Research ethics review is built primarily around the biomedical, hypothesis-testing model: a fixed protocol, a defined intervention, and consent obtained once before data collection begins. Qualitative research routinely does not work that way, and applying the generic model without adjustment leaves real ethical gaps. This guide covers the ethical issues that are distinctive to qualitative methodologies — interviews, focus groups, ethnography, narrative and case-study work — rather than restating the general framework already covered in CASRAI’s guides to the Belmont Report and IRB protocol writing.
The issues below matter to researchers twice over: once during data collection, and again when writing the manuscript, since qualitative reporting standards (COREQ, SRQR) expect authors to document how consent, confidentiality, and the researcher’s role in the data were actually handled. See CASRAI’s guide to writing the methodology section of a qualitative paper for the reporting side of this; this guide focuses on the ethical reasoning that has to happen before you can write that section honestly.
Why qualitative studies need a different ethical lens
Standard human-subjects frameworks — the U.S. Common Rule (45 CFR 46), built on the Belmont Report’s three principles of respect for persons, beneficence, and justice — were designed around studies with a specified procedure and a predictable set of risks, known at the time consent is obtained. Qualitative research is frequently emergent: the interview guide evolves, the researcher follows unanticipated leads, and what a participant discloses partway through a study can be more sensitive, or more identifying, than anything anticipated at the design stage. Canada’s Tri-Council Policy Statement (TCPS 2), which devotes a full chapter specifically to qualitative research, states this directly — qualitative inquiry is often iterative, and ethical review needs to accommodate a research plan that develops as the study proceeds, rather than assuming every question and every risk can be fixed in advance.
This does not mean qualitative studies get less ethical scrutiny. It means the scrutiny has to be exercised differently: through the researcher’s ongoing judgment during data collection, not only through what was approved on a protocol document months before fieldwork began.
Informed consent in emergent, iterative designs
A single signature on a form at intake assumes the researcher can describe, upfront, what will actually happen to the participant and to their data. In an emergent design, that assumption often does not hold: a life-history interview can move into territory neither party anticipated, a focus group can surface disclosures about other named individuals, and an ethnographic fieldwork period can extend, narrow, or redirect its focus based on what the researcher observes.
The methods literature’s answer to this is usually described as process consent (sometimes “consent as an ongoing negotiation” or “continuous consent”): treating informed consent not as a single event but as something revisited at meaningful points across the study — when the focus shifts, when a follow-up interview is proposed, when a participant’s own account raises questions about what they’re comfortable including. In longitudinal qualitative work in particular, checking in about continued willingness to participate, and about how prior disclosures should be used, is now widely treated as better practice than relying solely on the initial consent form.
Practical implications for a protocol or consent process:
- Describe the emergent nature of the design honestly at intake. Rather than overstating precision the study doesn’t have, consent materials can explain the general topic and format, and that specific questions may evolve as the study progresses.
- Build in checkpoints for re-consent where the scope has genuinely changed — a new interview wave, a shift in research questions, or a change in how data will be shared or published.
- Separate consent to participate from consent to specific uses of the data (e.g., direct quotation, use in teaching materials, deposit in a qualitative data archive) — these are ethically distinct and often need to be asked about separately, since a participant may be comfortable with one and not the other.
- Document, in the protocol, how mid-study judgment calls will be handled — most IRBs and research ethics boards expect a description of the process for handling unanticipated issues, not a claim that none will arise.
None of this substitutes for your own IRB or research ethics board’s current requirements — confirm process-consent and re-consent expectations with your board before finalizing a protocol; practice varies by institution and jurisdiction.
Confidentiality with small or identifiable samples
Confidentiality in most quantitative human-subjects research assumes a large enough sample that removing direct identifiers (name, date of birth, address) is sufficient to prevent identification. Qualitative studies routinely violate that assumption: sample sizes are often small by design, participants are frequently drawn from a specific organization, profession, or community, and the data itself — a detailed narrative, a description of a specific role or event — can be identifying even with names removed.
TCPS 2’s qualitative-research chapter distinguishes two related but different risks, a distinction worth carrying into any qualitative protocol:
- External confidentiality — whether someone outside the study (a reader of the published output, a member of the public) can identify a participant from what’s published.
- Internal confidentiality — whether other participants or people within the same study, organization, or community can identify each other from what’s published, even when external readers cannot. A colleague quoted anonymously but describing a specific, recent, and unusual event at a small organization may still be immediately identifiable to co-workers who were also interviewed, even though a stranger reading the paper never could.
Internal confidentiality is easy to miss precisely because a de-identification checklist built for external readers looks complete while it is being applied. Mitigations worth documenting in a protocol include: aggregating or generalizing identifying details (job title, tenure, specific dates) rather than removing them piecemeal; being explicit with participants, at consent, about the limits of confidentiality in a small or closed sample; and, where the risk is high enough, considering whether individual attribution should be avoided altogether in favor of composite or thematic reporting. See CASRAI’s dictionary entries on informed consent and de-identification for the general mechanics these build on.
