Toward good practice in thematic analysis: Avoiding common problems and be(com)ing a knowing researcher

In response to the increasing use of thematic analysis (TA), and particularly the TA approach we have developed, in qualitative and mixed methods research published in IJTH, the editors of the journal have invited us to provide a commentary on good practice and common problems in TA research. The aim of this commentary is to guide researchers in producing and reporting methodologically coherent TA and reviewers in assessing what constitutes good practice, and ultimately to support IJTH in publishing high quality TA. This commentary is based on a review of 20 papers published in IJTH citing our work¹—typically Braun and Clarke (2006), the paper in which we first outlined our TA approach—with most authors claiming to have “followed” our approach, and a minority citing our work but using other approaches. It’s important to stress that our goal here is not to encourage or enforce strict adherence to the procedures we have outlined—or what has been dubbed methodolatry (Chamberlain, 2000) or proceduralism (King & Brooks, 2018), where procedures are prioritized over reflexivity and theoretical sensitivity. Rather, we want to encourage what we call knowing practice of TA. A knowing researcher is one who strives to “own” their perspectives (Elliott et al., 1999), both personal and theoretical, is deliberative in their decision-making, and reflexive in their practice of TA.

The 20 papers we reviewed provide an interesting “snapshot” of TA research in the field of transgender health research. Most papers reported qualitative studies—with the rest mixed-method designs—typically based on interview, focus group or qualitative survey data. Most exemplified an experiential orientation to qualitative research, with a focus on the lived experience and perspectives of transgender folx (and some interest in the views of others, such of parents of transgender adolescents), or the factors that influence and contextualize certain behaviors and choices, underpinned by a realist or critical realist ontology. We were heartened that some researchers in the field of transgender health are also using TA to interrogate the social construction of meaning within a critical orientation to qualitative research (for further discussion of the experiential/critical distinction, see Braun & Clarke, 2022b). The idea that TA—or indeed qualitative research—just offers a window into experience is unfortunately too common, when the full potential is far wider.

In our critical review, we highlight three important issues (1) the, often unacknowledged, diversity within TA and many researchers’ tendency to, paraphrasing Marecek (2003), “swim unknowingly in the waters of positivism”; (2) confusing themes-as-meaning-unified-interpretative-stories with themes-as-topic-summaries; and (3) (not) owning one’s perspective.

TA is a family of methods, not a singular method—there is no “standardised TA”!

An important step on the path to knowing TA practice is appreciating the diversity within TA and understanding what type of TA you are practising. Some of the papers reviewed implicitly or explicitly presented TA as a singular method (e.g., through references to “standardised TA”), whereas TA is better thought of as a family of methods. To capture some of the diversity within the TA family, we have developed a typology of similar approaches, which we designate coding reliability, codebook and reflexive TA, and thematic coding (for our most detailed discussion, see Braun & Clarke, 2022b). These approaches have some things in common: practices of coding and theme development; the possibility of capturing semantic and/or latent meaning, and orienting to data inductively and/or deductively; and the designation of TA as a theoretically flexible method,² rather than a theoretically informed and delimited methodology. However, they differ in the enactment of coding and theme development, underlying research values, and the conceptualization of key concepts such as the theme. Procedural differences should not be dismissed as trivial, as they reflect underlying research values. We found Kidder and Fine (1987) small q/Big Q qualitative distinction useful when developing our typology. We also like Finlay’s more recent (2021) distinction between scientifically descriptive (small q, positivist) and artfully interpretive (Big Q, non-positivist, reflexive) TA.

Small q qualitative research reflects the use of techniques of qualitative data collection and analysis within a framework of (post)positivism—in many disciplines, including our discipline of psychology, this is the dominant values framework for research. Coding reliability TA is an example of small q or positivist qualitative, as it emphasizes procedures for ensuring the objectivity, reliability or accuracy of coding and keeping “researcher bias” in check (e.g., such as through the use of structured codebooks, multiple coders who independently code the same data, the calculation of intercoder agreement and consensus coding).

