HAMILTON, Circuit Judge.

Under Federal Rule of Civil Procedure 23(b)(3), a class may be certified only if questions of law and fact common to members of the class predominate over questions affecting only individual members of the class. In this case, plaintiff-appellant Steven Messner and other named plaintiffs alleged that a merger between defendant-appellee Northshore University Health-System and Highland Park Hospital violated federal antitrust law. In fact, the Federal Trade Commission found that the merger violated section 7 of the Clayton Act, 15 U.S.C. § 18. Plaintiffs seek treble damages and injunctive relief under section 4 of the Clayton Act, 15 U.S.C. § 15, and they seek certification of a class of individual patients and third-party payors who allegedly paid higher prices for hospital care as a result of the merger.

One key issue on the merits will be proof that the merger had an antitrust impact on the plaintiff class, primarily in the form of higher prices. To show the predominance of common questions regarding the merger's antitrust impact on class members, plaintiffs proposed to rely on the same economic and statistical methods used by the Federal Trade Commission staff and Northshore's own economic experts to analyze antitrust impact in the merger litigation before the FTC. The basic method, called "difference-in-differences," is designed to estimate the amount of Northshore's price increases that resulted from exercise of market power rather than from other factors. This analysis, plaintiffs claimed, will show that Northshore leveraged its newfound market power to impose higher prices on insurers and patients.

The district court denied plaintiffs' motion for class certification, concluding that their expert had not shown that his proposed methodology could address the antitrust impact issue on a class-wide basis. The district court believed that plaintiffs' proposed methodology required proof that defendant raised its prices at uniform rates affecting all class members to the same degree. Finding a lack of uniformity in price increases, the district court concluded that plaintiffs could not show predominance and that class certification should be denied. The district court based this conclusion on its belief that plaintiffs' expert had conceded that the common methodological framework by which he proposed to demonstrate impact to members of the class was invalid absent uniform price increases. Plaintiffs sought interlocutory appeal under Rule 23(f).

Because of the importance of the issue for this case and for private antitrust enforcement, particularly with respect to hospitals and health care providers with complex pricing systems, we granted the petition for interlocutory appeal. We find that the district court's conclusion that a lack of uniform price increases required denial of class certification was erroneous as a matter of both fact and law, resulting in a denial that we must find was an abuse of discretion. Although plaintiffs' expert initially believed that Northshore did in fact increase its prices uniformly across all services, he acknowledged that it might not have done so, and he explained how his common methodology could show impact to the class despite such complications. Apart from the expert's supposed concession, the degree of uniformity the district court demanded simply is not required for class certification under Rule 23(b)(3). In essence, it is important not to let a quest for perfect evidence become the enemy of good evidence. We vacate the district court's denial of class certification and remand this matter for further proceedings.

We begin by reviewing Northshore's merger and the FTC proceedings that found it unlawful, and then turn to the proceedings in the district court on class certification. On the merits of the appeal, we consider first the district court's procedural handling of a challenge to the testimony of Northshore's expert witness, then the central substantive issue of proving antitrust impact on a class-wide basis. We conclude by considering some additional objections to class certification raised by Northshore.

I. The Merger and the FTC Proceedings

On January 1, 2000, defendant Northshore, then doing business as Evanston Northwestern Healthcare Corporation, merged with Highland Park Hospital, located in Highland Park, Illinois. Prior to the merger, Northshore owned Evanston Hospital in Evanston, Illinois, as well as Glenbrook Hospital in nearby Glenview, Illinois.

In February 2004, the Federal Trade Commission filed an administrative complaint against Northshore alleging that the merger violated section 7 of the Clayton Act, 15 U.S.C. § 18, by substantially lessening competition for general acute care inpatient hospital services in the "area directly proximate to the three [Northshore] hospitals and contiguous geographic areas in northeast Cook County and southeast Lake County, Illinois." In re Evanston Northwestern Healthcare Corp., 2007 WL 2286195, at *3 (F.T.C. Aug. 6, 2007). Following an eight-week trial, an administrative law judge concluded that the merger violated the Clayton Act. The judge ordered Northshore to divest Highland Park Hospital. Id. at *4.

On appeal in 2007, the Federal Trade Commission agreed with the ALJ that the merger enabled Northshore to exercise increased market power and that it used that power to increase its prices by a substantial amount. The FTC pointed out that Northshore's own economic expert found a price increase of nine to ten percent. Id. at *66. The FTC concluded that the evidence as a whole demonstrated that Northshore's "substantially higher-than-predicted merger-coincident price increases were due to market power, rather than competitively-benign factors." Id. None of Northshore's alternative explanations for those price increases, the FTC concluded, were supported by the record. Id. The FTC rejected Northshore's arguments that the merger made Highland Park a meaningful competitor in the market, that the merger's anticompetitive effects were outweighed by quality improvements at Highland Park resulting from the merger, and that Northshore's not-for-profit status reduced the potential for anticompetitive harm. Id. at *67-*73. The FTC affirmed the ALJ's ruling that the merger violated section 7 of the Clayton Act. Id. at *76.

When it came to the question of remedy, however, the FTC concluded that divestiture of Highland Park was not required. The FTC instead required Northshore to use "separate and independent" teams— one for Evanston and Glenbrook, and another for Highland Park—to negotiate contracts going forward. Id. at *77-*79. This remedy, the Commission concluded, would provide for effective competition between the hospitals and avoid the "complex, lengthy, and expensive process" of divestiture. Id. at *79.

II. Proceedings in the District Court

In April 2008, plaintiff Messner filed this class action suit against Northshore. (Dkt. 64, Ex. A.) Messner's suit was one of several similar actions challenging the merger, all of which were consolidated under the title In re Evanston Northwestern Healthcare Antitrust Litigation. In their consolidated class action complaint, Messner and the other named plaintiffs accuse Northshore of monopolization and attempted monopolization of the market for "general inpatient and hospital-based outpatient services" in the "the geographic triangle created by ... Evanston Hospital, Glenbrook Hospital, and Highland Park Hospital," in violation of section 2 of the Sherman Act, 15 U.S.C. § 2. (Dkt. 224.) They also allege that the merger substantially lessened competition in that market in violation of section 7 of the Clayton Act, 15 U.S.C. § 18. Plaintiffs bring their claims under sections 4 and 16 of the Clayton Act, 15 U.S.C. §§ 15 & 26, requesting injunctive relief and treble damages for injuries they suffered as a result of the alleged antitrust violations.

