Singleton v. Fifth Generation, Inc.
August T. Horvath and Rebecca Kirk Fair
Why It Made the List
For the past five years, class-action plaintiffs have struggled to construct damages models in advertising cases that comport with the Supreme Court’s decision in Comcast v. Behrend.1 In Comcast, the Supreme Court held that the predominance requirement for class certification of Federal Rule of Civil Procedure 23(b) requires that at the class certification stage, a plaintiff seeking class-wide damages on a common basis must proffer a damages case that (1) can demonstrably calculate class-wide damages by common proof and (2) is consistent with its liability case. The Supreme Court directed that District Courts “must conduct a rigorous analysis to determine whether that is so.”2
The Comcast showing has proven challenging to satisfy in consumer class actions whose theory of injury and damages is that a class of consumers paid a price premium because they were deceived by the advertising or marketing of a product, relative to what they would have paid if they had known the truth about the product. Foods and beverages have been among the most common subjects of such suits in recent years. In 2017, several District Courts issued important decisions on this question of whether the damages case at issue satisfies the Rule 23(b)(3) predominance requirement, and any of three or four of them could have been our Top Case. We chose Singleton v. Fifth Generation, Inc.,3 because it was one of the better elucidations of the issues. Below, we also discuss the other key 2017 and early 2018 decisions in this area, both in cases involving foods and beverages and those dealing with other consumer products where similar price-premium class actions have been litigated.
In Comcast, the Supreme Court held that “a model purporting to serve as evidence of damages in [a] class action must measure only those damages attributable to [plaintiff’s theory of liability]. If the model does not even attempt to do that, it cannot possibly establish that damages are susceptible of measurement across the entire class for purposes of the predominance requirement of Rule 23(b)(3).”4 Thus, as one requirement, “any model supporting a plaintiff’s damages case must be consistent with its liability case.”5 The other requirement of the damages model under Comcast is some level of specificity. Comcast advised that at the class certification stage, “calculations need not be exact.”6 However, subsequent cases have held that, while the plaintiff “does not need to ‘implement or test his methodology at the class certification stage, he must still provide sufficient detail about the proposed methodology to permit a court to determine whether the methodology is suitable to the task at hand.’”7
Comcast was an antitrust case. Later cases held that Comcast applied in cases where the plaintiffs seek restitutionary damages.8 In a false-advertising case where plaintiffs seek restitution of an alleged price premium for the falsely advertised product attribute, post-Comcast cases have held that “plaintiffs must be able to show that their damages stemmed from the defendant’s actions that created the liability” 9—i.e., from the specific advertising or marketing claim that is challenged. Some courts have held that the price premium is the difference between the market price actually paid by consumers and the market price that would have prevailed if consumers had not been deceived.10 Where this claim promotes a product attribute that is allegedly not as advertised, the plaintiffs’ damages methodology must convincingly isolate the falsely advertised attribute from any other reason why consumers might pay more for the product.11 Damages for the entire class must be calculable by a common method, although it is still permissible for each individual’s damages to require individual calculation under that common formula.12
To satisfy the Comcast requirements, plaintiffs in consumer class actions have proposed several different damages estimation methods including conjoint and regression analyses. Conjoint analysis, used alone as a means of estimating restitutionary damages in class actions, has been rejected by some courts on the basis that it estimates only the amount that consumers say they were willing to pay for the falsely promoted feature, and not how much more they actually paid.13 Other courts, however, have accepted conjoint analysis, even without supplementation by some other method to calibrate it to real-world prices, as a means of estimating a price premium.14 Others have approved conjoint analysis when used “in conjunction with” a marketplace-data-based method such as a regression analysis on real-world prices to tie the conjoint results to actual market price.15
The defendant, Fifth Generation, Inc., does business as, and markets a product named, Tito’s Handmade Vodka. The case concerns the defendant’s representation that its vodka is “handmade” and “crafted in an old fashioned pot still.” The plaintiff, a New York resident, sued in 2015, purporting to represent a class of all persons in New York who purchased Tito’s Handmade Vodka after April 12, 2012. The plaintiff asserted that Fifth Generation’s characterizations of its vodka are false because the vodka actually is made in large, automated modern stills and bottling facilities at the rate of 500 cases per hour, with little or no direct human involvement.
Key allegations in the case are that consumers prefer products made in small quantities and with direct human involvement, believing them to be of higher quality. Accordingly, it is alleged, consumers will pay more for a product having those attributes than they would pay for an otherwise indistinguishable product that does not have them. This allegedly enabled Fifth Generation to sell Tito’s Handmade Vodka at higher prices in comparison to competing products. Consumers who purchased Tito’s Handmade Vodka allegedly would not have purchased the vodka, or would have paid less for it, if they did not see and believe the defendant’s express claims that it was handmade in an old-fashioned pot still and the associated implied claim that it was made in small batches.
