You have finished your interviews. You have hours of recordings, dozens of pages of transcripts and a growing sense that you do not know where to begin. This is the point where many qualitative dissertations stall. The data feel enormous, your supervisor has said “find the themes”, and the word “themes” suddenly seems impossible to pin down.
Thematic analysis is the most widely used method for analysing qualitative data in UK dissertations, partly because it is flexible and partly because it is learnable. This guide explains what thematic analysis is, walks through the six phases, shows how to carry them out in NVivo, and covers writing up, quality and mistakes to avoid. For support across the whole project, see our main guide to dissertation help UK.
What is thematic analysis?
Thematic analysis is a method for identifying, analysing and reporting patterns of meaning, called themes, across a qualitative dataset. It can be applied to interview transcripts, focus groups, open-ended survey responses, diaries, documents and social media text.
Its great advantage is that it is not tied to one theory. You can use it with an interpretivist, critical realist or constructionist approach, and with inductive or deductive coding. Its great risk is that this flexibility lets weak analysis hide behind a vague label. Markers want to see a clear, systematic process and a genuine analytic argument.
The approach most commonly cited in UK dissertations is reflexive thematic analysis, developed by Virginia Braun and Victoria Clarke, whose six-phase framework (first set out in a 2006 paper in Qualitative Research in Psychology) is widely used and later refined in their own writing. Their thematic analysis resource site is a trustworthy place to read their guidance directly, and you should cite their work rather than a summary of it. Other versions exist, such as codebook and coding-reliability approaches, and your choice should match your philosophy and your question.
Is thematic analysis right for your dissertation?
It suits projects that ask how or why questions about experiences, perceptions, practices or meanings, for example:
- How do newly qualified nurses describe their first six months on acute wards?
- How do employees experience motivation under hybrid working?
- How do first-year students perceive online peer feedback?
It is less suited to questions about the structure of discourse (consider discourse analysis), the detailed lived meaning of a single experience (consider interpretative phenomenological analysis) or theory building from scratch (consider grounded theory). If you are still deciding on your overall approach, read qualitative vs quantitative dissertations and our guide to the dissertation methodology chapter.
Key terms you must be able to use precisely
| Term | Meaning | Example |
|---|---|---|
| Data item | One piece of data | One interview transcript |
| Data set | All the data collected | All twelve transcripts |
| Code | A short label capturing something meaningful in the data | “Missing informal chat” |
| Theme | A pattern of shared meaning built around a central idea | “Connection has to be deliberately made” |
| Subtheme | A distinct facet within a theme | “Scheduled social time feels forced” |
| Codebook | A record of codes with definitions | Table in your appendix |
| Reflexivity | Reflecting on how your position shapes analysis | Research diary entries |
The most important distinction is between codes and themes. A code labels a piece of data. A theme is a bigger idea you construct by bringing codes together around a shared meaning. Themes do not sit in the data waiting to be found; you build them through interpretation.
Decisions to make before you start
Inductive or deductive?
- Inductive: codes are developed from the data, with no pre-set framework. Common in exploratory projects.
- Deductive: codes are guided by existing theory or a framework. Useful when testing or applying a model.
- Many projects combine both, but say so.
Semantic or latent?
- Semantic: coding the explicit, surface meaning of what participants say.
- Latent: coding underlying assumptions, ideas and meanings beneath the surface.
Essentialist or constructionist?
Are you treating the data as reflecting participants’ reality, or as showing how reality is socially constructed? Your answer should align with your methodology chapter.
Write these decisions down early. They shape every later step and you will need to justify them.
The six phases of thematic analysis
| Phase | What you do | Output |
|---|---|---|
| 1. Familiarisation | Read and reread the data; note initial ideas | Notes, memos, transcripts checked |
| 2. Generating codes | Code systematically across the whole dataset | A set of codes with data extracts |
| 3. Generating initial themes | Cluster codes into candidate themes | A first thematic map |
| 4. Reviewing themes | Test themes against the coded data and the full dataset | Refined themes, some merged, split or dropped |
| 5. Defining and naming themes | Write what each theme is about and name it | Theme definitions and names |
| 6. Writing up | Weave narrative and extracts into an argument | Findings chapter |
The process is not strictly linear. You will loop back, and that is part of the method.
