The big idea: Good interview data starts with how you treat people: when they understand, agree and feel safe, they tell you what is really going on.
In 60 seconds
- Good interview data starts with how you treat people. Informed consent means they join freely, understand the project and know the risks.
- Confidential is not anonymous. Use codes or roles, keep files in one team-only place and never paste names into a public AI tool.
- Trustworthy data is valid and reliable: one shared guide, one note format, and exact quotes kept apart from your interpretation.
Do this: set up Deliverable #3 with a consent opener, one shared question guide and a log that records consent for every interview.
How ethical, trustworthy interviews work
Your interviews are the raw material for everything that follows: themes, insights, prototypes and, later, your senior project. Two questions decide whether that material is any good. Did you treat people right? And can anyone trust what you collected?
Plan
Pick a mix of stakeholders, write one shared question guide and check whether you need approval.
Ask: who could be harmed or pressured by this?
Ask
Explain who you are, what the project is and that saying no is fine.
Say: “You can skip any question or stop at any time.”
Listen and record
Ask before recording or taking a photo. Capture exact words and what you see.
Keep: quotes in quotation marks, your thoughts labeled as notes.
Protect
Use codes or roles instead of names. Keep files in one team-only place.
Never: paste names or private details into a public AI tool.
Close the loop
Thank people, keep every promise you made and share back what you learned if you offered to.
Result: people would talk to your team again.
Two jobs at once
Ethics: treat people right
- Informed consent. People join of their own free will, understand the project and know about any risks before they say yes.
- Confidentiality. You know who said it, but you do not release anything that identifies them without permission.
- Freedom from harm. No one should be put at risk: socially, at work, emotionally or legally.
- No deception. Never hide the purpose or pretend to be someone you are not. The class slides put it plainly: there is no place for it in action research.
Quality: data you can trust
- Validity. Are you capturing what is really happening, or what people think you want to hear?
- Reliability. Would your team collect the same kind of data again, no matter who did the interview?
- Generalizability. Not the goal here. Action research aims to understand this setting and improve it.
- Bias check. Watch for confirmation bias: hearing only what proves the idea you already like.
Validity and reliability in plain language
Validity: is it true?
Your method measures what you say it measures. If you ask a nurse “Is the discharge process frustrating?” you planted the answer. Ask “Walk me through yesterday’s last discharge” instead.
Watch out: leading questions, interviewing only fans, and quoting the one dramatic story.
Reliability: is it consistent?
Your data would come out the same way if a teammate ran the interview. Same core guide, same note format, notes written the same day.
Watch out: four interviewers with four different question lists. You cannot compare what you collected.
The catch
Reliable is not the same as valid. A scale that always reads five pounds heavy is consistent, and consistently wrong. Your guide can be the same every time and still ask the wrong thing.
Example: a city permit office survey that asks only about staff friendliness will reliably miss the real problem: wait times.
Confidential is not anonymous
Anonymous means even your team does not know who said it, like an unsigned comment card. Confidential means you know, but you protect it. Almost every interview you do is confidential, so do not promise anonymity.
A quick ethics gut check
Would you be comfortable if the person you interviewed read your notes, your slides and your report? If not, fix it before it leaves your team.
Informed consent, step by stepWhat to say before the first question
Consent is an ongoing conversation, not a box you tick once. In the class slides it has three parts: people take part of their own free will, they understand the study, and they know about any possible risks.
- Who you are. “We are Haslam honors students working on a class consulting project.”
- What it is for. One sentence on the challenge, and where the results will go (class, a client, a poster).
- Voluntary. “You can skip any question or stop at any time, no problem.”
- Recording and photos. Ask separately for each. A yes to notes is not a yes to audio. A yes to audio is not a yes to a photo.
- Naming. “How would you like to be described: by name, by role only, or not at all?”
- Check again. If the talk turns personal, pause: “Are you still comfortable with us using this?”
Written permission is sometimes needed, for example when results will be published. Ask your instructors which form to use.
