Research Methodology

Writing Informed Consent for Qualitative Interviews: What IRBs Actually Want to See

A practical guide to writing informed consent forms for qualitative research — covering audio recording, transcription, direct quotes, withdrawal, and confidentiality in small-community studies.

Angel Reyes·April 14, 2026·15 min read

Writing Informed Consent for Qualitative Interviews: What IRBs Actually Want to See

Most consent forms attached to qualitative interview protocols are written in the voice of a Phase II drug trial. They describe "study procedures" that are actually a ninety-minute conversation and warn about "side effects" when the real risk is that a participant's employer might recognize her quote in a published article. Generic medical-style templates applied to interview research produce forms that participants skim without comprehension and that IRB reviewers recognize instantly as copy-paste jobs. The result is a document that protects no one and delays approval by weeks.

The fix is to write a form that accurately describes what happens in qualitative research — recording, transcription, quotation, re-identification risk in small communities, the near-impossibility of full withdrawal after data collection — using the specific language IRBs look for under 45 CFR 46.116. This article walks through that language section by section.

The federal consent regulation — 45 CFR 46.116 — was written in a biomedical frame. Its default assumptions are that risks are physical and predictable, that the researcher controls the encounter, and that identity can be protected by assigning a subject number. Every one of those assumptions breaks down in a qualitative interview.

Interview risk is rarely physical. It is psychological, social, economic, and reputational — embarrassment about topics raised, distress when discussing trauma, damage to a professional relationship if an employer hears about participation, retaliation in small communities where the researcher cannot guarantee that participation itself stays private. These risks are also non-generalizable: what counts as a risk depends on who the participant is and where they live.

The researcher also controls far less of the encounter than in a clinical trial. Participants raise topics the researcher did not ask about, name other people, and volunteer information about illegal activity or private family matters the protocol did not anticipate. A consent form that promises the conversation will "stay within the study topics" is already lying at the point of signature.

Finally, identity is not reducible to a subject number. Direct quotes — the core output of qualitative analysis — carry voice, cadence, regional phrasing, and references that re-identify the speaker to anyone in their community. The question for qualitative consent is not whether you can protect anonymity but how honestly you can describe the limits of the protection you actually offer.

The Required Elements Under 45 CFR 46.116 (Translated for Interview Research)

Before drafting, list every required element of consent from the Common Rule and write one paragraph for each, framed for your interview study. Use the Free Research Ethics Checklist to track completeness. The core elements an IRB expects to see:

  1. A statement that the study involves research — named as research, not as "a conversation" or "a project."
  2. The purpose, duration, and procedures — written in plain language. "One interview of approximately 60 to 90 minutes, audio-recorded, conducted via Zoom or in person at a location of your choosing."
  3. Reasonably foreseeable risks — including psychological and social risks specific to qualitative research.
  4. Reasonably expected benefits — which for interview studies are usually minimal and should not be overstated.
  5. Confidentiality provisions — including the limits of confidentiality when quotes are published.
  6. Voluntariness and the right to withdraw — with honest language about what withdrawal can and cannot accomplish after data has been analyzed.
  7. Contact information for the researcher and the IRB.
  8. For audio or video recording — explicit, separable consent for recording, with the option to participate without being recorded when feasible.

Under the 2018 revised Common Rule, consent documents must also include a concise summary of key information at the top — what a reasonable person would want to know in order to decide whether to participate. For interview studies, that summary should name the three things participants most often underestimate: the recording, the quotation, and the limits of confidentiality.

For adaptable consent, assent, and parental-permission templates — plus annotated sample text for interview studies, focus groups, and research with vulnerable populations — see IRB Ready, which maintains templates keyed to the three review categories.

Recording is not a minor procedural detail. It is the aspect of interview research that most often generates IRB stipulations, because a recording is a copy of a participant's voice — a biometric identifier that cannot be de-identified by removing a name.

Three principles govern recording consent. First, the consent to be recorded must be separable from the consent to participate — a participant who agrees to the interview but not to the recording should still be able to take part (with the researcher taking notes). Second, the form must specify who will hear the recording, who will transcribe it, where it is stored, how long it is retained, and when it is destroyed. Third, the form must name any third-party transcription service and confirm a confidentiality agreement is in place.

Sample language:

This interview will be audio-recorded if you agree. The recording will be stored on an encrypted, password-protected drive accessible only to the principal investigator and the transcription service listed below. A professional transcription service ([Service Name], which has signed a confidentiality agreement with our university) will produce a written transcript. Your name will be replaced with a pseudonym in the transcript, and any identifying details you mention (employer, neighborhood, family members' names) will be redacted or generalized. The original audio file will be destroyed [within 90 days of transcription / at the end of the study / on [DATE]]. The de-identified transcript will be retained for [X years] as required by [institution/funder].