Plaintiffs moved for class certification pursuant to Rule 23(b)(3) on behalf of "All persons or entities ... who purchased or paid for inpatient hospital services or hospital-based outpatient services directly from Northshore ..., its wholly-owned hospitals, predecessors, subsidiaries, or affiliates... from at least as early as January 1, 2000 to the present."1 In support of that motion, plaintiffs offered the expert report of Dr. David Dranove, an economist on the faculty of Northwestern University who specializes in the health care industry. He would use common economic and econometric methods to prove the antitrust impact of Northshore's actions on the class and to estimate damages. Dranove would do so with the difference-in-differences method, by comparing "the percentage change in [Northshore's] prices between the pre- and post-merger periods ... to the percent change in prices at a control group of local hospitals during the same period." App. 126. "If the percentage change at [Northshore] is higher than the change at the control group by a statistically significant amount," plaintiffs said, "impact can be demonstrated." App. 126-27. Plaintiffs said that this same method could also provide an estimate of damages to individual class members.

In light of the FTC's findings that the merger had violated the law and enabled Northshore to raise its prices at least nine or ten percent above competitive prices, it is understandable that Northshore put up a determined opposition to class certification. The central issue under Rule 23(b)(3) became whether plaintiffs could show on a class-wide basis the antitrust impact of Northshore's actions on the proposed class. Northshore argued that plaintiffs' proposed class included a number of members who, for a variety of reasons, were not affected by the alleged price increases, and that plaintiffs had failed to propose "a common methodology for identifying purported class members... included within these `no impact' categories." In support of this argument, Northshore relied primarily on the expert testimony and report of Dr. Monica Noether, an expert on whom it had also relied during the FTC proceedings.

After extensive briefing and an evidentiary hearing, the district court denied plaintiffs' motion for class certification. In re Evanston Northwestern Healthcare Corp. Antitrust Litig., 268 F.R.D. 56 (N.D.Ill.2010). Although the district court found that plaintiffs' proposed class satisfied the requirements of Rule 23(a), it concluded that questions of law and fact individual to proposed class members regarding the antitrust impact of the merger predominated over questions common to the class as a whole. Id. at 61-65, 87.2

III. Analysis

A. Requirements for Class Certification

To be certified, a proposed class must satisfy the requirements of Federal Rule of Civil Procedure 23(a), as well as one of the three alternatives in Rule 23(b). Siegel v. Shell Oil Co., 612 F.3d 932, 935 (7th Cir.2010). As a threshold matter, a proposed class must always meet the Rule 23(a) requirements of numerosity, typicality, commonality, and adequacy of representation. When certification is sought under Rule 23(b)(3), as it is here, proponents of the class must also show: (1) that the questions of law or fact common to the members of the proposed class predominate over questions affecting only individual class members; and (2) that a class action is superior to other available methods of resolving the controversy. Id.

In conducting this analysis, the court should not turn the class certification proceedings into a dress rehearsal for the trial on the merits. See, e.g., Schleicher v. Wendt, 618 F.3d 679, 685 (7th Cir.2010); Kohen v. Pacific Investment Management Co., 571 F.3d 672, 677 (7th Cir.2009); Payton v. County of Kane, 308 F.3d 673, 677 (7th Cir.2002). On issues affecting class certification, however, a court may not simply assume the truth of the matters as asserted by the plaintiff. If there are material factual disputes, the court must "receive evidence ... and resolve the disputes before deciding whether to certify the class." Szabo v. Bridgeport Machines, Inc., 249 F.3d 672, 676 (7th Cir.2001). Plaintiffs bear the burden of showing that a proposed class satisfies the Rule 23 requirements, see, e.g., Trotter v. Klincar, 748 F.2d 1177, 1184 (7th Cir.1984), but they need not make that showing to a degree of absolute certainty. It is sufficient if each disputed requirement has been proven by a preponderance of evidence. Teamsters Local 445 Freight Div. Pension Fund v. Bombardier Inc., 546 F.3d 196, 202 (2d Cir.2008).

We review the denial of plaintiffs' motion for class certification for an abuse of discretion. See Arreola v. Godinez, 546 F.3d 788, 794 (7th Cir.2008). If, however, the district court bases its discretionary decision on an erroneous view of the law or a clearly erroneous assessment of the evidence, then it has necessarily abused its discretion. Cooter & Gell v. Hartmarx Corp., 496 U.S. 384, 405, 110 S.Ct. 2447, 110 L.Ed.2d 359 (1990); accord, Ervin v. OS Restaurant Services, Inc., 632 F.3d 971, 976 (7th Cir.2011) (reversing denial of class certification).

Plaintiffs argue here that the district court made two reversible errors, one procedural, the other substantive. First, they contend that the district court failed to determine whether defense expert Noether's report and opinions were admissible under Federal Rule of Evidence 702 before ruling on the motion for class certification. Second, they argue that the district court incorrectly applied Rule 23(b)(3)'s predominance requirement. We agree on both points, turning our attention first to the procedural issue, then to the substantive issue, and finally to additional arguments that Northshore makes in support of the denial.

B. Daubert and Class Certification

Under Rule 702 and Daubert v. Merrell Dow Pharmaceuticals Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), expert testimony is admissible only if (1) the expert testifies to valid technical, scientific, or other specialized knowledge; and (2) that testimony will assist the trier of fact. NutraSweet Co. v. X-L Eng'g Co., 227 F.3d 776, 787-88 (7th Cir.2000). Before the hearing on class certification, plaintiffs moved to exclude the report of defendant's expert, Dr. Monica Noether, a private consulting economist. Plaintiffs argued that Noether's "economic analyses are fundamentally defective" and that her opinion "should be stricken as a whole."3

The district court denied plaintiffs' motion. Although it agreed that "Noether's report ... include[s] some misleading information and analysis," the court concluded that plaintiffs' "two opportunities—in their reply brief and at oral argument—to respond to the conclusions contained in Noether's report" were sufficient to address the report's failings. Evanston Northwestern Healthcare, 268 F.R.D. at 77. For this reason, the district court declined to "undertake a Daubert analysis at this procedural juncture," explaining that it was giving "Noether's report the weight it believes it is due." Id.

