RECOMMENDED FOR PUBLICATION Pursuant to Sixth Circuit I.O.P. 32.1(b) File Name: 24a0260p.06
UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT
┐ IN RE: NISSAN NORTH AMERICA, INC. LITIGATION. │ ___________________________________________ │ ROBERT GARNEAU; NANCY HOUSELL; JEFFREY │ OLKOWSKI; VAUGHN KERKORIAN; DAVID TURNER; │ COURTNEY JOHNSON; SCOTT REEVES; LISA > No. 23-5950 HENDRICKSON; RHONDA PERRY; JANE REEVES; │ MORELA JOVA; KIMBERLY WRIGHT; TODD BURROWS; │ HOSEA BARTLETT; AURELIA FOWLER; JOHN │ HARTWELL; KEITH HUDDLESTON; LAKEITA KEMP; │ MICHELLE BEREDA; ANGELENE HOEFFKEN; SCOTT │ NERI, │ Plaintiffs-Appellees, │ │ │ v. │ │ NISSAN NORTH AMERICA, INC.; NISSAN MOTOR │ COMPANY, LTD., │ │ Defendants-Appellants. ┘
Appeal from the United States District Court for the Middle District of Tennessee at Nashville. Nos. 3:19-cv-00843; 3:19-cv-00854; 3:22-cv-00098; William Lynn Campbell, Jr., District Judge.
Argued: October 31, 2024
Decided and Filed: November 22, 2024
Before: SUTTON, Chief Judge; LARSEN and MURPHY, Circuit Judges.
_________________
COUNSEL
ARGUED: Aaron D. Van Oort, FAEGRE DRINKER BIDDLE & REATH LLP, Minneapolis, Minnesota, for Appellants. John E. Tangren, DICELLO LEVITT LLC, Chicago, Illinois, for Appellees. ON BRIEF: Aaron D. Van Oort, John L. Rockenbach, Anderson C. Tuggle, FAEGRE DRINKER BIDDLE & REATH LLP, Minneapolis, Minnesota, E. Paul Cauley, Jr., No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 2
S. Vance Wittie, FAEGRE DRINKER BIDDLE & REATH LLP, Dallas, Texas, for Appellants. John E. Tangren, Adam J. Levitt, Daniel R. Ferri, Adam Prom, DICELLO LEVITT LLC, Chicago, Illinois, J. Gerard Stranch, IV, Michael C. Iadevaia, STRANCH, JENNINGS & GARVEY, PLLC, Nashville, Tennessee, Benjamin L. Bailey, Jonathan D. Boggs, BAILEY & GLASSER LLP, Charleston, West Virginia, W. Daniel “Dee” Miles, III, BEASLEY, ALLEN, CROW, METHVIN, PORTIS & MILES, P.C., Montgomery, Alabama, Joel D. Smith, SMITH KRIVOSHEY, P.C., Boston, Massachusetts, for Appellees. Philip S. Goldberg, SHOOK, HARDY & BACON L.L.P., Washington, D.C., John M. Thomas, DYKEMA GOSSETT PLLC, Ann Arbor, Michigan, Kyle M. Asher, DYKEMA GOSSETT PLLC, Lansing, Michigan, Donald M. Falk, SCHAERR ǀ JAFFE LLP, San Francisco, California, Jeffrey R. White, AMERCAN ASSOCIATION FOR JUSTICE, Washington, D.C., for Amici Curiae.
OPINION _________________
SUTTON, Chief Judge. Many modern cars, indeed more and more each year, provide automatic alerts when they come too close to another car or an obstacle and will automatically brake to avoid a collision. Some models of Nissan cars, according to the complaint in this case, occasionally activate these alerts at the wrong time, say at a railroad crossing or in a parking garage. A group of car owners sued Nissan for various state-law claims. The district court certified ten statewide classes under Civil Rule 23(b)(3). Because the classes do not meet the material requirements for certification, we vacate and remand for further proceedings.
I.
In 2016, Nissan began equipping its cars with automatic electronic braking systems. Each system has three components: a radar, a control unit, and the brakes. The radar measures the distance to and speed of nearby obstacles and notifies the control unit, which interprets the data. If the data suggest a potential collision, the control unit issues a visual and audible warning to the driver. If the driver brakes, the system helps him by increasing the braking force. If he does not, the control unit alerts the driver again and may brake automatically. The control unit brakes harder as the risk of a crash becomes imminent.
