Reflexive Thematic Analysis: The Six Phases, Quality and Reporting

Thematic analysis (TA) is a method for developing, analyzing and interpreting patterns of meaning across a qualitative dataset, such as interview transcripts, focus group discussions, open-ended survey responses or documents. It is flexible: it is not tied to one theoretical framework, and it can be used for questions about experiences, perceptions, practices, and the ways meaning is constructed in talk and text.

Virginia Braun and Victoria Clarke's 2006 paper made TA widely known and gave it a clear procedure. In later work they renamed their approach reflexive thematic analysis, to emphasize that the researcher's interpretation, positioning and theoretical assumptions are central resources for the analysis, not biases to be eliminated. They also clarified that TA is better understood as a family of methods, with at least three broad clusters that differ in philosophy and procedure.

Thematic analysis is not simply sorting quotes into categories, counting how often words appear, or summarizing what participants said about each interview question. Those activities can be part of the work, but a thematic analysis produces an interpretive account: themes that capture shared meaning and tell the reader something about the research question that the raw data alone would not.

Choosing a version of TA is a methodological decision that should match your research paradigm. Mixing procedures from different clusters, such as calculating inter-rater reliability in a study that claims to use reflexive TA, is one of the most common problems examiners identify.

Familiarization means immersing yourself in the whole dataset before coding. Read and re-read transcripts, listen to recordings if you have them, and note initial thoughts, questions and reactions. If you transcribed the data yourself, that process is already part of familiarization.

Keep familiarization notes separate from coding. They might record an image a participant used, a tension between two interviews, or your own emotional response to a story. These notes are the beginning of reflexive practice: they help you notice how your background shapes what stands out to you.

A common mistake is to rush this phase to get to coding quickly. The quality of later interpretation depends on knowing the data well enough to see connections across accounts.

Coding means working systematically through the dataset and attaching short labels to segments that are relevant to your research question. A code captures a single idea. Codes can be semantic, staying close to what participants explicitly said, or latent, capturing underlying assumptions, ideas or discourses that shape the explicit content.

Code the entire dataset with equal attention, not just the parts that feel most interesting. Expect your codes to change: in reflexive TA, you might merge, split or rename codes as your understanding develops, and many researchers code the dataset more than once. There is no requirement to reach a fixed codebook or to measure agreement with another coder.

You can code on paper, in a word processor or spreadsheet, or in qualitative data analysis software (CAQDAS) such as NVivo, ATLAS.ti or MAXQDA. Software helps manage and retrieve data, but it does not analyze for you; the interpretive work is still yours.

• Too descriptive: 'talks about work' says little. Better: 'work as escape from caring duties'.
• Too broad: a code applied to half the dataset is probably a topic, not a code.
• Keep a coding journal noting why you created or changed codes.

In this phase you look across your codes and coded data for broader patterns of shared meaning. Group codes that seem to express a common idea, and draft candidate themes. A thematic map, a visual sketch of candidate themes and how they relate, often helps.

Braun and Clarke describe a theme as a pattern of shared meaning organized around a central organizing concept. That concept is what holds the theme together. Themes do not emerge from the data on their own; you construct them through active engagement with the data, your research question and relevant theory. Avoid writing 'themes emerged' in your report, because it hides the analytic work you did.

At this stage themes are provisional. It is normal to have too many candidates, some overlapping, some thin.

Now test your candidate themes against the data. First, check each theme against the coded extracts: do they fit the central organizing concept, or does the theme try to hold too much? Second, check themes against the full dataset: does the set of themes tell a convincing story about the data in relation to the research question?

Expect to restructure. A theme may split into two, collapse into a subtheme of another, or be discarded because it does not have enough depth. Themes should be internally coherent and distinct from one another. If two themes are hard to tell apart, their central concepts probably need sharpening.

Write a short definition for each theme that states its central organizing concept, its scope and boundaries, and how it contributes to the overall analysis. If you cannot summarize a theme in a few sentences, it may need further work.

Choose names that convey the analytic point rather than a topic. A name such as 'Support' signals only a topic; a name like 'Help that comes with strings attached' tells the reader what you are claiming. Some researchers use a short participant quote as part of a theme name, which can be vivid if the quote captures the central idea.

Writing is part of analysis, not something that happens after it. The report weaves together an analytic narrative, data extracts and, depending on your approach, connections to existing literature. Extracts can be used illustratively, to show what a theme looks like, or analytically, where you interpret the specific language of the extract.

The key is balance. A report with long quotes and little commentary reads as description. A report with confident claims and few extracts gives readers no way to evaluate them. Aim for each extract to be followed by interpretation that says what it shows and why it matters for the research question. Present themes in an order that builds a coherent argument, and consider whether your discussion belongs within the results (common in reflexive TA) or in a separate chapter, as many dissertations require; a well-structured results chapter can accommodate either.

Quality in reflexive TA is not demonstrated by reliability statistics or by claiming saturation. It is shown through coherence between your research question, theoretical assumptions and procedures; through depth of interpretation; and through transparent reflexivity about your role.

One quality problem stands out: confusing topic summaries with themes. A topic summary reports everything participants said about a domain, often mirroring an interview question such as 'Experiences of training'. It contains diverse and sometimes contradictory views with no single shared meaning. A theme, by contrast, is unified by a central concept. Topic summaries are legitimate in some codebook approaches but are a weakness when presented as reflexive TA themes.

• Your paradigm, version of TA and orientation (inductive or deductive, semantic or latent) are stated and consistent.
• Themes are patterns of shared meaning, not topics or interview questions.
• Each theme has a clear central organizing concept and is distinct from the others.
• Claims are supported by extracts, and extracts are interpreted rather than simply presented.
• Reflexivity is concrete: how your position shaped the analysis, not a generic disclaimer.
• Sample size is justified in terms of information power or richness, not only a claim of saturation.
• The procedure described matches what was actually done.

In your methods section, name the specific version of TA and cite the sources you followed. Explain your theoretical position (for example critical realist or constructionist), whether coding was mainly inductive or deductive, and whether you focused on semantic or latent meaning. Describe how you moved through the phases, how many rounds of coding you did, what tools you used, and how you kept a reflexive record.

Braun and Clarke have published guidance on evaluating and reporting TA, and general qualitative reporting standards, such as the APA's Journal Article Reporting Standards for qualitative research (JARS-Qual) and the COREQ checklist for interview and focus group studies in health research, may also be required by your department or journal. Check which apply before you write.

If you collaborated with others, describe what collaboration involved, such as discussing interpretations to deepen reflexivity, rather than presenting it as reliability checking. If you used software, name it and the version, and describe its role accurately.