Thematic analysis is a method for analysing qualitative data that entails searching across a data set to identify, analyse, and report repeated patterns (Braun & Clarke, 2006). It is considered a good first analytic method for novice qualitative researchers to master. It is an appropriate and powerful method to use when seeking to understand a set of experiences, thoughts, or behaviours across a data set (Braun & Clarke, 2012). It is designed to search for common or shared meanings, and therefore less suited for examining unique meanings or experiences from a single person or data item (Kiger & Varpio, 2020, p. 2). Through a thematic analysis, the researcher constructs themes to reframe, reinterpret, and/or connect elements of the data, which means that themes are not merely organisational tools, used to classify and label data, and that the thematic analysis goes further into the interpretation and data transformation processes, but not to the point of developing a theory, which is the main goal of the grounded theory (Glaser & Strauss, 1967).
In thematic analysis, the theme is a patterned response or meaning derived from the data that informs the research question (Braun & Clarke, 2006, p. 82). It is more abstract than a category, and researchers can identify themes irrespective of the number of times a particular idea or item related to that theme appears in a data set. So, according to the above-mentioned authors, a theme might be given a considerable space in some data items, and little or none in others, or it might appear in relatively little of the data set. Therefore, it is upon the researcher to determine what a theme is, and researchers have great flexibility when identifying themes. They can use an inductive or deductive approach to theme identification (Braun & Clarke, 2012, p. 12). The former derives themes from the researcher’s data, and is not necessarily reflective of the researcher’s theoretical interests or beliefs on the subject. On the other hand, a deductive approach uses a pre-existing theory, framework, or other researcher-driven focus to identify themes of interest (Braun & Clarke, 2012, p. 12).
According to Boyatzis (1998), there are two levels at which themes can be identified: semantic or manifest, addressing more explicit or surface meanings of data items, and latent or interpretative, reflecting deeper meanings, assumptions or ideologies. So, semantic themes are identified without looking for anything beyond what a participant has said or what has been written, whereas latent themes are the result of examination of underlying ideas, assumptions, ideologies that shape or inform the semantic content of the data.
The most widely adopted method of thematic analysis consists of six steps (Clarke and Braun, 2017). It is a recursive, rather than linear process, in which subsequent steps may prompt the researcher to circle back to earlier steps in light of new data or newly emerging themes that merit further investigation (Kiger & Varpio, 2020, p. 3). The steps are as follows:
- familiarising oneself with the data – with the entire data set, which requires repeated and active reading through the data. Though time-consuming, the transcription of audio recordings is an excellent way of getting familiar with the data.
- generating initial codes – a code is the most basic segment of raw data that can be assessed in a meaningful way regarding the phenomenon (Boyatzis, 1998, p. 63). A code must be sufficiently well-defined in order not to overlap with other codes, and should fit logically within a larger coding framework/template. Once the coding template is defined, researchers apply the same codes to the entire data set by labeling data extracts with relevant codes, making notes of any potential patterns or connections between items that might inform subsequent theme development. A single extract can be labeled with multiple codes if relevant (Braun & Clarke, 2006).
- searching for themes – the coded extracts are examined in order to find potential themes of broader significance. Braun and Clarke (2012) offer an analogy that, if an entire analysis is seen as a house, the individual codes are bricks and tiles, and themes are the walls and the roof. So, themes do not simply emerge from the data – they are constructed by the researcher through analysing, combining, comparing, and even graphically mapping how codes relate to one another. In an inductive analysis, researchers derive themes from the coded data, and they are reflective of the data set, whereas in a deductive analysis, the theme development is informed by predefined theories, and these themes focus more on a particular aspect of the data set or a specific question of interest (Braun & Clarke, 2006). Thematic maps help to visually demonstrate cross-connections between themes and sub-themes. The researcher should be inclusive at this point, and make note of any theme of potential significance regardless of whether it is directly related to the research question, and regardless of the quantity of data that falls under them (Kiger & Varpio, 2020, p. 5). Researchers can even create a miscellaneous theme to incorporate the codes that do not fit well within the theme template.
- reviewing themes – the researcher looks at coded data placed within each theme to ensure that they are coherent in supporting the theme, that they have enough commonality, but are different enough to merit separation. So, data extracts can be re-sorted at this point, and themes can be modified, added, combined, divided, discarded, to better reflect the coded data (Kiger & Varpio, 2020, p. 6). The researcher should keep detailed notes regarding their thought processes and decisions made on how themes were developed, modified, removed. When the researcher decides that the thematic map properly covers all of the coded data, they start checking whether individual themes fit meaningfully within the data set, and whether the thematic map accurately represents the entire data set (Braun & Clarke, 2006). The thematic map should clearly demonstrate how themes interrelate. The researcher has to re-read the entire data set to re-examine the themes and re-code for additional data newly created or modified in this phase, and then revise the thematic map accordingly (Braun & Clarke, 2006), thus confirming the recursive nature of the thematic analysis (Kiger & Varpio, 2020, p. 7).
- defining and naming themes – a narrative description of each theme is created. The names of themes are then reviewed to ensure they are brief and sufficiently descriptive. The overlapping areas of themes are identified, as well as sub-themes. Data extracts that illustrate the key features of themes, and are to be presented in the final report, should be selected at this stage, and narratives about them should be created (Braun & Clarke, 2012).
- producing the report/manuscript – writing up the final analysis and description of findings, which is the continuation of the analysis and interpretation that have already been done (King, 2004, p. 267). Both narrative descriptions and representative data extracts (direct quotations from participants) should be used. The discussion section can broaden the analysis by relating themes to larger questions, discussing implications of findings, and questioning the assumptions and preconditions that gave rise to the themes (Braun & Clarke, 2016). Referencing the related literature can also add to the strength of the analysis by building support for why particular themes were selected, and situating findings within the existing body of literature.
Thematic analysis is simple to master and apply. It is a powerful method for analysing data, which allows researchers to interpret a wide range of data sets. The flexible nature of this analysis can make it difficult for some researchers to determine which aspects of data to focus on, and which theoretical frameworks to use for their analysis.