Writing · Intermediate
Write a qualitative research paper with a transparent analysis
Report qualitative work through a coherent approach, evidence trail and reflexive interpretation.
13 min read · Oct 26, 2026
Key points
- Explain sampling and researcher position.
- Show how interpretation developed.
- Protect quotations and contextual identities.
Contents
Report qualitative work through a coherent approach, evidence trail and reflexive interpretation.
Qualitative research examines meaning, experience, interaction and context. A good paper explains how the researchers developed an interpretation from relevant material and why that interpretation answers the question. Its strength is not determined by whether it contains a large sample, a statistical test or a list of quotations. The question, theoretical position, sampling, data generation and analysis must fit together.
This tutorial uses a fictional project about students seeking help in a research group. Every excerpt and participant description in the example is invented for teaching. None is an interview record or evidence about actual students. In a real paper, never invent quotations, fill gaps in a transcript from imagination, or present an AI generated passage as participant speech.
Ask a question that qualitative evidence can answer
“Do students who attend office hours obtain higher grades?” asks about a numerical association. “How do students decide whether a question is appropriate to raise in office hours?” asks about experience and meaning. Interviews might help with the second question, while observations could reveal how help seeking unfolds in practice. Documents could show how an institution officially describes available support.
Specify what kind of understanding you seek. A descriptive account of common concerns, an explanation of a social process, an analysis of language, and an interpretation of an individual experience are different projects. Choose an approach because it fits that purpose. “We used qualitative methods because there were too few participants for statistics” does not establish a coherent rationale.
For the fictional project, a suitable question is: “How do students in their first research placement interpret opportunities to ask for help?” The unit of analysis might be the patterned meanings in interview accounts, rather than a frequency of students who are anxious. The paper should explain this distinction early so readers know what its findings represent.
State your methodological position in understandable language
Methodology is the reasoning that connects your question to your approach. Methods are the practical procedures, such as interviewing, observing or coding. Epistemology concerns what counts as knowledge and how it can be developed. You do not need a dense philosophical paragraph, but you do need to avoid incompatible assumptions.
For example, an interview can be treated as an account of experience shaped through the conversation, rather than a perfectly transparent measurement of an inner state. If this is your position, explain how question wording, the interviewer's role and the setting may shape what is said. If your goal is a structured content analysis of categories, explain the category definitions and coding logic instead. Different approaches can produce useful research while making different claims.
Thematic analysis identifies and interprets patterns across a dataset, but it is not one uniform procedure. Reflexive thematic analysis, approaches using a structured codebook, and approaches focused on coding reliability have different assumptions. Braun and Clarke caution against claiming their reflexive approach while importing incompatible requirements such as treating agreement between independent coders as the test of one objectively correct interpretation. S506
Explain your relationship to the setting
Reflexivity means examining how your position, assumptions and interactions shape the research. A student interviewing classmates, a lecturer interviewing students, and an outside researcher interviewing volunteers occupy different relationships. Those relationships can affect recruitment, what participants disclose and what the analyst notices.
A useful reflexive statement is specific: “The interviewer had previously worked in the same laboratory and initially expected students to describe practical barriers. During analysis, we considered whether that expectation led us to overlook accounts about belonging and perceived competence.” A list of demographic characteristics without explaining their analytical relevance is less informative.
Reflexivity does not mean pretending that every influence can be removed. It means making relevant interpretive choices visible. Keep notes on assumptions before data generation, moments of uncertainty during interviews, and changes in interpretation during analysis. Explain how those reflections affected a decision rather than adding them as a ceremonial paragraph at the end.
Justify sampling through the question
Purposive sampling selects people, settings or materials because they can inform the research question. You might seek students with different placement experiences, disciplines or routes into a programme. Explain the intended variation and the practical recruitment process. A convenience sample may be necessary, but its accessibility should not be disguised as a principled sampling strategy.
