concurring in part and dissenting in part.
As the majority describes, to prove their Unlawful Trade Practices Act claim, plaintiffs have to prove three elements: an unlawful trade practice, causation, and damages. 257 Or App at 117. The majority holds that each of those elements can be litigated on a class-wide basis. I respectfully disagree. In my view, the element of causation cannot be litigated on a class-wide basis.
As the majority holds, to prove the causation element of their particular claim — viz., that they and the putative class members suffered ascertainable losses “as a result of’ the alleged unlawful trade practice, ORS 646.638(1) — plaintiffs have to prove that they, and each putative class member, relied on defendant’s representations about Marlboro Lights. 257 Or App at 146. For the reasons explained below, I do not believe that the issue of reliance can be litigated on a class-wide basis. Therefore, I dissent from the majority’s holdings that (1) plaintiffs could litigate reliance on a class-wide basis, 257 Or App at 161-66; (2) the trial court erred in concluding that common issues predominated, 257 Or App at 166-68; and (3) remand is necessary for the trial court to reconsider whether to certify the action as a class action, 257 Or App at 169. However, because I agree with the majority’s holdings that plaintiffs could litigate the other two elements of their claim — viz., whether defendant engaged in an unlawful trade practice and whether plaintiffs and the putative class members suffered ascertainable losses — on a class-wide basis, 257 Or App at 138-39, I concur in the majority’s holding that the case must be remanded to the trial court to reconsider its decision regarding the certification of an issue class because that decision was predicated on the erroneous conclusion that none of the three elements of plaintiffs’ claims could be litigated on a class-wide basis, 257 Or App at 172.
Under the Oregon Supreme Court’s case law, whether an issue can be litigated on a class-wide basis depends on the cohesiveness of the class. Bernard v. First Nat’l. Bank, 275 Or 145, 159-60, 550 P2d 1203 (1976); Derenco v. Benj. Franklin Fed. Sav. and Loan, 281 Or 533, 570, 577 P2d 477, *174cert den, 439 US 1051 (1978). That is, it depends on the likelihood that the class members share the characteristic necessary for the issue to be resolved in their favor. If it is likely that numerous members of the class do not share that characteristic, then the issue cannot be litigated on a class-wide basis; it is an individual issue. Thus, in Bernard and Derenco, where the plaintiff-borrowers’ claims depended on whether they and the class members lacked knowledge or notice of the defendant-banks’ lending practices, the Supreme Court focused on whether it was likely that numerous members of the classes did not lack knowledge or notice of the practices.
Accordingly, in this case, where plaintiffs’ claim depends on whether they and the putative class members relied on defendant’s alleged representation that Marlboro Lights were inherently light, we must focus on whether it is likely that numerous members of the class did not rely on that representation. More precisely, because the party seeking class certification bears the burden of proving that the requirements for class certification have been met with respect to the elements of its claim, we must ask whether plaintiffs established “that the number of [class members] who would be legitimately subject to challenge on the issue of [reliance] is less than ‘numerous.’” Bernard, 275 Or at 162 (quoting former ORS 13.220(2)(c) (1973), repealed by Or Laws 1979, ch 284, § 199); see also Bernard, 275 Or at 159 (focusing on the likelihood that a “considerable number” of the class members lacked the required characteristic); Derenco, 281 Or at 572 (focusing on whether a “substantial number” of the class members lacked the required characteristic).
Under certain circumstances, class members’ reliance on a defendant’s representation can be litigated on a class-wide basis. As the Supreme Court held in Strawn v. Farmers Ins. Co., 350 Or 336, 358, 258 P3d 1199, adh’d to on recons, 351 Or 521, 256 P3d 100 (2011), cert den, ___ US ___, 132 S Ct 1142 (2012), “[d]irect evidence of reliance by each of the individual class members is not always necessary [.]” Reliance on the part of all class members “can, in an appropriate case, be inferred from circumstantial evidence.” Id. For that inference to arise in a case like this one, “the same misrepresentation must have been made without *175material variation” and “the misrepresentation must be of a nature that the class members logically would have had a common understanding of the misrepresentation, and naturally would have relied on it to the same degree and in the same way.” Id. at 358-59. Thus, in this case — where there is no dispute that defendant made the same representations to all of the putative class members — the questions, for class certification purposes, relate to the likelihood that the putative class members had a common understanding of the representations and, if they did, the likelihood that the representations played a substantial role in their decisions to purchase Marlboro Lights.1
The representations at issue were that Marlboro Lights were “Lights” and had “Lowered Tar and Nicotine.” In my view, reliance on the part of all putative class members cannot be inferred from circumstantial evidence in this case because of the likely variations in both how putative class members understood defendant’s representations and, relatedly, whether the representations played a substantial role in their decisions to purchase Marlboro Lights.
Some putative class members may have understood the representations to mean that Marlboro Lights were inherently light; that is, they may have understood the representations to mean that Marlboro Lights would deliver less tar and nicotine than Marlboro Regulars, regardless of how the Marlboro Lights were smoked. Such putative class members may have believed that either the contents or the design of the cigarettes made it impossible for Marlboro Lights to deliver the same amount of tar and nicotine as Marlboro Regulars. But other putative class members may have understood the representations to mean only that Marlboro Lights would deliver less tar and nicotine if smoked the same way as Marlboro Regulars. In other words, they may have understood that, in a side-by-side comparison, with all other variables controlled, Marlboro Lights would deliver less tar and nicotine than Marlboro regulars, a fact which plaintiffs do not dispute.
*176It may well be that, given defendant’s representations and its advertising, many putative class members who purchased Marlboro Lights early in the class period did not know or have reason to know that Marlboro Lights were not inherently light. But the class period is long: It runs from 1971 to 2001. In that 30-year period, information about dilution filters and the phenomena of titration and compensation was increasingly available.
