n March 2016, the U.S. Supreme Court decided Tyson Foods, Inc. v. Bouaphakeo,1 which was initially panned by some as a win for the plaintiff bar and a walk back from the Court’s landmark decisions in Wal-Mart v. Dukes2 and Comcast v. Behrend3 which seemed to reject “trial by formula” in class action cases. In Tyson Foods, the Supreme Court held that a class of workers in a food processing facility could rely on representative sampling and statistical analysis regarding hours worked “to fill an evidentiary gap” created by Tyson’s failure to keep adequate records to establish class-wide liability for alleged violations of the Fair Labor Standards Act (FLSA) and a similar Iowa state law. In so holding, the Supreme Court declined to address the other issue raised by Tyson’s petition for certiorari—whether plaintiffs must demonstrate a mechanism for ensuring that uninjured class members do not receive damages. While technically a victory for plaintiffs, the Tyson Foods majority opinion, authored by Justice Kennedy, made clear that it was not a broad ruling sanctioning the use of representative proof, often called “trial by formula,” in class actions. Indeed, as detailed below, the Supreme Court laid the foundation for defendants to challenge the introduction of sampling evidence in class actions in several ways. First, by deciding the case on its facts, the Supreme Court in Tyson Foods left open whether its holding is limited to FSLA collective actions. Second, Tyson Foods emphasized that the underlying admissibility question is whether the representative evidence could have been used to establish liability in an individual action, reinforcing that pursuant to the Rules Enabling Act, there can be no special rules relaxing the standards for class actions. Third, Justice Roberts’ concurrence suggested that plaintiffs must present a trial plan explaining how they intend to use the representative sampling and their proposed method for allocating damages, which provides defendants with yet another tool to expose due process, manageability and other issues related to the use of sampling evidence early in the case. Finally, now, a few months after the decision, it has become apparent that lower courts are beginning to interpret and apply Tyson Foods in ways that are often favorable to defendants. Plaintiffs Sought to Prove Tyson’s Liability Through Statistical Averages First the facts: the plaintiffs in Tyson Foods were hourly workers who alleged that Tyson failed to compensate them for time spent donning and doffing protective equipment in violation of the FLSA and an Iowa state wage and hour law. The district court permitted plaintiffs to prove liability and damages by employing statistical evidence that presumed all class members were identical to an “average” employee and spent equal time on the tasks at issue. In addition, the district court certified a class that included members whom plaintiffs’ own expert conceded were not underpaid (and who were thus not injured). After denial of Tyson’s motion to decertify the class,4 the case went to trial and resulted in a $2.9 million jury verdict in favor of the plaintiff class. The district court then denied Tyson’s motion for judgment as a matter of law, and Tyson appealed to the Eighth Circuit. On appeal, Tyson argued that the Supreme Court’s 2011 decision in Wal-Mart v. Dukes should have precluded the district court from certifying the putative class.5 In Dukes, 6 a sex discrimination class action, the Supreme Court rejected the use of representative proof whereby plaintiffs offered a plan to try a sample set of class members’ claims and, if the alleged sex While technically a victory for plaintiffs, the Tyson Foods majority opinion … made clear that it was not a broad ruling sanctioning the use of representative proof, often called “trial by formula,” in class actions. Weil, Gotshal & Manges LLP 6 Employer Update August 2016 discrimination at issue in the case was proved, to then multiply the average back-pay award to determine the class-wide recovery. The Dukes Court, in an opinion penned by the late Justice Scalia, expressed its view that plaintiffs could not extrapolate evidence from one subset of plaintiffs and apply it to the class as a whole absent additional individualized proceedings.7 Tyson also argued that class certification was improper because the putative class included many individuals who were not underpaid and who have no legal right to damages. The Eighth Circuit rejected Tyson’s arguments and affirmed the district court’s rulings. When Tyson further appealed to the Supreme Court, it granted certiorari on two issues: (1) whether differences among individual class members may be ignored and a class certified under Federal Rule of Civil Procedure 23(b)(3) or as a collective action where liability and damages will be determined with statistical techniques that presume all class members are identical to the average observed sample and (2) whether a class action may be certified or maintained under Rule 23(b)(3) or as a collective action when the class contains hundreds of members who were not injured and have no legal right to any damages.8 In its merits brief, however, Tyson reframed the second issue by arguing only that plaintiffs needed to show how they planned to exclude uninjured class members “prior to judgment” for damage allocation purposes—not prior to class certification. A Limited Majority Opinion on the Permissibility of “Trial by Formula” In holding that the plaintiffs in Tyson Foods were permitted to use statistical averages to show Tyson’s liability for unpaid overtime, Justice Kennedy, joined by Justices Roberts, Ginsburg, Breyer, Sotomayor and Kagan, noted that the ability for future plaintiffs to use statistical analysis “depends on the purpose for which the evidence is being introduced and on the elements of the underlying cause of action.”9 The Court further explained that “[t]he fairness and utility of statistical methods in contexts other than those presented here will depend on facts and circumstances particular to those cases.”10 Underlying the Court’s decision to permit the use of statistical analysis in Tyson Foods were the facts that the case was an FSLA collective action and that Tyson had failed to keep adequate records of the hours each employee worked. First, citing Anderson v. Mt. Clemens Pottery Co.,11 an FSLA case that was discussed at length during oral argument, the Court noted that it had previously allowed employees to use statistical analysis in FLSA collective actions to show hours worked.12 Second, because Tyson had failed to keep adequate records regarding its employees, each employee would have likely needed to rely on statistics to prove the amount of time he or she worked had each class member brought an individual lawsuit.13 The Court wrote that “[i]f the sample could have sustained a reasonable jury finding as to hours worked in each employee’s individual action, that sample is a permissible means of establishing the employees’ hours worked in a class action.”14 The Supreme Court also rejected Tyson’s argument that the use of statistical averages contravened its decision in Wal-Mart Stores, Inc. v. Dukes. 15 The Court explained that Dukes did not stand for the broad proposition that representative samples could never be used to establish a defendant’s liability.16 Such evidence may be used to establish liability where it could have been used to establish liability had the class members brought individual actions. In Dukes, the plaintiffs did not provide significant proof of a common policy of discrimination to which each employee was subject and, therefore, none of them could have prevailed in an individual suit by relying on evidence detailing the ways in which other employees were discriminated against by their particular store managers. In contrast, in Tyson Foods, class members worked at the same plant, used similar equipment and were paid under the same policy and, under these circumstances, statistics regarding a subset of employees was probative of the experience of all class members. Thus, the Supreme Court held that calculation of class-wide damages based on the average time class members spent donning and doffing equipment was permissible and not inconsistent with Dukes. 17 In dissent, Justice Thomas, joined by Justice Alito, stated that the district court erred in certifying the class because the class of workers did not satisfy the Weil, Gotshal & Manges LLP 7 Employer Update August 2016 predominance requirement of Federal Rule of Procedure 23(b)(3).18 Justice Thomas wrote that the district court failed to undertake a rigorous analysis to ensure that the statistical evidence offered by the workers was sufficiently probative of the individual issue—whether each employee worked over 40 hours without receiving full overtime pay—to make it susceptible to class-wide proof.19 The Question of Uninjured Class Members Remains Unresolved The Supreme Court declined to address the second issue on which it granted certiorari—whether a class could be certified with uninjured class members— reasoning that it would be inappropriate to decide the issue since Tyson reframed the argument in its merits brief where Tyson argued that plaintiffs needed to set forth a more complete trial plan, i.e. articulate a feasible method for identifying uninjured class members to ensure that they would not receive damages. The Court similarly declined to decide the issue as reframed by Tyson on the record before it because it was unclear whether plaintiffs’ proposed damages allocation method would successfully identify uninjured class members.20 But the Court noted that Tyson could raise its challenge to the trial plan when the case is remanded to the district court for disbursal of the jury award. Justice Roberts further emphasized this point in a concurrence, expressing his concern that plaintiffs would be unable on remand to the district court to devise a means of distributing the jury award to only injured class members: “[i]f there is no way to ensure that the jury’s damages award goes only to injured class members, that award cannot stand. This issue should be considered by the District Court in the first instance.”21 The Supreme Court granted certiorari on a similar issue in Spokeo, Inc. v. Robins. 