# Lesson 10: Collecting Qualitative Data: Interviews and Focus Groups

*Companion-podcast transcript, Sarah and Kiffer*

---

**Sarah:** Welcome back to Office Hours. I'm Sarah.

**Kiffer:** And I'm Kiffer. Today we are talking about interviews and focus groups, which is where many students first collect data that they have to make sense of themselves.

**Sarah:** The last two lessons were about numbers, from surveys and chart reviews to administrative records and data linkage. Why switch to words now?

**Kiffer:** Because numbers leave some questions unanswered. Our running example, the Cedar Valley Social Connection Study, is fictional but built like a real mixed-methods study. The team surveyed sixteen hundred adults aged sixty-five and older, and three hundred and ninety-two of them, about one in four, scored in the lonely range on a short loneliness scale. That tells the health authority how common loneliness is. It cannot tell them what loneliness feels like for a woman who has moved from Cedar City to a small town like Kestrel Lake, or what helped her. Those are questions about experience, and that is the job of the qualitative strand.

**Sarah:** Remind me what that strand involves.

**Kiffer:** It involves twenty-four semi-structured interviews with older adults who live alone, and four focus groups. Two of the groups are with older adults, one is with family caregivers, and one is with clinic staff and community connectors. The interviews answer the qualitative question the team wrote back in Lesson two, using the SPIDER framework, which stands for sample, phenomenon of interest, design, evaluation and research type. The question asks how older adults living alone experience social connection after a move to a smaller town.

**Sarah:** Let's start where the lesson starts. What exactly is a qualitative interview?

**Kiffer:** The simplest definition is a conversation with a purpose. The researcher asks open questions and follows up on the answers, so the participant can describe their experiences and views in their own words. The data are what the person says, how they say it, and what the researcher notices about the setting and the exchange.

**Sarah:** The lesson mentions miners and travellers. That sounded a bit poetic for a methods course.

**Kiffer:** It is poetic, but it is useful. Steinar Kvale described two ways of thinking about the interviewer. In the first, the interviewer is a miner who digs out facts that are already sitting in the participant's head. In the second, the interviewer is a traveller who walks alongside the participant and comes home with an account that both of them shaped. Most qualitative health researchers work closer to the traveller image, which is why so much of this lesson is about question wording, the interviewer's conduct and field notes.

**Sarah:** Then there's the question of structure, from structured to semi-structured to unstructured interviews. Walk me through those.

**Kiffer:** A structured interview asks everyone the same questions in the same order, often with fixed answer options. Honestly, it is a survey read aloud, so it belongs with Lesson eight. An unstructured interview opens with one broad request, something like "tell me the story of your move," and lets the participant lead. Semi-structured sits in between. The interviewer has a guide of open questions but can change the order, skip what has already been answered, and follow something new.

**Sarah:** Let's take the three methods in turn, starting with in-depth interviews.

**Kiffer:** An in-depth interview is one participant and one interviewer, usually for somewhere between forty-five and ninety minutes, exploring that person's own experience in detail. It is private, so it suits sensitive topics like loneliness, grief, money or sexual health. The twenty-four Cedar Valley interviews are in-depth interviews.

**Sarah:** How is a key informant interview different?

**Kiffer:** The person is chosen because their role gives them specialized knowledge of a community, a program or a system, and you ask them mostly about what they know and see. Before writing its guide, the Cedar Valley team interviewed four key informants, including a primary care nurse in North Bench and, under the partnership agreement, the community health director at the Cedar Valley First Nations Health Centre. They learned which roads close in winter, where the internet is poor and where older adults gather.

**Sarah:** Let's turn to focus groups. I think most people picture a market research room with a one-way mirror.

**Kiffer:** That picture is not far off historically. The method grew out of the focused interview that Robert Merton and colleagues developed in the nineteen forties to study how audiences responded to wartime radio and film. Market researchers took it up, and later health researchers did too. Today a focus group is a moderated discussion among about five to ten people who share something relevant to the question.

**Sarah:** What do you get from a group that you can't get from eight separate interviews?

**Kiffer:** The interaction among participants is the difference. Jenny Kitzinger made this argument in the nineteen nineties. When participants question one another, agree, disagree, even tease each other, you see the group's shared language and its unspoken rules. A group of caregivers might reveal a shared norm about never asking the neighbours for help, which no single caregiver would think to mention on their own.

