In Duran v. U.S. Bank National Association, the California Court of Appeal, First Appellate District, overturned a $15 million judgment against U.S. Bank ("USB") entered in a case tried before Alameda County Superior Court Judge Robert Freedman. In its lengthy and very detailed opinion, the court shredded all the major trial management and evidentiary rulings made by the trial court, holding that its use of flawed statistical evidence and refusal to admit relevant testimony in support of USB's defense of exempt status denied USB its right to due process. In the first California appellate decision to apply the U.S. Supreme Court's 2011 Wal-Mart Stores v. Dukes decision, the court determined that the trial management plan was a fatally flawed exercise in "Trial by Formula." As a final repudiation of the trial court's rulings, the Duran court also ruled that the class should be decertified.

What Happened in the Trial Court

Two named plaintiffs brought a class action seeking overtime pay on behalf of a class of 260 business banking officers, claiming they were misclassified as exempt outside sales personnel. The trial court certified the class and notice was sent allowing class members to opt out of the lawsuit.

Acting without the agreement of the parties or expert advice, the trial court decided that the case would be tried by choosing a "random" sample of 20 class members to testify at trial. Five alternates were chosen to substitute in if any in the group of 20 could not or would not participate. After this process, attorneys for the class dismissed some claims and added others. Over USB's objections, the court ordered a second notice to the class and a second opportunity to opt out of the lawsuit. Four of the 20 "random witnesses" (20% of the group) and only 5 of the remaining 250 absent class members (2% of that group) exercised their second chance to opt out of the lawsuit.

USB protested. It asked that the four "random witnesses" who had opted out of the lawsuit be permitted to come back. It claimed they had had opted out because they did not support the class claim, did not believe they were misclassified, and because the lawyers for the class had urged them to avoid involvement in the lawsuit. It argued their absence would skew the "random" sample in favor of the plaintiffs. USB also supplied an expert opinion that the significant difference in opt-out rates between the "random" witnesses and the remaining class members would undermine the validity of any extrapolation from the sample to the class as a whole. Richard Drogin, the expert statistician hired by the class, opined that allowing the alternate random witnesses to substitute for the 4 new opt-outs would be statistically acceptable and that there was no reason to think there was any bias in the sample. The trial court accepted these assertions. The appellate court rejected them, branding them "remarkable."

Throughout the proceedings, USB sought to introduce testimony from class members outside of the "random" group, including 70 class members who had signed declarations stating that they spent more than half their working time away from the bank's premises – the key issue in determining their exempt status. The trial court rejected every attempt by USB to present this evidence in support of its affirmative defense. It also rejected the bank's motion to decertify the class on the basis of the declarations, while simultaneously reminding USB that it bore the burden of proving that its business banking officers were properly classified as exempt.

After hearing the testimony of the two named plaintiffs and 19 "random" witnesses, the court concluded they had all been misclassified and had worked overtime. It also found they were typical and representative of the class as a whole, and that the process used in the case had validated the use of the "random witness" process as part of the trial management plan for a wage and hour class action. Remarkably, the trial court also found that USB failed to negate or rebut the testimony of the random witnesses. This finding was "critical" in the view of the appellate court – because every effort by USB to introduce evidence from any class member who was not among the "random" witnesses was rejected, and because that evidence "arguably would have shown that some class members were properly classified as exempt or did not work overtime."

The Court of Appeal Overturns the Judgment and Decertifies the Class

The Trial Plan Improperly Relied on Representative Sampling

The Duran court concluded that "the procedural flaws anticipated [in earlier decisions] appear to have been fully realized in the present case." The statistical methods adopted by the trial court – including the selection of a sample size of 20 class members – were not based on any expert opinion. The margin of error was too great. The high rate of opt-outs among the "random witnesses" and the inclusion of the named plaintiffs in the "sample" destroyed any possibility that the sample was random. And the trial management plan "was lacking in any expert input or principled statistical foundation."

The court drew distinctions between its own decision in an earlier case, Bell v. Farmers Ins. Exchange ("Bell III") and Duran. In Bell III, damages were proved through statistics, not liability. In Bell III, the defendant agreed to the use of statistical proofs and the standards to be used, and the process was at all times random. The margins of error were much lower, and the defendant did not seek to introduce any other evidence. In Duran, these factors were not present. Worse, the trial court repeatedly refused to allow USB to introduce "manifestly relevant evidence" that "could have greatly mitigated the damages awarded and possibly could have defeated plaintiffs' class action claim entirely."

The Duran court noted that no court had ever allowed a finding of liability (as opposed to damages) on the basis of statistics. It drew support from earlier decisions denying class certification where defenses to overtime claims turned on the specific circumstances of individual employees. The Duran court found In re Wells Fargo Home Mortg. Overtime Pay Lit. ("Wells Fargo II") "particularly instructive." That case recognized the problem of identifying individuals who had been harmed, versus those who had not, even where the proper use of statistical methods could reliably predict what percentage of a class had been improperly classified as exempt. Quoting Wells Fargo II, the Duran court observed: "Plaintiff has not identified a single case in which a court certified an overbroad class that included both injured and uninjured parties . . . . In fact, the court has been unable to locate any case in which a court permitted a plaintiff to establish the nonexempt status of class members, especially with respect to the outside sales exemption, through statistical evidence or representative testimony."