Some parameters constrain the system. The control unit limits heavy braking at speeds above 25 miles per hour for safety reasons. It also stops automatic braking if the driver steers, if No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 3
the driver accelerates, or if the obstacle moves. Drivers may deactivate the automatic braking system at any time.
Fourteen Nissan models at issue here use the ARS410 model radar: the 2017–20 Rogue; the 2017–21 Rogue Sport; the 2019–21 Altima; and the 2020–21 Kicks.
In 2017, some drivers reported “phantom activations” of the automatic braking system at low overpasses, railroad crossings, and parking garages. R.243-37 at 2; R.263-8 at 8; R.263-11 at 9. Engineers deduced that the radar hardware sometimes misread the road ahead when drivers approached an incline or turned on a curve. On a curve, the radar might treat a car in the adjacent lane as an obstacle directly ahead. If driving uphill in, say, a multi-tiered parking garage, the radar might perceive a low-hanging overpass as an incoming wall.
Bound by hardware limitations, Nissan turned to software solutions. In 2018, it released the “S1” update to both the radar’s software and the control unit’s software, notifying dealers and owners of the available modification. In 2019, Nissan released the “S2” software update to the radar and control unit to refine target recognition in parking garages. By August 2022, around 63% of ARS410 radar-equipped Rogue, Rogue Sport, Altima, and Kicks owners had upgraded to the S2 software.
In June 2020, individuals from ten states—California, Connecticut, Florida, Illinois, Massachusetts, Missouri, New York, Ohio, Pennsylvania, and Texas—sued Nissan, alleging that it had sold them defective cars because they contained faulty automatic braking systems. They argued that the defects breached their warranties, constituted fraud, violated their states’ consumer protection statutes, and unjustly enriched Nissan. The plaintiffs moved to certify ten statewide classes of owners or lessees of the fourteen models with this radar.
Two perspectives emerged before the district court. From the plaintiffs’ vantage point, a single radar problem plagues each Nissan car and lies at the center of each state-law claim. Nissan acknowledged that an incorrect prompt from the radar could pose a safety hazard. And the plaintiffs presented the expert report of Steve Loudon, a control systems engineer who identified the radars as the “root cause” of the phantom brake activations. R.241 at 4. The class action process, as they see it, could quickly confirm or deny the alleged malfunction. No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 4
From Nissan’s vantage point, this case involves dozens of distinct products and many versions of those products—and at any rate does not concern one product liability tort action about one alleged defect but dozens of distinct consumer claims, including claims for express warranty, implied warranty, fraudulent concealment, consumer protection, and unjust enrichment. It produced the expert report of Dr. Nathan Soderborg, a statistician who observed that brake-related warranty claim rates varied widely across vehicle model and software update level. Owners of newer models were significantly less likely to seek repairs for their vehicles, which indicated that different claimants experienced different defects based on different combinations of vehicle models and software versions. Because some of the plaintiffs purchased their vehicles before the release of later model versions and updates, Nissan’s knowledge of these defects would vary depending on when each claimant purchased the vehicle. And because the case did not involve one product liability claim but a range of warranty, fraud, and consumer claims, each with distinct individualized elements, it did not lend itself to the types of common questions and answers that Rule 23 was designed to facilitate. As for the plaintiffs’ expert, Loudon, Nissan thought that he lacked the qualifications to opine on automated braking systems and could not “identify any particular defect” anyway. R.264 at 30.
The district court agreed with the plaintiffs and certified all ten classes. A motions panel of this Court granted Nissan’s petition for interlocutory appeal under Rule 23(f). In re Nissan N. Am., Inc., No. 23-0501 (6th Cir. Oct. 24, 2024).
II.
Did the district court properly certify these ten classes? Abuse-of-discretion review applies to the court’s decision. In re Whirlpool Corp. Front-Loading Washer Prods. Liab. Litig., 722 F.3d 838, 850 (6th Cir. 2013). A court abuses that discretion if it relies on clearly erroneous facts, misapprehends the law, or makes a “clear error of judgment.” In re Ford Motor Co., 86 F.4th 723, 727 (6th Cir. 2023) (per curiam).