22 While holding that a named plaintiff must show that he has suffered a “particularized” and “concrete” harm, Spokeo also left open the issue of whether a class can be certified if it is clear that some potential class members are uninjured and would not themselves meet Article III’s standing requirements.23 Practical Defense Strategies Post-Tyson Foods In the wake of Tyson Foods, defendants will likely see an increase in the number of class action plaintiffs proffering statistical evidence in an attempt to establish liability and damages. However, as noted above and discussed in further detail below, the Tyson Foods opinion laid the foundation for various strategies for challenging such evidence and defeating class certification. Tyson Foods May Be Limited to Its Facts Defendants should argue that the decision in Tyson Foods authorizing the use of statistical sampling was limited to its facts in the context of FSLA collective actions and, therefore, is not a radical endorsement of “trial by formula.” The Supreme Court was clear that the permissibility of representative evidence outside of the FSLA context “depend[s] on the purpose for which the evidence is being introduced and on the elements of the underlying cause of action.”24 In Tyson Foods, the underlying cause of action was an FSLA collective action, giving plaintiffs the benefit of clear prior Supreme Court precedent—the Mt. Clemens decision—endorsing the use of representative sampling to prove liability in collective actions where the employer failed to keep adequate records in accordance with its statutory duties. To date, few lower courts have applied Tyson Foods, leaving the question of what circumstances outside of FSLA collective actions with similarly situated class members permit the use of representative sampling unanswered. In fact, only one reported opinion, in a case that also involved wage and hour claims, has approved of the use of statistical sampling since Tyson Foods. In In the wake of Tyson Foods, defendants will likely see an increase in the number of class action plaintiffs proffering statistical evidence in an attempt to establish liability and damages. Weil, Gotshal & Manges LLP 8 Employer Update August 2016 Villalpando v. Exel Direct Inc.,25 defendant Exel Direct Inc. (“Exel”) moved to decertify a class of delivery drivers asserting wage and hour claims under the California Labor Code, arguing that the Mt. Clemensrule permitting the use of statistical evidence of hours worked by the average employee to show liability did not apply where the employer maintained adequate time records.26 The District Court for the Northern District of California disagreed, finding that the burden was on the employer to show that its employee time records were complete and accurate and that Exel had failed to meet this burden even though it had produced over four million pages of timesheets and other records.27 Representative sampling, the district court held, could be used in cases where an employer’s records were simply unorganized or missing certain information and even where there was no statutory duty for an employer to keep the specific records at issue.28 In short, Villalpando seemed to expand on the circumstances offered in Tyson Foods where statistical sampling could be used. On the other hand, the rule in favor of allowing statistical sampling in FLSA and wage and hour class actions does not appear to be absolute, and defendants should consider challenging the use of sample evidence even when faced with an FSLA collective action on the basis that the class members are not “similarly situated” like the workers in Tyson Foods. In Senne v. Kansas City Royals Baseball Corp.,29 for example, the District Court for the Northern District of California decertified a FSLA class of minor league baseball players on the ground that plaintiffs’ trial plan brought to light “the variations among class members as to the types of activities in which they engaged and the time spent performing those activities [that] ma[d]e th[e] case distinguishable from Tyson Foods.”30 There, plaintiffs’ work activities were not significantly uniform; many players testified that they had discretion as to their work schedule, including deciding when they arrived at the stadium before games and when they left after games.31 Under those circumstances, the district court found that “[r]ather than merely filling in ‘evidentiary gaps’ in a situation where all of the employees were similarly affected by a uniform policy, Plaintiffs here are attempting to paper over