How to do thematic analysis

By Dr. Rafiq Muhammad, MD, PhD · Updated June 2026

Thematic analysis (TA) is the most widely used method for finding patterns of meaning — themes — across qualitative data. Braun & Clarke’s six phases are the standard recipe. The key mindset: themes are actively constructed by you around a central idea, not nuggets waiting to be found — and the process is recursive, not a one-way march.

The six phases (Braun & Clarke)

1. Familiarise yourself with the data. Transcribe, read and re-read, and jot first impressions.
2. Generate initial codes. Systematically label features of interest across the whole dataset.
3. Search for themes. Collate codes into candidate themes — patterns of shared meaning.
4. Review themes. Check each theme against its coded extracts and the whole dataset; split, merge, or drop.
5. Define and name themes. Pin down what each theme captures and its boundary; give it a clear name.
6. Produce the report. Weave the themes into an analytic narrative, evidenced with vivid quotes.

Code vs theme — don’t confuse them

A code is a short label for one feature of the data (“fear of judgement”); a theme is a broader pattern organised around a central organising concept, built by grouping codes (“the social cost of asking for help”). The most common TA mistake is using your interview-guide topics as themes — that’s a summary of what you asked, not a finding about what it means.

Reflexive vs coding-reliability TA

Two flavours, two philosophies — choose and justify one: reflexive TA treats your subjectivity as a resource and doesn’t use a fixed codebook or inter-coder reliability; coding-reliability TA uses a structured codebook and reports agreement (e.g. Cohen’s κ). See trustworthiness for how rigor is judged in each.