Big Q qualitative involves the use of techniques of qualitative data generation and analysis within a non-positivist framework informed by qualitative research values. There is no one set of research values that all qualitative researchers agree on, but many emphasize researcher subjectivity as a resource for research, rather than a threat to be contained, and meaning and knowledge as contextually situated, partial and provisional. Big Q researchers typically conceptualize mind-dependent truths, rather than a mind-independent truth (Tebes, 2005). Our approach, which we now call reflexive TA to acknowledge the plurality of TA and better distinguish it from other approaches (see Braun & Clarke, 2019), is an example of a Big Q or non-positivist qualitative approach (other reflexive approaches include Hayes, 2000; Langdridge, 2004). Reflexive TA approaches embrace researcher subjectivity as a resource for research (rejecting positivist notions of researcher bias, see Varpio et al., 2021), view the practice of TA as inherently subjective, emphasize researcher reflexivity, and reject the notion that coding can ever be accurate—as it is an inherently interpretative practice, and meaning is not fixed within data.

Codebook approaches to TA—such as framework (Gale et al., 2013) or template (King & Brooks, 2018) analysis—combine some of the more structured procedures of coding reliability TA³ with some of the qualitative research values of reflexive TA. Finally, thematic coding involves the use of grounding theory coding procedures to develop themes from data. Although thematic coding remains in use—including in a few of the papers in our review⁴—and discussed in some methodological texts (e.g., Flick, 2018; Rivas, 2018), it was more common before TA was widely recognized as a distinct method.

Methodological incoherence beckons when researchers seemingly unknowingly mash together different approaches to TA (see Braun & Clarke, 2022a). There were several examples in the papers we reviewed of researchers using Big Q reflexive TA approach and procedures, with conceptually incoherent additions, such as small q consensus coding and measuring intercoder agreement, or adding a codebook development/recoding the data phase, or referencing both positivist notions of researcher bias and Big Q notions of reflexivity, or expressing concern for the accuracy and objectivity of the coding or the potential for misinterpreting the data (implying that correct interpretation is possible). The papers evidenced little recognition of the potential for methodological incoherence when drawing on concepts and procedures from both positivist/small q and non-positivist/Big Q TA/qualitative research, or any justification or rationale for these “mash-ups.” In pointing this out, we’re not arguing that analytic procedures should be followed precisely like baking recipes, and that this is what constitutes good research. Rather, a knowing TA researcher would acknowledge their divergence from established procedures, including conceptual incoherence, and provide a rationale for their innovative approach. It’s not the case that “anything goes” in TA research, as the procedures have broad paradigmatic and conceptual foundations. Researchers cannot coherently be both a descriptive scientist and an interpretative artist (Finlay, 2021). The procedures we and other TA methodologists have developed are thought-through manifestations of underlying research values, meaning divergences and mash-ups should be equally thoughtful.

Telling meaning-united stories or summarizing topics?

Another divergence across the TA family of methods relates to the conceptualization of themes and whether themes are understood as a) summaries of topics or categories (what is shared and unites the observations in the theme is the topic, such as “good experiences in healthcare”); or b) capturing a core idea or meaning (what is shared and unites the observations in the theme is meaning), and the telling of an interpretative story about it. This latter type of theme can draw together data connected to even seemingly unrelated topics, if the core idea or meaning is evident. Topic summaries as themes are common in coding reliability and some codebook TA, and thematic coding; themes in reflexive, and some codebook, TA are conceptualized as meaning-based, interpretative stories. But these conceptualisations often don’t map onto what happens in practice. Topic summary themes are so widely used in reflexive TA that we have identified this as the most “common problem” in reflexive TA (see Braun & Clarke, 2021a)—alongside what we call “positivism creep,” where positivism slips unknowingly into reflexive TA (through the use of concepts like researcher bias; see Braun & Clarke, 2022b). Both problems—positivism creep, and the use of topic summary themes in reflexive TA—were evident in the papers we reviewed. For us, the distinction between these two “types” of theme is very clear, but we know researchers sometimes struggle with the distinction. So, if you’re doing reflexive TA, how might you check if your theme is a topic summary or a meaning-based interpretive story? If you could conceivably have developed this theme before analyzing your data, or if it maps closely on to a data collection question, then it is quite likely to be a topic summary. Similarly, if it summarizes the different or main things participants said about a particular issue or topic. Theme names can suggest a topic summary through, for instance, a one-word name that identifies the topic, such as “Doctors,” or something like “Experiences of…,” “Barriers to…,” “Influences on…,” suggesting diverse experiences, barriers and influences will be discussed (sometimes meaning-based themes may just be badly named, see Braun & Clarke, 2022b). By contrast, themes as interpretative stories built around uniting meaning cannot be developed in advance of analysis. They contain diversity, but they have a central idea that unifies the diversity (instead of “good experiences of healthcare” you might have the theme “validation of my personhood”).