significant material variations that make application of the survey results to the class as a whole improper.”32 Other post-Tyson Foods decisions also suggest that defendants may have a strong argument against the use of statistical evidence in class actions based on allegations of fraud or deceit. For instance, in United States v. Vista Hospice Care, Inc.,33 the District Court of the Northern District of Texas, declined to allow plaintiffs to present representative sampling evidence to prove fraud in a False Claims Act case. There, the court rejected plaintiffs’ attempt to use representative sampling and extrapolation to show liability where plaintiffs alleged that a chain of nursing homes submitted claims for Medicare patients who did not qualify for hospice care.34 The United States sought to introduce evidence of the total number of false claims based on a sample of 291 hospice patients. Citing Tyson Foods, the court noted that permissibility of statistical sampling turns on the degree to which the evidence is reliable in proving or disproving the elements of the relevant cause of action.35 Accordingly, because the underlying determination of eligibility for hospice is inherently subjective, patientspecific, and dependent on the judgment of physicians, the court held that statistical sampling was not permissible to establish fraud in submitting claims for ineligible patients.36 Sampling Evidence Must Be Admissible in an Individual Action to Establish Class Liability Tyson Foods also reaffirmed another maxim that is highly favorable to defendants — that class actions are no different than individual actions. Indeed, in the majority opinion, Justice Kennedy emphasized that, under the Rules Enabling Act, the class action device cannot be used to enlarge the substantive rights of class members thereby depriving defendants of their right to litigate individual defenses. The result in Tyson Foods is also in line with that premise—representative and statistical evidence was admissible to prove Tyson’s liability, because, and only because, sampling would have been admissible in an individual lawsuit to prove liability and damages. District courts after Tyson Foods have declined to certify classes of plaintiffs where the statistical Weil, Gotshal & Manges LLP 9 Employer Update August 2016 evidence offered to prove liability would not have been admissible in an individual action. In In re Celexa & Lexapro Marketing & Sales Practices Litigation,37 for example, the District Court for the Northern District of Massachusetts rejected a class action plaintiff’s attempt to rely on statistical sampling to prove liability in a case involving allegations that that defendants, prescription drug companies, had induced doctors into prescribing antidepressant drugs for off-label uses which resulted in plaintiffs, several different insurance companies, overpaying benefits for unnecessary prescriptions.38 During the class certification stage, the insurance companies argued that the case did not require individualized determinations about whether a particular prescription was induced by fraudulent off-label marketing and cited Tyson Foods in support of the use of statistical evidence to establish but-for causation and liability.39 The district court, however, refused to certify the class for lack of predominance, finding that the insurance companies could not rely on statistical evidence because they could not make the threshold showing identified in Tyson Foods—that each plaintiff in the proposed classes could have relied on the aggregate statistical evidence to prove but-for causation in an individual action.40 The district court found the regression analysis that the expert used suggested that all promotional activities, i.e., both fraudulent and non-fraudulent, increase drug sales but it did not specifically suggest that such a correlation is a reasonable proxy for the relationship between fraudulent promotions and fraudulently induced sales. In the district court’s view, that flaw precluded both an individual plaintiff and the putative class from using the statistics as evidence of but-for causation.41 Therefore, the argument that courts should not relax the rules for class actions could prove indispensable to defendants in challenging sampling evidence. Defendants Should Insist on a Trial Plan In his Tyson Foods concurrence, Justice Roberts suggested that, on remand to the district court, plaintiffs needed to set forth a viable plan to allocate the jury’s award to ensure that uninjured class members would not recover—and if plaintiffs could not set forth such a plan, the class should be decertified.42 This underscores the importance of pushing plaintiffs to submit a trial plan early in the proceedings. The trial plan should describe, among other things, what statistical evidence, if any, plaintiffs plan to present and how damages will be allocated among class members if they prevail at trial. Although there is