How to Conduct Semi-Structured Interviews for Your Thesis: A Complete Step-by-Step Guide (2026)

Semi-structured interviews are the most widely used qualitative data collection method in social science, education, health, and business research — and for good reason. They give you the structure to answer your research question consistently across all participants while preserving the flexibility to follow unexpected threads that often yield the richest insights. If your thesis involves human participants and you want to understand experiences, perspectives, or processes in depth, this is likely the method for you.

Yet many students underestimate what a rigorous semi-structured interview study actually requires. Without a carefully designed interview guide, a sound sampling strategy, proper ethical safeguards, and a systematic approach to analysis, even hours of recorded conversation can produce data that an examiner will reject as anecdotal. This guide walks you through every stage — from framing your research question to writing up your findings — so your interview methodology stands up to scrutiny.

Whether you are writing a bachelor’s dissertation, a master’s thesis, or a PhD, learning how to conduct semi-structured interviews properly is one of the most transferable research skills you can develop.

What Are Semi-Structured Interviews?

A semi-structured interview sits between a fully structured questionnaire — where every question is fixed and responses are coded — and an unstructured conversation with no predetermined agenda. The researcher prepares an interview guide: a set of open-ended questions and topic prompts that ensure every participant is asked about the same core themes. However, the sequence can change based on the flow of conversation, and follow-up probes are used to explore emerging ideas in real time.

This balance makes semi-structured interviews uniquely powerful for thesis research. You generate comparable data across participants (which satisfies your examiner’s need for methodological rigour) while allowing individuals to express their views in their own words (which produces the nuanced, contextualised understanding that quantitative methods cannot deliver).

Semi-Structured vs Structured vs Unstructured Interviews

When Should You Use Semi-Structured Interviews in Your Thesis?

Semi-structured interviews are the right choice when you need to understand how or why something happens from the perspective of those who experience it. They are particularly well suited to:

• Exploring lived experiences (e.g., “How do first-generation students navigate university culture?”)
• Understanding decision-making processes in professional or organisational contexts
• Investigating under-researched phenomena where theory is underdeveloped
• Generating hypotheses that can later be tested quantitatively (exploratory phase of mixed-methods research)
• Studying sensitive topics where standardised questionnaires feel cold or reductive

They are less suited to situations where you need statistically generalisable findings, where large samples (100+) are required, or where you are testing a specific hypothesis under controlled conditions. For those goals, consider a quantitative or mixed methods design.

Step 1: Align Your Interview Design with Your Research Question

Every methodological decision flows from your research question. Before you write a single interview question, confirm that your research question is genuinely exploratory or interpretive. Questions that begin with “What is the experience of…”, “How do participants understand…”, or “What factors influence…” are well suited to semi-structured interviews. Questions that begin with “How many…”, “Does X cause Y…”, or “To what extent does…” usually call for a different method.

Next, identify the specific information you need from participants to answer your research question. Cluster this information into 4–6 topic domains. These domains become the skeleton of your interview guide. For example, a research question about the barriers to AI adoption in NHS primary care might generate topic domains of: current workflows, awareness of AI tools, institutional culture, data privacy concerns, and training needs.

Your supervisor should review your research question and proposed domains before you build the guide. Misalignment at this stage wastes weeks of data collection.

Step 2: Obtain Ethics Approval

In the UK, most universities require dissertation students whose research involves human participants to obtain approval from a departmental or faculty Research Ethics Committee (REC) before recruiting. In the US and Canada, the equivalent body is the Institutional Review Board (IRB). Never begin recruiting or collecting data before your ethics approval is confirmed. Starting early violates research governance rules and can result in your data being unusable — seriously jeopardising your submission.

