Case Study 1: Daubert v. Merrell Dow Pharmaceuticals — the Case That Made the Judge a Gatekeeper

A structured analysis of the 1993 U.S. Supreme Court decision that replaced "general acceptance" with judicial reliability gatekeeping. Facts here are drawn from the published opinion and the public record of the litigation (Tier-1). The point of studying it is not the drug at the center — it is the rule the case produced, and the irony that the standard meant to keep junk science out was born in a fight over plaintiffs' science, not forensic science.

Background: a morning-sickness drug and two children

Bendectin was a prescription drug widely taken in the United States to treat nausea in pregnancy. Two children, Jason Daubert and Eric Schuller, were born with serious birth defects, and their families sued the drug's maker, Merrell Dow Pharmaceuticals, alleging that the mothers' use of Bendectin during pregnancy had caused the defects. The case was a civil product-liability suit — no crime, no defendant facing prison. The standard of proof was the civil one: preponderance of the evidence, more likely than not. Keep that civil setting in mind; it is the chapter's "bitter irony" in seed form.

The legal battle came down to a single scientific question: can Bendectin cause birth defects in humans? And that question came down to whose experts the jury would be allowed to hear.

The evidence fight: published epidemiology vs. re-analysis

The two sides offered fundamentally different kinds of scientific evidence, and the contrast is a clinic in what the four Daubert factors are for.

Merrell Dow's evidence rested on the published epidemiological literature. The company put forward an expert who reviewed a large body of existing human studies — studies that, taken together, had not found a statistically significant link between Bendectin and birth defects. This was peer-reviewed, published, generally accepted work in the ordinary sense: the mainstream epidemiology of the day did not show the drug to be a human teratogen.

The families' evidence was different in character. Their experts proposed to testify that Bendectin could cause birth defects, based on: animal ("in vivo") studies in which high doses produced defects; chemical-structure ("in vitro") analyses suggesting a mechanism; and — critically — a reanalysis ("recalculation") of the published human data that, the experts argued, did reveal a link the original studies had missed. This reanalysis had largely not been published or subjected to peer review; it had been prepared for the litigation.

🔬 At the Bench Notice the shape of the dispute, because it recurs throughout forensic science. One side offers published, externally scrutinized science that reached a negative conclusion. The other offers a litigation-driven reanalysis of the same data that reached the conclusion the party needed. The reanalysis might, in principle, have been correct — published science is not infallible and minority findings are sometimes right. But the markers of reliability the four Daubert factors look for — testability, a known error rate, peer review, independent acceptance — were largely absent from it. The case forced courts to decide: do those markers matter for admissibility, or is it enough that a credentialed expert is willing to say it?

The procedural path and the question presented

The trial court granted summary judgment for Merrell Dow, and the Court of Appeals affirmed, both relying on the Frye standard: the families' reanalysis was not generally accepted in the scientific community, so it was inadmissible, and without it the families could not prove causation. Under Frye, that was the end of the matter — the methodology had not won over its field, so the jury would never hear it.

The Supreme Court took the case to resolve a question that had been simmering since Congress enacted the Federal Rules of Evidence in 1975: did FRE 702 displace the old Frye "general acceptance" test, or incorporate it? The text of FRE 702 did not mention Frye or general acceptance at all. Did its silence mean Frye survived, or that something new had replaced it?

The holding: reliability gatekeeping replaces "general acceptance"

In 1993, the Court held that the Federal Rules of Evidence superseded the Frye test. "General acceptance" was no longer the sole gate for scientific evidence in federal court. In its place, the Court announced what this book calls the Daubert standard: under FRE 702, the trial judge must serve as a gatekeeper, making a preliminary assessment of whether the reasoning and methodology underlying proposed expert testimony are scientifically valid and properly applicable to the facts — that is, whether the testimony is both reliable and relevant — before the jury hears it.

