The Daubert Reliability-Factor Analysis Under Rule 702

What the Reliability Factors Are

Daubert gave federal judges a working list of questions for testing whether expert methodology is reliable: Can it be tested? Has it survived peer review? What is its error rate? Do standards control it? Is it generally accepted? The list guides the inquiry — it was never meant to be a scoreboard.

The five inquiries come straight from the Supreme Court’s opinion and its progeny: whether the theory or technique can be, and has been, tested; whether it has been subjected to peer review and publication; the known or potential rate of error; the existence and maintenance of standards controlling the technique’s operation; and whether it enjoys general acceptance within a relevant scientific community. The Fifth Circuit’s criminal cases repeat the same framework. In United States v. Norris, the court confirmed both halves of the doctrine at once: the gatekeeping obligation applies to all expert testimony, not just laboratory science, and the four classic factors are “non-exclusive” and do not constitute a definitive checklist or test.

Two consequences follow for motion practice. First, no single factor is dispositive in either direction — a technique with an unknown error rate is not automatically excluded, and a generally accepted technique is not automatically admitted. Second, the factor list is a floor for questioning, not a ceiling: courts also ask whether the opinion was developed for litigation rather than grown out of independent work, whether the expert accounted for obvious alternative explanations, and whether the field itself is known to reach reliable results. The factor analysis, in other words, is a structured cross-examination of the method conducted by the judge before the jury is ever exposed to it — and the defense’s job is to give the court specific, factor-shaped reasons to ask hard questions. The Fifth Circuit has also emphasized that the inquiry is tied to the task at hand: the factors matter only insofar as they help the court decide whether this witness’s reasoning can be trusted on this question, which is why a motion should frame each factor around the specific opinion the government intends to elicit rather than around the discipline in the abstract.

The Rule 702 Framework the Factors Serve

The factors are not freestanding law; they serve the four requirements in the text of Rule 702. As amended December 1, 2023, the rule admits an opinion only if the proponent demonstrates to the court that it is more likely than not that all four requirements are satisfied.

The four requirements track the rule’s subsections: (a) the expert’s scientific, technical, or other specialized knowledge will help the trier of fact; (b) the testimony is based on sufficient facts or data; (c) the testimony is the product of reliable principles and methods; and (d) the expert’s opinion reflects a reliable application of those principles and methods to the facts of the case. The classic Daubert factors do most of their work under subsection (c), but the 2023 amendment deliberately moved the spotlight to (b) and (d) — the sufficiency of the basis and the reliability of the application — where most real-world expert failures happen.

The amendment made two textual changes. It wrote the preponderance burden into the rule — “the proponent demonstrates to the court that it is more likely than not” — to correct decisions that had waved reliability disputes through to the jury as questions of weight. And it rephrased subsection (d) to require that the “opinion” itself reflect a reliable application, which the advisory committee explained means every conclusion must stay within the bounds of what the basis and methodology can reliably support. The committee note singles out overstatement — opinions expressed with certainty a subjective method cannot deliver — as a principal target.

The forum for all of this is Rule 104(a): the judge, not the jury, decides the preliminary questions, and in doing so is not bound by the evidence rules except privilege. That is why the reliability fight is won or lost in motion papers and pretrial hearings rather than in front of the jury.

Flexible, Not Optional: How Courts Choose Which Factors Apply

Trial judges have considerable leeway in deciding how to test reliability — which factors fit the discipline, and what substitutes make sense when a factor does not translate. What they do not have is the option to skip the inquiry. Flexibility governs the how, never the whether.

The Fifth Circuit’s formulation in Johnson v. Arkema, Inc. captures the latitude: judges have “considerable leeway in both how to test the reliability of evidence and determining whether such evidence is reliable,” and broad discretion to decide whether the body of evidence an expert relies on is sufficient to support the opinion. Norris supplies the criminal-side illustration: for an agent’s videotaped fire re-creation, the court did not march through peer-review literature; it asked the factor-adjacent question that actually fit — whether the experiment was conducted under conditions substantially similar to the events at issue.

But the same cases mark the boundary. In Black v. Food Lion, Inc., the Fifth Circuit distilled the overarching standard: the gatekeeper must ensure that the expert “employs in the courtroom the same level of intellectual rigor that characterizes the practice of an expert in the relevant field.” Black shows what failing that standard looks like — a physician’s causation theory that had never been tested, never been published, and was not accepted in the field was excluded, and the court called her elimination-based reasoning the fallacy of post-hoc logic rather than science.

For the defense, the practical move is to propose the test. A motion that tells the court which factors fit this discipline, which do not and why, and what questions should replace them gives the judge a road map — and makes an unexplained refusal to engage with the factors far harder to defend on appeal.

Applying the Factors Beyond Laboratory Science

Most contested experts in criminal cases are not bench scientists — they are examiners, officers, physicians, and technicians whose method is trained judgment. In the Fifth Circuit, the reliability analysis reaches all of them: experience-based and technical testimony gets no pass around the factor analysis.