**Sarah:** What are the downsides?

**Kiffer:** Each person speaks for only part of the session, so you get less depth on any one person's story. A confident speaker can pull the group toward their view. And confidentiality is limited. The research team can promise to keep what people say confidential, but it cannot promise that the other participants will.

**Sarah:** Is that why the groups are kept separate?

**Kiffer:** That's part of it. Researchers usually segment groups, meaning that people who share a key characteristic meet together. Cedar Valley keeps caregivers in a different group from clinic staff, because a caregiver may hesitate to criticize the clinic with clinic staff sitting across the table.

**Sarah:** So when a student has a question in front of them, how do they choose?

**Kiffer:** I'd ask four things. Is the question about individual experience or about shared views and norms? How sensitive is the topic? Who actually holds the knowledge you need? And what is practical for the participants, given travel, hearing, schedules and internet service? Work through those, and the method usually becomes clear.

**Sarah:** Then the question everyone asks. How many interviews do I need?

**Kiffer:** Qualitative studies have no formula of the kind you meet for surveys in the epidemiology courses. Researchers lean on two ideas instead. The first is saturation, the point where new interviews add little new information relevant to the question. In one well-known study of sixty interviews, Greg Guest and colleagues found that most of the codes they eventually used had appeared within the first twelve.

**Sarah:** So twelve is the magic number?

**Kiffer:** No, and that is a common misreading of one fairly focused study. The number you need depends on how varied your participants are and how broad your question is. That's where the second idea, information power, comes in. Kirsti Malterud and colleagues argued that a study needs fewer participants when its aim is narrow, its participants have very specific experience, the interviews are rich, and the analysis looks closely at a few cases. Cedar Valley planned twenty-four interviews and will check after sixteen whether new ones are still adding something.

**Sarah:** Let's move to section two, writing the guide. What's actually in an interview guide?

**Kiffer:** It has an opening script, a short list of open questions in a suggested order, probes and prompts for each question, a closing, and short instructions for the interviewer. It does three jobs. It keeps interviews comparable across interviewers, it reminds the interviewer of every topic the research question needs, and it shows the research ethics board what participants will be asked.

**Sarah:** Where do the questions themselves come from?

**Kiffer:** They come from the research question, though only indirectly. Nobody can answer "how do you experience social connection after a move to a smaller town" in one go. So the team breaks it into sub-questions. What did the move involve? How does everyday contact happen now? What makes it harder? What helps? How does it relate to health? Then they write guide questions for each and record the links in a question matrix. The matrix is a simple table that shows whether any part of the question has been missed.

**Sarah:** I noticed one question in the Cedar Valley guide asks about health and the emergency department. That seemed to come from the other strand.

**Kiffer:** It does, deliberately. The quantitative strand asks whether loneliness is associated with emergency department visits. If interview participants describe going to emergency because they had nobody to call at night, that story helps the team interpret the linked administrative data. That is how a mixed-methods study makes its two halves talk to each other.

**Sarah:** Let's talk about the shape of the guide.

**Kiffer:** A guide has four parts. The opening explains the study and confirms consent and recording. A warm-up question asks for something easy, like how long the person has lived in their community. The key questions take most of the time and move from concrete description toward feelings, with the most sensitive questions in the middle once some trust has built up. The closing invites anything else and ends on a forward-looking question.

**Sarah:** Now the craft of writing questions. What makes a good one?

**Kiffer:** A good question is open, so it can't be answered with one word. It asks about one thing at a time, it uses the words participants use, and it is neutral, so it doesn't signal the answer you are hoping for. Questions about concrete experience usually work better than requests for opinions, because people find it easier to describe what happened than to state a general view.

**Sarah:** Give me an example of a bad question turning into a good one.

**Kiffer:** "Do you feel lonely?" It invites a yes or no, and the word carries stigma. The Cedar Valley version reads, "Some people tell us they feel lonely at times, and others say they rarely do. What has it been like for you since the move?" The first sentence tells the participant that either answer is fine. The second invites a description.

**Sarah:** What about leading questions?

**Kiffer:** "Don't you think the health authority should provide more transport for seniors?" You can hear the answer the interviewer wants. The revision asks, "What, if anything, would make it easier for you to get to the places you want to go?" Now transport might matter, or it might not, and the participant decides.

