The whole point of the Supreme Court’s decision in Wal-Mart Stores, Inc. v. Dukes was to put an end to “trial by formula” class actions that stack the deck against defendants. Lower courts, unfortunately, haven’t gotten the message. And that is a serious threat to defendants’ due process rights. Fortunately, the Supreme Court will return to the issue next term in Tyson Foods, Inc. v. Bouaphakeo.

There is a whiff of parody to the case. The plaintiff class consists of about 1,300 workers at Tyson’s Storm Lake, Iowa, pork-processing plant who say that Tyson failed to compensate them for the overtime that they spent putting on and taking off protective gear before and after their shifts. The class was certified as presenting “common” fact questions despite that the plant has some 420 job classifications, each of which has different protective requirements, not to mention that Tyson provides additional gear that employees may choose to wear—so even workers in the same department or at the same position may wind up wearing different equipment.

Logically, one would expect the plaintiffs to present evidence of the amount of time that they each spent putting on and taking off gear, compare that to the work and pay records kept by Tyson, and then show that they weren’t properly compensated for any time they worked over 40 hours in a given week, which the Fair Labor Standards Act sets as the trigger for overtime. After all, that’s how it would work in an individual suit.

But that’s nothing like what happened here.

Rather than put on evidence of individual employees’ changing times, an expert for the plaintiffs measured several workers, whose changing times ranged from a few seconds to ten minutes—reflecting what even the expert acknowledged was “a lot of variation.” No matter, the expert averaged everything out, added in a few minutes to account for various factors, and arrived at “average” changing times of 18 minutes for “processing” workers and 21.25 for “slaughter” workers. At this point, a second expert added those averages to individual employees’ time records, identified weeks in which workers were due overtime pay, and then calculated damages, arriving at a class-wide figure of over $6 million in unpaid overtime wages. But as the second expert conceded, that figure was very sensitive to changes in the averages; reduce them by just a small amount, and hundreds of workers drop out of the class altogether, because their time drops below 40 hours for every week or they were already adequately compensated for any overtime. Indeed, although the jury ultimately found for the employees, it didn’t buy the average times and reduced the damages award to less than $3 million.

In case that’s not clear: based on the plaintiffs’ formula evidence, the jury necessarily believes that a substantial number of class members suffered no injury at all. Yet as class members, they’ll still receive damages—at the expense of other workers who might actually have been underpaid. So the jury verdict is incoherent, but you can’t really blame the jury: the class itself is the real problem, throwing together workers with so many different kinds of jobs and such different circumstances that there aren’t any truly common fact questions among them. Meanwhile, due to the use of common, formula-based evidence, Tyson was denied any ability to challenge its liability to and the damages of individual class members, only the plaintiffs’ formula. Even with most of its defenses off the table, Tyson’s challenge to the formula was mostly successful. Absent the formula, it’s possible that Tyson might have prevailed on many individual claims or even all of them—there’s no way to know.

That’s not how class actions are supposed to work. Rule 23, which sets the procedure for class actions, is a procedural device and is not supposed to alter the substantive rights of plaintiffs or defendants. In particular, that means that a class action can’t be used as a shortcut by “extrapolating” damages and liability based on a formula. The Court explained in Dukes that this requires that plaintiffs’ claims turn on a common contention “that it is capable of classwide resolution—which means that determination of its truth or falsity will resolve an issue that is central to the validity of each one of the claims in one stroke.” (For example, a common issue in a product-liability class action might be whether a given product was defectively designed.) When that requirement is loosened, the result is trial by formula, as individual issues are decided based on class-wide evidence, whether or not that evidence says anything at all about particular class members’ claims. The quaint notion of having to support legal claims with probative evidence goes out the window. Relieving plaintiffs of the burden of proving their cases is, needless to say, something more than merely “procedural.”

Trial by formula not only violates Rule 23 and its statutory basis, the Rules Enabling Act, but also raises serious due process concerns. “Due process requires that there be an opportunity to present every available defense,” the Supreme Court has explained. It follows that “[a] defendant in a class action has a due process right to raise individual challenges and defenses to claims, and a class action cannot be certified in a way that eviscerates this right or masks individual issues.” When aggregate litigation procedures abridge a defendant’s ability to mount the same defenses that it could bring in individual suits, something is seriously wrong.

The Court needs to make clear that trial by formula is off-limits, whether a case is brought as a class action or as a “collective action” under the Fair Labor Standards Act or a combination of the two. Defendants have a right to defend themselves, and that’s possible only when a class action satisfies Dukes’s “one stroke” rule. That’s as true in FLSA wage and hour cases—which many of the lower courts seem to believe are an exception to due process requirements—as in any other. Skipping past that requirement offends fundamental principles of justice and perverts the substantive law being enforced by extending liability to persons and circumstances that Congress never intended.