Thematic analysis is a family, not a method

Thematic analysis is the most taught and least specified method in qualitative research. Students learn the name in their first week, and they can recite a set of phases, but no single agreed procedure sits behind either. Braun and Clarke (2006) called it poorly demarcated and rarely acknowledged. Fifteen years on, the same authors report that its flexibility has been mistaken for homogeneity, by editors, reviewers and authors alike (Braun and Clarke, 2021).

This fourth of ten units argues that thematic analysis is not one procedure but a family. Its members differ in what they assume about knowledge. A report that does not name its version cannot be evaluated, and the sentence "themes emerged" misdescribes work the analyst did. The version decides which claims a study can carry. It also decides whether a reviewer's request for an agreement figure can be answered at all. The unit that follows finds the same pluralism in grounded theory.

A name that has never covered one thing

Braun and Clarke (2006) set out what thematic analysis is. They located it among other methods that search for patterns, and placed it in relation to competing epistemological and ontological positions. They gave guidelines and named the pitfalls: extracts strung together with comment, and research questions reported as themes. A method that fits many theoretical positions, they show, is not one method with many users.

Braun and Clarke (2021) state it plainly. Thematic analysis refers not to a singular approach but to a cluster of sometimes conflicting approaches, divergent in procedure and philosophy. Assuming that it is one approach is the first of ten common problems they find in work claiming to follow their guidance. They name its institutional form: coding reliability measures promoted as universal quality requirements.

A dissertation reports that its data were analysed using thematic analysis, cites Braun and Clarke, and tabulates themes with participant counts. Nothing is wrong except the label. The label conceals a coding reliability approach.

The versions are not interchangeable

Joffe (2011) treats thematic analysis as epistemologically positioned rather than neutral. Her chapter sets out the method's origins, its key epistemological assumptions, the questions and data it suits, and how to conduct it. Assumptions come before procedure. The same label covers a realist reading in which themes are latent truths, and a constructionist reading in which they are products of the analyst's engagement.

Braun and Clarke (2022) carry the point into the textbook. They develop reflexive thematic analysis across a worked data set, and then ask how it differs from its siblings and cousins. Terry and colleagues (2017) give the handbook version of those distinctions.

The opposing position deserves a fair statement. Kiger and Varpio (2020) describe thematic analysis as widely used but often misunderstood. They define a theme carefully, set out the six-step framework, and say when it suits a variety of epistemological frameworks. That is one method, one procedure, many epistemologies. It is also, in Braun and Clarke's (2021) diagnosis, the assumption of homogeneity itself.

"Themes emerged" describes an act, and describes it wrongly

Braun and Clarke (2021) quote their 2006 paper to make their sharpest point. An account of themes emerging or being discovered is passive. It denies the active role the researcher plays in identifying patterns, selecting which are of interest, and reporting them. Themes must be analytic outputs, not inputs. A theme that was found cannot have been built.

Byrne (2022) demonstrates the alternative, working a real data set through the phases. Her data are eleven interviews of twenty-five to thirty minutes from a study of Irish post-primary educators and the promotion of student wellbeing. Subject teachers, wellbeing teachers, pastoral staff and senior managers were among the participants, all of working age, so a theme about ageing needs a sample of its own. Her worked phases write the analyst's reasoning between the steps rather than assuming it. Themes are produced by organising codes around a central organising concept the researcher interprets from the data, not predefined in order to find codes. Clarke and Braun (2017) give the compact form: thematic analysis interprets patterns of meaning, and its aim is not simply to summarise the data.

Vaismoradi, Jones, Turunen and Snelgrove (2016) reach a compatible conclusion from outside that tradition. The meaning of a theme and its development remain unclear, they observe, and they respond with a staged model of theme construction. A theme is something a researcher builds by a described procedure, not something a data set gives up.

What a version claim has to make checkable

Nowell, Norris, White and Moules (2017) argue that an analysis is trusted when it is conducted consistently. The report must also disclose the method in enough detail for a reader to judge whether the process is credible. Their auditable decision trail is what makes a version claim checkable. They illustrate it through a case study of strategic clinical networks in Alberta, Canada. For a version claim the trail is the evidence. It has to record the decision taken at each phase, so a reader can see where a category was formed, merged or dropped.

