Table of Contents
Thematic analysis is a qualitative method for identifying, analysing, and reporting patterns (themes) within data. Braun and Clarke’s widely used six-phase framework moves from familiarisation through coding, theme searching, reviewing, defining, and writing up, turning transcripts into structured findings.
Developed by psychologists Virginia Braun and Victoria Clarke in 2006, the method is now standard across psychology, health research, education, and social science dissertations. It offers a flexible, theoretically adaptable approach rather than a rigid formula, useful for first-time qualitative researchers.
This guide walks through each of the six steps in order, with a worked example, a copyable coding template, and practical tips for avoiding the mistakes markers flag most often in dissertation methodology chapters.
Thematic analysis examines qualitative data, such as interview transcripts, focus group recordings, open-ended survey responses, or documents, to find recurring patterns of meaning. It interprets what themes mean for the research question, rather than only counting frequency.
It sits within the broader landscape of qualitative approaches described in our guide to research methodology, alongside grounded theory, discourse analysis, and narrative analysis. Thematic analysis is often chosen for its flexibility and relative accessibility to new researchers.
Before analysis can begin, researchers need well-organised data. If you are still planning interviews, surveys, or observations, see our guide to data collection for methods, sampling, and ethical considerations to follow.
Braun and Clarke (2006) set out six phases that move iteratively, not strictly in a straight line, from raw data to a written report. Each step builds on the last, and researchers often revisit earlier phases as understanding deepens.
| Step | Name | What Happens |
|---|---|---|
| 1 | Familiarisation | Read and re-read data; note initial ideas |
| 2 | Generating codes | Label interesting features systematically across the dataset |
| 3 | Searching for themes | Group codes into candidate themes and sub-themes |
| 4 | Reviewing themes | Check themes against coded extracts and the full dataset |
| 5 | Defining and naming | Refine the specifics and essence of each theme |
| 6 | Writing up | Produce a report with evidence and analytic narrative |
Read and re-read the entire dataset before coding anything. Note initial ideas, surprising quotes, or recurring language in the margins. Transcribing your own audio recordings, rather than outsourcing this, often deepens familiarity with tone and emphasis.
This phase has no shortcuts. Skimming transcripts leads to shallow coding later, so budget real time here, particularly for longer interview sets or multiple focus groups collected during fieldwork.
Familiarisation is also when researchers first notice tensions or contradictions across participants, which often signal richer analytic territory. Keep a simple log of first impressions to compare against later, more structured coding decisions.
Systematically work through the dataset, labelling features relevant to your research question. Codes should be short, descriptive tags, such as ‘fear of failure’ or ‘lack of support’, applied consistently across every transcript or document.
Coding can be done by hand with highlighters and margin notes, or in qualitative software such as NVivo or ATLAS.ti. Code as many potential patterns as possible; irrelevant codes can be dropped later.
Aim for depth over speed. Braun and Clarke recommend coding as inclusively as possible in this phase, capturing anything potentially relevant, since a code dismissed too early can mean losing a genuinely important pattern.
Once coding is complete, sort codes into broader candidate themes and sub-themes. A theme captures something important about the data in relation to the research question, not simply a repeated word or phrase.
Visual tools help here: thematic maps, mind maps, or simple tables grouping codes under tentative theme headings. Some codes may form a standalone theme; others may become sub-themes within a larger pattern.
Some analysts print every coded extract and physically sort them into piles by candidate theme. Others use software’s node or code-grouping features. Either method works, provided you stay systematic and keep a decision record.
Review candidate themes against the coded extracts first, then against the entire dataset. Ask whether each theme is supported by enough data, whether themes overlap too closely, and whether any distinct pattern has been missed.
This is an iterative check, not a single pass. Themes may need splitting, merging, or discarding entirely if they do not hold up once re-examined against the full transcript set.
Braun and Clarke describe two levels of review: checking coherence within each theme, then checking that themes together tell an accurate, coherent story of the entire dataset without excessive overlap or repetition.
Write a detailed analysis for each theme, identifying its ‘essence’, what it captures, and how it relates to the wider research question. Give each theme a short, informative name that readers understand immediately.
Avoid vague labels like ‘Theme One’. A name such as ‘Navigating Isolation as a First-Year Student’ tells a reader far more than a generic heading and signals the analytic story you are building.
Write a short paragraph under each theme name summarising its scope and boundaries. This becomes the backbone of your findings chapter and clarifies, for you and your supervisor, exactly what each theme includes.
The final report weaves theme definitions together with vivid, compelling data extracts and an analytic narrative that goes beyond description. Link findings back to the research question and existing literature throughout.
For dissertations, this typically becomes the Findings or Analysis chapter, followed by a separate Discussion chapter interpreting what the themes mean for theory, practice, or policy in your field.