no trial plan requirement in Rule 23, the advisory committee notes to the 2003 amendments emphasize that “[a] critical need is to determine how the case will be tried,” and observe further that an “increasing number of courts require a party requesting class certification to present a ‘trial plan’ that describes the issues likely to be presented at trial and tests whether they are susceptible to class-wide proof.”43 Some jurisdictions have held that a trial plan is a requirement prior to certification,44 but in other jurisdictions a trial plan is simply something that be requested via interrogatory45 or that a judge may order in her discretion as part of her inherent authority over management of the case. A trial plan may illustrate to the judge that the use of sampling evidence would prevent the defendant from raising individualized defenses and thus poses due process concerns or manageability issues. Indeed, demanding plaintiffs produce a trial plan based on Justice Robert’s concurrence in Tyson Foods can be a powerful tool to bring to light arbitrary sampling evidence or practical manageability problems in adjudicating their claims on a class wide basis.46 Conclusion While at first glance Tyson Foods appeared to be a win for the plaintiff bar, a deeper analysis of the opinion shows that it actually laid the foundation for defendants to challenge the use of statistical evidence in class actions to defeat class certification. Therefore, in the long run, this narrow victory for plaintiffs may prove a big win for defendants. Weil will continue to monitor lower court interpretations of Tyson Foods and will continue to report any new developments concerning the use of sampling evidence in class actions. 1. Tyson Foods, Inc. v. Bouaphakeo, 136 S. Ct. 1036 (2016). 2. Wal-Mart Stores, Inc. v. Dukes, 131 S.Ct. 2541 (2011). 3. Comcast Corp. v. Behrend, 133 S. Ct. 1426 ( 2013). Weil, Gotshal & Manges LLP 10 Employer Update August 2016 4. Bouaphakeo v. Tyson Foods, Inc., No. 5:07-CV-04009-JAJ, 2011 WL 3793962 (N.D. Iowa Aug. 25, 2011). 5. Brief for Appellant at 20, 2013 WL 663846 (No. 12-3753). 6. Wal-Mart Stores, Inc. v. Dukes, 131 S.Ct. 2541 (2011). 7. Id. at 2561. 8. Tyson Foods, Inc. v. Bouaphakeo, 135 S.Ct. 2806 (2015). 9. Id. at 1041 (internal quotation marks and citations omitted). 10. Id. at 1049. 11. Anderson v. Mt. Clemens Pottery Co., 328 U.S. 680 (1946). 12. Tyson Foods, Inc., 136 S. Ct. at 1047-49 (2016). 13. Id. at 1047. 14. Id. at 1046. 15. Id. at 1048. 16. Id. 17. Id. 18. Tyson Foods, Inc., 136 S. Ct. at 1057 (Thomas, C., dissenting). 19. Id. 20. Tyson Foods, Inc., 136 S. Ct. at 1049. 21. Tyson Foods, Inc., 136 S. Ct. at 1053 (Roberts, J. concurring). 22. Spokeo, Inc. v. Robins, 136 S. Ct. 1540, 1543 (2016), as revised (May 24, 2016). 23. See id. 24. See id. at 1036 (internal quotation marks and citations omitted). 25. Villalpando v. Exel Direct Inc., No. 12-CV-04137-JCS, 2016 WL 1598663 (N.D. Cal. Apr. 21, 2016). 26. See id. at *8. 27. See id. 28. See id. 29. Senne v. Kansas City Royals Baseball Corp., No. 14-CV- 00608-JCS, 2016 WL 3940761 (N.D. Cal. July 21, 2016). 30. See id. at *51. 31. See id. 32. See id. 33. United States v. Vista Hospice Care, Inc., No. 3:07-CV- 00604-M, 2016 WL 3449833, at *11 (N.D. Tex. June 20, 2016). 34. See id. 35. See id. 36. See id. 37. In re Celexa & Lexapro Mktg. & Sales Practices Litig., No. CV 13-13113-NMG, 2016 WL 3102004 (D. Mass. June 2, 2016). 38. See id. at *9-10. 39. See id. at *10. 40. See id. 41. See id. 42. Tyson Foods, Inc., 136 S. Ct. at 1053 (Roberts, J. concurring). 43. Fed. R. Civ. P. 23(b)(3) advisory committee’s notes (2003). 44. E.g., Sandwich Chef of Texas, Inc. v. Reliance Nat. Indem. Ins. Co., 319 F.3d 205, 220 (5th Cir. 2003) (“Certification of a class under Rule 23(b)(3) requires that the district court consider how the plaintiffs’ claims would be tried.”); see also In re MTB Prods. Liab. Litig., 209 F.R.D. 323, 351-53 (S.D.N.Y. 2002) (refusing to certify class based on unmanageable trial plan); but see Chamberlan v. Ford Motor Co., 402 F.3d 952, 961 n. 4 (9th Cir.2005) (declining “Ford’s suggestion that the district court’s failure to adopt a trial plan or to articulate how the class action would be tried was an abuse of discretion ... [since] [n]othing in the Advisory Committee Notes suggests grafting a requirement for a trial plan onto the rule.”); Olson v. Tesoro Refining & Marketing Co., 2007 WL 2703053, *7 (W.D. Wash. Sep. 12, 2007) (“[T]he Court rejects defendant’s contention here that a trial plan is required for class certification”). 45. See 1 McLaughlin on Class Actions § 3.7 (5th ed. 2009) (providing sample interrogatory). 46. See e.g., Senne v. Kansas City Royals Baseball Corp., No. 14-CV-00608-JCS, 2016 WL 3940761, at *51 (N.D. Cal. July 21, 2016) (rejecting plaintiffs’ proposed trial plan and decertifying the class where the statistical sampling method identified in the plan was not a viable option to show liability).