How to conduct research interviews
A research interview is a guided conversation to gather qualitative data. The semi-structured interview — a guide of open questions you can follow and reorder — is by far the most common, because it keeps you covering the same ground across participants while leaving room to chase what matters to each one.
The three types
- Structured — fixed questions in fixed order (closer to a spoken survey).
- Semi-structured — a topic guide of open questions + probes; flexible order. The default.
- Unstructured — a single opening prompt, then follow the participant; most demanding to analyse.
Writing the questions
- Open, not closed: “Tell me about…”, “Walk me through…” — not yes/no.
- Non-leading: don’t embed the answer (“How stressful was it?” presumes stress).
- One thing at a time: avoid double-barrelled questions.
- Prepare probes: “Can you say more?”, “What was that like?”, “Can you give an example?” — and don’t fear silence.
Running it well
Open with the purpose, consent, and recording reminder; warm up with easy questions; move through your topics with open questions and probes; close by inviting anything you missed. Build rapport, listen more than you talk, and let pauses breathe. Always get ethics approval and informed consent, anonymise, and store recordings securely.
Get the free Qualitative Research toolkit
Interview-guide templates, consent scripts, and a probing-question bank from QDA with ChatGPT and QualCoder. We’ll email you the download link.
Frequently asked questions
What is a semi-structured interview?
A guide of open questions on set topics that you can follow up and reorder — consistency plus flexibility. The qualitative default.
How do I write good questions?
Open, non-leading, one thing at a time, with prepared probes. Fewer rich questions beat many narrow ones.
What is a probe?
A short non-leading follow-up (“Say more?”, “An example?”) — or silence — that draws out depth.
How many interviews?
No fixed number — until new interviews stop adding insight (information power / saturation). Justify against your design.