Thematic Analysis
Thematic analysis is a qualitative data analysis method used to identify, analyse and interpret patterns of meaning, known as themes, across a dataset. It is commonly applied to interview transcripts, focus-group discussions, open-ended survey responses and other textual or qualitative data. The method is highly flexible, but this flexibility also means that researchers need to make and justify important methodological choices about how themes are developed, what level of meaning is analysed and what role the researcher plays in interpretation. Braun and Clarke describe thematic analysis as theoretically flexible, while their later work stresses that different forms of thematic analysis should not be treated as if they follow one universal set of procedures or quality criteria.
On this page
- Thematic analysis explained simply
- What thematic analysis is
- Types of thematic analysis
- Inductive vs deductive thematic analysis
- Semantic vs latent thematic analysis
- Codes vs themes
- The six phases of thematic analysis
- Thematic analysis vs content analysis
- How to report thematic analysis in a dissertation
- Application example
- Advantages and limitations
- Common mistakes
- Thematic analysis in business research
- Thematic analysis in the age of AI
- When to use thematic analysis
- Dissertation example
- Exam tip
Thematic Analysis at a Glance
| Decision | Question to ask | Main possibilities |
|---|---|---|
| Research purpose | What patterns of meaning need to be understood? | Experiences, perceptions, meanings, practices or responses |
| Analytical direction | What guides attention to the data? | More inductive ↔ more deductive |
| Level of meaning | At what level will meaning be interpreted? | Semantic ↔ latent |
| Researcher role | How is researcher interpretation understood? | Active/reflexive ↔ more structured and standardised coding |
| Approach to TA | What form of thematic analysis fits the study? | Reflexive, codebook or coding-reliability approaches |
| Analytical output | What should the analysis ultimately produce? | Coherent themes that address the research question |
Thematic Analysis Explained Simply
Imagine a hotel conducts 40 interviews with guests about their experiences. Reading the transcripts reveals repeated comments about slow check-in, helpful employees, room cleanliness, noise and how complaints were handled.
The researcher does not simply count how often each word occurs. Instead, related pieces of meaning are coded and examined for broader patterns. Comments about employees listening carefully, resolving problems quickly and following up afterwards might contribute to a theme such as “service recovery restores customer confidence.”
The theme is therefore more than a topic such as customer service. It expresses a meaningful pattern in the data that helps answer the research question.
What Is Thematic Analysis?
Thematic analysis is a method of systematically engaging with qualitative data in order to develop and interpret patterns of meaning relevant to a research question. Braun and Clarke originally presented it as an accessible and theoretically flexible approach and provided guidelines for conducting it rigorously.
This flexibility is one of the method’s greatest strengths, but it can also cause methodological confusion. Thematic analysis is not a single homogeneous technique. Different approaches make different assumptions about coding, researcher subjectivity, theme development and quality. Applying criteria from one approach indiscriminately to another can therefore produce an incoherent methodology.
For this reason, writing simply “the data were analysed using thematic analysis” is usually insufficient in a dissertation. The researcher should be able to explain what kind of thematic analysis was conducted, how the analysis was oriented, how themes were developed and why those choices were appropriate for the study.
Types of Thematic Analysis
A useful distinction can be made between reflexive, codebook and coding-reliability forms of thematic analysis. These approaches share an interest in patterns across qualitative data, but they do not necessarily conceptualise the analytical process in the same way. Contemporary methodological guidance specifically warns against treating the diversity within thematic analysis as insignificant.
Reflexive thematic analysis treats the researcher as an active participant in knowledge production. Themes are developed through sustained and reflexive engagement between the researcher, data, research question and theoretical assumptions. Researcher subjectivity is therefore not regarded simply as contamination that must be removed.
Codebook approaches make greater use of an organised coding framework. This can provide structure when several researchers are working with a dataset or when a project requires greater procedural consistency.
Coding-reliability approaches place greater emphasis on coding agreement and consistency between coders. Such approaches may use predetermined coding frameworks and measures or procedures designed to establish consistency.
These approaches should not be casually combined. For example, requiring inter-rater reliability as evidence that a reflexive thematic analysis is rigorous can conflict with the assumptions underlying reflexive TA. Braun and Clarke specifically identify the inappropriate application of coding-reliability standards to reflexive thematic analysis as a quality problem.
