Scientific integrity is the commitment to conducting, analyzing, communicating, and reviewing research according to standards that support trustworthy scientific knowledge.
It applies throughout the research lifecycle.
Researchers are expected to represent data accurately, describe methods transparently, report findings responsibly, disclose relevant conflicts, acknowledge limitations, and correct the scientific record when significant errors are discovered.
Scientific integrity overlaps with academic integrity, but the concepts are not identical. Academic integrity can apply broadly to activities such as teaching, coursework, examinations, attribution, and scholarly conduct. Scientific integrity focuses more specifically on the practices required for trustworthy research.
Several principles underpin responsible scientific practice.
Honesty. Researchers should present methods, data, results, and uncertainties accurately.
Selective manipulation or intentional misrepresentation undermines the ability of others to evaluate research.
Transparency. Scientific conclusions are easier to evaluate when researchers clearly explain how the study was conducted and how analytical decisions were made.
Where appropriate, sharing data, protocols, code, and other research materials can further support transparency.
Accountability. Researchers remain responsible for the accuracy of the work associated with them.
Authorship, supervision, data management, and correction of errors all involve accountability.
Reproducibility. Other researchers should have enough methodological information to understand how results were generated and, where appropriate, attempt to reproduce or replicate them.
Objectivity. Scientific reasoning should be driven by evidence rather than by a preferred conclusion.
Definitions vary somewhat between institutions and jurisdictions, but research misconduct commonly focuses on fabrication, falsification, and plagiarism.
Data Fabrication. Data fabrication involves making up data, observations, participants, experiments, or results that did not actually occur.
This directly undermines the evidentiary basis of the research.
Data Falsification. Data falsification involves manipulating research materials, equipment, processes, data, or results in a way that misrepresents the research record.
This can include inappropriate alteration, omission, or presentation of data when performed deceptively.
Plagiarism. Plagiarism involves presenting another person's words, ideas, methods, or intellectual contributions as one's own without appropriate attribution.
Not every problem in a research paper constitutes misconduct.
Researchers can make calculation errors, choose an imperfect method, misinterpret findings, or reach conclusions that later prove wrong.
The distinction often depends on factors such as intention, recklessness, severity, and whether the research record was deliberately misrepresented.
No single mechanism protects the scientific record.
Researchers and Institutions. Research institutions typically maintain policies covering responsible conduct, data management, authorship, conflicts of interest, and misconduct investigations.
Peer Review. Peer reviewers may identify methodological inconsistencies, unsupported conclusions, suspicious data patterns, or reporting problems before publication.
However, peer review is not designed to detect every form of misconduct.
Journals and Publishers. Editors may issue corrections, expressions of concern, or retractions when significant problems are discovered.
Funding Agencies. Funding organizations can establish research integrity requirements and, depending on their authority and jurisdiction, investigate or respond to alleged violations associated with funded research.
Post-Publication Scrutiny. Researchers, readers, statistical reviewers, integrity specialists, and automated tools increasingly evaluate published papers after publication.
Scientific integrity therefore depends on both prevention and correction.
Related concepts such as scientific bias and submission readiness also affect how reliably research is designed, assessed, and communicated.