The Voucher, the Principal, and the Free-Text Box

Opt-out consent can be defensible for low-risk anonymous surveys, but this protocol is not low-risk: it asks about self-harm, involves minors, and is conducted in a school where participation may feel expected. The $50 voucher is substantial for a low-decile community and may act as an undue inducement, especially if it is contingent on completion rather than participation or if it pressures caregivers who are struggling. The principal's enthusiasm and the classroom setting add power pressure: students may not feel able to decline. Before collection, the researcher should secure active caregiver consent or a carefully justified opt-out with clear, accessible information and genuine student assent; make the voucher proportionate and not contingent on answering every item; separate participation from withdrawal; and ensure a private way to decline. The self-harm items require an explicit protocol for responding to distress, including a named contact and follow-up, because the school counselling team is understaffed. The principal should not receive class-level results; small classes make re-identification likely, and results could stigmatise students or teachers. Aggregate school-level reporting with suppression thresholds and whānau advisory group input would be more defensible.

The researcher has a duty of care that cannot be discharged by deleting the response. Although confidentiality is important for valid research, it is not absolute: participants should have been told in advance that disclosures indicating serious risk may be acted on. The obligation is to follow the pre-specified distress protocol, which should include immediate review by a clinician, contact with the student at a private and safe time, risk assessment, and a warm handover to an appropriate service. Because the school counselling team is understaffed, the protocol must identify a backup pathway, not simply 'refer to the counsellor'. The researcher should not promise secrecy, should not confront the student in class, and should document decisions and rationale. If there is imminent risk, emergency procedures override ordinary confidentiality. The pilot also reveals a design problem: free-text boxes invite disclosures that the team may not be resourced to handle. The protocol should be revised before the main study, with the ethics committee notified, and the incident should inform consent materials and staff training.

Identifiable data on a university server can be appropriate if access is role-based, encrypted, and minimised, but the researcher should store identifiers separately from survey responses and set a destruction date. Sharing de-identified data overseas is not automatically safe: in a small school and a low-decile community, combinations of ethnicity, year level, and sleep or anxiety items may permit re-identification, and data about Māori students engage Māori data sovereignty expectations. The whānau advisory group should have been consulted before design, not after; their absence is an ethical and cultural safety failing, not a paperwork issue. The researcher should pause sharing, consult the group and the school, establish a data-sharing agreement that specifies purpose, access, retention, and benefit, and consider whether overseas transfer is necessary. If the collaborator only needs aggregate estimates, share summary outputs rather than unit-record data. The ethics application should reflect these arrangements, and the school should receive a plain-language summary that does not identify classes or individuals.