Anonymizing narrative and verbatim quotes
Qualitative findings are often reported as direct quotations, because the participant’s own words — not a paraphrase or a coded category — are the evidence. That creates a genuine tension: the more a quote is altered to protect identity, the more it risks misrepresenting what was actually said; the more faithfully it is reproduced, the more identifying detail it can retain.
Approaches researchers use, each with a real trade-off to disclose in the write-up:
- Pseudonyms for participants and for third parties they mention — straightforward, but does not address identifying content within the quote itself.
- Redaction or generalization of identifying details within a quote (a specific job title, location, or date replaced with a more general description) — reduces identifiability but changes the text from a strict verbatim transcript to an edited one, which should be flagged to readers.
- Composite quotes or scenarios, built from more than one participant’s account and explicitly labeled as such — used where individual attribution carries meaningful risk, but a method that has to be disclosed transparently rather than presented as a single person’s verbatim words.
- Member checking or participant review of quotes before publication — lets the participant flag anything they did not realize was identifying, though it adds a step to the timeline and is not always feasible.
Whichever approach is used, the methods section is where this needs to be made explicit: readers and reviewers cannot evaluate a quote’s evidentiary weight without knowing whether it is verbatim, lightly edited, or composite. Reporting frameworks for qualitative work — the Standards for Reporting Qualitative Research (SRQR) and the Consolidated Criteria for Reporting Qualitative Research (COREQ) — both expect this kind of methodological transparency; see CASRAI’s guide to the qualitative methods section for how to write it up.
Researcher positionality and power dynamics
Quantitative instruments are designed to minimize the effect of who administers them. Qualitative methods — especially the interview — do not have that insulation: the researcher’s identity, institutional position, and relationship (or lack of one) to the participant’s community actively shape what is said and how. This is usually discussed under two related terms:
- Positionality — the researcher’s own social location (discipline, institutional affiliation, and characteristics relevant to the study population) and how it may shape access, rapport, interpretation, and the questions asked.
- Reflexivity — the ongoing practice of examining and documenting how that positionality is influencing the research, rather than treating the researcher as a neutral instrument.
Power asymmetries show up concretely in qualitative fieldwork in ways a consent form alone doesn’t resolve: a researcher interviewing people they supervise, treat clinically, or otherwise hold institutional authority over; a researcher from a well-resourced institution interviewing participants from a marginalized or under-resourced community; or an interviewer whose presence changes what an interviewee is willing to disclose about a shared workplace or program. Ethical practice does not eliminate these dynamics — it requires naming them, considering how they affect voluntariness of participation and honesty of disclosure, and building safeguards (for example, using a third party to recruit participants who report to the researcher, or being explicit at consent about how the researcher’s role could affect what is shared).
This is also, again, a reporting obligation as much as a fieldwork one: SRQR explicitly calls for authors to describe the researcher’s characteristics and relationship with participants, and many qualitative journals now expect a positionality statement in the methods section. A protocol or manuscript that never names the researcher’s relationship to the data is missing information a reader needs to evaluate the findings, not just an ethical nicety.
Putting it into a protocol and a manuscript
In practice, these four issues are addressed together, not separately: a protocol’s consent process should anticipate emergent design and internal-confidentiality risk before fieldwork starts, and the eventual manuscript’s methods section should report, transparently, how consent, confidentiality, quote handling, and positionality were actually managed — not just what was originally planned. CASRAI’s guides to what a research protocol is and writing one for IRB submission cover the drafting mechanics this feeds into; the qualitative research and qualitative vs. quantitative pages give the broader methodological context.
Frequently asked questions
What is process consent?
Process consent (also called continuous or ongoing consent) treats informed consent as something revisited at meaningful points across a study, rather than a single form signed at intake — particularly important in emergent qualitative designs where the scope or focus can shift after data collection begins.
What is the difference between internal and external confidentiality?
External confidentiality is whether someone outside the study can identify a participant from published material; internal confidentiality is whether other participants, or people within the same organization or community, can identify each other — a risk that survives standard de-identification and is specific to small or closed-sample qualitative studies.
Can I edit a participant’s quote to protect their identity?
Yes, but the edit and its rationale should be disclosed in the manuscript — whether the quote is verbatim, lightly redacted for identifying detail, or a labeled composite changes its evidentiary status, and reporting frameworks like SRQR and COREQ expect that distinction to be made explicit rather than left implicit.
What is a positionality statement, and is it required?
A positionality statement is a description, usually in the methods section, of the researcher’s relevant social location and relationship to the study population and how it may have shaped the research. It is not universally mandated, but SRQR calls for describing researcher characteristics and relationships with participants, and an increasing number of qualitative journals expect it.
Does a qualitative study need re-consent if the research questions change mid-study?
Where the change is substantive — a materially different focus, new categories of data collected, or a new intended use of the data — re-consent or at minimum a documented re-briefing is generally expected. Confirm the specific threshold with your IRB or research ethics board, since practice varies by institution.
