Thematic analysis Victoria Clarke & Virginia Braun Published as: Clarke, V. & Braun, V. (2014) Thematic analysis. In T. Teo (Ed.), Encyclopedia of Critical Psychology (pp. 1947-1952). New York: Springer. Introduction Empirical research within critical psychology is strongly associated with the use of qualitative methods. In the field of qualitative psychology a distinction can be made between experiential and critical approaches (Braun & Clarke, 2012a, Reicher, 2000), both of which involve some kind of critique of mainstream psychology. Experiential approaches aim to capture participants’ experiences and perspectives and ground research in participants’ accounts, rather than researcher’s categories, but view language as a reflection of “internal categories of understanding” (Reicher, 2000: 3), and so assume it is possible to ‘read off’ participants’ thoughts, feelings and practices from their use of language. By contrast, critical approaches challenge what experiential approaches have in common with mainstream psychology – the assumption that language is only of interest as a description of inner states. Critical approaches (usually some version of discourse analysis) understand and analyse language as something that is constitutive, rather than reflective, of our social and psychological words, as a form of social action. One of the most common qualitative methods is (some form of) thematic analysis (TA); however, there is debate about whether TA is an experiential or a critical approach, and thus an appropriate analytic method for critical psychology. History The term ‘thematic analysis’ was first introduced by the physicist, philosopher and historian of science Gerald Holton in the 1970s (Merton, 1975). Since then a number of different versions of TA, as a method for analysing qualitative data, have been proposed within psychology and the social sciences (e.g., Aronson, 1994, Boyatzis, 1998, Joffe & Yardley, 2004), but none of these approaches have achieved the ‘brand recognition’ of grounded theory, discourse analysis, and most recently, interpretative phenomenological analysis (IPA). TA arguably developed out of content analysis and could be regarded as a form of 1 qualitative content analysis (Joffe, 2011). TA is often associated with an experiential approach to qualitative psychology, and many versions of TA are located within a phenomenological framework (e.g., Guest, MacQueen & Namey, 2012, Joffe, 2011). However, in 2006, critical psychologists Virginia Braun and Victoria Clarke proposed a theoretically flexible approach to TA (see also Braun & Clarke, 2012a, 2012b), which is ideally suited to critical psychology and provides an alternative to discourse analysis, which is often regarded as the (only) method of critical psychology. Since their 2006 publication, TA has increased in popularity; it is often included in the latest editions of qualitative teaching texts and edited collections. TA is now well on the way to achieving the ‘brand recognition’ of approaches such as grounded theory, IPA and discourse analysis. Definitions TA is a method for identifying and analysing patterns of meaning (themes) in qualitative data. Braun and Clarke (2006) present TA as just an analytic method, not a methodology; it does not prescribe theoretical assumptions, appropriate research questions and ideal methods of data collection. They argue that TA can be used within a wide range of theoretical frameworks: from essentialist, experiential TA to constructionist, critical TA; even thematic discourse analysis is possible (see Taylor & Ussher, 2001). Fitting with its poststructuralist framework, thematic discourse analysis involves attention to the constructive role of language, and multiple and shifting meanings (Taylor & Ussher, 2001), but retains a specific interest in patterned meaning (discourses) within the dataset. TA can be used to address most types of research question, from questions about people’s practices and views and opinions to questions about the representation and construction of particular social and psychological objects and subjects in particular contexts, with the exception of questions about language practice (which require a discursive psychological or conversation analytic orientation). TA can be used to analyse most types of qualitative data, from secondary sources (such as men’s magazines or television talk shows) to textual data (qualitative surveys or story completion tasks); from interactive (transcripts of interviews and focus groups) to naturalistic data (transcripts of audio recordings of counselling sessions). TA can also be used to analyse large or small data sets (from 2 to twenty focus 2 groups) and to produce both data-driven (bottom-up) and theory-driven (top-down) analyses. TA involves the generation of codes and then themes from qualitative data (in