Research Misconduct: Preventing Fabrication, Falsification and Plagiarism

Understand the conduct at issue

Research integrity depends on an accurate account of what was planned, done and found. Fabrication involves making up research data or results. Falsification involves altering or omitting material in a way that misrepresents the research record. Plagiarism involves using another person's contribution without appropriate acknowledgement.

The US Office of Research Integrity defines these categories within its own framework and distinguishes them from honest error and differences of opinion. ^[1] Other institutions and jurisdictions may use broader definitions or procedures. Researchers should know the policy that applies to their work rather than treating any single international definition as a universal legal rule.

Identify pressures and weak points

Problems can arise where expectations reward a result more than a careful process. Tight deadlines, unclear supervision, poor records and fear of reporting mistakes may create conditions in which questionable practices go unchallenged. Prevention therefore requires more than asking researchers to sign an integrity declaration.

Map vulnerable points in the workflow: manual data entry, undocumented exclusions, image editing, reuse of text, collaborative authorship and the transfer of files between team members. Establish checks that are proportionate to the work. A small project may need a clear analysis log and independent review of key outputs; a larger programme may require more formal quality controls.

Preserve an accurate research record

Retain original data securely and distinguish it from cleaned or derived versions. Document transformations, exclusions and corrections. Keep the reasoning behind analysis decisions so another authorised person can understand how reported results were produced. Do not delete inconvenient observations simply because they weaken a preferred conclusion.

For written work, maintain notes that separate quotations, paraphrases and original ideas. Verify citations against the actual source. Similarity software can identify text overlap but cannot, on its own, determine whether plagiarism or misconduct occurred. Interpretation requires context, attribution, disciplinary conventions and a fair opportunity to explain the material.

Make supervision and training practical

Discuss integrity through real workflow decisions: how to handle an unexpected value, when to report a mistake, how to agree authorship and what constitutes acceptable image adjustment. Give junior researchers a route to ask questions without embarrassment or retaliation.

UKRIO's Code of Practice provides guidance for researchers and organisations on good research practice and integrity. ^[2] Institutions can use such resources to inform local procedures, but should translate expectations into responsibilities, training and accessible advice. Rules that exist only in a policy folder are unlikely to help a researcher facing a difficult decision late in a project.

Illustrative example: an unexplained spreadsheet change

Imagine a research assistant notices that several low scores have disappeared from the analysis file. The project leader says they were invalid responses, but no exclusion rule or decision record exists. The assistant preserves the relevant files and raises the concern through the institution's designated process.

The institution should establish what happened without assuming guilt from the initial observation. There may be a documented error elsewhere, a misunderstanding or a serious integrity problem. The review should consider original records, applicable rules and explanations from those involved. If published conclusions are affected, correction of the record needs consideration alongside any personnel or disciplinary process.

Respond to concerns through fair procedures

Provide confidential reporting routes, explain their limits and separate initial assessment from a full investigation where policy requires it. Protect evidence, manage conflicts and avoid informal public accusations. Both the person raising a concern and those whose work is questioned deserve fair treatment.

Do not promise an outcome before review. Record decisions and communicate through authorised channels. Where journals, funders or partners must be informed, coordinate the response according to applicable obligations. Correcting the scholarly record and determining individual responsibility are related but distinct tasks; one should not be neglected because the other is unresolved.

Put prevention into everyday practice

• Agree data, authorship and supervision responsibilities at project start.
• Preserve originals and document transformations.
• Review exclusions and unusual results before reporting.
• Check source attribution and publication permissions.
• Provide confidential advice and reporting routes.
• Train teams using realistic examples.
• Encourage prompt correction of honest mistakes.

An institution's response to error shapes its research culture. Treating every mistake as misconduct can discourage disclosure, while ignoring serious concerns damages trust. A credible approach combines careful documentation, proportionate scrutiny, fair process and a commitment to an accurate research record.