Do litigants in civil and criminal proceedings need statistical expert witnesses to present statistical analyses? Or, can lawyers take the data that are in evidence, and present their own statistical analyses?
Surely, lawyers could add figures to arrive at a sum, which is relevant to the issues in dispute. Some lawyers and judges might be able to take model assumptions, and compare two means or two proportions, to show that they the statistics did not likely come from the same populations. Indeed, some lawyers may be able to do such analyses better than some expert witnesses, but this begs the question: is it legally permissible?
In In re Pfizer Inc. Securities Litig., 584 F.Supp. 2d 621 (S.D.N.Y. 2008), defendant Pfizer filed a motion to dismiss a securities class action complaint. The court found that Pfizer’s motion would require it to interpret statistical significance, and that it could not accept the parties’ non-expert assertions of the meaning of the concept; nor could the court take judicial notice of the meaning:
“The Court declines to take judicial notice of the meaning of statistical significance or of the data interpretations proffered by Defendants in the context of this motion practice. Rule 201 of the Federal Rules of Evidence provides that courts may only take notice of facts ‘either (1) generally known . . . or (2) capable of accurate and ready determination by resort to sources whose accuracy cannot reasonably be questioned’. Fed. R. Evid. 201(b). While statistical significance may have certain characteristics capable of general abstraction, it is far beyond the scope of Rule 201 to accept as fact the particular definitions of statistical significance proffered by Defendants as either facts generally known or as drawn from sources whose accuracy cannot reasonably be questioned. It is one thing to take notice of the fact that an author has written that 5% is the threshold for statistical significance. It is quite another thing entirely to use that 5% figure as a basis for rejecting the significance of complicated medical studies.”
Id. at 634. Similarly, the court refused to look at specific studies and conclude that they failed to find a statistically significant association between Celebrex and cardiovascular adverse events:
“A motion to dismiss a complaint is not an appropriate vehicle for determination as to the weight of the evidence, expert or otherwise. Clearly, the Court cannot take judicial notice that the three studies show a lack of any statistically significant link between Celebrex/Bextra and adverse cardiovascular events because that supposed fact is neither generally known nor capable of accurate and ready determination by reference to unquestionably accurate sources. Moreover, the Court cannot determine as a matter of law whether such links were statistically insignificant because statistical significance is a question of fact.”
Id. at 635.
In Bristol-Myers Squibb v. AIU Insurance Co., et al., Cause No. A-145,672, Jefferson County, 58th Judicial District, Texas, plaintiff’s counsel made a Batson challenge to the defendants’ exercise of peremptory challenges. See Daily Transcript in (May 13, 1997). Not having expected the defense counsel to exercise their peremptory challenges in an apparently discriminatory fashion, the plaintiff’s counsel did not have a statistician ready to analyze the pattern of challenges. One of the plaintiff’s counsel presented the analysis in his oral argument to the court. The venire panel was made up of 49 persons, 18 black, and 31 white. The defense exercised 6 of their 7 peremptory challenges to black veniremen. Based upon these numbers, plaintiff’s counsel presented a calculation of the probability that defense counsel would have exercised their challenges in such an extreme fashion if they made their choices independent of race. The defense objected to plaintiff’s counsel’s calculations, but the trial court overruled the objection and noted that the laws of probability were subject to judicial notice. Id. at 828-30.
The Texas trial court found a prima facie case of discrimination, and permitted plaintiff’s counsel to cross-examine defense counsel about their peremptory and selection decisions. Id. at 858. The case settled shortly afterwards. See also Andrew T. Berry, “Selecting Jurors,” 24 Litigation 8, 9 (Fall 1997)(“For example, in a recent (unreported) large civil case in the Southwest, the defendants successfully defeated a Batson challenge to their use of 85 percent of their peremptory challenges against protected class members. The successful defense? That the two dozen eminent counsel …, given less than a quarter-hour to exercise their peremptories were simply too disorganized to have struck jurors in violation of Batson.”)
