Reference Chapter 87 · Generating evidence in schools without turning children into an unprotected research population
A Ministry of Education wants to know whether a new mathematics programme improves understanding. A university proposes interviewing students about bullying. A software provider offers free access to its learning platform in exchange for pupil activity data. A school introduces a new timetable and plans to publish the outcomes as research. Each idea may create useful knowledge, but it also raises a question more important than a completed report: what authority permits the activity, what happens to the children involved and who is responsible if the activity causes harm?
The gold standard is a national school-based research ethics and education-trials system that protects participants, ensures meaningful consent or other lawful authorisation, maintains educational continuity, preserves independent review and requires honest interpretation of evidence. Research should improve teaching and public decisions without making children pay the hidden cost of experimentation.
This is a Ministry of Education reference for university studies in schools, ministerial policy trials, school improvement evaluations, surveys, classroom observations, digital experiments, student interviews and the use of routine education data. It complements Education Research, Evidence Synthesis and Policy Learning, which concerns learning from evidence. This chapter concentrates on the ethical permission and operational protections needed before evidence is produced.
1. Define the difference between teaching, evaluation and research
Schools routinely examine student work to improve teaching. Ministries evaluate funded programmes, while researchers may conduct studies intended to produce knowledge that extends beyond one institution. These activities can overlap but are not interchangeable.
A National Education Evidence Activity Register should classify: – Ordinary teaching and formative assessment, carried out within legitimate education duties. – Internal quality assurance and school improvement. – Programme evaluation commissioned by a public authority. – Formal research involving pupils, staff or caregivers. – Experiments or trials introducing a new curriculum, tool or treatment. – Surveys, interviews and observation outside ordinary teaching. – Secondary analysis of identifiable or sensitive school records. – Digital analytics, AI model development or vendor product tests. – Publication and external sharing of research results.
Classification helps determine which law, ethics review, consent, safeguarding and data protection requirements apply. Labelling a project “routine improvement” should not provide a way to bypass essential protections where research approval is required.
2. Place children’s rights and welfare before project convenience
The UNICEF Innocenti compendium, Ethical Research Involving Children (2013), stresses children’s dignity, rights and well-being across differing research contexts. The associated ERIC ethical guidance addresses harms and benefits, privacy, consent and compensation as ongoing questions, not merely forms to sign.
A Child-Centred Evidence Charter should require a justified educational or public purpose, proportionate risks, competent oversight, appropriate participation and credible plans for responding to unexpected harm.
The researcher’s career, a university’s publication target or a vendor’s commercial interest is not sufficient justification for imposing unnecessary burdens on learners.
The ministry must also consider the effect on children who do not participate, including possible loss of instruction, unequal access to new resources or disruption caused by trial allocations.
3. Establish a national school research permission gate
Schools should know which projects they may approve locally and which require ministry, institutional ethics committee, regulator or other competent authorisation.
A School Research Approval Record should contain:
- Research question and claimed public benefit.
- Principal investigator and accountable institution.
- Participating schools and intended populations.
- Activity type and reason for involving children.
- Ethics approval or lawful determination, where required.
- School and ministry permission within their actual authority.
- Consent, assent and withdrawal arrangements.
- Safeguarding and distress-response plan.
- Data sources, privacy, retention and disclosure.
- Instructional time and resource burden.
- Study design and analysis plan.
- Conflicts, funding and sponsor influence.
- Conditions for pausing or stopping the study.
- Results reporting and participant communication.
- Record of approved modifications.
School access permission is not automatically a substitute for independent research ethics approval. Conversely, approval from an outside university does not confer a right to interrupt school operations without proper educational authorisation.
4. Require meaningful, age-appropriate consent and assent
Children may not have the same legal capacity to consent as adults, depending on age and jurisdiction. Where parental or guardian permission is required, the child’s own understanding and voluntary agreement can still be ethically essential.
The UNICEF Policy on Ethics in Evidence Activities Involving People, issued in 2026, emphasises ethics review, meaningful consent or assent, confidentiality, safeguarding and risk management across UNICEF evidence activities. Its rules apply to UNICEF’s relevant work; national ministries must use their own lawful requirements.