Two of the reviewed papers provided clear examples of themes as meaning-based interpretative stories. Fraser et al. (2021) explored transgender adults’ experiences of gender affirming healthcare readiness assessments in New Zealand. In contrast to most of the papers reviewed, which reported higher numbers of themes and subthemes, Fraser et al. reported two themes: proving gender and the trans narrative. Proving gender centered around participants’ experiences of the assessment process as an aversive gatekeeping practice designed to test whether they were adequately or truly trans. The trans narrative centered around the pressure participants felt—because of the testing if they were “properly trans” character of the assessment process—to present their gender in a particular way, in order to gain access to gender affirming healthcare. The trans narrative required a binary trans identity, knowing they were trans from a very young age and wanting “full” medical transition. Frohard-Dourlent et al. (2020) explored experiences of surgical readiness assessments in Canada and reported three themes and seven sub-themes 1) Assessments as gatekeeping (incorporating three subthemes: assessments as outdated and irrelevant; power asymmetry undermining care; assessments as discriminatory); 2) Assessments as a barrier to care (subthemes: assessments as confusing; and inaccessible); and 3) Assessments as useful (subthemes: assessments as effective and clarifying; and affirming). Similar to experiences reported by Fraser et al., participants often experienced the assessment process as aversive gatekeeping, with health professionals having the power to deny access to care, and compel conformity to an archetypal and outdated binary trans narrative. Assessments were experienced as difficult to access because of opaqueness, bureaucracy and a lack of socio-economic privilege. Assessment were also conversely experienced as positive by some, with participants feeling supported by the assessor and prepared for the next steps. Frohard-Dourlent et al. also provide an example of a clear overview of the thematic structure (in the form of a table), and a brief but effective account of the authors’ analytic process and engagement with reflexive TA, both important elements of a high-quality TA report (Braun & Clarke, 2022b).

Why does this distinction between topic summary and meaning-based interpretive story themes matter, and why is it a problem if researchers use reflexive TA but produce a set of topic summaries? Simply put, topic summaries make no conceptual sense in reflexive TA, and the procedures have been designed to support the development of deep understanding and the telling of interpretative stories about meanings (sometimes obvious, sometimes subtle) that cut across a dataset and capture an important aspect of whatever you are trying to understand! The practice requires depth of engagement, thinking creatively and reflexively about the data, an intensive and organic coding process designed to parse out different facets of data meaning, and to help the researcher move beyond the most obvious or superficial meanings in the data. There is little point engaging in this laborious (but hopefully rewarding) process to then summarize data under headings that could have been determined before beginning the analysis. If your goal is to develop a set of topic summary type themes, select an approach, determined by your research values, that has that as its analytic purpose. Both coding reliability and some codebook approaches conceptualize themes as topic summaries (template analysis notably allows for the possibility of developing themes during or from coding). They are developed early in the analytic process, sometimes lifted from data collection questions, and coding is a process for allocating data to these early/pre-determined themes.