When an expert's report or testimony is "critical to class certification," we have held that a district court must make a conclusive ruling on any challenge to that expert's qualifications or submissions before it may rule on a motion for class certification. American Honda Motor Co. v. Allen, 600 F.3d 813, 815-16 (7th Cir. 2010); see also Wal-Mart Stores, Inc. v. Dukes, ___ U.S. ___, 131 S.Ct. 2541, 2553-54, 180 L.Ed.2d 374 (2011) (expressing doubts regarding district court's conclusion that "Daubert did not apply to expert testimony at the certification stage of class-action proceedings").4

In American Honda, we used the word "critical" broadly to describe expert testimony important to an issue decisive for the motion for class certification. If a district court has doubts about whether an expert's opinions may be critical for a class certification decision, the court should make an explicit Daubert ruling. An erroneous Daubert ruling excluding non-critical expert testimony would result, at worst, in the exclusion of expert testimony that did not matter. Failure to conduct such an analysis when necessary, however, would mean that the unreliable testimony remains in the record, a result that could easily lead to reversal on appeal.

The district court's refusal to rule on plaintiffs' Daubert motion was an error under American Honda. Noether's opinions were undoubtedly "critical" to the district court's decision. Her report and testimony laid the foundation for Northshore's entire argument in opposition to class certification, and the district court obviously relied on Noether's reasoning when making its decision, quoting and discussing it many times. E.g., Evanston Northwestern Healthcare, 268 F.R.D. at 86 (noting that Noether's "analysis ... cast[s] doubt" on Dranove's contract analysis); id. (observing that Noether's supplemental report "suggest[s]" errors in Dranove's contract analysis). Given the importance of Noether's opinions, the district court needed to rule conclusively on plaintiffs' challenge to her opinions before it turned to the merits of plaintiffs' motion for class certification.

Instead of ruling on the admissibility of Noether's report, the court said it would give the report "the weight ... it is due." Id. at 77. We recognize that this is a time-honored and often acceptable approach toward many difficult evidentiary issues when the judge is the trier of fact. This approach does not suffice, however, when expert testimony is in fact critical to class certification. As we explained in American Honda, a district court cannot merely "leave[ ] open the questions of what portions of [the expert's] testimony it may have decided (or will decide) to exclude." American Honda, 600 F.3d at 816. Those tough questions must be faced and squarely decided. Id. at 817, citing West v. Prudential Securities, Inc., 282 F.3d 935, 938 (7th Cir.2002); see also Szabo, 249 F.3d at 676 ("Before deciding whether to allow a case to proceed as a class action ... a judge should make whatever factual and legal inquiries are necessary under Rule 23.").

To avoid this conclusion, Northshore proposes that we adopt the asymmetric rule that a definitive Daubert ruling is necessary only when a district court grants class certification, as in American Honda, but not when the court denies certification, as here. In effect, Northshore argues that a plaintiff should be allowed to rely on an expert's opinion in support of class certification only if that opinion is backed by reliable methods and information, but that a defendant may rely on unqualified or unhelpful "expert" opinions.

This result-oriented attempt to narrow American Honda finds support in neither the irrelevant cases cited by Northshore nor anything in American Honda itself. We did not suggest in American Honda that denials of class certification should be exempt from the strictures of Daubert and Rule 702. We made clear that whenever an expert's report or testimony is critical to a class certification decision, a district court must rule conclusively on a challenge to the expert's qualifications or opinions before ruling on class certification, without regard to whether the district court ultimately grants or denies that motion. See American Honda, 600 F.3d at 815-16. The ruling is just as important to the plaintiffs as it is to the defendants. Northshore's proposed rule would also create an unworkable logical conundrum, requiring a court to determine first whether to certify a class before considering the admissibility of the evidence it relied upon in making that determination.

We also reject two secondary arguments Northshore makes for its proposed limitation of American Honda. First, Northshore emphasizes that such a limitation must be read into American Honda because only plaintiffs bear the burden of satisfying Rule 23's requirements while defendants may present no evidence if they so choose. Northshore Br. 29, citing Carnegie v. Household Int'l, Inc., 376 F.3d 656, 662 (7th Cir.2004); In re American Medical Systems, Inc., 75 F.3d 1069, 1086 (6th Cir.1996). The general point about the burden of proof is correct but has no bearing on Rule 702, which applies to plaintiffs and defendants alike, regardless of which side bears the burden of proof. The fact that a defendant is not required to present evidence to defeat class certification does not give that defendant license to offer irrelevant and unreliable evidence. Second, Northshore argues that we must have meant for American Honda to apply only to decisions granting class certification because a Daubert hearing is unnecessary when certification is denied on grounds not addressed by the expert in dispute. (Northshore Br. 30). But a Daubert hearing is necessary under American Honda only if the witness's opinion is "critical" to class certification. That requirement is not met if the court decides the motion for class certification on grounds not addressed by the witness.

To conclude on this procedural issue, we decline Northshore's invitation to cut the holding of American Honda in half with a new exception for denials of class certification. The district court should have ruled definitively on plaintiffs' Daubert motion and objections before ruling on their motion for class certification. Northshore also argues that any error under American Honda was harmless. We disagree. As explained in the following section, the district court frequently discussed Noether's opinions in reaching the substantive decision that we find erroneous. We proceed to the primary substantive dispute between the parties regarding the proper application of Rule 23(b)(3) to the facts of this case.

C. Predominance and Antitrust Impact

Rule 23(b)(3) permits class certification only if the questions of law or fact common to class members "pre-dominate" over questions that are individual to members of the class.5 There is no mathematical or mechanical test for evaluating predominance. See 7AA Wright & Miller, Federal Practice & Procedure § 1778 (3d ed. 2011). The Supreme Court has discussed predominance in broad terms, explaining that the Rule 23(b)(3) "inquiry trains on the legal or factual questions that qualify each class member's case as a genuine controversy," with the purpose being to determine whether a proposed class is "sufficiently cohesive to warrant adjudication by representation." Amchem Products, Inc. v. Windsor, 521 U.S. 591, 623, 117 S.Ct. 2231, 138 L.Ed.2d 689 (1997). While similar to Rule 23(a)'s requirements for typicality and commonality, "the predominance criterion is far more demanding." Id. at 623-24, 117 S.Ct. 2231. At the same time, the Supreme Court also commented in Amchem: "Predominance is a test readily met in certain cases alleging consumer or securities fraud or violations of the antitrust laws." Id. at 625, 117 S.Ct. 2231. We understand the comment to mean that careful application of Rule 23 is necessary in antitrust cases, as in all cases, and that in antitrust cases, "Rule 23, when applied rigorously, will frequently lead to certification." Robert H. Klonoff, Antitrust Class Actions: Chaos in the Courts, 11 Stan. J.L. Bus. & Fin. 1, 7 (2005) (discussing Amchem); accord, Behrend v. Comcast Corp., 655 F.3d 182, 191 (3d Cir. 2011).