The plaintiff’s theory of liability in cases like Singleton is that consumers actually paid more for the falsely marketed product than they would have paid for that product, or for a competing product, absent the false marketing. Courts have rejected the theory that it constitutes cognizable economic damages for consumers merely to have received less perceived value from a falsely marketed product than they thought they were getting, even though they would have paid just as much for the product if their perception of it was accurate.16 Plaintiffs must therefore prove an actual price premium paid by consumers that would not have been paid in the but-for world of fully truthful marketing. The task of establishing this actual price premium falls to social scientist expert witnesses working in the marketing, economics, and/or survey research fields. These experts are assigned to isolate and quantify the economic value associated with the feature or features that are allegedly falsely promoted. In this case, the hypothesis they test is that because consumers believe Tito’s Handmade Vodka is made in small batches with direct human intervention in old-fashioned stills, the vodka commands a market premium over a hypothetical product that is otherwise identical, but that consumers understand is made in vast quantities in massive, automated, largely unattended distilling plants.
Expert witnesses face several methodological challenges in performing these studies. First, experts must ensure that their proposed methodologies follow best practices with respect to sampling, recruitment, question design, and analyses. Therefore, at the class certification phase of the case, experts must give a complete description of their proposed approach and demonstrate that their approach and design complies with best practices in social science research. In particular, when designing a survey, the expert must detail several survey elements such as the target population, sample selection, screening questions, survey questionnaire, data collection and analysis, and other survey-related procedures. The failure to properly detail such considerations implies such considerations may not be made in the final survey.
There are further methodological challenges in both conjoint analyses and in other methods, such as hedonic modeling, which must be addressed when examining false labeling claims. The proposed research design must be able to measure both the materiality of the alleged misrepresentation(s) and be able to quantify the damages (price-premium paid) as a result of the alleged misrepresentation(s). The proposed design must allow for causal inferences and isolate the price premium paid specifically due to the alleged misrepresentations. Specifically, an expert witness must be able to clearly isolate the allegedly false misrepresentations in her or his survey design. In the case of Singleton, the survey design must be sufficiently precise to isolate the effect of Tito’s “handmade” claims without causing respondents to focus on a claim they otherwise would not have considered as part of their purchasing decision. Similarly, if an expert intends to rely upon hedonic modeling or analyses of transaction data, one needs to ensure that the comparisons being made are appropriately tied to the allegations and the results are not conflated by other changes in the marketplace. Ultimately the evidence proffered by an economist or survey expert must be carefully designed in a manner consistent with academic standards.
Although cases like Singleton are decided on the basis of an evaluation of the expert damages evidence proffered by the plaintiff, courts caution that the analysis is not a consideration of the admissibility of the expert’s testimony under Daubert v. Merrill Dow Pharmaceuticals, Inc.17 Some courts, including District Judge Brenda K. Sannes in Singleton, separately address Daubert admissibility motions within the same decisions in which they address the Comcast Rule 23 analysis, but the criteria are different. An expert opinion can be of adequate reliability and comport with the practice of experts in the field, and thus be admissible under Daubert, but not match the plaintiff’s theory of liability or calculate damages for all class members with a common proof, and thus miss the mark set in Comcast. Thus, in addition to complying with best practices of social science research, the expert must adequately describe their proposed research design and provide enough information to demonstrate its applicability to the plaintiff’s theory of liability.18
Simultaneously before the court in September 2017 were the plaintiff’s motion for class certification, filed in February 2017, defendant’s motion for summary judgment, filed in May, and the parties’ challenges to each other’s expert witness testimony under Daubert in connection with the class certification motion. The court noted that it is not completely settled whether District Courts need to conduct a Daubert evaluation when expert testimony is presented at the class certification stage, but it seems likely that the Supreme Court expects courts to do so as part of the Rule 23 analysis. The court therefore undertook a Daubert evaluation of the proffered expert testimony, limited to the purpose of whether the expert testimony was admissible for purposes of establishing the Rule 23 requirements.
