Regression analysis is a plaintiff's best friend

Werdebaugh v. Blue Diamond Growers, 2014 WL 2191901,  No. 12–CV–2724 (N.D. Cal. May 23, 2014)

I’m going to try, with probably limited success, to summarize results rather than reasoning for most of this consumer class action case in which certification of a California class was granted, and focus on the interesting damages model bits.  Werdebaugh sued Blue Diamond for listing the sweetener on its almond milk as “evaporated cane juice” instead of sugar, and using “All Natural” when in fact the products allegedly contained synthetic ingredients.

Werdebaugh had standing to bring his claims for damages, but not for injunctive relief.  His deposition testimony clearly showed that he wouldn’t have bought Blue Diamond almond milk had he known about the alleged misbranding, which sufficed.  He testified that he was unfamiliar with the product, but “the ‘all natural’ label stood out” to him. He “stood at the shelf and saw this packaging and picked it up, read the labels, and made the purchase,” and the “all natural” label was a substantial reason why.  Werdebaugh further testified that he was concerned about seeing “evaporated cane juice” on the ingredients list, but did not know that it is “the equivalent of table sugar.”

Blue Diamond argued that it was implausible for Werdebaugh to believe that the products didn’t contain added sugars, since the nutrition facts panel that the almond milk contained 20 grams of sugars, and anyway he didn’t know what “evaporated cane syrup” or “dried cane syrup” was either so his purchase decision wouldn’t be affected by using those terms either.  The court disagreed.  The ingredients wouldn’t necessarily have indicated the presence of added sugars; Werdebaugh testified that he thought the 20 grams of sugar in the product he purchased was not added sugar, but rather was “an actual squeezed element of an almond that naturally occurs.”  Though he testified that “dried cane syrup” wouldn’t have affected his purchase decision, that just showed that he didn’t know what it was.  He’d still introduced sufficient evidence of reliance on both statements.

Ascertainability: not a problem, especially limited to a California class.  The class was defined based on objective criteria, and that was enough.

Numerosity: but of course.

Commonality:  Blue Diamond argued that what’s material varies from consumer to consumer.  “The law is to the contrary…. Whether Blue Diamond’s label statements constitute material misrepresentations does not depend on the subjective motivations of individual purchasers, and the particular mix of motivations that compelled each class member to purchase the products in the first place is irrelevant.”  Blue Diamond also argued that the allegedly deceptive labeling statements were not specifically regulated and, therefore, were not material, since the only official guidance came from “non-binding FDA policy statements.” But at this stage, the only question was whether materiality was a common question.  Finally, Blue Diamond argued that “All Natural” had no common definition and thus was not susceptible to common proof.  But cases using that reasoning generally involved representations that differed for each class member, such as representations made by doctors prescribing drugs. Here, the alleged misrepresentations were the same to each calss member, so the objective inquiry into whether “a reasonable consumer would attach importance” to Blue Diamond’s label statements was a question common to the class.  And, unlike the Astiana case cited by Blue Diamond, where over 90 different products with different ingredients and different ad campaigns were challenged, Werdebaugh was only challenging seven products, all based on their inclusion of the same ingredient (potassium citrate).  Blue Diamond didn’t contend that differences in its products’ labels would cause prospective consumers to understand the representations differently.

Typicality: Blue Diamond objected to including products Werdebaugh didn’t buy—he only bought Blue Diamond Almond Breeze Shelf Stable Chocolate Almond Milk.  But every other product included in the class definition was an almond milk product, and each bore one or both of the same misbranded label statements. They were different flavors, but the legal theory was identical for all claims. That’s typicality.

Blue Diamond then argued that Werdebaugh’s claims were atypical because he didn’tread or review the back label, which contained two of three “All Natural” statements.  But Blue Diamond didn’t persuade the court that he needed to read and rely on all alleged misrepresentations; even if he didn’t read two repetitions of “All Natural,” “he read the third, and Defendant provides no reason to distinguish between the three statements.”  Reading it on the package once was enough for typicality.

Nor were defenses unique to Werdebaugh likely to become the focus, since under California consumer protection law “individual experience with a product is irrelevant” because “the injury under the UCL, FAL and CLRA is established by an objective test.” Regardless of his particular motivations for purchase, “he shares with the proposed class the same interests in determining whether Blue Diamond products were deceptively advertised and labeled.”