As defendant points out, there were articles in the lay press throughout the class period that explained that light cigarettes could deliver the same amount of tar and nicotine as regular cigarettes. As described in the majority’s opinion, as early as 1976, within five years of the 1971 introduction of Marlboro Lights, Consumer Reports published an article explaining compensation, stating that “[n]icotine is an addicting agent for most smokers. When cigarette smoke contains less nicotine than such smokers are accustomed to, their bodies simply contrive ways to get more smoke [,]” and that human smokers “do not necessarily smoke a low-nicotine cigarette in the same way they smoke a high-nicotine cigarette[.]” 257 Or App at 125. Such warnings continued to be published in the national and local press, including in a 1983 Newsweek article which stated that “[t]he widely touted notion that low-tar-and-nicotine cigarettes are safer than stronger brands is a pipe dream,” and in a 1994 Oregonian article, which described the FTC Method and stated that “[sjmokers of cigarettes labeled low in tar and nicotine may be getting more of those substances than they think.”2 257 Or App at 126.
Thus, information about the fact that Marlboro Lights are not inherently light was in the press for many years of the class period. Given the availability of that information, it is possible that a considerable number of the putative class members knew, or at least were on notice, that Marlboro Lights were not inherently light.
*177Indeed, some of the survey evidence in the record suggests that a considerable number of people, including smokers of light and ultra-light cigarettes, did not believe that light and ultra-light cigarettes were safer than regular cigarettes. As described in the majority’s opinion, in one survey, 51 percent of the smokers surveyed agreed that smoking low-tar cigarettes was safer than smoking high-tar cigarettes and, in another survey, 35 percent of light and ultra-light smokers said that they chose their brand for health-related reasons. 257 Or App at 125. Those numbers suggest that, although many people may have believed that light cigarettes were safer than regulars, many other people did not. And even those who believed that light and ultralight cigarettes were safer could have understood that they were not inherently safer.
Thus, this case is like Bernard, where the Supreme Court concluded that the information it had regarding the nature of the loans and the characteristics of the class members gave rise to a legitimate question about whether a “considerable number” of the class members knew about the defendants’ lending practices. 275 Or at 159, 162. Here, there is a legitimate question about whether a considerable number of the putative class members knew, or were on notice, that Marlboro Lights were not inherently light and, therefore, did not understand defendant’s representations to mean that Marlboro Lights were inherently light or did not rely on the representations
Given that, I would conclude that the trial court did not err in ruling that the element of causation could not be litigated based on evidence common to the class. I would further conclude that the trial court did not err in ruling that, because the element of causation would have to be litigated based on evidence specific to each class member, common questions did not predominate and, consequently, a class action would not be superior to other methods for adjudicating the controversy.
Regarding predominance, I would conclude that, in light of the importance to the action of putative class members’ reliance, the trial court did not err in ruling that common questions did not predominate. In determining *178whether common issues predominate, a court must consider the relative importance of common and individual questions to the action, not their numbers. Cf. Moore v. PaineWebber Inc., 306 F3d 1247, 1252 (2d Cir 2002) (under FRCP 23(b)(3), common issues predominate if the issues that can be resolved through generalized proof “are more substantial than the issues subject only to individualized proof”). Here, because the question of causation is an individual question and resolution of it could involve questioning thousands of class members, common issues do not predominate. See Newman v. Tualatin Development Co., 287 Or 47, 53-54, 597 P2d 800 (1979) (no predominance where “individual determinations of reliance would be required”); Bernard, 275 Or at 162 (no predominance where substantial issue of class members’ knowledge of banks’ practices could not be litigated through common evidence).
Regarding superiority, I would affirm the trial court’s ruling, which we review for abuse of discretion. Newman, 287 Or at 51; Joachim v. Crater Lake Lodge, 48 Or App 379, 393, 617 P2d 632, rev den, 290 Or 211 (1980). The trial court assessed the manageability of a class action trial, which would involve individual inquiries of thousands of class members. Considerations relevant to that assessment include the difficulties of maintaining a jury for such a trial and the possibility that the risks of a mistrial would increase with the duration of the trial and number of witnesses. The trial court could reasonably decide to avoid the possibility that, perhaps after the presentation of a few hundred class-member witnesses, something would trigger a mistrial, with the result that the case would have to be tried again and all of those witnesses would have to be recalled. Thus, although individual trials would be repetitive, the trial court could conclude that they would be more manageable, more likely to reach a conclusion and, therefore, superior to a single class action.
Accordingly, I would affirm the trial court’s decision denying certification of the action as a class action. However, because, as mentioned, I agree with the majority’s determination that the trial court’s decision denying certification of an issue class was based on the erroneous conclusion that none of the three elements of plaintiffs’ claims could be *179litigated on a class-wide basis, I concur in the majority’s conclusion that reversal and remand is required for the trial court to reconsider that decision.
Haselton, C. J., and Schuman and Hadlock, JJ., join in this dissent.It bears emphasizing that, at the certification stage, the questions relate to whether to the likelihood that the putative class contains a “considerable number,” Bernard, 275 Or at 159, or a “substantial number,” Derenco, 281 Or at 572, of members who did not rely on the representations, which is distinct from a determination of the merits of plaintiffs’ claim.
In addition, in 1990, defendant began to include information about titration in its advertising (although not on its Marlboro Lights packages). The advertisements stated, “The amount of tar and nicotine you inhale will vary depending on how you smoke the cigarette.” Admittedly, that statement does not explain that the amount of tar and nicotine that a smoker can receive from a Marlboro Light can be the same as the smoker could receive from a Marlboro Regular, but it does provide notice that tar and nicotine yields can be affected by smoking methods.