03 · What You Need to Know
Good Supervision Is Part of Research Integrity
Supervision is more than approving a thesis at the end
A supervisor’s research-integrity role begins long before a manuscript or dissertation reaches its final draft.
Students may be learning how to design studies, maintain research records, manage data, analyze evidence, work with participants, collaborate, assign authorship, report limitations, and respond when something goes wrong. They cannot reasonably be expected to infer every professional norm through osmosis.
The National Academies identifies neglectful or exploitative research supervision as a detrimental research practice. The European Code of Conduct for Research Integrity likewise emphasizes responsibilities for training, supervision, and mentoring within research environments.
Good supervision therefore includes helping students understand not merely how to finish the project, but how to conduct it responsibly.
Supervisors should establish expectations early
Many integrity problems begin as unspoken assumptions.
Who owns and stores the research data? What records must be retained? Can files be kept on personal devices? Who can access identifiable participant information? How should deviations from the protocol be documented? Who checks an analysis? When should the supervisor be consulted? How will authorship be decided?
Leaving these questions until a dispute occurs makes everyone’s life unnecessarily interesting.
Supervisors should discuss applicable expectations early and revisit them as the project changes. Relevant requirements may come from the institution, ethics or regulatory approvals, funders, disciplinary norms, laboratories, data-management plans, journals, or collaborative agreements.
Training should match the work the student is expected to perform
A student should not simply be assigned a technically or ethically demanding research task and expected to discover the necessary standards independently.
If a student will conduct interviews involving sensitive information, use laboratory equipment, manage confidential data, write analysis code, conduct specialized statistical procedures, work with animals, or perform another unfamiliar research activity, appropriate training should precede or accompany that responsibility.
This does not mean that supervisors must personally possess every skill required by a multidisciplinary project. Responsible supervision can involve bringing in a statistician, laboratory specialist, qualitative methodologist, data steward, ethics expert, or other qualified collaborator when necessary.
Knowing when additional expertise is needed is itself part of competent supervision.
The appropriate level of oversight should change with experience and risk
Supervision should be proportionate rather than mechanically identical for every researcher.
A novice student performing an unfamiliar procedure may need close review of early work. An experienced doctoral researcher using an established method may appropriately work with greater independence. High-risk or irreversible procedures may require more oversight than routine, easily checked tasks.
| Situation |
Supervisory implication |
| Student is new to the method |
Provide closer instruction, observation, and early verification until competence is established. |
| Student demonstrates reliable competence |
Greater independence may be appropriate while maintaining agreed checkpoints and documentation. |
| Task has substantial ethical, safety, confidentiality, or validity consequences |
Use stronger controls, clear procedures, and appropriate review regardless of the student’s general experience. |
| Unexpected discrepancies or warning signs emerge |
Increase scrutiny and investigate the specific concern rather than continuing routine oversight unchanged. |
| Required expertise falls outside the supervisor’s competence |
Seek appropriate additional expertise rather than pretending supervisory status confers methodological omniscience. |
Supervisors should support research records that allow the work to be understood
The Singapore Statement calls on researchers to keep clear and accurate records that permit verification and replication of their work by others. For student research, supervisors can help establish what adequate records mean in the relevant discipline.
That might involve laboratory notebooks, interview logs, data dictionaries, code repositories, analysis scripts, field notes, consent documentation, file-naming conventions, version control, decision logs, or other records appropriate to the project.
The supervisor does not necessarily need to inspect every file continuously. But a project should not reach submission with nobody knowing where the primary data are, which dataset generated the tables, or how the final analysis was produced.
Supervisors should review enough to exercise meaningful oversight
There is no universal rule stating that a supervisor must personally verify every data point. In many projects that would be unrealistic and would defeat the purpose of training independent researchers.
Yet supervision becomes hollow if the supervisor reviews only polished conclusions without any reasonable engagement with how those conclusions were produced.
Depending on the project, meaningful oversight might involve discussing raw or intermediate results, reviewing analysis procedures, checking selected records, examining unexpected findings, asking the student to explain methodological decisions, or ensuring that appropriate quality-control procedures are functioning.
The goal is not surveillance. It is enough engagement to teach, detect foreseeable problems, and assess whether the work is progressing according to appropriate research standards.
Supervisors should make it safe to report mistakes
A research environment that treats every mistake as evidence of incompetence can inadvertently encourage concealment.
Students need to know that they should report mislabeled samples, coding errors, protocol deviations, lost files, unexpected results, and other problems promptly rather than quietly trying to repair them.
