Opinion by JUDGE WEBB

¶ 1 This putative class action concerns uninsured/underinsured motorist (UM/UIM) coverage. Plaintiffs, James and Janet Maxwell and Leon Hill, individually and on behalf of all others similarly situated, assert that defendants, United Services Automobile Association and USAA Casualty Insurance Company (collectively, USAA), fraudulently concealed information necessary for USAA insureds to make informed decisions about purchasing UM/UIM coverage on their additional vehicles. They pleaded claims for fraudulent concealment, insurer bad faith, and violation of the Colorado Consumer Protection Act (CCPA), §§ 6-1-101 to -1001, C.R.S.2013.

¶ 2 In this interlocutory appeal of the order denying class certification, plaintiffs raise two main contentions. First, the trial court abused its discretion by concluding that plaintiffs failed to satisfy the predominance requirement of C.R.C.P. 23(b)(3), because it improperly required them to prove reliance and invaded the merits by considering USAA's circumstantial evidence of some putative class members' likely nonreliance. Second, the court erred in holding that the filed rate doctrine applies to the insurance industry and, as applied here, precludes refund of UM/UIM premiums on additional vehicles as damages for fraudulent concealment. Both contentions raise unresolved questions of law in Colorado.

¶ 3 We discern no abuse of discretion in the decision denying class certification and we agree with the trial court's legal conclusions concerning the filed rate doctrine. Therefore, we affirm the order denying class certification and remand the case for further proceedings consistent with this opinion.

I. Background

¶ 4 In 2003, twenty-seven plaintiffs sued twenty-five insurance companies, including USAA. The trial court severed the case into separate proceedings against specific insurers. This is one of those severed cases.

¶ 5 After severance, the trial court granted USAA's summary judgment motion. The court held that because USAA required its insureds to purchase UM/UIM coverage for either every vehicle they insured with USAA or none of them, the putative class could not have been misled about purchasing such coverage on additional vehicles. The summary judgment was reversed in Maxwell v. USAA (Maxwell I), (Colo.App. No. 07CA1611, 2008 WL 5104227, Dec. 4, 2008) (not published pursuant to C.A.R. 35(f)). In Maxwell I, the division:

¶ 6 On remand, plaintiffs moved for class certification. According to plaintiffs, USAA fraudulently concealed that after DeHerrera, the OBNI exclusion was not enforceable as to named insureds or their resident relatives, and that buying UM/UIM coverage on additional vehicles did not increase the protection of these insureds. (Although plaintiffs did not coin the terminology, through the litigation, named insured and their resident relatives came to be termed "class one insureds"; nonresident relatives and guests were termed "class two insureds.") They contend this harmed the putative class in two ways:

¶ 7 After holding a five-day evidentiary hearing on class certification, the trial court denied the motion in a lengthy and well-reasoned order. Plaintiffs appeal that order based on the court's C.R.C.P. 54(b) certification.

II. The Trial Court Did Not Abuse its Discretion in Admitting Retention Data of Another Insurer at the Class Certification Hearing

¶ 8 Plaintiffs contend the trial court erred in admitting data compiled by State Farm Mutual Insurance Company (State Farm) about its insureds' retention of UM/ UIM coverage on additional vehicles after having been notified of DeHerrera (SF Data). This data showed that the majority of State Farm's insureds who were notified of DeHerrera chose to retain UM/UIM coverage on all vehicles. We begin with this issue because error would require remand for further findings without regard to the SF Data, which the trial court noted in the order denying certification. We discern no abuse of discretion.

¶ 9 The court denied plaintiffs' motion in limine to exclude the SF Data on three grounds: (1) the SF Data is admissible hearsay under the business records exception; (2) this data is relevant and any argument to the contrary goes to weight, not admissibility; and (3) reversal of the summary judgment order in Young v. State Farm Mut. Auto. Ins. Co., (Colo.App. No. 10CA1405, 2011 WL 2650736, July 7, 2011) (not published pursuant to C.A.R. 35(f)), vacated & remanded, (Colo. No. 11SC814, 2012 WL 1194239, Apr. 9, 2012) (unpublished order granting certiorari), does not require a different result.

A. Standard of Review and Preservation

¶ 10 Evidentiary rulings are reviewed for an abuse of discretion, meaning they "are reversible only if they are manifestly arbitrary, unreasonable, or unfair." Chavez v. Parkview Episcopal Med. Ctr., 32 P.3d 609, 611 (Colo.App.2001). Plaintiffs' motion in limine preserved this issue. C.R.E. 103(a)(2).

B. Application

¶ 11 None of the three grounds on which the trial court ruled to admit the SF Data constitutes an abuse of discretion.

¶ 12 First, before the trial court admitted the SF Data under the business records exception to the hearsay rule, C.R.E. 803(6), it considered the five requirements for admissibility under Schmutz v. Bolles, 800 P.2d 1307, 1312 (Colo.1990). It found record support for the court's reasoning in Health Alliance Network, Inc. v. Cont'l Cas. Co., 245 F.R.D. 121, 129-30 (S.D.N.Y.2007), which analyzed the admissibility of a subset of data extracted from a larger database. Although plaintiffs assert that the SF Data had "obvious hearsay... issues," plaintiffs do not provide any basis for reversing the trial court's findings as to the hearsay exception. We decline to "address the contention based on a hypothetical construction of what [plaintiffs'] argument might be." People v. Cordova, 293 P.3d 114, 118 (Colo.App.2011).

