Left to right: Partner Kristin Linsley Myles and Associate David Feder, Munger Tolles & Olson
Left to right: Partner Kristin Linsley Myles and Associate David Feder, Munger Tolles & Olson ()

When can statistical evidence be introduced to prove injury in a class-action proceeding? In Tyson Foods, Inc. v. Bouaphakeo, the U.S. Supreme Court offered modest guidance on the issue, holding that statistical evidence sometimes may be used to prove class-wide liability—at least where the class members are similarly situated. Though the court declined to adopt a bright-line rule prohibiting such evidence, its opinion leaves open significant questions about exactly when class-action plaintiffs can rely on this sort of evidence and when class certification in such a case is warranted. These questions are likely to be a major focus of litigation moving forward.

In Tyson, a group of employees alleged that the company violated the Fair Labor Standards Act by failing to compensate them for overtime worked. (They also alleged an independent state-law violation but the court considered the claims together.) The employees worked in a pork plant’s kill, cut, and retrim departments. For safety and sanitation, each employee must wear certain protective gear on the job (varying by department and task). Tyson, the employees contended, failed to compensate its employees for the time it took to put on and take off this gear—which would push their hours over the threshold for overtime under the law.

The employees sought to proceed as a class. To earn class certification under the Federal Rules of Civil Procedure, the plaintiffs must demonstrate, among other things, a common issue of law or fact. And the rules prescribe a more stringent showing if the plaintiffs seek certification as a damages class—as did the Tyson plaintiffs. In that case the common issue must also “predominate” over any individual issues.

As it turned out, only if the employees could use statistical evidence to prove liability on a class-wide basis would the predominance requirement be satisfied. The critical piece of evidence was a study by Dr. Kenneth Mericle. He conducted 744 observations and estimated how long it took employees to put on and take off their work gear. Another expert, Dr. Liesl Fox, used this study to calculate the amount of uncompensated work performed by each employee.

Justice Anthony Kennedy, writing for the court, held that Mericle’s study could be used to prove Tyson’s liability. The court’s analysis was straightforward: Under the Rules Enabling Act, the class mechanism cannot diminish or expand an individual’s claim. And Mericle’s study could have been used to prove each employee’s individual claim. Thus, the court concluded, that same study could also be used to support liability in a class-wide proceeding, or else the class-action vehicle would have served (impermissibly) to diminish the class-members’ claims.

In reaching this conclusion, the court rejected Tyson’s reading of the court’s 2011 opinion in Wal-Mart Stores, Inc. v. Dukes. Tyson argued that Wal-Mart‘s language decrying “trial by formula” pointed toward a bright-line rule against the use of statistical evidence to establish a defendant’s class-wide liability. The court disagreed. It held that a representative sample, like all evidence, is but a means to establish or rebut liability and its permissibility turns not at all on the form a proceeding takes but instead on evidentiary rules.

As the court saw it, the difference between Wal-Mart and Tyson turned on what evidence the class members could have introduced in an individual action. In Wal-Mart, the individual employees could not have used the representative evidence to prove their injuries. Each employee’s experiences there bore little similarity to another’s: an employee could not have proved her individual discrimination claim by introducing statistical evidence that other employees’ supervisors engaged in discrimination. By contrast, in Tyson, each employee worked in the same facility, did similar work, and was paid under the same policy. Accordingly, Mericle’s study would have been admissible and sufficient to support a jury finding as to the hours worked if it were introduced in each employee’s individual action. The court held that the same must be true in a class action.

The court then turned to Tyson’s second argument: The employees needed to demonstrate some mechanism to remove uninjured parties from the damages pool or else the jury verdict could not stand. As the lower courts had not ruled on this issue, however, the court left it for the district court to decide in the first instance—though the court hinted that Tyson might be deemed to have waived this argument by insisting on a class-wide, lump-sum jury verdict at trial.

Chief Justice John Roberts concurred. He wrote separately primarily to suggest that it would likely prove impossible to remove uninjured class members from the damage pool—and thus the verdict remained in jeopardy on remand.

Justice Clarence Thomas dissented, joined by Justice Samuel Alito. He explained that the plaintiffs could not maintain a class action when an important element of liability depended on facts that vary among individual class members.

Tyson is striking as much for what it does not say as for what it does. The opinion offers little in the way of guidance about when certification is appropriate generally (and how the predominance inquiry works) in cases where the plaintiffs propose to rely on statistical

evidence. The opinion also says remarkably little about when class members are sufficiently “similarly situated” to take their case out of the realm of Wal-Mart and into the realm of Tyson. It’s a safe bet that these issues will be the focus of future litigation and may well return to the court sooner rather than later.

Still, Tyson settles a couple of important points. The court characterized its 1946 precedent in Anderson v. Mt. Clemens Pottery Co. as consistent with the rule that each employee must prove her individual injury—suggesting it rejected reading the case as authorizing the use of otherwise inadmissible or inadequate statistical evidence in Fair Labor Standards Act cases where the employer lacks adequate records of hours worked. By the same token, the court also rejected a reading that would have flatly precluded the use of representative evidence to prove that individual injury in the class context. Rather, the court held that the evidentiary inquiry must be the same in Fair Labor Standards Act cases as in other cases, and the same in class actions as in individual actions.

Perhaps its most significant contribution, Tyson offers a safe harbor—albeit one with ill-defined contours—for plaintiffs seeking class certification using statistical evidence. That safe harbor ensures that, if statistical evidence could have been offered by each employee proceeding individually, then it must also be allowed to prove liability on a class-wide basis. This inquiry, in turn, will depend on the similarity of the class. Defendants should seek to characterize the class as more dissimilar and requiring more individualized inquiries (as in Wal-Mart) and rebut plaintiffs’ efforts to paint the class as one that is similarly situated (as in Tyson). Defendants should also seek to rely more on evidentiary challenges (especially Daubert challenges) to combat the use of representative evidence.

By offering only modest guidance, Tyson seemingly leaves district courts with significant discretion to determine the appropriateness of class certification when it comes to classes proposing to rely on representative evidence for certification. With much left unsaid, only time will tell the size of the opening Tyson cuts in Wal-Mart’s wall against “trial by formula.”