Erased by Policy: The Institutional Destruction of Raw Research Data and the Death of Scientific Verification

When a peer-reviewed study is published in a major scientific journal, most readers reasonably assume that the underlying evidence—the raw numbers, the handwritten lab logs, the instrument readings, the participant records—is preserved somewhere, available for independent inspection if the conclusions are ever challenged. That assumption, it turns out, is increasingly wrong.

A growing body of evidence, drawn from public records requests, whistleblower testimony, and a careful reading of institutional data governance policies, suggests that American universities and research centers are routinely destroying the foundational materials of published science. Not through negligence. Through policy.

The Clause Nobody Reads

Every major research institution in the United States operates under a patchwork of data retention policies, many of them shaped by federal funding requirements, journal mandates, and internal administrative procedures. On paper, these policies often require that data be retained for a minimum period—commonly five to seven years following publication. In practice, what happens at the end of that window is where the story becomes troubling.

Embedded within the research agreements that govern federally funded projects are provisions allowing—and in some cases requiring—the disposal of physical and digital research materials once retention minimums are satisfied. Attorneys and compliance officers at major research universities refer to these as 'disposition clauses.' Scientists who have watched their own experimental records destroyed call them something less polite.

One former research associate at a large Midwestern land-grant university, who requested anonymity citing ongoing professional relationships, described the process plainly: 'You get a notice from the research office that your retention period is expiring. You fill out a form. And then the files are gone. Nobody asks whether the science has been replicated. Nobody asks whether someone might need those records to check the work. The clock runs out, and that's the end of it.'

What the FOIA Requests Revealed

Full Disclosure Project submitted public records requests to fourteen major public research universities across nine states, seeking documentation of their data retention and destruction schedules, any internal audits of compliance with those schedules, and records of communications between research compliance offices and external funders regarding data disposal.

The responses were instructive—not primarily for what they contained, but for what they withheld.

Several institutions declined to produce destruction logs on the grounds that such records were themselves subject to routine disposal. Others provided heavily redacted versions of their data governance frameworks, citing proprietary agreements with corporate research partners. Three institutions did not respond within statutory deadlines.

Of the substantive documents obtained, the most revealing were internal policy memoranda from two large public universities in the South and Mid-Atlantic regions. Both documents explicitly authorized the destruction of 'supplementary research materials, including raw data files, instrument calibration records, and laboratory notebooks' following the expiration of the applicable retention period, with no requirement that the disposing researcher notify the journal that published the associated findings.

In other words: a paper can remain in the scientific literature, cited by other researchers and referenced in regulatory decisions, long after every piece of evidence that supported it has been legally incinerated.

The Replication Crisis and the Missing Evidence

This practice does not exist in a vacuum. It intersects directly with what scientists have spent the better part of two decades calling the 'replication crisis'—the disturbing discovery that a significant proportion of published findings in fields ranging from psychology to oncology cannot be independently reproduced.

Replication requires data. When researchers attempt to verify a published result, they typically need access to the original dataset, the precise methodology, and ideally the raw instrument outputs. If those materials have been destroyed pursuant to institutional policy, replication becomes structurally impossible—not because the science was necessarily fraudulent, but because the evidence no longer exists to settle the question either way.

Dr. Mariana Reyes, a research integrity consultant who has worked with several federal agencies, framed the problem in direct terms during a 2023 conference presentation: 'We have built a scientific communication system that treats the published paper as the permanent record and the underlying evidence as temporary. That is exactly backwards from how verification is supposed to work.'

The implications extend beyond academic dispute. Published research informs drug approvals at the Food and Drug Administration, environmental risk assessments at the Environmental Protection Agency, and safety standards across a range of federal regulatory frameworks. When the data behind those papers is gone, regulators and the public are left with conclusions they cannot independently examine.

Corporate Agreements and the Confidentiality Shield

The problem is compounded in cases where university research is conducted under agreement with private industry. Sponsored research contracts—arrangements in which corporations fund academic studies in exchange for intellectual property rights or publication approval—frequently include confidentiality provisions that restrict what data can be retained, shared, or disclosed.

In several instances examined by Full Disclosure Project, these corporate agreements explicitly accelerated data destruction timelines, requiring disposal of certain research materials well before the standard institutional retention period would have expired. The justification offered in contract language reviewed for this article cited 'proprietary methodology protection' and 'competitive sensitivity of experimental parameters.'

The effect is that some of the most consequential research—studies on pharmaceutical efficacy, agricultural chemical safety, and materials toxicology—may be subject to the most aggressive data destruction, precisely because the findings carry the highest commercial stakes.

The Whistleblowers Who Tried to Preserve the Record

Not everyone complies quietly. Several researchers who spoke with Full Disclosure Project described attempts to preserve data beyond institutional disposal windows, only to face formal pressure to comply with destruction schedules.

One former laboratory director at a private research university in New England described retaining a hard drive containing raw data from a contested nutrition study after being instructed to delete the files. 'I kept it because I knew someone would eventually come asking,' she said. 'Six months later, a journalist did. By then, the institution had already certified to the journal that all materials had been disposed of per policy.'

The legal exposure for researchers who preserve data against institutional instruction is not trivial. Depending on the terms of their employment agreements and the nature of the research funding, unauthorized retention of research materials can constitute a breach of contract or, in federally funded contexts, a compliance violation with serious professional consequences.

The incentive structure, in other words, runs entirely in one direction.

What Accountability Requires

The path toward a more transparent research ecosystem is neither technically nor logistically impossible. Several European research funding bodies, including those operating under the framework of Horizon Europe, now require that raw datasets be deposited in publicly accessible repositories as a condition of funding—before publication, not after. The National Institutes of Health introduced a data management and sharing policy in 2023 that nudges American federally funded researchers in a similar direction, though enforcement mechanisms remain limited and exceptions are broad.

What is missing is not a model. It is the institutional will to apply one.

Universities have financial relationships with corporate sponsors that depend on data remaining controlled. Journal publishers have no formal role in monitoring what happens to evidence after a paper appears in their pages. Federal funding agencies impose retention requirements but rarely audit compliance. And the researchers who might object most loudly are also the most professionally vulnerable.

Until the disposal of raw scientific data is treated not as routine administrative housekeeping but as a matter of public accountability, the published scientific record will remain exactly what current policy makes it: a collection of conclusions without a verifiable past. For a nation that funds much of this research through public taxation and relies on its findings to make regulatory decisions affecting public health and safety, that is not a technical problem. It is a transparency failure.