Methodological Failures That Get Opinions Struck

Where the recurring failures come from

Exclusion arguments are drawn from two sources, and keeping them apart is the difference between a brief that is accurate and one that a court corrects in a footnote.

The first is FRE 702 itself, whose four elements supply the categories: helpfulness under (a), sufficient facts or data under (b), reliable principles and methods under (c), and reliable application of those principles and methods to the facts of the case under (d). Since 1 December 2023 the proponent must establish each by a preponderance.

The second is the 2000 Advisory Committee note, which compiled five further considerations from lower-court practice: whether the opinion grew out of independent research or was developed expressly for testifying; whether the expert unjustifiably extrapolated from an accepted premise to an unfounded conclusion; whether obvious alternative explanations were adequately accounted for; whether the expert is being as careful as in regular professional work outside paid litigation consulting; and whether the field is known to reach reliable results for the type of opinion offered.

Those five are frequently miscited as “the Daubert factors.” They are not. The Daubert factors are testability, peer review and publication, known or potential error rate together with the existence of standards controlling the technique's operation, and general acceptance. The five above are the Advisory Committee's; the first quotes Daubert on remand, 43 F.3d 1311 (9th Cir. 1995), and the fourth draws on Kumho Tire's requirement that an expert employ “the same level of intellectual rigor that characterizes the practice of an expert in the relevant field.” For a non-scientific expert they are usually the more workable list — but they are not Supreme Court holdings.

Ipse dixit — the conclusion rests on the expert's authority

The governing language is from General Electric Co. v. Joiner, 522 U.S. 136, 146 (1997): nothing in Daubert or the Federal Rules requires a district court to admit opinion evidence 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.”

Two points about that passage are commonly lost. Joiner's holding is that abuse of discretion is the standard of appellate review for a decision to admit or exclude expert evidence; the Court reversed the Eleventh Circuit for applying a more searching review of exclusions. The analytical-gap language is the Court's explanation of why the district court had not abused its discretion on those facts — enormously influential and quoted in thousands of decisions, but reasoning rather than a test with elements. That matters, because briefs treating it as a test invite the response that no such test exists.

The failure also appears inside an otherwise accepted method. In Elcock v. Kmart Corp., 233 F.3d 734 (3d Cir. 2000), a vocational rehabilitation expert's stated methodology would logically have produced a finding of total disability, but he opined 50 to 60 percent and never explained the difference. That unexplained step is ipse dixit located in the middle of the analysis rather than at the end of it.

Unsupported extrapolation

Joiner is also the extrapolation case on its facts, and the facts are the useful part. The plaintiff's experts relied on studies of infant mice injected with massive doses of PCBs directly into the peritoneum, and on epidemiological studies whose own authors had declined to draw a causal inference, to support the opinion that workplace PCB exposure had promoted the plaintiff's small-cell lung cancer. The Court held the district court did not abuse its discretion in finding the gap between that data and that opinion too great.

The structure generalizes well beyond toxic tort. Extrapolation becomes a Rule 702 problem when the distance between the supporting material and the conclusion is one the expert cannot explain in terms the method supplies: a different dose, a different population, a different route of exposure, a different product or product generation, a different time period, or a source that reports association where the opinion asserts causation. The 2000 Committee note captures it as whether the expert “has unjustifiably extrapolated from an accepted premise to an unfounded conclusion.”

Trained experts extrapolate constantly, and Joiner says so. What the rule reaches is extrapolation the expert cannot justify from within the discipline.

Cherry-picked data

Selective reliance is now generally treated as a reliability problem rather than a credibility one, which is a meaningful shift. In In re Onglyza (Saxagliptin) and Kombiglyze XR Products Liability Litigation, 93 F.4th 339 (6th Cir. 2024), the Sixth Circuit affirmed exclusion of the MDL plaintiffs' general-causation expert whose opinion rested heavily on a single finding within one trial while discounting contrary evidence.

The most quotable treatment is Judge Agee's concurrence in EEOC v. Freeman, No. 13-2365 (4th Cir. 20 Feb. 2015), which affirmed exclusion of the agency's statistical expert and the resulting summary judgment. The concurrence described the reports as containing a “plethora” of “analytical fallacies” and a “mind-boggling number of errors,” and quoted the definition that matters: “‘[c]herry-picking' data produces a ‘misleadingly favorable result by looking only to “good” outcomes.'” The concurrence also catalogued the expert's history of problematic work across multiple circuits — the prior-exclusions line of cross-examination made visible in a published opinion.

The line worth respecting is that an expert who acknowledges a contrary study and explains why it does not change the analysis is doing normal work; the Advisory Committee said in terms that failing to read every existing study raises a question of weight. What is different is unexplained selection that drives the result.

A method built for the case

The first of the five additional reliability considerations asks whether the expert is “proposing to testify about matters growing naturally and directly out of research they have conducted independent of the litigation, or whether they have developed their opinions expressly for purposes of testifying.” It is the least technical of the arguments and often the most effective, because it does not require the court to evaluate the discipline — only to notice that the method has no existence outside the case.

Elcock is the paradigm. The Third Circuit described the vocational expert's combination of two accepted approaches as “nothing more than a hodgepodge of the Fields and Gamboa approaches” — a synthesis that was subjective, unreproducible, and unsupported by any showing of general acceptance. Neither component method was the problem; the assembly was.

Related and separately available is the fourth consideration, drawn from Kumho Tire: whether the expert “is being as careful as he would be in his regular professional work outside his paid litigation consulting.” Where a discipline has protocols, standards or accreditation requirements that the litigation work did not follow, that comparison is usually easier to establish than an attack on the underlying science.

