Daubert’s Error Rate

In Daubert, the Supreme Court came to the realization that expert witness opinion testimony was allowed under the governing statute, Federal Rule of Evidence 702, only when that witness’s “scientific, technical, or other specialized knowledge” would help the fact finder. Knowledge clearly connotes epistemic warrant, and some of the Court’s “factors” speak directly to this warrant, such as whether the claim has been tested, and whether the opinion has an acceptable rate of error. The Court, however, continued to allow some proxies for that warrant, in the form of “general acceptance,” or “peer review.”

The “rate of error” factor has befuddled some courts in their attempt to apply the statutory requirements of Rule 702, especially when statistical evidence is involved. Some litigants have tried to suggest that a statistically significant result suffices alone to meet the demands of Rule 702, but this argument is clearly wrong. See, e.g., United States v. Vitek Supply Corp., 144 F.3d 476, 480, 485–86 (7th Cir. 1998) (stating that the purpose of the inquiry into rate of error is to determine whether tests are “accurate and reliable”) (emphasis added). See also Judicial Control of the Rate of Error in Expert Witness Testimony” (May 28, 2015). The magnitude of tolerable actual or potential error rate remains, however, a judicial mystery[1].

Sir Austin Bradford Hill described ruling out bias, confounding, and chance (or random error) as essential prerequisites to considering his nine factors used to assess whether an association is causal:

“Disregarding then any such problem in semantics we have this situation. Our observations reveal an association between two variables, perfectly clear-cut and beyond what we would care to attribute to the play of chance. What aspects of that association should we especially consider before deciding that the most likely interpretation of it is causation.”

Austin Bradford Hill, “The Environment and Disease: Association or Causation?” 58 Proc. Royal Soc’y Med. 295, 295 (1965). The better reasoned cases agree. See, e.g., Frischhertz v. SmithKline Beecham Corp., 2012 U.S. Dist. LEXIS 181507, *6 (E.D.La. 2012) (“The Bradford-Hill criteria can only be applied after a statistically significant association has been identified.”) (citing and quoting among other sources, Federal Judicial Center, Reference Manual on Scientific Evidence, 599 & n.141 (3d. ed. 2011)).

Citing the dictum in Matrixx Initiatives[2] as though it were a holding is not only ethically dubious, but also ignores the legal and judicial context of the Court’s statements[3]. There are, after all, some circumstances such as cases of death by blunt-force trauma, or bullet wounds, when epidemiological and statistical evidence is not needed. The Court did not purport to speak to all causation assessments; nor did it claim that it was addressing only instances in which there were “expected cases,” and “base-line risks,” in diseases that have an accepted occurrence and incidence among unexposed persons. It is, of course, in exactly those cases that statistical consideration of bias, confounding, and chance are essential before Bradford Hill’s factors can be parsed.

Lord Rutherford[4] is often quoted as having said that “[i]f your experiment needs statistics, you ought to have done a better experiment.” Today, physics and chemistry have dropped their haughty disdain for statistics in the face of their recognition that some processes can be understood only as stochastic and rate driven. In biology, we are a long way from being able to describe the most common disease outcomes as mechanistic genetic or epigenetic events. Statistical analyses, with considerations of random and systematic error, will be with us for a long time, whether the federal judiciary acknowledges this fact or not.

*        *        *        *        *        *        *        *        *        *        *         *        *       

Cases Discussing Error Rates in Rule 702 Decisions


Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 593 (1993) (specifying the “the known or potential rate of error” as one of several factors in assessing the scientific reliability or validity of proffered expert witness’s opinion)

Kumho Tire Co. v. Carmichael, 526 U.S. 137, 151 (1999) (suggesting that reliability in the form of a known and an acceptable error rate is an important consideration for admissibility)

US Court of Appeals


United States v. Shea, 957 F. Supp. 331, 334–45 (D.N.H. 1997) (rejecting criminal defendant’s objection to government witness’s providing separate match and error probability rates)


Rabozzi v. Bombardier, Inc., No. 5:03-CV-1397 (NAM/DEP), 2007 U.S. Dist. LEXIS 21724, at *7, *8, *20 (N.D.N.Y. Mar. 27, 2007) (excluding testimony from civil engineer about boat design, in part because witness failed to provide rate of error)