Boyatzis (1998) made an earlier attempt at the same systematisation, distinguishing themes at a manifest level from those at a latent level. What was once a distinction within one method is now a choice between approaches. Braun and Clarke (2023), writing as editors, add that readers who do not share the author's assumptions judge the report. The assumptions therefore have to be written down.

The boundary problem runs in two directions

Vaismoradi, Turunen and Bondas (2013) find that the boundaries between qualitative content analysis and thematic analysis have not been clearly specified. The terms are used interchangeably, and their one difference is the opportunity for quantification.

A version claim therefore has to answer two questions. Which member of the family is this, and which neighbouring method is it not? A study that reports theme frequencies and cites Braun and Clarke has answered neither. The frequencies mark it as content analysis, while the reflexive version treats counting as an alien import.

Contested question: one method with many uses, or a family of incompatible members?

The first position is that thematic analysis is a family whose members differ in ways that matter. Braun and Clarke (2021) state it, and the strength of their evidence deserves notice. They are the method's owners arguing in a journal article, and their ten problems were found by reading work that cites their own guidance. They insist that a plurality of approaches differing paradigmatically, philosophically and procedurally is often unrecognised. On this view a report that does not name its version cannot be evaluated, because the criteria depend on the version. Asking a reflexive analysis for an agreement figure is then a category mistake.

The second position is that thematic analysis is one flexible method with a teachable procedure, usable under a range of epistemological frameworks. Kiger and Varpio (2020) state it in a teaching guide, and Finlay (2021) supports it from a review of published analyses rather than from a systematic sample. She finds a span from the systematic to the intuitive, gives examples of good, bad and ugly practice, and proposes four criteria: rigour, resonance, reflexivity and relevance. She states that there is no one way to do it, yet offers her four Rs to the family as a whole. That is a universal quality standard, which the first position holds cannot be framed while the plurality goes unrecognised.

What would settle it is an audit neither side has run. Take published thematic analyses, code each for whether a version is named and for the criteria it invokes, then test whether expert judgements of quality track the version. Kiger and Varpio (2020) supply the counter-prediction: if one procedure travels across epistemologies, naming a version should not predict how a study is judged. The strongest counter-argument to this unit's own claim follows from that prediction. Even if the family is plural, naming the member may predict nothing about quality, so the demand this unit makes of every report could be a demand its readers cannot use.

What to do while the dispute is open

Name the version, and name the criteria you expect to be judged by. When a reviewer asks for a count, say which member would warrant one and note that yours does not. Keep a decision trail for each phase, so a reader can see where a category was formed, merged or dropped, and can apply the standard for the version you claimed.

Where the argument is heading

The direction of travel is towards version-specific quality. Braun and Clarke's 2022 guide devotes a chapter to comparing reflexive thematic analysis with its methodological siblings and cousins. Their 2023 editorial addresses editors directly. Finlay (2021) shows where the pressure will come from next. If a reviewer or examiner applies one set of criteria to every thematic analysis, the plurality Braun and Clarke (2021) describe becomes invisible again. Students then learn to write the version the rubric rewards. The work the area needs next is that audit.

