Should Researchers Alter Details in Qualitative Quotes to Protect Identity?

Can you change a participant's words when the original details might reveal who they are?

A participant gives you an excellent quotation, but one detail makes them recognizable. Perhaps they name their hospital, describe a rare occupation, identify their municipality, mention the exact age of a child, or refer to a distinctive event.

Could you simply change the identifying detail?

Sometimes researchers modify, generalize, omit, or otherwise mask identifying information in qualitative material. But these are not trivial editorial changes. A direct quotation is presented as evidence of what a participant said, so altering substantive details can affect meaning, interpretation, transparency, and the participant's voice.

Protect identity without quietly changing the evidence

Anonymising a quotation is also an editorial decision about evidence

Start by distinguishing different kinds of modification

"Changing a quote" can describe several very different practices. Treating them as equivalent makes it difficult to judge whether the modification preserves the evidence.

Each approach has different implications. Removing a hospital's name from a quotation is not the same methodological act as changing "cardiologist" to "dermatologist."

Omission may be enough when the identifier contributes nothing to the analysis

Suppose a participant says, "When I transferred to St. Example Hospital, I finally received proper training."

If the hospital's identity is irrelevant to the finding, the researcher may be able to omit or mask the name while preserving the substantive claim about training. Qualitative researchers commonly use brackets or ellipses according to relevant editorial conventions to indicate omissions, clarifications, or substitutions.

UK Data Service guidance recommends replacing identifying information in qualitative text with pseudonyms or broader categories where appropriate and maintaining a clear anonymisation strategy. The important point is to alter only what is necessary and to do so systematically.

Generalization can reduce disclosure while preserving truthful context

Some details are analytically useful but unnecessarily precise. An exact village may be represented by a broader geographic area. An exact age may sometimes become an age range. A highly specific organizational unit may become a broader department or sector.

This approach retains a truthful category while reducing precision. It is often preferable to inventing a different factual detail when the broader category still supports the analysis.

Generalization does have a cost. The more information is generalized, the less contextual precision remains. Researchers should therefore decide which level of specificity is needed to support interpretation.

Substituting a false detail can create a methodological problem

Suppose a participant says, "As the only neurosurgeon in the province, I could never take leave."

Changing "neurosurgeon" to "cardiologist" may make identification harder, but it also creates a statement the participant did not make. If specialty is relevant to workload, scarcity, professional status, or the interpretation of the finding, the substitution may materially alter the evidence.

Some qualitative researchers use disguise strategies that deliberately change identifying details while attempting to preserve the underlying meaning. This approach can sometimes be defensible, but it requires careful judgment and transparency because disguise itself can introduce distortion.

Researchers should therefore distinguish protecting identity from manufacturing convenient facts. A quotation should not become more publishable by quietly becoming less true.

Meaning can depend on precisely the detail you want to remove

Identifiers and analytical content are not always separable.

A participant's gender may be central to an account of discrimination. Their occupation may explain exposure to a particular risk. Their location may be essential to understanding access to services. Their relationship to another person may explain a power imbalance.

Removing or changing those characteristics can make the quotation safer while simultaneously making the analysis weaker or even misleading.

This is why the problem of what to do when removing identifying details changes the meaning of qualitative data cannot be solved by a mechanical anonymisation rule.

Paraphrase is different from an altered direct quotation

If protecting identity requires substantial rewriting, researchers should consider whether the material should still be presented inside quotation marks.

A paraphrase allows the researcher to communicate the participant's substantive point without implying that the displayed wording is verbatim. This can sometimes provide more methodological clarity than extensively modifying a quotation until only its grammatical skeleton remains original.

Paraphrase is not automatically risk-free. A distinctive event or combination of contextual details can remain identifying even after the wording changes. But it provides greater freedom to summarize the evidence without falsely presenting reconstructed language as the participant's exact words.

Choosing another quotation may be the cleaner solution

Researchers often have several excerpts supporting the same analytical theme. If one is highly identifying and another conveys substantially the same finding with lower disclosure risk, the second quotation may be preferable.

This can be methodologically cleaner than heavily disguising the more memorable quotation. The most vivid excerpt is not automatically the most responsible one to publish.

Anonymisation should be planned across the dataset

UK Data Service guidance recommends planning anonymisation early, identifying direct and indirect identifiers, applying replacements consistently, and keeping unanonymised source data separate and secure rather than overwriting them.

This is important because ad hoc changes made only during manuscript writing can create inconsistencies. The same hospital may become "a hospital" in one quotation, "a regional medical center" in another, and remain fully named in a third.

A documented anonymisation scheme helps researchers apply comparable treatment to comparable information while preserving the original research record under appropriate access controls.

Do not overwrite the original data merely to create a publication-safe version

The research record and the dissemination version serve different purposes. A quotation used in a manuscript may need identifying details removed or generalized, while the original transcript may need to preserve the participant's actual words for analysis, audit, or other approved purposes.

UK Data Service guidance specifically recommends retaining original source data securely and creating anonymised versions rather than irreversibly replacing the source material.

This approach also allows researchers to revisit whether a modification has inadvertently changed the meaning.

Changes should not conceal analytical contradictions

Anonymisation should protect participants, not tidy the data.

Changing wording because it is awkward, contradictory, politically inconvenient, or inconsistent with the researcher's interpretation is not a confidentiality technique. Researchers should separate legitimate identity protection from ordinary quotation editing and from alterations that would change the evidentiary record.

Participants may sometimes help assess disclosure risk

In some qualitative designs, researchers use participant-centered approaches to confidentiality in which participants are consulted about quotations or potentially identifying information. Kaiser describes such an approach as one possible response to the limitations of assuming that researchers alone can predict what participants consider sensitive.

This is not a universal requirement, and returning quotations to participants can introduce practical and methodological complications of its own. Participants may withdraw useful material, request substantive rewriting, become unreachable, or have different preferences over time.

Researchers should therefore use participant review when it fits the methodology and approved ethics arrangements rather than treating it as an automatic solution.

One identifying quotation, four possible responses

There is no automatic winner among these strategies. The appropriate choice depends on what the detail contributes to the analysis, how identifying it is, what confidentiality was promised, and how much modification the quotation can tolerate before it ceases to represent the original evidence faithfully.

Common mistakes when disguising identifying details in quotations

Change the minimum necessary and protect the meaning first

When a quotation creates identification risk, begin with the least intrusive response that adequately protects the participant. The objective is not to make every quotation generic. It is to reduce disclosure without quietly changing what the evidence means.

Above all, be consistent. Similar identifiers should receive similar treatment unless there is a defensible reason for different handling. A documented anonymisation strategy is much easier to defend than dozens of improvisations made at 2:00 a.m. while revising Reviewer 2's comments.

Before changing a detail inside a qualitative quotation

Questions about changing identifying information in participant quotations

Protect confidentiality without rewriting what the participant meant

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.