Rule 23(b)(3)'s predominance requirement is satisfied when "common questions represent a significant aspect of [a] case and ... can be resolved for all members of [a] class in a single adjudication." Wright & Miller, supra, § 1778. Or, to put it another way, common questions can predominate if a "common nucleus of operative facts and issues" underlies the claims brought by the proposed class. In re Nassau County Strip Search Cases, 461 F.3d 219, 228 (2d Cir.2006), quoting Waste Mgmt. Holdings, Inc. v. Mowbray, 208 F.3d 288, 299 (1st Cir.2000). "If, to make a prima facie showing on a given question, the members of a proposed class will need to present evidence that varies from member to member, then it is an individual question. If the same evidence will suffice for each member to make a prima facie showing, then it becomes a common question." Blades v. Monsanto Co., 400 F.3d 562, 566 (8th Cir.2005). Individual questions need not be absent. The text of Rule 23(b)(3) itself contemplates that such individual questions will be present. The rule requires only that those questions not predominate over the common questions affecting the class as a whole.

Analysis of predominance under Rule 23(b)(3) "begins, of course, with the elements of the underlying cause of action." Erica P. John Fund, Inc. v. Halliburton Co., ___ U.S. ___, 131 S.Ct. 2179, 2184, 180 L.Ed.2d 24 (2011). Section 4 of the Clayton Act, 15 U.S.C. § 15, requires plaintiffs to prove: (1) that Northshore violated federal antitrust law; and (2) that the antitrust violation caused them some injury. In re Hydrogen Peroxide Antitrust Litig., 552 F.3d 305, 311 (3d Cir. 2008); Blades, 400 F.3d at 566; Bell Atlantic Corp. v. AT&T Corp., 339 F.3d 294, 302 (5th Cir.2003). The same cases, and many others, also show that plaintiffs also must show damages, but individual proof of this element of a claim under the Clayton Act is not an obstacle to a showing of predominance. It is well established that the presence of individualized questions regarding damages does not prevent certification under Rule 23(b)(3). See Wal-Mart v. Dukes, 131 S.Ct. at 2558 (deeming it "clear that individualized monetary claims belong in Rule 23(b)(3)"); Arreola, 546 F.3d at 801 (recognizing that "the need for individual damages determinations does not, in and of itself, require denial of [a] motion for certification" under Rule 23(b)(3)); Hardy v. City Optical, Inc., 39 F.3d 765, 771 (7th Cir.1994) ("There have been many antitrust class actions in which the relief sought was damages, and the fact that the damages would generally be different for each member of the class was not deemed an insuperable obstacle."); Allapattah Servs. v. Exxon Corp., 333 F.3d 1248, 1261 (11th Cir.2003) ("numerous courts have recognized that the presence of individualized damages issues does not prevent a finding that the common issues in the case predominate"), aff'd, 545 U.S. 546, 125 S.Ct. 2611, 162 L.Ed.2d 502 (2005); see also Klay v. Humana, Inc., 382 F.3d 1241, 1260 (11th Cir.2004) (only in rare, extreme cases would individual issues of damages be so complex as to defeat class certification under Rule 23(b)(3)).

In this case, common questions clearly predominate in regard to whether Northshore's merger violated federal antitrust law. The focus of the dispute here is on the second element, referred to as "antitrust impact," Hydrogen Peroxide, 552 F.3d at 311, or "fact of damage," Bell Atlantic, 339 F.3d at 302 n. 12. Under Rule 23(b)(3), plaintiffs had to show that it was possible to use common evidence to prove that Northshore's merger injured the members of the proposed class. To do so, plaintiffs presented Dranove's expert report and opinion. Dranove claimed that "if [Northshore] overcharged an insurer by a certain percentage, all or substantially all class members covered by that insurer will be overcharged by approximately the same percentage." App. 1900. As a result, he said, "Overcharges to an insurer result in injury to that insurer as well as to all or substantially all other class members who are covered by that insurer." App. 1901. As the issue was developed further in the district court, however, it became considerably more complex.

If the market for health care services functioned like a market for a generic, undifferentiated commodity (i.e., corn, wheat, or pork bellies) traded on an exchange with standard contract terms and little opportunity for individual bargaining, showing antitrust impact through such overcharges would have been relatively simple. In such a market, one can in theory, at least, estimate simple supply and demand curves to show that an acquisition of market power raised price and lowered supply. That's antitrust impact from monopolization.

Real markets are not as simple and elegant as the classic economic model, and the market for hospital services seems to be particularly complex. Insurers and other third-party payors negotiate sophisticated contracts with health care providers. Through multi-year contracts for health care services, the parties may lock those prices in place or negotiate for long-term price changes significantly different than would have been agreed if the prices were renegotiated each year. Factors such as a hospital's location, quality of services, and reputation also can affect the price of a particular service.

Adding even more complexity is the fact that insurers and health care providers negotiate contracts that cover not a single service but complex bundles of many different services and products. See, e.g., John C. Render & Neal A. Cooper, Survey of Recent Developments in Health Care Law, 37 Ind. L.Rev. 1161, 1189 (2004). A hypothetical bill for a Caesarian section, for example, might consist of a variety of bundled items: anesthesia, operating room use, surgeon's fee, post-operative care for the mother, newborn care for the baby, etc. A hospital could unbundle and re-bundle those items in different ways, adding some items in the overall charge and removing others, so that a later bill for a service still called a Caesarian section would charge for a different set of items and have a very different overall price. The record here reflects such complexity. For example, one contract repriced cardiology services after the physicians' fees were "unbundled" from the prices and were charged separately. The nominal prices of the hospital's charges for cardiology services dropped, but after accounting for the unbundling to ensure an apples-to-apples comparison, the overall prices increased significantly.