Phase 1: Familiarise yourself with the data
Transcribe carefully. Transcription is already an analytic act. Decide on a level of detail (verbatim, including pauses and hesitations, or cleaned) and apply it consistently. If you use automated transcription software, check every transcript against the audio, and make sure the tool meets your ethics approval and data protection commitments, since recordings may contain personal data.
Read actively. Read each transcript at least twice. On the first pass, just absorb it. On the second, make margin notes about what seems interesting or surprising.
Keep a research diary or memos. Record your reactions, assumptions and early ideas. This supports reflexivity and gives you material for your methodology chapter and limitations section.
In NVivo:
- Create a new project and save it in a secure, backed-up location.
- Import transcripts through Import then Files, and give each a clear, anonymised name (for example P01, P02).
- Create cases for participants and add attributes such as role, years of experience or sector, if they are relevant and ethically appropriate.
- Use memos to record early reflections.
Phase 2: Generate initial codes
Coding means working through the dataset line by line or segment by segment and labelling anything relevant to your research question.
How to code well
- Code for meaning, not just topic. “Remote working” is a topic. “Feeling invisible to managers” is a code.
- Keep codes close to the data, using participants’ words where helpful.
- Code inclusively. Keep some surrounding context so extracts still make sense later.
- Code systematically. Give each transcript the same attention. Do not code only the parts that fit your hunches.
- Allow multiple codes on one segment where it carries several meanings.
- Code the unexpected, including data that contradicts your assumptions.
Example
| Extract | Code |
|---|---|
| “I never see anyone pass by my desk now, so nobody knows what I’m working on.” | Invisibility of effort |
| “The Friday video call is the only time we chat, and it feels forced.” | Scheduled social time feels artificial |
| “At home I can finally concentrate for hours.” | Uninterrupted focus at home |
In NVivo:
- Open a transcript. Select the text.
- Right-click and choose Code Selection at New Code, or drag the selection onto an existing code in the Codes area.
- Use clear, consistent names. Add a short description to each code as you go; this becomes your codebook.
- Check the Coding Stripes (View, then Coding Stripes) to see how coding overlaps across a document.
- Merge or rename codes as your understanding changes. Keep a record of major changes in a memo.
A caution on auto-coding. NVivo and similar software offer automated coding and AI-assisted features. These can help you explore data, but they cannot do interpretive analysis for you. Your analysis must be your own, and you should check what your university permits and report any tool use honestly.
Phase 3: Generate initial themes
Now shift from codes to patterns. Look at your full list of codes and ask: which belong together around a shared idea?
- Print or list your codes and physically group them, or use a mind map.
- Look for a central organising concept that unites a cluster, not just a shared topic.
- Think about relationships: some codes may become themes, some subthemes, some may be set aside.
- Keep a “miscellaneous” group for codes that do not yet fit. Do not force them.
Topic summary versus theme
This is the most common weakness in student work.
| Topic summary (weak) | Theme (stronger) |
|---|---|
| “Communication” | “Connection no longer happens by accident, and must be deliberately engineered” |
| “Work-life balance” | “Flexibility is experienced as freedom and as pressure to be always available” |
| “Technology” | “Digital tools both enable and erode a sense of being seen” |
A good theme makes a point. It has a story to tell about the data, and it connects to your research question.
In NVivo:
- Drag codes under parent codes in the Codes area to create a hierarchy of candidate themes and subthemes.
- Use the Mind Map or Concept Map visualisations to sketch relationships.
- Use Memos to write a short paragraph about each candidate theme.
Phase 4: Review themes
Here you test your candidate themes against the data at two levels.
Level 1: coded extracts. Read all the extracts under each theme. Do they form a coherent pattern? Are some off-topic? Is the theme too thin (not enough data), too broad (covers several ideas) or overlapping with another?
Level 2: the whole dataset. Return to the full transcripts. Does your thematic map accurately reflect the data overall? Have you missed something important? Does any theme dominate because of one or two participants?