¶ 13 Second, the trial court recognized that the question of sufficiency of notice remains a factual issue to be resolved at trial, but found "the overwhelming majority of State Farm insureds who had multiple vehicles purchased UM/UIM coverage on all their vehicles after the DeHerrera notice" and this "is circumstantial evidence of non-reliance." Plaintiffs dispute relevance, arguing that "[a]bsent the assumption that State Farm provided proper disclosure, the [SF] Data were irrelevant and had no evidentiary value as to the conduct of informed insureds." But the court correctly differentiated between weight and admissibility when assessing the relevance of the SF Data. See People v. Trefethen, 751 P.2d 657, 659 (Colo. App.1987) ("Whether evidence is too remote to be relevant is within the trial court's discretion... [and] affects only the weight to be given to evidence, not its admissibility.").

¶ 14 Third, we agree with the trial court that the reversal of summary judgment in Young does not preclude admission of the SF Data, for two reasons. First, the heightened standard for summary judgment and the lower standard for admission of evidence require different analyses. Second, Young did not mention the SF Data in the part of its opinion reversing the summary judgment order, but only identified, as a disputed issue of material fact, the adequacy of the insurer's DeHerrera notice to its insureds.

¶ 15 Accordingly, we conclude that the trial court did not abuse its discretion in admitting the SF Data and considering this data when denying class certification.

III. The Trial Court Acted within Its Discretion in Concluding that Plaintiffs Failed to Establish Predominance under C.R.C.P. 23(b)(3)

A. Standard of Review

¶ 16 An order denying class certification is reviewed for an abuse of discretion. Farmers Ins. Exch. v. Benzing, 206 P.3d 812, 818 (Colo.2009). Under this highly deferential standard, "so long as the trial court rigorously analyzes the evidence, it retains discretion to find to its satisfaction whether the evidence supports each C.R.C.P. 23 requirement." Jackson v. Unocal Corp., 262 P.3d 874, 884 (Colo.2011). But the trial court's legal conclusions underlying the order denying class certification are reviewed de novo. See BP Am. Prod. Co. v. Patterson, 263 P.3d 103, 108 (Colo.2011).

B. C.R.C.P. 23(b)(3) Predominance

¶ 17 To be certified as a class under C.R.C.P. 23(b)(3), plaintiffs must establish that common questions of law or fact predominate over individual questions. This inquiry begins with the elements of the underlying cause of action, see Erica P. John Fund, Inc. v. Halliburton Co., ___ U.S. ___, 131 S.Ct. 2179, 180 L.Ed.2d 24 (2011), to test whether the proposed class "is sufficiently cohesive to warrant adjudication by representation." Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 623, 117 S.Ct. 2231, 138 L.Ed.2d 689 (1997).

¶ 18 When assessing predominance, a court may neither "decide the substantive claims and defenses of the parties" nor "determine whether the class will ultimately succeed in establishing each element necessary to prove its claims." Benzing, 206 P.3d at 820. Rather, it must engage in a "fact-driven, pragmatic inquiry," Medina v. Conseco Annuity Assur. Co., 121 P.3d 345, 348 (Colo.App.2005), to determine whether "the plaintiff advances a theory by which to prove or disprove an element on a simultaneous, class-wide basis," Benzing, 206 P.3d at 820 (internal quotation marks omitted). For example, in affirming denial of class certification, the court in Benzing explained that the record "appeared to support the defendant's argument that individual inquiries as to whether the defendants caused injury to each of the class members might be required, and thus that issues individual to the class might predominate over common ones." Id. at 819. But if a plaintiff successfully advances such a theory, then classwide "proof obviates the need to examine each class member's individual position." Id.

C. Application to Fraudulent Concealment Claim

¶ 19 If plaintiffs satisfy the predominance requirement, their "fraudulent concealment claim [would be otherwise] amenable to class-wide adjudication." Patterson, 263 P.3d at 112. To do so, plaintiffs must show that questions of law or fact common to the class would predominate in resolving each of the five elements of fraudulent concealment:

First Interstate Bank of Fort Collins, N.A. v. Piper Aircraft Corp., 744 P.2d 1197, 1200 (Colo.1987).

¶ 20 Here, the trial court found that the first three elements could be proven on a classwide basis.4 But it concluded that plaintiffs had failed to establish predominance because individualized inquiry would be required as to the fifth element — reliance. ("[T]he court finds that USAA has raised sufficient evidence of non-reliance to cause a need for individualized inquiry to determine whether class members would want to purchase class two coverage.") Plaintiffs argue that in doing so, the trial court made two errors: first, requiring them to "prove [that]... they would have made `different,' `uniform,' and `predictable' purchase decisions had USAA" disclosed the scope of UM/UIM coverage (emphasis in original); and, second, holding that USAA had refuted an inference of classwide reliance with circumstantial evidence of reasons why putative class members may not have relied.

