Statistician as Star Chamber

McCleery v. Allstate Ins. Co., No. B282851 (D2d1 Dec. 14, 2018)

Plaintiffs in this wage and hour class action put forward a trial plan where the only evidence on liability would be statistical analyses of results obtained from an anonymous, double-blind survey of a sampling of class members. The survey seems like it was analytically rigorous, but that’s still not going to fly.

First, it didn’t take into account important factual aspects of the case, such as the fact that plaintiffs were employed by contracting companies whose services were used by, among others, the two insurance companies that are defendants in the case. Without any breakdown of time worked for each defendant (or some other non-defendant customer of the contractor) core liability questions—questions like “did this employee work more than eight hours for any particular defendant?”—can’t be answered. 

More fundamentally, even if the expert was doing a good job as a statistician, the survey still depended on questions like: From 2005 to 2008, how many times did you fail to take a meal break less than thirty minutes long? How many weeks in 2008 did you work more than forty hours? The veracity of the respondents and the quality of their recollection were, essentially, presumed. And because all the survey participants were anonymous, even to the expert, the multiple hearsay inside his opinions can’t be impeached or cross-examined. Which is a pretty big problem; it flies in the face the way trials are supposed work in the common law adversarial system. 

No doubt, that experts can sometimes rely on inadmissible hearsay. And it is also true that statistical techniques have a valid role to play in wage-and-hour class actions. But an expert can’t be a conduit that dumps huge amounts of hearsay into evidence while shielding it from any meaningful adversarial testing. As the court explains, no case “suggests a trial may be conducted solely on the evidence of an expert witness relying on an anonymous double-blind survey, no matter how scientific the survey may be.”

Denial of class cert affirmed.