpreting the data. For example, the basic approach could be used with discourse analysis to explore the use of language.

They also discuss the difference in approaches between the similarly named and quite conceptually similar ‘thematic coding’ approach, described prominently by Gibbs (2007). A lot of the terminology maps well to the stages of TA, but usually has a greater stress on creating memos, and links closer to grounded theory. You can see a lot more about the distinctions between different approaches in the excellent FAQs on Thematic Analysis on Bruan and Clarke’s website.

There is a lot of confusion and overlap between these different approaches, and the Big Two: Grounded Theory and Thematic Coding are often mis-applied and misdescribed as one sentence descriptions of the analysis process in published research. The classic “We applied grounded theory / thematic analysis to analyse the data” is not nearly detailed enough to know exactly how researchers applied these expansive approaches. Braun and Clarke argue that while there is little reason to nitpick about exactly how a procedure is followed, it is important to understand the difference between applying the different approaches. For me this is especially crucial when writing up research methodologies, and when teaching qualitative analysis. Pedagogy suffers when “We applied grounded theory / thematic analysis to analyse the data” is all the detail students can learn about a piece of research that is methodologically interesting to them.

However, at the risk of creating yet-another-intepretation of Thematic Analysis, I would suggest that anyone wanting to learn it go to the resources linked on Braun and Clarke’s website. However, there are also some good resources not linked there, including video tutorials by Clarke and obviously critiques and issues observed by dozens of other authors such as Nowell et al. (2017).

In Quirkos, there are different ways that users can apply a TA-type approach. The first is to create a series of codes (Quirks/bubbles created on the fly) and then later group them together by creating a theme or bucket as a separate bubble, and group by dragging and dropping codes onto the top-level theme. You can also code directly at both levels. It’s also possible to use the Groups function to group codes together into themes, with the name of the themes matching the codes. If you decide that a lot of codes are going to fall under multiple themes, this is probably a neater way. Users can also use the memo functions to keep reflexive text, and address reliability/trustworthiness issues using multiple coders, and comparing interpretations.

Quirkos gives you a 4 week free trial to see for yourself how flexible it can be for different analytical approaches, and if you end up liking it, some of the cheapest and most generous licences around. See if a colourful and simple qualitative software tool can help you apply Reflexive Thematic Analysis!

References:

Braun, V., Clarke, V., 2019, Reflecting on reflexive thematic analysis, Qualitative Research in Sport, Exercise and Health, 11:2, 589-597

Braun, V., Clarke, V., 2019, Answers to frequently asked questions about thematic analysis, https://cdn.auckland.ac.nz/assets/psych/about/our-research/documents/Answers%20to%20frequently%20asked%20questions%20about%20thematic%20analysis%20April%202019.pdf

Braun, V., Clarke, V., 2006, Using Thematic analysis in Psychology, Qualitative Research in Psuchology, 3, 77-101
https://eprints.uwe.ac.uk/11735/2/thematic_analysis_revised_-_final.pdf

Gibbs, G., 2007, Analyzing Qualitative Data, 4. Thematic Coding and Categorizing, Sage, London
https://methods.sagepub.com/book/analyzing-qualitative-data