Last week, the US Supreme Court affirmed a $5.8 million judgment and upheld the District Court’s decision to certify a class action on behalf of pork-processing workers employed by Tyson Foods Inc., the second largest meat producer in the world. Of note, the Court ruled that workers could use statistical sampling, averages and other statistical analyses to support certification. Given that some of these methods are not currently permitted on certification in Ontario and Canada, those practicing in class actions should consider the possible relevance of the US Supreme Court’s decision to their practices.
In 2011, an Iowa federal jury had awarded workers $2.9 million, which was subsequently raised to $5.8 million, for uncompensated time spent putting on (i.e. donning) and taking off (i.e. doffing) required uniforms and protective clothing or equipment, as well as time spent walking. This judgment was upheld in 2014 by an Eighth Circuit panel. In both instances, the workers relied on a study that analyzed how long various donning and doffing activities took, averaged the time, and then added the estimates to the timesheets of each employee to ascertain which class members worked more than 40 hours a week and the value of damages across the class.
Tyson Foods appealed to the Supreme Court, arguing that the decisions of the lower courts ran counter to such landmark cases as Wal-Mart Stores Inc. v. Dukes, in which Justice Scalia found that a class action should only be allowed where the damages suffered by the class are identical to those of the named plaintiff. The judgment of the lower courts in Tyson relied on “extrapolation and averaging” where there was in fact some variance in the amount of time it took various workers to put on and take off their protective equipment. Thus not all cases were identical.
However, Justice Kennedy, writing for a 6-2 majority, found that the statistical analyses were necessary to fill an evidentiary gap left by the lack of records kept by Tyson Foods regarding don-doff times. If any worker had commenced an individual action, Justice Kennedy found that they would have had to rely on similar studies to prove the additional hours worked and thus damages suffered. The fact that this action was brought on behalf of a class should not bar the plaintiffs from using statistical methods if appropriate.
Justice Kennedy did caution that the acceptance of statistical evidence in cases that do not involve inferences of hours worked within the context of the FLSA should depend on the facts and circumstances of that particular case. Further, Justice Kennedy noted that this decision did not contradict Wal-Mart v Dukes, which did not state that a representative sample could never be used for establishing class-wide liability.
The Supreme Court of Canada has rejected the approach in Wal-Mart v Dukes on two separate occasions. In Pro-Sys Consultants Ltd v Microsoft Corporation, Justice Rothstein, writing for the majority, found that, at the certification stage, expert methodology need only offer a realistic prospect of establishing loss on a class-wide basis. To force Canadian courts to weigh and review conflicting expert testimony in a “robust” and “rigorous” manner, as required by the US Supreme Court in Wal-Mart v Dukes, would be inappropriate given that discoveries in Canada typically occur after certification, rather than before in the US. Further, in Vivendi Canada Inc v Dell-Aniello, Justices LeBel and Wagner, writing for the majority, found that authorization of class proceedings in Quebec is more flexible that the current American approach identified in Wal-Mart v Dukes.
Although Canadian courts rely on the less stringent “some basis in fact” test on certification, it is interesting to note developments south of the border arising out of Wal-Mart v Dukes. Although many will likely see this as a big win for plaintiffs counsel (and the catalyst for a new wave of lawsuits) the decision appears to be quite narrow. Yes, plaintiffs may use statistics to fill evidentiary gaps, but only where similar sampling studies “could have been used to establish liability in an individual action.”
In Canada, most provincial class proceedings statutes allow for the admissibility of statistical evidence in respect of remedies where it would otherwise be inadmissible. In Ontario, where damages are certified as a common issue, sections 23 and 24 of the Class Proceedings Act, 2002 (“CPA”) allow the court to admit statistical information to determine issues relating to the quantum or distribution of a monetary award, and to undertake an aggregate assessment of the quantum of monetary damages owed to the class members.
Although statistical information is permitted, according to the Ontario Court of Appeal, section 24(1)(c) of the CPA does not permit the sampling of class members in order to conduct an aggregate assessment of damages, as was done in Tyson Foods. Although Justice Belobaba recently followed such authority in Nolevaux, he left a detailed critique of the Court of Appeal’s conclusion on random sampling to provide the basis for a partial or aggregate assessment of damages:
In other words, even if random sampling of a handful of class members is the very method that would be used by a trial judge in a parallel mass tort or contract action (to try to monetize the intangible “loss of use” claims), this same approach – the random sampling of the same handful of class member claimants – cannot be used by the common issues trial judge in a class action because this would be in breach of s. 24(1)(c).
I frankly do not understand this interpretation of s. 24(1)(c). With respect, this is not a generous or purposive reading of the CPA.
Justice Belobaba’s critique is fully consistent with the reasoning in Tyson Foods. Canadian class action counsel might wish to consider the effect of Tyson Foods on the certification of common issues concerning aggregate damages going forward.
Class proceedings aside, Tyson Foods does provide a good reminder to Canadian employers of their duty to keep adequate records under both federal and provincial legislation.