Typical outcomes include:
- Merge two themes that say the same thing.
- Split a theme that holds two different ideas.
- Discard a theme with too little support.
- Promote a subtheme to a full theme.
- Recode parts of the data.
In NVivo:
- Open a code to view all coded references. Read them in context.
- Use Coding Query or Matrix Coding Query to see how themes relate to cases or attributes (for example how a theme is distributed across roles). Use this to explore, not to turn your analysis into counts.
- Use Compound Query or a Text Search to check whether a concept appears outside your coded areas.
Phase 5: Define and name themes
For each theme, write a short definition stating:
- What the theme is about, in one or two sentences.
- What is unique about it compared with other themes.
- How it answers, or contributes to answering, your research question.
Then name it. Good names are informative and often phrased as a short claim: “Visibility has to be earned remotely”, rather than “Visibility”.
Check the overall thematic map. Do the themes fit together as a coherent story? Is there a central organising idea? Aim for a manageable number. For most undergraduate and masters dissertations, three to five themes with subthemes is typical, but follow the guidance of your department and your data.
Example thematic map (illustrative)
| Theme | Central idea | Subthemes |
|---|---|---|
| 1. Autonomy as double-edged | Control over time brings freedom and pressure | Freedom to focus; Always-on expectations |
| 2. Connection must be engineered | Informal contact no longer happens by chance | Scheduled social time feels forced; Missing corridor conversations |
| 3. Visibility has to be earned | Effort is harder to see remotely | Proving productivity; Managers’ reliance on presence |
| 4. Recognition and belonging | Feeling valued depends on explicit acknowledgement | Praise from managers; Peer support |
These are placeholders to show structure. Your own themes must come from your own data.
In NVivo:
- Add a description to each parent code for your theme definition.
- Use Export Codebook (under Explore or Export options, depending on your version) to create a record for your appendix.
Phase 6: Write up the analysis
The findings chapter is where your analysis becomes visible. Its job is to tell a convincing story about the data, not to list themes.
A reliable structure for the chapter
- Introduction: remind the reader of the research question, give a brief overview of participants and present an overview of themes (a table or figure helps).
- One section per theme: a clear heading, a short framing paragraph, then analysis supported by extracts.
- Links between themes: show how they relate.
- Summary: pull the findings together and point to the discussion.
How to write each theme
A good pattern for each theme or subtheme is:
- State the point in your own words (the analytic claim).
- Present evidence: one or two well-chosen extracts, introduced and identified (participant code).
- Interpret: explain what the extract shows and why it matters.
- Connect: relate it to other themes and, in your discussion chapter, to the literature.
Example
Weak (descriptive, extract-led):
Many participants said they missed the office. P3 said, “I miss seeing people.” P7 said, “It’s lonely at home.”
Stronger (analytic):
Participants described the loss of informal contact not as a lack of social time but as a loss of the incidental ways in which they were noticed and included. As P3 put it, “I never see anyone pass by my desk now, so nobody knows what I’m working on.” Here, informal visibility is bound up with recognition: without chance encounters, effort has to be made visible deliberately, which several participants found tiring.
The second version makes an argument, uses the extract as evidence and moves toward interpretation. (Both examples are invented for illustration.)
Using quotations
- Choose extracts that are vivid and clearly illustrate the point.
- Keep them reasonably short, and trim with ellipses where appropriate, without distorting meaning.
- Introduce and interpret every quotation. Never leave one to speak for itself.
- Use quotations from a range of participants, not just one or two.
- Anonymise: use codes or pseudonyms, and remove identifying details.
Avoid counting where it does not belong
Reflexive thematic analysis does not rely on frequency to determine importance. Avoid claims like “60 percent of participants said…” with small qualitative samples, unless your approach explicitly supports it. Words such as “most”, “several” and “some” can be used carefully to indicate patterns, with a note on how you used them.
Showing quality and trustworthiness
Markers want evidence that your analysis was careful and credible. Useful strategies include:
- A clear audit trail: codebook, memos, screenshots of the coding structure and a description of how themes developed.
