Distinguish a code, topic summary and theme

A code, a topic summary and a theme can all look like short labels in a spreadsheet or software panel. That visual similarity causes a practical problem: researchers may rename a broad interview topic and report it as a theme, or promote a frequently used code without showing the pattern of meaning that makes it analytically useful.

This guide provides a diagnostic for Braun and Clarke’s reflexive thematic analysis. It is not a universal dictionary for every form of thematic analysis. Coding-reliability, codebook and reflexive approaches can differ in what they treat as a theme, how codes are developed and what counts as quality. Name the approach you are using before applying the tests below.

The worked example is wholly fictional. It traces extracts into codes, compares a topic summary with a developed theme, checks a contradictory extract and shows why a candidate may need revision rather than automatic promotion.

Start with the unit each label describes

A quick distinction is:

A code captures an analytically relevant idea associated with a particular data segment.
A topic summary gathers what participants said about a subject, domain or interview question.
A theme, in reflexive thematic analysis, presents a pattern of shared meaning organised around a central concept and relevant to the research question.

The distinction is functional, not a rule about label length. A code may be conceptually rich. A topic summary may be carefully written. A theme may have a short title. What matters is the analytic work each output performs.

On a narrow screen, scroll the table sideways to read every column.

Output	Primary unit	Main question	Typical warning sign
Code	A meaning-relevant segment	What idea is analytically useful here?	The label merely repeats the participant’s words without a reason for selecting them
Topic summary	Material about one subject or question	What did participants say about this topic?	The grouping has no idea connecting its different observations beyond the subject itself
Developed theme	A pattern across relevant data	What shared meaning does this pattern present in relation to the research question?	The title sounds polished, but the extracts do not cohere around a central concept

The thematic analysis FAQs maintained by Braun and Clarke make two qualifications important. First, the code–theme distinction is clear but not absolute: a complex code can become the starting point for a candidate theme. Second, confusing a topic summary with a theme is especially problematic when a researcher claims to use their reflexive approach, because that approach conceptualises themes as patterns of shared meaning organised by a central concept.

Worked example: from extracts to codes

Northbridge Thesis Exchange, every participant and every extract below are wholly fictional. They were written for this guide and are not research data.

A fictional study asks:

How do part-time postgraduate students make written supervisor feedback usable during dissertation revision?

Six short extracts are selected from the invented dataset.

P1: “If I read the comments as soon as they arrive, I hear them as criticism. The next morning I can turn each one into something I need to change.”
P2: “I circle the directive words—justify, compare, define—and put them into a checklist beside the chapter.”
P3: “Sometimes I send a voice note to a course friend. Saying the comment aloud helps me work out what the supervisor is actually asking.”
P4: “The last comments were unusually clear. I changed the two paragraphs immediately and did not need to interpret anything.”
P5: “One draft says to narrow the argument and another says to show more context, so I keep a decision log to stop myself undoing earlier changes.”
P6: “A positive margin note settles me, but ‘good point’ does not tell me what should happen next.”

Coding does not mean compressing each entire extract into one official answer. The researcher selects features relevant to the question and can attach more than one code to a segment.

On a narrow screen, scroll the table sideways to read every column.

Extract	Possible code	Why this code is relevant
P1	Creating emotional distance before deciding what to revise	Connects timing and emotional response to the work of making feedback actionable
P2	Translating directive verbs into a revision checklist	Identifies an active conversion from comment to task
P3	Using peer talk to decode an ambiguous request	Shows interpretation being distributed through conversation
P4	Acting immediately when the requested change is explicit	Challenges any claim that all feedback requires extended translation
P5	Stabilising competing comments through a decision record	Links contradiction to a strategy for maintaining continuity
P6	Receiving reassurance without action guidance	Separates emotional value from practical direction

These are defensible possibilities, not a correct codebook. In reflexive thematic analysis, coding is shaped by the research question, theoretical position, analytic orientation and the researcher’s engagement with the data. Another coherent analysis might code some segments more semantically or more conceptually. The obligation is to explain and enact the chosen approach consistently.

See how a topic summary forms

Suppose the analyst groups the extracts under these headings:

emotional responses to feedback;
strategies for understanding comments;
clear and unclear feedback; and
ways students organise revision.

