What the Scientific Record Leaves Out: Unreported Evidence and Expert Reliability Under Daubert

This article is the first in a two-part “Law Meets Science” series on how the state of the scientific record can affect the reliability of expert evidence. This installment examines the litigation significance of unpublished or selectively reported research findings. Part 2 will address a related concern: how changes in scientific knowledge over time can independently bear on expert reliability.

In a recent news release, the FDA reminded more than 2,200 medical product companies and researchers of their obligation to submit certain clinical trial results to ClinicalTrials.gov, the federal registry and results database for studies involving human participants. According to the agency, 29.6% of studies highly likely to be subject to mandatory reporting still have no results information on file. The agency warned that the absence of unfavorable findings can leave the public record incomplete, fuel publication bias, and distort perceptions of medical product safety and efficacy.

That concern extends beyond clinical care. When positive findings are more visible than negative or null results, the published literature can present a skewed account of the evidence. In litigation, that imbalance can shape how experts evaluate causation, safety, and a product’s regulatory history.

That is where Daubert and Rule 702 assume particular importance. Trial courts serve as gatekeepers and may admit expert testimony only when the proponent shows, more likely than not, that the opinion is grounded in sufficient facts or data and reflects a reliable application of sound principles and methods to the facts of the case. When the published literature is incomplete because unfavorable results were never disclosed or fully reported, counsel may have reason to test whether an expert’s opinion rests on the full evidentiary record or on a selectively reported subset of it.

For that reason, counsel should scrutinize whether the published record is complete. One practical step is to compare published articles with study registrations, posted results, and other available research records to identify gaps, including completed studies with no reported results, discrepancies between registered and published endpoints, or publications that appear to present only part of the underlying data. That inquiry fits squarely within Daubert and Rule 702. To be sure, not every undisclosed study or analysis will be identifiable. Some research may never be registered, some results may never be posted, and some discrepancies may not be apparent from the published record alone. Even so, careful attention to these issues can help build a stronger evidentiary record and support more disciplined challenges to expert reliability.

As this article suggests, gaps in the scientific record can materially affect the foundation for expert analysis. But an incomplete record is not the only concern. In Part 2, we examine a distinct but related issue: how changes in the scientific landscape over time can raise similar reliability questions under Daubert and Rule 702.