Because class certification “magnifies the stakes of litigation” by resolving hundreds, if not thousands, of cases at one time, Civil Rule 23 imposes stringent requirements on the process. Id. at 726; Wal-Mart v. Dukes, 564 U.S. 338, 349 (2011). The named claimants must No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 5
“affirmatively demonstrate” the four threshold imperatives of certification. Dukes, 564 U.S. at 350. The class must be sufficiently “numerous” that joinder is unrealistic. Fed. R. Civ. P. 23(a)(1). The combined claims must raise “common” questions of law or fact. Fed. R. Civ. P. 23(a)(2). The class representative’s claims must be “typical” of the claims of other members in the class. Fed. R. Civ. P. 23(a)(3). And the named representative must be in a position to “fairly and adequately” look after the interests of the other class members. Fed. R. Civ. P. 23(a)(4). After that, the plaintiff must satisfy the requirements of one of the types of class actions permitted under Rule 23(b): actions where separate lawsuits would risk inconsistent verdicts or impede third-party interests ((b)(1)), actions where injunctive or declaratory relief is appropriate ((b)(2)), and, the one at issue here, actions where common questions predominate over issues affecting individual plaintiffs ((b)(3)). Pilgrim v. Univ. Health Card, LLC, 660 F.3d 943, 945–46 (6th Cir. 2011).
Plaintiffs “must actually prove—not simply plead—that their proposed class satisfies each requirement.” Halliburton Co. v. Erica P. John Fund, Inc., 573 U.S. 258, 275 (2014). In assessing this proof, the district court must conduct a “rigorous analysis” to determine whether the plaintiff has met each of these requirements. Dukes, 564 U.S. at 351–52; Comcast Corp. v. Behrend, 569 U.S. 27, 33 (2013). A district court abuses its discretion by failing to do so. Ford, 86 F.4th at 727.
As this interlocutory appeal comes to us, it raises three questions. Did the district court properly identify a common question of law or fact under Rule 23(a)(1)? Did the court properly determine that common questions predominate over individual ones under Rule 23(b)(3)? Did the court properly rely on the plaintiffs’ expert in granting class certification without ensuring that he was qualified under Daubert to offer his report? Each question deserves a turn.
A. Common Questions of Law or Fact
Civil Rule 23(a) requires the plaintiff to identify “questions of law or fact common to the class.” In asking whether the trial court conducted a “rigorous analysis” of commonality, Dukes, 564 U.S. at 351, the answer does not turn on whether the district court wrote a long opinion or held multiple hearings. It turns on whether the court examined the material elements of each No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 6
claim and determined which ones, if any, yield a common answer. Else, how could a court know precisely which common factual questions—which elements of the cause of action—it could submit to the jury? And how else could it know whether any common questions predominate over any individualized questions? Commonality drives the initial Rule 23 inquiry. It either establishes the first building block of a proposed class action or exposes an inadequate foundation.
The claimant at a minimum must show that the action will “resolve an issue that is central to the validity of each one of the claims.” Dukes, 563 U.S. at 350. A factual issue will not be “central” to an aggregate dispute if it does not “affect at least one element” of the joined claims. Doster v. Kendall, 54 F.4th 398, 430 (6th Cir. 2022) (quotation omitted), judgment vacated on other grounds, 144 S. Ct. 481 (2023). The district court thus must walk through each cause of action, identify the relevant elements, and evaluate which elements, if any, submit to common answers. To be common, the question must “allow a decisionmaker to reach a yes-or-no answer for the class in ‘one stroke.’” Id. at 430–31 (quotation omitted). If the court can answer “yes” for some class members and “no” for others, no commonality exists. Id. at 431. So too if different class members must use different pieces of evidence to answer the question. Tyson Foods, Inc. v. Bouaphakeo, 577 U.S. 442, 453 (2016). To identify such variances, a careful analysis of commonality must assess potential “[d]issimilarities within the proposed class” and explain why these apparent differences do not defeat class certification. Dukes, 564 U.S. at 350 (quotation omitted). The court must consider opposing arguments to ensure that the plaintiffs “actually prove” a common answer exists. Halliburton, 578 U.S. at 275; Ford, 86 F.4th at 728.
Our decision in Doster shows how one imperative of this analysis works—that the trial court must examine the elements of the relevant claims to determine whether they yield yes-or- no answers. 54 F.4th at 430–31. In that case, service members alleged that the Air Force systematically denied their applications for religious exemptions from COVID-19 vaccine mandates, violating the Religious Freedom Restoration Act of 1993 and the Free Exercise Clause. Id. at 405. We held that two common questions existed for these pattern-or-practice claims: whether the Air Force followed a policy of rejecting religious exemptions and whether it treated religious exemptions worse than other exemptions. Id. at 433. We noted that both No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 7
questions directly answered central elements of the pattern-or-practice claims. The first question would inform whether the Air Force took the “least restrictive means” to achieve its public- health interests under the statute. Id. at 433–34 (quotation omitted). The second question would answer whether the Air Force’s policy was “neutral and generally applicable” under the First Amendment. Id. at 434; see also Dukes, 564 U.S. at 349–50 (assessing the elements of an employment discrimination claim to determine whether the proposed class action would yield common yes-or-no answers); Whirlpool, 722 F.3d at 853–54 (assessing the elements of a breach- of-warranty claim to determine whether the proposed class action would yield common yes-or- no answers).