thematic discourse analysis, codes, themes and discourses - “underlying systems of meaning”- are identified; Taylor & Ussher, 2001: 297). Codes capture interesting features of the data of potential relevance to the research question. Coding is not simply a method of data reduction; it is an analytic process that captures both semantic (surface) meaning within the data and latent (underlying) meaning. Themes are analytically constructed from coding, and the dataset, and capture broader patterns of meaning within the data. They are used to structure the presentation of results. Theoretical flexible TA emphasises the active (and reflexive) role of the researcher in constructing codes and themes; themes do not simply ‘emerge’ from data, the researcher does not simply search the data for the themes residing within them. Rather, the researcher makes active, interpretative, choices in generating codes and in constructing themes. Keywords Code, thematic analysis, thematic discourse analysis, theme, qualitative methods Traditional debates: Is thematic analysis a realist method? Just as IPA (Smith, Flowers & Larkin, 2009) could be regarded as a form of phenomenological TA, many proponents of TA demarcate it as a phenomenological (contextualist) or a realist method (Guest, et al., 2012, Joffe, 2011). For example, although Joffe (2011: 211) acknowledges that TA “is not tied to a particular theoretical outlook”, she describes it as fitting well with the assumptions of social phenomenology. Furthermore, Joffe’s version of TA involves the development of a ‘coding frame’, which two or more independent coders then apply to the dataset; the researcher seeks to determine the reliability of the coding frame by calculating inter-rater reliability scores. Likewise, Guest et al.’s (2012: 18) ‘applied’ TA is “more comfortably applied within a positivist framework” and involves the use of a ‘code book’ (similar to Joffe’s coding frame) and the calculation of ‘intercoder agreement’ (another term for inter-rater reliability). Boyatzis’ (1998) TA aims to bridge the divide between positivist and interpretative social science and again advocates 3 the use of a codebook, multiple independent coders and the calculation of inter-rater reliability scores. All are realist in assuming there is an accurate reality in the data that can be captured through coding. By contrast, Braun and Clarke’s (2006) theoretically flexible TA can be applied equally within an essentialist or a constructionist framework. Coding is a more organic process, not guided by the development of a coding frame. Reflecting a critical approach to coding, it is understood as an active and reflexive process that inevitably and inescapably bears the mark of the researcher(s). With no one ‘accurate’ way to code data, the logic behind inter-rater reliability disappears (it can be argued that it shows that two researchers have been trained to code data in the same way, but not that coding is accurate). A comparison of two studies (Lavie & Willig, 2005; Nicolson & Burr, 2003) on a similar topic (women’s experiences of [lack of] orgasm) illustrates the differences between theoretically flexible TA and phenomenological approaches. Nicolson and Burr’s (unclaimed) TA is explicitly located within a critical feminist framework (with the aim of challenging the medicalisation of women’s sexual problems) and draws on the conceptual lens of symbolic interactionism and that of psychodynamism in analysing 33 (mainly heterosexual) women’s interview accounts. The data are clearly located within (a feminist interpretation of) the wider socio-cultural context. The authors identified three (latent) themes that underpinned the women’s accounts: male sexuality as active and demanding of sexual satisfaction; female sexuality as active but different from male sexuality; and the notion that there is a ‘normal sexuality’, which exists as a standard for women to judge themselves against. The analysis examined the ways in which the women’s accounts constitute the social ‘reality’ of orgasm in particular ways. By contrast Lavie and Willig’s (2005) IPA, or phenomenological TA, analysed the psychological experience of inorgasmia for 6 heterosexual women and focused (at the semantic level) on the ways in which inorgasmia impacted on the women’s self-image, their partner relationships and their enjoyment of sex. Although, Lavie and Willig note the symbolic interactionist underpinnings of IPA and recognise that “meanings are negotiated within a social context” (p. 118), references to wider discourses of sexuality and gendered norms and expectations are limited (for example, there is “a stigma attached to inorgasmic women in our society”, p. 120). Although, Lavie and Willig’s analysis could be regarded as a departure from mainstream psychology in the sense that it prioritises the 4 