Semi-structured interviews: a practical guide

How to design, run and analyse semi-structured interviews for a dissertation, with example questions, probes and pitfalls to avoid.

9 min read · Reviewed 2026 · Written and reviewed by Chris McGarrigle

A semi-structured interview uses a fixed set of open-ended questions as a guide, but allows the interviewer to reorder, probe and follow up depending on what the participant says. It sits between a rigid structured survey and a free-flowing unstructured conversation, which is why it is the most common qualitative method in dissertations.

Used well, it produces rich, comparable data because every participant answers the same core questions, while still leaving room for unexpected but relevant detail to emerge. Used badly, it produces a transcript full of leading questions and no analytic depth.

This guide covers building the interview guide, running the interview itself, and getting from transcript to usable data.

Building the interview guide

Start from your research question and objectives, then write four to six open questions that each map onto one objective. Avoid yes/no phrasing: instead of 'Did you find the training useful?', ask 'How did the training change the way you approach X?'

Order questions from broad and easy (background, context) to more specific and sensitive, and write two or three probes under each main question — 'Can you give me an example?', 'What made you feel that way?' — so you are not improvising probes under pressure.

  • Opening question: low-stakes, gets the participant talking (their role, background, experience).
  • Core questions: 3-5 questions directly tied to your objectives.
  • Closing question: 'Is there anything we haven't covered that you think is important?'

Piloting before you commit

Run your guide with one or two people who resemble your sample — a coursemate in a similar role works if you cannot access a real participant. Time it, check for leading or double-barrelled questions, and revise wording that produced short, closed answers.

A pilot also tells you whether 45 minutes is realistic or whether you need to cut two questions to keep participants engaged.

Running the interview

State consent and recording arrangements again at the start even if a form was signed earlier, and remind participants they can skip a question or stop at any time. Use silence deliberately: after an answer, wait three seconds before your next question — participants often add the most useful detail in that gap.

Keep your own talking to a minimum. If you find yourself explaining your own opinion mid-interview, you are shaping the data rather than collecting it.

From transcript to data

Transcribe verbatim, including hesitations if you plan to do close discourse work, or a cleaned-up version if you are doing thematic analysis. Anonymise names and identifying details as you go rather than leaving it until the end.

Once transcribed, the interview becomes raw data for whichever analysis method you specified in your methodology — usually thematic analysis. Keep the interview guide, transcripts and your reflective notes as your audit trail for a credibility/trustworthiness section.

Common mistakes

  • Writing closed questions ('Did you like...?') that shut down elaboration.
  • Not piloting the guide, so the first real interview is also the first test of the wording.
  • Talking too much or agreeing/disagreeing with participants during the interview.
  • Losing track of which objective each question serves, producing a guide that wanders.
  • Forgetting to record consent, date and pseudonym for each transcript, causing chaos at analysis stage.

Try this: Build and stress-test your guide

  1. Write one open question per research objective, avoiding yes/no phrasing.
  2. Add two probes under each question in case the first answer is thin.
  3. Read the guide aloud to a coursemate and time how long it takes.
  4. Revise any question that produced a one-word or closed answer.

Do it in SuperDiss

Draft and justify your interview guide alongside your sampling and ethics decisions.

Open the Methodology workspace

Students also ask

What is a semi-structured interview?
A semi-structured interview uses a fixed set of open-ended questions as a guide, but lets the interviewer reorder, probe and follow up depending on what the participant says, producing data that is both comparable across participants and rich in unexpected detail.
How many interview questions should I prepare?
Four to six core open questions is usually enough, each tied to one research objective, plus two or three probes under each in case the first answer is short. More than this and interviews run long and lose focus.
Should I pilot my interview guide?
Yes. Piloting with one or two people similar to your sample reveals leading or double-barrelled questions and gives you a realistic sense of timing before you use up access to your real participants.

Choose philosophy, approach, sampling and ethics you can defend in the viva.

How to choose a methodology

A decision path from research question to methodology, plus how to write the justification examiners are looking for.

Qualitative vs quantitative research

The real difference between qualitative and quantitative dissertation research, and how to choose based on your question rather than your comfort zone.

How to get ethical approval

What ethics committees actually look for, the documents you need, and the mistakes that cause applications to be returned.

Defend your methodology

Continue your dissertation journey

Browse every guide in research methodology.

Read next

This guide is general educational content. Your institution's regulations, handbook and supervisor's instructions always take precedence. Where a named framework is mentioned, consult the original source text rather than relying on this summary.

Free account

Start your dissertation in the next two minutes

£19.99/month after a free trial. Cancel whenever you like.

Takes under a minute · Works with your university email · Cancel or delete anytime