Suspicion Without Proof: How Reviewers Should Handle Suspected Data Fabrication

Peer review is, at its core, an exercise in professional judgment. Reviewers are asked to assess the validity of methods, the soundness of conclusions, and the overall contribution of a manuscript to its field. Most of the time, that judgment operates within well-understood parameters. But occasionally, a reviewer encounters something that defies easy categorization—data that feels wrong, figures that appear too clean, statistical distributions that seem statistically improbable. The work may not be demonstrably fraudulent, but something is off.

This is the reviewer's dilemma: what do you do when suspicion is real but evidence is insufficient?

It is a question that sits at the intersection of professional ethics, institutional protocol, and personal risk. Handled poorly, it can damage the reputation of an innocent researcher. Handled too cautiously, it can allow fabricated science to enter the permanent scholarly record. Understanding how to navigate this gray zone is one of the most important—and least discussed—competencies in the peer review ecosystem.

Understanding the Nature of Suspicion in Peer Review

Suspicion of data fabrication rarely arrives as a single, dramatic red flag. More often, it accumulates. A reviewer may notice that error bars in a figure appear implausibly uniform. Or that a dataset's distribution lacks the natural variability one would expect from real-world collection. Or that results across multiple experimental conditions align with theoretical predictions so precisely that chance alone seems an inadequate explanation.

None of these observations, in isolation, constitutes proof. Statistical improbability is not the same as fabrication. Unusually clean data can reflect rigorous experimental control rather than manipulation. Reviewers must resist the cognitive pull toward premature conclusion while also resisting the opposite temptation—to dismiss discomfort in favor of a smoother review process.

The first step is recognizing that suspicion, when grounded in specific, articulable observations, is itself a form of evidence. It may be insufficient to establish misconduct, but it is not nothing. A reviewer who can point to concrete anomalies—rather than simply expressing a vague unease—is in a far stronger position to act responsibly.

What Reviewers Are and Are Not Equipped to Do

It bears stating clearly: peer reviewers are not investigators. The review process does not grant reviewers access to raw data files, laboratory notebooks, or institutional records. Reviewers work from what authors submit, and that limitation is structural. Attempting to conduct an independent investigation—reaching out to co-authors, querying institutional affiliations, or making direct accusations in the review itself—is both inappropriate and potentially harmful.

What reviewers can do is document their observations with precision and communicate them through appropriate channels. This distinction matters enormously. A reviewer who writes, "The standard deviations in Table 3 appear unusually consistent across all twelve experimental conditions, which warrants closer examination," is doing something fundamentally different from one who writes, "The authors have fabricated their data." The former is a professional observation. The latter is an allegation that demands a level of certainty the review process cannot support.

The reviewer's role, when suspicion arises, is to flag—not to adjudicate.

Escalating Concerns to Editors: When and How

Most major journals operating in the United States follow guidelines established by the Committee on Publication Ethics (COPE), which provides explicit frameworks for handling suspected misconduct. Reviewers who have concerns about data integrity are generally encouraged to communicate those concerns directly to the handling editor, separate from the formal review itself.

This separation is deliberate and important. The review document is typically shared with authors; a confidential communication to the editor is not. A reviewer who includes unsubstantiated suspicions in the review text risks exposing themselves to reputational and legal liability while also potentially contaminating the editorial process.

When reaching out to an editor, reviewers should:

• Be specific. Describe the exact figures, tables, or data points that prompted concern. Vague expressions of unease are difficult for editors to act upon.
• Use measured language. Frame observations as anomalies that merit further scrutiny, not as conclusions about intent or culpability.
• Acknowledge uncertainty. Editors are experienced professionals who understand the difference between a genuine red flag and a misunderstanding. Being transparent about the limits of your knowledge strengthens rather than weakens your credibility.
• Suggest rather than demand. A reviewer might note that independent statistical verification or raw data access could resolve the concern, without insisting on a particular editorial outcome.

Editors who receive such communications are not obligated to act on them in any particular way, but reputable journals will take them seriously and may request additional documentation from authors or involve their institutions.

The Role of Documentation

If a reviewer decides to escalate a concern, documentation is the foundation of a credible communication. Screenshots of figures, notations about specific data values, and references to published literature that establishes expected variance in a given research domain all serve to ground the concern in observable fact.

Reviewers should also consider whether the anomalies they have identified are consistent with known forms of data manipulation—such as data rounding, figure duplication, or implausible sample sizes—or whether alternative explanations are equally plausible. Tools such as image analysis software and statistical forensics resources have become more accessible in recent years, though reviewers should be cautious about venturing beyond their own areas of expertise.

Keeping a personal record of the communication—including dates, the editor's name, and the substance of the exchange—protects the reviewer if questions arise later.

Institutional Variation and the Policy Landscape

One of the more frustrating realities for reviewers operating in good faith is that institutional and journal policies governing suspected misconduct vary considerably. Some journals have detailed protocols for handling integrity concerns; others offer reviewers little guidance at all. Research universities and academic medical centers in the United States are generally subject to federal regulations governing research misconduct, particularly those funded by the National Institutes of Health or the National Science Foundation. But the reviewer, as an external party, may have limited visibility into how those mechanisms operate.

For reviewers affiliated with institutions, consulting with a research integrity officer before escalating concerns externally can be a prudent step—particularly if the manuscript involves federally funded research or high-profile findings. This is not about seeking permission to raise a concern; it is about ensuring that the concern is raised effectively and through channels that carry institutional weight.

Acting Responsibly Without Becoming an Accuser

The professional and ethical stakes of this situation are real in both directions. A fabricated paper that passes through review unchallenged can distort subsequent research, waste public funding, and in some fields—clinical medicine, public health, environmental science—cause direct harm. At the same time, an unfounded accusation can derail a researcher's career and expose the accuser to significant professional consequences.

The framework that best serves reviewers in this situation is one built on proportionality. The strength of the action taken should correspond to the strength of the evidence available. A mild anomaly warrants a quiet note to the editor. A pattern of anomalies across multiple figures, combined with implausible statistical results, warrants a more formal communication. In rare cases where evidence is compelling and the potential harm is significant, reviewers may have an obligation to persist even if an initial concern is dismissed.

PRR Society encourages its members to approach these situations with the same rigor they bring to the review itself—careful, evidence-based, and oriented toward the integrity of the scholarly record rather than the comfort of any individual party. Suspicion, when handled well, is not an accusation. It is an act of professional responsibility.

The peer review system depends on reviewers who are willing to ask hard questions. It also depends on them asking those questions in the right way.