Even without such unbundling or re-bundling, the prices of the individual component items are themselves subject to a wide variety of market influences. For example, an advance in anesthesia technology might result in a sharp decrease in the cost of anesthesia at the same time that a new and higher standard of care for a related service requires expensive new machinery. Without any exercise of market power, therefore, the price for the bundled service (say, a Caesarian section or a cardiac catheterization) might go up, go down, or stay level, despite substantial changes in the prices of the components.

As a result of these complexities, changes in the nominal prices charged for particular services might actually conceal rather than reveal a health care provider's exercise of unlawful market power. The price for a service may remain nominally the same or even decrease, but only because changes in the prices of the underlying components of that service have changed or because the service has been restructured in a way that conceals the anticompetitive price increase.

Dranove proposed to account for these complexities by conducting what is known as a "difference-in-differences" or "DID" analysis. He would compare prices at Northshore's hospitals with prices at a control group of comparable area hospitals not party to the merger but otherwise presumably subject to the same market forces affecting prices in hospitals. App. 1901, 1904. The difference between price changes for Northshore and the control group, he explained, would estimate the average overcharge imposed on Northshore's patients due to Northshore's exercise of increased market power after the merger. App. 1904. "For example," he explained, if Northshore's hospitals "raise prices by 30% after the merger and a control group of hospitals raises price by 10%, ... the `difference-in-differences' is approximately 20%" and represents the approximate amount of Northshore's overcharges. App. 1921. If Northshore overcharged an insurer by this percentage, he explained, "all or substantially all class members covered by that insurer will be overcharged by approximately the same percentage." App. 1926. Accordingly, Dranove concluded, a contract's "increase in average price will have a common impact on all or substantially all class members." Id.

As things turned out, however, even that approach does not deal sufficiently with all of the relevant variations that could confound the antitrust impact analysis. In denying class certification, the district court concluded that the viability of Dranove's methodology turned on "whether [Northshore] increased prices at a uniform rate across services." Evanston Northwestern Healthcare, 268 F.R.D. at 85. The court added: "Dranove's method of proving classwide impact ... rests on an assumption that [Northshore] increased prices at a uniform rate across services." Id. Such uniformity was absent, the district court concluded, noting for example that "even a cursory examination of the 2000 Payor A contract and the September 22, 2002 Payor A contract makes clear that the prices of some services changed at a variable rate." Id. at 86.6 To the extent that Dranove claimed that the prices increased uniformly, the court believed that he "focused primarily on price changes within contracts—changes that are usually attributable to escalator clauses." Id. Price changes controlled by escalator clauses "were not due to an exercise of market power," Northshore's expert testified, because Northshore "had the opportunity to exercise market power not within a contract period, but only at the time of renegotiation." Id.7 Because plaintiffs' proposed method for proving class-wide impact "relies on an assumption that [Dranove has] not been able to validate," the district court concluded, plaintiffs failed to establish predominance in regard to antitrust impact, so class certification was denied. Id. at 87.

On appeal, the parties raise two general arguments regarding the district court's denial of class certification. For their part, plaintiffs contend that the district court applied an incorrect standard of predominance under Rule 23(b)(3) when it made uniformity of price increases a condition for class certification. In response, Northshore contends that the district court was correct to require uniformity of price increases because plaintiffs conceded that such uniformity was necessary to Dranove's methodology for showing impact to members of the class. We explain first that the district court applied too stringent a standard in evaluating predominance. We explain second that plaintiffs did not agree to or invite the use of the wrong standard.

1. Predominance and Non-Uniform Price Increases

The district court misapplied Rule 23(b)(3)'s predominance standard when it made uniformity of nominal price increases a condition for class certification. Under the proper standard, plaintiffs' "burden at the class certification stage [was] not to prove the element of antitrust impact," but only to "demonstrate that the element of antitrust impact is capable of proof at trial through evidence that is common to the class rather than individual to its members." Behrend v. Comcast Corp., 655 F.3d at 197, quoting with emphasis Hydrogen Peroxide, 552 F.3d at 311-12; accord, Schleicher, 618 F.3d at 686 (noting that Rule 23's present structure is the result of a "decision ... to separate class certification from the decision on the merits"); Blades, 400 F.3d at 566 ("To determine whether common questions predominate... the court must look only so far as to determine whether ... common evidence could suffice to make out a prima facie case for the class.").

Through his proposed difference-in-differences or DID analysis of the contracts between Northshore and its insurers, Dranove claimed that he could show whether and to what extent Northshore's post-merger price increases were the result of increased market power resulting from the merger. In other words, Dranove claimed that he could use common evidence—the post-merger price increases Northshore negotiated with insurers—to show that all or most of the insurers and individuals who received coverage through those insurers suffered some antitrust injury as a result of the merger. App. 2541. That was all that was necessary to show predominance for purposes of Rule 23(b)(3). See Hydrogen Peroxide, 552 F.3d at 311-12.

Contrary to Northshore's view, Dranove's ability to use common evidence to show impact on the class did not ultimately depend on assuming the uniformity of the nominal price increases imposed under any individual contract. For reasons we explained above, such uniformity would certainly simplify matters. It would allow Dranove to plug a single percentage—the uniform price increase imposed on all patients covered under an individual contract—into his DID analysis to calculate the antitrust impact on those patients covered by that contract. But as Dranove explained in his report, a lack of uniformity would only require him to do more DID analyses for each contract—one analysis for each individual non-uniform price increase imposed in the contract being analyzed. App. 2540. As a simple example, if one post-merger contract raised the cost of hypodermic needles by 30 percent but increased the cost of saline solution by only 20 percent, DID comparison to price changes for the control group for those individual price increases could still be used to show any antitrust impact those price increases had on all patients who paid for hypodermic needles, saline solution, or both under that contract. App. 2540-41. In a more complex world, multiple analyses would be needed to show more accurately a contract's precise impact on class members. That need does not change the fact that those analyses all rely on common evidence—the contract setting out the non-uniform price increases—and a common methodology to show that impact. The ability to use such common evidence and common methodology to prove a class's claims is sufficient to support a finding of predominance on the issue of antitrust impact for certification under Rule 23(b)(3). See Hydrogen Peroxide, 552 F.3d at 311-12.