Those headings may be useful for data management. They answer, “What areas did participants discuss?” An interview guide might even have asked about those areas directly.

But a heading such as Strategies for understanding feedback remains a topic summary if it simply gathers everything said about strategies. Its members share a subject. They do not yet necessarily share an analytic meaning.

A fuller topic summary might say:

Participants described waiting before rereading comments, extracting directive verbs, talking with peers and keeping decision records. They also distinguished clear comments from ambiguous or encouraging ones.

That is an accurate synopsis of the invented material. It does not yet explain what pattern the researcher sees across those observations or how that pattern answers the study question.

A topic summary is not automatically poor research. Some thematic-analysis approaches, research purposes and reporting formats use topic or domain summaries deliberately. The problem is a mismatch: claiming the shared-meaning logic of reflexive thematic analysis while reporting interview-question headings as if they were developed themes.

Develop and test a central concept

The analyst next proposes a candidate theme:

Making feedback actionable is interpretive work, not receipt alone.

Its central concept is that written comments become usable through acts of translation, distancing, discussion and record-keeping. The candidate links P1, P2, P3 and P5. P6 adds a useful distinction: reassurance can matter without specifying an action.

This statement does more than name the topic “feedback strategies.” It proposes a patterned meaning: receiving feedback does not itself determine revision; students often transform comments into decisions through additional interpretive work.

A central concept should be expressible in one or two sentences without merely listing subtopics. Test it with five questions:

What is the unifying idea?
Which codes and extracts instantiate that idea?
How does the pattern answer the research question?
What does the theme claim—and not claim—about the dataset?
What evidence would make the theme need revision, division or rejection?

If the answer to the first question is only “all of these extracts concern feedback,” the candidate is still organised by topic.

Use contradictory extracts as analytic tests

P4 complicates the candidate. The participant reports acting immediately because the requested changes were explicit. Treating that extract as irrelevant would protect a neat story at the cost of the dataset.

The contradiction does not automatically destroy the theme. It tests its scope.

On a narrow screen, scroll the table sideways to read every column.

Possible decision	When it would be defensible	Application to P4
Retain and refine	The extract identifies a boundary or condition within the same central concept	Revise the claim so that interpretive work becomes especially visible when comments are ambiguous, affectively difficult or inconsistent
Split the candidate	Two internally coherent patterns answer different aspects of the question	Possible only if clear-directive responses form a substantial, conceptually distinct pattern elsewhere in the dataset
Move the extract	Another candidate theme explains the extract more coherently without hiding its relevance	Appropriate if a second theme concerns conditions that reduce or increase revision work
Reject the theme	The central concept does not hold across its supposed evidence or adds little beyond the topic heading	Necessary if most relevant extracts show no translation process and the claim depends on isolated quotations

For this mini-dataset, a reasonable refinement is:

When feedback is not self-explanatory, students make it actionable through translation, distance and dialogue.

P4 now marks a condition: explicit comments can reduce the need for extended interpretation. P6 also prevents overclaiming that every valued comment supplies practical guidance.

This is still only a candidate theme. In a real study, it must be reviewed against all relevant coded extracts and the wider dataset. The analyst may combine it with another candidate, divide it, rewrite its boundaries or discard it.

Do not use size as the deciding rule

A code is not a theme merely because it appears often. A theme is not stronger merely because it contains the largest number of extracts. Frequency can matter for some questions, but reflexive thematic analysis is not a counting contest.

Similarly:

one long code is not automatically a theme;
one interview question is not automatically a theme;
a heading with subheadings is not automatically a theme;
every participant need not mention a pattern for it to be analytically useful; and
an unusual or contradictory extract should not be deleted only because it disrupts a tidy account.

The original Braun and Clarke paper describes thematic analysis as theoretically flexible and its phases as recursive rather than a rigid linear procedure. Their later APA handbook chapter likewise presents themes as patterns of meaning across a dataset and illustrates movement between coding, candidate themes and review. These full primary texts support the iterative process; they should not be used to pretend that all later forms of thematic analysis share one definition.

Run a code–summary–theme audit

For each proposed output, complete this audit.