Sorto-Romero v. Delta Int’l Mach. Corp., No. 05-CV-5172 (SJF) (AKT), 2007 U.S. Dist. LEXIS 71588, at *22–23 (E.D.N.Y. Sept. 24, 2007) (excluding engineering opinion that defective wood-carving tool caused injury because of lack of error rate)

In re Ephedra Products Liability Litigation, 393 F. Supp. 2d 181, 184 (S.D.N.Y. 2005) (confusing assessment of random error with probability that statistical estimate of true risk ratio was correct)

Roane v. Greenwich Swim Comm., 330 F. Supp. 2d 306, 309, 319 (S.D.N.Y. 2004) (excluding mechanical engineer, in part because witness failed to provide rate of error)

Nook v. Long Island R.R., 190 F. Supp. 2d 639, 641–42 (S.D.N.Y. 2002) (excluding industrial hygienist’s opinion in part because witness was unable to provide a known rate of error).

United States v. Towns, 19 F. Supp. 2d 67, 70–72 (W.D.N.Y. 1998) (permitting clinical psychologist to opine about defendant’s mens rea and claimed mental illness causing his attempted bank robbery, in part because the proffer of opinion maintained that the psychologist would provide an error rate)  

Meyers v. Arcudi, 947 F. Supp. 581 (D. Conn. 1996) (excluding polygraph in civil action in part because of error rate)


United States v. Ewell, 252 F. Supp. 2d 104, 113–14 (D.N.J. 2003) (rejecting criminal defendant’s objection to government’s failure to quantify laboratory error rate)

Soldo v. Sandoz Pharmaceuticals Corp., 244 F. Supp. 2d 434, 568 (W.D. Pa. 2003) (excluding plaintiffs’ expert witnesses in part because court, and court-appointed expert witnesses, were unable to determine error rate).

Pharmacia Corp. v. Alcon Labs., Inc., 201 F. Supp. 2d 335, 360 (D.N.J. 2002) (excluding ; error too high).


United States v. Moreland, 437 F.3d 424, 427–28, 430–31 (4th Cir. 2006) (affirming district court’s allowance of forensic chemist’s testimony that could not provide error rate because reviews of witness’s work found it to be free of error)

Buckman v. Bombardier Corp., 893 F. Supp. 547, 556–57 (E.D.N.C. 1995) (ruling that an expert witness may opine about comparisons between boat engines in rough water but only as a lay witness, because the comparison tests were unreliable, with a high estimated rate of error)


Albert v. Jordan, Nos. 05CV516, 05CV517, 05CV518, 05CV519, 2007 U.S. Dist. LEXIS 92025, at *2–3 (W.D. La. Dec. 14, 2007) (allowing testimony of vocational rehabilitation expert witness, over objection, because witness provided “reliable” information, with known rate of error)


United States v. Leblanc, 45 F. App’x 393, 398, 400 (6th Cir. 2002) (affirming exclusion of child psychologist, whose testimony about children’s susceptibility to coercive interrogation was based upon “‘soft science’ . . . in which ‘error is . . . rampant’.” (quoting the district court))

United States v. Sullivan, 246 F. Supp. 2d 696, 698–99 (E.D. Ky. 2003) (admitting expert witness’s opinion on the unreliability of eyewitness identification; confusing error rate of witness’s opinion with accuracy of observations made based upon order of presentation of photographs of suspect)


United States v. Vitek Supply Corp., 144 F.3d 476, 480, 485–86 (7th Cir. 1998) (affirming denial of defendant’s Rule 702 challenge based in part upon error rates; the purpose of the inquiry into rate of error is to determine whether tests are “accurate and reliable”; here the government’s expert witnesses used adequate controls and replication to ensure an acceptably low rate of error)

Phillips v. Raymond Corp., 364 F. Supp. 2d 730, 732–33, 740-41 (N.D. Ill. 2005) (excluding biomechanics expert witness who had not reliably tested his claims in a way to produce an accurate rate of error)


Bone Shirt v. Hazeltine, 461 F.3d 1011, 1020 (8th Cir. 2006) (affirming district court’s ruling to admit testimony of expert witness’s regression analysis in vote redistricting case); see id. at 1026 (Gruender, J., concurring) (expressing concern with the questioned testimony’s potential rate of error because it is “difficult to weigh this factor in Daubert’s analysis if ‘the effect of that error is unknown’.” (quoting court below, Bone Shirt v. Hazeltine, 336 F. Supp. 2d 976, 1002 (D.S.D. 2004))