Confidentiality in practiceCodes, folders, quotes and AI
| Situation | Do this |
|---|---|
| Writing notes | Use a code and a role (“B4, small manufacturer, owner”) instead of a name. |
| Storing files | One team-only folder for notes, audio and photos. Delete what you no longer need when the project ends. |
| Quoting in slides | Quote the role, not the person, unless they agreed to be named. Remove details that give someone away (“the only night-shift pharmacist”). |
| Using AI | The course AI policy says not to enter confidential client, company or personal data into public AI tools. Strip names and identifying details first, or do that part by hand. |
| Talking between interviews | Never tell one person what another person said. In a small office, a “harmless” repeat can cost someone their job. |
Power, pressure and people under 18When saying no is harder than it looks
Consent only counts if people feel free to say no. Watch for these situations:
- Power gaps. Interviewing your own employee, your boss, a professor who grades you, or staff while their manager stands nearby. Offer a private setting and make it easy to decline.
- Gatekeepers. If a manager “volunteers” their team for you, ask each person yourself whether they want to take part.
- Under 18. The class slides are clear: if participants are underage, you need parent or guardian permission first, stating what data may be shared, for what purpose and with whom.
- Sensitive topics. Health, money, job security, family. Plan how you will pause, skip or end kindly if someone gets upset.
Trustworthy data: the deeper criteriaGuba, Maxwell and Anderson from the class slides
Qualitative research judges quality differently from a survey with numbers. The class slides list three sets of criteria. You do not need to memorize them. Use them as questions.
Trustworthiness (Guba)
Credibility: are the findings believable? Transferability: could they apply in similar settings? Dependability: is the data stable and well kept? Confirmability: would a neutral person reach the same reading?
Understanding (Maxwell)
Descriptive: are the facts accurate? Interpretive: did you capture the person’s meaning, not yours? Theoretical: does it explain what is going on? Evaluative: can you present it without judging people?
Action research (Anderson)
Democratic: did you hear all the people affected? Outcome: did it lead to a resolution? Process: was it done in a dependable, competent way? Catalytic: did it spark action? Dialogic: did peers review it?
Habits that protect validity (Wolcott)
Talk little and listen a lot. Record accurately. Begin writing early. Let readers see for themselves. Report fully. Be candid. Seek feedback. Write accurately.
Confronting your own biasFour protections from class
- Be systematic. Same guide, same note format, every interview logged.
- Write down your hunches first. List what you expect to find before you start, so you can see when the data disagrees.
- Watch for confirmation bias. Look hard for the quote that breaks your favorite theory.
- Keep a beginner’s mindset. The class slides call it the best protection: assume you do not know yet.
Do you need approval?Codes of conduct and the IRB
Researchers answer to professional codes (such as the American Psychological Association), national guidelines (the National Research Act of 1974) and their institution’s Institutional Review Board, the IRB. The class checklist starts with: determine whether you need approval, and get it if you do.
Class projects that stay in class are often handled differently from research that will be published or presented publicly. Before you plan to share results beyond class (for example a thesis or a EUReCA poster), ask your instructors what review your project needs. This is not legal advice.
If review is needed, plan to show clear data collection plans, proof that taking part is not coerced, an explanation of the study, permission forms and your question guide.
Try it: Interview Readiness and Ethics Check
Tool · BUAD 317
Interview Readiness and Ethics Check
Run this before your team’s first interview. Must-dos change with your situation.
- 1Set situation
- 2Tick
- 3Share
All flags
Put it to work: Deliverable #3 Stakeholder Interviews
Ethics and data quality show up first in Deliverable #3, Stakeholder Interviews: 10 or more solid interviews with a real mix of stakeholders, logged in the SmithGLS Interview Log, with a photo of each participant taken with permission. They come back in Deliverable #6, when you take your prototype to the street. Set them up once and use them all semester.
Set up your team in six steps
Run the readiness check
Use the tool above with your team’s real situation. Fix every red flag before anyone books an interview.
Your deliverable: a short “how we protect people” note at the top of your interview log, listing the flags you found and how you fixed them.
Add a consent opener to your guide
Put three or four sentences at the top of your Deliverable #2 question list: who you are, what it is for, that it is voluntary, and how they want to be named.
Your deliverable: the opener every interviewer reads, word for word, at the start of each interview.
