Thesis or Dissertation Interviews: How to Plan, Conduct, and Report Them

Interviews produce the evidence in most qualitative dissertations, and the quality of the interviews sets the ceiling on the quality of Chapter 4. A weak interview cannot be repaired during analysis. Questions that suggest the answer, a protocol that does not connect to the research questions, or a transcript that is not accurate will show up later as findings the committee does not trust.

This article follows an interview study from planning to reporting. Chapter 3 shows where each step appears, and the guide to Chapter 3 covers the chapter as a whole.

Choosing the type of interview

Interviews run on a spectrum of structure.

Structured interviews use the same questions in the same order for every participant. They suit studies that need answers to be comparable, and they leave little room for what participants consider important.

Semi-structured interviews follow a protocol of main questions, with follow-up probes that vary by participant. Most dissertations use this type, because it keeps every interview on your research questions while leaving room to pursue what a participant raises.

Unstructured interviews work from a topic and a few prompts, and the conversation follows the participant. They fit some phenomenological and narrative designs, and they demand skill and a long time.

Choose by asking what your research questions need. Name the type in Chapter 3 and say why it fits your design.

Writing the protocol from your research questions

The protocol is a short document: an opening script, the main questions with probes, and a closing. Its structure comes from your research questions, and a committee will check that link.

Build it in a table before you write the final wording.

Research question	Interview question	Probe
How do principals describe the practices they use to keep teachers?	"Tell me about what you do to keep your teachers at the school."	"Can you give me an example?"
Which practices do principals judge most effective?	"Which of those has made the biggest difference, in your experience?"	"What makes you say that?"
What conditions limit those practices?	"What gets in the way of doing more?"	"How does that show up day to day?"

Every research question should have at least one interview question feeding it. Every interview question should serve a research question. If one does not, remove it or find out which question it answers. Include the table, or a version of it, in Chapter 3 or an appendix, because it shows the committee that the instrument can answer your questions.

Writing the questions themselves

Ask open questions that invite description. "Tell me about," "Walk me through," and "What was that like" produce accounts. Questions that can be answered yes or no produce little.

Avoid leading questions. Compare "Tell me how you decided to stay at your school" with "Don't you agree that heavy workload is why most teachers leave?" The second tells the participant what you expect and tends to produce agreement. It also adds your assumption to the data.

Ask one thing at a time. A question with three parts gets an answer to the last part only. Use words your participants use, not the terms of your literature review, unless they are the same.

Order the questions from easy to demanding. Begin with background and easy descriptions, move to the central questions, and end by asking whether the participant wants to add anything. That final question often gives a useful result.

Piloting

Try the protocol before the real interviews, with one or two people similar to your participants but outside your sample. Listen for questions that confuse, questions that get the same answer, and timing that runs too long. Revise the protocol after the pilot.

Piloting involves people, so it may need IRB approval first. Many boards treat a pilot as part of the study. Ask your board before you invite anyone. Our guide to IRB approval step by step explains the order of events.

How many participants

The number depends on the design, the population, and how much each interview yields. A phenomenological study often uses fewer participants, each interviewed in depth, than a study that compares groups. Qualitative researchers often judge sufficiency by whether new interviews stopped adding new themes, a point usually called saturation, though traditions define it differently. Justify your number in Chapter 3. Our article on justifying a smaller sample size shows how to do it.

Conducting the interview

Prepare the practical side first. Test your recorder or recording software before every session. Choose a setting the participant can speak in freely, and agree how to handle interruptions. For online interviews, check the platform's recording and storage settings against your IRB approval.

Begin by reading the consent information again and checking that the participant is still willing. Then say what the interview covers and how long it should take.

During the interview, listen more than you speak. Use the probes in your protocol, and add your own where the participant says something unclear. "What do you mean by that?" and "Can you say more?" are neutral and productive. Do not correct a participant, argue with a view, or offer your own experience in the middle of an answer, because it changes what they say next.

Write notes immediately after each interview. Record what you noticed about the setting, the participant's manner, and any ideas for analysis. These notes become part of your audit trail.

Recording and transcribing

Record with consent, store the files as your IRB approved, and transcribe each interview in full. Check the transcript against the recording. Automatic transcription tools make errors, especially with accents, technical terms, and overlapping speech, and an error in a quotation is an error in your evidence. Our article on presenting themes clearly explains why quotations carry so much weight in Chapter 4.

Replace names and identifying details with pseudonyms or codes in the working transcripts, and keep the key separately. Do not change what participants said when you use their words in the dissertation, apart from the minimal edits your program permits and discloses.

If you send transcripts or summaries back to participants to check, which is often called member checking, describe how you did it and what you changed. Report it only if you did it.

Reporting interviews in Chapter 3 and Chapter 4

Chapter 3 describes the process. State the type of interview and why, how participants were selected and how many took part, how long interviews lasted, how and where they were held, how they were recorded and transcribed, and how you protected confidentiality. Report what happened, including any change from the plan. Our article on what to write about IRB approval in Chapter 3 covers the ethics part.

Chapter 4 reports what participants said, organized around your research questions, with quotations that illustrate each theme. The protocol table makes this organization easy to follow.

What committees flag most often

• Interview questions that do not map to any research question.
• Leading questions in the protocol.
• No mention of a pilot, or a pilot that happened before approval.
• A sample size with no justification.
• Transcripts not checked against recordings.
• A description of member checking, triangulation, or reflexivity that did not happen.
• Participants who are not described, so that the reader cannot judge whom the findings represent.

If you want an independent read on whether your interview design and protocol can answer your research questions, our research design and methodology review examines them. The questions, the participants, and the findings stay yours.