United States v. Beasley, 102 F.3d 1440, 1444, 1446–48 (8th Cir. 1996) (confusing random error with general error rate) (affirming admissibility of expert witness testimony based upon DNA testing, because such testing followed acceptable standards in testing for contamination and “double reading”)


United States v. Chischilly, 30 F.3d 1144, 1148, 1152, 1154–55 (9th Cir. 1994) (affirming admissibility of testimony based upon DNA match in sex crime, noting that although error rate of error was unquantified, the government had made a sufficient showing of rarity of false positives to support an inference of low error rate)

Cascade Yarns, Inc. v. Knitting Fever, Inc., No. C10–861RSM, 2012 WL 5194085, at *7 (W.D. Wash. Oct. 18. 2012) (excluding expert witness opinion because error rate was too high)

United States v. Microtek Int’l Dev. Sys. Div., Inc., No. 99-298-KI, 2000 U.S. Dist. LEXIS 2771, at *2, *10–13, *15 (D. Or. Mar. 10, 2000) (excluding polygraph data based upon showing that claimed error rate came from highly controlled situations, and that “real world” situations led to much higher error (10%) false positive error rates)


Miller v. Pfizer, Inc., 356 F.3d 1326, 1330, 1334 (10th Cir. 2004) (affirming exclusion of plaintiffs’ expert witness, Dr. David Healy, based upon district court’s findings, made with the assistance of court-appointed expert witnesses, that Healy’s opinion was based upon studies that lacked sufficient sample size, adequate controls, and freedom from study bias, and thus prone to unacceptable error rate)


Quiet Tech. DC-8, Inc. v. Hurel-Duboi U.K., Ltd., 326 F.3d 1333, 1343–45 (11th Cir. 2003) (affirming trial court’s admission of defendant’s aerospace engineer’s testimony, when the lower court had found that the error rate involved was “relatively low”; rejecting plaintiff’s argument that the witness had entered data incorrectly on ground that the asserted error would not affect the validity of the witness’s opinions)

Wright v. Case Corp., No. 1:03-CV-1618-JEC, 2006 U.S. Dist. LEXIS 7683, at *14 (N.D. Ga. Feb. 1, 2006) (granting defendant’s motion to exclude plaintiff’s mechanical engineering expert, because the expert’s alternative designs for the seat safety bar were not reliable due to potential feasibility issues, and because the associated error rate was therefore unquantifiable but potentially very high)

Benkwith v. Matrixx Initiatives, Inc., 467 F. Supp. 2d 1316, 1326, 1330, 1332 (M.D. Ala. 2006) (granting defendant’s motion to exclude testimony of an expert in the field of epidemiology regarding Zicam nasal spray’s causing plaintiff’s anosmia, because the opinions had not been tested and a rate of error could not be provided).


Ambrosini v. Upjohn Co., No. 84-3483 (NHJ), 1995 U.S. Dist. LEXIS 21318, at *16, *22–24 (D.D.C. Oct. 18, 1995) (finding that plaintiff’s teratology expert was not permitted to testify, because the methodology used was found to be unreliable and could not yield an accurate error rate)

[1] Jed S. Rakoff, “Science and the Law: Uncomfortable Bedfellows,” 38 Seton Hall L. Rev. 1379, 1382–83 (2008) (observing that an error rate of 13 percent in polygraph interpretation would likely be insufficiently reliable to support admissibility of testimony based upon polygraph results).

[2] Matrixx Initiatives, Inc. v. Siracusano, 131 S. Ct. 1309, 1319 (2011) (suggesting that courts “frequently permit expert testimony on causation based on evidence other than statistical significance”).

[3] See, e.g., WLF Legal Backgrounder on Matrixx Initiatives (June 20, 2011); “The Matrixx – A Comedy of Errors”; Matrixx Unloaded (Mar. 29, 2011)”; “The Matrixx Oversold” (April 4, 2011); “De-Zincing the Matrixx.”

[4] Ernest Rutherford, a British chemist, investigated radioactivity. He won the Nobel Prize in chemistry, in 1908.

Print Friendly, PDF & Email

Comments are closed.