(Not) owning one’s perspective

The papers we reviewed exemplified both good and bad practice with regard to researchers striving to own their perspective. All included some kind of statement of the researchers’ personal positioning and/or professional experience with regard to gender identity, which is particularly important when researching socially minoritized groups and when researchers may be more socially powerful and privileged outsiders. However, reflexivity rarely extended beyond this more personal framing (Wilkinson, 1988). Reflexive practice, which goes beyond a “shopping list” of identities (Folkes, 2022), is particularly important for reflexive TA. We look forward to reading more instances of researchers linking their personal positioning to their analytic process and more detailed discussions of how researchers engaged in reflexivity, and how this shaped the analysis they produced (for excellent examples of this, see Ho et al., 2017; Trainor & Bundon, 2021). Language around theme development is also important to signal the researcher’s active role in generating their themes (and also to clearly signal that themes are not implicitly conceptualized as real things that exist within data prior to analysis). In reflexive TA, themes are generated, created or constructed (for example), they are not identified, found or discovered, and they definitely don’t just “emerge” from data like a fully-grown Venus arising from the sea and arriving at the shore in Botticelli’s famous painting (see Braun & Clarke, 2006, 2016).

Reflexivity, personal or otherwise, is one aspect of a researcher striving to own their perspectives; discussing and coherently enacting their theoretical assumptions is another. Some of the papers we reviewed included a statement of the theoretical assumptions informing the use of TA. But some didn’t. Because TA is better understood as closer to a method than a methodology, and because of its theoretical flexibility, it’s vital that researchers locate their use of TA theoretically (see Braun & Clarke, 2022a, 2022b). TA cannot be conducted in a theoretical vacuum—researchers inescapably bring in assumptions about the nature of reality, about what constitutes meaningful knowledge and knowledge production, and what their data represent or give them access to, even if these are not discussed. Ideally, the reader should not be left to detect what the researcher’s assumptions are—they would explicitly be discussed in the paper. We originally stated that reflexive TA could be underpinned by (simple) realism (Braun & Clarke, 2006), but we now think the Big Q research values of reflexive TA make a simple or naive realist reflexive TA a tricky proposition. This raises questions about whether widely used realist/positivist quality practices, like saturation, triangulation and member checking (see Varpio et al., 2017), all referenced in the reviewed papers and in published TA more broadly, are coherent with reflexive TA.

We’ll use member checking or participant validation of analysis to explore this—where participants are asked to input on whether an analysis faithfully or fairly represents their experience. This practice in theory controls for or corrects any subjective bias—misinterpretation, misemphasis—of the researcher (Smith & McGannon, 2018). Ethically/politically, this practice is unquestionably important when researching and claiming to represent the experiences of socially marginalized groups, especially so if the researchers are all privileged “outsiders” to that group, as is often the case for trans health research. However, the use of member checking is infused with assumptions about reality and knowledge production that sit (conceptually) uncomfortably with reflexive TA—including that there is a truth of participants’ experiences that we can access if we can keep the potentially distorting effects of researcher influence in check (see Smith & McGannon, 2018). Reflexive TA is premised on the researcher always shaping their research; it will always be infused with their subjectivity, and they are never a neutral conduit, simply conveying a directly-accessed truth of participants’ experience. Tracy (2010) highlighted the concept of member reflections as a Big Q alternative to member checking, which is not about verification, or accessing truth or reality. With this process, participants are invited to reflect on the analysis to offer additional insight and generate further data on the topic at hand. This can mean exploring gaps in understanding, recognizing and reflecting on contradictions and differences in understanding, and considering how to acknowledge and present these in the written report (see Smith & McGannon, 2018). Within a wide diversity of Indigenous and participatory approaches (e.g., Barlo et al., 2020; Cammock et al., 2021; Fine et al., 2021; Ware et al., 2018), differently-conceptualized relationships between “researchers” and “participants,” and/or understandings of (the purpose of) knowledge and the (primary) obligations and purposes of research, also render considerations of “member-checking” quite differently from more conventional western frameworks of (qualitative) knowledge production.

The key to selecting methodologically coherent quality standards and tools is knowing practice—reasoning through the assumptions embedded in particular concepts and practices. Luckily, qualitative methodologists have already done a lot of the heavy lifting here (e.g., Braun & Clarke, 2021b; Smith & McGannon, 2018; Varpio et al., 2017, 2021), presenting thoughtful considerations of what’s assumed and at stake in various supposedly universal criteria. Your task is to think, and reflect, and ensure any quality measures you use are coherent with your approach to TA and underpinning theoretical assumptions.

