Paul Scott
To use thematic analysis for dissertation interviews, read your transcripts closely, code relevant passages, and develop patterns that answer your research question. Then support each theme with interview evidence and explain your interpretation.
The hardest part is turning a collection of comments into a defensible analytical argument. Braun and Clarke’s six phases provide a practical route, starting with sustained engagement with your interviews.
Key Takeaways
- Codes identify relevant features of interview passages; themes connect codes through a shared meaning.
- Choose an approach that fits your research question, and explain how it shaped your decisions.
- Revisit earlier phases as your understanding develops.
- Present themes with contextualised quotations, contradictory evidence and clear limits on your claims.
Understand What Thematic Analysis Does
Thematic analysis identifies, analyses and reports patterns within qualitative data. It works with interview transcripts, focus groups, open-ended survey responses and product reviews; each is a distinct source of qualitative data.
Virginia Braun and Victoria Clarke’s 2006 methodological paper{target=”_blank” rel=”nofollow”} appeared in Qualitative Research in Psychology, 3(2), pages 77-101. Their framework remains a widely used starting point.
For Braun and Clarke, a theme connects shared, multifaceted meanings across at least some of a dataset through a central organising concept. Therefore, “university experiences” names a topic, but doesn’t explain a pattern.
Terminology also matters. Thematic content analysis has competing definitions{target=”_blank” rel=”nofollow”}: it can mean qualitative content analysis, thematic analysis itself (associated with Anderson, 1998), or a narrower approach using McClelland’s thematic units (1975).
Name the method you actually use. If you’re following reflexive thematic analysis, explain that approach rather than treating these labels as interchangeable.
Choose Your Analytical Approach
Decide how you’ll engage with the data before coding, whilst remaining open to revising your assumptions.
Inductive or Deductive Coding
Inductive analysis works bottom-up. You develop codes and themes through engagement with the interviews rather than starting with predetermined categories. Choose it when your question concerns participants’ experiences without requiring a particular theoretical framework.
Deductive analysis works top-down, guided by existing theory, research questions or a predefined codebook. It suits questions about how a particular concept operates in participants’ accounts.
Eval Academy describes deductive analysis as more common in evaluation and suggests a catch-all code such as “Unexpected info”. This gives material outside your initial framework somewhere to go.
However, actively search for evidence that challenges the framework. Otherwise, confirmation bias can turn coding into an exercise in finding expected answers.
Semantic or Latent Interpretation
Semantic analysis focuses on what participants explicitly say. Latent analysis examines underlying assumptions, ideas and conceptual subtexts.
These choices are separate from the inductive versus deductive distinction. You can conduct theory-guided semantic analysis or develop latent interpretations inductively.
Reflexive thematic analysis{target=”_blank” rel=”nofollow”} involves active interpretation. Even inductive work reflects your question, disciplinary background and decisions about what matters.
Begin with the Interviews: Phases 1 and 2
The six phases are recursive. Later interpretations often send you back to earlier transcripts, codes or provisional themes.
1. Familiarisation
Familiarisation (US spelling: familiarization) begins with reading every transcript closely. Listen again where tone or context affects meaning. Correct transcription errors, anonymise identifying details and preserve a secure original.
During rereading, note recurring concerns, tensions and differences between accounts. Separate what a participant said from your initial explanation.
Keep the research question visible, but avoid selecting only passages that support your preferred answer. Familiarisation covers the whole dataset, including material that complicates your expectations.
2. Generating Initial Codes
A code captures something relevant in a passage. Select enough surrounding text to preserve its meaning, and apply several codes to one extract when appropriate.
Code systematically across every interview. Compare new passages with earlier ones, revising labels when their meanings overlap or differ.
Maintain a master code list with definitions and linked extracts. In reflexive analysis, this list can change as your interpretation develops. A fixed codebook belongs to a different analytical arrangement.
Manual coding in a spreadsheet or a CAQDAS tool such as NVivo can help organise transcripts and retrieve extracts, but neither replaces your interpretation.
Also record transcript identifiers and line numbers. These connections let you return to the original account when reviewing a theme or selecting quotations.
Build Meaningful Patterns: Phases 3 and 4
A group of related codes becomes useful when you can explain what connects them.
3. Searching for Themes
Gather related codes and their extracts, then write a sentence describing their shared meaning.
Paul Willis’s Learning to Labour (1977) offers a published example for distinguishing analytical levels. His ethnography describes working-class boys’ opposition to school authority and rejection of academic expectations.