Quote extracts should be chosen for clarity and representativeness, not simply because they are dramatic. Always anonymise participants with pseudonyms or codes, and check your extracts against your institution’s ethics approval conditions.
Inductive thematic analysis lets themes emerge directly from the data, with no pre-existing coding framework. Deductive (or theoretical) thematic analysis starts from a theory or research question and codes data against it specifically.
Most dissertations use a hybrid: broadly inductive coding, refined against existing theory during the reviewing and defining stages. State your approach explicitly in your methodology chapter, and justify the choice against your research question.
In 2019, Braun and Clarke reframed their approach as ‘reflexive thematic analysis’, emphasising the researcher’s active, subjective role in generating themes rather than themes simply ’emerging’ neutrally from data.
This update pushed back against treating coding as a mechanical, checklist exercise with fixed reliability measures. Reflexivity means acknowledging your own assumptions and positionality shape which patterns you notice and prioritise.
For dissertations, briefly stating which version, and how you approached coding reliability or reflexive engagement, strengthens your methodology chapter and shows examiners you understand current debates in qualitative research.
NVivo and ATLAS.ti are the most widely used paid tools in UK universities, offering coding, memo-writing, and visualisation features, often available through institutional licences via your university’s software portal.
Free alternatives include Taguette and simple spreadsheet-based coding in Excel or Google Sheets, which work well for smaller datasets or first dissertations where a paid licence is not accessible.
Software does not perform the analytic thinking for you. It organises codes and extracts; the interpretive work of naming and defining themes remains a human, judgement-based task throughout the process.
Keep a codebook recording each code’s definition and an example extract, so labelling stays consistent across a large dataset or when multiple coders are involved in the same project.
Distinguish semantic codes, which stay close to the explicit content, from latent codes, which capture underlying assumptions or ideology. Braun and Clarke’s method supports both, but be consistent about which level you are working at.
If working with a supervisor or research team, consider inter-rater reliability: have a second person code a sample independently, then compare and discuss discrepancies before finalising the coding frame.
Reflect briefly in a research diary after each coding session. Noting decisions and doubts as you go makes the audit trail required for rigorous, transparent qualitative research much easier to write up later.
Treating themes as simple summaries of interview questions, rather than patterns of meaning across the dataset, is the most common marker complaint in qualitative dissertation chapters submitted for review.
Other frequent issues include incomplete data familiarisation, themes that overlap too heavily, insufficient supporting quotes, and skipping the review phase entirely under time pressure near a submission deadline.
Relying on a single quote to support an entire theme weakens credibility. Aim for multiple extracts, ideally from different participants, showing the theme recurs meaningfully across the dataset rather than one outlier.
Keep an audit trail: dated versions of your codebook and thematic map demonstrate rigour to examiners and make the analysis process defensible if questioned at viva or during marking.
Qualitative dissertation chapters are demanding to structure clearly under word-count pressure. Our team supports UK students with methodology guidance, data analysis feedback, and structuring advice across every stage of a dissertation.
For quantitative chapters alongside qualitative coding, our statistical analysis service supports SPSS, R, and Excel-based work. Browse further guidance in our dissertation guide and study skills categories for related resources.
Thematic analysis is a qualitative research method for identifying, analysing, and reporting recurring patterns (themes) within data such as interview transcripts or open-ended survey responses. Braun and Clarke’s six-step framework, developed in 2006, is the most widely taught version across UK psychology, health, and social science dissertations.
The six steps are: familiarising yourself with the data, generating initial codes, searching for themes, reviewing themes, defining and naming themes, and writing up. The phases are iterative rather than strictly linear, so researchers commonly revisit earlier steps as their analytic understanding deepens through the process.
Inductive thematic analysis lets themes emerge directly from the data with no pre-existing coding framework, suiting exploratory questions. Deductive, or theoretical, analysis codes data against an existing theory or research question instead. Many UK dissertations combine both approaches, coding inductively then refining findings against relevant literature.
Work systematically through every transcript, labelling meaningful features with short, consistent codes such as ‘fear of failure’ or ‘lack of support’. Keep a codebook defining each code with an example extract, then group related codes into candidate themes during the searching-for-themes phase of the process.
A code is a short label attached to a specific data extract, such as a phrase or sentence describing one idea. A theme is a broader pattern built from several related codes, capturing something meaningful about the research question across the whole dataset rather than one isolated moment.
NVivo and ATLAS.ti are the most common paid tools, with institutional licences available at most UK universities through the software portal. Free alternatives include Taguette or manual coding in Excel and spreadsheets, which work well for smaller datasets, first dissertations, or when a paid licence is not accessible.
You May Also Like
All work is written by human writers. 100% AI free, guaranteed.
100% money back guarantee if you find plagiarism in our work.
COMPANY DETAILS
