Thematic Analysis Write-Up With Evidence

Thematic analysis is a family of methods, and your write-up has to match the one you used. Psychologists Virginia Braun and Victoria Clarke, who set out the six-phase approach in 2006, distinguish three main types¹, and each changes what you report.

• Reflexive: themes are patterns of shared meaning you build from your own codes, with no codebook and no agreement score.
• Codebook, such as framework or template analysis: themes are set early, often around topics, and the data are charted against them.
• Coding reliability: independent coders apply a structured codebook, and you report their agreement with a statistic such as Cohen's kappa.

Mixing approaches without saying so is common. In a 2024 review of 20 papers in the journal Palliative Medicine that cited their work, Braun and Clarke found some degree of methodological incoherence in every one².

Short answers to an open survey question are often coded with a codebook and counted instead. Coding open-ended survey responses walks through that job.

Braun and Clarke's fourth phase is checking the candidate themes against the coded data and the entire dataset³. Do it in a sheet, not in your head, so you can show your work.

1. 1

   State the central idea

   Write what the theme claims. If you can't describe its scope and content in a couple of sentences⁴, Braun and Clarke say it may need more refining.

2. 2

   Count who is behind it

   Note how many participants contributed, and how many extracts came from the top contributor. Braun and Clarke warn against themes built from a few vivid examples¹.

3. 3

   Look for contradictions

   Search the rest of the data for extracts that complicate or limit the claim. If you find none, record where you looked.

4. 4

   Decide what it is

   Keep it, make it a subtheme, rename it as a claim, dissolve it into its codes, or report it as one person's account.

A grid of participants by candidate themes makes this quick. Nicola Gale and colleagues note that the framework method's matrix of participants by codes can draw attention to contradictory data, deviant cases or empty cells⁵.

Begin with an overview of the themes and subthemes. APA's reporting standards for qualitative research suggest synthesizing illustrations such as diagrams, tables or models⁶ where they help.

For a journal article, a master's dissertation or one analytic chapter of a doctoral thesis, Braun and Clarke consider two to six themes, including subthemes¹, about right. Use subthemes sparingly, for one facet of a theme that deserves emphasis.

Order the themes so each builds on the last, and keep background that isn't a theme in the participant description.

Give each theme a section: the claim as its heading, its central idea in a sentence or two, then a narrative carried by extracts. Braun and Clarke's 2006 paper says extracts need to be embedded within an analytic narrative⁴ that argues something about your research question.

Choose extracts that show the claim plainly, spread them across participants, and label each with an ID such as P7. Then say what each one shows. APA's standards say quotes should not replace the description of the findings⁶.

Put contradicting extracts inside the theme they complicate, not in a limitations paragraph at the end. APA's standards ask for findings that make sense of contradictions or disconfirming evidence⁶, for instance by explaining why a conflict might exist.

A candidate resting on one or two people can still matter. Report it under its own heading as an individual account, stating its narrow base plainly.

Write the methods from your notes, not from the textbook. In the paper introducing their reporting guidelines (RTARG), Braun and Clarke ask authors to report not the generic phases but their specific engagement, process and decisions². Say who coded, how you tested candidates, and which you merged, renamed or dissolved, and why.

Add a few sentences on your background and assumptions, and how they shaped the analysis. In reflexive work, don't say themes "emerged"; you developed them. Skip saturation and worries about researcher bias too: Braun and Clarke describe both as founded in (post)positivism and incoherent with reflexive thematic analysis².

Many journals require a reporting checklist at submission. COREQ, a 32-item checklist for interviews and focus groups⁷, is widely used. For reflexive work, Braun and Clarke propose RTARG instead.

If a reviewer asks for something your approach doesn't use, such as a kappa, explain why in your response to reviewers instead of bolting it on.

Your discussion then sets each theme against earlier studies, which a literature review synthesis helps you map. River, the tool at the top of this page, tests each candidate theme against your coded extracts, then drafts your results and methods.