Those behaviours can inform codes concerning resistance to teachers and rejection of school success. Bringing them together requires an organising idea, such as the role of opposition in shared group identity.
The move towards a theme depends on that explanatory connection. Simply placing both codes under “School Experiences” leaves the analytical work unfinished.
4. Reviewing Themes
Review each candidate theme against its extracts and then against the complete dataset. Its evidence should form a coherent pattern, whilst its central idea should differ from neighbouring themes.
Merge overlapping themes, divide overloaded ones and discard those that don’t answer your question. Also examine contradictory accounts, which can reveal a theme’s boundaries.
A topic’s absence from an interview may reflect your questions rather than its absence from the participant’s experience.
Retain context when comparing accounts. Apparent disagreement may reflect different circumstances rather than incompatible views. Frequency alone doesn’t establish a theme’s importance.
Turn Themes into Findings: Phases 5 and 6
Writing helps sharpen the analysis, so begin writing up findings before every theme is settled.
5. Defining and Naming Themes
Write a short definition explaining each theme’s central concept, scope and contribution to your research question. Identify what belongs elsewhere.
Then choose a name that communicates the finding. “Relationships with Staff” identifies a subject; a stronger name states what the relationship means within your analysis.
Use subthemes when they develop distinct aspects of the same central concept. Avoid creating them merely to accommodate every remaining code.
A useful test is whether you can explain why the theme matters without listing its codes. If you can’t, its organising idea needs further work.
6. Writing the Report
In your methodology chapter, state which form of thematic analysis you used and describe what you actually did. Our dissertation methodology guidance explains how analysis fits alongside recruitment, data collection and interpretive limits.
In the findings chapter, introduce the organising principle, then build each section around a theme. State the analytical claim, contextualise an extract, and explain how it supports that claim.
Choose quotations for their analytical contribution. Identify speakers consistently, preserve anonymity and include relevant disagreement. Don’t leave quotations to explain themselves.
Our guide to organising dissertation findings covers thematic chapter structures. Separate findings and discussion chapters, or combine them where appropriate. Either way, distinguish interview evidence from broader theoretical explanation.
Remember that qualitative findings already involve interpretation. A separate discussion develops wider implications and comparisons with literature; it doesn’t make the findings interpretation-free.
Keep Your Analysis Transparent and Coherent
Reflexive thematic analysis treats subjectivity as a resource; reflexivity means examining how your position shapes interpretation.
Use analytical memos to record why you merged codes, changed a theme or rejected an explanation. Note where your expectations influenced an early reading and how contradictory evidence changed it.
These records create a traceable analytical process. Keep dated memos, evolving code lists and links between claims and extracts.
Intercoder reliability means asking a colleague to code the same dataset using your codebook, then assessing whether you captured the same features. That fits coding-reliability approaches, but isn’t a universal quality requirement for reflexive analysis.
Braun and Clarke’s guidance on good practice{target=”_blank” rel=”nofollow”} stresses methodological coherence. Collaborative discussion can enrich reflexive interpretation without seeking one objectively correct coding.
Before submission, address three common weaknesses:
- Replace interview-question headings with themes that communicate an analytical finding.
- Include evidence that challenges your preferred explanation, particularly in deductive work.
- Keep claims within the design’s limits: detailed interview accounts don’t establish population prevalence.
Frequently Asked Questions About Thematic Analysis
How Many Themes and Codes Should I Have?
Eval Academy suggests six to ten broad themes with subthemes and no more than four or five codes per theme. Treat these as practical suggestions for its approach, rather than requirements for every dissertation.
In reflexive analysis, the organising concept and analytical depth matter more than a numerical target. One theme may connect many codes. Don’t split a coherent theme merely to satisfy a suggested count.
Do I Need to Include Every Interview Quotation?
Include the evidence needed to support your claims, alongside relevant variation and contradictory accounts. A memorable quotation earns its place only when it contributes to the analysis.
Extensive supporting material can go in appendices where appropriate. However, the main chapter must contain enough evidence for readers to understand each theme without reconstructing your analysis themselves. Keep the speaker’s context clear wherever shortening an extract could alter its meaning.
Make Every Theme Answer Your Research Question
Strong thematic analysis connects careful coding with a clear explanation of what the interviews mean. The six phases support that work, but your decisions and evidence give it substance.
Before submitting, trace each theme back to its extracts and forward to your research question. A defensible theme has a clear organising concept, acknowledges variation and supports a proportionate conclusion.