1. Purchase Decisions

¶ 21 Initially, plaintiffs misstate the trial court's analysis of purchase decisions by putative class members. The court did not impose a categorical legal rule requiring plaintiffs to prove that with proper disclosure, USAA insureds would have made different decisions concerning UM/ UIM coverage. Rather, the court distinguished between cases such as Patterson, involving circumstances where "no reasonable person would not want to receive the full amount of money they were entitled to receive" (emphasis in original), and the evidence presented at the hearing, from which it found that "class two coverage has value for some class members and little or no value for others." This finding is sufficiently supported by the record. Plaintiffs do not argue otherwise.5

¶ 22 Based on this distinction, the court also found that "one cannot predict what any individual class member's purchasing decision might be with any degree of certainty." And it concluded that "USAA has raised sufficient evidence of nonreliance to cause a need for individualized inquiry to determine how class members would have responded to the material information that was not disclosed," which defeated C.R.C.P. 23(b)(3) predominance.6

¶ 23 But the trial court's findings subsume two legal questions: whether, in a fraudulent concealment case where, as here, materiality of the undisclosed information and uniformity of distribution have been found, any evidence of reliance must be presented at the class certification stage; and if so, whether circumstantial evidence of nonreliance should be rejected because it presents a merits issue beyond the scope of class certification. Plaintiffs raised both questions below and argue them on appeal. We address these questions as follows: whether plaintiffs were required to establish reliance and could do so by inference; whether the inference can be rebutted with evidence of other explanations for putative class member's behavior; and whether the evidence used to rebut the inference could be circumstantial.

2. Plaintiffs Were Required to Establish Reliance at the Class Certification Stage as Part of Their Fraudulent Concealment Claim, But Could Do So By Inference

¶ 24 Plaintiffs argue that the fifth element of fraudulent concealment — reliance — does not require evidence that members of the putative class took action which they otherwise would not have taken. They rely on the statement that "material inducement is not whether the plaintiff's action would, but whether it might, have been different if the misrepresentation had not been made." Morrison v. Goodspeed, 100 Colo. 470, 478, 68 P.2d 458, 462 (1937). The trial court rejected this argument, concluding that without a requirement of reliance — inferred or otherwise — "the fifth element would be a nullity." We agree.

¶ 25 This element requires plaintiffs to prove their "reliance on ... the assumption that the concealed fact d[id] not exist." Nielson v. Scott, 53 P.3d 777, 780 (Colo.App. 2002). Typically, this is done by establishing causation in the form of "action on the concealment resulting in damage." Berger v. Sec. Pac. Info. Sys., Inc., 795 P.2d 1380, 1385 (Colo.App.1990).

¶ 26 But Colorado courts have also held that "[d]irect evidence of reliance, one of the elements of fraudulent concealment, is not required." Kopeikin v. Merch. Mortg. & Trust Corp., 679 P.2d 599, 602 (Colo.1984) (citing generally Morrison, 100 Colo. 470, 68 P.2d 458). While many Colorado cases have recognized that "reliance may be inferred from circumstantial evidence where the defendant concealed a material fact from the plaintiff," Patterson, 263 P.3d at 110, plaintiffs do not cite a Colorado case that has entirely removed reliance from the nondisclosure calculus. Nor have we found one. Thus, we conclude that some evidence of reliance was required and turn to whether plaintiffs met this burden by showing that they were entitled to an inference of reliance.

¶ 27 Both Maxwell I and the trial court's findings support such an inference. The division in Maxwell I held that the undisclosed information regarding UM/UIM coverage could be material, which the trial court found on remand. ("[I]t is undisputed that USAA chose not to disclose material information to class members during the class period."; "The Court finds that Plaintiffs can prove materiality on a class-wide basis.") The trial court also found "uniform nondisclosure." (Emphasis in original.)

¶ 28 But the court held that the inference of reliance was limited to cases where "there is predictable behavior on the part of class members." And it concluded "that is not the case here," because "there is no uniform method to predict what class members would have done if there had been disclosure."7

¶ 29 To the extent the court held that plaintiffs were not entitled to an inference of reliance, we disagree based on Morrison and Kopeikin. But whether, after considering USAA's evidence, plaintiffs were not entitled to an inference of reliance, or they were entitled to the inference but it was rebutted, is a difference lacking practical effect in a proceeding without a jury. And in any event, plaintiffs are correct that either conclusion raises two legal questions: first, whether a putative class should be entitled to an inference where reliance is "a logical explanation," despite evidence showing that it is not the only one, especially where the reliance is the type of behavior the defendant sought to induce; and, second, regardless of which standard applies, whether a defendant may use circumstantial evidence to refute the inference of reliance without invading the merits.

3. The Inference of Reliance Based on Uniform Nondisclosure Can Be Rebutted with Evidence of Other Explanations for the Putative Class Members' Behavior

¶ 30 The strength of a reliance inference at the class certification stage based on evidence of uniform nondisclosure has not been tested in Colorado. And few courts elsewhere have addressed this issue.

¶ 31 According to plaintiffs, testing the inference by asking whether it is the only explanation for behavior of putative class members that allegedly resulted in harm to them — the trial court's approach — would preclude class certification wherever, as here, those class members received a benefit, albeit a benefit different from what they reasonably had been led to believe they were receiving, based on the defendant's concealment. This argument is unpersuasive because it ignores the trial court's finding that "reasonable minds can differ about the value of class two coverage."

¶ 32 As the court further explained, such coverage "would be valuable to members with elderly family members who do not drive or with adult children in the service who do not own cars and return home on leave." The court found that "it is reasonable for some members to want the assurance that any guest passenger in their car will be covered in the event of a UM/UIM accident," while also recognizing that it is "reasonable for members to want to reduce their costs of insurance [by] eliminating class two coverage."