None of the key expert witnesses was excluded on Daubert grounds. The court permitted the testimony of a food industry consultant who reviewed available vodkas and opined that the main reason for the price premium of Tito’s over two selected competitors was the handmade representation, even though the expert’s method was “somewhat shaky.” It also found the plaintiff’s key statistical expert, who performed the conjoint survey and hedonic regression analysis, adequate in reliability. Only one of the plaintiff’s experts, whose testimony simply expressed support for the methods of the other two, was excluded. The opinions of all of the defendant’s experts, primarily critiquing those of the plaintiff, were admitted. On the class certification motion, the court quickly found that the requirements of Rule 23 other than predominance—numerosity, commonality, typicality, adequacy of representation, and ascertainability—were met. The court rejected the plaintiff’s claim for injunctive relief on the ground that because the representative plaintiff now knows about the allegedly deceptive representation, he is not at any risk of future injury and lacks standing to seek an injunction. These preliminaries set up the main event: the evaluation of whether the plaintiff’s damages claim satisfied the Rule 23(b)(3) predominance requirement.
As a warm-up, before addressing damages, the court considered the defense argument that the materiality of an advertising claim, which is a necessary element of a New York GBL 349/350 violation, is inherently subjective and therefore not predominant. The court ruled that materiality in advertising cases is an “objective inquiry,” and that materiality is a property of an advertising claim. Either a claim is “likely to mislead a reasonable consumer acting reasonably under the circumstances” or it is not, the court concluded.19
Rule 23 subsection 23(b)(3) has become the focus of the suitability for an alleged course of conduct and type of damages for class treatment because, courts have found, it sets up a predominance requirement “far more demanding” than Rule 23(a)’s requirement of commonality, which can be satisfied just by showing a generally similar pattern of injury caused by common alleged conduct.20 Under Rule 23(b)(3), the court inquires whether “the common, aggregation-enabling, issues in the case are more prevalent or important than the non-common, aggregation-defeating, individual issues.”21 The plaintiff’s expert witness in Singleton presented three different damages estimation methods, an increasingly common strategy in price-overcharge class actions. Plaintiffs and expert witnesses in recent cases have proposed to use multiple methods either in the alternative or in combination as a way of solving the difficulties associated with estimating a class-wide real price premium. The court in Singleton named the three methods the Comparator Model, the Conjoint Analysis, and the Hedonic Regression.
In the Comparator Model, the expert testifying for the plaintiff proposed to compare the defendant’s Tito’s vodka against high-end competitors Skyy and Smirnoff. The court rejected this analysis because it was not satisfied that the expert had established that the competitor vodkas were comparable with Tito’s on dimensions of quality and other measures. These differences would be entangled with the “handmade” claim, which would be impossible to isolate for purposes of computing a price premium.
In the Conjoint Analysis, a different expert testifying for the plaintiff proposed to conduct a specialized survey and appropriate statistical analysis to estimate consumers’ willingness to pay a premium for the Tito’s “handmade” claim. Survey respondents would be asked which they would purchase among competing sets of hypothetical vodkas, varying on different attributes including the “handmade” claim and various price points. By analyzing a large number of survey respondents’ comparisons of these sets, conjoint analysis can (one might say) distill each respondent’s willingness to pay a premium for the “handmade” claim, and compute an average for the entire class. The court was skeptical of this approach. Interestingly, the court did not object to the use of a conjoint analysis, as some prior courts have done, on the basis that a conjoint analysis measures only willingness to pay and not an actual market premium; on the contrary, this court held that under the Ninth Circuit’s conception of restitutionary damages, the economic harm to the plaintiff is that “the consumer has purchased a product that he or she paid more for than he or she otherwise might have been willing to pay if the product had been labeled accurately.”22 Instead, the court was concerned that “a conjoint analysis with two hypothetical products is too detached from the facts of this case to measure damages tied to Plaintiff’s theory of liability.”23 The court also criticized the expert’s “vague” elucidation of the means by which real-world pricing information would be tied to the willingness-to-pay analysis.24
In Singleton the Court noted that the plaintiff’s expert’s proposed design failed to sufficiently isolate the causal relationship between the “handmade representation” and price premium paid by consumers. By not specifying whether his conjoint survey respondents would view generic bottles of vodka or the actual Tito’s product, the expert failed to account for the premium respondents may attribute to the Tito’s brand name with or without the corrective statement. The failure to propose a research design that addresses the specific claims of harm and the plaintiff’s causal theory of injury suggest that the issues of injury and damages must be assessed based on individualized inquiry.25 The Singleton court also determined that the expert’s description of his proposed conjoint analysis “lacks the sort of step-by-step detail necessary to evaluate whether the methodology is workable.”26
Finally, the court rejected the proposed hedonic regression analysis, in which the expert for the plaintiff proposed to analyze data on purchases of all vodkas in New York, using attributes of the vodkas as predictors of the final prices. The outcome of the regression analysis would be regression coefficients that predict how much each attribute, including the “handmade” claim, contributes to a vodka’s final price. Here, the Singleton court acknowledged that hedonic regression analysis had been approved by courts in similar cases, but it criticized the expert for making “little attempt to identify a relevant set of product attributes” beyond a few suggestions, and in particular, for providing no means of quantifying the product quality.27
Based on these critiques of the proposed expert analysis, the Singleton court held that common issues do not predominate over individual ones because “although materiality may be proven with class-wide evidence, the issues of injury and damages will devolve into individualized inquiries because Plaintiff has failed to propose a sufficiently-detailed and suitable model to measure the alleged price premium for Tito’s vodka due to the ‘handmade representation.’”28 Judge Sannes accordingly denied the request for class certification.29
On March 22, 2018, the mediator in the Singleton case advised the court that the matter “has been settled, or is in the process of being settled,” and accordingly the court dismissed the case.30 The amount of any monetary payment associated with the settlement was not disclosed. The court’s September 27, 2017, decision will thus be the final word on class certification in that case.