In the welding fume MDL 1535, the plaintiffs persisted in challenges to a particular industry-funded, published epidemiologic study, which reported findings of no increased risks for Parkinson’s disease and parkinsonism among non-shipyard Danish welders. Jon Fryzek, J. Hansen, S. Cohen, J. Bonde, et al., “A cohort study of Parkinson’s disease and other neurodegenerative disorders in Danish welders,” 47 J. Occup. & Envt’l Med. 466 (2005). Plaintiffs’ counsel went to the extreme of traveling to Denmark, with one of their expert witnesses in tow, to analyze the underlying data for this study. Upon returning to the United States, the plaintiffs moved to bar reliance upon the Fryzek study, on the theory that the statistical analysis concerning the article’s finding of no statistically significant difference in the age of onset was incorrect. In support of their argument, one of the plaintiff’s counsel, a law professor who was assisting plaintiffs in the welding fume litigation, submitted an affidavit in support of the motion in limine to bar defense witness’s testimony. See Affidavit of Theodore Eisenberg, in In re Welding Fume Products Liability Litigation, Case No.: 1:03-cv-17000, MDL No. 1535, Document 1862 Filed 08/07/2006.
Eisenberg’s affidavit reported analyses of the Danish data, apparently based upon work done by an unnamed “programmer” at the Danish Cancer Society. The affidavit included truncated computer program output, without identification of the statistical tests, or of the statistical software, used. Eisenberg interpreted the p-value result of the attached statistical analysis to show that there was a statistically significant difference in the age of onset of Parkinson’s disease between welders and non-welders.
The defense opposed the motion on grounds that Eisenberg’s affidavit was an ethically impermissible attempt by a lawyer in the case to present an expert witness opinion. The defense also countered substantively with an affidavit from one of its expert witnesses, who analyzed the affidavit and realized that Eisenberg and the anonymous programmer had not presented the complete software output from their analyses, and that they had used a different test from that used in the published paper. Eisenberg’s affidavit therefore had not identified an error in the published paper. Declaration of Timothy L. Lash (Sept. 11, 2006), filed in In re Welding Fume Products Liability Litigation, Case No.: 1:03-cv-17000, MDL No. 1535. The trial court denied the plaintiff’s motion to bar reliance upon the Fryzek study, without comment on the propriety of Eisenberg’s affidavit.
The MTBE mass tort litigation gave rise a peculiar instance in which a trial court held that a real estate value appraiser had departed from the level of intellectual rigor used in assessing property value changes, claimed to have resulted from a gas station’s pollution of the ground water in a small town in Orange County, New York. The witness opined that the plaintiffs’ property suffered a 15% decline in market value, but he failed to identify the methods he used to arrive at his opinion. In re Methyl Tertiary Butyl Ether (“MTBE”) Prods. Liab. Litig., 2008 U.S. Dist. LEXIS 44216 (S.D.N.Y. June 4, 2008)(Scheindlin, J.). The expert witness did explain that there were so few sales in the affected town that he could not use regression analysis, and that it was thus necessary to look at “trend data on sales by sub-markets, sales/list price analysis and days on the market comparisons.” Id. at *5. Even so, the trial court could not otherwise discern what method the witness did use:
“In this case, I am unable to discern any method — much less a reliable method — that Langer used to reach his conclusion that the value of plaintiffs’ property decreased by fifteen percent because of MTBE contamination. Rather, Langer has merely compiled market data and then offered his conclusions, yet he has failed to explain the relationship between the two.”
Id. at *11.
Although the expert witness’s departure from the professional standard of care rendered his opinion inadmissible, the trial court decided that the would-be expert witness could still testify as a fact witness to the facts that he had collected about sales trends in the affected community and elsewhere. According to the court, the statistics gathered by this witness were relevant, and the plaintiffs’ counsel could argue plausible inferences to the jury, from the sales figures. Id. at *16-17. The court thus remarkably permitted the plaintiffs’ counsel to provide the statistical analysis that his designated expert witness had failed to give in a legally reliably form.