- Reflexivity: a short statement of how your background and assumptions may have shaped the analysis.
- Thick description: enough context for readers to judge transferability.
- Negative or deviant cases: acknowledge data that does not fit.
- Peer or supervisor discussion: talking through coding and themes with a supervisor or critical friend. In reflexive approaches, this is for deepening interpretation rather than testing coding agreement.
- Member reflections where appropriate and ethical.
Use the vocabulary of your approach: credibility, dependability, transferability and confirmability for qualitative work, as covered in our methodology chapter guide.
Ethics and data protection
Thematic analysis involves detailed handling of personal data. Make sure you:
- Follow the consent and confidentiality commitments in your ethics approval.
- Store recordings, transcripts and NVivo files securely, in line with UK GDPR and the Data Protection Act 2018.
- Anonymise transcripts and remove identifying details before sharing extracts or files with anyone.
- Check any transcription or analysis software against your ethics approval and university policy before using it.
- Follow retention and deletion plans.
Useful references include the BERA ethical guidelines, the British Psychological Society and the UK Research Integrity Office. See also our guide to dissertation ethics approval in the UK.
Getting NVivo set up and working well
Your university may provide NVivo through its software catalogue. The publisher’s own NVivo product pages give current features and learning resources. Practical tips:
- Back up often and keep a dated copy of your project file at key stages.
- Name things consistently. Participant codes, code names and memo titles should follow a pattern.
- Do not over-code. Hundreds of near-duplicate codes make theme building harder. Merge and tidy as you go.
- Use memos generously. They are the best source of analytic thinking for your write-up.
- Screenshot your coding structure at different stages. They make good appendix evidence of your process.
- Learn only the features you need. Coding, memos, hierarchy and a few queries are enough for most dissertations.
- NVivo does not analyse for you. It organises and retrieves; you interpret.
If you prefer manual coding in Word, Excel or on paper, that is also acceptable. Software is a tool, not a requirement.
Common mistakes
- Describing instead of interpreting. Extracts without analysis.
- Themes that are just topics, such as “Communication” or “Technology”.
- Using interview questions as themes. The structure of your interview guide is not an analysis.
- “Themes emerged from the data.” Themes do not emerge passively; you develop them. Say “I developed” or “themes were developed”.
- Too many themes, thinly supported.
- Cherry-picking quotations that fit a story while ignoring contradictions.
- Counting to claim importance in a small sample.
- Overlapping themes that say the same thing in different words.
- Ignoring your research question. Interesting themes that do not answer it.
- Poor process reporting. Not explaining how coding and theme development actually worked.
- Leaving analysis too late. It takes longer than most students expect.
- Claiming generalisability from a small purposive sample.
A realistic timeline
Thematic analysis is iterative and slow. As a rough guide for a dozen interviews:
| Stage | Typical time |
|---|---|
| Transcription and checking | Several hours per interview hour |
| Familiarisation and initial coding | 2 to 3 weeks |
| Theme development and review | 2 to 3 weeks |
| Defining themes and drafting findings | 2 to 3 weeks |
| Revision with supervisor feedback | 1 to 2 weeks |
Build buffers into your plan. If you are short on time, see what to do if you are behind on your dissertation deadline.
Thematic analysis advice for international and Nigerian students
Many students from Nigeria and across Africa study with UK universities, in person or through transnational programmes. A few points are worth noting.
- If interviews were conducted in another language, decide how you will transcribe and translate, and report how you checked that meaning was preserved. Where possible keep key terms in the original language with a translation.
- Reflect on positionality, especially if you are researching your own community or workplace.
- Treat context as data. Cultural, institutional and economic context may shape how participants talk, and your analysis can reflect that.
- Be explicit about your analytic choices. UK markers reward reasoning, not just description.
- Plan for time zones and connectivity when discussing your coding with a supervisor or mentor.
Our online dissertation support services are designed with these situations in mind.
