Heagney v. John Paul Mitchell Systems

NOT FOR PUBLICATION FILED UNITED STATES COURT OF APPEALS OCT 7 2026 MOLLY C. DWYER, CLERK U.S. COURT OF APPEALS FOR THE NINTH CIRCUIT RANDALL HEAGNEY; RICA No. 25-6818 GUERRERO; KERRIE GONNELLA; D.C. No. JOHN ROHLOFF; JEWEL RULE, 3:23-cv-00687-VC individually and on behalf of themselves and all others similarly situated, MEMORANDUM* Plaintiffs - Appellees, v. JOHN PAUL MITCHELL SYSTEMS, Defendant - Appellant. Appeal from the United States District Court for the Northern District of California Vince Chhabria, District Judge, Presiding Argued and Submitted August 31, 2026 San Francisco, California Before: GOULD, BYBEE, and BRESS, Circuit Judges; Dissent by Judge BYBEE. Randall Heagney and other plaintiffs (collectively, Heagney) brought this class action lawsuit against John Paul Mitchell Systems (JPMS) under California consumer protection laws, alleging that labels on various JPMS hair care products * This disposition is not appropriate for publication and is not precedent except as provided by Ninth Circuit Rule 36-3. misrepresented that JPMS did not conduct animal testing on its cosmetics. The district court certified a class under Federal Rule of Civil Procedure 23(b)(3). We granted JPMS’s petition for permission to file an interlocutory appeal of the class certification order under Federal Rule of Civil Procedure 23(f). We review the “decision to certify a class and ‘any particular underlying Rule 23 determination involving a discretionary determination’ for an abuse of discretion.” Olean Wholesale Grocery Coop. v. Bumble Bee Foods, 31 F.4th 651, 663 (9th Cir. 2022) (en banc) (quoting Yokoyama v. Midland Nat’l Life Ins. Co., 594 F.3d 1087, 1091 (9th Cir. 2010)). “A district court abuses its discretion if it commits legal error, ‘relies on an improper factor,’ ‘omits a substantial factor,’ or ‘commits a clear error of judgment’ when weighing factors.” Black Lives Matter L.A. v. City of Los Angeles, 113 F.4th 1249, 1257 (9th Cir. 2024) (quoting Abdullah v. U.S. Sec. Assocs. Inc., 731 F.3d 952, 956 (9th Cir. 2013)). We have jurisdiction under 28 U.S.C. § 1292(e) and Rule 23(f). Although we discern no error in much of the district court’s reasoning, we conclude that plaintiffs’ proposed classwide damages model was insufficient on the present record. We accordingly vacate the grant of class certification and remand for further proceedings. 1. The district court did not err in applying a classwide inference of reliance because the representations in question appeared on the products 2 themselves, even though on the side or back of the products. Under California’s consumer protection statutes, “[w]hether a statement is likely to deceive members of the public is decided by reference to an objective ‘reasonable consumer’ standard.” Noohi v. Johnson & Johnson Consumer, Inc., 146 F.4th 854, 868 (9th Cir. 2025) (quoting Ebner v. Fresh, Inc., 838 F.3d 958, 965 (9th Cir. 2016)). This standard “requires a probability ‘that a significant portion of the general consuming public or of targeted consumers, acting reasonably in the circumstances, could be misled.’” Ebner, 838 F.3d at 965 (quoting Lavie v. Procter & Gamble Co., 129 Cal. Rptr. 2d 486, 495 (Cal. Ct. App. 2003)). Under California law, when a defendant “communicated the same misrepresentation to all class members . . . the class is entitled to an inference of reliance.” DZ Reserve v. Meta Platforms, 96 F.4th 1223, 1237 (9th Cir. 2024); see also Noohi, 146 F.4th at 869 (noting that classwide exposure was “undisputed” because “the ‘oil-free’ term appears in the Product’s name prominently displayed on the front of the packaging”). Here, JPMS’s no-animal-testing misrepresentations were prominently featured on the product labels and were part of a broader marketing campaign against animal testing. See Kwikset Corp. v. Superior Ct., 246 P.3d 877, 889 (Cal. 2011) (noting that “the marketing industry is based on the premise that labels matter”). Despite JPMS’s argument to the contrary, this case is not more analogous to 3 cases concerning forms of consumer advertising other than product labeling. See, e.g., Mazza v. American Honda Motor Co., 666 F.3d 581, 585–86 (9th Cir. 2012), overruled in part on other grounds by Olean, 31 F.4th at 682 n.32. JPMS further points out we have said that “if a product’s front label is plausibly misleading to reasonable customers, then the court does not consider the back label at the pleadings stage.” Whiteside v. Kimberly Clark Corp., 108 F.4th 771, 778 (9th Cir. 2024). But this line of authority does not address the relevant issue and is inapposite here. 2. The district court did not err in concluding that Heagney’s evidence of materiality was sufficient for purposes of class certification. See Downey v. Pub. Storage, Inc., 258 Cal. Rptr. 3d 290, 300 (Cal. Ct. App. 2020) (citation modified) (“[A] falsity or omission in an advertisement is ‘material’ if a reasonable person would attach importance to that falsity or omission in determining his or her choice of action in the transaction in question.”). A representation “may also be material if ‘the maker of the representation knows or has reason to know that its recipient regards or is likely to regard the matter as important in determining his choice of action.’” Kwikset, 246 P.3d at 892 (quotations omitted). In this case, internal JPMS documents show that branding products as not involving animal testing was a key marketing strategy for the company. Further, an internal JPMS study reported that 71% of surveyed participants said statements that a product was made with “cruelty- free ingredients” was a “most important callout.” 