**Sarah:** And grand tour questions?

**Kiffer:** That term comes from the anthropologist James Spradley. A grand tour question is a broad descriptive request, like "could you walk me through a typical week now?" It hands the participant control of the account and gives you plenty to follow up on. The third Cedar Valley question is a grand tour question.

**Sarah:** Which brings us to probes and prompts. I always mix those up.

**Kiffer:** Most people do. A probe follows up on something the participant has already said. "Can you tell me more about that?" "Can you give me an example?" "When you say it goes around, what do you mean?" Repeating the participant's words back to them is a probe, and so is silence. A prompt is a planned reminder written into the guide, which you raise only if the participant has not mentioned that topic.

**Sarah:** So if the guide lists video calls as a prompt and the participant has already talked about video calls, I skip it.

**Kiffer:** That's right. If they have not mentioned it, you raise it. Reading every prompt aloud at the start of the question would steer the answer toward your list.

**Sarah:** How is a focus group guide different?

**Kiffer:** Richard Krueger and Mary Anne Casey call it a question route, and it has five kinds of question. An opening question goes round the circle so everybody speaks in the first few minutes. Introductory and transition questions bring the group toward the topic. A small number of key questions take most of the time. And ending questions ask what matters most, check a summary, and invite anything the group thinks was missed.

**Sarah:** And then piloting. Is that really necessary for a small study?

**Kiffer:** Even one pilot interview with a volunteer will teach you something. Hanna Kallio and colleagues describe three kinds of testing: within the team, review by experts or by people like your participants, and field testing in real conditions. Cedar Valley did all three, and the interviewers found that the draft ran about eighty minutes against a target of sixty.

**Sarah:** What changed as a result?

**Kiffer:** The pilot changed five things. "Social connection" became "keeping in touch with people," because the advisory group found the research term vague. The loneliness question moved from second to fifth place, because one pilot participant became guarded when it came too early. Prompts about winter roads and internet service went in. Two questions that overlapped with others came out, which brought the interview down to about an hour. And the team printed a large-print card of the main topics, because one pilot participant asked what was coming next.

**Sarah:** Section three takes us into the room. What does preparation look like?

**Kiffer:** It starts with the setting. Private, quiet, accessible and convenient for the participant. For many older adults that's their own kitchen table. Others prefer a private room at the library, where family members won't overhear. Consent follows the process from Lesson five, confirmed before the recorder starts. And always two recorders, because a recording that fails cannot be repeated.

**Sarah:** Once the recorder's running, what separates a good interviewer from an average one?

**Kiffer:** The difference is listening. Rapport is the trust that lets someone speak candidly, and it builds from small things like arriving on time and showing real interest. Active listening means you hear the phrase that deserves a follow-up and you ask about it before moving on. When the Cedar Valley participant says her daughter was the one who found her this place, a good interviewer hears that there is something more there.

**Sarah:** What habits should people avoid?

**Kiffer:** Filling every pause is the big one. New interviewers find silence uncomfortable, so they rush to the next question. If you wait five seconds, people very often add the most thoughtful part of their answer. Other habits are approving answers with "that's great," sharing your own story, asking two questions at once, and reading the guide word for word.

**Sarah:** Let's talk about the distress protocol. That sounds serious.

**Kiffer:** It is a planned set of steps, approved by the research ethics board, for when a participant becomes upset. Talking about loneliness, bereavement or losing independence can bring up strong feelings. The interviewer pauses, acknowledges what the person has shared, and offers a break, the chance to skip the topic, or the choice to stop. The interview continues only if the participant wants it to. At the end, the interviewer checks how they are and gives them a sheet of local supports.

**Sarah:** What if it's more than sadness?

**Kiffer:** If something suggests a risk of serious harm, the interviewer follows the steps set out in the protocol and the consent form, which already explained the limits of confidentiality. Then the interviewer debriefs with Doctor Hart, who leads the study. The protocol keeps the interviewer in a research role and makes sure the participant is connected with people trained to help.

**Sarah:** Let's move to focus groups. Who's in the room besides the participants?

**Kiffer:** There are two people from the team. The moderator welcomes everyone, sets the ground rules, asks the questions and manages the discussion. The assistant moderator runs the recorders, handles latecomers, takes notes on what is said and on body language, and logs who speaks in what order, which helps enormously when someone later has to transcribe seven voices.

