When Numbers Lie: The Limits of Statistical Methodology in California Class Action Management

Courts that oversee class actions can use class sampling and other statistical methods to manage litigation involving large numbers of plaintiffs and the vast amount of data associated with them. In California, however, those methods must be reliable, and cannot strip defendants of the right to litigate affirmative defenses.

The California Supreme Court recently announced its decision in Duran v. U.S. Bank National Association, 2014 WL 2219042, finding that the trial court had abused its discretion in managing a class action employee misclassification case. The Court criticized various aspects of the trial court’s plan, but focused significant attention on the faulty statistical methods utilized by the trial court to assess both liability and damages. Additionally, the Supreme Court found that the trial court’s plan prevented U.S. Bank (“USB”) from litigating its affirmative defenses.   In a 43-page opinion that will likely have implications in class action case management beyond the employment context, the Supreme Court held that “[a] trial plan that relies on statistical sampling must be developed with expert input and must afford the defendant an opportunity to impeach the model or otherwise show its liability is reduced.”


In Duran, USB business banking officers (BBOs) sued their employer, asserting that they had been misclassified as exempt employees who were not entitled to overtime. USB had classified them as outside sales employees exempt under California Labor Code Section 1170, which requires such employees to spend more than 50% of their workday in sales outside of the office. The trial court certified a class of 260 BBOs.

After certification, USB proposed dividing the class members into groups and appointing special masters to conduct individual hearings on liability and damages. Plaintiffs, on the other hand, proposed using a class-wide survey and random sampling.  Rejecting both USB’s and Plaintiffs’ proposal, the trial court devised its own plan to select a random group of 20 class members plus the 2 class representatives (the “Random Witness Group” or “RWG”) who would testify at trial and determine both liability and damages for USB, and to then extrapolate those outcomes to the class as a whole.

USB objected repeatedly to the trial court’s management of the case, and unsuccessfully moved to decertify the class due to the predomination of individual issues. Once trial began on liability, the trial court refused to accept any evidence related to the classification of any class member not in the RWG. On the issue of liability, USB sought to offer evidence that some class members worked outside the office more than 50% of the time, and therefore had been properly classified. Because those class members were not in the RWG, however, the court refused to allow USB to present any of that evidence.

In criticizing the trial court’s approach, the Supreme Court focused first on the certification and trial management plan. Not only does the trial court have to consider the predominance of common issues, it also must “conclude that litigation of individual issues, including those arising from affirmative defenses can be managed fairly and efficiently.” If a class is certified and then proves unmanageable, the trial court has a duty to decertify. 

The Supreme Court also criticized the trial court for “rigidly adhering to its flawed trial plan and excluding relevant evidence central to the defense.” By using a small statistical sampling to determine liability – an individual issue driven by the number of hours spent in the office – not just damages, it glossed over the potential that USB was not liable to some of the BBOs. In short, the trial court “did not manage individual issues.  It ignored them.”

As to the flawed trial plan, the Supreme Court highlighted several rulings that compromised the randomness of the RWG. First, while 20 of the class members were chosen by the court, the RWG also included the two named plaintiffs, who had been selected by class counsel. In fact, the named plaintiffs had been substituted several times, based on friendliness to the class’s position. The Supreme Court noted that the inclusion of the 2 named plaintiffs was the opposite of random, and skewed the sample in favor of the plaintiffs. Additionally, there was no explanation by the trial court of whether or how it had determined that twenty plaintiffs was an appropriate sample size for the RWG.

The Court also pointed out that, after the RWG was selected, the Plaintiffs amended the complaint. This in turn led the trial court to allow an additional opt-out opportunity for class members who no longer wanted to be a part of the class under the amended complaint.   In the RWG, 4 out of 20 opted out (20%), while only 5 of the remaining 250 members opted out (2%).  Such a large discrepancy in opt-out rates was “very unlikely to be attributable to random chance,” according to USB’s expert. When USB investigated the RWG class opt-outs, some of the RWG members who had opted out said that class counsel had encouraged them to do so, further calling the randomness of the sample into question.

As for the use of statistical sampling, the Supreme Court noted that “the court’s attempt to implement random sampling was beset by numerous problems.”  While not going so far as to say that sampling is never permissible, the Supreme Court laid out how the trial court failed to use sampling properly and protect parties’ rights. Specifically, the sample size was too small, not random, and had intolerably large margins of error – for example, 43.3% as to estimated overtime.

The Duran opinion makes clear that – whatever the methods used by courts to make class actions manageable – individual issues must be fairly managed, and, when a court utilizes statistical sampling, the sample “must be representative and the results obtained must be sufficiently reliable to satisfy concerns of fundamental fairness.” This focus on the fairness and reliability of class action management methods raises parallels to the United States Supreme Court’s Daubert opinion’s focus on the relevance and reliability of expert testimony. Just as Daubert seeks to avoid undue intrusion into the parties’ rights to call whichever scientific expert they see fit while ensuring that the resulting testimony is still scientific, so Duran seeks to avoid unnecessarily limiting trial court’s discretion to manage unwieldy litigation, while ensuring that the methods employed are still fundamentally fair. And while Daubert and Duran apply to plaintiff and defense equally, the nature of the two sides approaches to litigation suggest that Duran will evolve into authority widely perceived as defense-friendly.

Image courtesy of Flickr by LendingMemo.com.

Argument Report: When Should a Court Lose the Power to Decertify?