Ford shows how another imperative of this analysis works and does so in the context of a product-liability case—that the yes-or-no inquiry must apply to materially similar car models. Owners of Ford F-150 pickup trucks alleged that their trucks contained faulty brake cylinders. Ford, 86 F.4th at 727. Ford contended that “key changes” to brake units over the years remedied those defects in some of the model years. Id. at 728. That suggested that some plaintiffs drove cars without a common defect. Id. The district court disregarded those changes because the claims were “based on the same Brake System Defect across all Class Vehicles.” Id. In rejecting this explanation, we reasoned that Rule 23 demands that the district court grapple with evidence of material changes to the allegedly defective product. Id. at 727–28. Only by coming to grips with changes to the challenged component of the car could the court decide whether any central questions “can each be answered in one stroke.” Id. at 727 (quotation omitted). By failing to do so, the trial court did not give the defect’s alleged commonality the scrutiny required. Id. at 728.
In applying these requirements to this class-certification order, we see two over-arching flaws. One turns on the distinct software upgrades created for different model cars. The other turns on the failure to analyze the elements of each state law claim to determine which, if any, elements would yield common yes-or-no answers. In fairness to the district court, it did not have the benefit of our Ford decision at the time it certified these classes.
Software upgrades for different models. All ten classes of Nissan buyers allege that their cars contain the same problematic automatic braking systems based on the same faulty radar. No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 8
Even if we assume for the sake of argument that one “defect” links all of these claims (more on that later), software updates to the system radar and control units remedied those alleged flaws in some cars. The district court must grapple with those updates to answer whether the existence of a defect can be established in one stroke. Id. at 727.
That same problem infiltrates the other possibilities for commonality. We cannot determine whether common evidence establishes Nissan’s knowledge as to the defects if the software updates remedied those defects for some cars. Id. at 728. Nor can we determine whether, for each state class, a common question of liability exists for all claims. The plaintiffs’ argument reduces to the bare assertion that common evidence establishes each cause of action. But without a common defect, that cannot be the case.
The district court did not consider evidence about these differences because it viewed them as rooted in a “level of specificity” not required at the certification stage. R.306 at 7. But Rule 23(a)(1) does not discount product differences based on their specificity. It asks only whether generalized evidence can resolve common questions “in one stroke.” Ford, 86 F.4th at 727 (quotation omitted). Assessing the materiality of these differences is critical to analyzing whether these questions will yield common answers, even if that analysis requires specificity in some circumstances.
Element-by-element assessment of commonality. The lack of a common answer to this threshold factual question—which version of the software was included in the car each consumer drove?—alone justifies vacating this order under Ford. But on remand, the district court must do more than identify whether the plaintiffs’ questions have some common answers. That’s because not every question with a common answer meets Civil Rule 23(a). It’s only “central” issues that matter. Dukes, 564 U.S. at 350. Otherwise, frivolous inquiries such as whether everyone sued the same company would suffice. Id. at 349–50.
The only way to determine whether an issue is central is to deal with the material elements of each claim. Doster, 54 F.4th at 430. The district court must examine each cause of action, identify the relevant elements, and evaluate how the common answer at hand helps to No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 9
resolve at least one of them. And it must repeat this analysis for all ten classes. That is the only way to assess whether the proposed class questions can resolve each claim in each class.
In an element-by-element analysis of each state claim, notably, it does not suffice simply to allege a common “defect” in a car—whether, in the plaintiffs’ words, “the [braking] systems are defective.” R.306 at 7. “[D]efective[ness]” is a term of art that, in this case at least, nobody has rigorously applied to the various state law claims. See, e.g., Plas-Tex, Inc. v. U.S. Steel Corp., 772 S.W.2d 442, 444 (Tex. 1989). In some States, including one of the States at issue (Texas), a “defect in a strict products liability case” differs from a “defect in an implied warranty” case. Id. In holding that the plaintiffs’ claims all “center around proof of a defect in the specific sensor,” R.306 at 7, the district court did not explain how that term matches with, or does not match with, each state law claim.