Other 2017-18 Decisions
Several other cases decided in 2017 and thus far in 2018 bear on the same issues as Singleton. We focus first on the cases involving marketing of foods and dietary supplements.
Hughes v. The Ester C Company.31
In Hughes v. Ester C, the plaintiffs challenged the marketing of Ester-C vitamin supplements as “The Better Vitamin C.” Judge Pamela Chen of the Eastern District of New York first denied the plaintiffs’ motion for class certification on September 30, 2016, and then denied it again upon reconsideration on July 21, 2017. The plaintiffs in the case proposed a damages model that combined a conjoint analysis with a hedonic regression analysis. In the 2016 decision, the court had ruled that the proposed analysis was not viable because “The Better Vitamin C,” standing alone, is puffery, and becomes actionable only in the context of more substantive claims, also present on the product label, that provide consumers with the factual basis of the “better” claim. The plaintiffs asserted that approximately 150 other Ester-C vitamin products exist that contain, individually or in combination, all ingredients and attributes of the challenged product except that they did not make the “Better Vitamin C” claim. The plaintiffs claimed that this group of competing products furnished adequate comparators for use in their conjoint analysis and/or hedonic regression. The court disagreed, finding that the plaintiffs’ analysis could then only find the purported value of the “Better Vitamin C” claim alone. This, the court found, was precisely what the plaintiff did not need to establish, because the “Better Vitamin C” claim became actionable only in the context of substantive factual allegations.
Both the litigants’ and the courts’ reasoning in the Hughes case is somewhat tangled because of the plaintiffs’ attempts to recast their analysis between the two decisions, but the bottom line is that the court considered the plaintiffs’ damages methodology not to match their theory of injury. The plaintiffs failed to consider exactly what made their case actionable (i.e., not merely the “Better Vitamin C” claim, but the combination of that claim with the claims of other, less subjective attributes) and to tailor their damages methodology to finding the impact of that combination.
The plaintiffs sought an immediate appeal of the denial of class certification to the Second Circuit Court of Appeals, but this was denied, so the case has proceeded as an individual action. In early 2018, the parties briefed NBTY’s motion for summary judgment, with the plaintiffs apparently planning to continue the case until some final resolution is reached that will permit them to appeal the class certification ruling.
Briseno v. ConAgra Foods, Inc.32
The plaintiffs in Briseno v. ConAgra alleged that ConAgra misleadingly marketed its Wesson brand cooking oils as “100% Natural” when, in fact, the oils were extracted from genetically-modified organisms (GMOs), which they contended are not “natural.” Judge Margaret Morrow of the Central District of California denied the plaintiffs’ initial motion for class certification on August 1, 2014, but granted leave to file an amended class certification motion. She then granted the amended motion and certified eleven statewide damages classes on February 23, 2015, and the Ninth Circuit affirmed the grant of class certification on January 3, 2017, with scant discussion of the damages methodology. The plaintiffs in the case proposed a damages methodology that combined a hedonic regression with a conjoint analysis. First, the plaintiffs proposed to calculate the price premium associated with the “100% Natural” claim with a hedonic regression. Then the plaintiffs proposed a conjoint survey and analysis conducted by another expert to determine the percentage of the “100% Natural” price premium specifically attributable to consumers’ beliefs that the Wesson oils are not derived from GMOs.
In a lengthy opinion, with significant discussion about the damages methodologies in both the context of Daubert challenges and Rule 23(b)(3) requirements, the court concluded that the hedonic regression in combination with the other expert’s conjoint analysis would satisfy Comcast. The court determined that the expert’s previously inadmissible proposed hedonic regression model was now admissible because it contained a preliminary regression model with more details including the twenty product attributes used in the model, which included the brand of oil, oil variety, the size of the bottle, promotional prices, the time period, the “natural” claims at issue in the case, and other product label claims. The court, however, again found that the hedonic regression model on its own was incapable of calculating the price premium associated with consumers’ belief that the oils are not derived from GMOs, the theory of liability, because it only measured the price premium attributable to the “100% Natural” label.