**Sarah:** What's the hardest part of moderating?

**Kiffer:** Usually it is managing the balance of voices. Picture the caregiver group. One man answers every question first and at length. Two women haven't spoken since the opening round. You thank him, turn your body toward the others, and say you'd like to hear from people who haven't had a chance yet. You might invite one of the quieter participants gently by name.

**Sarah:** What about when everyone agrees too quickly?

**Kiffer:** You get suspicious, politely. You might say, "Some people have told us the opposite. Has anyone here found that?" Early agreement can just mean people are being polite to the first confident speaker. If nobody disagrees after a real invitation, then the agreement means more.

**Sarah:** That leads us to field notes.

**Kiffer:** Field notes record what the audio cannot. The setting, who else was there, body language, interruptions, anything said before the recorder started, and your first impressions. The Cedar Valley team writes them up the same day on a contact summary form, a one-page sheet listing the main points, anything new, and what to follow up next time.

**Sarah:** The lesson includes an example from interview fourteen.

**Kiffer:** That is the interview you meet again as a transcript in section four. The interviewer noted that they sat at the participant's kitchen table, that her phone rang once and she let it ring, and that she paused for several seconds before describing her first months in Kestrel Lake. The most surprising point was that a personal phone call from the librarian brought her to the coffee morning after a notice on the library board had failed to.

**Sarah:** Then there's reflexivity, which I think intimidates students.

**Kiffer:** It sounds abstract, but the idea is simple. Reflexivity means paying ongoing attention to how your own background, assumptions and position shape what participants tell you and how you read it. Positionality is your social position relative to the participants. The Cedar Valley community research associate grew up in North Bench and knows many residents. That helps with trust, and it may also make some people more guarded about local matters. The graduate research assistant grew up in a large city and moved to the region for school, so she is an outsider to small-town life.

**Sarah:** The reflexive memo in the reading is quite personal.

**Kiffer:** It is, and that is what makes it useful. After interview fourteen, the graduate research assistant wrote that when the participant joked about being put in a care home by Christmas, she felt the urge to reassure her and moved on instead. On reflection, she thought she had moved on too quickly, because the joke may have carried a real fear about losing independence. She also noticed that she had expected loneliness to be about distance from family, when the participant described losing casual daily contact. Memos like this later inform the methods section of a report.

**Sarah:** In section four, we have an audio file. What happens to it?

**Kiffer:** The first thing to recognize is that a recording is identifiable personal information. A voice can identify someone, and participants mention names and places. So the team names files with study codes only, moves them the same day to encrypted university storage, checks that the copy plays, and then deletes them from the recorders. If an interview was on video and the analysis needs only the words, they keep the audio and delete the video.

**Sarah:** And then transcription. Isn't that just typing?

**Kiffer:** It looks like typing, but every transcript reflects decisions about what to keep. Spoken language is full of hesitations, false starts, laughter, interruptions and people talking over each other. Oliver and colleagues described a range from naturalized transcription, which records every detail, to denaturalized transcription, which strips those features out so the content reads clearly. Where you land on that range depends on your analysis.

**Sarah:** Now for the three conventions: verbatim, intelligent verbatim and clean verbatim.

**Kiffer:** Verbatim keeps everything, including every um and uh, every false start, the interviewer's mm-hmms, pauses with their length, sighs and laughter. Intelligent verbatim removes the fillers, false starts and those little interviewer responses, but keeps the speaker's own grammar and marks noticeable pauses and laughter. Clean verbatim goes one step further. It corrects slips of grammar and keeps a bracketed note only where it changes how the words should be read. Transcription services use these labels inconsistently, so write down exactly what your convention keeps and removes, whatever you call it.

**Sarah:** Let's talk about the excerpt. It's invented, right?

**Kiffer:** It is completely invented, written for this course, and labelled that way. It comes from interview fourteen with a participant we call Ruth, aged seventy-eight, who moved from Cedar City to Kestrel Lake. In under four minutes she describes losing casual contact, hiding her loneliness from her daughter, a hard first winter without a car, the library coffee morning, and the risks of gossip in a small town.

**Sarah:** Give me one place where the three versions differ.

**Kiffer:** Talking about her daughter, she says, "If I'd said I was lonely, she'd of had me in a care home by Christmas," and then she laughs. Verbatim and intelligent verbatim keep her grammar exactly as spoken. Clean verbatim corrects it. All three keep the laugh, because without it you might read the line as a plain complaint about her daughter. The laugh tells you she is presenting a real fear as a joke.