Dudovskiy Thematic Analysis Decision Framework
The following framework is designed to help researchers move beyond the vague statement “I will use thematic analysis” and identify a more coherent analytical strategy.
| Decision | Ask yourself | Methodological implication |
|---|---|---|
| 1. Research purpose | Am I trying to understand patterned meanings, experiences, perceptions or practices across qualitative data? | Establishes whether thematic analysis fits the analytical purpose |
| 2. Analytical direction | Should analysis be primarily data-driven or informed by existing concepts/theory? | Positions the study toward inductive or deductive analysis |
| 3. Level of meaning | Am I primarily analysing explicit meanings or underlying assumptions? | Positions the analysis toward semantic or latent interpretation |
| 4. Researcher role | Do I understand interpretation as an active part of knowledge production, or does the study require greater coding standardisation? | Helps clarify assumptions about subjectivity and coding |
| 5. TA approach | Which form of thematic analysis is consistent with the previous decisions? | Guides selection of reflexive, codebook or coding-reliability TA |
| 6. Quality criteria | What would count as good analysis within that particular approach? | Prevents incompatible quality criteria from being applied mechanically |
The framework does not imply that these six decisions are independent boxes that can be combined arbitrarily. Research philosophy, research questions, theoretical positioning, analytical procedures and quality criteria need to form a coherent whole.
The central principle is:
Do not choose a thematic analysis technique first and justify it afterwards. Start with the research purpose and methodological assumptions, then choose an approach that is consistent with them.
Inductive vs Deductive Thematic Analysis
Inductive thematic analysis develops patterns primarily through engagement with the data, whereas deductive thematic analysis is more explicitly directed by existing theory, concepts or analytical interests.
Suppose a researcher interviews employees about hybrid working.
With a more inductive orientation, the researcher may approach the transcripts without imposing a predetermined theoretical coding structure and develop patterns from issues that become analytically significant through engagement with employees’ accounts.
With a more deductive orientation, the researcher might analyse the same interviews using concepts derived from job demands-resources theory. Coding would then be more explicitly sensitised to issues such as workload, autonomy, social support and resources.
The distinction should not be treated as absolute. Real qualitative analysis can contain both inductive and deductive elements. What matters is transparency about what guided analytical attention and how theory interacted with interpretation. Recent worked methodological research likewise stresses the importance of aligning inductive/deductive choices with methodological positioning rather than treating them as merely technical coding options.
Semantic vs Latent Thematic Analysis
Semantic thematic analysis focuses primarily on meanings expressed relatively explicitly in the data, while latent thematic analysis moves further towards interpreting underlying ideas, assumptions or conceptual structures.
Consider an employee saying:
“Whenever the new system makes a decision, I check it myself because I don’t completely trust it.”
A semantic analysis might code this as lack of trust in automated decisions.
A more latent analysis might consider how the statement reflects underlying assumptions about human expertise, technological authority or perceived loss of professional control.
Neither level is automatically superior. The appropriate analytical depth depends on the research question, theoretical position and purpose of the study. Researchers should therefore avoid claiming latent interpretation merely because it appears more sophisticated; deeper interpretation needs methodological justification.
Codes vs Themes in Thematic Analysis
A code identifies something analytically relevant in a piece of data; a theme represents a broader pattern of shared meaning organised around a central idea.
The difference is important.
Imagine employees discussing the introduction of AI at work. Initial codes might include:
- fear of job loss
- uncertainty about future responsibilities
- reduced control over decisions
- pressure to learn new skills
- distrust of automated recommendations
Several related codes might contribute to a theme such as “AI adoption destabilises employees’ sense of occupational security.”
The theme does not merely group similar subjects. It makes an analytical claim about a patterned meaning across the dataset.
This distinction prevents a common problem: presenting broad interview topics—training, technology, management, communication—as if they were automatically themes. In reflexive thematic analysis particularly, themes are understood as analytical constructions developed through researcher engagement rather than entities waiting in the dataset to be discovered.