The sources this unit rests on

• Boyatzis, R. E. (1998). Transforming Qualitative Information: Thematic Analysis and Code Development. Thousand Oaks: Sage. https://archive.org/details/transformingqual0000boya Boyatzis systematises thematic analysis and code development, arguing that themes can be identified at manifest and latent levels. It is the unit's earlier settlement of the same questions.
• Braun, V. and Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology 3(2), 77-101. doi:10.1191/1478088706qp063oa Braun and Clarke argue that thematic analysis is poorly demarcated but widely used and theoretically flexible, give guidelines, and name the pitfalls, including extracts strung together with commentary. They call themes emerging a passive account that denies the researcher's role.
• Braun, V. and Clarke, V. (2021). One size fits all? What counts as quality practice in (reflexive) thematic analysis?. Qualitative Research in Psychology 18(3), 328-352. doi:10.1080/14780887.2020.1769238 Braun and Clarke state that thematic analysis is not a singular approach but a cluster of sometimes conflicting ones, identify ten common problems in work claiming to follow their guidance, and make assuming homogeneity the first. It is the unit's central source for the family claim.
• Braun, V. and Clarke, V. (2022). Thematic Analysis: A Practical Guide. London: Sage. https://uk.sagepub.com/en-gb/eur/thematic-analysis/book248481 The guide develops reflexive thematic analysis from familiarisation through coding and theme development to reporting, then locates it theoretically and compares it with its siblings and cousins. It is the unit's account of one member of the family.
• Braun, V. and Clarke, V. (2023). Toward good practice in thematic analysis: Avoiding common problems in coding and reporting. International Journal of Transgender Health 24(1), 1-6. doi:10.1080/26895269.2022.2129597 Writing as journal editors, Braun and Clarke set out the problems they see most often in thematic analysis submissions to their journal. It shows that the audience for a version claim includes publishers.
• Byrne, D. (2022). A worked example of Braun and Clarke's approach to reflexive thematic analysis. Quality and Quantity 56(3), 1391-1412. doi:10.1007/s11135-021-01182-y Byrne works a real data set through the phases of reflexive thematic analysis, noting that many who claim to follow the approach fail to adhere fully to its principles, and shows themes built by organising codes around a central organising concept. It demonstrates that a theme is made rather than found.
• Clarke, V. and Braun, V. (2017). Thematic analysis. The Journal of Positive Psychology 12(3), 297-298. doi:10.1080/17439760.2016.1262613 Clarke and Braun give a two-page statement of thematic analysis for readers outside qualitative psychology, defining it as a method for interpreting patterns of meaning. Its aim, they insist, is not simply to summarise the data.
• Finlay, L. (2021). Thematic analysis: The 'good', the 'bad' and the 'ugly'. European Journal for Qualitative Research in Psychotherapy 11, 103-116. https://openjournals.ljmu.ac.uk/ejqrp/article/view/3062 Finlay contrasts the systematic and the intuitive forms of thematic analysis, gives examples of good, bad and ugly practice, and proposes four criteria, rigour, resonance, reflexivity and relevance. It is the unit's strongest case for a universal standard.
• Joffe, H. (2011). Thematic analysis. In D. Harper and A. R. Thompson (eds), Qualitative Research Methods in Mental Health and Psychotherapy. Chichester: Wiley, 209-223. doi:10.1002/9781119973249.ch15 Joffe covers the method's origins, its key epistemological assumptions, the research questions and data it suits, and how to conduct it. It orders the unit's argument, placing assumptions first.
• Kiger, M. E. and Varpio, L. (2020). Thematic analysis of qualitative data: AMEE Guide No. 131. Medical Teacher 42(8), 846-854. doi:10.1080/0142159x.2020.1755030 Kiger and Varpio describe thematic analysis as widely used but often misunderstood, define a theme carefully, set out the six-step framework, and state when it suits a variety of epistemological frameworks. It is the unit's clearest statement of the homogeneity position.
• Nowell, L. S., Norris, J. M., White, D. E. and Moules, N. J. (2017). Thematic analysis: Striving to meet the trustworthiness criteria. International Journal of Qualitative Methods 16(1), article 1609406917733847. doi:10.1177/1609406917733847 Nowell and colleagues argue that an analysis is trusted when it is conducted consistently and the method is disclosed in enough detail for a reader to judge credibility. Their auditable decision trail, from a study of strategic clinical networks in Alberta, Canada, makes a version claim checkable.
• Terry, G., Hayfield, N., Clarke, V. and Braun, V. (2017). Thematic analysis. In C. Willig and W. Stainton Rogers (eds), The SAGE Handbook of Qualitative Research in Psychology. London: Sage, 17-36. doi:10.4135/9781526405555.n2 Terry, Hayfield, Clarke and Braun set out thematic analysis for qualitative psychology in the field's reference handbook, distinguishing the versions that have developed. The distinctions are explicit.
• Vaismoradi, M., Turunen, H. and Bondas, T. (2013). Content analysis and thematic analysis: Implications for conducting a qualitative descriptive study. Nursing and Health Sciences 15(3), 398-405. doi:10.1111/nhs.12048 Vaismoradi, Turunen and Bondas find that the boundaries between the two methods have not been clearly specified, that the terms are used interchangeably, and that their one difference is the opportunity for quantification. It shows the unit that the family's edges are contested too.
• Vaismoradi, M., Jones, J., Turunen, H. and Snelgrove, S. (2016). Theme development in qualitative content analysis and thematic analysis. Journal of Nursing Education and Practice 6(5), 100-110. doi:10.5430/jnep.v6n5p100 Vaismoradi and colleagues observe that the meaning of a theme and its development remain unclear, and respond with a staged model of theme construction for both methods. It shows the unit that a theme is an achievement with a procedure.