4  JPMS disputes the district court’s interpretation of that figure, arguing that the survey more accurately shows that only 60% of survey participants considered cruelty-free to be a “most important” callout. But even if a somewhat greater percentage of survey participants did not state that cruelty-free was one of the “most important” qualities of a product, a misrepresentation can still be material if a reasonable person would deem the statement important in her decision to buy the product. See Downey, 258 Cal. Rptr. 3d at 300. Indeed, “a misrepresentation need not be ‘the sole or even the decisive cause of the injury-producing conduct.’” Lytle v. Nutramax Lab’ys., Inc., 114 F.4th 1011, 1036 (9th Cir. 2024) (quoting Moore v. Mars Petcare US, Inc., 966 F.3d 1007, 1020 (9th Cir. 2020)) (emphasis in the original). Therefore, even if there was some ambiguity in the survey figures, the district court did not err in finding the materiality requirement met. JPMS’s other efforts to cast doubt on the reliability of the survey it commissioned are likewise without merit. Similarly, the district court did not err in finding that Dr. Hanssens, JPMS’s survey expert, failed to rebut Heagney’s showing of materiality, considering that Dr. Hanssens’s survey used open-ended questions and relied on vague terms like “acceptable.” 3. Based on the showing before the district court at the time it granted class certification, plaintiffs failed to provide sufficient evidence that damages could be measured on a classwide basis. Under our precedent, “class action plaintiffs may 5 rely on a reliable though not-yet-executed damages model to demonstrate that damages are susceptible to common proof so long as the district court finds that the model is reliable and, if applied to the proposed class, will be able to calculate damages in a manner common to the class at trial.” Lytle, 114 F.4th at 1019. But “[t]he fact that a model is underdeveloped may weigh against a finding that it will provide a reliable form of proof,” and “[m]erely gesturing at a model or describing a general method will not suffice to meet this standard.” Id. at 1032. We conclude that based on the record before the district court at class certification, Heagney’s expert, Dr. Macartney, did not put forward a sufficiently reliable damages model. Although Macartney had substantial experience performing hedonic regressions and had previously analyzed conjoint surveys, he had limited experience designing and executing consumer surveys. And his proposed pricing and conjoint analysis models in this case were generic, lacked details, and were insufficiently designed. Among other things, Macartney did not meaningfully analyze economic or market data, determine the variables or data he would use in his models, or select the analytical method—conjoint analysis vs. hedonic pricing—that he would apply. He also did not identify the target population beyond broad generalities, nor did he identify any comparator brands to be shown in his survey. His proposed methods for analyzing damages were skeletal and substantially preliminary in nature. 6  This case is therefore unlike Lytle, where the plaintiffs’ expert had already “fully designed the conjoint analysis and the methodology behind it, including by identifying the target population, analyzing economic data to determine the structure of the market, and specifying the mathematical analysis he will perform on the survey results.” 114 F.4th at 1031. It is also unlike Noohi, in which the expert had similarly “designed the survey methodology and identified target respondent populations.” 146 F.4th at 864–65. Here, we need not consider whether Lytle or Noohi set a “floor” for acceptable damages models because the models proposed by Macartney were substantially more preliminary in nature than the proposed models in those cases. Heagney represents that through further work in connection with expert discovery, Macartney has now fully designed a damages model and calculated classwide damages. But the district court has not considered this revised report. We leave that for the district court to consider in the first instance. VACATED AND REMANDED. 7  FILED OCT 7 2026 Heagney v. John Paul Mitchell Systems, 25-6818 MOLLY C. DWYER, CLERK BYBEE, Circuit Judge, dissenting: U.S. COURT OF APPEALS Although I agree with the majority opinion with respect to the exposure, reliance, and materiality issues raised, I do not agree that “plaintiffs failed to provide sufficient evidence that damages could be measured on a classwide basis.” Majority Op. at 5. Plaintiffs’ expert, Dr. Gareth Macartney, presented a report and deposition testimony describing two methods that could be used to model classwide damages: hedonic regression and conjoint analysis. Dr. Macartney did more than generically describe these well-established methodologies; he identified sources of data that would likely enable him to execute the models and described the varied considerations that would inform the models’ final iterations. Because these showings are sufficient to demonstrate damages’ susceptibility to