Finally, the Duran court applied the principles announced in Wal-Mart Stores v. Dukes, which rejected the "novel project" of "Trial by Formula" and held that "a class cannot be certified on the premise that Wal-Mart will not be entitled to litigate its statutory defenses to individual claims." Calling particular attention to the fact that this portion of the Wal-Mart opinion "was the expression of a unanimous court," the Duran court agreed that "we find this approach to be untenable."

The Trial Court Denied the Bank Its Right to Due Process

USB offered evidence that could have proved that at least one-third of the absent class members were entitled to no recovery at all. Though it was "unquestionably relevant and therefore admissible," the trial court rejected this evidence because it did not fit within the trial plan. This "unfortunate" result ignored the fact that the relevancy of the evidence should have been judged "by the trial itself as it unfolded in the courtroom" and not on the flawed trial plan. Speed and efficiency cannot outweigh the obligation of the courts to provide fair and accessible justice, and the trial court "erred in foreclosing USB the opportunity to raise individualized challenges to the absent class members' claims." The "deprivation was profound" and the error a prejudicial miscarriage of justice – as a different result would have been probable in the absence of the trial court's error. Indeed, Duran cites an earlier case as providing clear support for the "premise that due process principles require individualized inquiries where the applicability of an exemption turns on the specific circumstances of each employee, even where the employer's misclassification may be willful."

The Duran court also rejected the trial court's reliance on the absence of a policy requiring class members to spend more than half their time away from the bank, and on the fact that USB did not monitor or track the time that class members spent away from the bank. In fact, the Duran court ruled that the trial court had things backwards – the absence of a policy or monitoring of class members' time supports the need for individualized testimony from class members. Exactly the sort of evidence that USB offered repeatedly and that the trial court entirely rejected.

The trial court's use of flawed statistical evidence and its repeated refusal to allow USB to provide testimony from class members who were not part of the "random" sample, despite the existence of 70 declarations demonstrating that class members were properly classified, created a "high risk that USB will be compelled to pay money to absent plaintiffs who may not be entitled to any recovery . . . ."

The Duran court rejected all arguments that the trial court's use of statistics was acceptable or compatible with its prior ruling in the Bell III litigation. The trial plan did not use a true random sample because it included the two named plaintiffs and there was no evidence from any expert that the use of a 20-person sample was statistically significant or representative for the purpose of extrapolating findings to the entire class. The "response rate" was not "extremely high" because six of the original "random witnesses" did not respond: four took the second opportunity to opt out; one was removed by the court; and one did not testify. There was a "measurement error" because several witnesses testified to working a range of hours without specifying how often they worked at the top of the range or at the bottom of the range, and there was no evidence as to the probable distribution of hours worked by class members who were not among the randomly selected witnesses.

The court was particularly troubled by the 43.3 percent margin of error, which it found to be a due process violation on its own. It had the potential to increase the bank's aggregate liability by "close to double that which would be warranted if the low end of the margin were applied." The restitution calculations used in the second phase of the trial were necessarily flawed because they were based on the court's statement of decision which was, itself, based on constitutionally suspect data.

The Duran court acknowledged the decision of the California Supreme Court in Sav-On Drug Stores v. Superior Court and its call for trial courts to fashion creative tools to manage class actions. It recognized that in class action litigation, "efficiencies must be maintained, sometimes resulting in imperfect results." But, it concluded, the trial court impermissibly favored expediency and "constructed a set of ground rules that unfairly prevented USB from defending itself" by "flatly foreclosing the admission of potentially relevant evidence."

The Class Is Decertified

The Duran court completed its repudiation of the trial court's rulings by finding that it abused its discretion when it denied USB's second motion to decertify the class. This denial was based on the "erroneous legal assumption that a finding of liability due to misclassification could be determined by extrapolating the findings based on the ['random witnesses'] to the entire class." That "conclusion was legally unsound because the methodology employed by the trial court was legally unsound; it violated USB's right to present relevant evidence in its defense."

The Lessons of the Duran Decision

The decision in Duran provides much needed support for what would seem to be some very basic principles of common sense and fairness – principles that sometimes seem to get lost in class action litigation. First, a case presenting a defense of exempt status or other issues turning on individualized facts – like how or where employees spend their time – should not be certified for class treatment. Second, if such a case is certified, employers must be able to present evidence by or about individual members of the class to meet their burden of proving the exemption. Third, Duran provides a reminder of the serious – and as yet insurmountable – obstacles to proving liability on the basis of statistical extrapolations. Lastly, it provides a detailed primer on what can go wrong when courts put speedy and efficient results ahead of more laborious justice by using flawed statistical models.

The appellate court's reasoning in Duran was foreshadowed in a number of articles and briefs by Littler attorneys written in the past few years. For further explanation of the flaws inherent in trial plans that rely on random sampling to solve the difficulties presented when individual claims are aggregated in a single case, see the Littler Report, Dukes v. Wal-Mart, Some Closed Doors and Open Issues.

Diane Kimberlin is a Shareholder in Littler Mendelson's Los Angeles office. If you would like further information, please contact your Littler attorney at 1.888.Littler, info@littler.com, or Ms. Kimberlin at dkimberlin@littler.com.