BERZON

Michael S. Sorgen (argued), Andrea Adam Brott, and Ryan L. Hicks, Law Offices of Michael S. Sorgen, and Richard A. Hoyer, Hoyer and Associates, San Francisco, CA, for Plaintiffs–Appellants. Christine Van Aken (argued), Deputy City Attorney, Dennis J. Herrera, City Attorney, Elizabeth Salveson, Chief Labor Attorney, and Jonathan C. Rolnick, Deputy City Attorney, City of San Francisco, San Francisco, CA, for Defendant–Appellee.
Appeal from the United States District Court for the Northern District of California, Phyllis J. Hamilton, District Judge, Presiding. D.C. No. 4:08–cv–05180–PJH.
Before: J. CLIFFORD WALLACE, RAYMOND C. FISHER, and MARSHA S. BERZON, Circuit Judges.

OPINION

BERZON, Circuit Judge:

Several San Francisco police officers (“the plaintiffs” or “the officers”) over the age of forty performed well enough on an examination in 1998 to qualify for consideration for promotion to Assistant Inspector. They allege that a new policy of the San Francisco Police Department (“SFPD”) abandoning the examination as a basis for certain assignments worked a disparate impact based on age. The plaintiffs sought certification of a class composed of all SFPD officers over forty who had qualified on the 1998 examination.

The district court denied certification for want of commonality. We permitted the officers to appeal the denial of class certification under Federal Rule of Civil Procedure 23(f), and now reverse, holding improper the district court's reliance on merits issues unrelated to the pertinent commonality inquiry.

I.

Nearly three and a half decades ago, the Civil Service Commission of the City and County of San Francisco (“City”) entered into a consent decree to settle allegations of employment discrimination in the SFPD on the basis of race, sex, and national origin, in violation of Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e et seq.

In late 1998, the district court terminated that consent decree. By stipulation of the parties, however, the district court retained jurisdiction over issues related to a recently administered promotional examination—the Q–35 Assistant Inspector examination for promotion to the Investigations Bureau—which the parties agreed would be governed by the terms of the consent decree. On the same day, the district court issued an order outlining promotion procedures for officers who had sat for the Q–35 Assistant Inspector examination. That order required a minimum of 175 Assistant Inspector appointments from the list of police officers who had passed the examination, arranged in order of exam performance (“Q–35 List”). The first 110 promotions were to be made in rank order of performance. Subsequent appointments, the court ordered, “will be made from a sliding band that will start at rank 111. The band width is 84 points.” The district court authorized the City to review various “secondary criteria” when making selections within that sliding band. Between 1998 and 2006, the City selected 229 police officers from the Q–35 List for promotion to Assistant Inspector.

In 2005, the Chief of Police announced a change in promotion policy: To improve operational flexibility and rationalize the promotional progression, the SFPD would no longer promote Assistant Inspectors to the Investigations Bureau from the Q–35 List. Instead, the SFPD would administer a new Q–50 Sergeants examination and assign some newly promoted Sergeants, selected on the basis of their performance on the new exam, to the Investigations Bureau. Those Sergeants assigned to the Investigations Bureau would have duties previously assigned to Assistant Inspectors.

The City administered the Q–50 Sergeants Examination in 2006 and created a list of eligible officers (“Q–50 List”) the following year. Soon thereafter, the SFPD began promoting Sergeants from the recently compiled Q–50 List, many of whom were given investigative duties. Assistant Chief Morris Tabak agreed, in a deposition taken in this case, that he knew of no reason “that those appointments could not have been made from the Q–35 list[ ] that resulted from the 1998 exam.”

The officers initially filed this action in late 2008, alleging that SFPD's decision to use the Q–50 list instead of the Q–35 list for investigative assignments both constituted a pattern or practice of discrimination and generated a disparate impact on older officers in violation of the Age Discrimination in Employment Act (“ADEA”), 29 U.S.C. §§ 621–634. They also alleged parallel violations of California's Fair Employment and Housing Act (“FEHA”), Cal. Gov't Code §§ 12940–12951.