The conjoint expert testifying for the plaintiffs proposed to use consumer surveys to isolate the percentage of the price premium specifically attributable to a customer’s belief that “100% Natural” means that the oils contain no GMOs and then she proposed to take the total price premium generated in the hedonic regression and multiply it by the percentage determined from the conjoint analysis to produce a damages figure that would be solely attributable to ConAgra’s misleading marketing. The court found that this hybrid damages model would quantify damages associated with the plaintiffs’ theory that they were misled to believe that the oils contained no GMOs. Of particular interest is that the court accepted that in this hybrid damages model it is sufficient to account for supply and market factors in the hedonic regression model; the conjoint analysis does not need to as well. In addition, the court was satisfied with the conjoint expert’s explanation as to her decision to limit the survey to six attributes, with the safeguards provided by pilot tests and focus groups which would ensure that each attribute selected reflects a significant meaning and is an attribute that consumers would consider in making purchasing decisions, and with the idea that the expert could adjust the attributes if necessary following the focus groups and pilot tests.
Morales v. Kraft Foods Group, Inc.33
In the Morales case, the plaintiffs alleged that Kraft’s Natural Cheese Fat Free Shredded Cheddar Cheese was not, in fact, natural cheese. The court had certified a class in 2015, which the defendants moved to decertify along with their motion for summary judgment in early 2017. By that time, an expert witness testifying for the plaintiffs had performed a conjoint analysis that tested various combinations of cheese attributes, including the label claim that Kraft’s product was “natural cheese,” to estimate the value to consumers of the “natural cheese” claim. The analysis assigned an average value of 74.7 cents to the claim, finding that 26% of respondents valued the “natural cheese” attribute at greater than $1. On its motion to decertify and to strike the testimony of the expert for the plaintiffs, the defendants largely launched a Daubert challenge, critiquing various aspects of the plaintiffs’ methodology. Holding, as several courts have done, that a conjoint analysis is a form of “survey,” and that deficiencies in a survey’s methodology go only to the weight of a survey and not its admissibility, the court rejected the Daubert argument.34 Just as in Singleton, however, the court differentiated the Daubert analysis from the Rule 23(b)(3) predominance inquiry, and the predominance analysis came out very differently.
The court had originally held that the proposed damages methodology was appropriately tied to the plaintiffs’ theory of liability and supported the proposed class. The defendants ultimately prevailed on their argument that the expert for plaintiffs “did not determine the price premium that Kraft charged for the ‘natural cheese’ label, but rather measured customers’ subjective willingness to pay, an academic and irrelevant exercise that is not consistent with Plaintiffs’ theory of liability.”35 After an extensive analysis of past cases that had used conjoint analysis in various applications, the court concluded that a conjoint analysis “does not provide any insight into the money received by the Defendant in connection with the sale of the Product. Rather, it bears only on the claimed loss to Plaintiffs. Thus, the evidence provided by Plaintiffs about their potential willingness to pay a premium due to the use of the ‘natural cheese’ label is insufficient to establish a basis for calculating restitution.”36 After the decision, the plaintiffs moved for recertification of the class, while at the same time, the parties entered into settlement discussions. The case docket went quiet in August of 2017, after several sealed filings, and it appears that the parties may have settled.
Zakaria v. Gerber Products Co.37
In the Zakaria case, the plaintiffs alleged that Gerber falsely marketed its “Good Start Gentle” infant formula as the “1st and only routine formula to reduce risk of developing allergies.” On March 23, 2016, the court certified a California damages class. On November 29, 2016, the defendant filed a motion to decertify the damages class as well as a motion for summary judgement. The defendant’s motions for decertification and summary judgement were granted. The court found that the proposed damages methodology did not measure the price premium paid by the class members due specifically to the “1st and Only” claim and other misleading statements concerning reducing the risk of developing allergies.
The plaintiff’s proposed damages methodology consisted of a “choice-based conjoint analysis” where the survey participants were presented with several hypothetical packages of Gerber infant formula with corresponding, hypothetical prices which did not directly correspond with the actual market prices for the at-issue product. Given the use of hypothetical prices instead of actual market prices, the court found that the analysis did not sufficiently reflect the actual price premium or the actual market conditions in which the product was sold. Additionally, the court took issue with the limited sample size and the lack of confirming studies (including hedonic regression) or market data. Ultimately, the court decided that the plaintiff’s analysis did not show the amount of money the defendant received as a result of the alleged misrepresentations and was insufficient to calculate damages. The plaintiff appealed these decisions to the Ninth Circuit Court of Appeals.38 Briefing took place in early 2018.