An Age-Appropriate Participation Explanation should state what will happen, what information will be collected, foreseeable discomfort, potential benefits, whether participation is optional, who can see results and how participants may ask questions or withdraw where applicable.
A signature does not prove that a child understood. Schools should avoid obtaining assent through pressure from the classroom teacher, compulsory grades or an implied promise of better educational treatment.
5. Separate voluntary research from compulsory education
Pupils may feel unable to decline a research activity when the request comes from their principal or teacher. A student should not need to risk their grade, access to ordinary lessons or relationship with school staff to refuse an optional study.
A Research Participation Independence Check should examine recruitment language, the role of authority figures, alternative learning activities and what happens to nonparticipants.
A school may legitimately introduce an authorised educational reform to all enrolled pupils without asking each child to volunteer for ordinary teaching. That implementation decision is different from seeking consent to collect extra private research data or participate in optional interviews.
The ministry should have competent officers determine the applicable boundary and explain it honestly. Do not claim every school evaluation always requires the same research consent arrangement, or that ordinary enrolment implies unlimited consent for commercial experimentation.
6. Review both expected benefit and foreseeable harm
Research may generate valuable knowledge while exposing participants to privacy risk, distress, stigma, exclusion or loss of teaching time.
A Study Risk and Benefit Matrix should distinguish educational disruption, sensitive questions, identity exposure, peer consequences, digital risks, unequal access to programme benefits and the possibility that findings will be misinterpreted.
The UNICEF Innocenti 2025 paper, Researching Sensitive Topics Involving Children, emphasises contextual understanding, trained researchers and actionable responses when studies may evoke distress or sensitive disclosures.
A project asking about serious violence, self-harm or abuse requires more than a generic school survey template. It may need specialist expertise, a lawful reporting plan and clear protocols that participants can understand before the activity begins.
7. Keep safeguarding investigation separate from research
A child may disclose harm during an interview, survey or focus group. The researcher must know the limits of confidentiality and the competent route for responding.
A Sensitive Disclosure Response Protocol should identify who receives reports, which disclosures trigger mandatory referral, immediate protective steps, secure recording and how the researcher avoids inappropriate questioning.
Researchers must not promise absolute confidentiality where applicable law requires reporting. They should also avoid collecting more sensitive narrative detail than necessary for protection or the justified study.
Do not treat an unverified allegation as a final factual finding, and do not replace a specialist investigation with a research interview. Competent statutory agencies and authorised safeguarding officers must perform their respective functions.
This interface is covered in Student Safeguarding, Well-being and Protection.
8. Protect the learning opportunities of trial participants
A study may compare two curricula, intervention durations, feedback strategies or digital learning modes. A valid design still needs to protect statutory and educational entitlements.
A Trial Learning-Floor Review should establish the ordinary instruction pupils must receive, the intervention’s anticipated effect, available alternative support and the conditions under which the trial must be modified or stopped.
Where random assignment is proposed, the ethics review should examine whether the design is justified, whether a credible benefit or harm is known, and which supports remain available to each group.
A ministry should not use an experiment as a reason to withhold established essential teaching indefinitely. Nor should researchers claim benefit simply because pupils used an appealing new platform.
Protecting valid public education and generating credible evidence must be considered together.
9. Minimise sensitive data and specify future use
Researchers may ask to link test scores, attendance, household income, disability-related supports or student interviews. Linking data can create powerful insights but also increase privacy risks.
A School Research Data Management Plan should state exact variables, purpose, legal basis, authorised access, identifiability, storage, encryption where appropriate, retention, transfer, publication and secure disposal.
Pseudonymisation and anonymisation are not the same. Small school groups can be identifiable through combinations of grades, roles or events even after names are removed.
A vendor should not gain unlimited rights to train future commercial models from children’s work simply because it offered a school a no-cost pilot. Secondary uses need their own legitimate legal and ethical basis.
See Education Data Governance, Privacy and Interoperability.
10. Establish independent review of high-consequence studies
Some school research can be evaluated through approved expedited or proportionate processes, while sensitive or high-risk studies require more detailed independent review.
A School Research Ethics Panel should include suitable expertise in child protection, educational methods, subject teaching, privacy, law and research ethics according to the project. Reviewers need independence from the study’s material interests.
The panel should document approval, modifications, conditions, continuing review and decisions after a serious adverse event. Membership and procedures depend on national institutions and laws.