The Six Phases of Thematic Analysis
Braun and Clarke’s influential approach describes six phases of thematic analysis. These should be understood as recursive rather than a rigid six-step algorithm: analysis may require movement backwards and forwards as interpretations develop.
| Phase | What the researcher does | Main output | Critical judgement |
|---|---|---|---|
| 1. Familiarisation | Reads, rereads and engages closely with the dataset | Notes and initial observations | What appears analytically significant? |
| 2. Coding | Identifies and labels relevant features | Initial codes | Why is this feature relevant to the research question? |
| 3. Developing candidate themes | Examines relationships among codes and patterns | Candidate themes | What broader patterned meanings are being constructed? |
| 4. Developing and reviewing themes | Tests, revises, combines or separates themes | Coherent thematic structure | Does each theme work in relation to both the data and research question? |
| 5. Refining, defining and naming themes | Clarifies each theme’s central organising concept | Clearly defined themes | What exactly is this theme saying? |
| 6. Writing up | Integrates interpretation, evidence and literature into an analytical narrative | Research findings | How does the analysis answer the research question? |
Familiarisation is analytical rather than merely administrative. Repeated engagement with the data helps the researcher notice contradictions, relationships and potentially significant patterns.
Coding then reduces analytically relevant segments into concise labels, but codes are not final findings. During theme development, the researcher examines how codes relate and whether they can contribute to broader patterns of shared meaning.
Reviewing and refining themes is especially important. Candidate themes may turn out to be too broad, too thin, internally inconsistent or insufficiently relevant to the research question. Some may need to be merged, divided or abandoned.
Finally, writing is itself part of analysis rather than merely the presentation of decisions made earlier. The researcher selects appropriate extracts, develops interpretation and demonstrates how the themes contribute to answering the research question.
Thematic Analysis vs Content Analysis
Thematic analysis and content analysis can both be used to analyse qualitative material, but they should not be distinguished simply by saying that content analysis counts things while thematic analysis interprets them. Qualitative content analysis can also involve interpretation.
| Dimension | Thematic Analysis | Content Analysis |
|---|---|---|
| Primary focus | Patterns of meaning across a dataset | Systematic categorisation of content |
| Typical analytical output | Themes organised around central meanings | Categories, concepts and sometimes frequencies |
| Role of interpretation | Usually central | Varies according to form of content analysis |
| Quantification | Usually not the central objective | May be incorporated |
| Flexibility | Can operate across different theoretical positions | Can be qualitative, quantitative or mixed in orientation |
| Typical question | What patterned meanings characterize participants’ experiences? | What types of content occur and how can they be systematically classified? |
For example, if a researcher analyses 5,000 online customer complaints, content analysis might be particularly useful for systematically categorising complaint types and examining their prevalence. Thematic analysis may instead be chosen when the objective is to understand patterned meanings in how customers experience and interpret service failure.
The appropriate choice therefore follows from the research question and desired analytical output, rather than from the assumption that one method is inherently deeper or more rigorous.
How to Report Thematic Analysis in a Dissertation
A methodology chapter should provide enough information for the reader to understand what version of thematic analysis was used, why it was appropriate and how the analysis was actually conducted.
A strong account normally clarifies the analytical approach adopted, whether analysis was predominantly inductive or deductive, whether attention was primarily semantic or latent where relevant, how coding and theme development proceeded, the role of the researcher in interpretation, and how the chosen analytical practices were consistent with the broader methodology.
This matters because methodological transparency contributes to the reader’s ability to judge the credibility and coherence of qualitative analysis. Research on rigorous thematic analysis similarly emphasises documenting and disclosing the analytical process sufficiently for readers to understand how interpretations were produced.
Avoid a description such as:
“Interview transcripts were coded and themes emerged from the data.”
It conceals most of the important methodological decisions.
A stronger account explains who developed the themes, through what analytical process, under what methodological assumptions and in relation to which research question.
Application of Thematic Analysis: an Example
Consider a study investigating how employees experience the introduction of AI-based decision systems in financial-services organisations. The researcher conducts semi-structured interviews with 25 employees from different organisational roles.
During familiarisation and initial coding, analytically relevant extracts generate codes such as fear of replacement, uncertainty about accountability, AI saves routine work, distrust of recommendations, need to verify outputs, loss of decision autonomy and opportunity to focus on complex tasks.
These codes are not treated as findings in themselves. The researcher explores patterns among them. Fear of replacement, uncertainty about future roles and pressure to acquire new skills, for example, may contribute to a candidate theme concerning occupational insecurity during AI-enabled change.
Meanwhile, distrust of recommendations, need to verify outputs and uncertainty about accountability may contribute to another theme concerning negotiating authority between human judgement and automated decisions.
