How long should a semi-structured interview last?

Many PhD interviews run 45 to 60 minutes, though anything from 30 to 90 is common. Your pilots will show how long your guide really takes. Interviews with officials are often shorter.

Can I change my interview guide after I start?

Yes. Adding probes or rewording a confusing question is normal in qualitative work. Record what changed and when, and check whether your ethics committee must be told about major changes.

Do I have to transcribe every interview in full?

For most PhD analysis methods, yes. Partial transcription makes it hard to show you looked at all your data. If you transcribe selectively, agree the rule with your supervisor and state it in the methodology.

Should participants check their transcripts?

Some researchers return transcripts, and COREQ asks whether you did. It can improve accuracy, but participants sometimes remove things they regret saying. Decide in advance, cover it in the consent form, and explain your choice. For the analysis that follows, see our thematic analysis guide.

Semi-Structured Interviews in Qualitative Research

A semi-structured interview is a one-to-one conversation built around a prepared interview guide: you ask every participant the same main questions, tied to your research questions, but you change the order and follow up with probes depending on what each person says. It is the most common way to collect qualitative data for a PhD in India, in fields from education and nursing to management and sociology.

This page covers the collection side: the guide, how many interviews, consent, interviewing in Tamil, Hindi or another language, running the interview, and turning recordings into transcripts. Analysing the transcripts is a separate job, covered in our thematic analysis guide. Last reviewed September 2026.

Structured vs semi-structured vs unstructured interviews: which one fits a PhD?

Three kinds of research interview compared
	Structured	Semi-structured	Unstructured
What you prepare	Fixed questions, often with fixed answer options	An interview guide: main questions, probes and follow-ups	A topic, perhaps one opening question
Order and wording	Same for everyone	Main questions asked of everyone, order can change	Led by the conversation
What the data look like	Short answers, often coded as numbers	Long accounts you can still compare across participants	Long, varied accounts that are hard to compare
Suits	Surveys read aloud to respondents who cannot fill a form	Most interview-based PhD studies	Life histories, ethnography, early exploration
Main risk	Missing what you did not think to ask	Drifting into a chat that never reaches your research questions	Data you cannot analyse against your objectives

A structured interview is really a questionnaire read aloud (see our questionnaire design guide). Unstructured interviews are harder to defend at a DC meeting than scholars expect. The committee will ask how you will compare one participant with another, and “I will see what comes up” won’t satisfy them.

So most scholars land on semi-structured interviews. If you are unsure whether interviews belong in your design at all, settle that first with our guide to research design types.

How do you write an interview guide for qualitative research?

Start from your research questions. Write each one at the top of a sheet and list what you would need to hear from a participant to answer it. Those become your main questions.

A usable guide has three layers:

Main questions: open, one idea each, in everyday words. Six to ten is usually plenty for 45 minutes to an hour.
Probes: short prompts for detail. “Can you give me an example?” “What happened next?”
Follow-ups: specific sub-questions you ask only if the participant doesn’t raise the point.

Open with easy, factual questions and keep sensitive ones for the middle, once some trust has built up. Close by asking whether there is anything they expected you to ask and you didn’t. The best material often turns up there.

A short illustrative interview guide

Part of a guide for a made-up study, how government school teachers in rural Tamil Nadu use a state teaching app, for the research question “What makes teachers continue or stop using the app?”

Tell me about the first time you used the app in a class. Probe: Which class? What were you trying to do?
Walk me through the last time you used it. Probe: What worked, and what didn’t? Follow-up (if not raised): Was the network or the device a problem that day?
Have there been weeks when you didn’t use it at all? What was going on then?
Who else in the school has a view on the app? Probe: How does the headmaster’s view affect what you do?
If a new teacher joined tomorrow, what would you tell them about it?

Look at what is missing. No “Do you find the app useful?”, which invites a polite yes. No “Don’t you think the training was too short?”, which hands the teacher your answer.

How many interviews are enough for a PhD study?

There is no correct number. What you need is a number you can justify.

A widely cited study is Guest, Bunce and Johnson’s 2006 paper “How many interviews are enough?” in Field Methods. Across 60 in-depth interviews with women in two West African countries, saturation (new interviews no longer adding new themes) came within the first 12, with basic elements present by six. Scholars quote “12 interviews” as a rule, but it was one dataset; a varied sample on a broad topic may take longer.

Hennink, Kaiser and Marconi (2017, Qualitative Health Research) split the idea in two. In their 25 interviews, “code saturation” (hearing the range of issues) came at nine, but “meaning saturation” (understanding them in depth) needed 16 to 24.

Malterud, Siersma and Guassora (2016) proposed information power instead: the more relevant information your sample holds, the fewer participants you need. It depends on five things: the aim of the study, how specific the sample is, use of established theory, the quality of the dialogue, and the analysis strategy. Twelve district officials who actually ran a scheme may tell you more than forty loosely chosen beneficiaries. If you use reflexive thematic analysis, Braun and Clarke argue that saturation doesn’t fit that method at all (our thematic analysis guide explains why).