¶ 33 These findings, which have record support, distinguish cases where reasonableness could not be found because the benefit actually received was either de minimis or qualitatively different from the benefit that putative class members expected. But here, plaintiffs did not present any evidence that class two coverage had only minimal value. And both coverages include the same risks.

¶ 34 Plaintiffs' assertion that the reference to "might" in CJI — Civ. 4th 19:2 (2013) precludes a reasonable reliance inquiry fails for two reasons. First, "might" appears in paragraph four, which deals with the defendant's "intent that the plaintiff take a course of action (he)(she) might not take if (he)(she) knew the actual facts," not in paragraph five, which deals with reliance. Second, an inquiry guided by only what putative class members "might" do would allow class certification, even though evidence showed that at least some reasonable class members would do otherwise. Plaintiffs cite no Colorado authority, nor are we aware of any, predicating relief on unreasonable action.

¶ 35 Nor does the trial court's reasonableness approach lack precedential support. In Garcia v. Medved Chevrolet, Inc., 263 P.3d 92, 99, 99 n. 5 (Colo.2011), the supreme court cited with approval Peterson v. H & R Block Tax Servs., Inc., 174 F.R.D. 78, 85 (N.D.Ill. 1997) (holding inference appropriate when reliance is "the only logical explanation" to explain behavior), and Negrete v. Allianz Life Ins. Co. of N. Am., 238 F.R.D. 482, 491 (C.D.Cal.2006) (holding inference appropriate where "no rational class member would" have acted if there had been "adequate disclosure"). Evidence of a reasonable alternative explanation for putative class members' reliance would rebut the inference under either standard.

¶ 36 Contrary to plaintiffs' assertion, Patterson does not suggest otherwise. There, "the evidence BP purported to use to demonstrate that Plaintiffs should have known that post-production costs were being deducted from their royalty payments, namely the Royalty Brochures and Royalty Reports, was common to the class as a whole." 263 P.3d at 114. Here, in contrast, the trial court credited USAA's evidence that "the decision whether to purchase UM/UIM coverage for class two insureds is based on a value judgment, which depends on individual values." And this approach conforms to the warning that "[w]e generally should be cautious in permitting classwide proof of reliance and ignorance in cases of alleged fraudulent concealment." Id. at 115 (Eid, J., concurring in the judgment).

¶ 37 Likewise unhelpful is plaintiffs' emphasis on the following statement in Thompson v. Budget Rent-A-Car Sys., Inc., 940 P.2d 987, 990 (Colo.App.1996): "We further conclude that the driver's after-the-fact statement that he would have refused the additional coverage if it had been offered does not require a different result." Thompson was not a class action. And plaintiffs' focus on this language mistakenly assumes that they could have satisfied the trial court's reasonableness test only with testimony from putative class members of what they might have done differently years earlier.8

¶ 38 But USAA did not offer such testimony. Like USAA, plaintiffs could have — but did not — presented evidence about how economic, familial, and other factors would influence reasonable insureds' purchase decisions concerning class two coverage.9 Thus, a reasonableness analysis does not require proof of what members of the putative class would have done differently.

¶ 39 The few insurance cases that support plaintiffs are distinguishable. For example, in Stanich v. Travelers Indem. Co., 249 F.R.D. 506, 521 (N.D.Ohio 2008), the court held:

(Emphasis in original.) But, unlike here, the defendant insurer did not present any evidence of why putative class members might reasonably have made a different decision. The same is true of Baughman v. State Farm Mut. Auto. Ins. Co., 88 Ohio St.3d 480, 727 N.E.2d 1265 (2000), a class action that was certified based on nondisclosure of the Ohio equivalent of DeHerrera.

¶ 40 Plaintiffs cite no case, nor have we found one, juxtaposing a presumption of reliance arising from uniform nondisclosure of material information against circumstantial evidence of why a reasonable consumer would have made the same decision, had the information been disclosed.10 In the absence of such authority, the citation to Peterson and Negrete in Garcia weighs against accepting plaintiff's invitation to adopt contrary out-of-state cases that find lower standards sufficient. See, e.g., Hale v. Enerco Grp., 288 F.R.D. 139, 148 (N.D.Ohio 2012) (looking to considerations of "common-sense judgments," "an obvious link," and "where it is logical to do so" (citations and internal quotation marks omitted)).

¶ 41 Further, we decline to consider whether Peterson and Negrete should be tempered where the defendant succeeded in inducing exactly the behavior that was its objective. This case is not an appropriate one in which to do so because the trial court's treatment of USAA's intent is unclear. See supra n. 4.

¶ 42 Therefore, we conclude that while uniform concealment of material information creates an inference of reliance, that inference may be rebutted by evidence that a reasonable consumer would have made the same decision, even if the information had been disclosed.11 This conclusion leads to plaintiffs' second question about not prejudging the merits.

4. The Trial Court Properly Considered USAA's Circumstantial Evidence of Nonreliance Because It Did Not Encroach on the Merits

¶ 43 As part of its C.R.C.P. 23(b)(3) analysis, "the trial court must consider not only whether the circumstantial evidence common to the class supports an inference of causation, but also whether any individual evidence refutes such an inference." Garcia, 263 P.3d at 100. Because the defendant in Garcia presented only individual evidence, this statement does not resolve whether USAA could instead use circumstantial or indirect evidence to refute the inference at the class certification stage, without forcing the trial court to improperly prejudge the merits.