The concrete illustrates the abstract. Here are a few representative examples of the state- law claims filed in this class action and the kinds of questions the parties and district court must ask and answer on remand as to whether those questions are “common” or “individualized” ones.
An implied-warranty claim under Pennsylvania law, for example, comprises four elements: (1) an implied warranty of merchantability, (2) breach, (3) causation, and (4) damages. In re Gen. Motors LLC Ignition Switch Litig., 257 F. Supp. 3d 372, 439 (S.D.N.Y. 2017). A common yes-or-no answer might resolve the first element: Is a vehicle merchantable if it might unexpectedly brake based on imperfections in a warning device? But that will turn on Pennsylvania law and the idiosyncrasies of its law with respect to the implied warranty of merchantability. What of causation or damages? These elements require inquiries into whether the defect manifested, and the amount of damages (economic or otherwise) each individual plaintiff suffered. Id. Pennsylvania measures the harm caused by the merchant by looking to “a manifestation of the breaching defect.” Zwiercan v. Gen. Motors Corp., 2002 WL 1472335, at *3 (Pa. Com. Pl. May 22, 2002). Because any sudden braking likely manifested only for some Pennsylvania drivers—those who did not disengage the feature and who live near low-hanging overpasses or use elevated parking garages—the district court will have to determine whether these elements are amenable to a common answer. No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 10
A breach of express warranty claim under California law, for example, requires proof (1) that the plaintiff received a written warranty from the defendant, (2) for which the product did not perform as promised, (3) that the plaintiff gave reasonable notice to the defendant, (4) that the defendant failed to repair the product or provide another remedy per the warranty, and (5) that the plaintiff was harmed. Cal. Com. Code § 2313; Deitsch Plastics Co. v. Gredale LLC, 602 F. Supp. 3d 1331, 1336 (C.D. Cal. 2022). The first and second elements could lend themselves to common proof. Because Nissan warranted to “repair or replace defective” brake components—presumably through the same written warranty—the plaintiffs may be able to determine the warranty’s scope in one stroke. R.148 at 59. And the class might be able to speak with one voice as to whether the braking system and software version in each Nissan vehicle are defective within the meaning of this contract language. Because the breach-of-warranty claims allege that Nissan failed to repair defective parts, a common answer as to defect could implicate the second element.
But the last three elements may present more individualized questions. Consider reasonable notice. Whether a plaintiff gave reasonable notice rests on whether he brought his Nissan in for repairs. This answer likely differs across the class members, potentially spoiling commonality. Consider harm. Some class members (perhaps many class members) never experienced sudden braking. This element, too, may require an individualized inquiry.
Now consider the inquiry over the defendant’s repairs. Nissan offered repairs to class members who requested them. But given that many members never sought repairs, it likely follows that Nissan did not offer repairs to some class members, potentially dashing commonality for this element. A customer, it is true, need not seek a repair if it wouldn’t solve the problem. See Benkle v. Ford Motor Co., No. SA CV 16-1569-DOC, 2017 WL 9486154, at *11–12 (C.D. Cal. Dec. 22, 2017). This reality may implicate a common question over the fourth element: whether the class vehicles’ braking systems contained irreparable defects. But the district court must analyze this possibility carefully, which includes considering the possibility that updates to the radar and control unit software materially improved the product and made any remaining problems fixable. No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 11
Under Illinois law, for example, the Consumer Fraud Act requires a plaintiff to show (1) a defendant’s deceptive act or practice in commerce, (2) an intent that the plaintiff rely on it, (3) actual damages, and (4) proximate cause. De Bouse v. Bayer, 922 N.E.2d 309, 313 (Ill. 2009). Common answers might resolve the questions presented by the first element. The plaintiffs contend that Nissan sold vehicles with faulty braking systems to Illinois residents without disclosing their limitations. That may implicate a common question about the deceptiveness of a uniform statement to customers purchasing (or leasing) Nissan vehicles about whether the automated brakes worked, assuming each vehicle contains the same one. And that argument for like reasons may involve a common answer to the question whether it occurred in trade or commerce.