The officers subsequently sought certification of a class under their FEHA disparate impact claim. The district court denied the motion for certification, explaining that the officers failed to satisfy the requirements of both Federal Rules of Civil Procedure 23(a)(2) and 23(b)(3).

After denial of certification, the officers filed a Second Amended Complaint alleging only a disparate impact theory of liability, again under both the ADEA and FEHA. Several months later, the officers renewed their motion for class certification of the FEHA claim, proposing as the class to be certified SFPD “officers who were aged forty and older as of each of the dates of the challenged appointments in 2007, 2008, and 2009 and who could have been appointed to investigative positions had the City properly made appointments from the Q–35 List.” The putative class defined an officer “who could have been appointed,” as an officer within the 84–point selection band at the time SFPD assigned a sergeant to investigative work. The officers further proposed the creation of separate subclasses for those who could have been appointed at each date. Certification was sought under Rule 23(b)(3). As the common question required for certification, Fed.R.Civ.P. 23(a)(2), the officers asserted a common question of fact: “[W]hether the City's employment practice of appointing Sergeants exclusively from the Q–50 List to investigative positions traditionally performed by Assistant Inspectors and refusing to make any of the appointments from the active Q–35 List, had a disparate impact on qualified applicants for the investigative positions who were aged forty or above at the time of those appointments, in violation of FEHA.”

The district court denied the renewed motion for class certification for want of commonality. Having disposed of the motionon that ground, the district court expressly declined to rule on the officers's argument that the putative class satisfied the requirement of Rule 23(b)(3), which requires both that common questions “predominate over any questions affecting only individual members” of the class and “that a class action is superior to other available methods for fairly and efficiently adjudicating the controversy.”

The officers timely sought permission to appeal under Rule 23(f), which authorizes discretionary review of an interlocutory order granting or denying class certification. SeeFed.R.Civ.P. 23(f); see also Chamberlan v. Ford Motor Co., 402 F.3d 952, 959 (9th Cir.2005) (per curiam) (describing the standards we employ to guide our discretionary authorization of appeals under Rule 23(f)). This Court granted permission for the appeal, and the officers timely perfected it. SeeFed. R.App. P. 5(d).

II.

Federal Rule of Civil Procedure 23(a)(2) conditions class certification on demonstrating that members of the proposed class share common “questions of law or fact.” Such commonality is one of four “threshold requirements” contained in Rule 23(a); the other three are numerosity, typicality, and adequacy of representation. See, e.g., Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 613, 117 S.Ct. 2231, 138 L.Ed.2d 689 (1997). Together, these requirements seek to “limit the class claims to those fairly encompassed by the named plaintiff's claims.” Wal–Mart Stores, Inc. v. Dukes, ––– U.S. ––––, 131 S.Ct. 2541, 2550, 180 L.Ed.2d 374 (2011) (quoting Gen. Tel. Co. of Sw. v. Falcon, 457 U.S. 147, 156, 102 S.Ct. 2364, 72 L.Ed.2d 740 (1982)) (internal quotation marks omitted).

Wal–Mart instructed that, although Rule 23(a)(2) refers to common “questions of law or fact” in the plural, “ ‘[e]ven a single [common] question’ will do.” Id. at 2556 (alterations in original) (quoting id. at 2566 n. 9 (Ginsburg, J., dissenting)). We have since clarified “that Rule 23(a)(2) requires ... ‘a single significant question of law or fact.’ ” Abdullah v. U.S. Sec. Assocs., Inc., 731 F.3d 952, 957 (9th Cir.2013) (emphasis in original) (quoting Mazza v. Am. Honda Motor Co., 666 F.3d 581, 589 (9th Cir.2012)).