¶ 44 The parties have not cited authority in Colorado or elsewhere, nor have we found any, addressing this issue. For the following reasons, we first conclude that circumstantial evidence may be introduced at the class certification stage to refute an inference of reliance in a fraudulent concealment case.

¶ 45 Two competing principles frame the analysis. As indicated, "a trial court may not resolve factual or legal disputes to screen out or prejudge the merits of the case." Jackson, 262 P.3d at 877. But Colorado trial courts have been required to "rigorously analyze the evidence presented to determine whether a class-wide inference is appropriate given the facts and circumstances of the case." Garcia, 263 P.3d at 99. And "[t]here is often an overlap between the class certification decision and the merits of the case." Jackson, 262 P.3d at 884. As relevant here, "[t]his overlap is particularly apparent in the context of C.R.C.P. 23(b)(3)'s predominance inquiry." Id. at 885.

¶ 46 Where possible nonreliance turns on "what representations were received," In re St. Jude Med., Inc., 522 F.3d 836, 838 (8th Cir.2008), or "the circumstances surrounding the sale," McManus v. Fleetwood Enters., 320 F.3d 545, 550 (5th Cir.2003), the focus will necessarily be on "whether any individual evidence refutes such an inference," Garcia, 263 P.3d at 100. But this statement is uninformative where, as here, the defendant dealt uniformly with all putative class members.12

¶ 47 USAA's circumstantial evidence of nonreliance focused on why some putative class members might have kept their UM/ UIM coverage with USAA, despite being informed by USAA of DeHerrera. For example, based on undisputed evidence of low premiums, high customer loyalty, and customer satisfaction, USAA presented expert opinion testimony that even if putative class members had been fully informed, some of them might have stayed with USAA out of loyalty or to avoid the inconvenience of having auto policies with two carriers.

¶ 48 As a result, the court concluded that "[t]here is no way to predict" which class members would want to purchase UM/UIM coverage for class two insureds and those who would not after a satisfactory DeHerrera notice. This conclusion does not, as plaintiffs assert, assume that any putative class members insured additional vehicles with USAA to obtain class two coverage. Rather, it addresses what a reasonable class member might have done after proper disclosure. Thus, the trial court "rigorously analyze[d] the evidence" presented. Jackson, 262 P.3d at 884.

¶ 49 We discern no way to adhere to this rigorous analysis mandate while precluding the trial court from considering otherwise relevant circumstantial evidence. Adopting plaintiffs' proposed limitation to individualized evidence of nonreliance would create several problems. For example, it would require courts to decide how many individual instances of nonreliance would suffice to defeat certification of classes that may number in the hundreds or thousands, when another basis for resolving class certification may exist. And it would tie the hands of a defendant that lacked access to such individualized evidence.13 Here, because of the uniform nondisclosure, the UM/UIM purchase decisions did not generate significant individual contacts between class members and USAA's representatives on this subject, about which these representatives could testify.

¶ 50 Further, the Jackson court recognized that "[e]xpert opinions may be and often are offered to establish or refute the facts relevant to each C.R.C.P. 23 requirement." 262 P.3d at 885. The court offered as an example an expert who testified "that every plaintiff has suffered injury is in effect testifying that injury may be established by common proof." Id. Because such testimony is indirect, it is like circumstantial evidence. And here, USAA presented expert testimony, which the trial court credited, that even with proper notification of DeHerrera, some putative class members might have stayed with USAA.

¶ 51 Our supreme court "has permitted trial courts to rigorously analyze factual and legal disputes for the purposes of making a C.R.C.P. 23 determination, even where those disputes overlap with the merits." Jackson, 262 P.3d at 885. In other words, trial courts are precluded only from resolving "a factual or legal dispute that goes solely to the merits of the case," while considering such issues "to the extent necessary to satisfy itself that the requirements of C.R.C.P. 23 have been met." Id.

¶ 52 Applying this limitation weighs against finding an abuse of discretion in the trial court's analysis. The court carefully avoided determining "whether class members would have wanted to purchase class two coverage or whether they would not have wanted to purchase class two coverage." Its findings as to what any putative class members might have done would not preclude trial testimony of individual reliance or prejudge USAA's or plaintiffs' ultimate success on the merits.

¶ 53 Rather, the court narrowed its inquiry to determining whether USAA had presented sufficient evidence to conclude that the unresolved questions of fact would require individualized inquiry in a trial on the merits. Thus, settling the conflict between plaintiffs' inference of reliance and USAA's circumstantial evidence of nonreliance did not go "solely to the merits of the case." Id.

¶ 54 Again, the analysis of expert testimony in Jackson is informative. The supreme court explained that "the issue is not whether the plaintiff's expert will ultimately prevail on the merits, but rather whether the plaintiff's expert offers evidence that can be intelligently presented and evaluated within the procedural mechanism of a class action." Id. at 886 (citing In re Urethane Antitrust Litig., 251 F.R.D. 629, 637 (D.Kan.2008) ("The recent trend of authority is to permit the district court to compare the relative weight of expert opinions in ruling on a motion for class certification to the extent necessary to resolve the independent question of whether the plaintiff has shown that common questions will predominate.")). Similarly here, the trial court credited USAA's circumstantial evidence, including expert testimony, in concluding that the inference of reliance on which plaintiffs relied to show commonality did not obviate the need for individualized inquiry, which would be inconsistent with the class action mechanism.