Agree on one guide and one note format
Everyone uses the same core questions and the same log columns. Follow-ups can vary. The core cannot.
Your deliverable: one shared question guide and a consistent row in the interview log for each person.
Ask for photos and names out loud
Ask separately for a photo and for how they want to be named. Record the answer in the log.
Your deliverable: a photo of each participant taken with permission, and a naming choice for every quote.
Keep evidence apart from interpretation
Exact words go in quotation marks. Your observations and guesses go in a separate notes column.
Your deliverable: sticky notes with one quote or insight each, ready for your in-class share-out.
Close the loop
Thank each person. Keep any promise you made, such as sending a summary.
Your deliverable: a team list of thank-you notes sent and promises kept before Deliverable #6.
Sample consent openerA script you can adapt in one minute
“Hi, we are Haslam honors students working on a class project about [your team’s challenge]. We would like to hear about your experience. It takes about 20 minutes. You can skip any question or stop at any time. We will not use your name unless you say we can. How would you like to be described: by name, by your role, or not at all? Is it OK if we take notes? Could we record audio so we get your words right? We will only use it for this project.”
Ask about a photo at the end, once trust is built, and accept a no without pushing.
What a clean log row looks likeOne example from a non-marketing setting
| Column | Example: a small manufacturer’s onboarding |
|---|---|
| Code and role | M3, line supervisor, 6 years |
| Consent | Notes yes, audio no, photo yes (hands at workstation only), role only |
| Quote | “Nobody shows the new folks where the real checklist is.” |
| Observation | Kept a handwritten checklist taped inside the toolbox lid. |
| Our interpretation | The official training may not match how the work is really done. Check with other supervisors. |
Before you submitCheck these before Deliverable #3
- [ ] Every interview has a consent note in the log (notes, audio, photo, naming).
- [ ] Every photo was taken with permission.
- [ ] Notes and slides use codes or roles unless someone agreed to be named.
- [ ] No names or identifying details went into a public AI tool.
- [ ] All interviewers used the same core question guide.
- [ ] We have a real mix of stakeholders, not ten versions of one person.
- [ ] Quotes are exact and kept apart from our interpretation.
- [ ] If results will go beyond class, we asked our instructors about approval first.
Go deeper
Back to all BUAD 317 topics · Next: From Interview to Insight: Analysis and Synthesis
In-class team activity · 25 min
Red Flag Rehearsal
By the end: A tested consent opener, a fixed list of your team's top ethics and quality risks, and one improved question, ready for your Stakeholder Interviews. Your instructor will give you the password in class.
Part of the Smith GLS human-centered leadership sequence: IB 407, BUAD 317 and BUAD 497, toward the Human Centered Consulting Certificate.
Frequently asked questions
What is informed consent in research interviews?
Informed consent means people take part of their own free will, understand the study and know about any possible risks before they say yes. Before the first question, say who you are, what the project is for and that they can skip any question or stop at any time, then ask separately about notes, audio and photos. Consent is an ongoing conversation, not a box you tick once.
What is the difference between confidential and anonymous research?
Anonymous means even your team does not know who said something, like an unsigned comment card. Confidential means you know who said it but you protect it and do not release anything that identifies them without permission. Almost every student interview is confidential, not anonymous, so do not promise anonymity to the people you interview.
What is the difference between validity and reliability?
Validity asks whether your data is true: are you capturing what is really happening, or what people think you want to hear? Reliability asks whether it is consistent: would a teammate collect the same kind of data? Reliable is not the same as valid. A scale that always reads five pounds heavy is consistent, and consistently wrong.
How do you protect interview participants' confidentiality?
Use a code and a role in your notes instead of a name, keep notes, audio and photos in one team-only folder, and quote roles rather than people unless they agreed to be named. Remove details that give someone away, such as "the only night-shift pharmacist." Never tell one person what another said, and never paste names or identifying details into a public AI tool.
Do student research projects need IRB approval?
It depends on how the results will be shared. Class projects that stay in class are often handled differently from research that will be published or presented publicly, such as a thesis or a EUReCA poster. The class checklist starts with determining whether you need approval and getting it if you do, so ask your instructors what review your project needs.