Cases in Other Industries:
In re Dial Complete Marketing and Sales Practices Litigation.39
In the Dial case, the plaintiffs challenged marketing statements about the antibacterial properties of “Dial Complete” soap: that it “Kills 99.99% of Germs,” is “#1 Doctor Recommended,” and “Kills more germs than any other liquid hand soap.” In December 2015, the court denied the plaintiffs’ initial motion for class certification finding that the plaintiffs’ proposed damages methodology failed to give sufficient detail for the court to determine whether damages could be calculated on a class-wide basis. However, the court allowed the plaintiffs to file an amended motion to address the deficiencies, which they did on June 24, 2016.
As part of the amended motion for class certification, the plaintiffs’ expert proposed a choice-based conjoint analysis where participants were shown hand soap profiles with five attributes including price and the at-issue attributes. The prices reflected the real-world prices observed in preliminary research including market research data. The results of the conjoint survey were then inputted into a market simulation in order to determine the difference between the equilibrium market price of the soap with the allegedly misleading claims and the soap without the allegedly misleading claims. Although the court noted that the plaintiffs’ methodology was “imperfect in some respects, weak in others, and subject to challenge on cross-examination,” it found the model capable of calculating the price premium for the allegedly falsely-claimed features and establishing the full extent of damages on a class-wide basis. Therefore, the court granted the plaintiffs’ amended motion for class certification. Dial Corporation sought leave for interlocutory appeal to the First Circuit Court of Appeals. This request was denied on July 31, 2017.40 As of early 2018, the case continues to be litigated.
Kurtz v. Kimberly-Clark Corp.41
The plaintiffs in the Kimberly-Clark matter, which includes several consolidated actions brought against Kimberly-Clark and other vendors and manufacturers of flushable toilet wipes, challenge the misleading characterization of the wipes as “flushable” because the wipes allegedly clogged household plumbing. The plaintiffs filed a motion for class certification in February 2015. The court found that the plaintiffs provided enough evidence to allow the certification of an injunctive class noting that “an injunction prohibiting defendants from labeling their products as ‘flushable’ and ‘safe for sewer and septic systems’ would provide a single solution, applicable to each class member.” The court also certified two New York damages classes finding that the “[p]laintiffs have shown that proposed methodologies can probably be used to learn common answers to common questions.”42
In order to demonstrate that damages could be calculated on a class-wide basis, the plaintiffs’ expert proposed a hedonic regression to calculate the proportion of the price attributable to the “flushable” characteristic, and he provided a preliminary list of the product attributes on which he would rely. He also stated in his report that his analysis would include evidence from the defendants’ business records, industry resources, and independent market research data from companies like Nielsen. The expert also provided details of two alternative approaches, contingent valuation and conjoint analysis, to determine how much consumers value product attributes. The court found that all three methods proposed by the expert to determine the price premium attributable to the “flushable” representation were adequate for class certification. The decision to accept the plaintiffs’ damages methodologies may have been easier for the court given that New York law provides for statutory damages of $50 to each class member for each time a defendant violates the New York General Business Law by a sale. As a result, the court noted that the “instant cases’ battle of experts on price differential is largely beside the point. Once an injury is established, statutory damages can be precisely calculated for each class member.”43 In June 2017, the Second Circuit Court of Appeals granted the defendants’ petition for interlocutory appeal. Briefing of the appeal was completed in early 2018.44
The plaintiff in Singleton failed to clear the Rule 23(b)(3) hurdle despite proposing a combination of damages methodologies—in particular, conjoint analysis and hedonic regression analysis—that has passed muster in some other false-advertising class actions. Obviously, given the Singleton decision and other cases above, such as Zakaria, that also proposed to use one or both of these methods, merely saying the words “conjoint analysis” and “hedonic regression” does not, at least in all instances, persuade courts that plaintiffs can estimate damages on a class-wide basis. What more is needed?
First, the analysis needs to be carefully thought out and proposed in some detail. Judge Sannes in Singleton repeatedly referred to the slapdash quality of the damages analysis proffered in the case. The court noted that the expert witness who proposed both the conjoint analysis and the hedonic regression had not originally been offered as a damages expert and had not been relied upon in the plaintiff’s original motion for class certification; rather, the relevant expert opinions “are tacked on to the end of his second expert report, which was submitted with Plaintiff’s reply in further support of the class motion.”45 The expert’s descriptions of the methodologies were at various points characterized by the court as lacking sufficient detail, vague, “skeletal analysis,” “bare-bones,” and “superficial.” The court was left without a clear idea of how the analyses were supposed to work or to do what they were supposed to. Indeed, the court’s criticism of conjoint analysis for its use of “hypothetical products,” which actually are a fundamental tool in conjoint analysis, suggests that the expert witness for the plaintiff failed to educate the court about the nature of the proposed analysis. In contrast, in both Dial and Briseno v. ConAgra Foods, following initial failures and the denial of initial certification motions, the courts found the plaintiffs’ experts provided sufficient details in subsequent filings. Specifically, incremental details on conjoint design and preliminary regression results, which helped to isolate the claims, were found to provide sufficient evidence that both classes were certified.