¶ 55 Further, trial courts are granted "substantial discretion in deciding whether to admit evidence." Masters v. People, 58 P.3d 979, 996 (Colo.2002). And the trial court's C.R.C.P. 23 discretion includes determining, under the particular circumstances of the case, whether "the evidence is sufficient to ensure that C.R.C.P. 23's requirements are satisfied." Jackson, 262 P.3d at 882. Just as "[l]eaving class certification to the discretion of the trial court without requiring a specific burden of proof squares with the pragmatic and flexible nature of the class certification decision," id. so too these broad grants of discretion suggest that trial courts also have discretion to determine when "rigorous analysis" has become an impermissible merits inquiry.

¶ 56 Thus, we conclude that the trial court did not err in considering USAA's circumstantial evidence to refute plaintiffs' reliance inference. Accordingly, we further conclude that the court did not abuse its discretion in denying class certification.

IV. The Filed Rate Doctrine Applies to the Insurance Industry and Bars Plaintiffs From Obtaining a Refund of UM/UIM Premiums as Damages on Their Initial Purchase Theory

¶ 57 Next, plaintiffs contend the trial court erred in holding that the filed rate doctrine applies to the insurance industry.14 Alternatively, they contend that even if this doctrine applies, the trial court erred in concluding that it precludes a partial or complete refund of UM/UIM premiums for additional vehicles as the measure of damages on their initial purchase theory. These contentions raise unresolved questions of law in Colorado.15 Before addressing them, we reject USAA's jurisdictional challenge.

A. The Filed Rate Doctrine Ruling Is Properly Before the Division

¶ 58 USAA contends that because the trial court ruled on the filed rate doctrine in an order entered under C.R.C.P. 56(h) and the court did not certify this order as final and appealable, we lack jurisdiction to review it. We conclude that jurisdiction exists to review this order based on our undisputed jurisdiction over the order denying class certification.

¶ 59 In the class certification order, the court referenced its C.R.C.P. 56(h) ruling regarding the filed rated doctrine: "[a] significant legal development occurred when this Court ruled that a refund of premiums paid for additional vehicles could not be the proper measure of damages, based on the filed rate doctrine." The court concluded that such a "finding implicates the manner in which [p]laintiffs can prove damages on a class-wide basis."

¶ 60 All underlying legal rulings that influenced a trial court's ultimate decision are subject to review on appeal of the final judgment. In Levine v. Empire Sav. & Loan Ass'n, 192 Colo. 188, 189, 557 P.2d 386, 387 (1976), the court held that an order determining the legal insufficiency of a complaint to proceed as a class action "is tantamount to a dismissal of the action as to all members of the class." (Internal quotation marks omitted.) For purposes of appellate review, the court treated such an order as having the legal effect of a final judgment, because otherwise "review will have been effectively foreclosed." Id. at 190, 557 P.2d at 387. Applying a similar analysis here, review of the order denying class certification, without addressing the filed rate doctrine order, would lead to incomplete review of class certification.16 See also In re Flag Telecom Holdings, Ltd. Sec. Litig., 574 F.3d 29, 38 (2d Cir.2009) (scope of interlocutory review under Fed.R. Civ. P. 23(f) included causation issue which lower court had "clearly considered... as relevant to Plaintiffs' certification motion.").

B. Standard of Review

¶ 61 A trial court's resolution of questions of law under C.R.C.P. 56(h) is reviewed de novo. W. Elk Ranch, L.L.C. v. United States, 65 P.3d 479, 481 (Colo.2002).

C. Application

¶ 62 The filed rate doctrine, which limits judicial review of rates approved by regulatory agencies, is often traced to Keogh v. Chicago & Northwestern Railway Co., 260 U.S. 156, 163, 43 S.Ct. 47, 67 L.Ed. 183 (1922).17 The doctrine was first applied concerning federal agencies and "was originally established to ensure strict adherence to filed rates and to prevent utilities from intentionally misquoting rates to preferred customers who could then enforce a lower rate than the filed tariff." U.S. W. Commc'ns, Inc. v. City of Longmont, 948 P.2d 509, 525 (Colo.1997).

¶ 63 But the doctrine has expanded beyond both utilities and federal agencies. Now, it precludes a challenge to a regulated entity's rates filed with any governmental agency — state or federal — having regulatory authority over the entity. See, e.g., Wegoland Ltd. v. NYNEX Corp., 27 F.3d 17, 20 (2d Cir.1994) ("[C]ourts have uniformly held, and we agree, that the rationales underlying the filed rate doctrine apply equally strongly to regulation by state agencies."). And the clear majority of courts have concluded that the doctrine bars claims challenging insurance industry rates.18

¶ 64 Cases in the majority express two primary rationales for broadly applying the filed rate doctrine: first, the nondiscrimination view, which "prevent[s] carriers from engaging in price discrimination as between ratepayers"; and, second, the nonjusticiability view, which recognizes "the exclusive role of ... agencies in approving rates ... by keeping courts out of the rate-making process." Marcus v. AT & T Corp., 138 F.3d 46, 58 (2d Cir.1998).

