03 · What You Need to Know
What Makes the Evidence Behind a Research Integrity Concern Sufficient?
Reporting a Concern Is Different From Proving Misconduct
A research misconduct process ordinarily involves different evidentiary questions at different stages.
At the reporting stage, the question is whether there is a sufficiently identifiable concern that warrants assessment. At later stages, authorized officials examine whether the allegation falls within the applicable definition of misconduct and whether the evidence supports further proceedings or a finding.
Under the United States Public Health Service research misconduct framework, a formal finding involves specified requirements concerning the conduct, culpability, and evidentiary standard. The regulation uses a preponderance-of-the-evidence standard for determining whether research misconduct occurred.
That standard concerns the formal determination, not a requirement that every complainant independently establish misconduct before making a report.
Researchers should therefore avoid confusing three distinct questions:
Is there a concern worth reporting?
Whether identifiable information provides a reasonable basis for seeking assessment.
Does the allegation warrant further proceedings?
Whether the applicable procedural criteria are satisfied after authorized assessment.
Has misconduct been established?
Whether the evidence satisfies the relevant substantive and procedural requirements for a formal finding.
These questions should not be collapsed into a single demand for proof at the earliest stage.
What Does a Reasonable Factual Basis Look Like?
A factual basis consists of identifiable information that can be examined, rather than a general impression that a researcher has behaved improperly.
For example, a report might identify two published figures containing apparently identical image regions despite being labeled as different experiments. It might identify a dataset in which particular values differ from the corresponding original instrument records.
The reporting person need not know why the discrepancy occurred. They should be able to explain what they observed and where the relevant evidence may be found.
Useful characteristics include:
-
Specificity: The concern identifies particular records, results, activities, or publications.
-
Verifiability: The relevant information can potentially be examined by an authorized reviewer.
-
Relevance: The discrepancy concerns research conduct, evidence, or reporting rather than an unrelated personal dispute.
-
Accuracy: The report distinguishes direct observations from assumptions and secondhand information.
-
Context: Available information that may explain or qualify the discrepancy is not deliberately omitted.
These characteristics provide a practical basis for preparing a report. They are not a universal formal test imposed by every institution.
How Much Evidence Is Enough to Request Confidential Advice?
The threshold for seeking advice may be lower than the threshold for submitting a formal allegation.
A researcher who is uncertain whether an observed discrepancy constitutes a research-integrity concern may consult an appropriate institutional official, research integrity officer, or authorized advisory service.
For example, a junior researcher might ask whether an unexplained change to a dataset should be handled as a routine documentation problem or referred for formal assessment.
Confidential consultation can help clarify the relevant policy and reporting procedure before the researcher takes potentially consequential action.
However, confidentiality arrangements vary. Some offices may be required to act on information indicating serious misconduct or immediate harm. Researchers should establish the scope of confidentiality before disclosing sensitive details.
What Types of Evidence May Support a Misconduct Concern?
Different concerns require different kinds of evidence.
| Possible concern |
Potential supporting information |
Important limitation |
| Fabricated data |
Discrepancies between reported observations and authentic source records, or evidence that reported activities did not occur. |
Unusual statistical patterns alone do not establish fabrication. |
| Falsified results |
Documented changes to measurements, images, or analytical outputs that may misrepresent the research. |
A legitimate correction or transformation may explain the change. |
| Plagiarism |
Identifiable similarities between earlier source material and later unattributed work. |
Quotation, attribution, common terminology, and disciplinary conventions require examination. |
| Undocumented exclusions |
Differences between source datasets, exclusion logs, and reported sample sizes. |
Missing documentation may reflect error or poor recordkeeping rather than misconduct. |
| Manipulated figures |
Apparent duplication, alteration, or inconsistent representation of images. |
Image assembly mistakes and legitimate processing must be distinguished from falsification. |
The presence of one of these indicators may justify assessment, but its meaning depends on the surrounding evidence.
Is Statistical Improbability Enough to Report Fabrication?
Statistical irregularities can raise legitimate questions, but unusual numerical patterns are not conclusive evidence of fabricated data.
For example, a dataset may contain unexpectedly low variability, repeated observations, or distributions inconsistent with an assumed model.
These patterns may arise from data processing, measurement constraints, coding conventions, sampling characteristics, or other legitimate causes. They may also indicate errors or deliberate manipulation.
A researcher who identifies a specific and consequential anomaly may reasonably seek clarification or institutional advice. The report should explain the observed pattern and the assumptions underlying the concern.
It should not claim that a low probability under a particular statistical model establishes that a researcher fabricated observations.
The National Academies' work on reproducibility and replicability emphasizes the importance of examining research methods, evidence, and sources of uncertainty rather than treating isolated indicators as definitive judgments about scientific validity.
Can You Report a Concern Based on Information From Another Person?
Secondhand information may warrant attention, particularly when it identifies specific records or activities that can be examined.
However, the report should distinguish information personally observed from information received through another person.
For example, a research assistant may tell a collaborator that certain observations were entered without corresponding experiments. The collaborator may have a legitimate reason to seek advice, but should not present the assistant's account as a firsthand observation.
Relevant details may include when the information was received, what was communicated, and whether the original source can be identified through appropriate procedures.
Researchers should avoid pressuring potential witnesses to provide statements or conducting unauthorized interviews. Such activities may affect the reliability of subsequent evidence and interfere with formal procedures.
What if the Evidence Has an Innocent Explanation?