By requiring uniformity of nominal price increases within and across contracts, the district court misread Rule 23(b)(3) to require a greater showing of common evidence than is contemplated by that rule. Under the district court's approach, Rule 23(b)(3) would require not only common evidence and methodology, but also common results for members of the class. That approach would come very close to requiring common proof of damages for class members, which is not required. To put it another way, the district court asked not for a showing of common questions, but for a showing of common answers to those questions. Rule 23(b)(3) does not impose such a heavy burden. See Blades, 400 F.3d at 566 ("The nature of the evidence that will suffice to resolve a question determines whether the question is common or individual."); see also Rule 23(b)(3) (requiring that common "questions" predominate). Because the district court applied the wrong legal standard when analyzing plaintiffs' motion for class certification, the district court abused its discretion when it denied the motion. See Ervin, 632 F.3d at 976; Hydrogen Peroxide, 552 F.3d at 312.

2. The Concession Issue

Northshore argues that these principles do not matter in the end because, it says, plaintiffs conceded that their case for common impact depended on uniform nominal price increases. In support, Northshore relies primarily on Dranove's confirmation at the hearing on the motion for class certification that "the viability of [his] method" came down to whether Northshore "really [did] increase prices at a uniform rate across services." Dkt. 418 at 41. On cross-examination, however, Dranove clarified that his "DID analysis can be performed with or without the uniformity." Id. at. 57. These statements, seemingly contradictory on their face, are easily reconciled once it is remembered that Dranove proposed two alternative methodologies: one in which the uniformity of merger-related price increases was presumed, and another in which such uniformity was absent. Which method to use depended on the degree of uniformity.

If price increases were, as Dranove initially believed, entirely or largely uniform, then he proposed to show the merger's impact on the individual class members by simply plugging the average price increase imposed by any given contract into his DID analysis. App. 1904-06, 1909, 1926, 1931, 2523-25, 2530, 2584. In those circumstances, the average price would accurately reflect the individual price increases found in that contract. If the average contract price went up an average of 20 percent, and all of the services in that contract experienced uniform price increases, each individual service also went up 20 percent in price. App. 1909, 2523, 2571.

This specific methodology relied on uniform nominal price increases, but the actual evidence was not that simple. As Dranove implicitly acknowledged in his reply report, if a contract's individual service prices went up at non-uniform rates due to Northshore's unequal exercise of market power, then DID analysis using that contract's overall average price increase would reveal little about the merger's antitrust impact on individual class members covered by that contract. App. 2523 n. 1. According to Dranove, however, it would be most unusual for a firm possessing market power in a geographic market to exercise that power selectively to increase the prices of only some of its services. App. 1931-33, 2539.

For this reason, Dranove believed that any non-uniform price increases observed in Northshore's contracts with insurers could be explained by what he called "restructuring," or changes in price resulting from variations in Northshore's marginal costs or the re-bundling of component services discussed above. App. 2530, 2543-45. Such restructuring, he said, was unrelated to market power, meaning that services exhibiting non-uniform price increases could be treated as if they were subject to the same percentage price increase imposed on all other services covered in the same contract. App. 2543-44. If, for example, the price for one service went up 30 percent while all other services in that same contract went up only 20 percent, that additional 10 percent increase would not be treated as an exercise of market power for purposes of his DID analysis. Only the 20 percent increase shared by all other services in the same contract would reflect the use of market power. App. 2544. As a result, he explained, any non-uniform price increases imposed in a single contract with an insurance provider did not foreclose his use of that contract's average price increase to calculate accurately the impact to all patients covered by that contract. App. 2545.

Even if non-uniform price increases in a contract resulted not from restructuring but from Northshore's differential exercise of market power across different services, Dranove explained that he could still use those contracts to show impact on the class members. At his deposition, Dranove explained that if his review of documents revealed a lack of uniformity unexplained by restructuring, he would simply "adapt the methodology." Dkt. 284, Ex. I at 113, 157-58. In his initial report, Dranove explained that he could adapt his analysis if needed to accommodate "selective" price increases regarding certain services. App. 1932. And his reply report showed exactly how he would do that. The reply report emphasized that the DID analysis was fully capable of addressing non-uniform price increases: "it is still possible to apply a common methodological framework to measure impact even [when Northshore] increased prices for different services at different rates." App. 2539. As noted above, he would do so simply by conducting as many DID analyses as there were non-uniform price increases in a particular insurer's contract with Northshore. App. 2540. In this way, Dranove explained, he would be able to calculate "different over-charges across different service categories" despite any non-uniform increase in the prices charged for those services. App. 2540-41.

In other words, uniformity of nominal price increases was not necessary to Dranove's proposed methodology for determining antitrust impact to the proposed class. This explains why Dranove was willing (though perhaps a little too reluctant for his own good) to concede the non-uniformity of the price increases in Northshore's post-merger contracts. In fact, Dranove acknowledged several times that Northshore's prices did not always increase uniformly, explaining that Northshore "almost invariably increase[d] prices at the same rate," App. 2523, and that price increases "are applied across-the-board for all or nearly all services," App. 2524, in the "vast majority of cases." App. 2525 (emphases added). He acknowledged that one contract called for "a dramatic decrease in the price" for some services at the same time it "impose[d] a significant increase in the price of other service[s]." App. 2543.

The data in Appendix D of Dranove's reply report became the focus of attention during the hearing, in the district court's decision, and on appeal. The parties make much of the evidence regarding "Payor A" contained in Exhibits 161 and 162 admitted under seal in the class certification hearing. Dranove had included the price changes in Payor A's contracts in his analysis showing generally uniform rates of price changes. Noether used Payor A's contracts to show he was wrong. The district judge focused on the issue, and counsel for the plaintiffs told the judge: "I think you're just going to have to look at the numbers yourself." Dkt. 418 at 127-28. The judge did so, and in her opinion denying class certification asserted that "of the 18 prices listed in the renegotiated September 22, 2002 contract, 6 increased at a uniform rate, 9 increased at variable rates, and 3 changed pricing methodologies from the previous contract, making it difficult to draw a comparison." 268 F.R.D. at 86.