But individualized questions may affect the other elements. Reliance tends to turn on individual circumstances. To establish proximate causation, the plaintiff must “actually be deceived by a statement or omission.” Id. at 316. If a consumer “has neither seen nor heard” the statement or ignored it, it may be difficult to show they have been harmed by it (or for that matter relied on it). See id. Even if each consumer experiences the same deceptive statement, they may experience it differently in making individual purchasing decisions. Skeptical purchasers who do “not rely” on a promotional statement have no claim. See id. That demands individual determinations because—unlike securities fraud—Illinois lacks a “market theory of causation” for these claims. Id.
An unjust enrichment claim under New York law, for example, requires proof that (1) the defendant was enriched, (2) at the plaintiff’s expense, and (3) “it is against equity and good conscience” for the defendant “to retain what is sought to be recovered.” Mandarin Trading Ltd. v Wildenstein, 944 N.E.2d 1104, 1110 (N.Y. 2011) (quotation omitted). All three elements examine the details of the transaction between the parties, the purchaser’s expectations, reliance on the seller’s statements, and other individualized details. Id. at 1110–11. That reflects the nature of equitable actions, which draw on “broad considerations of equity and justice” that frequently change in context. Paramount Film Distrib. Corp. v. State, 285 N.E.2d 695, 698 (N.Y. 1972). Common details about the braking system would seem to do little to resolve the first two elements. The same likely goes for the third element. Perhaps one could ask whether No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 12
selling any system that unexpectedly brakes is necessarily against good conscience. Even so, this would not resolve the third element, which examines a plaintiff’s individual damages.
A fraudulent concealment claim under Illinois law, for example, requires (1) a “false statement” or omission “of material fact,” (2) a defendant’s knowledge of its falsity, (3) a defendant’s intent to induce the plaintiff to act, (4) the plaintiff’s reliance, and (5) some damages resulting from that reliance. Connick v. Suzuki Motor Co., 675 N.E.2d 584, 591–93 (Ill. 1996).
Reliance looms large in fraud claims. In this instance, it’s hard to see how the class could assert with one voice that they had all relied on Nissan’s statements or suffered the same damages because of that reliance. Illinois law asks “which, if any, of the plaintiffs heard these representations and relied on them.” Id. at 591. A similar problem haunts the damages inquiry, which turns on the extent to which each plaintiff relied on each representation. All of this explains why the Supreme Court characterizes reliance as often “an insuperable barrier to class certification.” Dukes, 564 U.S. at 351 n.6. The district court will need to examine these elements closely—as well as consider the possibility that a common question exists over whether Nissan knew that the class vehicles had faulty braking systems but said nothing.
We provide these examples to show the nature of the “commonality” question and the need for the question to resolve something that matters to the plaintiffs’ claims. But our analysis should not be taken too far. Technically, the plaintiffs need only identify one common question to satisfy Rule 23(a)(2)’s commonality element. See id. at 359. If other questions raise individual issues, the court’s decision to certify a class must then depend on Rule 23(b)(3)’s predominance element. See id. Here, however, the district court’s analysis and the plaintiffs’ briefing failed to identify a single legally significant common question because they did not attempt to match any common question to any element of any of the plaintiffs’ claims.
The plaintiffs resist these conclusions for several reasons. They argue that another case involving Ford, Daffin v. Ford Motor Co., 458 F.3d 549 (6th Cir. 2009), supports the class- certification order. But Daffin hurts more than helps them. There, a class of Mercury Villager buyers sued Ford on the ground that a defective throttle caused the accelerator to stick. Id. at 550. We approved and certified the class because a single product—an identical throttle body in No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 13
each car—presented a common question of defect amenable to classwide resolution. Id. at 552. Here, in marked contrast, Nissan has provided evidence that software upgrades to the radar and brake unit stopped the defect for some claimants. That reality shows that each buyer does not face a common defect.
The plaintiffs insist that Nissan admitted that the radars in each vehicle had a common defect. But such an admission does not come to grips with Nissan’s efforts to remedy it. The company presented evidence that it had updated the radar, the control units, and the relevant software. At issue by the terms of the plaintiffs’ complaint, moreover, is whether a common question exists over the defectiveness of the braking “systems” and their associated updates, not just whether the radars are defective. R.241 at 18. A court cannot assess the commonality of those alleged defective systems without investigating how their components changed.
The plaintiffs claim that the software updates imposed negligible performance improvements. But Nissan produced evidence that the S1 and S2 software updates, which improved the radar’s detection function, mitigated the problems experienced by vehicles with the S0 software. As proof, Nissan identified a decrease in warranty claims for vehicles with the updated software. The district court must closely examine the effect of these changes on remand. If a reasonable jury could answer “yes” to the defect question for those members of the class who did not receive updates and “no” for those who did, the question does not represent a “common” one within the meaning of Rule 23. See Doster, 54 F.4th at 430–31.
