How Much Identifying Information Should Researchers Collect From Participants?

How Do You Know Whether You Are Collecting Too Much Identifying Information?

A registration form asks for a participant's full name, email address, telephone number, exact age, department, job title, and location. Each field seems potentially useful. Taken together, however, they create a detailed identity profile.

Researchers often focus on protecting identifiers after collection. An earlier question may be even more useful: did the study need to collect each identifier in the first place?

The appropriate amount is not always zero. Follow-up studies, longitudinal designs, incentive distribution, clinical procedures, record linkage, withdrawal mechanisms, and other legitimate research functions may require identifying information. The task is to distinguish what the study needs from what is merely convenient to have.

Collect What the Study Needs, but No More

Data Minimisation Starts Before the First Participant Enrols

Begin With Purpose, Not With a Standard Demographic Form

The easiest way to collect unnecessary personal information is to begin with last year's questionnaire and leave every field in place.

A better approach starts with purpose. For every item that identifies or helps identify a participant, ask what function it serves in this particular study.

A name might be needed for consent documentation or participant communication. An email address might be necessary for follow-up. A telephone number might be unnecessary if all communication occurs by email. Exact date of birth might be excessive if the analysis needs only broad age categories.

Under the UK GDPR's data-minimisation principle, for example, personal data should be adequate, relevant, and limited to what is necessary for the purposes for which they are processed. ICO guidance translates this into a practical instruction: identify the minimum amount of personal data needed to fulfil the purpose and hold that much, but no more.

That is a useful design question even where UK law does not apply. The applicable legal obligations will vary by jurisdiction, but unnecessary identifiers generally create privacy and governance burdens without improving the science.

"Necessary" Does Not Mean "Might Be Useful Someday"

Research produces uncertainty, and researchers understandably want flexibility. That does not make every potentially useful variable necessary.

The ICO's research guidance makes a useful distinction: necessary processing need not be absolutely indispensable, but it should be a targeted and proportionate way of achieving the research purpose. If the same purpose can reasonably be achieved through less intrusive means, collecting additional personal data requires stronger justification.

For example, collecting participants' personal mobile numbers because researchers might decide to conduct follow-up interviews later is weaker justification than collecting them because follow-up interviews are an approved component of the protocol.

The difference is between a defined purpose and speculative convenience.

Collecting No Identifier at All May Sometimes Be Possible

If a study does not require follow-up, record linkage, verification, incentives tied to identity, or another participant-specific function, researchers should consider whether identifying information is needed at all.

ICO research guidance explicitly advises researchers first to consider whether research can be conducted without personal data and, where possible, using anonymous information.

This does not mean every study should be anonymous. Many cannot be. It means researchers should not assume that identity is a default variable.

The distinction between anonymous and confidential research becomes important here. If identity is genuinely unnecessary, designing the study so the researcher never receives it may eliminate an entire category of confidentiality risk.

Collect the Least Identifying Version That Answers the Research Question

Data minimisation is not only about whether a variable is collected. It also concerns precision.

Suppose age matters analytically. Do you need:

• date of birth;
• exact age;
• five-year age bands; or
• broad age categories?

The scientifically appropriate answer depends on the analysis. A study examining age-related developmental change may legitimately require greater precision than a study using age only to describe the sample.

The same question applies to geography. An exact home address may be essential for environmental exposure modelling but excessive for a study that needs only region of residence.

Data minimisation therefore does not mean mechanically choosing the least detailed variable. The data must remain adequate for the research purpose. The objective is the least identifying level of detail that still permits the study to do what it legitimately needs to do.

Distinguish Research Variables From Administrative Identifiers

Some identifying information is analytically important. Other information exists only to operate the study.

This distinction matters because administrative identifiers often do not need to travel with the research dataset. If email addresses are needed only to schedule interviews, there may be little reason for analysts to receive them alongside interview responses.

Researchers can therefore ask not only whether an identifier must be collected, but also whether it must be stored with the substantive research data.

Every Additional Identifier Expands the Identification Surface

An identifier does not have to be a name to increase identifiability. Exact age, occupation, geographic location, institutional role, event dates, and demographic details may narrow the possible participants.

Collecting several such variables can create distinctive combinations even when none identifies anyone alone. This is why researchers should consider direct and indirect identifiers together.

A useful question is therefore not merely, "Is this field sensitive?" It is also, "What does this field add to the possibility of identifying someone when combined with everything else I collect?"

Small Samples Need Particular Restraint

Detailed demographics may be scientifically useful, but their identifying power changes with population size.

Recording exact age, specialised occupation, department, seniority, gender, and institution may be unremarkable in a national survey involving thousands of participants. The same variables in a study of 12 employees from one organisation may describe individuals almost by name.

Researchers planning small or distinctive samples should therefore examine whether every demographic detail is necessary and whether broader categories would still answer the research question.

This does not justify altering scientifically necessary variables simply to make a dataset look safer. It means identification risk should form part of the study-design decision rather than being discovered at publication.

Collecting an Identifier Creates Obligations Beyond Secure Storage

Once identifying information enters the research system, researchers may need to consider lawful processing, transparency, access control, retention, disclosure, participant expectations, security, and eventual deletion or archival treatment under the rules applicable to their study.

Data minimisation can therefore simplify governance. Information that was never collected cannot be leaked from the research database, accidentally included in an export, unnecessarily shared with collaborators, or retained long after its purpose has ended.

That does not mean "collect nothing." Inadequate data can also compromise research validity. The UK GDPR formulation is deliberately balanced: personal data should be adequate and relevant as well as limited to what is necessary.

More Variables Can Also Create More Re-Identification Opportunities

A detailed dataset may become easier to link with information available elsewhere. Exact dates, locations, job titles, and demographic characteristics can provide matching variables for re-identification through other datasets.

Researchers planning eventual data sharing should therefore consider the downstream consequences of collection choices. It is easier to avoid collecting an unnecessary highly identifying variable than to preserve its full analytical detail later while somehow making its identifying power disappear.

There Is No Universal Maximum Number of Identifiers

Data minimisation is not a numerical quota. Collecting three unnecessary identifiers is not acceptable merely because another study collects ten, and a study requiring several identifiers is not automatically excessive.

The appropriate amount depends on purpose, proportionality, participant population, study procedures, analytical requirements, applicable law, ethics requirements, and available alternatives.

From a Convenient Form to a Minimised Data Collection Plan

The revised design does not minimise data by deleting useful research variables indiscriminately. It distinguishes analytical necessity from administrative convenience and removes information that serves neither.

Common Mistakes When Deciding What Identifying Information to Collect

Make Every Identifier Earn Its Place in the Dataset

Before finalising a questionnaire, interview form, case-report form, registration system, or data-extraction template, review every field that identifies or helps identify participants.

A useful protocol table can document each identifying variable, its purpose, level of precision, who needs access, where it is stored, and when that need ends. That small exercise tends to expose unnecessary fields remarkably quickly. Forms, like literature reviews, have a habit of accumulating material nobody remembers adding.

Audit Identifying Information Before Data Collection Begins

Frequently Asked Questions About Collecting Participant Identifiers

Collect Identifiers Because You Need Them, Not Because the Form Has Space

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.