voices of inorgasmic women, rather than researcher’s categories, it is not sufficiently located within wider discourses of (hetero)sexuality nor a productive role of language to be regarded as an example of a critical psychology of orgasm. Critical debates: Critical thematic analysis or discourse analysis? What are the distinctive features of critical TA compared to discourse analysis, and why would a critical psychologist chose critical TA over discourse analysis? For the novice critical researcher, TA is a far more accessible method; it has a clearly defined set of procedures (see below) rather than relying on more ill-defined analytic practices such as ‘craft skills’. Furthermore, although critical TA recognises the constitutive nature of language and discourse, it does not generally involve a micro analysis of language use; therefore it does not require a technical knowledge of language practice (unlike discursive psychology and conversation analysis). Critical TA can be applied flexibly within a wide range of critical frameworks (from symbolic interactionism and psychodynamism [Nicolson & Burr, 2003] to feminist poststructuralism [Farvid & Braun, 2006]) and, as noted above, can be used to analyse almost all forms of qualitative data, both small and large data-sets, and to address most of the different types of research question posed by critical psychologists. Six-phase process Braun and Clarke’s theoretically flexible TA involves a six-phase process preceded by critical reflection on the ‘many questions’ of TA. The ‘many questions’ include whether TA will be used in a (broadly) experiential framework or a critical one, and whether analysis will be ‘inductive’ (working bottom-up from the data but recognising that pure induction is never possible; the researcher always brings their prior knowledge and epistemological assumptions to the data) or guided by existing theoretical concepts (top-down). Coding and theme development in TA is always guided by a research question, but this question can be developed and refined throughout the analytic process. Phase one of the analytic process involves familiarisation with the data. The researcher must become intimately familiar with their data, reading and re-reading the data (and listening to any audio data at least once). This process should be accompanied by note-taking, recording initial analytic observations about the data and potential avenues for further exploration. Phase two requires a thorough and systematic coding of the data. Codes can evolve across the coding process and 5 should capture features of the data that are potentially relevant to the research question. As noted above, codes can be developed at both the semantic and latent level, encapsulating both the surface meaning of the data (interpreted through the lens of the researcher’s epistemological and theoretical assumptions) and the “underlying systems of meaning” (Taylor & Ussher, 2001: 297). The code itself should be a pithy phrase that evokes the relevant features of the data and (potentially) the researcher’s interpretative lens. Because themes are initially developed from the coded data, codes should ‘work’ independently of the data and the researcher should end this phase by collating all their codes, and all the data relevant to each code. The process of theme development commences in phase three with the generation of a set of potential or ‘candidate’ themes. The codes and coded data are examined to identify similarity and overlap – patterns of semantic meaning and/or underlying concepts; codes are clustered together to form themes, or larger, more complex codes can be ‘promoted’ to themes. The researcher should also reflect on the relationships between potential themes. Themes must work together to tell a rich and complex story about the data in relation to the research question. The researcher ends this phase with a set of candidate themes, all the relevant coded data collated for each theme, and, ideally, with a figurative representation of the relationship between themes (and codes) – a ‘thematic map’ of the analysis. Phase four is particularly important for the novice researcher or the researcher who is working with a larger data set and involves two levels of review of the candidate themes. First the themes are reviewed against the coded data to check if there is a good ‘fit’ between the themes and the coded data and the themes tell a convincing and coherent story about the coded data. If the researcher judges the themes to ‘work’ at the level of the coded data then they progress to a review of the themes against the full data-set. As the researcher gains confidence, theme development becomes a more recursive process (particularly when working with smaller data sets), flip-flopping between theme development and review until the researcher is satisfied