Please initial one: ___ I agree to be audio-recorded. ___ I do not agree to be audio-recorded. I understand the researcher will take written notes instead.

For video interviews and remote platforms, disclose whether cloud recording is used, whether auto-generated transcripts are saved, and whether the recording is downloaded and deleted from the platform immediately after the session. IRBs increasingly ask about this explicitly.

Confidentiality vs. Anonymity: A Distinction Your Form Must Name

One of the most common errors in qualitative consent forms is using the word anonymous when the study is only confidential. A study is anonymous only if the researcher cannot link data to the participant's identity — never true of an interview, because the researcher met the participant. Interview studies are, by definition, confidential, not anonymous. IRB reviewers notice this error immediately. The fix is simple:

This study is confidential, not anonymous. The researcher will know your identity because you will be speaking with her directly. However, your name will not appear in any transcript, field notes, published article, or presentation. A pseudonym will be used instead. Identifying details you mention during the interview — such as your employer, the names of colleagues or family members, or the specific neighborhood where you live — will be changed or generalized to protect your identity.

Then name the limits honestly. Mandatory reporting obligations (child abuse, elder abuse, imminent harm) must be disclosed. Legal compulsion (subpoena) must be disclosed if your data type does not have a Certificate of Confidentiality. And re-identification risk in small communities must be disclosed even when it feels like it might deter participation. Disclosure is the point.

The publishable output of qualitative analysis is usually quotation. This is also how participants are most often re-identified — by a colleague who recognizes a turn of phrase, a family member who recognizes a story, or a small-community reader who recognizes the only person it could possibly be. Your consent form should treat quotation as a distinct consentable act, and it should treat member checking as another one.

For direct quotes, consider a tiered consent:

With your permission, direct quotes from this interview may appear in academic publications, conference presentations, and the dissertation itself. Quotes will always be attributed to a pseudonym, never to your real name. Before any quote is published, you may review it in context and request changes or removal.

Please initial: ___ I agree that my de-identified quotes may be used in publications and presentations. ___ I agree to be quoted only if I review and approve the specific quote in writing before use. ___ I do not wish to be quoted. My interview may inform the study's themes but no direct quotes from me will be published.

For member checking — returning transcripts, summaries, or draft findings to participants for verification — the consent form should state that the material will be returned and the expected turnaround. Participants have the right to decline member checking, and they have the right to be told honestly that member checking is a credibility practice, not an invitation to rewrite the analysis. This is also the place to briefly mention researcher positionality — participants have a right to know who is interpreting their words and from what vantage point.

Withdrawal After Data Collection: The Hardest Clause

The biomedical template usually says: "You may withdraw from the study at any time without penalty." In a qualitative interview study, that sentence is partly fiction, and the fiction is what gets consent forms returned by IRBs who read carefully. A participant can always stop an interview or end future contact. What a participant cannot always do is withdraw data that has already been analyzed, coded, and woven into published findings.

Sample honest language:

You may stop the interview at any time. You may also ask the researcher to destroy your interview recording and transcript at any point up until [DATE — typically when analysis begins, or when the manuscript is submitted]. After that date, your de-identified contributions may be part of analyses that cannot be fully disentangled, and complete withdrawal will no longer be possible. You may still ask that specific quotes attributed to you not be used.

Name the cutoff. Name the reason. Give the participant a way to withdraw partial material (specific quotes) even after full withdrawal is no longer feasible. IRBs reward this specificity because it matches reality.

Sensitive Topics: Trauma, Stigma, Illegal Activity

When the interview touches trauma, stigma, or behavior that could expose the participant to legal or social consequences, the consent form must do three things generic templates miss.

First, warn specifically about the topics — not in the abstract ("you may discuss difficult topics") but with accurate naming ("the interview includes questions about your experience of domestic violence"). Second, provide real resources — the name and number of a local crisis line or counseling service, not a 1-800 placeholder. Third, when the topic involves illegal behavior (undocumented status, drug use, sex work), consider applying for a Certificate of Confidentiality and state that protection on the form.

This is where the Four Principles of Biomedical Ethics — autonomy, beneficence, non-maleficence, and justice — stop being abstract. A trauma interview that runs longer than the participant has energy for violates non-maleficence even when it respects autonomy.