Adequacy: yep.

Predominance: not for a nationwide class per Mazza, but for a California class.

Blue Diamond then argued that there was no predominance because Werdebaugh hadn’t identified an appropriate damages model, an argument the court analyzed in detail.  An appropriate damages model under the Supreme Court’s recent Comcast ruling must measure only damages attributable to the defendant’s conduct.

Restitution was available under California consumer protection law to compensate the purchaser for the difference between a product as labeled and the product as received.  For this, Werdebaugh needed a damages methodology that could determine the price premium attributable to Blue Diamond’s use of the labeling statements “All Natural” and “Evaporated Cane Juice.” 

The court rejected a full refund model, because it was wrongly based on the assumption that consumers receive no benefit whatsoever from purchasing the accused products.  

Then, the court turned to Werdebaugh’s expert’s price premium model, which compared the price of the accused products to allegedly comparable products without the challenged statements and calculated the entire price difference as restitution.  The court also rejected this model. “[The expert] has no way of linking the price difference, if any, to the allegedly unlawful or deceptive label statements or controlling for other reasons why allegedly comparable products may have different prices.”  The comparison product (a Whole Foods house brand) itself included the objectionable ingredient potassium citrate; the label listed “organic evaporated cane juice” as an ingredient until 2013; and the alternative was currently priced the same as the challenged Blue Diamond version at Whole Foods stores in San Francisco and Palo Alto. Werdebaugh’s deposition testimony also indicated that consumers typically pay a premium simply by buying from Whole Foods, which the model didn’t account for.  The price premium model just calculated what the price difference was, without tying it to a legal theory; it didn’t account for factors that might lead consumers to prefer Blue Diamond over “other identical products.”  These factors could include “brand loyalty or quality differences between brand and generic products.”  (If they’re identical, why the brand loyalty?)

However, a regression model saved the day (or ruined it, if you’re Blue Diamond).  By controlling for commonly recognized factors associated with sales—“price of the product, prices of competing and complementary products, income, advertising, seasonality, and regional differences”—as well as by taking into account differences in sales of the products before and after the “All Natural”/“evaporated cane juice” labeling, a more precise measure of damages could be calculated.

Blue Diamond argued that this model would raise individual issues not subject to common proof because consumers experience price variation across “products, sales channels, retailers, geographic regions, and time.”  For example, “the price for the top selling shelfstable Blue Diamond almondmilk product in 2012 varies by 26% across sales channels, by 31% across the top three retailers, and by 40% across cities in California.” But that’s kind of what the regression controls for, and Blue Diamond didn’t explain how these differences would affect the measure of damages—“price changes within regions that correspond to the introduction and/or removal of the allegedly misleading label statements.”  Even if a carton costs $4 in San Francisco and $3 in Sacramento, that wouldn’t necessarily affect damages—the price might have risen by a constant amount or by a percentage, but either way it could be calculated classwide.  (This would also be easier with a class limited to California instead of nationwide.)  If Blue Diamond introduced evidence that the price increase attributable to the allegedly false labeling would be 20% in San Francisco and 2% in Sacramento, the model might well fail under Comcast, but Blue Diamond didn’t explain why that would happen.

Blue Diamond also argued that retail prices varied and that it didn’t set retail prices, and that using a weighted average price measure would undercompensate some consumers and overcompensate other, but that didn’t address the way the model purported to measure price changes from the allegedly false advertising.  The model controlled for regional differences.  Comcast doesn’t require calculations to be exact, only that a model supporting a damages case has to be consistent with the theory of liability, both at certification and at trial. Even with regional differences, the regression model was sufficiently precise under Comcast, and could control for non-liability-producing factors.

Blue Diamond said that the methodology was too vague, but at this stage what is needed is a workable model, not necessarily a model that actually works. “Comcast did not articulate any requirement that a damage calculation be performed at the class certification stage,” so the fact that the expert had yet to actually run the regressions and provide results, and would need discovery to do so, wasn’t necessarily fatal.  Blue Diamond couldn’t succeed by attacking the specific variables of the regression the expert posed, if the tool of regression analysis itself was appropriate, as it was.

Superiority: yep.