Pomerantz LLP

Supremes: Statistical Averages Can Provide a Basis for Class-Wide Liability


In Tyson Foods v. Bouaphakeo, the Supreme Court upheld the use of statistical sampling evidence in class actions, at least where such evidence would have been admissible in an individual action. Defendants had argued that such statistical methods improperly treated each individual as if he or she matched a statistical average, thus manufacturing predominance by assuming away the very individualized differences that made class-wide litigation inappropriate in the first place. The Court rejected this premise and focused instead on the relevance of the statistical evidence to the substantive claim at issue. It held that, if a given class member could have used the statistical evidence to obtain a favorable jury verdict in an individual action, then the class could use it the same way. It was up to the jury to decide, in light of all of the evidence presented, whether the statistical average was probative of the situation of each class member.

Particularly under the specific facts of the case, this ruling was a straightforward application of evidentiary common sense. Nevertheless, it was generally seen as a significant victory for the plaintiffs’ bar.

The case involved workers at a pork processing plant who claimed they were not paid overtime for time spent putting on and taking off protective gear, in violation of federal law requiring compensation for such “donning and doffing”  time if it is “integral and indispensable” to their regular work. After the district court certified two classes of employees, the case went to trial and the jury awarded the classes $29 million in damages. On appeal, the defendant argued that the verdict should be thrown out because the classes never should have been certified.

To be certified as a class, the worker-plaintiffs had to prove that they could establish key elements of their claims through generalized, class-wide proof. This was easy for some elements: they all worked in the same plant, had similar job responsibilities, and were subject to essentially the same compensation policies. But the defendant insisted that individualized inquiries into each employee’s total donning and doffing time were necessary because different employees wore different gear and took varying amounts of time to don and doff the gear. It also argued that no class could be certified without proof that every member was injured, which required individualized inquiries into each employee’s time.

Federal law, to some degree anticipating this evidentiary problem, has long required employers to keep accurate records of employee work hours. But despite a 1998 federal court injunction against the very same slaughterhouse requiring it to record employee time donning and doffing protective gear, Tyson Foods had never done so. Instead, it had been compensating workers based on its own approximations of how long those activities should take.

Because there was no good individualized evidence, the worker-plaintiffs used what they called “representative evidence” to show how long workers in each department generally took to don and doff protective gear. Most significantly, they presented a study by an industrial relations expert who drew on a representative sample of 774 videotaped observations of workers and calculated the average time for workers in each department to don and doff their gear.

There are procedural mechanisms to ensure the reliability of this kind of evidence, but the defendant largely ignored them. It did not challenge the expert’s qualifications or statistical methodology. It rejected the workers’ proposal to bifurcate the trial into separate proceedings for liability and damages. While it argued at trial that the expert’s calculations were too high, it did not present a rebuttal expert with different calculations. Instead, in opposing class certification and also at trial, it insisted that it was fundamentally improper to assume that each employee donned and doffed for the same time as the average in the sample. It decried being subjected to a “trial by formula” and barred from raising unspecified “defenses to individualized claims.” And on appeal, it called for a categorical rule barring the use of representative sampling evidence in class actions.

The Supreme Court rebuffed this effort and categorically rejected the idea that class actions required their own special set of evidentiary rules. It emphasized that statistical sampling evidence is routinely used in all kinds of litigation and is often the only practicable means, for plaintiffs and defendants in individual as well as class actions, to collect and present relevant data. Thus, it held that class certification was proper as long as a reasonable jury could have believed that the employees spent roughly equal time donning and doffing. If so, it was for a trial jury to weigh the expert’s average-time calculations against the other evidence presented and to decide whether the statistical average was probative of the time actually worked by each employee.

While the utility of statistical averages in other class actions will vary, the main takeaway is that the issue must be considered in practical terms of how a reasonable jury resolving the underlying substantive claim would view the evidence. In many cases, statistical averages will be the most compelling evidence available and will say a great deal about each member of the class. This was particularly true in Tyson Foods because the defendant had never bothered to keep individualized records (despite being legally mandated to do so), and instead simply paid workers based on its own approximations of how long donning and doffing should take. But in other cases with stronger evidence of meaningful individualized variations, a jury might find statistical averages less useful.