Cal. Supreme Court Holds That Statistical Sampling In Class Actions May Deprive Defendants Of Due Process

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 on the ground that it was based on a flawed statistical sampling methodology. While the Court did not foreclose the possibility of using statistical sampling to establish class-wide liability, it made clear that (1) a trial plan that includes the proposed sampling should be presented to the trial court before class certification, (2) the proposed sampling should be statistically reliable, and (3) the trial plan should not deprive defendants of their due process right to present affirmative defenses. 

Duran is significant because, in recognizing the due process concerns raised by the use of representative evidence, and in requiring trial courts to meaningfully address those concerns early in litigation, it provides welcome ammunition for employers in opposing class certification.

Lower Court Proceedings

Plaintiffs filed a class action alleging that their employer, U.S. Bank, misclassified its Business Banking Officers (“BBO”) as exempt outside sales employees. After certification, the parties submitted their respective trial plans, crafted with the aid of their experts. Over U.S. Bank’s objections, the trial court adopted its own trial plan, under which a purportedly random sample of twenty class members—plus two of the named plaintiffs—would testify at trial, and the liability and damages findings based on the sample group would be extrapolated to the entire class.

After the trial court denied U.S. Bank’s decertification motion, it held a bench trial on U.S. Bank’s exemption defense. During the liability phase, the trial court excluded all evidence concerning BBOs who were not part of the sample group, including U.S. Bank’s evidence showing that some class members were properly classified as exempt. Based primarily on the sample group’s testimony, the court found that the entire class of 260 BBOs had been misclassified.

During the damages phase, the trial court adopted the determination of plaintiffs’ expert that class members worked on average 11.87 hours of overtime per week, subject to a 43% margin of error—meaning the actual amount of overtime worked by each BBO could range from 6.7 hours to almost 17 hours per week. Based on that extrapolation, the court entered judgment against U.S. Bank in the amount of approximately $15 million.

The Court of Appeal reversed, holding that the trial court’s reliance on flawed and unreliable statistical sampling to extrapolate class-wide liability denied U.S. Bank its right to litigate affirmative defenses, and that the high margin of error underlying the damages calculations implicated due process concerns. Additionally, the Court of Appeal held, the trial court abused its discretion in denying U.S. Bank’s decertification motion and ordered the class decertified.

The Supreme Court Decision

The California Supreme Court, in a unanimous decision, affirmed the Court of Appeal’s judgment in its entirety and ordered a new trial. In so doing, the Supreme Court articulated several principles that are likely to have a significant impact on certification and trial proceedings in all class actions, particularly those in the wage-hour arena.

First, and perhaps most significantly, the Supreme Court recognized a defendant’s due process right to “litigate its statutory defenses to individual claims,” a proposition on which there had been disagreement in the lower courts. Thus, “any trial must allow for the litigation of affirmative defenses, even in a class action case where the defense touches upon individual issues.” 

Second, while the Supreme Court was careful not to reach a sweeping conclusion regarding whether or when statistical sampling should be available as a tool for proving liability in a class action, it did set forth some concrete guidelines. As an initial matter, any trial plan involving statistical proof must allow the defendant to litigate relevant affirmative defenses, even when they turn on individualized questions, and if it cannot do so, then the statistical proof may not be appropriate. Moreover, the trial plan must employ valid statistical methodology, which means, 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 sample must yield results within a reasonable margin of error. Further, the defendant “must be given a chance to impeach that [statistical] model or otherwise show that its liability is reduced because some plaintiffs were properly classified as exempt.”

Third, the Supreme Court advised lower courts to consider at the certification stage whether a trial plan has been developed to address the use of statistical evidence, rather than “accepting assurances that [one] will eventually be developed.” A trial plan must show how individual issues can be managed at trial, and if it proves “unworkable,” the class must be decertified.

Turning to the facts before it, the Supreme Court held that the lower court’s trial plan met none of these basic requirements. Among other things, the plan deprived U.S. Bank of its right to litigate its affirmative defenses by excluding relevant evidence relating to BBOs outside the sample group, and by extrapolating liability based on a flawed statistical model. That model, the Supreme Court held, was fatally flawed because, among other things, the 22-member sample group was too small relative to the 260-member class; the supposed randomness of the sample group was undermined by the inclusion of the named plaintiffs and the later exclusion of others who had opted out, were replaced, or were unavailable. As a result, the sample was “biased in plaintiffs’ favor.”  The Court also found the 43% margin of error to be “intolerably high,” potentially yielding a judgment twice the size of U.S. Bank’s actual liability.

What Duran Means For Employers

While the Supreme Court stopped short of establishing a bright-line rule that statistical sampling cannot be used to prove class-wide liability, Duran nonetheless makes it clear that class counsel face an uphill battle if they wish to rely on statistical evidence. Any proposed statistical methodology must allow a defendant to litigate its affirmative defenses. And, in cases involving questions unique to each class member, statistical evidence cannot create commonality where it does not otherwise exist, nor can liability be extrapolated where commonality is absent.

Duran is also significant because it requires trial courts to consider—at the class certification stage—whether a workable trial plan involving statistical evidence can be developed. When opposing class certification, therefore, employers should be prepared to challenge the class counsel’s proposed trial plan, or their failure to identify one, based on the principles set forth in Duran.

Finally, Duran is particularly useful to employers defending misclassification cases, as it affirms that such claims—unless they turn on standardized job duties or policies that compel employees to uniformly spend their time on nonexempt work—have “the potential to raise numerous individual questions that may be difficult, or even impossible to litigate on a class-wide basis.”