The possibility of an innocent explanation does not necessarily eliminate the basis for a concern.
A duplicated image may result from an assembly mistake. A discrepancy between datasets may reflect an undocumented but legitimate transformation. Similar wording may arise from properly attributed source material or standardized methodological descriptions.
Where a routine clarification can be sought safely and appropriately, it may resolve the issue without formal misconduct proceedings.
However, researchers should not assume that they must obtain an explanation directly from the implicated person before seeking institutional guidance.
In situations involving possible evidence destruction, serious power imbalances, or retaliation risks, direct confrontation may be inappropriate.
The considerations involved in confronting a researcher before reporting suspected misconduct should be evaluated separately from the factual basis of the allegation.
Should You Gather More Evidence Before Reporting?
Gathering additional information may be reasonable when it involves checking publicly available materials or reviewing records within your authorized responsibilities.
For example, confirming that two figures come from the same published article and are labeled as different experimental conditions can make a report more precise.
However, researchers should not delay reporting a sufficiently specific concern merely because they cannot independently obtain restricted source data.
They should also avoid searching colleagues' private accounts, removing original records, copying confidential participant information, or questioning witnesses without authorization.
The evidence needed for formal assessment may be held by the institution and should be secured through its established procedures.
Watch Out
Do not attempt to strengthen an allegation through unauthorized access, covert alteration of research systems, or disclosure of confidential information. Evidence obtained or handled improperly can create additional risks for participants, researchers, and the integrity of subsequent proceedings.
What Does Good Faith Mean in Research Misconduct Reporting?
Good faith generally concerns the sincerity and reasonable basis of the reporting person's actions, although the precise definition varies by policy.
A researcher may raise a concern in good faith and later discover that the suspected misconduct did not occur.
Conversely, a report containing genuine documents is not necessarily made in good faith if the reporting person knowingly misrepresents their meaning or deliberately conceals information that changes the allegation.
The relevant distinction is not whether the allegation ultimately succeeds. It is whether the concern was raised responsibly, with an appropriate factual basis and without knowingly false or misleading claims.
Institutional protections for complainants may depend on the applicable definition of good faith and other procedural requirements.
The consequences of an allegation that proves unfounded should therefore be distinguished from those of a deliberately false report.
Does the Evidence Threshold Change When the Allegation Is Serious?
The seriousness of a possible violation affects the consequences of the concern and the need for careful handling, but it does not mean that a reporting person must independently prove a serious allegation before seeking authorized assessment.
For example, suspected manipulation of clinical research data may have implications for participant safety and require prompt attention. Delaying a report while attempting to obtain definitive proof could create additional risk.
At the same time, serious allegations can cause substantial reputational harm. Reports should therefore remain specific, restrained, and grounded in available information.
The appropriate balance is to avoid both unsupported accusations and unnecessary delay in bringing credible concerns to qualified authorities.
How Do Institutions Evaluate Whether an Allegation Warrants Further Action?
Institutions generally apply their governing policies to determine whether a reported concern falls within the relevant definition of research misconduct and warrants further proceedings.
Under the U.S. Public Health Service framework, an allegation must satisfy applicable jurisdictional and substantive criteria before proceeding through the formal process.
The institution may examine whether the allegation is sufficiently specific and credible, whether the relevant research falls within the framework, and whether the available information warrants an inquiry.
An inquiry is not the same as a final determination of misconduct. Its purpose and procedural requirements are established by the applicable regulation.
Different jurisdictions and institutions may use different stages, terminology, and thresholds. Researchers should consult the current policy rather than assuming that the U.S. framework applies universally.
The distinction between a preliminary inquiry and a formal misconduct investigation is important when interpreting what an institution's decision to proceed actually means.
What if You Have Only a Strong Suspicion but No Identifiable Evidence?
A general belief that a researcher is dishonest, without a specific factual basis, is usually insufficient to support a formal allegation.
Personal conflicts, unusual productivity, unexpected results, or disagreement with a researcher's interpretation do not independently establish misconduct.
Nevertheless, a researcher who is uncertain about the significance of what they have observed may seek confidential advice from an appropriate institutional office.
During that consultation, the researcher can explain the concern accurately and ask whether it falls within the office's responsibilities.
The important distinction is between seeking advice about uncertainty and presenting unsupported speculation as an established allegation.
How Should Evidence Be Organized for a Report?
A concise factual report is generally more useful than a lengthy narrative containing repeated accusations.
Where possible, organize the information around the research activity involved, the specific discrepancy, the supporting materials, and the limits of what is known.
For each important item, identify its source and explain how it relates to the concern.
If the issue involves conflicting research records, preserve the original versions and explain the precise difference rather than replacing them with a single preferred account.
Where evidence is missing, identify what cannot presently be verified. Do not fill gaps with assumptions about what the absent material would have shown.
The report should enable the authorized authority to assess the allegation without requiring the reporting person to conduct the investigation independently.
When Should You Stop Gathering Evidence and Report?
There is no universal quantity of evidence that marks the appropriate reporting point.
A practical question is whether you can identify a consequential concern, explain its factual basis, and direct an authorized authority toward information that may permit assessment.
If those conditions are met, additional independent evidence gathering may offer little benefit and may create unnecessary procedural risks.
Where there is a credible possibility of evidence destruction, participant harm, or retaliation, early confidential consultation may be especially appropriate.
The broader procedures for responding to suspected research misconduct should guide the reporting person's actions once a sufficiently specific concern has been identified.