The plaintiffs contend that the details of the various versions of the radar system—and whether they were defective—amount to a forbidden merits inquiry at the class-certification stage. Not so. The court must consider the merits to the extent “relevant to determining whether the Rule 23 prerequisites” are met. Amgen Inc. v. Conn. Ret. Plans & Tr. Funds, 568 U.S. 455, 466 (2013). That analysis often “will entail some overlap with the merits” and in situations like this one “cannot be helped.” Dukes, 564 U.S. at 351. Analyzing the various manifestations of the alleged defect is necessary to assess whether common evidence could vindicate the plaintiffs or Nissan on a classwide basis. No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 14
Plaintiffs’ arguments about a common defect exhausted, they turn to Nissan’s knowledge of the defect. They claim that common evidence can establish that knowledge even if the radar system changed over time. But Ford forecloses that line of reasoning. 86 F.4th at 728. As in that case, Nissan’s understanding of the defect changed with each update. Nissan could reasonably believe that each software update fixed, or at least ameliorated, any defect for some plaintiffs. And later claimants, who purchased cars with S2 brake systems, may not have faced any defect at all.
The plaintiffs offer other reasons why common knowledge might exist. They contend that Nissan’s knowledge didn’t change over time because it continued to disclose the risk of false activations in its manuals and before the district court. And they argue that Nissan’s witnesses and expert testimony fail to establish that the upgrades prevented phantom breaking. But none of this addresses the question left unanswered below: Does awareness of a general risk of product limitations—true for any automatic braking system, as no perfect one yet exists—establish knowledge of a common defect? And none of this, most critically, comes to grips with the requirements of looking at these issues through the specific lens of each cause of action and each of its elements. See Doster, 54 F.4th at 430–31; Dukes, 564 U.S. at 350–52.
B. Predominance
Commonality issues taint the predominance inquiry from the get-go. For classes certified under Civil Rule 23(b)(3), as this one was, common questions of law or fact must “predominate over any questions affecting only individual members.” The rule requires a court to put the common issues on one side, the individual issues on the other, then “qualitatively evaluate which side ‘predominates.’” Fox v. Saginaw County, 67 F.4th 284, 300 (6th Cir. 2023). This balancing works only if the district court properly identified at least one common (non-individualized) issue. Having determined that the district court failed to do so, especially in light of Ford and Doster, it follows that the predominance inquiry requires a second look.
In taking that second look, the district court must consider, and account for, some of the “individualized determinations” that might be “ill-equipped for classwide proof.” Tarrify Props., LLC v. Cuyahoga County, 37 F.4th 1101, 1106 (6th Cir. 2022). Accounting for these No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 15
individualized determinations in connection with any truly common questions of law and fact is indispensable to the analysis demanded.
C. Expert Witnesses Under Daubert
The Supreme Court requires parties to “satisfy through evidentiary proof” that they “in fact” meet the elements in Civil Rule 23. Behrend, 569 U.S. at 33. Which prompts this question: When challenged expert testimony is relevant to class certification, must a district court perform a Daubert analysis of the evidence? The Third, Fifth, Seventh, and Eleventh Circuits say yes. See, e.g., In re Blood Reagents Antitrust Litig., 783 F.3d 183, 187 (3d Cir. 2015); Prantil v. Arkema Inc., 986 F.3d 570, 574–76 (5th Cir. 2021); Am. Honda Motor Co. v. Allen, 600 F.3d 813, 815 (7th Cir. 2010); Loc. 703, I.B. of T. Grocery & Food Emps. Welfare Fund v. Regions Fin. Corp., 762 F.3d 1248, 1258 n.7 (11th Cir. 2014). The Eighth and Ninth Circuits perform a more limited Daubert analysis. See, e.g., In re Zurn Pex Plumbing Prods. Liab. Litig., 644 F.3d 604, 614–15 (8th Cir. 2011) (assessing a subset of the factors under Daubert); Sali v. Corona Reg’l Med. Ctr., 909 F.3d 996, 1003–06 (9th Cir. 2018) (using Daubert to assess the weight of evidence, rather than to exclude evidence).
We agree with the majority view. If challenged expert testimony is material to a class certification motion, the district court must demonstrate the expert’s credibility under Daubert.