Focus Groups: One Participant Cannot Guarantee Another's Confidentiality

Focus groups are a particular challenge for consent, because the researcher cannot guarantee what participants will say outside the session. You can remind the group of confidentiality norms; you cannot enforce them. The consent form must reflect this limit.

Sample language:

Because this is a group conversation, the researcher cannot guarantee that other participants will keep what is said in the session confidential. We ask all participants to respect each other's privacy, but we cannot control what they do after the session ends. If you would prefer not to share information you consider highly sensitive, you are welcome to decline to answer any question or to leave the session at any time.

Some studies add a separate "group norms" agreement each participant signs — not as a legally binding promise but as a visible commitment. Use a structured interview protocol that opens the session with an explicit review of those norms before the first question.

Research with minors requires parental permission and, typically, child assent — two separate documents. The parental permission form follows adult consent conventions and must state what information will be shared back with the parent. Assent forms are written at the child's developmental level: short sentences, concrete descriptions, an invitation to ask questions. Assent is revocable at any moment. A child who signs and then refuses to answer questions is exercising assent refusal, and the researcher must honor it regardless of parental permission.

Gatekeepers — school principals, clinic directors, community leaders — often serve as institutional permission-givers, but their permission is not a substitute for individual consent. Make sure your form and recruitment script make clear that the gatekeeper's endorsement does not obligate the individual to participate.

Remote interviewing adds consent elements in-person interviews do not require. Name the platform (Zoom, Teams, phone). Disclose whether the session uses cloud or local recording, whether auto-generated transcripts are created, and whether a business associate agreement is in place. Signing is also more complicated remotely. Options include electronic signature platforms (DocuSign, Adobe Sign, Qualtrics) that capture IP address and time stamp; emailed PDF returns of a signed form prior to the interview; and verbal consent recorded at the start with a waiver of documentation. Each option requires IRB approval of the specific procedure — do not improvise.

When you use AI tools as part of remote interview workflows — automated transcription, thematic analysis support — the consent form should disclose this too. For background, see our guide on AI for qualitative research.

An IRB can waive the requirement for a signed consent form under 45 CFR 46.117(c): when the only record linking participant to study would be the consent form itself and the principal risk is breach of confidentiality; when the research is minimal risk and involves no procedures requiring written consent outside the study; or when the participant belongs to a cultural group in which signing documents is not the norm.

When documentation is waived, the consent conversation still happens — it just ends with verbal agreement instead of a signature. The researcher reads an information sheet aloud (or provides it in writing), answers questions, and records the participant's verbal "yes." A common structure:

"Before we begin, I'm going to read a short description of the study and what participating involves. I'll then ask whether you agree to take part. If you say yes, I'll turn on the recorder and we'll start. Is that okay with you?"

The first thing captured on the recording is often: "And just to confirm on the recording — do I have your permission to begin the interview and record it?" This pattern matters especially in community-based research with populations for whom a signature is burdensome, stigmatizing, or legally dangerous. The Free Consent Form Generator includes a waiver-of-documentation variant designed around this exact case.

When oral consent is captured on the recording, that file is now both research data and consent documentation — creating a storage problem, since the same file that must be kept to satisfy consent-documentation requirements is also the file that must be destroyed to satisfy the data-minimization promise.

The cleanest solution is to split the recording: capture the consent portion as a separate file stored in the consent-documentation archive, then begin a new recording for the interview itself. When that is not feasible, extract the consent segment at transcription time, store that clip on the consent-documentation retention schedule, and destroy the interview portion on the data-destruction schedule. IRBs almost always approve this procedure when it is described clearly. They rarely approve vague language about "managing the recording appropriately."

A Working Drafting Sequence

A consent form for a qualitative interview study benefits from being written in a specific order:

  1. Draft the concise summary — three to five sentences naming the recording, the quotation, and the limits of confidentiality.
  2. Draft the procedures paragraph — what happens, how long, on what platform, who is present.
  3. Draft the risks paragraph — grounded in your specific population, not copy-pasted.
  4. Draft the confidentiality paragraph — using "confidential," not "anonymous," and naming the limits.
  5. Add separable checkboxes for recording, quotation, and member checking.
  6. Draft the withdrawal paragraph with a real cutoff date or milestone.
  7. Add contact information for the researcher and IRB.
  8. Reread the form as a participant would. Where does it feel like a contract? Where does it feel like a conversation? The goal is the latter.

Draft a Consent Form Built for Interviews

Generate a starting consent draft for your qualitative study — with language for recording, transcription, quotations, and withdrawal — ready to adapt to your IRB's template.

Try the Consent Form Generator →