On September 11, the Central District of California granted summary judgment in a putative class action involving an alleged defect in Honda vehicles. The court’s order provides a thorough dissection of the methodological flaws underpinning the plaintiffs’ expert’s “choice-based conjoint analysis”—a survey-based damages model that plaintiffs’ economist experts commonly employ to artificially inflate alleged damages. [Disclosure: King & Spalding represents Defendants in this litigation.]

• Plaintiffs alleged that all U.S. model year 2011–2016 Honda Odyssey vehicles manufactured, sold, and/or leased by American Honda Motor Company, Inc. and Honda North America, Inc. had flawed transmissions that caused the vehicles to shudder and hesitate during operation.
• In addition to opposing class certification, Honda moved for summary judgment and to strike plaintiffs’ experts, including their damages expert. The damages expert relied on a methodology known as “choice-based conjoint analysis,” which uses “survey questions to ask respondents their willingness to purchase certain products and product attributes for certain prices.” Plaintiffs relied on this expert to establish the difference in economic value between the vehicles with and without the alleged transmission defect, which in turn would be used to calculate the damages allegedly owed to plaintiffs.
• The court agreed with Honda that the expert’s analysis was unreliable and should be excluded under Daubert for two overarching reasons: (i) it accounted only for demand-side considerations and (ii) it used an invalid design that led to irrational results.
  • First, because the model showed only the price consumers purportedly were willing to pay—not the market price that consumers actually pay—it failed to account for supply-side factors, such as costs and the competitive structure of the industry. The court observed that “measuring damages solely through the lens of a consumer’s [willingness to pay] overestimates a consumer’s damages” because “the consumer does not suffer any loss if their reduced [willingness to pay] remains greater than the actual cost of the product.”
    • Moreover, the expert’s attempt to incorporate supply-side variables into his model—such as by using price options as survey choices—was fundamentally flawed because he was not an expert in the valuation of automotive components and his price options were untethered to real-world prices.
  • Second, the court held that the damages model contained numerous other methodological flaws. For instance, while the expert’s “pretest” survey (used to ensure that the questions on the final survey are not misleading, confusing, or suggestive) included thirty-two vehicle attributes, his final survey included just four vehicle attributes. This disparity undermined the study in several respects:
    • Because the pretest survey varied markedly from the final survey, the expert failed to eliminate choices and questions that respondents found confusing or which generated illogical responses.
    • The expert’s reduction of vehicle attributes both failed to track real-world considerations (where consumers face a plethora of choices) and improperly skewed preferences (by artificially inflating the importance of the remaining features). Indeed, the most preferred feature in the survey concerned a fictitious fuel injection system with no real-world counterpart.
    • The high rate of irrational responses in the study (roughly 55%) meant that the survey produced unreliable results that failed to reflect consumers’ actual willingness to pay. For example, one respondent’s choices indicated a willingness to pay $3 million more for a defective vehicle than for a vehicle without the alleged transmission defect. While plaintiffs argued that the expert’s composite average of valuations was reliable—even if certain individual responses were not—the court rejected the notion that the expert’s composite average could be reliable “if the majority of individual responses constituting his average are irrational and unreliable.”
• Because plaintiffs produced no evidence of damages apart from the stricken report, the court held that there was no triable issue of fact on damages and, thus, Honda was entitled to summary judgment.
• The court’s order will be helpful to class action defendants seeking to establish the unreliability of a damages model derived from a conjoint analysis, as many of the same methodological flaws pervade conjoint analyses used in other class actions. In addition to its rigorous scrutiny of the damages model at issue, the court’s order contains a useful survey of case law rejecting analogous conjoint analyses.
• The decision is MacDougall v. American Honda Motor Co., Inc., and you can read more here.