Describe eligibility, recruitment routes, who approached participants, nonparticipation information when available and the relevant setting. Consider whether a person with authority over potential participants should be involved in recruitment. Obtain the ethics review, permissions and consent required for the actual project before collecting data. Local requirements differ; this guide does not determine whether a specific project is exempt.
Justify sample adequacy using the study's aim, the relevance and richness of the material, the specificity of the sample and the intended analysis. Malterud and colleagues propose information power as one way to reason about the amount of data needed, considering factors such as the focus of the question, sample specificity, theory, dialogue quality and analytic strategy. It is a judgement framework, not a formula that generates a universally adequate interview count. S508
Saturation is used differently across qualitative traditions. If you invoke it, state what was considered saturated and how this fits the chosen methodology. Braun and Clarke argue that saturation understood as discovering all possible codes or meanings is not a coherent stopping rule for reflexive thematic analysis. Do not automatically write “saturation was reached” because you completed a familiar number of interviews. S507
Make data generation visible
For interviews, describe who conducted them, the mode and setting, approximate duration, guide development, recording, field notes and transcription. Explain whether questions changed during the project and why. A guide is a structure for inquiry, not a script that prevents following up an unexpected but relevant account.
Compare two fictional prompts. “Why were you afraid to ask your supervisor questions?” assumes fear and identifies the supervisor as its target. “Tell me about a time you considered asking for help during the placement” invites an account. Follow up with “What influenced that decision?” or “What happened next?” A neutral question does not guarantee an unbiased conversation, but it avoids embedding the desired finding in the prompt.
Choose transcription detail to match the analysis. Research about conversational interaction may require pauses, overlap and timing that a study focused on broad experiential accounts does not. State whether transcripts were checked against recordings and how identifying details were removed. Retain secure access to original material according to the approved plan. Publishing a quotation with a pseudonym does not guarantee anonymity if the context identifies the speaker.
SRQR offers broad reporting recommendations for qualitative studies. COREQ specifically addresses interviews and focus groups. These can help check that readers receive information about researchers, context, methods and interpretation; use them with the methodological approach rather than treating every item as a universal design requirement. S22 S39
Show how analysis developed
Begin by reading and rereading the material while recording observations and questions. A code is a concise label attached to a meaningful feature of the material. A theme, in reflexive thematic analysis, is a developed pattern of meaning organised around a central idea. A topic such as “supervisors” is not automatically a theme. It tells readers what participants discussed, but not what the analysis explains.
Consider the following fictional excerpts:
| Teaching excerpt | Possible initial code | Question for further interpretation |
|---|---|---|
| “I knew whom to ask, but I was worried that my question was too basic.” | Protecting an appearance of competence | How is a legitimate question defined? |
| “When a senior student said they also got stuck, I felt I could ask.” | Permission through shared difficulty | What changes the perceived cost of asking? |
| “I asked about equipment, but kept questions about the project to myself.” | Separating acceptable and risky questions | Does the topic affect willingness to disclose uncertainty? |
These codes are possible interpretations, not a correct answer key. Examine the surrounding account and the wider dataset. The first excerpt could concern belonging, fear of evaluation, previous treatment or another context dependent explanation. Do not decide from one attractive sentence alone.
A provisional theme might be “Help seeking as a display of competence”. The organising idea is that asking for help can feel like a public statement about ability, rather than simply a way to obtain information. Test the developing theme against cases where students ask freely, situations where they avoid help for practical reasons, and passages that do not fit. Revising a theme is part of analysis rather than evidence that the initial coding failed.
Software can store transcripts, retrieve coded extracts and organise memos. It does not establish the meaning of a theme. Describe what you did with the software and how decisions developed. “The software identified three themes” hides the analyst's role and is usually an inadequate account of interpretive work.
Match quality practices to the approach
In an analysis based on a predefined coding framework, independent coding and agreement measures may answer a relevant question about the consistent application of categories. In reflexive thematic analysis, discussion among analysts can deepen interpretation without aiming to discover one correct code through consensus. Specify the purpose of collaboration rather than borrowing a familiar reliability statistic without considering its assumptions.
