Thematic Analysis for Your Dissertation: A Step-by-Step Guide

Thematic analysis is one of the most widely used qualitative data analysis methods in dissertation research. Its flexibility, accessibility, and compatibility with a range of epistemological positions make it a strong choice for doctoral students across disciplines – particularly in education, health sciences, psychology, and social work.

Yet despite its apparent simplicity, thematic analysis is frequently done poorly in dissertations. Students often confuse it with content analysis, skip critical phases of the process, or produce themes that are little more than summaries of their interview questions. This guide walks you through the six-phase approach outlined by Braun and Clarke, the most cited framework for thematic analysis, with practical guidance tailored specifically to the dissertation context.

What Thematic Analysis Is (and Is Not)

Thematic analysis is a method for identifying, analyzing, and reporting patterns (themes) within qualitative data. It involves systematically coding your data set, searching for patterns across those codes, and then organizing those patterns into themes that address your research questions.

Thematic analysis is not:

A summary of what participants said, organized by interview question
A frequency count of how often topics appear in your data
An automated process that software does for you
A single reading of your transcripts with themes that “emerge” fully formed

Themes do not emerge from data like butterflies from cocoons. They are actively constructed by the researcher through a rigorous, iterative process of reading, coding, comparing, and refining. Your role is not passive. You are making analytical decisions at every stage, and those decisions must be transparent and defensible.

Before You Begin: Epistemological Positioning

One of the most common committee criticisms of thematic analysis in dissertations is the failure to articulate the epistemological position guiding the analysis. Before you begin coding, you need to address two questions in your methodology chapter:

Inductive or deductive? An inductive approach generates codes and themes from the data itself, without trying to fit them into a pre-existing framework. A deductive approach uses an existing theory or framework to guide coding. Many dissertations use a combination, starting inductively and then connecting findings to existing theory in the discussion chapter.

Semantic or latent? A semantic approach focuses on the explicit, surface-level meaning of what participants said. A latent approach looks beneath the surface to identify underlying assumptions, ideologies, or conceptualized meanings. Most dissertation-level thematic analysis operates at the semantic level, though your committee may expect you to move toward latent analysis in your interpretation.

Phase 1: Familiarization With the Data

Before you code a single line of text, you need to know your data intimately. This means reading through your entire data set – every transcript, every field note, every document – at least twice before you begin formal coding.

During familiarization:

Read actively, not passively. Take notes in the margins or in a separate document about initial impressions, surprising statements, and recurring ideas.
Resist the urge to start coding immediately. Premature coding leads to shallow, surface-level themes because you are coding individual statements without understanding the broader patterns.
If you conducted the interviews yourself, you already have some familiarity. But reading transcripts is different from conducting interviews. You will notice things in text that you missed in conversation.

Practical tip for dissertations: Create a one-page summary of each transcript. Include the participant’s key characteristics, the main topics discussed, notable quotes, and your initial impressions. These summaries become invaluable reference documents during later phases.

Phase 2: Generating Initial Codes

Coding is the process of labeling segments of data that are relevant to your research questions. A code is a brief descriptor attached to a piece of data – a sentence, a paragraph, or even a single phrase – that captures something analytically interesting about that segment.

How to Code

Code everything relevant, not just things that seem important. You do not know yet which codes will become part of your themes. Code broadly in this phase and refine later.

Code for meaning, not topic. “Talked about advisor” is a topic label, not a code. “Felt unsupported by advisor during data collection” is a code. Codes should capture what the participant was saying about the topic, not just that the topic was mentioned.

Use your participants’ language when possible. In vivo codes – codes that use the participant’s exact words – preserve the richness of your data and are useful for grounding your analysis.

Allow data segments to have multiple codes. A single statement can be relevant to multiple aspects of your research questions. Do not force each segment into a single code.

Tools for Coding

You can code using qualitative data analysis software (NVivo, ATLAS.ti, Dedoose, MAXQDA), using a spreadsheet, or even using colored highlighters on printed transcripts. The tool matters less than the rigor of the process. If you use software, it organizes your codes and allows you to retrieve all data segments associated with a particular code quickly. If you use a spreadsheet, create columns for the participant ID, the data segment, the code, and any analytical notes.

Practical tip for dissertations: Maintain a codebook from the start. For each code, record the code name, a brief definition, an example from the data, and any notes about when to apply it versus a similar code. Your committee will likely ask to see your codebook, and building it as you go is far easier than reconstructing it later.

Phase 3: Searching for Themes

Once your data is coded, you shift from working with individual data segments to looking for patterns across your codes. A theme is a patterned response or meaning that captures something important about the data in relation to your research questions.

How to Identify Themes

Sort your codes into potential groups. Look for codes that share a common thread or that relate to each other conceptually. You might use sticky notes on a wall, a mind map, or the grouping features in your QDA software.

A theme is not just a code that appears frequently. Frequency matters, but a theme’s significance comes from its relevance to your research questions and its ability to capture something meaningful about the data. A code that appears in only three transcripts can be part of an important theme if those three instances are analytically significant.