¶ 65 As applied in proceedings that do not directly involve rate-making, the nondiscrimination view precludes "an action [that] may result in different prices being paid by victorious plaintiffs than non-suing ratepayers, which violates the statutory scheme of uniform rates." Edge v. State Farm Mut. Auto. Ins. Co., 366 S.C. 511, 623 S.E.2d 387, 391-92 (2005). The nonjusticiability view performs three functions:

Id. This view better explains applying the doctrine in the insurance context, given Colorado's extensive regulation of that industry.

¶ 66 The General Assembly created the Department of Insurance (DOI) to regulate and oversee the insurance industry. See §§ 10-4-401 to -421, C.R.S.2013. Regulation ensures that rates are "not ... excessive, inadequate, or unfairly discriminatory." § 10-4-401(1). All such rates must be filed with and approved by the Insurance Commissioner, who has the "power to investigate violations, issue cease and desist orders, levy limited monetary penalties, and order the suspension or revocation of insurance licenses." Showpiece Homes Corp. v. Assurance Co. of Am., 38 P.3d 47, 53 (Colo.2001); see also §§ 10-4-404, -405, -418, C.R.S.2013.

¶ 67 In the face of this broad authority, not applying the filed rate doctrine would invite judicial second-guessing of agency-approved premium rates and could produce disharmony between judicial determinations and those of the DOI. And declining to apply this doctrine would disregard the DOI's greater expertise. Cf. Egle v. City & Cnty. of Denver, 93 P.3d 609, 613 (Colo.App.2004) (exhaustion doctrine recognizes that administrative agencies "have expertise in the subject matter before them and are in a better position than the court to adjudicate matters involving that expertise").

¶ 68 Thus, to preserve "the stability, uniformity, and finality inherent in rates filed with" the DOI, Edge, 623 S.E.2d at 392, we conclude that the filed rate doctrine applies to Colorado's insurance industry. In so holding, we "align our decision with the considerable weight of authority from other jurisdictions that have applied the filed rate doctrine to rate-making in the insurance industry." Richardson v. Standard Guar. Ins. Co., 371 N.J.Super. 449, 853 A.2d 955, 963 (App.Div. 2004). However, this conclusion does not resolve whether the filed rate doctrine also applies to consumer fraud actions, and if so, whether here it bars plaintiffs from obtaining a refund of UM/UIM premiums for additional vehicles as damages on their initial purchase theory.

1. The Filed Rate Doctrine Applies to Consumer Fraud Claims

¶ 69 Consistent with the majority rule applying the filed rate doctrine to the insurance industry, a majority of courts have also held that the doctrine applies to fraud claims against an insurer by an insured. See, e.g., Weinberg v. Sprint Corp., 173 N.J. 233, 801 A.2d 281, 287 (2002) ("[T]he filed rate doctrine bars money damages ... where the damage claims are premised on state contract principles, consumer fraud, or other bases on which plaintiffs seek to enforce a rate other than the filed rate.")19

¶ 70 The limited contrary authority cited by plaintiffs is unpersuasive. The statement that "[t]he filed tariff doctrine ... is of no help to a defendant which fraudulently induces a plaintiff to pay a filed rate" in Nordlicht v. N.Y. Tel. Co., 617 F.Supp. 220, 227-28 (S.D.N.Y.1985), aff'd 799 F.2d 859 (2d Cir. 1986), is dictum. And in Tenore v. AT & T Wireless Servs., 136 Wn.2d 322, 962 P.2d 104, 109-110 (1998), the court held that "this case does not implicate the `filed rate' doctrine" because the defendants, as "mobile radio service providers ... are specifically exempted from tariff filing requirements by the FCC." (Footnote omitted.) Thus, again, we side with the majority.

2. The Doctrine Precludes Only Damages that Implicate the Reasonableness of Filed Rates

¶ 71 The harder question is how the doctrine affects fraud claims. Contrary to plaintiffs' implication, the trial court's C.R.C.P. 56(h) ruling addressed the impact of the filed rate doctrine on damages, not whether that doctrine precluded an entire claim. ("Plaintiffs cannot ask for a refund of the premium or any portion of the premium based on an argument that the coverage was not worth the amount charged.") Hence, this appeal does not require us to address when, if ever, the doctrine would bar an entire claim, and we do not do so.

¶ 72 Instead, the trial court held that the doctrine precludes only those damage theories that directly or indirectly involve the reasonableness of filed rates. But the trial court also recognized that where another damage theory does not implicate reasonableness of rates, it may be advanced. We agree because this distinction reflects the rationale behind the doctrine: "the impact [that] the court's decision will have on agency procedures and rate determinations." Rios v. State Farm Fire & Cas. Co., 469 F.Supp.2d 727, 737 (S.D.Iowa 2007) (internal quotation marks omitted).

¶ 73 Plaintiffs' two arguments in support of refunding UM/UIM premiums as damages for fraudulent inducement are unpersuasive.

¶ 74 First, plaintiffs cite to several rescission cases allowing recovery of insurance premiums. E.g. Cent. Life Assur. Soc'y of U.S. v. Mulford, 45 Colo. 240, 100 P. 423, 424 (Colo.1909). But plaintiffs do not seek to rescind their USAA policies. And the law does not allow partial rescission, even in cases of fraud. See, e.g., Trimble v. City & Cnty. of Denver, 697 P.2d 716, 723 (Colo. 1985) ("Although ... [the defendant's] intention not to honor the settlement agreement constituted fraud, partial rescission of the agreement is not a proper remedy for that misconduct.").

