Back in February of 2012, I wrote a blog post about a California Court of Appeal decision addressing the use of statistical sampling in class actions. The California Supreme Court recently granted review and affirmed the Court of Appeal’s decision that the trial court improperly allowed the case to be tried based on statistical evidence that failed to meet appropriate standards for reliability from a statistics point of view, and also refused to allow the defendant to present its individualized defenses. The California Supreme Court’s opinion in Duran v. U.S. Bank National Association, 2014 Cal. LEXIS 3758 (Cal. May 29, 2014) – a rare decision following a class action trial – has received a fair amount of attention. The key takeaways are that defendants cannot be deprived of litigating their defenses, even when they are individualized defenses, and that if sampling is used in some fashion it must done correctly and with a small margin of error.
Duran is an employment class action claiming that the defendant improperly classified certain employees as exempt based on a salesperson exemption, applicable to employees who spend 50% of the workday outside the office in sales activities. The class was relatively small (260 employees). The first phase of the case (liability) was initially tried by a method that involved randomly selecting 20 class members (one of whom refused to appear), adding the two named plaintiffs, and taking testimony from them. The court refused to allow the defendant to introduce other evidence from absent class members. The court concluded that the entire class was misclassified as exempt employees, although there was evidence that some were not. During the second phase of the trial (on damages), the court calculated overtime hours for the class by an extrapolation from the testimony of the class members, a calculation which the plaintiffs’ own expert testified had a 43% margin of error.
The California Supreme Court’s decision is lengthy. Here are the central points: First, the court emphasized the importance of a trial plan prior to deciding class certification:
If statistical evidence will comprise part of the proof on class action claims, the court should consider at the certification stage whether a trial plan has been developed to address its use. A trial plan describing the statistical proof a party anticipates will weigh in favor of granting class certification if it shows how individual issues can be managed at trial. Rather than accepting assurances that a statistical plan will eventually be developed, trial courts would be well advised to obtain such a plan before deciding to certify a class action. In any event, decertification must be ordered whenever a trial plan proves unworkable.
Id. at *48.
Second, the trial plan must allow the defendant to litigate its affirmative defenses, regardless of whether they are individual defenses:
We need not reach a sweeping conclusion as to whether or when sampling should be available as a tool for proving liability in a class action. It suffices to note that any class action trial plan, including those involving statistical methods of proof, must allow the defendant to litigate its affirmative defenses. If a defense depends upon questions individual to each class member, the statistical model must be designed to accommodate these case-specific deviations. If statistical methods are ultimately incompatible with the nature of the plaintiffs’ claims or the defendant’s defenses, resort to statistical proof may not be appropriate. Procedural innovation must conform to the substantive rights of the parties.
Id. at *68. The court also quoted the U.S. Supreme Court’s statement in Wal-Mart v. Dukes that “a class cannot be certified on the premises that [the defendant] will not be entitled to litigate its statutory defenses to individual claims.” Id. at *56.
Third, if statistical sampling is used, “[w]ith input from the parties’ experts, the court must determine that a chosen sample size is statistically appropriate and capable of producing valid results within a reasonable margin of error.” Id. at *74. This analysis can involve evaluating the extent of variability within the sample, ensuring that a sample is randomly selected, that there is no selection bias (the court found it improper to use the named plaintiffs in the sample, and to allow opt outs without re-sampling), and a large margin of error is improper. One problem that will arise here is that in any class action you will have a significant portion of class members who decline to participate, and that can create selection bias in any sample.
From the defense perspective, some key takeaways here, as I see them, are: (1) try to force the plaintiff to present a detailed trial plan at the time of class certification; (2) emphasize your affirmative defenses and insist on presenting them; and (3) in challenging sampling, focus on variability and selection bias.