Recent Third Circuit Opinion Reinforces That Daubert Requires Courts to Examine the Data Underlying an Expert’s Analysis

Types : Alerts

Meta Platforms, Inc. recently defeated certification of a class of consumers who claim the company lied about its user privacy safeguards and violated antitrust laws. The court sided with Meta based on a successful Daubert challenge to the plaintiffs’ economics expert’s theory of antitrust injury, concluding the expert’s data did not support his theory. Klein v. Meta Platforms, Inc., No. 20-CV-08570-JD, 2025 WL 295361, at *4 (N.D. Cal. Jan. 24, 2025). In framing its standard, the court cited a recently published (but less publicized) Third Circuit opinion, Cohen v. Cohen, 125 F.4th 454 (3d Cir. 2025)—a case which illustrates the importance of closely examining expert opinions and the data they rely on for their opinions.

In Cohen, the plaintiff sued her father, alleging he had sexually abused her in childhood based on “emotional sensations and physical sensations” she experienced eighteen years later that eventually “crystalized” into memories. Id. at 458. Ms. Cohen’s expert testified that her experience of “recovered memories” was both a clinically recognized phenomenon and a reliable form of recollection. Id. at 459. Mr. Cohen’s expert, on the other hand, testified that there is no scientific support for the proposition that “trauma victims can repress and then later recover memory” containing “all the clarity and detail of the original experience.”  Id. at 458-59.

Mr. Cohen moved to exclude Ms. Cohen’s expert, arguing that “repressed memory is no longer a generally accepted theory among the current scientific community.” Id. at 459. The district court denied the motion because it wanted to keep the conflicting testimony on equal footing. Id. It explained that if Mr. Cohen’s expert could testify “that repressed memories aren’t real,” then Ms. Cohen’s expert could testify that they sometimes are. Id. A jury that heard both experts found for Ms. Cohen and awarded her $1.5 million in damages. Id.

On appeal, the Third Circuit reversed the trial court’s Daubert decision. It held that the lower court abused its discretion when it admitted the testimony of both experts without conducting an individualized Daubert analysis on the record.  And that error was not harmless because, when properly analyzed under Daubert, critical aspects of Ms. Cohen’s expert testimony expressed opinions that were either insufficiently reliable or did not fit the facts of the case. That is, they were inadmissible. Id. at 460–61.

Upon close review of Ms. Cohen’s expert’s opinion and his data, the Third Circuit highlighted three problems with the expert’s testimony:

Relying on old, distinguishable studies with minuscule sample sizes. The panel rejected as unreliable the expert’s conclusion that “there is no evidence that recovered memories are less accurate than continuous memories, but there is evidence for no difference in accuracy,” because it relied on two unpersuasive studies that drew inapt comparisons. Id. at 462-63. The studies themselves were unpersuasive because they were thirty years old and relied on sample sizes that were too small. Id. Although it avoided drawing broad statistical standards, the Third Circuit explained that a “seventeen-person sample in the 1996 study [was] insufficient.” Id. Moreover, neither study compared individuals whose recovered memories were sufficiently like Ms. Cohen’s experience. Id.

Taking the DSM out of context. The Third Circuit next rejected as unreliable the expert’s assertion that “the phenomena of what are known colloquially as ‘repressed’ and ‘recovered’ memories are well-established facts” because he relied solely on the DSM’s inclusion of a broader, distinct condition called “dissociative amnesia.” Id. at 462. The panel equated that assertion to saying one could “demonstrate the acceptance of unicorns” merely by observing a general acceptance of “the diagnosis of ‘hoofed mammals.’” Id.

Making statements that did not match the facts. Other assertions lacked an adequate fit to the facts of the case. Id. at 464. For example, Ms. Cohen’s expert testified at trial that “research over and over again shows that memories freely recalled are 85 to 95 percent accurate.” Id. But Ms. Cohen’s experiences of physical and emotional sensations (as opposed to recalling the alleged abuse because of someone else’s questions) were not emblematic of “free recall” at all. Id.

More fundamentally, the Court held that Ms. Cohen’s expert’s “general theory of memory repression” did not fit what happened in her case. Id. That theory was premised upon his opinion that abuse victims repress memories of their abuse so they can “connect with and survive with the caregiver.” Id. But Ms. Cohen not only did not try to “connect with” her father, she was “disgusted and revolted by him” from an early age. Id.

Cohen is significant in part because it is the first time any federal court has fully examined whether the theory of “repressed memories” should be permitted under Rule 702. But it also offers important, more general guidance: litigants must probe experts’ claims and supporting data—including the statistical soundness of underlying studies—when evaluating the admissibility of their testimony. As Klein demonstrates, trial courts have already begun to take note.