Mass Tort Damages: A Statistical Alternative
by Jerome M. Staller, Ph.D and Lawrence M. Santucci, M.A.
Traditionally, courts have been reluctant to consider the use of statistical tools such as sampling to determine damages in class-action litigation and other matters involving large groups of plaintiffs. Arguments against the practice include the fact that it seemingly flies in the face of the Seventh Amendment, and that damages, by their very nature, are peculiar and specific to each individual plaintiff.
However, as Federal District Judge Jack B. Weinstein observed in a ruling in a huge consumer-fraud action against Big Tobacco, “mass-tort cases have stripped the ability of the common law, with its relatively rigid adherence to precedent, to fashion remedies that adequately redress the harms of modern technological society.” In re: Simon II Litigation, No. 00-CVBB5332, 2002 U.S. Dist Lexis 19773 (E.D.N.Y. 2002).
In an exhaustive analysis of the role of statistical evidence in mass torts, Weinstein wrote: “Requiring individual proof as to each claim would unnecessarily intrude upon the lives of hundreds of thousands of people. Examining each grain of sand is too burdensome in a survey of a beach.” He concluded that consolidation and an aggregate determination of damages may be the only realistic course available.
To that end, the technique known as statistical sampling is particularly apt. Sampling can be applied to a wide variety of questions arising in litigation, and, if properly applied, can yield astonishingly accurate results.
An early example of the accuracy of sampling techniques in litigation took place in Los Angles in 1955: Sears, Roebuck & Company determined, through a sampling of sales records, that it had overpaid the city of Inglewood $27,000 in sales taxes. The city disagreed. In the ensuing litigation in Los Angeles Superior Court, Sears’ expert statistician determined, via a sampling of 33 days’ worth of sales records out of an 826-day period, that Sears had overpaid sales taxes to the tune of $28,250, with a 95% confidence interval that the overpayment was between $24,000 and $32,400.
The court refused to accept the statistical evidence and ordered an audit of sales records for all 826 days in the disputed period. The laborious and expensive audit revealed that the overpayment was $26,750.22, a few dollars short of the $27,000 Sears originally claimed. (See “The Admissibility of Sample Data Into a Court of Law: A Case History,” Sprowls, C., 4 UCLA LR 222 (1957)).
Since then, several courts have sought to apply quantitative analysis to damages questions in mass-tort litigation . In an excellent article on the topic, authors Laurens Walker and John Monahan, both professors of law at the University of Virginia, examined three such cases. “Sampling Damages,” 83 Iowa LR 545 (1998).
In one case, a mass-tort matter involving 2,298 asbestos claimants, the court decided to determine damages by trying 160 cases picked randomly from the pool of claimants, which had been divided in five subgroups based on disease categories. Cimino v. Raymark Industries Inc., 751 F. Supp. 649 (E.D. Tex 1990), (rev’d. 1998 U.S. App. Lexis 20096 (5th Cir. 1998)).
Two juries heard medical and damages testimony and returned individual verdicts in each case in the pool of sample cases; the court then assigned damages to the remaining claimants based on average verdicts from the pool in each of the five disease categories (total damages amounted to more than $1 billion). On appeal, the Fifth Circuit ruled that the methodology employed in Cimino was contrary to Texas law and also violated the defendants’ Seventh Amendment rights.
Statistical analysis of damages fared better in Estate of Marcos Human Rights Litigation, 910 F. Supp. 1460 (D. Haw. 1995) (aff’d in Hilao v. Estate of Marcos, 103 F.3d 767 (9th Cir. 1996)). This massive litigation addressed claims of human-rights abuses by the regime of Philippine President Ferdinand Marcos that occurred between 1972 and 1986.
A class was certified consisting of victims and families of victims, subdivided into three categories: victims of torture, families of those who had been summarily executed by the regime, and families of those who had “disappeared” after being taken into custody by the regime. Proofs of claims were filed by 10,059 individuals.
The court then heard testimony from a statistician who devised a plan to determine the total amount of compensatory damages using statistical sampling techniques. Under the plan, 137 claimants were randomly selected using a computer program. A special master appointed by the court was then dispatched to the Philippines to depose the 137 claimants in the sample.
The special master made recommendations as to compensatory damages for each individual member of the sample group, and then made recommendations as to the total compensatory damages amounts due each of the three sub categories of claimants. The trial judge then told the jury that it could accept, reject or modify the special master’s recommendations and make its own decision as to individual damages for the sample-group members and the total compensatory damages due to the class.
The jury, after five days’ deliberations, returned the individual verdicts and a total verdict of over $766 million. The scheme withstood appellate scrutiny.
In 1997, 3,000 residents of a Houston subdivision sought compensation for exposure to hazardous chemicals. The court proposed a plan to determine damages via random sampling of 30 claimants. Under the plan, 15 sample plaintiffs were to be named by the plaintiffs’ side and 15 sample plaintiffs were to be named by the defendant, Chevron.
Chevron (quite correctly, from a statistical point of view) objected on the grounds that the selection process was not random and therefore would yield biased results. The circuit-court panel agreed. The panel allowed the 30 sample trials to commence, but limited the verdicts to those individual cases, ruling that they could not be used as the basis for determining damages to the other plaintiffs.
The sample, not being randomly selected, would not reliably represent those remaining claimants. In re Chevron U.S.A. Inc., 109 F.3d 1016 (5th Cir. 1997).
One area where many procedural kinks in the quantitative determination of damages have been worked out is wage-and-hour claims brought under the federal Fair Labor Standards Act. For a discussion of procedure and sample-size requirements in FLSA damages matters, see Reich v. Southern Maryland Hospital et al., 43 F. 3d 949 (4th Cir. 1995); 1995 U.S. App. LEXIS 697.
While an ideal procedural approach to the quantitative determination of damages in other types mass torts has yet to be developed, it seems inevitable, given the amount of mass torts and the proven accuracy of sampling techniques, that the statistical determination of mass-tort damages will eventually become the norm.
Jerome M. Staller, Ph.D., an economist, is president of the Center for Forensic Economic Studies. Lawrence M. Santucci, M.A. is a senior economist / statistician at the Center. This article is reprinted with permission from the Nov / Dec 2004 issue of Product Liability Law & Strategy © 2004 ALM Properties Inc. All rights reserved. Further duplication without permission is prohibited.