The circuit settled the point early. Watkins v. Telsmith, Inc. held that the Daubert analysis applies to expert testimony grounded in engineering principles and practical experience, expressly rejecting the idea that only novel or controversial science needs screening; the factors — testing, peer review, general acceptance — remain germane to deciding whether the witness is “a hired gun or a person whose opinion in the courtroom will withstand the same scrutiny that it would among his professional peers.” The en banc court then anchored the framework in Moore v. Ashland Chemical Inc.: the proponent must prove reliability by a preponderance, and that showing requires “some objective, independent validation of the expert’s methodology” — the expert’s own assurance that he used generally accepted methods is not enough.

Translated to the experts who appear in federal criminal trials, the factor questions become concrete. For a feature-comparison examiner: has the discipline been tested by blind proficiency studies, and what error rates did they show? For an officer interpreting behavior or language: can the interpretive method be checked against anything outside the officer’s own conclusions? For a physician or toxicologist: does published literature actually connect the data to the conclusion, or does the opinion rest on clinical intuition dressed as method?

The unifying test is Moore’s: independent validation. Whenever the only evidence of reliability is the witness vouching for his own technique, the proponent has not carried the burden the rule now states expressly — and the defense should say exactly that, factor by factor.

The Analytical-Gap and Sufficiency Attacks

A method can be flawless and the opinion still inadmissible. The analytical-gap doctrine targets the space between the data and the conclusion: when the expert’s leap from one to the other rests only on his own authority, the court may exclude the opinion as connected to the data by nothing but ipse dixit.

The Fifth Circuit’s en banc Moore opinion adopted the principle in the Supreme Court’s words: nothing in Daubert or the rules “requires a district court to admit opinion evidence which is connected to existing data only by the ipse dixit of the expert,” and a court “may conclude that there is simply too great an analytical gap between the data and the opinion proffered.” Johnson v. Arkema shows the doctrine at work: experts relied on studies of related chemicals and broad chemical classes, and the court held they were “free to reject a theory based on extrapolation” where the bridge from those studies to the specific substance and injury was never built.

The gap attack maps onto two subsections of amended Rule 702. Subsection (b) — sufficient facts or data — asks whether the inputs exist at all: an opinion built on cherry-picked observations, incomplete records, or assumptions contradicted by the case file fails at the foundation. Subsection (d) — reliable application — asks whether the route from inputs to conclusion follows the method’s own rules: an examiner who skips steps his protocol requires, or a witness whose stated certainty outruns what the technique can deliver, fails in the application even if the technique is sound.

In briefing, the gap argument has a distinctive virtue: it concedes nothing. The defense can stipulate that the discipline is respectable and still show that this opinion, on these facts, was reached by extrapolation the method does not license. Courts that would hesitate to condemn a whole field will still close the gap in a single case. The doctrine also polices the opposite maneuver — an expert who starts from a sound conclusion in the literature and stretches it to cover facts the underlying studies never examined. Either way, the question the court must answer is the same: does the stated reasoning actually carry the weight the opinion places on it, or does the witness’s authority do the carrying?

Briefing the Factors: A Working Checklist

A reliability motion earns a hearing by being specific. The strongest format walks the court through each factor, states what the government’s disclosure shows on that factor, and identifies exactly what is missing — then asks for findings on the record, factor by contested factor.

The table below is the working checklist the motion should track.

Around the checklist, three procedural moves carry the motion. First, tie every factor argument to the government’s own expert disclosure, so the court sees the gaps in the proponent’s proof rather than a war of adjectives. Second, remind the court where the burden sits: under Moore and the amended rule, the proponent must establish each requirement by a preponderance, with objective, independent validation. Third, request explicit Rule 104(a) findings on each contested requirement — findings force engagement, sharpen any limiting order, and preserve the record. Pair the motion with a Rule 403 objection where the method’s aura of precision outruns its substance.

Where This Fits

This guide is one of four situations that arise under the Daubert Motion (Expert Testimony). Start with the parent motion for the overall framework, or move to a related fact pattern:

What the Case Law Says

These decisions—verified against primary sources—control how this issue is litigated. Every case still turns on its own facts.

• Moore v. Ashland Chemical Inc., 151 F.3d 269 (5th Cir. 1998) (en banc) — The proponent must prove reliability by a preponderance; an expert's own assurances are insufficient without objective, independent validation of the methodology.
• Black v. Food Lion, Inc., 171 F.3d 308 (5th Cir. 1999) — Gatekeeping ensures the expert employs the same intellectual rigor in court as in the field; an untested, unaccepted causation theory was properly excluded.
• Watkins v. Telsmith, Inc., 121 F.3d 984 (5th Cir. 1997) — The Daubert reliability analysis applies to technical and experience-based expert testimony, including engineering opinions, not only to hard science.
• United States v. Norris, 217 F.3d 262 (5th Cir. 2000) — The gatekeeping obligation applies to all expert testimony; the four reliability factors are non-exclusive and not a definitive checklist.
• Johnson v. Arkema, Inc., 685 F.3d 452 (5th Cir. 2012) — Courts may reject extrapolated opinions where the analytical gap between data and opinion is too great; judges have leeway in testing reliability.

General summaries of published opinions for information only — not predictions about any specific case.