A standard ethics application for a semi-structured interview study typically requires:

1. A participant information sheet (PIS) — written in plain English (no jargon), explaining the study purpose, what participation involves, how data will be stored, and the right to withdraw at any time without penalty
2. An informed consent form — signed by each participant before the interview begins, confirming they have read the PIS and agree to participate
3. Your interview guide — submitted for review so the ethics committee can assess question sensitivity
4. A data management plan — describing how recordings and transcripts will be stored, anonymised, and eventually deleted
5. A risk assessment — identifying any potential harms (psychological distress, reputational risk, etc.) and how you will mitigate them

For guidance on how to write the ethics statement that goes into your methodology chapter, see our dedicated article on citing and documenting your research governance decisions. You can also read about the full landscape of research ethics guidelines every student must know.

Step 3: Design Your Interview Guide

Your interview guide is not a questionnaire. It is a navigation tool — a series of open-ended questions and follow-up prompts that keep the conversation focused without constraining the participant’s language or framing.

Structure of a Good Interview Guide

A well-designed guide for a 45–60 minute interview typically contains:

• Introduction (5 minutes): Restate the study purpose, confirm consent, explain recording, invite questions. Keep this scripted — consistency matters.
• Warm-up questions (5 minutes): Easy, non-threatening questions about the participant’s background or role. These build rapport and get them talking before you reach sensitive themes.
• Core questions (30–40 minutes): 6–10 open-ended questions, each targeting one of your topic domains. Use “grand tour” questions (“Can you walk me through…”) before drilling into specifics.
• Closing (5–10 minutes): Summarise what you have heard, ask if there is anything they would like to add, and end with a forward-looking question (“Is there anything you feel researchers should be asking about this topic that you weren’t asked today?”).

Example Question Stems That Work Well

• “Can you describe a time when…”
• “How did you come to understand…”
• “What does [concept] mean to you in your daily work?”
• “How do you decide between…”
• “What would have to change for…”

Probes to Use in the Moment

• Clarification probe: “When you say X, what do you mean by that?”
• Extension probe: “Can you tell me more about that?”
• Example probe: “Can you give me a specific example?”
• Contrast probe: “How was that different from…?”
• Silence: Simply waiting — participants often add the most valuable material after a pause.

Pilot your guide with 1–2 participants before committing to full data collection. A pilot interview typically reveals questions that are too abstract, too leading, or too long — problems that are easy to fix at the design stage but impossible to correct once you have 15 transcripts.

Step 4: Sample and Recruit Participants

Qualitative research does not use random sampling in pursuit of statistical representativeness. Instead, you use purposive sampling — deliberately selecting participants who have the knowledge, experiences, or perspectives most relevant to your research question. Within purposive sampling, several subtypes are available:

How Many Participants Do You Need?

This is the question every student asks — and the honest answer is that sample size in qualitative research is determined by data saturation, not a formula. Saturation is reached when additional interviews stop generating new themes or categories. In practice, most master’s theses achieve saturation with 8–15 participants, and many PhD studies with 20–30. If your participants are highly homogeneous (e.g., all experienced practitioners in a narrow specialism), saturation typically comes sooner.

Discuss your anticipated sample size with your supervisor and justify it in your methodology chapter by referencing the saturation concept. Citing Lincoln and Guba (1985) or Guest et al. (2006) — who found saturation at 12 interviews in their landmark study — is standard practice.

Step 5: Conduct the Interviews

Most semi-structured interviews in 2026 are conducted via video call (Zoom, Teams, or Google Meet), which dramatically widens your accessible participant pool. Face-to-face interviews remain preferable for particularly sensitive topics or when non-verbal communication is analytically important, but video is now fully accepted by examiners when justified in your methodology.

