Allstate Insurance Company “insured” a last week in an off the clock class action pending in Los Angeles Superior Court, vindicating employers’ argument that plaintiffs cannot simply intone the magical incantation of “statistical sampling” as a means of collective proof in a class action. Rather, plaintiffs must proffer a detailed and manageable trial plan that relies on sound statistical science. Likening Plaintiff’s trial plan to a house built on a poor foundation, Judge John Shepard Wiley rejected the statistically unsound trial plan because it would be “an enduring source of grief.”
After almost nine years of litigation, Judge Wiley granted Allstate’s motion to decertify both an off-the-clock and wage statement class because none of the multiple trial plans suggested by Plaintiff complied with the requirements in the California Supreme Court’s 2014 decision in Duran v. U.S. Bank National Association or last month’s U.S. Supreme Court decision in Tyson Foods, Inc. v. Bouaphakeo.
Over the past two years, Plaintiff offered several trial plans based on statistical sampling and extrapolation suggested by two different experts. The court, however, found that each of the plans failed to comply with sound statistical methodology and were “premised on invalid logic.” Recognizing that a 95% confidence interval and a 5% margin of error is the common convention, the court roundly criticized Plaintiff’s expert who proposed an 84% confidence interval and anywhere from a 10-20% margin of error. The court also rejected Plaintiff’s plea to let him proceed with trial and enter a directed verdict if he could not prove his claims because such a plan was “doomed to be an expensive waste of time.” Under proper sampling methodologies, the case would be unmanageable at trial as the sample size would require testimony from at least 495 class members.
Significantly, the court’s decision implicitly rejects the Plaintiff’s argument that not all members of his proposed survey need to testify at trial. The decision is therefore powerful ammunition to counter plaintiffs’ oft repeated argument that a “sample of a sample” is sufficient to testify at trial. If sound statistical methodology requires a sample of 495 class members in order to extrapolate the results to the larger class consistent with the proper confidence interval and margin of error, then all 495 class members need to testify at trial so that the jury can determine their credibility and assess their testimony. Plaintiffs cannot simply propose that their expert will testify at trial as to the results of a survey of the sample. If this means that the trial will be unmanageable, then the case should be decertified.
Although Plaintiff argued that Tyson Foods was a “game changer,” the court found Tyson Foods to be entirely consistent with Duran. The court recognized that Tyson Foods and Duran prohibit the use of statistically inadequate evidence such as that presented by Plaintiff. Although representative proof sometimes can be used in a certified class action, statistical evidence only can be used if the proof is reliable.
This case provides employers with several important “take-aways.” Defense counsel should aggressively challenge a plaintiff’s proposed trial plan to ensure that the trial plan is statistically reliable. Additionally, neither Tyson Foods nor Duran stands for the proposition that statistical sampling and surveys can be used to prove liability in every case. Whatever the supposed benefits of a class action may be, they cannot defeat a defendant’s right to due process. Trial plans must be tailored to the specific facts of the claims alleged and an unmanageable trial plan that is not scientifically sound should be rejected.