Ten recommendations for producing and reporting methodologically coherent TA and being a knowing TA researcher

We now distill this commentary into ten snappy recommendations for TA researchers to help with producing and reporting methodologically coherent TA. We encourage reviewers to use these recommendations to inform their assessments of TA manuscripts. These should not be treated a checklist in the narrow sense, but as important things to reflect on, and reason through—consider them provocations for knowing practice.

1. Recognize the plurality of TA; determine where your chosen TA approach is located on the scientifically descriptive (small q)—artfully interpretive (Big Q) spectrum.

2. Determine your underlying research values and philosophical assumptions; locate your use of TA theoretically.

3. Consider your analytic practice; ensure all methodological procedures and concepts cohere with your research values and TA approach.

4. Justify divergences from established practice and “mashups;” ensure these are theoretically coherent.

5. If using reflexive TA, link personal reflexivity to your analytic practice; don’t mention bias.

6. Discuss how exactly you engaged with your chosen approach to produce your analysis.

7. Recognize the differences between topic summary and meaning-based interpretative story conceptualisations of themes; ensure your type of theme is coherent with your TA approach (and justify any divergences).

8. Ensure your language around theme development is coherent with your TA approach.

9. Provide a clear overview of your themes/thematic structure in the form of a list, table or thematic map.

10. Ensure the quality standards and practices used cohere with your TA approach and underlying theoretical assumptions.

Everything changes…

We end with a final recommendation for readers wanting to pursue good practice in reflexive TA—read beyond our 2006 paper! Our thinking around TA has evolved since 2006, including now using the specific name reflexive TA (Braun & Clarke, 2019). Our recent book Thematic analysis: A practical guide (Braun & Clarke, 2022b) provides the most comprehensive guidance for both doing reflexive TA, and thinking about TA. We have developed a website—www.thematicanalysis.net—which links to all of the resources we have produced. We have published numerous papers addressing, among other things, misconceptions and confusions around reflexive TA (Braun & Clarke, 2021a), whether saturation is a meaningful concept for reflexive TA (TL; DR no it isn’t!; Braun & Clarke, 2021b), designing methodologically coherent TA research (Braun & Clarke, 2022a), and when and why to use reflexive TA (Braun & Clarke, 2021c). We particularly encourage reviewers to read the tool for evaluating TA manuscripts for publication in Braun and Clarke (2021a).

Virginia Braun
School of Psychology, The University of Auckland, Auckland, New Zealand
[email protected] https://orcid.org/0000-0002-3435-091X Victoria Clarke
School of Social Sciences, University of the West of England, Bristol, UK
https://orcid.org/0000-0001-9405-7363

Funding

The author(s) reported there is no funding associated with the work featured in this article.

Notes

1 As we are discussing bad practice and have no wish to “name and shame” authors, we have chosen not to include a list of papers reviewed and only reference examples of bad practice in general terms. However, as we also recognise the value of concrete examples, we do identify the authors of two papers that exemplify good practice in various ways.

2 TA was occasionally described as a methodology in the papers reviewed—we don’t think it fulfils all of the requirements of a methodology because it doesn’t inherently provide researchers with a theoretically-informed framework for research. Unlike methodologies such as grounded theory or discourse analysis, TA offers few directives around philosophical and theoretical positioning, appropriate research questions, data collection methods, or the size or constitution of the participant group/dataset (see Braun & Clarke, 2022a). Therefore, we think TA is better understood as closer to a method. We say closer to because the procedures and conceptualisations of key concepts associated with different iterations of TA reflect particular underlying paradigmatic assumptions or research values.

3 In line with qualitative values, codebooks are more used to chart the developing analysis, instead of being a tool to measure whether coding is reliable.

4 One paper justified the use of grounded theory coding techniques to do TA on the grounds that Braun and Clarke (2006) don’t provide a plan for doing TA. Given that the purpose of our 2006 paper was precisely to do that, and it was the first place we outlined a clear six-phase approach to reflexive TA—and critiqued the use of grounded theory techniques to do TA—this stands as one of the many spurious claims that exist about TA (see Braun & Clarke, 2021a). We are troubled how such basically incorrect information survives the peer review and editorial process.