Practical Checklist Before Each Interview

• Confirm the participant has received and read the Participant Information Sheet
• Obtain signed consent (digital signature or email confirmation is accepted at most institutions)
• Start your recording software (Otter.ai, Zoom built-in, or a separate voice recorder) and confirm the participant is aware they are being recorded
• Have your interview guide on screen, but do not read questions verbatim — treat it as a map, not a script
• Take brief notes during the interview to capture non-verbal cues or ideas for follow-up

During the Interview

Your role is to listen actively, not to fill silence or guide participants towards your hypotheses. The most common interviewing mistake is “leading” — framing questions in ways that suggest a desired answer (“Would you say that management resistance was the main barrier?”) rather than inviting open description (“What obstacles did you encounter?”). Ask one question at a time. Avoid double-barrelled questions (“How did you feel about the process, and do you think it was fair?”).

After each interview, write a brief reflexive memo (1–2 paragraphs) noting your impressions, any unexpected themes, and your own emotional or intellectual reactions. These memos are valuable when you begin analysis.

Step 6: Transcribe and Manage Your Data

Verbatim transcription — capturing every word, hesitation, and interruption — is standard for most qualitative analysis. AI transcription tools (Otter.ai, Whisper, Microsoft Word’s transcribe feature) can produce a first draft in minutes, but always review and correct the output manually before analysis. Automated transcripts have error rates of 10–20% for non-standard accents or technical vocabulary.

Assign each participant a pseudonym or code (P1, P2… or names from a consistent theme) at the transcription stage. Never use real names in your transcript files. Store all audio recordings and transcripts in encrypted, password-protected folders on your institution’s approved cloud storage. Most UK universities require research data to be stored on university systems, not personal Google Drive or Dropbox accounts.

Step 7: Analyse the Data

The most commonly used analysis approach for semi-structured interview data is thematic analysis, particularly Braun and Clarke’s (2006, revised 2019) reflexive thematic analysis framework. Their six-phase process — familiarisation, generating initial codes, searching for themes, reviewing themes, defining and naming themes, and producing the report — provides a transparent, auditable trail from raw data to findings that examiners can follow.

Other appropriate analysis frameworks include:

• Framework analysis — particularly used in applied health and policy research; structured matrix approach
• Interpretative Phenomenological Analysis (IPA) — for studies focused on individual lived experience; smaller samples (3–8)
• Grounded theory coding — open, axial, and selective coding to generate substantive theory from data
• Template analysis — a priori coding template informed by existing theory, adapted as analysis proceeds

Whichever framework you choose, use it consistently and name it explicitly in your methodology chapter. “I analysed my interviews thematically” is not enough — specify which version of thematic analysis, by which authors, and how you applied it.

Software tools such as NVivo, ATLAS.ti, MAXQDA, or the free Taguette can help you manage and code large transcript datasets, but they are analytical aids, not analysis methods. The intellectual work of interpretation is always yours.

For a deeper dive into data analysis frameworks, read our guide on research methodology types and when to use each.

Step 8: Write Up Your Methodology Chapter

Your methodology chapter must justify every design decision you made. Examiners read it looking for the logic connecting your research question to your data collection and analysis choices. For a semi-structured interview study, your methodology chapter should address:

1. Research paradigm and philosophical stance — are you working within interpretivism? Constructivism? State this explicitly and cite the appropriate underpinning theorists (Denzin & Lincoln; Bryman; Creswell).
2. Methodological approach — qualitative research and why; why semi-structured interviews rather than focus groups, surveys, or observation
3. Interview design — how the guide was developed, piloted, and revised
4. Sampling strategy — type of purposive sampling, inclusion/exclusion criteria, how participants were recruited
5. Data collection procedure — format (face-to-face/video), duration, recording process, your role as interviewer
6. Ethical considerations — ethics approval reference number, consent procedures, anonymisation, data security
7. Analysis method — named framework, phases, how coding was conducted, whether you used software
8. Quality and rigour — how you ensured trustworthiness: credibility (member checking, peer debriefing), transferability (thick description), dependability (audit trail), confirmability (reflexivity statement)

See our full guide on how to write a research methodology chapter for chapter structure templates, annotated examples, and advice on justifying your philosophical position.

Common Mistakes and How to Avoid Them

Frequently Asked Questions