**Sarah:** And what gets lost in the clean version?

**Kiffer:** The clean version loses a sigh after the word "quiet," the interviewer's own laugh, and a phone that rang in the middle of an answer. Some of that matters, and some of it doesn't. The interviewer's laugh connects to the reflexive memo, so a careful analyst would want to know about it. That is one reason the audio stays available during analysis.

**Sarah:** Which convention did the Cedar Valley team choose?

**Kiffer:** They chose clean verbatim, with a written rule that pauses, laughter and other sounds stay in wherever they change how the words should be read. Their question is about what participants experienced, and their analysis works with the content of what people say. A team studying how people take turns in conversation would need full verbatim with much more detailed notation, like the system Gail Jefferson developed for conversation analysis.

**Sarah:** Let's get to automated transcription, because I suspect every student listening has used it.

**Kiffer:** They probably have. Speech recognition software turns audio into a draft, often with timestamps and a guess at who is speaking. It saves a lot of time, since typing up an hour of audio by hand takes several hours. But the drafts need correcting against the recording. Accuracy drops with background noise, people talking over each other, soft voices, accents and local place names, and error rates have been shown to differ across groups of speakers.

**Sarah:** The reading also says some tools invent words.

**Kiffer:** Some speech recognition models have been found to insert words or phrases that nobody said, often during pauses. Koenecke and colleagues documented this in twenty twenty-four. So the rule is simple. Someone who knows the interview reads every draft against the full audio, paying special attention to names, numbers, quiet passages and anything that will be quoted.

**Sarah:** What about privacy?

**Kiffer:** That's the bigger issue. Uploading an interview to a cloud service sends identifiable voices to a company's servers, which may be outside Canada, and the company may keep the files or use them to improve its products. In British Columbia, universities and health authorities are subject to the provincial Freedom of Information and Protection of Privacy Act, and the national research ethics policy has a chapter on privacy and confidentiality. Moderators should also switch off automatic meeting transcripts before an online focus group.

**Sarah:** So what does Cedar Valley actually do?

**Kiffer:** For the twenty-four interviews, a speech recognition program installed on an encrypted university laptop produces a draft without sending any audio over the internet. The interviewer who conducted the interview corrects that draft against the recording. The four focus groups go to a professional transcriber who has signed a confidentiality agreement and stores files in Canada, because overlapping voices defeat most automated tools. A second team member checks one transcript in four.

**Sarah:** Then comes de-identification.

**Kiffer:** Names and identifying details are replaced with bracketed descriptions, like "daughter's name" in square brackets. A separate, protected linking key connects participant codes to real names, and one named team member holds it. In a small town you have to think harder, because ordinary details add up. A seventy-eight-year-old widow who moved to Kestrel Lake two years ago, stopped driving after cataract surgery and goes to the library coffee morning may be recognizable to half the town.

**Sarah:** So what does the team do with those details?

**Kiffer:** They keep community names in the analysis transcripts, because place matters to the question, and they generalize them in anything published. So she becomes a woman in her late seventies who moved to a small town in the region. Recordings and transcripts from First Nations participants follow the data governance agreement with the health centre, which sets out who may transcribe them, who may access them, and where they are stored.

**Sarah:** Let's finish by putting a guide together. What does a complete draft contain?

**Kiffer:** A complete draft follows the structure from section two: an opening script, one warm-up question, six to eight key questions with probes and prompts, and a closing. A question matrix links every key question to part of the research question. Then the team pilots it, even with one colleague, and notes what it changed.

**Sarah:** What goes in the transcription plan?

**Kiffer:** It should be about half a page. It says how sessions will be recorded, which convention will be used and why, whether automated transcription will be used and how privacy is protected if it is, and how the transcripts will be checked and de-identified. Together with the recruitment plan and the data management plan, it completes the data collection plan for an interview study.

**Sarah:** Any last advice before people start writing questions?

**Kiffer:** Read your questions aloud before you pilot them. If a question sounds like something you would never say to a real person across a kitchen table, rewrite it until it does.

**Sarah:** That's a good test. Thanks, Kiffer.

**Kiffer:** Thanks, Sarah. Next time, in Lesson eleven, we take the first steps in analysis, and that includes building a first codebook and coding the clean version of the Kestrel Lake excerpt.

**Sarah:** See you then on Office Hours.