The researcher then returns repeatedly to the transcripts, checking whether the candidate themes meaningfully represent patterns across participants’ accounts, looking for contradictions and refining the boundaries between themes.
The final analysis might show that employee responses cannot adequately be described as simply resistance to AI. Instead, acceptance may depend on how employees perceive changes to professional identity, control, accountability and the value of their expertise.
The analytical contribution comes from that interpretation—not from the mere presence of frequently mentioned words.
Advantages and Limitations of Thematic Analysis
Thematic analysis is attractive because it can accommodate a wide variety of research questions and qualitative datasets without requiring commitment to one narrowly defined theoretical tradition. Its accessibility also makes it particularly useful for researchers learning qualitative analysis, while its flexibility allows experienced researchers to conduct theoretically sophisticated interpretation. Braun and Clarke’s original treatment specifically identified accessibility and theoretical flexibility among its strengths.
The same flexibility creates an important limitation. Researchers can claim to conduct thematic analysis while making incompatible assumptions about themes, coding, researcher subjectivity and quality. The label thematic analysis alone therefore provides surprisingly little methodological information. Later methodological work explicitly identifies failure to recognise diversity within TA as a recurring problem in research practice.
The method can also produce superficial findings when coding remains descriptive and researchers simply convert interview questions into themes. Strong thematic analysis requires movement from organising data toward constructing meaningful patterns that answer the research question.
Another limitation is the substantial interpretive responsibility placed on the researcher. This is not necessarily a methodological weakness—particularly within reflexive approaches—but it requires researchers to understand and communicate how their assumptions and analytical decisions shaped the resulting knowledge.
Common Mistakes When Using Thematic Analysis
A particularly damaging mistake is methodological mixing without justification. A dissertation may claim to use reflexive thematic analysis while simultaneously treating researcher subjectivity as contamination, requiring consensus coding and presenting inter-rater agreement as the main proof of quality. These practices belong to different methodological traditions and should not simply be assembled because each appears rigorous. Braun and Clarke’s quality guidance specifically warns against imposing coding-reliability assumptions universally across thematic analysis.
A second problem is treating themes as topics rather than patterns of meaning. Headings such as Leadership, Training and Communication may merely reproduce interview subjects. A strong theme should normally communicate something analytically meaningful about those subjects.
Researchers can also describe themes as having simply “emerged” from the data, obscuring their own analytical role. This is particularly inconsistent with reflexive thematic analysis, in which theme development involves active researcher interpretation.
Another error is using the six phases as a mechanical checklist. Completing coding software functions in the correct sequence does not guarantee good analysis. The phases involve recursive analytical judgement, and researchers may need to revisit codes, themes, data and theoretical assumptions repeatedly.
Finally, researchers sometimes choose thematic analysis simply because they have collected interviews. Data collection does not determine analysis automatically. Interviews can be analysed using numerous qualitative methods, and thematic analysis should be selected because it suits the research question and analytical purpose.
Thematic Analysis in Business Research
Thematic analysis is particularly useful in business and management research because many organisational questions concern how people interpret experiences, decisions, relationships and change. Contemporary qualitative business-methods literature treats thematic analysis as an established approach in organisational research.
It can be applied to employee interviews examining organisational culture, leadership or workplace change; customer interviews exploring service experiences and brand perceptions; open-ended survey responses concerning satisfaction; stakeholder accounts of sustainability initiatives; and qualitative material concerning innovation or digital transformation.
For example, a retailer investigating declining customer loyalty might use thematic analysis to move beyond individual complaints and identify broader patterns concerning perceived value, trust, service consistency and emotional attachment to the brand.
Its value in business research is therefore not merely that it can organise large amounts of text. It can help explain how organisational actors make sense of phenomena that cannot be adequately understood through numerical indicators alone.
Thematic Analysis in the Age of AI and Digital Research
Generative AI creates a particularly important methodological issue for thematic analysis because contemporary systems can now propose codes, cluster text and generate candidate themes at a scale that would be difficult for a researcher to achieve manually.
Recent research demonstrates that generative models can support aspects of codebook development and theme organisation, particularly with large text corpora. However, such workflows do not automatically reproduce reflexive thematic analysis, and researchers need to distinguish computational assistance from the methodological act of interpretation.
The central question is therefore not simply:
Can AI identify themes?
A more important question is:
If AI proposes the codes, decides which patterns matter and constructs the themes, whose interpretation does the final thematic analysis represent?