classwide treatment at the class certification stage, I respectfully dissent. “Before it can certify a class, a district court must conduct a ‘rigorous analysis’ to ensure that the requirements of Federal Rule of Civil Procedure 23 are satisfied.” Noohi v. Johnson & Johnson Consumer Inc., 146 F.4th 854, 862 (9th Cir. 2025) (quoting Olean Wholesale Grocery Coop., Inc. v. Bumble Bee Foods LLC, 31 F.4th 651, 664 (9th Cir. 2022) (en banc)). To facilitate this “rigorous analysis” with respect to whether common damages questions predominate under Rule 23(b)(3), plaintiffs must present the district court with a sufficiently reliable damages model. See Lytle v. Nutramax Lab’ys, Inc., 114 F.4th 1011, 1023–24 (9th Cir. 2024). The model need not be fully executed, but plaintiffs or their experts must do more than merely “gestur[e]” at a model or describe a methodology. Id. at 1024, 1032. District courts’ ultimate goal for purposes of class certification is to decide whether the proposed model is likely to generate common answers at trial. See id. at 1032. In Lytle, we outlined several factors that weigh on this inquiry when an unexecuted damages model is presented to the court. District courts may consider the plaintiffs or their experts’ ability to obtain necessary data, to apply the proposed method to the specific facts of the case, and to eventually execute the model without error. See id. at 1031–33. We also emphasized the novelty of the expert’s proposed method and the timing of the class certification decision as relevant considerations. See id. at 1031, 1033; cf. id. at 1023 (“[I]t is critical to keep in mind that class certification is different from summary judgment.”). Here, Dr. Macartney outlined two potential methods for assessing damages on a classwide basis that are commonly invoked in class actions and that he had significant prior experience with. See id. at 1033 (“As the district court observed, conjoint analysis is a well-accepted technique that is frequently used to establish damages in CLRA actions.”); Briseño v. Henderson, 998 F.3d 1014, 1019 (9th Cir. 2021) (describing a damages expert’s proposed use of hedonic regression). He identified his anticipated data sources, including proprietary datasets, his own market research, and preliminary surveys that he could conduct. And he described 2 the variables he would control for and others he may control for in the hedonic regressions in addition to highlighting potential attributes to be assessed in the conjoint analysis. Ultimately, he did more than simply point to a model or describe general methods. See Lytle, 114 F.4th at 1032. Despite ongoing discovery, he competently “chart[ed] out a path to obtain all necessary data and demonstrate that the proposed method[s] will be viable as applied to the facts of [this] case.” Id. Dr. Macartney’s damages model is unexecuted and “substantially preliminary.” Majority Op. at 6. Nonetheless, Dr. Macartney laid out a viable path to eventually execute finalized damages models, centering his analysis around well- established empirical methods, all while discovery was still underway. Each of these considerations suggests the models presented to the district court were sufficiently reliable for purposes of class certification.1 See Lytle, 114 F.4th at 1031–33. At the very least, these considerations demonstrate that the district court did not abuse its discretion by concluding that damages were susceptible to classwide proof.2 1 The majority distinguishes the models at issue in Noohi and Lytle, arguing that Dr. Macartney’s models here “were substantially more preliminary in nature than the proposed models in those cases.” Majority Op. at 7. However, in those cases, we held that the district courts’ reliance on the incomplete models was appropriate. Noohi, 146 F.4th at 865; Lytle, 114 F.4th at 1034. Consequently, even if Dr. Macartney’s models were substantially less developed than the models in Noohi or Lytle, those cases do not clearly establish the minimum showing necessary for an unexecuted damages model to satisfy Rule 23(b)(3). 2 “We review the question of whether an expert’s damages model ‘is capable of showing class-wide impact, thus satisfying one of the prerequisites of Rule 23(b)(3) 3 Admittedly, it may be the case that Dr. Macartney will ultimately fail to successfully execute the models, but “[t]he speculative possibility that [Dr. Macartney] might slip up in executing his model[s], standing alone, is insufficient to defeat class certification.” Id. at 1033. The majority’s decision today is unlikely to end the parties’ disagreements concerning classwide damages. Once a damages model is fully executed on remand, Defendant will have other opportunities to raise its reliability arguments. Since a “plaintiff may not avoid ultimate scrutiny of the admissibility of their experts’ final opinions” through a renewed Daubert and/or summary judgment motion “simply by declining to develop those opinions in advance of class certification,” it is unnecessary to require heightened scrutiny of damages models at class certification, as the majority does here. Noohi, 146 F.4th at 867 (quoting id. at 1034). For the foregoing reasons, I would affirm the district court’s certification order. I respectfully dissent. of the Federal Rules of Civil Procedure, for an abuse of discretion.’” Id. at 1024 (quoting Olean, 31 F.4th at 663). 4