Look for relationships between potential themes. Are some themes subsets of larger themes? Are there themes that contrast with or contradict each other? Mapping these relationships helps you develop a thematic structure rather than a flat list.

Keep a miscellaneous category. Some codes will not fit neatly into any theme. Set them aside rather than forcing them in. You can revisit them in the next phase.

Phase 4: Reviewing Themes

This is the quality control phase. You are checking whether your candidate themes actually work – whether they are supported by the data and whether they form a coherent, non-overlapping set that addresses your research questions.

Review at the code level. Read all the data segments associated with each theme. Do they actually fit together? Are there segments that do not belong? Are there codes that need to be moved to a different theme?

Review at the data set level. Reread your entire data set (or your transcript summaries) with your thematic map in mind. Does the thematic structure accurately represent the data set as a whole? Are there important aspects of the data that your themes do not capture?

Refine, merge, split, or discard themes. This is an iterative process. Themes that seemed distinct may turn out to overlap. Themes that seemed coherent may turn out to contain two separate ideas that need to be split. Themes that do not have enough data to support them may need to be discarded or absorbed into other themes.

Practical tip for dissertations: Aim for four to seven themes for a typical dissertation. Fewer than four suggests your analysis may be too broad. More than seven suggests you may not have sufficiently abstracted from your codes. These are guidelines, not rules – your data should drive the number of themes, not an arbitrary target.

Phase 5: Defining and Naming Themes

For each theme, write a detailed description that captures:

What the theme is about – its central organizing concept
What the theme is not about – its boundaries
How it relates to other themes – its position in your thematic structure
How it addresses your research questions – its analytical contribution

Naming Themes

Theme names should be concise, informative, and analytically descriptive. Avoid single-word theme names (“Communication”) because they describe topics, not themes. Aim for names that convey the specific pattern you identified: “Navigating Contradictory Feedback From Committee Members” tells the reader something specific. “Committee Communication” does not.

Some researchers use a participant quote as the theme name (e.g., “You Just Have to Figure It Out Yourself”). This can be effective, but make sure the quote is genuinely representative of the theme and not just a memorable statement.

Phase 6: Reporting Your Findings

The results chapter of a qualitative dissertation using thematic analysis typically presents each theme as a separate section, with a structured narrative that:

Introduces the theme and explains its significance
Presents evidence in the form of participant quotes (typically two to four quotes per theme, from different participants)
Provides analytical commentary that explains how the quotes illustrate the theme and connects back to the research questions

Common Reporting Mistakes

Too much description, not enough analysis. Your results section should not read like a transcript summary. After each quote, provide analytical commentary that explains what the quote shows and why it matters.

Using quotes as stand-alone evidence. Quotes should illustrate points you are making, not make points on their own. The analytical work is in your narrative, not in the quotes.

Presenting themes in isolation. Show how your themes relate to each other. A thematic map or diagram at the beginning of your results section helps readers see the overall structure before diving into individual themes.

Attributing claims to “the data.” The data does not speak. You interpret it. Use language that reflects your analytical role: “The analysis revealed…” or “Participants described…” rather than “The data showed…”

Using Thematic Analysis Software

While thematic analysis can be done manually, qualitative data analysis (QDA) software can significantly streamline the process for dissertation-scale data sets:

NVivo is the most widely used in academic settings. It handles large data sets well, supports multiple data formats, and offers visualization tools for exploring code relationships. Many universities provide student licenses.

ATLAS.ti offers strong visual network mapping tools that are particularly useful for Phase 3 (searching for themes). Its interface is intuitive for visual thinkers.

Dedoose is web-based, which makes it accessible without installation. It also handles mixed methods projects well if your dissertation combines qualitative and quantitative data.

MAXQDA is popular in European and health sciences research. It offers strong tools for team-based coding if you have a research assistant.

Regardless of which tool you use, remember that the software does not analyze your data. It organizes it. The analytical work – deciding what to code, how to group codes, and what constitutes a theme – is yours.

Ensuring Rigor and Trustworthiness

Your committee will evaluate the rigor of your thematic analysis. Common strategies for establishing trustworthiness include:

Member checking. Share your themes with participants and ask whether they accurately represent their experiences.
Peer debriefing. Have a colleague or fellow doctoral student review a subset of your coded data and discuss any discrepancies.
Audit trail. Document every analytical decision – why you created a code, why you merged two themes, why you discarded a potential theme. This trail demonstrates that your analysis was systematic, not arbitrary.
Reflexivity. Acknowledge your own position, assumptions, and potential biases in a reflexivity statement in your methodology chapter. Your perspective inevitably shapes your analysis, and transparency about that perspective strengthens rather than weakens your work.

Moving From Themes to Discussion

Your results chapter presents what you found. Your discussion chapter interprets what it means. This is where you connect your themes to existing literature, to your theoretical framework, and to the broader implications of your research.

A well-conducted thematic analysis gives you a solid foundation for discussion because your themes are analytically grounded in your data. The discussion chapter is where your themes become contributions to your field – and where the value of all that careful coding and refining becomes clear to your committee.