That does not mean mistakes have no consequences. Some may be serious. But early disclosure makes correction more feasible and helps distinguish honest error from misconduct.
Supervisors should therefore reward transparency about problems rather than creating incentives to hide anything that makes the project look imperfect.
Supervisors should not pressure students toward preferred results
Supervisory power creates a particular integrity risk when the supervisor has a strong investment in a hypothesis, publication, grant, or project outcome.
A supervisor should not pressure a student to remove inconvenient data without methodological justification, repeat analyses until significance appears and report only the favorable result, suppress findings merely because they are disappointing, or present exploratory decisions as though they had been planned in advance.
Students should be encouraged to report what the evidence supports, including uncertainty and results that do not confirm the original expectation.
A senior researcher’s claim that “everyone does it” is not a substitute for methodological justification.
Authorship should not be extracted as payment for supervision
Supervisors often make substantial intellectual contributions that legitimately qualify them for authorship. But supervisory status alone should not automatically settle authorship.
The National Academies recommends authorship standards based on significant intellectual contribution and identifies honorary and coercive authorship as detrimental practices. It specifically distinguishes general supervisory or administrative support from the substantial intellectual contributions that should ground authorship under its recommended model.
The applicable criteria still depend on the discipline and journal. Supervisors should discuss authorship early, revisit it as contributions evolve, and apply the relevant criteria consistently rather than using authority over the student to demand credit.
Students remain responsible for their own conduct
Strong supervision does not turn students into passive instruments of their supervisors.
Students remain responsible for conducting their work honestly, maintaining appropriate records, following applicable requirements, accurately representing results, acknowledging contributions, and raising problems when they occur.
A supervisor should therefore teach responsibility rather than creating a system in which the student assumes, "If my supervisor approved it, anything I did must be acceptable."
Student responsibility
The student remains accountable for their own research conduct, records, representations, decisions, and compliance with responsibilities appropriate to their role.
Supervisor responsibility
The supervisor is responsible for the training, oversight, research environment, decisions, and responses reasonably required by the supervisory role.
A student’s misconduct does not automatically establish misconduct by the supervisor
A supervisor can provide reasonable training, appropriate procedures, regular oversight, and accessible support and still be deceived by a student who deliberately fabricates research records.
The student’s misconduct should therefore not automatically be imputed to the supervisor.
Conversely, the supervisor’s title should not provide immunity if evidence shows direct participation, knowledge, reckless disregard, concealment, or another relevant breach. The question of whether someone can be responsible when another team member commits misconduct must be answered from each person’s own conduct and obligations.
Warning signs should change the level of supervision
Routine oversight may be adequate while a project proceeds normally. Specific warning signs should trigger closer attention.
Examples might include repeated inability to produce underlying records, results that cannot be reproduced from the stored data, unexplained inconsistencies between notebooks and datasets, recurring image problems, implausibly rapid data collection, undocumented changes to analyses, or conflicting accounts of how results were obtained.
None of these automatically proves misconduct. They do justify questions.
A supervisor who encounters credible warning signs should seek clarification, preserve relevant records where appropriate, and escalate the matter through institutional processes when the concern cannot be satisfactorily resolved.
Supervisors have responsibilities when errors are discovered
When a student identifies a serious error, the supervisor should help determine its scope rather than focusing first on blame or reputation.
The relevant data and records should be preserved. The analysis may need to be repeated. Collaborators, ethics bodies, funders, institutions, or journals may need to be informed depending on the circumstances. If the research record is affected, an appropriate correction may be necessary.
The student’s willingness to disclose the problem should not become a reason to suppress it. When a consequential mistake is confirmed, correcting an honest mistake may itself become an integrity obligation.
Watch Out
A supervisor’s desire to protect a student, laboratory, thesis timeline, grant, or publication does not justify concealing a material problem. Support the student while addressing the research record honestly. Those objectives are not mutually exclusive.
Institutions must support supervisors too
Supervisory responsibility does not remove institutional responsibility.
Institutions should provide clear research policies, responsible-conduct training, appropriate infrastructure, accessible research integrity expertise, fair procedures for addressing concerns, and systems through which students can seek advice outside the immediate supervisory relationship.
The National Academies treats research integrity as an institutional as well as individual responsibility and identifies inadequate institutional capacity for fostering integrity as a detrimental research practice.
This matters particularly when the supervisor is the source of the concern. A student cannot meaningfully be told to "raise it with your supervisor" when the supervisor is the person allegedly pressuring the student to act improperly.
