On May 29, 2014, the California Supreme Court issued its much-anticipated opinion in Duran v. U.S. Bank, vacating a $15 million judgment in a wage-hour class action where the court found liability based on a flawed statistical sampling. Although the Court did not completely reject the use of statistical sampling to establish class-wide liability, it significantly limited the circumstances in which sampling is available. Most importantly, the Court held that sampling may not be used when it deprives a defendant of its due process right to present affirmative defenses as to all class members. It also reaffirmed plaintiffs’ burden of establishing manageability in order to certify a class, as well as the obligation of courts to decertify classes that prove to be unmanageable.
The plaintiffs in Duran alleged that U.S. Bank had misclassified certain employees as “outside salespersons” who are exempt from overtime and other requirements because they spend more than 50% of their working time making sales outside the office. Although declarations from putative class members showed that some class members met the 50% requirement and others did not, the trial court certified a class. It then adopted a trial plan in which a purportedly random sample of 20 class members plus two of the named plaintiffs would testify at trial, and both liability and damages would be extrapolated to the entire class from findings based on the sample group. At trial, the court excluded all evidence that class members outside the sample group were in fact exempt, and concluded based only on the sample evidence that all class members had been misclassified. It also set damages by extrapolating from the sample despite a 43% margin of error, awarding approximately $15 million to the class.
Affirming the Court of Appeal, which had reversed the trial court, the California Supreme Court agreed that the class should have been decertified and ordered a new trial. The Court confirmed that a defendant has a due process right to “litigate its statutory defenses to individual claims.” Thus, “any trial must allow for the litigation of affirmative defenses, even in a class action case where the defense touches upon individual issues.” If statistical sampling is to be used at all for proving liability in a class action, the Court held, it must comport with due process and use sound methodologies. In particular, among other things: (a) the sample size must be “sufficiently large to provide reliable information about the larger group,” (b) the sample must be random and free of selection bias, and (c) the approach must yield results within a reasonable margin of error. Furthermore, trial courts should evaluate any proposals for statistical sampling before certifying a class, and they must decertify classes that prove unmanageable.
Although Duran unfortunately did not completely reject the use of statistical sampling to establish classwide liability, it does make it significantly more difficult for plaintiffs to use that approach. It also will help defendants opposing class certification, as it reaffirms the requirement that class litigation be manageable, which too often is ignored in lower courts.