What I’d put in the synopsis is a planned range, say 15 to 25, with the reasoning, and then in the thesis a plain account of why you stopped where you did. Recruitment is a sampling decision too, so name your method; purposive, snowball and theoretical sampling are covered in our sampling techniques guide.

What should you prepare before the first interview?

Ethics approval and consent

Get ethics committee approval before you contact anyone. Committees usually want the interview guide, the participant information sheet and the consent form, so draft those early; our guide to research ethics approval covers the process. Give participants these documents in a language they read. Consent to be recorded is separate from consent to take part, so plan for people who will talk but not on tape (detailed notes, written up the same day). For participants who cannot read, ask your committee what form of consent it accepts before you reach the field.

Language

Decide which language each interview will be in. If your thesis is in English and the interviews are in Hindi or Telugu, translate the guide and have someone else translate it back into English. Where the back-translation differs from your original, the question probably means something different from what you intended. Words like “stress” or “harassment” rarely have neat equivalents, and many participants will use the English word anyway.

Pilot interviews

Run two or three pilots with people like your participants who won’t be in the final sample. Listen back for one-line answers, misunderstood questions and places where you talked too much. Almost every guide changes after piloting, and reporting what changed is a strength.

How do you conduct a semi-structured interview well?

Most of the skill is listening. If the recording has you explaining your study, finishing their sentences or agreeing loudly, you are shaping the data.

Spend a few minutes on tea and small talk before recording, then confirm consent on the recording.
Use neutral probes. “Tell me more about that” is neutral; “That must have been frustrating” puts words in their mouth.
Allow silence. People often add the important part after a pause.
Ask about specific events (“the last time…”) rather than general opinions. Stories are harder to polish.

Officials, elites and power

A Block Development Officer or a hospital superintendent may give you twenty minutes, a gatekeeper in the room and the official line. Carry a trimmed guide with only the questions you can’t do without, know the scheme well enough that they can’t talk around you, and ask for examples. With contract workers, village women or students in your own department, power runs the other way. They may tell you what they think you want, especially if you came through their employer. Say plainly that there are no right answers and that nothing they say will reach their boss.

Phone and video interviews

Most committees accept them. They tend to be shorter and thinner, and work best with people you have already met. Check that the platform’s storage fits your ethics approval.

How should you record, store and transcribe interviews?

Record on a dedicated recorder or a phone in flight mode, and copy files off the device the same day into encrypted or university storage, never a WhatsApp chat. Keep the key linking codes (P01, P02) to names in a separate protected file. India’s Digital Personal Data Protection Act, 2023 may apply to participants’ personal data, so ask your ethics committee or research cell what your university expects.

Anonymise while transcribing: in a small setting, “the only woman medical officer in the block” identifies her as clearly as her name.

Verbatim or clean transcription?

Verbatim keeps everything: fillers (“ah”, “matlab”), repetitions, false starts, laughter, pauses. Clean (intelligent verbatim) drops fillers and false starts but keeps the words and meaning. Clean is usually enough for thematic analysis; discourse or conversation analysis needs full verbatim. Write down your conventions for pauses, inaudible words and overlapping speech, and apply them to every transcript.

Code-switching and translation

Many participants move between English and their own language mid-sentence. Transcribe in the language spoken, in its script or one consistent romanisation. Then decide whether to translate whole transcripts or only the extracts you quote, and keep the original beside every translated quote in your files.

How long it takes, and automatic tools

Transcribing an hour of clear speech by hand typically takes several hours, and longer for mixed-language talk or poor audio. Build that into your timeline. Automatic transcription gives a first draft, but accuracy on Indian languages and mixed speech varies a lot, so check every line against the audio. Before uploading, check where the tool stores recordings and whether your consent form allows it.

What should you report about interviews in your methodology chapter?

Examiners read this section to decide whether to trust your findings chapters. Cover at least:

Why semi-structured interviews suit your research questions.
Who the participants were, how you found them (sampling method, gatekeepers), how many you approached and how many agreed.
How you decided the number of interviews was enough.
Where each interview happened (office, home, phone, video), who else was present, and how long it lasted.
The language of each interview, who translated and transcribed, and how you checked the translation.
How consent was taken, the ethics approval reference, and how you anonymised the data.
What the pilot changed. Put the final guide in an appendix.
Your own position: your relationship to participants and anything about you (gender, caste, institution, job) that may have shaped their answers.

For a fuller self-check, use COREQ, the Consolidated criteria for reporting qualitative research, published by Tong, Sainsbury and Craig in 2007 in the International Journal for Quality in Health Care. Its 32 items are grouped into three domains: research team and reflexivity, study design, and data analysis and reporting. It was written for health research but works as a self-check in any discipline. See our guide to writing the research methodology chapter for where this section sits.

Semi-structured interviews for qualitative PhD research