Where to get help with your analysis
Your supervisor is your first port of call. See working with your dissertation supervisor for advice on productive meetings. Many universities also run NVivo training and research methods clinics. For additional one-to-one guidance on coding, theme development and write-up, our dissertation data analysis help service supports students in understanding and carrying out their own analysis, within your institution’s rules. If your project is quantitative or mixed, our SPSS for dissertations guide covers the statistical side.
Academic integrity and your analysis
Your analysis must reflect your own interpretation of your own data, reported honestly. Describe the process you actually followed, cite the method you draw on, and do not present AI-generated or borrowed analysis as your own. Learning from model structures and seeking feedback are sensible, as long as you follow your university’s rules on outside help. For more, see how to avoid plagiarism in your dissertation and the QAA’s UK Quality Code for the standards UK institutions work to. For citation formats, Cite Them Right is a trusted reference.
Thematic analysis checklist
- I have chosen thematic analysis for reasons linked to my question and philosophy.
- I have decided and recorded whether coding is inductive or deductive, semantic or latent.
- Transcripts are accurate, anonymised and securely stored.
- I read every transcript closely and kept a research diary.
- I coded the whole dataset systematically.
- My codebook has clear names and descriptions.
- My themes are built around central ideas, not just topics.
- I tested themes against the coded extracts and the whole dataset.
- Each theme has a definition and an informative name.
- My findings chapter makes an argument and does not just list extracts.
- Quotations are introduced, interpreted and anonymised.
- I explain my process and show quality measures.
- Limitations and my own positionality are acknowledged.
- The analysis answers my research question.
Frequently asked questions
What is thematic analysis in a dissertation?
It is a method for identifying and interpreting patterns of meaning, called themes, across qualitative data such as interviews or open-ended survey responses. It is widely used in UK dissertations because it is flexible and systematic.
What are the steps of thematic analysis?
The commonly used six phases are familiarisation, generating codes, generating initial themes, reviewing themes, defining and naming themes, and writing up.
What is the difference between a code and a theme?
A code is a short label for a meaningful piece of data. A theme is a broader pattern of shared meaning that you build by bringing related codes together around a central idea.
How many themes should I have?
There is no fixed number. Three to five well-developed themes with subthemes is common for undergraduate and masters dissertations. Depth and coherence matter more than quantity.
Do I have to use NVivo?
No. NVivo is helpful for organising and retrieving data, but you can code manually or with other software. Choose what your university supports and what you can use confidently.
Can NVivo analyse my data for me?
No. It helps you organise, code, search and visualise, but interpretation is your job. Automated features can support exploration but should not replace your analysis.
How many interviews do I need for thematic analysis?
It depends on your design and aims. Many student projects use roughly 8 to 20 interviews, justified with reference to information power or sufficiency. Always follow your department’s guidance.
Should I count how often themes appear?
In reflexive thematic analysis, frequency does not determine importance. Avoid percentages with small samples. Use cautious language such as “several” or “most”, and explain how you used it.
How do I show that my analysis is trustworthy?
Keep an audit trail, be reflexive, provide thick description, discuss negative cases, and explain how themes developed. Use qualitative quality language such as credibility and dependability.
How long does thematic analysis take?
Longer than most students expect, often several weeks for a dozen interviews, including transcription. Start early and build in time for revision.
Can someone help me with my thematic analysis?
Yes, within your university’s rules. Supervisors, methods advisers and mentors can help you understand the process and check your reasoning. Writers.ng offers guidance and feedback so that the analysis you submit is your own.
Continue through the dissertation process
Related guides in our cluster:
- Dissertation methodology chapter
- Qualitative vs quantitative dissertations
- SPSS for dissertations: a step-by-step guide for beginners
- How to write a dissertation conclusion
- Dissertation ethics approval in the UK
Explore our wider support:
- Dissertation help UK: expert support from topic to submission (main hub)
- Dissertation data analysis help UK (SPSS, NVivo, R, Stata)
- Online dissertation support services UK
- Masters dissertation support UK
- Free dissertation resources UK
Stuck with your coding or themes?
Share your research question, your coding approach and where you are stuck. We will give honest, specific guidance on process, theme development and write-up, and leave the analysis in your hands.