03 · What You Need to Know
How to Investigate Missing Research Records Without Compromising the Evidence
What Counts as an Important Missing Research Record?
Not every missing file has the same significance. The importance of a record depends on what it documents and whether its absence affects the ability to understand, verify, or defend the research.
Important missing materials may include original observations, instrument outputs, laboratory notebooks, field notes, consent documentation, data dictionaries, analytical scripts, coding records, or evidence supporting consequential methodological decisions.
For example, losing a duplicate presentation file may have little effect if the authoritative version remains available. Losing the only original dataset underlying a published analysis may create a substantially more serious problem.
The distinction between research data and the broader research records matters because a study may retain its numerical results while losing documentation needed to establish their provenance.
What Should You Do Immediately After Discovering the Loss?
The first priority is to prevent additional loss or alteration while establishing what has happened.
Identify the affected records, their last known location, the people who had authorized access, and the circumstances in which they were discovered missing. Record the actual date of discovery rather than assuming that the loss occurred on that date.
If the problem may involve a storage failure, accidental overwrite, or security incident, avoid actions that could destroy recoverable information. Continuing to write data to an affected device, for example, may complicate recovery.
Contact the appropriate institutional information technology, records-management, or research administration personnel when the situation exceeds the researcher's technical authority or expertise.
If the missing records contain identifiable participant information, consider whether the circumstances also create a confidentiality or security incident. The response may need to follow the institution's incident-reporting procedures in addition to its research-integrity requirements.
Watch Out
Do not silently replace missing original records with reconstructed entries. A reconstruction may help explain what probably occurred, but it must not be represented as contemporaneous evidence that survived intact.
How Should You Search for Missing Research Records?
A systematic search is preferable to relying on a single folder or the recollection of one researcher.
Begin with authorized storage locations, including institutional repositories, approved shared drives, laboratory systems, version-controlled repositories, archived project folders, and backup systems.
Determine whether the record was moved, renamed, archived, superseded, or stored under a different identifier. File metadata, transfer logs, repository histories, and inventory records may help establish its location.
Where another researcher previously managed the material, consult that person through appropriate channels. A missing file may reflect an incomplete handover rather than destruction.
Recovery should respect access controls and privacy obligations. The fact that a file is important does not authorize accessing another person's private account or device without permission.
Keep a record of the search, particularly when the missing material is consequential. Documenting which sources were checked and what was recovered helps distinguish a genuinely unavailable record from one that has not yet been located.
What Is the Difference Between a Missing Record and a Destroyed Record?
A missing record is one whose location or availability cannot presently be established. A destroyed record is one that has been eliminated or rendered irrecoverable, to the extent that this can be established.
These conditions should not be conflated. A record absent from a working directory may still exist in an institutional archive or backup.
Conversely, a record may appear in a file inventory but be unreadable because of corruption, obsolete software, or missing encryption credentials.
The investigation should establish what is known about the record's status. If destruction cannot be confirmed, describe the material as missing or unavailable rather than asserting that it has been permanently deleted.
Where deletion is involved, the distinction between authorized disposal and inappropriate destruction should be examined using the relevant research record deletion requirements.
Can Missing Research Records Be Reconstructed?
Sometimes, but reconstruction must be distinguished from recovery.
Recovery restores an authentic existing record from a backup, archive, source system, or another reliable copy. Reconstruction creates a new account or derived record using evidence that remains available.
For example, an original dataset may be recovered from a verified institutional backup. By contrast, recreating a data table from published summary statistics produces a new reconstruction, not the original participant-level observations.
Similarly, an experimental procedure may be partially reconstructed from instrument logs and dated correspondence. The resulting account may be informative, but its evidentiary status differs from an original laboratory entry.
Reconstruction should identify its sources, assumptions, uncertainties, and date of creation. It should never conceal the absence of the original material.
The broader standard is whether the surviving evidence permits a defensible reconstruction of consequential research activities.
How Do You Determine Whether the Loss Affects Research Findings?