Our reports on the oral arguments of the Illinois Supreme Court's September term continue with Mashal v. City of Chicago. Mashal presents an issue of potentially enormous importance to class action practice in the Illinois state courts: when does the Circuit Court lose the power to decertify the class under Section 2-802 of the Code of Civil Procedure? To watch the video of the argument, click here.

For a detailed discussion of the facts and rulings below in Mashal, click here. The case arises from the City of Chicago's practice of issuing "fly-by" traffic citations to taxi drivers -- citations which were received by mail, rather than being personally served or placed on the vehicle. The City conceded that it issued such citations occasionally, principally when drivers either fled or became aggressive; the plaintiffs alleged that the practice was far more widespread than that. A class was certified in 2002. In 2005, partial summary judgment was entered, finding that the practice was illegal. The following year, the City obtained its own partial summary judgment, eliminating claims before 1995 on statute of limitations grounds.

According to Section 2-802, a court may amend a class certification order at any time "before a decision on the merits." The statute doesn't define what "decision on the merits" means. In 2007, the City decided to find out, moving to decertify the Mashal class on the grounds that the partial summary judgment on the legality of the practice eliminated the only common issue. The Circuit Court granted the motion to decertify, and the case rose to the Appellate Court on various certified questions. The Appellate Court held that "decision on the merits" meant something similar to the types of judgments and orders given res judicata effect -- a complete resolution of the liability claim.

Counsel for the plaintiffs argued that the need for individualized determinations didn't mean that a "decision on the merits" hadn't yet occurred, since every class action involves such determinations. Justice Thomas asked counsel to address the City's argument that the partial summary judgments were a decision on the merits. Counsel responded that if a decision resolving nothing more than the legality of the practice involved in the case became a vehicle for decertification, that exception would swallow the class action statute. Such a rule was antithetical to the underlying reason for class actions: that litigating the claims one by one was impractical. Justice Theis asked whether the Circuit Court's decision rejecting the City's affirmative defenses was a decision on the merits, and counsel responded that that order could have been appealed if Rule 304(a) language had been granted.   Justice Theis then suggested that "decision on the merits" and "final judgment" arguably sounded like the same thing, and asked counsel to explain his distinction. Counsel responded that a final judgment was a complete resolution of a claim, granting relief. A decision on the merits didn't need to be final; the order concluding that issuance of fly-by tickets violates the law was enough. Justice Thomas asked counsel whether counsel was saying that every order resolving affirmative defenses was a "decision on the merits," or only the defenses resolved in the particular order at issue. Counsel responded that the particular defenses involved in the court's order made the decision one on the merits.

Counsel for the defendant responded that the reason for setting the cutoff for decertification at the decision on the merits was to keep the option open until all possibly common issues have been decided. Justice Garman asked whether some determination of liability to a class member or members was required. Counsel responded "yes", and in the case at bar, such a determination was impossible given the need to determine whether each class member had received a "fly-by" citation. Justice Burke pointed out that the City had admitted to issuing such citations; counsel responded that the City had admitted the practice as a general matter, but the taxi driver class members had not linked themselves to the limited group who received such citations. Justice Thomas asked counsel to address plaintiffs' argument that there was no rebuttal to the proposition that the class members had received fly-by citations. Counsel answered that such affidavits were not admissible, and in fact there were many reasons to question the class members' credibility. Counsel argued that deposition testimony was needed to determine the credibility of all class members. Justice Garman wondered why issues like credibility and individual liability weren't part of ancillary proceedings. Counsel responded that under the circumstances presented, liability would have to be determined class member by class member, and this was the time to conclude, once and for all, whether class prerequisites were met.

On rebuttal, Justice Karmeier asked counsel for the plaintiffs to respond to the City's argument that plaintiffs' reading of the statute would eliminate courts' ability to reevaluate whether common issues continued to predominate. Counsel answered that it was necessary to set a cut-off point. Although not all summary judgments would bar decertification, this one -- holding that defendant's conduct was unlawful -- would. Justice Thomas asked how the Court should take into account the fact that individual plaintiffs would have to proceed with small liability cases, given that the rule announced by the Court would be used even outside the class action context. Counsel responded that the Court's ruling would be limited in impact -- even res judicata involved subtle differences that might limit the scope of the Court's decision. At any rate, counsel argued that this consideration should not affect the Court, since the legislature could step in and change any rule it didn't approve of.

U.S. Supreme Court Rejects Statistical Significance as Requirement to Plead Materiality

Matrixx Initiatives, Inc. v. Siracusano (.pdf) was a securities fraud class action where claimants alleged that Matrixx failed to disclose reports of a potential link between its cold medicine, Zicam, and loss of smell. The district court dismissed the claim because the reports lacked statistical significance and, therefore, could not have formed a “material” omission under the Securities Exchange Act. The Ninth Circuit reversed the dismissal.

In a unanimous opinion, authored by Justice Sotomayor, the Supreme Court held that the district erred in employing a statistical significance requirement. Instead, the court should have considered the “total mix” of information available to investors. No single bright-line test of materiality is controlling, and the allegation taken as a whole permitted the inference that the adverse reports could have affected a reasonable investor.

Although the opinion does not concern the admissibility of expert testimony, the Court’s discussion of the reliability of a causation inference in the absence of statistically significant findings will doubtless be cited in future Daubert battles.

Also of general interest is the Court’s explanation of how the facts pleaded met the Twombly plausibility requirements. Indeed, the case may serve as a template for pleading fraud by omission.