Keep an audit trail, meaning records of important analytic decisions, code revisions and theme development. Consider alternative interpretations and contradictory material. Explain what a reflexive discussion changed. Participant feedback can be valuable for some purposes, but it is not a universal truth test for every analytic interpretation. Participants may disagree with a broader interpretation for understandable reasons; explain how such feedback was considered.
Avoid a list of quality terms with no evidence that the procedures occurred. “Triangulation” should identify the sources or perspectives compared and what that comparison added. Agreement among sources may strengthen an interpretation, while disagreement can reveal a meaningful difference in context. The author guidance for reflexive thematic analysis emphasises methodological coherence and a developed analytic account rather than a generic checklist of quality rituals. S522
Write an analytic results section
A results section should explain each theme, show evidence and develop its relationship to the research question. A quotation illustrates or supports an interpretation; it should not perform all the analytical work. Introduce enough context to make the passage intelligible, then explain what it contributes.
Weak fictional result: “Theme 1 was confidence. One student said, ‘I knew whom to ask, but I was worried that my question was too basic.’ This shows that confidence is important.”
Developed fictional result: “Help seeking was described as a judgement about how uncertainty would be interpreted by others. In the teaching excerpt, knowing the appropriate person did not remove hesitation; the concern was whether the question would signal inadequate ability. A second fictional account, in which a senior student's admission of difficulty made asking easier, suggests that the perceived meaning of uncertainty could change through interaction. These passages support an interpretation of help seeking as a social display of competence, rather than only a problem of locating information. However, the account about asking equipment questions while withholding project questions indicates that the perceived risk also depended on the topic.”
The developed paragraph connects evidence, interpretation and variation. It does not claim that all students behave this way. In a real study, each quotation needs a genuine source identifier and the interpretation must be supported by the dataset, not assembled from three invented examples.
Use counts only when their meaning is clear. “Seven participants mentioned transport” can describe the material if counting is justified and consistently performed. It does not by itself estimate how common transport problems are in the wider population. A theme can be important because it explains a consequential process, even when it is not the most frequently mentioned topic.
Build a discussion with appropriate boundaries
Relate the findings to existing theory and research without forcing every theme to confirm a familiar model. Explain what your account adds, what alternative interpretations remain and how the setting shaped the findings. Transferability concerns whether readers can judge relevance to other contexts using a sufficiently detailed account; it is different from statistical generalisation from a probability sample.
For the fictional project, the conclusion might suggest studying how research groups communicate expectations about questions. It should not claim that one intervention will reduce anxiety or improve grades unless those outcomes were investigated with a design supporting that inference. A qualitative explanation can motivate an intervention study without serving as its result.
In research combining qualitative and quantitative methods, explain the relationship between components. Were interviews used to develop a survey, explain an unexpected numerical pattern, or provide a different perspective on the same process? A joint display can place numerical findings beside qualitative interpretations and identify agreement or tension. Putting two unrelated results sections in one manuscript does not automatically achieve integration.
Practice and model answer
Task: A draft says, “Two researchers achieved perfect coding agreement, proving that our reflexive thematic analysis was unbiased. Saturation occurred after twelve interviews.” Identify what needs revision.
Model answer: Explain which form of thematic analysis was actually used and whether coding agreement fits its assumptions. Describe the purpose of collaboration and how interpretations changed rather than claiming the absence of influence. Replace the unsupported saturation assertion with a rationale appropriate to the question, dataset and method. If saturation was used within a different approach, define it and show how it was assessed. Report the actual process, even if the original methodological label needs correction.
Before submitting, trace one central claim from the Results back to the relevant material, analytic notes and alternative interpretations. Then ask a reader whether the connection is understandable. Use Template 16 for the decision trail and adapt Template 7 to the approach rather than forcing every qualitative paper into an experimental template.