An institution should not treat an ethics committee as a rubber stamp. Review quality is shown when it identifies risks, improves consent language, rejects inadequate designs and imposes workable protections.
Researchers also need an accessible route to seek advice before submitting a complex protocol.
11. Manage commercial research and conflicts of interest
An education technology company may offer funding, devices or free software while seeking study data or endorsement. Its commercial objectives can conflict with academic independence and pupil privacy.
A Sponsored Research Integrity Agreement should disclose the sponsor, research design authority, data rights, potential benefits, staff incentives, publication freedom, adverse finding treatment and conditions for terminating the partnership.
A sponsor should not be permitted to suppress inconvenient results merely because they affect product sales. Nor should school leaders receive undisclosed personal rewards for enrolling pupils into a programme.
Students and caregivers should know when a product trial is sponsored and whether the vendor will process their information.
The procurement standards are in Education Procurement, Contracts and Supplier Accountability.
12. Require a research plan that can answer its question
Unethical research is not limited to direct harm. Poorly designed studies can waste children’s time and create misleading conclusions that influence future policy.
A Research Design Integrity Record should clarify the target outcome, starting conditions, comparison logic, selection processes, sample and practical constraints, analysis plan and likely uncertainty.
A mathematics programme trial should specify whether it seeks better arithmetic fluency, algebraic reasoning, retention or examination performance. These are not identical outcomes.
A study comparing volunteers with nonvolunteers cannot automatically infer that its programme caused the difference. Researchers should distinguish descriptive associations from justified causal claims.
Methodological honesty is part of respecting participants: their effort should contribute to the possibility of trustworthy knowledge.
13. Prepare schools and teachers for research participation
Schools need to know who will organise interviews, when lessons will be interrupted, how parents are contacted and what teachers must do if pupils withdraw or become distressed.
A School Research Operations Agreement should state scheduling, resource needs, school contacts, safeguarding roles, appropriate teacher involvement, materials, consent logistics, confidentiality and conditions for school withdrawal.
Teachers should not be pressured to recruit students merely to meet a researcher’s sample target. Their professional assessment and workload must be respected.
The school should preserve a named contact for families who have questions after a research team leaves. A study’s obligation to participants should not disappear when fieldwork ends.
14. Protect fair treatment of learners with access needs
Children with disabilities, differing instructional languages or unstable attendance may be routinely excluded from studies because their participation requires adaptation. The resulting research may then fail to represent the learners affected by the proposed policy.
An Inclusive Study Design Review should examine accessible consent, communication, sampling, appropriate accommodations and the reasons for any exclusion.
Excluding a group may sometimes be justified by the study question or risk, but it should be deliberate and explained rather than an administrative default.
Do not force a child to disclose sensitive disability or household information simply to participate in an ordinary classroom observation study.
The ministry should also evaluate whether the new intervention is workable in remote schools and other contexts beyond the pilot’s original sample.
15. Include a withdrawal and complaint route that genuinely works
A participant should know whom to contact with questions, distress, privacy concerns or a wish to stop participating, within the actual legal and research conditions.
A Participant Concern and Withdrawal Receipt should identify the request, relevant consent conditions, responsible reviewer and what happens to data already collected under applicable rules.
Withdrawal may not always require deletion of all earlier research data, depending on law and the study’s consent terms; these limitations should be explained truthfully before participation.
An independent complaint channel matters where a pupil fears that a teacher or researcher might react badly to refusal.
Serious breaches should have a procedure for investigation, corrective action and appropriate participant notification.
16. Preserve academic freedom while preventing misconduct
Independent education research can reveal that a national policy is ineffective. Its findings should not be altered because the ministry commissioned the study or dislikes its implications.
A Research Publication and Integrity Clause should identify legitimate confidentiality limits, ownership, peer or technical review, data availability, timing, correction and the right to report negative or inconclusive results.
Academic independence does not exempt researchers from privacy law, child protection, conflicts or methodological accountability.
Likewise, a government can reasonably require accurate presentation of methods without demanding favourable conclusions.
The balance is public trust: study participants are protected and the findings may still expose genuine institutional failures.
17. Evaluate a completed study’s service burden and outcomes
Research may be ethical at approval yet cause unexpected classroom disruption, excessive teacher workload or privacy issues during implementation.