The consequences of missing records should be assessed in relation to the claims they support.
Ask whether the remaining evidence establishes the original observations, relevant methodological procedures, data transformations, and analytical results.
A lost working copy may have no substantive effect if an authenticated authoritative version remains available. Missing source observations may create a serious limitation even when final statistical outputs survive.
The assessment should distinguish several possible consequences:
-
No material evidentiary loss: The necessary information remains available through verified authoritative records.
-
Partial verification limitation: Some procedures or observations can be established, but particular details remain uncertain.
-
Analytical limitation: The missing information prevents an analysis from being reproduced or adequately checked.
-
Substantive evidentiary limitation: The remaining records are insufficient to support one or more important findings.
-
Compliance or security concern: The loss may violate retention, confidentiality, or other applicable obligations.
These categories are practical distinctions, not a formal classification imposed by a particular regulatory authority.
Importantly, the existence of a final result does not establish the authenticity or accuracy of the original data. A statistical output may be mathematically reproducible from a retained analytical dataset while the underlying observations remain unverifiable.
Does Losing Research Records Mean Research Misconduct Occurred?
Not necessarily. Research records can be lost through accidental deletion, hardware failure, inadequate storage practices, natural disasters, or other circumstances that do not establish fabrication, falsification, or plagiarism.
Under the U.S. Public Health Service framework, research misconduct has a defined scope and requires assessment under specified standards. An honest error is not automatically research misconduct.
However, intentional destruction, concealment, or manipulation of records may raise serious concerns, particularly when the records are relevant to an investigation or when their removal causes the research to be inaccurately represented.
Researchers should avoid drawing conclusions about intent from missing records alone. The circumstances require evidence-based assessment under the applicable institutional or regulatory framework.
The appropriate response is to preserve relevant information and refer credible concerns through established procedures rather than conduct an informal investigation that could compromise evidence.
When Should Missing Records Be Reported?
Reporting obligations depend on the importance of the records, the nature of the loss, and the requirements governing the project.
A minor problem resolved through an authorized backup may be handled within routine project documentation. A substantial loss affecting participant confidentiality, regulated records, sponsor obligations, or published findings may require formal notification.
Potential recipients include the principal investigator, institutional research administration, research-integrity office, information-security team, data-protection officer, ethics committee, sponsor, or journal editor.
These offices have different responsibilities. A data-security incident should not automatically be treated as a misconduct allegation, and a research-integrity concern should not be assumed to be solely an information technology problem.
Follow the applicable reporting procedures and deadlines. Where an urgent privacy or security risk exists, notification should not be delayed while the research team completes its assessment of scientific consequences.
What if the Missing Records Support an Already Published Study?
Publication does not eliminate the importance of the underlying research evidence. Records may remain necessary for verification, subsequent analysis, institutional review, or compliance with retention requirements.
If important source records are missing after publication, assess whether the remaining evidence adequately supports the reported findings.
The appropriate editorial response depends on the significance of the loss and the reliability of the published conclusions. Some circumstances may require no public notice, while others may warrant discussion with the journal about correction, an expression of concern, or retraction under its applicable procedures.
COPE's guidance on post-publication concerns provides a framework for editorial responses, but the mere discovery of missing records does not automatically justify a particular editorial action.
Researchers should provide accurate information about the loss and avoid claiming that findings have been independently verified when the necessary evidence is unavailable.
How Can Researchers Reduce the Risk of Future Record Loss?
Prevention requires more than saving files occasionally. Research teams should establish authoritative storage locations, controlled access, appropriate backups, versioning, retention responsibilities, and procedures for transferring records when personnel change.
Backups should be tested for recoverability rather than assumed to be functional. Important records should not depend entirely on an individual researcher's personal device or account.
Projects should also preserve the documentation necessary to interpret their files, including data dictionaries, analytical scripts, methodological decisions, and version relationships.
These practices support long-term preservation of the evidence behind a study and reduce the risk that a personnel change will make records inaccessible.
