Tuesday, March 29, 2011

Oral Argument Transcript in Wal-Mart v. Dukes

Scotus The Supreme Court heard arguments today in Wal-Mart v. Dukes, the gender discrimination class action that has been pending for ten years. Based on the transcript, it's hard to predict what the Court will decide. The Justices seemed to split their questions between the procedural questions about the class action rules, questions about the practicalities of trying the case, and questions that go to the merits of the discrimination claim. I'm still processing what I've read, but here are my preliminary impressions.

Counsel for Wal-Mart argued that the only policy the company had was to not discriminate, that any policy which allowed for excess subjectivity by managers couldn't be a "practice," presumably for pattern and practice or disparate impact liability, and that to the extent that there might have been a policy towards that discretion, the policy affected individuals in different ways, so there was no commonality. He also argued that the claims for backpay were really monetary rather than equitable, and were the primary remedy sought, so certifying the class under the section of Rule 23 that allows for an easier process for claims for injunctive relief was improper.

Counsel for the class argued that parts of the policies on pay and promotion gave too much discretion to managers, but there was a strong corporate culture that filled in the exercise of that discretion in a way to prime stereotyping. Thus, the women in the class were all subject to the same policy, which made their claims sufficiently common. He also argued that what was really at stake was the claim for injunctive relief, including back pay, and that Wal-Mart's HR database was thorough enough that a formula could be devised to determine who in the class was entitled to back pay and the amount of that backpay so that individual hearings would not be necessary.

Many of the questions of counsel for Wal-Mart focused on when company headquarters should be on sufficient notice that it should have known that its policies were allowing discrimination to occur. More focused on the possibility of separating out the monetary claims into a separate class and what the effect would be on women who did not want to be part of the class.

Many of the questions for the plaintiff-class's counsel focused on the tension within the plaintiffs' theory of discrimination and commonality. How could a policy of too much managerial discretion affect all of the members of the class in a common way. And more fundamentally, how did the allegation of the strong corporate culture then mesh with the delegation of too much discretion. More questions were skeptical of the use of statistics in general, particularly in the remedial phase. Similarly, many questions focused on how Wal-Mart would be able to raise the pattern and practice defense that it didn't discriminate against a particular woman without individual hearings.

I'm terrible at predictions, and this one seems particularly challenging. It seems safe to say that Justices Ginsburg, Breyer, Sotomayor, and Kagan would hold that some formation of a class could move forward. It's equally safe to say that Justice Scalia, and probably Justice Alito would say the class could not. I couldn't tell conclusively how Justice Roberts or Justice Kennedy might be leaning. Justice Roberts was concerned about fairness to the women not before the court, but he also seemed to suggest that companies should be on notice at some point of a pattern of discriminatory actions by lower level managers such that inaction becomes a policy. He may have been skeptical, though, that the point had been reached here. Justice Kennedy, too, was thinking along those lines, and referenced Monell, a case about when action or inaction by a municipality becomes a policy the municipality is liable for. But both he and Justice Roberts said they had a hard time pinpointing what the plaintiffs were alleging the policy was.

I'll predict a win for Wal-Mart because of those concerns, but I could easily be wrong.



Employment Discrimination | Permalink

TrackBack URL for this entry:


Listed below are links to weblogs that reference Oral Argument Transcript in Wal-Mart v. Dukes:


A quick read of the transcript leaves me with a general conclusion that most of the members of the Court do not understand much about the use of probability statistical techniques, especially the power of multiple regression, in supporting the decision to reject the hypothesis that sex and salary/promotion are unrelated.

Justice Kennedy claims the plaintiffs are making two contradictory claims. But I don't think that is true. On one hand, Wal-Mart has become the huge enterprise that it is by having exquisite data collection that goes to Bentonville and that is analyzed to impose controls throughout every part of its operation. I have been told that Bentonville knows when a freezer turns on and off at a store in Shanghai and that if the performance of those freezers is out of line with a model for freezer use of electricity, the word goes down to do something about it. On the other hand, Wal-Mart collects an enormous amount of data about all of its employees everywhere but does not analyze but essentially leave the local managers to make pay and promotion decisions using their discretion without the input from Bentonville beyond the written policy not to discriminate. If multiple regression studies support rejection of the hypothesis that sex is unrealated to pay and promotion decision, then that supports, if not mandates, drawing the inference that sex was related to pay and promotion.

That is the underlying substnative theory of the plaintiffs'case. For Rule 23(a) purposes, that establishes commonality at a level sufficient to meet the first step to certify the class, regardless of size. If backpay is equitable relief available under Rule 23(b)(2), then the class should be certified.

If the case goes to trial and the plaintiffs establish defendant's liability, then all of that data that Wal-Mart collects centrally about every aspect of every employee's story can be used to determine whether or not each woman in the class is entitled to backpay.

So, it is having the data but the failure of Wal-Mart in Bentonville to do anything with all the statistical data in collects about employment because it gives discretion to local managers to make pay and promotion decisions while Wal-Mart exercise careful supervision and control of all other aspects of the operation of each store that establishes disparate treatment liability. All of that data can then be utilized to determine backpay as relief.

Posted by: Mike Zimmer | Mar 29, 2011 8:45:28 PM

My reading of the transcript is a bit more pessimistic than Marcia's and probably closer to Mike Zimmer's, though I would not so much characterize them as misunderstanding the role of statistics but rather they seemed quite unsympathetic to the statistical display. It also seemed that the civil procedure questions were coming from the more liberal members of the Court while the substantive questions primarily came from the conservative wing, which likely does not bode well for the plaintiffs. I think the possible sticking point will be whether the Court can dispose of this case without effectively ending employment discrimination class actions, as I suspect they may have trouble finding 5 votes for the proposition pushed by Wal-Mart that employment discrimination claims that include monetary demands cannot satisfy the class cert requirements.

Posted by: Mike Selmi | Mar 30, 2011 7:05:00 AM

I think the proposed use of statistical modeling in Phase II, rather than individualized hearings, will be the sticking point. The Court will be able to reverse without ending employment discrimination class actions. Under Teamsters, a statistical showing in Phase I - no matter how strong - only creates a rebuttable presumption that any particular individual was the victim of disparate treatment under that unlawful pattern or practice. According to the Teamsters framework, Wal-Mart then has the opportunity to disprove discrimination as to any particular individual. The statistical model for calculating backpay deprives Wal-Mart of that opportunity, and even Justice Sotomayor seemed concerned about that (Transcript, pp. 46-47).

Establishing the pattern or practice at Phase I, even assuming a flawless and powerful regression analysis, justifies only "prospective" relief, according to Teamsters. This is true even though the defendant is a proven wrongdoer after the statistical showing in Phase I. Even if you can categorize backpay as equitable relief that might be available under Rule 23(b)(2), it is not "prospective" relief, as envisioned by Teamsters, but rather "individual relief for the victims of the discriminatory practice." Teamsters says that the defendant has a chance to rebut the presumption as to any individual before that type of relief is awarded.

After reading Tristin Green's draft, I do think it is interesting that everyone involved seems to assume that pointing to a "policy" responsible for the discrimination is required - even if that "policy" is subjective, decentralized decision-making. Why could you not just establish Phase I by saying "we're not sure whether any particular 'policy' is responsible for this outcome, but look at the disparities shown by this regression analysis?"

Posted by: Jason Bent | Mar 30, 2011 8:21:49 AM

Post a comment