that the individual themes are coherent and the themes together address the research question in a meaningful way and capture the most relevant features of the data. This review process may entail some level of revision to the candidate themes, or the candidate themes may be completely discarded and the process of theme development begun again. Once the researcher has decided upon 6 a (relatively) definitive set of themes the analysis progresses to the final analytic phases of defining and naming themes and writing up (the process of revision, or rejection, of themes can continue into these final phases). The process of writing theme definitions is also particularly useful for novice researchers and facilitates the development of a rich and complex analytic narrative. Theme definitions should tell the story of each theme, its central concept, scope and boundaries and how it relates to the other themes and to the research question. The theme name should encapsulate the ‘essence’ of the theme and be concise and vivid (short data extracts can form excellent theme titles if they capture the central concept of the theme). ‘Writing up’ is the final phase in the analytic process; however there is no clear separation between analysis and writing. In practice, writing-up involves assembling, editing and (new) writing, and further analysis, organisation and re-organisation of the themes and relevant selected data extracts. Presented data extracts, whether treated as illustratively or analytically (Braun & Clarke, 2012b), should provide clear and compelling evidence to support analytic claims. International relevance Although some countries have stronger and more diverse traditions of qualitative and critical psychology than others (for example, Northern Europe and Australasia have a much stronger tradition than North America), TA is clearly an international method, with proponents in the UK (e.g., Braun & Clarke, 2006, Joffe, 2011) and in the US (e.g., Aronson, 1994, Boyatzis, 1998, Guest et al., 2012) and practitioners in these and many other countries. Future directions Although theoretically flexible TA is used by some critical psychologists (see, for example, Farvid & Braun, 2006, Nicolson & Burr, 2003), discourse analysis remains the dominant method of critical psychology and many continue to perceive TA as a realist method. Furthermore, TA is most often used within a realist framework and as a descriptive rather than an interpretative or conceptual method. Moving beyond these limitations, theoretically flexible TA has much to offer critical psychology. 7 References Aronson, J. (1994). A pragmatic view of thematic analysis. The Qualitative Report, 2(1), http://www.nova.edu/ssss/QR/BackIssues/QR2-1/aronson.html. Boyatzis, R. E. (1998). Transforming qualitative information: Thematic analysis and code development. Thousand Oaks, CA: Sage. Braun, V. & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3, 77-101. Braun, V. & Clarke, V. (2012a). Successful qualitative research: A practical guide for beginners. London: Sage. Braun, V. & Clarke, V. (2012b). Thematic analysis. In Cooper, H. (Ed.), The Handbook of Research Methods in Psychology. Washington, DC: American Psychological Association. Farvid, P. & Braun, V. (2006). ‘Most of us guys are raring to go anytime, anyplace, anywhere’: Male and female sexuality in Cleo and Cosmo. Sex Roles, 55, 295-310. Guest, G., MacQueen, K. M. & Namey, E. E. (2012). Applied thematic analysis. Thousand Oaks, CA: Sage. Joffe, H. (2011). Thematic analysis. In D. Harper & A. R. Thompson (Eds.), Qualitative methods in mental health and psychotherapy: A guide for students and practitioners (pp. 209-223). Chichester: Wiley. Joffe, H. & Yardley, L. (2004). Content and thematic analysis. In D. F. Marks & L. Yardley (Eds.), Research methods for clinical and health psychology (pp. 56-68). London: Sage. Lavie, M., & Willig, C. (2005). ‘I don’t feel like melting butter’: An interpretative phenomenological analysis of the experience of ‘inorgasmia’. Psychology & Health, 20(1), 115-128. Merton, R.K. (1975). Thematic analysis in science: notes on Holton’s concept. Science as Culture, 188(4186), 335-338. Nicolson, P., & Burr, J. (2003). What is 'normal' about women's (hetero)sexual desire and orgasm?: a report of an in-depth interview study. Social Science & Medicine, 57, 1735-1745. 8 Reicher, S. (2000). Against methodolatry: Some comments on Elliot, Fischer and Rennie. British Journal of Clinical Psychology, 39, 1-6. Smith, J. A., Flowers, P. and Larkin, M. (2009). Interpretative phenomenological analysis: Theory, method and research. London: Sage. Taylor, G. W. & Ussher, J. M. (2001). Making sense of S&M: A discourse analytic account. Sexualities, 4(3), 293-314. Online resources Braun and Clarke thematic analysis website [currently in development; url to be added ASAP] 9