03 · What You Need to Know
Safety Decisions Do Not Require You to Wait for Harm to Occur
Ethics approval does not obligate you to complete every encounter
An approved protocol describes research that may proceed under anticipated conditions. It is not an instruction to continue regardless of what actually happens.
UKRI's Economic and Social Research Council states that researchers should consider risks and risk management throughout the project lifecycle. Its guidance also recognizes that ethics issues may emerge after review and requires mechanisms for reporting unforeseen events that might challenge the ethical conduct of research or provide grounds for discontinuing a study.
Researcher safety is explicitly part of that ethical landscape. ESRC guidance asks ethics applications to address risks to researchers and how researchers will be protected or supported, particularly during fieldwork. Research ethics committees are also instructed to consider researcher safety, especially in situations such as lone fieldwork.
Approval therefore does not convert a foreseeable safeguard into a professional obligation to endure danger.
You do not need proof that harm will occur
Safety decisions are prospective. If researchers could act only after proving that an assault, accident, or other harm was about to happen, risk management would arrive rather late.
Researchers should instead respond to credible indicators in context. These may include threats, escalating aggression, blocked exits, unwanted physical contact, unsafe transportation, unexpected people at a site, severe environmental hazards, civil unrest, loss of reliable communication, or other material changes from the assessed conditions.
Sometimes the concern is less concrete. A researcher may notice several contextual signals that collectively suggest the situation is deteriorating. Professional judgment matters here, particularly because field researchers often have information that supervisors and ethics committees do not have in real time.
Feeling uneasy does not establish that another person is dangerous. It can, however, be a reason to pause and reassess rather than suppress the concern simply because the interview has begun.
Discomfort and danger are not identical
Research can legitimately be uncomfortable. A participant may criticize the researcher, express anger about the topic, describe disturbing events, challenge assumptions, or become emotional. Difficult conversations are not automatically unsafe conversations.
This distinction matters. Researchers should not use “I felt unsafe” casually to avoid perspectives they dislike or to characterize participants as threatening merely because they are distressed, marginalized, politically opposed to the researcher, or difficult to interview.
Research discomfort
May involve emotional difficulty, disagreement, uncertainty, awkwardness, or challenging material without a credible threat to safety.
Safety concern
Involves circumstances that reasonably suggest continued participation could expose the researcher or others to meaningful harm.
The boundary will not always be obvious. That is why protocols should provide decision criteria and support rather than expecting researchers to distinguish the two perfectly under pressure.
You can stop before behavior becomes overtly aggressive
A researcher does not need to wait for a participant to become physically violent. The setting itself may become unsafe even when the participant has done nothing threatening.
Imagine arriving at a home interview and discovering that the meeting will take place somewhere substantially more isolated than expected. Or a researcher conducting observation notices nearby violence developing. A fieldworker loses the communication needed for the agreed lone-working procedure. A researcher realizes that the only available transportation would leave them dependent on someone they do not feel safe accompanying.
These situations differ from responding to an aggressive or threatening participant. The common principle is that researchers may respond to material changes in risk before those changes result in actual harm.
Ending the encounter should itself be done safely
Announcing “I am leaving because you are dangerous” may not be the safest way to exit an escalating interaction. Depending on the circumstances, a brief neutral explanation, a pause, or a practical reason for ending the meeting may allow the researcher to disengage with less confrontation.
There is no universal script because the safest response depends on the situation. If explaining the decision would increase danger, the researcher should not prolong the encounter merely to provide a complete explanation.
Once an immediate threat exists, preserving rapport, collecting the final answer, formally debriefing the participant, or retrieving every research item should not take priority over reaching safety.
Researchers should know in advance that they have authority to leave
A safety protocol is weakened if a researcher must telephone a supervisor for permission while the danger is unfolding. Projects involving meaningful fieldwork risks should make clear what researchers can decide independently during an immediate safety concern.
This is particularly important for research assistants, students, junior staff, local fieldworkers, interpreters, and others who may feel that abandoning an interview will be interpreted as poor performance.
The team should therefore establish in advance who can stop an activity, when authorization is unnecessary because immediate safety is involved, whom the researcher contacts afterward, and who decides whether the activity can resume.
These arrangements are part of the broader question of who is responsible for assessing and managing researcher safety.
Leaving one encounter does not automatically mean abandoning the entire study
Stopping an interview, leaving a field site, and discontinuing a research project are different decisions.
A researcher may leave one encounter because of circumstances unique to that moment. The team can then determine whether the problem was isolated, whether the same activity can proceed with modified safeguards, or whether the incident reveals a broader problem with the study design.
ESRC guidance anticipates ongoing ethics review where circumstances require it. Research ethics committees should have procedures for unforeseen events that may challenge ethical conduct, and serious concerns can ultimately justify suspension or discontinuation of research.
The proportional response might therefore range from rescheduling one interview to changing the venue, introducing paired working, modifying recruitment, suspending a field site, submitting an ethics amendment, or stopping part or all of the study.
The participant's rights do not disappear when the researcher leaves
Ending an unsafe encounter does not permit researchers to disregard confidentiality, privacy, consent, or data-protection obligations afterward.
Questions may arise about recordings or data already collected. Whether those data can be retained depends on the approved consent arrangements, any withdrawal request, the stage of processing, applicable data-protection requirements, and institutional policy. The safety incident itself does not automatically answer the data-retention question.
Incident reports should also be factual and appropriately protected. A researcher can document threatening or concerning behavior without unnecessarily circulating identifiable participant information.
Stopping may protect participants as well as researchers
Continuing an encounter despite escalating risk can impair judgment and intensify conflict. A distressed or frightened researcher may become less able to maintain boundaries, respond appropriately, protect confidentiality, or make sound decisions.
Stopping can therefore be ethically protective rather than adversarial. Sometimes the safest decision for the researcher is also the decision least likely to allow a difficult interaction to deteriorate further.
Watch Out
Do not design a safety protocol that effectively requires researchers to remain until a predefined threshold such as physical violence occurs. Risk assessment should allow proportionate action before foreseeable harm becomes actual harm.
