Editor’s Note: This blog post was originally posted on ClassActionBlawg.com. It is reproduced with permissions.
The California Supreme Court issued its long-awaited decision in Duran v. U.S. Bank National Association yesterday, addressing the use of statistical sampling as a way of evaluating aggregate liability and damages in a class action. Although Duran is a wage and hour case, its analysis is pertinent to the use of statistical evidence in a variety of other class action contexts.
In the opening line of his majority opinion, Justice Corrigan referred to Duran “an exceedingly rare beast” because it was a wage and hour class action that had proceeded all the way through trial to verdict. In the trial court, the plaintiff had presented testimony from statistician Richard Drogin, who had also notably served as an expert for the plaintiffs in Walmart Stores Inc. v. Dukes. Drogin proposed a random sampling analysis that purported to estimate the percentage of the defendant’s employees that had been misclassified for purposes of entitlement to overtime pay. The trial court did not rely on Drogin’s analysis but instead came up with its own sampling approach, which involved pulling the names of 20 class members, hearing testimony from these witnesses along with the named plaintiffs, and then extrapolating the court’s factual findings across the entire class in order to determine the defendant’s liability.
The supreme court affirmed a decision by the Court of Appeal holding that this sampling approach violated due process and was a manifest abuse of discretion. Generally, there were two independent reasons for the supreme court’s conclusion: 1) the use of random sampling deprived the defendant of the opportunity to present individualized evidence supporting its defenses to the claims; and 2) the sampling method adopted by the court was inherently flawed and unreliable.
Without categorically rejecting the use of statistics as a tool in managing class action litigation, the supreme court identified numerous conceptual limitations on its use. First, “[s]tatistical methods cannot entirely substitute for common proof . . . . There must be some glue that binds class members together apart from statistical evidence.” So, while statistics may serve as circumstantial evidence to support a common issue–such as the existence of centralized policy or practice, they may not be used as a substitute for establishing commonality or for avoiding individualized determination of individual issues–such as by generalizing effects of a given policy or practice on large groups of claimants where the effects vary in actuality.
Second, a trial court cannot utilize statistical evidence in a way that prevents the individual adjudication of individual defenses. Although courts are encouraged to develop innovative procedures in managing individual issues, a court cannot ignore individual issues altogether or prevent them from being decided on an individual basis.
Third, if statistical evidence is to be used as part of a litigation plan for managing complex class action, the methods to be employed should be presented, evaluated, and scrutinized at the class certification stage. The court should not simply assume that statistical methods will permit class treatment and certify the class based on this hypothetical possibility.
Fourth, the court must ensure that the statistical method to be employed has to be reliable, based on statistically valid data, and not prone to a high margin of error. In other words, junk science or ad hoc, rough justice are not enough.
The Duran opinion is worthy of careful study for anyone considering the use of statistics in class certification proceedings, both in the wage and hour context and in class actions more generally. It also provides a colorful illustration of the due process and manageability problems posed by the “trial by formula” approach to class actions that the United States Supreme Court criticized in Dukes.