The probability is “not really”

Statistics are kind of a holy grail of class action litigation. Everyone seems to know that they exist, but their understanding is shadowy and the quest to find valid statistical models often proves elusive. Last month’s Supreme Court decision in Tyson Foods, Inc. v. Bouaphakeo, Case No. 14-1146 (Mar. 22, 2016), certainly addressed the topic, and while it will likely increase the number of attempts to use statistical data, their actual use will likely remain limited.

The Tyson Foods case (which is much easier to say than “Bouaphakeo”) was one of many donning and doffing cases brought over the past several years in the slaughterhouse industry. In this particular case, the claim was that the employer failed to compensate employees for the time they spent donning and doffing various pieces of protective gear. The type of gear varies among the types of work actually being done. In addition, the actual amount of time worked by the employees (such as whether they worked any overtime with or without the doffing time) also varied. The district court ultimately certified classes under the FLSA and Iowa state law, and the case proceeded to trial.

At trial, the plaintiffs introduced statistical evidence regarding the amount of time spent donning and doffing. In the gruesomely named “kill” department, the time was 21.25 minutes, and it was 18 minutes at the “cut and trim” departments. The defendant opposed the use of statistics generally, but apparently not the specifics of the methodology used by the plaintiffs’ expert. The jury awarded a single figure for the entire class of $2.9 million, without any breakdown.

Sixty years ago, in a case that predated and actually precipitated substantial amendments to the FLSA, the Supreme Court held that representative testimony could be used to establish the amount of liability in an FLSA action where, as was the case at Tyson Foods, the employer kept no records of the disputed time. Anderson v. Mt. Clemens Pottery Co., 328 U.S. 680 (1946). In Tyson Foods, the Supreme Court addressed a number of issues relating to the application of that holding, but due to the case’s unique procedural posture, its holding left more questions than it answered.

Without addressing the 1946 FLSA amendments, a majority of the Court affirmed the use of representative testimony to establish the amount of liability, but its holding left a gaping hole: what to do about putative class members who had no FLSA claim at all and whether the verdict could stand when it could not be determined who suffered a claim and who did not. The court ultimately “punted” on these issues as they were not adequately addressed in the briefing or before the trial court. The Court remanded the case for the trial court to come up with an appropriate methodology. As the concurrence noted, however, that task may prove impossible as there is no meaningful way to reverse-engineer a jury award when it does not match the expert testimony, does not break out employees with different time estimates, and does not consider the claims of those (here, over 200 employees) who mathematically had no FLSA claim.

The Tyson Foods decision stands for little more than that representative testimony may, if appropriate, theoretically be used to establish the amount of liability in FLSA cases where the employer has kept no records of the disputed time. The Court did not decide, however, what methodology would need to be used or even whether the methodology used in Tyson Foods itself was appropriate. The Court emphasized that it was not altering the holding in Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338 (2011), prohibiting the use of “Trial by Formula.” Moreover, the majority found no need to address (because it was premature) the issue of whether the verdict could even stand when there were hundreds of class members who suffered no injury at all. These two flaws render the Tyson Foods holding of only limited application.

The proof of that limited reach is reflected, surprisingly enough, in a pair of California cases that do address the two issues Tyson Foods left open. Two years ago, the California Supreme Court rejected a class action and $15 million verdict premised upon claimed statistical evidence. Duran v. U.S. Bank National Ass’n, 59 Cal. 4th 1 (May 29, 2014). We blogged that decision here. Among many other flaws, the court found that the small sample size, poor selection criteria, poor controls for uncooperative plaintiffs, questionable support for sampling, and methodology all undermined the statistics’ validity. Further, the court noted the issue not reached in Tyson Foods, and found that statistics and notions of rough justice could not overcome the issue of whether the defendant was liable to particular individual class members at all. Last week, a trial judge in Los Angeles decertified a case involving over 2,000 employees claiming various wage and hour violations, in large part because of gaps and unexplained processes used in the statistician’s calculation. Williams v. Allstate Insurance Co. Case No. BC382577 (Apr. 13, 2016).

The upshot is that while employers will not be able to argue that statistical evidence can never be used in wage and hour cases, most cases will still break down over how such evidence can be used and plaintiffs will face particular challenges when questions remain as to whether particular class members have any claim at all.

The bottom line: Statistical evidence can be used for some aspects of wage and hour litigation, but not across the board, and the need to use appropriate statistical methods will doom many attempts.