We have also examined the contract, by comparing the 2002 prices in Exhibit 162 to the 2000 prices in Exhibit 161. The prices for eight categories of inpatient services all increased by approximately 6.0 percent. In other words, those price increases were uniform.8 For three categories of outpatient services, the pricing methodology stayed the same for two (a discounted percentage of billed charges), while the third changed from a flat rate per case to a percentage of the billed charges.9 Where the superficial variation occurred was in the pricing for specified cardiac services. There were nine categories. Five showed decreases of 9.3 to 13.0 percent. Two showed increases of 14.8 and 60 percent, respectively, and two changed from flat rates to a percentage of billed charges. The cardiac price changes, both in terms of variations and the significant price reductions, appear to be inconsistent with Dranove's approach. When one looks more closely, however, one sees that there was a significant restructuring of these services and their pricing. The baseline prices from 2000 all included the professional services of physicians. App. 2725. In the 2002 contract, the professional services of physicians have been removed from the prices. App. 2728. These superficially non-uniform changes in prices therefore merely pose the sort of manageable challenge that Dranove's methodology can handle. They do not undermine the methodology itself.

If Dranove believed that his entire methodology was invalidated by non-uniform price increases, he expressed that belief in a most unusual way. He admitted the existence of non-uniform increases in nominal prices. He offered an economic explanation—restructuring—why this apparent lack of uniformity was misleading. He included data in Appendix D of his reply report showing that some non-uniformity appeared even in contracts that had not undergone any restructuring. And he explained in his reports how he would account for that lack of uniformity. Dranove did not concede away (and certainly not in a single statement at the class certification hearing) the viability of the very methodology that he had defended so vigorously over the course of two lengthy expert reports. The district court's conclusion to the contrary was a clear error.10

D. Class Members Who Did Not, or Could Not, Suffer Injury

The district court based its denial of class certification on two critical errors: (1) a misapplication of Rule 23(b)(3)'s predominance standard; and (2) an erroneous belief that Dranove's DID methodology would be valid only if Northshore's contracts with insurers uniformly increased prices across all services. On appeal, Northshore argues that, even absent these errors, plaintiffs' proposed class cannot be certified because it contains many individuals who were not injured by Northshore's alleged exercise of market power. First, Northshore argues that the evidence shows that Blue Cross Blue Shield of Illinois, plaintiffs' "largest putative class member," as well as a number of other individuals, suffered no injury at all. Second, Northshore argues that, for several reasons, a number of class members could not have been harmed by its post-merger price increases. We address each of these analytically distinct categories of individuals in turn.

1. Blue Cross and Other Allegedly Uninjured Parties

Northshore first contends that a number of members of the putative class were not harmed by any post-merger price increases. Northshore argues that Blue Cross was not actually harmed by any post-merger price increases, relying largely on Blue Cross's affidavit stating, without any real explanation, that it "did not pay artificially inflated prices" and did not suffer "any injury or damage," App. 722-23, as well as the FTC's conclusions that Blue Cross experienced no merger-related price increases between the merger in 2001 and the FTC proceedings in 2005.11 As a result, Northshore argues, none of the class members who paid prices negotiated by Blue Cross were harmed either. Northshore makes a similar argument regarding individuals who it says were not injured because "any price increases were passed on or borne by someone other than the class member."12

All of this is at best an argument that some class members' claims will fail on the merits if and when damages are decided, a fact generally irrelevant to the district court's decision on class certification. See, e.g., Schleicher, 618 F.3d at 687 ("The chance, even the certainty, that a class will lose on the merits does not prevent its certification."); Payton v. County of Kane, 308 F.3d 673, 677 (7th Cir.2002) (observing that "a determination on the propriety of class certification should not turn on [the] likelihood of success on the merits"). As we have previously explained:

Kohen, 571 F.3d at 677.

The reasons a court may not decline to certify a class merely because it believes the class's claims will fail on the merits should be clear. For one thing, it is unlikely that discovery regarding the merits of a claim will be complete by the time the court is called upon to certify a class. See Fed.R.Civ.P. 23(c)(1) (requiring ruling on class certification at "an early practicable time"). Any consideration of the merits at the class certification stage also runs the risk of supplanting the jury as the finder of fact. See Behrend, 655 F.3d at 199. This risk is particularly troubling because the procedural protections available for such early judicial evaluations of the merits—such as the assumption under Rule 12(b)(6) that allegations in the complaint are true and the Rule 56 requirement to give the non-moving party the benefit of conflicting evidence—are not available under Rule 23. See Szabo, 249 F.3d at 675 ("The proposition that a district judge must accept all of the complaint's allegations when deciding whether to certify a class cannot be found in Rule 23 and has nothing to recommend it.").

Perhaps Northshore could have used its evidence regarding Blue Cross to argue that Dranove's methodologies were flawed. That would be an appropriate and limited use of merits evidence at the certification stage. See Schleicher, 618 F.3d at 685 ("a court may take a peek at the merits before certifying a class," but the peek must be "limited to those aspects of the merits that affect the decisions essential under Rule 23"). But Northshore never developed such an argument in its briefs to this court, thus waiving that argument on appeal. E.g., Awe v. Ashcroft, 324 F.3d 509, 512-13 (7th Cir.2003). The argument that Northshore has made, however, is one that we have rejected time and again. See Schleicher, 618 F.3d at 687; Kohen, 571 F.3d at 677; Payton, 308 F.3d at 677. We now reject yet again. Also, even if we could reach the merits of plaintiffs' claims involving Blue Cross, the barebones affidavit on which Northshore relies did not so thoroughly disprove those claims as to render any further presentation of evidence to the contrary pointless.13

2. Class Members "Immune" From Injury

Northshore next argues that class certification is inappropriate because the class contains a number of individuals who could not have been harmed by any post-merger price increases. Among such individuals, Northshore says, are those putative class members who "met their annual plan out-of-pocket maximum or their deductible regardless of any price increase," as well as those individuals whose contracts "protect[] against any price increases."