¶ 75 Second, plaintiffs rely on Showpiece to argue that applying the filed rate doctrine to a fraud-based claim against an insurer would "require the plaintiff to establish damages without reference to the actual injury incurred and would frustrate the clearly stated legislative purpose" of the Unfair Competition — Deceptive Practices Act, §§ 10-3-1101 to -1116, C.R.S.2013 (Unfair Claims Act). Plaintiffs' reliance is misplaced.

¶ 76 Showpiece held that a claim under the CCPA was not preempted by the Unfair Claims Act because "[t]o exempt insurance companies from the purview of the CCPA would frustrate the broad remedial purposes of the Act." 38 P.3d at 53. It did so based on preemption principles and rules of statutory construction, without addressing either the filed rate doctrine or appropriate measures of damages. See id. at 52-55.

¶ 77 But here, as the trial court noted, plaintiffs can still recover damages under the split coverage theory, which is "that they would not have purchased the UM/UIM coverage on additional vehicles at all."20 This distinction between the CCPA claims upheld in Showpiece and the particular measure of damages precluded here undercuts plaintiffs' reliance on the case. It also addresses plaintiffs' assertion that further proceedings will disregard "the actual injury incurred."

¶ 78 As plaintiffs correctly point out, section 10-3-1114 of the Unfair Claims Act includes a carve-out for common law claims.21 But to the extent plaintiffs assert that the carve-out is a legislative judgment counseling against applying the filed rate doctrine, this provision speaks only to claims. As indicated, the trial court did not dismiss any claim as barred by this doctrine. And plaintiffs do not reference, nor have we found, any provision that weighs against applying the doctrine to limit damages. Thus, the structure of the Unfair Claims Act is uninformative in applying the filed rate doctrine to preclude a refund of premiums as damages.

¶ 79 Absent any such statutory bar, analysis of the trial court's distinction between plaintiffs' initial purchase theory that because "they were fraudulently induced to purchase UM/UIM coverage on additional vehicles," the premiums obtained by fraud should be refunded, and their split coverage theory, centers on the reason for applying the filed rate doctrine in fraud cases. Under the nonjusticiability view, as germane to the insurance industry, "[t]he dispositive question... is whether, if plaintiffs succeed on their damages claims, the court's determination will impact the agency's rate determinations." Coll v. First Am. Title Ins. Co., 642 F.3d 876, 890 (10th Cir.2011).

¶ 80 Here, plaintiffs' refund damage theory implicates reasonableness of rates because before DeHerrera, USAA's premiums for UM/UIM on additional vehicles covered two increased risks: claims by the insured and resident relatives, as well as claims by guests and nonresident relatives. But after DeHerrera, covering additional vehicles only increased the second risk.22 Hence, plaintiffs' argument that their fraudulent inducement claim does not challenge the reasonableness of USAA's premiums, but challenges only USAA's business practices, is unpersuasive.

¶ 81 Instead, as the trial court pointed out, plaintiffs' "underlying assertion is that the rates charged are unreasonable, given the benefits received. Plaintiffs' claims are essentially claims of overcharging for UM/UIM premiums on additional vehicles." However, the court explained:

¶ 82 We agree that plaintiffs' refund damage theory would require the trier of fact to determine the rate USAA could reasonably have charged for UM/UIM coverage on additional vehicles obtained post-DeHerrera. This is so because a total refund is not the measure of plaintiffs' actual injury for fraudulent inducement. As the trial court found, with record support, additional vehicle coverage provided class two coverage, which would have value to some USAA policy holders.

¶ 83 Such an assessment would invade the province of the DOI under section 10-4-403(1). And by continuing to approve USAA's rates after DeHerrera, the Insurance Commissioner has found that the premiums charged for such additional coverage were not excessive. Thus, this damage theory would involve judicial second-guessing of the approved insurance rates, which the doctrine's nonjusticiability rationale forbids. See McCray v. Fid. Nat'l Title Ins. Co., 682 F.3d 229, 242 (3d Cir.2012).

¶ 84 Plaintiffs' assertion of this damage theory based on an underlying claim of consumer fraud does not change the foregoing analysis because "[t]here is no fraud exception to the filed rate doctrine." AT & T Corp. v. JMC Telecom, LLC, 470 F.3d 525, 535 (3d Cir.2006); see also Horwitz ex rel. Gilbert v. Bankers Life & Cas. Co., 319 Ill.App.3d 390, 253 Ill.Dec. 468, 745 N.E.2d 591, 605 (2001) (holding that the filed rate doctrine barred private right of action for consumer fraud) (applying Colorado law); Richardson, 853 A.2d at 964-65 (holding that the filed rate doctrine precludes insured's fraudulent inducement claim).23

¶ 85 Accordingly, we conclude that the trial court did not err in holding that a complete or partial refund of UM/UIM premiums for additional vehicles was not a permissible theory of damages on the fraudulent inducement claim.

V. Conclusion

¶ 86 We affirm the trial court's order denying class certification and remand the case for further proceedings consistent with this opinion.

JUDGE BERNARD and JUDGE PLANK* concur.