A Post-Study Ethics and Value Review should examine approved versus actual activities, participant complaints, adverse events, educational time used, data management, responsible disclosure and whether the original question was meaningfully answered.
An intervention that failed academically may still produce valuable evidence if the study was fair and well conducted. The ministry should avoid equating negative findings with misconduct.
A study that obtained exciting results by violating participant rights is not a gold-standard example.
18. Case study: the free tutoring app that wants every child’s records
Imagine a commercial provider offering a district free AI mathematics practice for one year. In exchange, it asks for identifiable test scores, attendance, learner support categories and permission to use the data in later model development.
A weak education authority views the transaction as a cost-saving pilot. A stronger review asks why each data element is necessary, whether the product’s claim can be tested with less data, what research approval and consent apply, and whether nonparticipating students retain ordinary education.
The district limits or rejects unjustified data uses, ensures independent study methods and preserves human teaching. If the pilot proceeds, it measures independently demonstrated mathematics knowledge rather than application engagement alone.
The resulting evidence can inform public policy without treating children’s private information as the hidden price of a free product.
19. Common failure modes and corrective routes
School access mistaken for ethics permission
Failure: a principal grants classroom access without required independent review. Repair: enforce separate education, ethical and legal approval gates.
The compulsory consent illusion
Failure: pupils believe refusing research will affect grades. Repair: use genuine informed procedures and an appropriate alternative.
Privacy waived through a free product
Failure: a vendor acquires broad student data rights through promotional arrangements. Repair: minimise data and contractually limit authorised use.
Weak study design wastes learner time
Failure: an evaluation cannot support the claims made from it. Repair: require a viable method and honest limitations.
Safeguarding disclosure handled by researchers alone
Failure: unqualified investigators respond to serious harm reports. Repair: establish competent statutory and protection referral pathways.
Results withheld after an inconvenient finding
Failure: sponsors influence publication to avoid negative evidence. Repair: guarantee appropriate research integrity and reporting.
20. The first 180 days
Days 1–30: inventory research and commercial trials in schools, existing legal and ethics approval routes, and urgent privacy or safeguarding issues.
Days 31–60: examine representative projects for consent, data use, instructional disruption, review independence and analysis plans.
Days 61–90: publish a School Research Permission Gate, Child-Centred Evidence Charter and differentiated ethics-review pathway.
Days 91–120: train school research contacts and pilot the new review process, including a vendor data-use case and a sensitive interview scenario.
Days 121–180: verify participants’ ability to complain or withdraw, monitor study burdens, and publish an appropriately anonymised review of corrected practices.
21. Gold-standard acceptance tests
- Does every school-based study have a legitimate, clearly described purpose?
- Are ordinary teaching, evaluation and human-participant research appropriately distinguished?
- Can the authority determine the actual ethics and legal approval required?
- Are consent, assent and refusal meaningful and age-appropriate?
- Can children participate or decline without inappropriate academic pressure?
- Are risks, sensitive disclosures and safeguarding responses planned by competent people?
- Are personal data minimised and restricted to lawful research purposes?
- Are study methods credible enough to justify the time and burden imposed?
- Are commercial interests declared and results protected against improper suppression?
- Can the ministry demonstrate ethical participant treatment alongside truthful and useful educational evidence?
22. Reference sources and related chapters
Authoritative starting points are UNICEF Policy on Ethics in Evidence Activities Involving People (2026), UNICEF Innocenti, Ethical Research Involving Children (2013), ERIC Ethical Guidance and UNICEF Innocenti, Researching Sensitive Topics Involving Children (2025). They offer ethical principles and institutional examples; enforceable rules depend on each jurisdiction and research sponsor.
This is Chapter 87 of The Gold Standard of a Ministry of Education Reference Manual. It complements Ministry of Education V2.0 Reference 56: Research Ethics and Evidence Synthesis, Education Reform Pilots and Responsible Scaling and AI Governance and Human Review.
Final principle
The gold standard of school-based research is not the number of studies conducted or the prestige of the institution collecting data. It is a public system that asks worthwhile questions, protects learners who participate, uses sound methods, and can act responsibly on evidence without sacrificing the children whose education it exists to improve.