Second, the expert witness must address the difficult issues and unique circumstances in every case. One common example is the lack of availability of sufficient marketplace data to conduct a hedonic regression analysis, or the tendency of the challenged attribute to co-vary with some other attribute, making the challenged attribute difficult to isolate. Here, a special problem was presented that is relatively unique to this case: the role of the variable “quality” of vodkas. The court raised the expert’s failure to even suggest how vodka quality would be operationalized as among the fatal flaws with both the Comparator Analysis and the Hedonic Regression, while noting with some skepticism that the Conjoint Analysis proposed merely to sidestep this variable. Quality presents a special research issue for a product such as an alcoholic beverage, where quality is not merely the product of a bundle of on-off or quantifiable attributes. Quantifying overall vodka quality is not an impossible research task, but it is a challenge, and one that the court felt that the expert for the plaintiff made little effort to resolve. Similarly, in Briseno v. ConAgra Foods, the courts indicated that it was necessary to marry a hedonic regression, which could determine a price premium for an “All Natural” label, with a conjoint analysis, which could further isolate the value of no-GMO within that overall context.
Finally, to the extent that multiple methodologies are proposed to be used together, they must be integrated. The Singleton court criticized the proposed conjoint analysis for failing to calculate the actual amount that consumers overpaid using real-world pricing and sales data.46 That function might have been served by either of the two alternative analyses proposed by the plaintiffs—the Comparator Analysis and the Hedonic Regression, which is really just a more sophisticated form of the Comparator Analysis utilizing controls. Apparently, the plaintiff and the expert witness he retained failed to explain convincingly how the other analyses might be used to calibrate or triangulate the results from the conjoint analysis, lending it some real-world validation. To be sure, such an exercise is not easy, and it requires convincing the court of the feasibility of at least two different types of analysis, plus some way in which they could be married. This issue of the intersection between consumer willingness to pay and market outcomes was discussed in nearly every matter. Failure to demonstrate that an expert can determine but-for prices was generally found to be sufficient to deny certification, as in the Morales case, the Zakaria matter, and the outcomes in initial Dial and Briseno v. ConAgra Foods decisions. Even in Kimberly Clark, it was the availability of statutory damages in that matter which appears to have resolved the tension between consumer values and market prices.
August T. Horvath is a partner and co-chair of the Advertising & Marketing Law practice at Foley Hoag LLP in New York.
Rebecca Kirk Fair is a Managing Principal at Analysis Group in Boston.
- 569 U.S. 27, 133 S. Ct. 1426 (2013).
- Id., 133 S. Ct. at 1433.
- 5:15-CV-474 (BKS/TWD) (N.D.N.Y. Sept. 27, 2017).
- Comcast, 133 S. Ct. at 1433.
- Weiner, 2010 U.S. Dist. LEXIS 79647, at *27.
- See Pulaski & Middleman, LLC v. Google, Inc., 802 F.3d 979, 989 (9th Cir. 2015).
- See In re POM Wonderful LLC, No. ML 10-02199-DDP-RZX, 2014 U.S. Dist. LEXIS 40415 (C.D. Cal. Mar. 25, 2014).
- In re NJOY, Inc. Consumer Class Action Litig., No. 14-428, 2016 U.S. Dist. LEXIS 24235 (C.D. Cal. Feb. 2, 2016); Brazil v. Dole Packaged Foods, LLC, 660 F. App’x 531, 534 (9th Cir. 2016); Werdebaugh v. Blue Diamond Growers, No. 12-cv-02724-LHK, 2014 U.S. Dist. LEXIS 173789 (N.D. Cal. Dec. 15, 2014).
- See In re Scotts EZ Seed Litigation, 304 F.R.D. 397, 413 (S.D.N.Y 2015) (“Courts routinely reject price premium methodologies under Comcast when the proposed methodologies do not attempt to isolate the premium due only to the allegedly misleading marketing statement.”)
- See Levya v. Medline Indus., 716 F.3d 510, 514 (9th Cir. 2013).