To guide that assessment, the Court offered the now-famous, deliberately flexible and non-exclusive factors — testability/falsifiability, known or potential error rate (and the existence of standards controlling the technique's operation), peer review and publication, and general acceptance (now demoted from the test to a factor). The Court was emphatic that this was not a rigid checklist: the inquiry is "flexible," focused on principles and methodology rather than conclusions, and tailored to the evidence at hand.

⚖️ In the Courtroom The Court did not hold that the families' evidence was inadmissible. It vacated the lower decision and sent the case back to be re-evaluated under the new standard. (On remand, the Court of Appeals again found the families' evidence wanting — this time under Daubert's reliability lens rather than Frye's acceptance lens — and again affirmed judgment for Merrell Dow.) The Supreme Court's job was to fix the rule, not the result. This is the recurring lesson of the whole chapter: an admissibility decision is about what the jury may hear, not about what is ultimately true.

What the case did and did not establish

What it established:

• The trial judge, not the relevant scientific field, is the gatekeeper of scientific evidence in federal court. Reliability — not mere reputation — became the touchstone.
• A flexible set of reliability factors (the Daubert factors) that have since become the rough rubric for evaluating any scientific method offered in court, including every forensic discipline in this book.
• The principle, stated in Justice Blackmun's opinion and quoted in this chapter's epigraph, that the answer to "shaky but admissible" evidence is the adversary system itself — cross-examination, contrary evidence, and jury instruction.

What it did not establish — and the cautions:

• Daubert did not say how rigorously judges must apply the factors, nor make judges into scientists. It handed a harder question (real validity) to the same generalist gatekeeper. Six years later, Kumho Tire had to clarify that the duty extends to all expert testimony, not just self-described "science."
• The case arose in civil litigation, and it is in civil cases — especially against plaintiffs' novel experts — that Daubert has been applied most aggressively. The standard's forensic promise (keeping junk pattern-evidence away from criminal juries) has been far less fully realized. The very birth of the standard previews the asymmetry the chapter calls its "bitter irony."
• Blackmun's faith in cross-examination assumes a jury that can tell strong science from weak. The CSI effect (Chapter 1) is the reason to doubt that assumption, especially for confident, credentialed forensic testimony.

The lesson

Daubert is the moment the law formally adopted the scientific method as its admissibility test — at least on paper. The four factors are, almost line for line, the questions a scientist asks of a claim: Can it be tested? What is its error rate? Has it survived outside scrutiny? Who accepts it, and on what basis? That is an enormous achievement, and a world with Daubert gatekeeping is better than the world of pure Frye deference that preceded it.

But the case also seeds every limitation the chapter develops. It was born in a civil suit, where the gate is guarded most fiercely. It left the gate in the hands of a generalist judge who is not equipped to do the science. And it trusted the adversary system to catch what the gate let through — a trust that, as the next case study shows, can fail with fatal consequences. The standard is real. Whether it does its job depends entirely on the gatekeeper's willingness and ability to apply it, and that is exactly where the system is weakest in criminal court.

Discussion questions

1. The families' reanalysis "might in principle have been correct." Why, then, do the Daubert factors justify treating its absence of peer review and unknown error rate as grounds for skepticism? When could a litigation-driven reanalysis still be reliable?
2. The Supreme Court fixed the rule but not the result — and on remand the families lost again, this time under Daubert. What does this teach about the difference between an admissibility standard and a judgment about truth?
3. Daubert was decided in a civil case but is cited constantly in criminal forensic disputes. Using the chapter's "bitter irony," explain why the same standard has produced such different real-world rigor in the two settings.
4. Justice Blackmun wrote that cross-examination and jury instruction are the cure for "shaky but admissible evidence." State the assumption that sentence makes about juries, and use the CSI effect (Chapter 1) to argue why that assumption is fragile for forensic testimony.
5. Apply the four Daubert factors to both sides' evidence in this case — the published epidemiology and the families' reanalysis. Which side's evidence better satisfies the factors, and does that match which side won?
6. Connect this case to the Cold Case admissibility map. The chapter sorts future evidence into "clears an honest gate," "contestable," and "should not survive." Which pile would the families' unpublished reanalysis fall into, and why?