This is especially consequential in reflexive thematic analysis because researcher subjectivity and reflexive engagement are part of knowledge production. Delegating theme construction to an AI system changes that relationship rather than merely accelerating an administrative task.
AI can nevertheless be valuable for organising large datasets, exploring alternative patterns, supporting searches across transcripts or acting as an additional analytical prompt. Researchers remain responsible for checking outputs against original data, understanding context, evaluating contradictory cases and deciding whether an interpretation is methodologically defensible.
Empirical research comparing LLM-generated and human thematic analysis has found useful AI assistance but also problems including inappropriate theme construction, weak contextual understanding and altered or hallucinated quotations. The researchers concluded that AI could assist analysis but was not equivalent to experienced human qualitative analysis.
Accordingly, using AI well in thematic analysis requires more than declaring that ChatGPT or another tool was used. Researchers should consider which analytical decisions were delegated, which remained human, how AI outputs were verified and whether the resulting workflow remains coherent with the particular form of thematic analysis claimed.
When to Use Thematic Analysis
Thematic analysis is particularly appropriate when:
- the research question concerns patterned meanings, experiences, perceptions or practices across qualitative data;
- the study uses interviews, focus groups, open-ended responses or other textual datasets from which meaningful patterns can be developed;
- the researcher needs an analytical method that can operate across different theoretical or epistemological positions;
- patterns across participants or cases are more important than producing a detailed idiographic account of each individual case;
- the researcher can justify a coherent form of thematic analysis rather than merely selecting it because qualitative data have been collected;
- the intended findings are interpretive themes rather than primarily frequencies, statistical relationships or a formal theory generated through another methodological tradition.
Thematic analysis may be less suitable when the primary objective is to quantify categories, analyse detailed linguistic construction, generate formal grounded theory, or preserve highly individual case-level experience. In those situations, content analysis, discourse analysis, grounded theory or interpretative phenomenological analysis may be more appropriate.
Dissertation Example
Consider a dissertation titled “Customer Trust in Mobile Banking Applications: A Qualitative Study of Young Consumers.” Semi-structured interviews are conducted with 20 mobile-banking users to explore how they understand and develop trust when using banking applications.
The methodology chapter might explain that reflexive thematic analysis was selected because the study sought to interpret patterned meanings across participants’ accounts of trust rather than quantify predetermined categories. Analysis was predominantly inductive and focused initially on participants’ expressed experiences, while acknowledging the researcher’s active role in coding and theme development. The researcher moved recursively through familiarisation, coding, candidate theme development, theme refinement and writing rather than treating these phases as a strictly linear procedure.
The resulting analysis could develop themes such as “security cues substitute for technical understanding,” “service failure tests digital trust,” and “institutional reputation transfers to the app.” These themes would then be supported by carefully selected participant extracts and interpreted in relation to the research question and relevant literature.
The important methodological point is that the student has not merely stated that thematic analysis was used. The dissertation explains why it fits the research purpose, what form was adopted, how the analysis was conducted and how researcher interpretation contributed to the findings.
Exam Tip
If asked to explain thematic analysis in an exam, do not define it simply as “finding themes in qualitative data.” A stronger answer explains that thematic analysis identifies and interprets patterned meanings across a dataset and then demonstrates awareness that different forms of thematic analysis make different assumptions about coding, researcher subjectivity and theme development. If space permits, distinguish codes from themes and mention that Braun and Clarke’s six phases are recursive rather than a mechanical sequence. That shows methodological understanding rather than memorisation.
Still unsure if thematic analysis is the right choice for your dissertation—or which approach you should use?
Use Dudovskiy Research Assistant to get a methodology tailored to your research topic, with clear academic justification for the choices you need to explain and defend.
References
Braun, V. and Clarke, V. (2006) ‘Using thematic analysis in psychology’, Qualitative Research in Psychology, 3(2), pp. 77–101.
Braun, V. and Clarke, V. (2021) ‘One size fits all? What counts as quality practice in (reflexive) thematic analysis?’, Qualitative Research in Psychology, 18(3), pp. 328–352.
Nowell, L.S., Norris, J.M., White, D.E. and Moules, N.J. (2017) ‘Thematic Analysis: Striving to Meet the Trustworthiness Criteria’, International Journal of Qualitative Methods, 16, pp. 1–13.