At first glance, it would seem that Northshore is arguing, as it did in regard to Blue Cross, that certification must be denied because plaintiffs' proposed class contains members whose claims will fail on the merits. In actuality, however, Northshore is arguing that the class for which certification is requested is fatally overbroad because it contains members who could not have been harmed by any post-merger price increases—Blue Cross certainly could have been harmed, but arguably might not have been for one reason or another.14

This distinction is critical for class certification purposes. As explained above, if a proposed class consists largely (or entirely, for that matter) of members who are ultimately shown to have suffered no harm, that may not mean that the class was improperly certified but only that the class failed to meet its burden of proof on the merits. See Schleicher, 618 F.3d at 686 ("Rule 23 allows certification of classes that are fated to lose as well as classes that are sure to win."). If, however, a class is defined so broadly as to include a great number of members who for some reason could not have been harmed by the defendant's allegedly unlawful conduct, the class is defined too broadly to permit certification. See Kohen, 571 F.3d at 677 (explaining that "if the [class] definition is so broad that it sweeps within it persons who could not have been injured by the defendant's conduct, it is too broad" and the class should not be certified). For example, if plaintiffs had sought certification of a class shown to include an high percentage of individuals who paid for medical services at Northshore's hospitals after the merger but under Northshore's pre-merger contracts with insurers (i.e., a multi-year contract signed in 1999), that class obviously could not be certified—it would contain a vast number of people who could not have been harmed by the merger because they purchased medical services in the absence of the market power allegedly created by that merger. See, e.g., Oshana v. Coca-Cola Co., 472 F.3d 506, 514 (7th Cir.2006) (affirming, in class action alleging deceptive advertising, denial of certification of class defined so broadly that it included "millions" of individuals who were not deceived).

This distinction—between class members who were not harmed and those who could not have been harmed—stems in part from the "in terrorem character of a class action." Kohen, 571 F.3d at 678. Even if a class's claim is weak, the sheer number of class members and the potential payout that could be required if all members prove liability might force a defendant to settle a meritless claim in order to avoid breaking the company. Id. While that prospect is often feared with large classes, the effect can be magnified unfairly if it results from a class defined so broadly as to include many members who could not bring a valid claim even under the best of circumstances. E.g., Oshana, 472 F.3d at 514. For this reason, "a class should not be certified if it is apparent that it contains a great many persons who have suffered no injury at the hands of the defendant." Kohen, 571 F.3d at 677. There is no precise measure for "a great many." Such determinations are a matter of degree, and will turn on the facts as they appear from case to case.

The problem posed by class members whose claims may fail on the merits for individual reasons is the obverse of a different problem with class definition: the problem of the "fail-safe" class: one that is defined so that whether a person qualifies as a member depends on whether the person has a valid claim. Such a class definition is improper because a class member either wins or, by virtue of losing, is defined out of the class and is therefore not bound by the judgment. See Randleman v. Fidelity Nat'l Title Ins. Co., 646 F.3d 347, 352 (6th Cir.2011) (fail-safe class definition was one of two grounds for decertifying class); Premier Electrical Const. Co. v. National Elec. Contractors Ass'n, 814 F.2d 358, 361-63 (7th Cir.1987) (explaining that Rule 23 was amended in 1968 to prevent "one-way intervention"); Adashunas v. Negley, 626 F.2d 600, 604 (7th Cir.1980) (affirming denial of class certification), quoting Dafforn v. Rousseau Associates, Inc., 1976 WL 1358 (N.D.Ind.1976) (Eschbach, J.); Campbell v. First American Title Ins. Co., 269 F.R.D. 68, 73-74 (D.Me. 2010); Roe v. Bridgestone Corp., 492 F.Supp.2d 988, 992 n. 1 (S.D.Ind.2007); Genenbacher v. CenturyTel Fiber Co. II, 244 F.R.D. 485, 488 (C.D.Ill.2007); Indiana State Employees Ass'n. v. Indiana State Highway Comm'n, 78 F.R.D. 724, 725 (S.D.Ind.1978).

Defining a class so as to avoid, on one hand, being over-inclusive and, on the other hand, the fail-safe problem is more of an art than a science. Either problem can and often should be solved by refining the class definition rather than by flatly denying class certification on that basis. See, e.g., Campbell, 269 F.R.D. at 73-74 (court revised class definition to correct problem); Lewis v. First American Title Ins. Co., 265 F.R.D. 536, 551 (D.Idaho 2010) (same); Carson P. ex rel. Foreman v. Heineman, 240 F.R.D. 456, 492 (D.Neb. 2007) (same); Flanagan v. Allstate Ins. Co., 228 F.R.D. 617, 618-19 (N.D.Ill.2005) (same).

We are not persuaded that plaintiff's proposed class is so overly broad as to require denial of certification in this case. As for the individuals whose contracts purportedly protected them from price increases, Northshore has given us no indication how many such individuals actually exist. In fact, Northshore's brief does not call our attention to even a single contract actually containing such a provision, let alone provide any basis to believe that a "great many" putative class members entered into such contracts. And Northshore admits that only about 2.4 percent of the putative class members paid only their out-of-pocket maximums or deductibles. While this may prove, depending on the ultimate size of the class at issue here, to be a significant number of additional plaintiffs, a 2.4 percent decrease in the size of the class is certainly not significant enough to justify denial of certification. Cf. Oshana, 472 F.3d at 514 (affirming denial of certification and noting that millions of people were improperly included in the proposed class).15 Accordingly, we reject Northshore's argument that plaintiffs' proposed class is impermissibly overbroad. Of course, the district court is free to revisit this issue at a later time if discovery shows that the number of members who could not have been harmed by the merger was more significant than it appears at this time. See Kohen, 571 F.3d at 679 (noting that defendant was free to depose a random sample of the class to determine whether an impermissibly high portion of the class could not have been harmed by the defendant's actions and, if so, request decertification of the class).

IV. Conclusion

As explained above, the district court's denial of class certification was based on a misinterpretation of the factual record, namely, the court's erroneous conclusion that Dranove had conceded away the validity of the common method by which he proposed to show antitrust impact on members of the proposed class. Once that erroneous conclusion is set aside, the evidence shows that Dranove can use common evidence and his difference-in-differences methodology to estimate the antitrust impact, if any, of Northshore's merger on the members of that class. Together with the common questions and evidence on other liability issues, this was sufficient to show predominance under Rule 23(b)(3). Northshore's remaining arguments against class certification are not persuasive. We VACATE the district court's order denying plaintiffs' motion for class certification and REMAND this matter for further proceedings consistent with this opinion.