Class certification requires the plaintiffs to provide “evidentiary proof” that they meet the elements of Rule 23. Behrend, 569 U.S. at 33. Careful qualification of experts ensures that expert evidence “both rests on a reliable foundation and is relevant to the task at hand.” Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 597 (1993). If expert testimony is insufficiently reliable to satisfy Daubert, it “cannot ‘prove’ that the Rule 23(a) prerequisites have been met ‘in fact’” through acceptable evidentiary proof. Blood Reagents, 783 F.3d at 187 (quotation omitted). It’s no surprise that the Supreme Court and most of our sister circuits that have spoken suggest that Daubert—interpreting Evidence Rule 702—applies to expert testimony at the class- certification stage. Dukes, 564 U.S. at 354; see Am. Honda, 600 F.3d at 816; Blood Reagents, 783 F.3d at 187; Prantil, 986 F.3d at 576. No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 16
The challenged expert report is a critical piece of the plaintiffs’ effort to answer this commonality inquiry: Does each class vehicle contain the same allegedly deficient braking system, or do different cars have materially different automated brakes? Steve Loudon, an automotive control systems expert hired by the plaintiffs, says yes: The “primary defect . . . is the ARS410 radar’s inability to properly discriminate between” real and fake obstacles. R.245-5 at 32. In support, he reviewed documents obtained in discovery and drove some of the Class Vehicles under bridges and over railroad crossings. If reliable, Loudon’s testing helps to identify a common deficiency. Given these stakes, the district court must assess if his testimony reliably reflects the “knowledge and experience of his discipline.” Kumho Tire Co. v. Carmichael, 526 U.S. 137, 148 (1999) (quotation omitted).
Even if reliable, it deserves note, Loudon’s proffered evidence does not suffice by itself to show a flaw in all of the cars. He did not test vehicles with the S2 software update. And his paper review of the software updates asserts without any evidentiary support that each version shares “substantially similar software” and “basic system architecture.” R.245-5 at 20–21. That bald statement alone does not inform whether the S2 upgrade created materially different braking systems, let alone whether a common defect infiltrates each Class Vehicle. In the end, the plaintiffs face at least two hurdles, not one, on this score. They need to establish that their expert satisfies Daubert, and their expert needs to offer an evidence-backed opinion on each material aspect of the commonality inquiry.
The plaintiffs respond that we should not decide the Daubert question because the district court found that the parties’ experts were “not ‘critical’ to resolving the class certification motion.” Appellee’s Br. 42. But the court found only that the plaintiffs’ damages experts are not critical. The other experts provide a key insight on whether purported flaws differed among models. That explains why Nissan challenged Loudon’s expert report on Daubert grounds below and on appeal.
The plaintiffs argue that Lyngaas v. Curaden Ag forecloses this approach. 992 F.3d 412 (6th Cir. 2021). It does not. In Lyngaas, we held that the evidence rules governing authentication do not necessarily apply at the class-certification stage. Id. at 428–29. But we addressed “nonexpert evidence” in making the point. Id. Unlike the rules of authentication, No. 23-5950 In re Nissan N. Am., Inc. Litig. Page 17
Daubert “make[s] certain that an expert . . . employs in the courtroom the same level of intellectual rigor that characterizes the practice of an expert in the relevant field.” Kumho Tire, 526 U.S. at 152.
The plaintiffs point out that Daubert exists to “protect juries from being swayed by dubious scientific testimony” and argue that this consideration has no purchase here because the judge alone decides class certification. See In re Zurn, 644 F.3d at 613. Even so, the text and principles of Daubert still apply to this fact-based inquiry. Evidence Rule 702 does not distinguish between jury and bench trials. And Daubert “ensure[s] the reliability and relevancy of expert testimony,” a touchstone of a careful analysis of evidentiary proof. Kumho Tire, 526 U.S. at 152.
Last and least, the claimants argue that this approach prematurely decides the merits. That objection misapprehends Rule 23. The Daubert inquiry does not resolve who wins. It resolves whether “a decisionmaker can answer the Plaintiffs’ questions all at once through evidence common to the class” and, if need be, through evidence provided by a qualified expert. Doster, 54 F.4th at 435 (quotation omitted). When a court does consider a merits question, it is only “to the extent” that it is “relevant to determining whether the Rule 23 prerequisites” are met. Amgen, 568 U.S. at 466.
We vacate and remand for further proceedings consistent with this opinion.