Ethical Dilemmas in Data Collection
Ethical data collection is about more than just reading a consent form. Enumerators often face real time dilemmas. What do you do when a respondent's safety is at risk? When you witness something concerning? When a respondent asks for help you can't provide? What ethical challenges have you faced in the field and how did you handle them?
Ethical data collection goes beyond obtaining informed consent; it requires researchers to continuously protect participants’ dignity, safety, confidentiality, and wellbeing throughout the research process. In my field experience, I have encountered situations where participants disclosed sensitive health, social, or personal circumstances that were beyond the immediate scope of the study.
One important ethical challenge occurred when a participant became visibly distressed while discussing experiences related to illness and delays in accessing healthcare. Rather than continuing with the interview simply to complete the questionnaire, I paused the interview, assessed whether the participant wished to continue, and reminded them that participation was voluntary and that they could decline to answer any question or withdraw without consequences. I maintained privacy, avoided probing unnecessarily into sensitive issues, and followed the study’s referral and escalation procedures where additional support was warranted.
I have also encountered situations where respondents asked for medical advice or assistance that the research team could not directly provide. I believe it is unethical to create expectations that participation in research guarantees treatment or other services. I therefore explained the limits of my role honestly while, where appropriate and permitted by the study protocol, directing the participant to available health or social-support services. Where a situation presented an immediate safety concern, the priority was to follow the established safeguarding and referral pathway rather than treating the issue as ordinary research data.
Another recurring challenge is confidentiality, particularly when interviews involve sensitive information. I ensure that interviews are conducted in as private an environment as possible, avoid discussing participants’ information in public settings, use study identification codes rather than unnecessary personal identifiers, and follow approved procedures for data storage, access, and reporting.
These experiences have reinforced my view that ethical research requires judgement in real time. The researcher must balance the scientific objectives of the study with the fundamental principle of “do no harm.” When faced with an ethical dilemma, I consider participant safety first, consult the study protocol and supervisory or ethics structures, document the incident appropriately, and escalate serious concerns through the designated channels. For me, a successful field interview is not simply one that produces complete data; it is one in which the participant remains respected, safe, informed, and free to make their own decisions throughout the research process.
@Rhoda Nakhosi When a Respondent Asks for Unreachable Help: Be honest about your role from the start, clarify during informed consent that you are an information collector, not a service provider. Never make false promises of direct aid or financial support to secure an interview. Instead, carry a verified, up-to-date referral sheet listing local social services, emergency helplines, or community support resources to hand to the respondent.
@Rhoda Nakhosi Establish a Verification Schedule: Assign a dedicated team member to perform simple phone checks or field visits on a bi-weekly or monthly basis to verify operational status, services offered, and operating hours.
Thank you, Rhoda and Nonvicks, for raising what I consider one of the most important—and often underestimated—dimensions of ethical field research.
I strongly agree that a referral sheet is only useful if the information is accurate, context-specific, accessible and operationally linked to a functioning referral pathway. Simply handing a respondent a list of telephone numbers or facilities can become tokenistic if we have not established whether those services are currently operational, affordable, geographically accessible, confidential, and appropriate to the participant's specific need.
From a research governance perspective, I would approach this as a referral and safeguarding system rather than merely a referral sheet.
First, the research team should establish a verified referral directory before fieldwork begins, ideally mapped according to the study population and anticipated risks. This could include health facilities, psychosocial support, GBV services, child-protection mechanisms where relevant, emergency services, social protection programmes, community-based services and other appropriate providers. Each referral point should have a responsible focal person or mechanism for verification.
Second, the directory should have a defined verification and version-control process. In rapidly changing or remote settings, I would recommend assigning responsibility to a field supervisor or safeguarding focal person to periodically verify key information—telephone contacts, operating hours, service availability, eligibility requirements, costs and referral procedures. The date of last verification should be recorded. A simple version number can also help ensure that enumerators are not using outdated information.
Third, enumerators need more than a referral list; they require competency-based training and decision support. They should understand the difference between routine referral, urgent referral and emergency escalation. They should also know when they are permitted to act independently and when the situation must immediately be escalated to a supervisor, safeguarding lead, clinical team or other designated authority.
Importantly, the referral process should preserve participant autonomy and confidentiality. An enumerator should not assume that a participant wants a referral simply because a concern has been disclosed. Where there is no immediate danger, the participant should generally be given appropriate options and allowed to make an informed decision. The researcher should also avoid collecting additional sensitive information merely because it might be useful for referral unless that information is justified by the protocol and ethically approved.
For higher-risk situations, I think the protocol should contain a pre-specified escalation algorithm. For example:
Disclosure/observation → immediate safety assessment → classify level of risk → provide appropriate information/referral → escalate according to protocol → document the incident using minimum necessary information → follow up where ethically and operationally appropriate.
This is particularly important because the ethical response can differ considerably between a participant requesting general psychosocial support and a situation involving imminent risk of serious harm.
I would also distinguish between referral completion and referral provision. Giving someone information does not necessarily mean that a successful referral occurred. Where ethically permissible and consistent with the consent process, programmes can consider documenting whether the participant accepted the referral, whether contact was successfully established, and whether barriers prevented access. However, this should be carefully separated from research outcome data to avoid coercion or unnecessary disclosure of sensitive information.
From a data governance perspective, referral information itself can become sensitive. Researchers therefore need to consider data minimisation, access controls, secure storage, de-identification, retention periods and whether referral records are part of the research dataset or safeguarding documentation. These should not automatically be merged simply because they were generated during the same encounter.
I also agree with Rhoda that the approach must avoid making participants feel dismissed. The language used by enumerators matters. Instead of simply saying “Here is a number you can call,” an enumerator can acknowledge the concern, explain honestly what the research team can and cannot provide, offer the available options, and—where appropriate—ask whether the participant would like assistance understanding how to access the service.
Ultimately, ethical fieldwork requires operational readiness, not only ethical approval. An IRB/REC-approved protocol may define the safeguards, but the field team must be able to operationalise those safeguards under real-world conditions.
For me, the strongest ethical field systems therefore have four interconnected components:
1. Prevention – anticipate foreseeable risks during protocol development and enumerator training.
2. Recognition – equip field teams to identify distress, safeguarding concerns and urgent risks.
3. Response – establish clear referral and escalation pathways with verified service providers.
4. Accountability – document, review and learn from ethical incidents without compromising participant confidentiality.
This also raises an important question for research institutions and funders: How do we move from having an “ethical referral plan” on paper to measuring whether referral and safeguarding mechanisms actually function during field implementation?
That, in my view, is where data governance, implementation science and research ethics intersect in a particularly meaningful way.