Obvious alternative causes not ruled out

The third additional consideration asks whether the expert “has adequately accounted for obvious alternative explanations,” and it does the most work in causation cases. Where the stated method is differential diagnosis or differential etiology, the ruling-out step is not a supplement to the method — it is the method, which makes an unexplained failure to perform it a 702(d) failure of application rather than a debating point about thoroughness.

Tamraz v. Lincoln Electric Co., 620 F.3d 665 (6th Cir. 2010), is the standard citation. The Sixth Circuit reversed the admission of a treating neurologist's specific-causation opinion linking welding fumes to the plaintiff's parkinsonism, holding that the differential etiology had not been reliably performed and that the causal chain rested on speculation.

The argument is weakest when the alternative is theoretical, and strongest when it is in the record, the expert agrees it can produce the same outcome, and the transcript shows nothing done to exclude it.

Qualification mismatch — narrower than it is usually written

This argument is overstated in most practitioner writing, and the leading case is the reason. In Ralston v. Smith & Nephew Richards, Inc., 275 F.3d 965 (10th Cir. 2001), the Tenth Circuit affirmed the exclusion of an orthopedic surgeon's testimony on the adequacy of the warnings accompanying an intramedullary nail: her general orthopedic qualifications did not reach the specific subject, because she had conducted no research on intramedullary nailing and was not familiar with the particular device. The operative test, drawn from Compton v. Subaru of America, Inc., 82 F.3d 1513 (10th Cir. 1996), is whether the issue falls “within the reasonable confines” of the expert's subject area.

The same authority cuts the other way, and this is the part usually dropped. A mere lack of specialization goes to weight, so long as the expert stays within the reasonable confines of the subject area. The mismatch argument wins when the opinion lies outside those confines. It does not win because the opposing expert is board-certified in a narrower field, has published more, or is the more specialized of the two. That is a jury argument.

What makes the argument workable is that qualification is assessed opinion by opinion, not witness by witness. Gayton v. McCoy, 593 F.3d 610 (7th Cir. 2010), illustrates it: a physician's general medical qualifications did not extend to every causation opinion offered, and the court parsed them, admitting some and excluding others. Partial outcomes of that kind are far more common than wholesale exclusion.

Facts not in evidence, and assumptions the record contradicts

Elcock supplies the clearest published example. The economist's damages model rested on assumptions unsupported by — and in places contradicted by — the record: annual earnings assumed at $12,480 where the plaintiff had actually earned $5,774; total disability assumed against the vocational expert's own 50 to 60 percent figure; documented post-injury earnings ignored.

The structural point is about FRE 703, which is regularly invoked to defend exactly this defect. Rule 703 permits an expert to rely on inadmissible facts or data where “experts in the particular field would reasonably rely on those kinds of facts or data.” That gate is about the type of material — hearsay reports, industry data, statements of others — not about whether an assumed fact is true. Rule 703 does not license reliance on a fact the record contradicts.

Rule 703 carries a second constraint that runs the opposite way from the usual balance: inadmissible underlying material may be disclosed to the jury only if its probative value in helping the jury evaluate the opinion “substantially outweighs” its prejudicial effect. That is a reversed FRE 403 test, running against disclosure, and it is stated backwards more often than not. Meanwhile FRE 705 lets the cross-examiner compel disclosure of the same material — an asymmetry that is deliberate.

Overstatement, and the expert who did not do the work

Two grounds are worth naming that do not appear on the standard list.

Overstatement beyond what the method supports. The 2023 amendment to 702(d) was written for this: the Advisory Committee said each expert opinion “must stay within the bounds of what can be concluded from a reliable application of the expert's basis and methodology,” and that judicial gatekeeping is needed because jurors “may also lack the specialized knowledge to determine whether the conclusions of an expert go beyond what the expert's basis and methodology may reliably support.” The note is emphatic about forensic testimony: such experts “should avoid assertions of absolute or one hundred percent certainty—or to a reasonable degree of scientific certainty—if the methodology is subjective and thus potentially subject to error.” This is now an independent basis for limiting testimony even where the underlying method survives.

The expert who did not do the work. Where the analysis was performed by staff and the named expert cannot explain it, the opinion is exposed under 702(b) and (d), and separately under FRCP 26(a)(2)(B)'s requirement that the report be prepared and signed by the witness. Delegation is normal in technical work; an inability to describe what was delegated is not.

What does not get an opinion struck

The failures above have something in common: each identifies a step the expert's own stated method required and did not deliver. Arguments without that shape rarely succeed, and filing them costs credibility a later motion needs.

Disagreement between qualified experts is not a reliability defect. The Advisory Committee addressed it directly: where experts reach different conclusions on contested sets of facts, the Rule 104(a) standard “does not necessarily require exclusion of either side's experts,” because the jury can decide the disputed facts and which experts to credit. Nor is imperfection — the note says the standard “does not require perfection” and does not require a court “to nitpick an expert's opinion.” Nor is a minor computational error that does not drive the conclusion, an acknowledged contrary study the expert addresses, or a judgment call made inside an accepted method.

A prior exclusion is not preclusive either. Rule 702 rulings turn on the particular opinion, methodology and record, and no issue preclusion attaches to a ruling against a different party on a different record. It is powerful cross-examination; it is not a result.

And credentials are not method. An expert less distinguished than the opponent's, paid more, or testifying frequently for one side has a credibility problem, which FRE 104(e) preserves the right to put before the jury. That is not an admissibility problem.

Primary sources for this page: FRE 702 and the 2000 and 2023 Advisory Committee notes; General Electric Co. v. Joiner, 522 U.S. 136 (1997); EEOC v. Freeman, No. 13-2365 (4th Cir. 2015).

Frequently Asked Questions

Published 2026-08-12