- See In re NJOY, Inc. Consumer Class Action Litig., 120 F. Supp. 3d 1050, 1119 (C.D. Cal. 2015); Apple, Inc. v. Samsung Elecs. Co., No. 11-cv-01846-LHK, Savedra v. Eli Lilly & Co., 2014 U.S. Dist. LEXIS 179088 (C.D. Cal. Dec. 18, 2014).
- See Guido v. L’Oreal, No. 2:11-cv-01067-CAS, 2014 U.S. Dist. LEXIS 165777 (C.D. Cal. July 24, 2014); In re Myford Touch Consumer Litig., No. 13-cv-03072-EMC, 2016 U.S. Dist. LEXIS 179487 (N.D. Cal. Sept. 14, 2016); Odyssey Wireless, Inc. v. Apple, Inc., No. 15-cv-01735-H-RBB, 2016 WL 7644790, at *9 (S.D. Cal. Sept. 14, 2016); In re Lenovo Adware Litig., No. 15-MD-02624-RMW, 2016 U.S. Dist. LEXIS 149958 (N.D. Cal. Oct. 27, 2016).
- See In re ConAgra Foods, Inc., 90 F. Supp. 3d 919, 1026 (C.D. Cal. 2015).
- See, e.g., Weiner v. Snapple Bev. Corp., No. 07-civ-8742, 2010 U.S. Dist. LEXIS 79647, at *17 (S.D.N.Y. Aug. 5, 2010) (“Only by showing that plaintiffs paid more as a result of [defendant’s false advertising] can plaintiffs establish the requisite elements of causation and accurate injury under [N.Y. G.B.L.] § 349.
- 509 U.S. 579 (1993).
- 5:15-CV-474 (BKS/TWD) (N.D.N.Y. Sept. 27, 2017), p. 22.
- Arguably, in making this finding, the court mistook the definition of “materiality,” which is normally expressed as whether a represented fact is likely to influence a consumer’s purchasing decision, for the definition of deceptiveness, which is what the court quoted. In this instance, the distinction may not matter much, because the court could just as easily have found, using the standard definition of materiality, that the claim would likely influence the purchasing decision of a reasonable consumer, and therefore would still be an objective property of the claim, not a subjective perception of each consumer.
- Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 624 (1997).
- In re Petrobras Securities, 852 F.3d 250, 270 (2d Cir. 2017), quoting Tyson Foods, Inc., 136 S. Ct. 1036, 1045 (2016).
- Singleton, 2017 U.S. Dist. LEXIS 97433 at *74, quoting Pulaski & Middleman, 802 F.3d at 988-89, in turn quoting Kwikset Corp. v. Superior Court, 51 Cal. 4th 310, 329, 120 Cal. Rptr. 3d 741, 246 P.3d 877 (2011).
- 2017 U.S. Dist. LEXIS 170415 at *64, citing Ault v. J.M. Smucker Co., 310 F.R.D. 59, 67 (S.D.N.Y. 2015).
- Id. at *64-65.
- Id. at *69.
- Id. at *67-68.
- Id. at *69-70.
- Id. at *70.
- 5:15-CV-474 (BKS/TWD) (N.D.N.Y. March 22, 2018) (Dkt. 172, Order of Dismissal by Reason of Settlement).
- No. 2:2012-CV-0041 (PKC) (E.D.N.Y. July 21, 2017).
- 674 Fed. Appx. 654, No. 15-55727 (9th Cir. Jan. 3, 2017) and In re ConAgra Foods, Inc., 90 F. Supp. 3d 919, No. CV 11-CV-05379 MMM (AGRx) (C.D. Cal. Feb. 23, 2015). The Supreme Court denied the petition for writ of certiorari in the case on October 10, 2017. See 138 S. Ct. 313, No. 16-1221.
- No. 2:14-CV-04387, 2017 U.S. Dist. LEXIS 97433 (C.D. Cal. June 9, 2017).
- Id. at *43.
- Id. at *66.
- Id. at *75-76.
- No. 2:15-cv-00200-JAK-E (C.D. Cal. Aug. 9, 2017).
- No. 0:17-cv-56509 (9th Cir.).
- MDL Case No. 11-md-2263-SM (D.N.H. March 27, 2017).
- Carter v. Dial Corp., 17-8009 (1st Cir.).
- Nos. 14-CV-1142, 14-CV-4090, 15-CV-2909, 15-CV-2910, 15-CV-2928, 15-CV-4579 (E.D.N.Y. March 27, 2017).
- 2017 U.S. Dist. LEXIS 44576, at *554.
- Id. at *551.
- No. 17-1856 (2d Cir.).
- 2017 U.S. Dist. LEXIS 170415, at *66 n.24.
- Id. at *65, citing Ault, 310 F.R.D. at 67.