1. Name the claimed approach

State whether the study uses reflexive thematic analysis, coding-reliability thematic analysis, a codebook approach or another specified method. Do not rely on “thematic analysis” alone when the procedural commitments matter.

2. Return to the research question

Write the exact question beside the candidate. A theme should contribute an answer, interpretation or insight relevant to that question—not simply reproduce the interview schedule.

3. Locate the unit

For a code, identify the precise segment and the idea selected. For a topic summary, identify the subject that groups the material. For a candidate theme, identify the pattern and central concept.

4. Build an evidence map

List supporting codes and representative extracts. Preserve identifiers and context in the secure analytic record. Do not detach quotations from speakers, questions or surrounding meaning merely because they fit a heading.

5. Search for tensions

Actively review extracts that complicate, limit or contradict the candidate. Record whether they refine a boundary, support a split, belong elsewhere or expose a weak candidate.

6. Test internal coherence

Ask whether the extracts make sense together under the central concept. A candidate that needs several unrelated explanations may be a container of topics rather than one theme.

7. Test external distinction

Compare the candidate with the other themes. If two themes make the same analytic point under different nouns, combine or redraw them. If one theme holds multiple distinct concepts, consider dividing it.

8. Write a bounded theme statement

State the shared meaning, its relevance and its limits. Avoid universal claims when the evidence supports a conditional or situated pattern.

9. Recheck the whole dataset

Review whether the candidate fits the wider dataset, not only memorable quotations. In Braun and Clarke’s overview of doing reflexive thematic analysis, theme development includes reviewing, developing, defining and naming; candidates can be combined, split or discarded.

10. Preserve the decision trail

Record what changed and why. This is analytic accountability, not a claim that another researcher must reproduce identical themes.

Report the analysis coherently

The 2024 Reflexive Thematic Analysis Reporting Guidelines paper identifies recurring problems such as treating thematic analysis as one uniform method and confusing themes with topics or codes. Its wider lesson is methodological coherence: the theoretical position, analytic practices, quality criteria and reporting language should fit one another.

A methods section should therefore explain enough for readers to understand:

why the selected thematic-analysis approach fits the research question;
the analytic orientation, including relevant theoretical assumptions;
what counted as data and how the dataset was constituted;
how familiarisation and coding were undertaken;
how candidate themes were generated, reviewed and developed;
how researcher subjectivity and reflexivity were understood;
what records or tools supported the analysis; and
how themes were reported with evidence and appropriate limits.

Do not import inter-rater reliability, consensus coding, saturation or a fixed codebook as universal quality tests. They may be coherent in some approaches and incoherent in others. Follow the method you have actually chosen, the design it belongs to and any defensible local requirements.

Keep the analysis ethical and auditable

A worked table in a guide is not permission to place identifiable participant quotations into a public spreadsheet, general-purpose chatbot or analytics event.

For real data:

follow the consent, confidentiality, retention and access terms approved for the study;
keep pseudonymisation distinct from complete anonymity;
preserve enough context to interpret extracts without exposing unnecessary identifiers;
do not invent, smooth or silently combine quotations;
mark translations, omissions and edited speech transparently where required;
maintain the difference between participant words and researcher interpretation; and
use secure, approved tools for any sensitive material.

The examples here are synthetic. They must not be represented as empirical findings or copied into a dissertation as if they came from participants.

Final diagnostic

Before calling an output a theme in Braun and Clarke’s reflexive thematic analysis, ask:

Does it present shared meaning rather than only a shared subject?
Can I state its central concept clearly?
Does it answer the research question?
Which codes and extracts support it?
Which extracts complicate it?
Have I reviewed it against the wider dataset?
Is it internally coherent and distinct from other themes?
Have I refined, split, combined or rejected weak candidates?
Does my reporting match the reflexive approach I claim?
Have I preserved confidentiality, context and a decision trail?

A code can be analytically rich without being a theme. A topic summary can be useful without becoming a shared-meaning theme. The goal is not to force every label upward through a hierarchy. It is to know what analytic job each output performs and to report that job honestly.

If you need help checking methodological coherence, research support can review your own research question, analytic record and developing theme structure while you retain responsibility for the analysis, interpretations and final claims.