Rule 23(a)(2) is not “a mere pleading standard,” so establishing commonality sometimes requires affirmative evidence, which the courts must subject to “rigorous analysis.” Wal–Mart, 131 S.Ct. at 2551 (quoting Falcon, 457 U.S. at 161, 102 S.Ct. 2364). Such rigor often “will entail some overlap with the merits of the plaintiff's underlying claim.” Id. While some evaluation of the merits frequently “cannot be helped” in evaluating commonality, id., that likelihood of overlap with the merits is “no license to engage in free-ranging merits inquiries at the certification stage.” Amgen Inc. v. Conn. Ret. Plans & Trust Funds, ––– U.S. ––––, 133 S.Ct. 1184, 1194–95, 185 L.Ed.2d 308 (2013). Instead, as the Supreme Court clarified last year, “[m]erits questions may be considered to the extent— but only to the extent—that they are relevant to determining whether the Rule 23 prerequisites for class certification are satisfied.” Id. at 1195 (emphasis added). “[W]hether class members could actually prevail on the merits of their claims” is not a proper inquiry in determining the preliminary question “whether common questions exist.” Ellis v. Costco Wholesale Corp., 657 F.3d 970, 983 n. 8 (9th Cir.2011).

Further, a common contention need not be one that “will be answered, on the merits, in favor of the class.” Amgen, 133 S.Ct. at 1191. Instead, it only “must be of such a nature that it is capable of classwide resolution—which means that determination of its truth or falsity will resolve an issue that is central to the validity of each one of the claims in one stroke.” Wal–Mart, 131 S.Ct. at 2551 (emphases added). “To hold otherwise would turn class certification into a mini-trial” on the merits, Ellis, 657 F.3d at 983 n. 8, when the purpose of class certification is merely “to select the metho[d] best suited to adjudication of the controversy fairly and efficiently,” Amgen, 133 S.Ct. at 1191 (alteration in original) (internal quotation marks omitted).

Amgen, published over a year after the district court decision here, illustrates well the application of the principle that demonstrating commonality does not require proof that the putative class will prevail on whatever common questions it identifies. In Amgen, a class alleged securities fraud under § 10(b) of the Securities Exchange Act of 1934, 48 Stat. 891, as amended, 15 U.S.C. § 78j(b), and Securities and Exchange Commission Rule 10b–5, 17 C.F.R. § 240.10b–5. 133 S.Ct. at 1191. To succeed on the merits, plaintiffs alleging securities fraud under Rule 10b–5 “must prove ‘(1) a material misrepresentation or omission by the defendant; (2) scienter; (3) a connection between the misrepresentation or omission and the purchase or sale of a security; (4) reliance upon the misrepresentation or omission; (5) economic loss; and (6) loss causation.” ’ Id. at 1192 (quoting Matrixx Initiatives, Inc. v. Siracusano, ––– U.S. ––––, 131 S.Ct. 1309, 1317, 179 L.Ed.2d 398 (2011)) (internal quotation marks omitted). The fraud-on-the-market doctrine “permits certain Rule 10b–5 plaintiffs to invoke a rebuttable presumption of reliance on material misrepresentations aired to the general public,” when those misrepresentations concern securities traded in an efficient market. Id. To invoke the fraud-on-the-market presumption, “plaintiffs must demonstrate that the alleged misrepresentations were publicly known ..., that the stock traded in an efficient market, ... that the relevant transaction took place ‘between the time the misrepresentations were made and the time the truth was revealed,’ ” Erica P. John Fund, Inc. v. Halliburton Co., ––– U.S. ––––, 131 S.Ct. 2179, 2185, 180 L.Ed.2d 24 (2011) (quoting Basic Inc. v. Levinson, 485 U.S. 224, 248 n. 27, 108 S.Ct. 978, 99 L.Ed.2d 194 (1988)), and that the misrepresentations were material, Amgen, 133 S.Ct. at 1195.

Amgen held that where a class of investors seeks to rely on the fraud-on-the-market presumption, proof of materiality is unnecessary to certify a class under Rule 23(b)(3). Id. at 1191. The reasoning, in part, was that “[t]he alleged misrepresentations and omissions, whether material or immaterial, would be so equally for all investors composing the class.” Id. at 1191. If the factfinder ultimately held those representations immaterial, that determination would not amount to “some fatal dissimilarity among class members that would make use of the class-action device inefficient or unfair. Instead, [it would generate] a fatal similarity—failure of proof as to an element of the plaintiffs cause of action.” Id. at 1197 (emphases added) (alteration omitted) (quoting Richard A. Nagareda, Class Certification in the Age of Aggregate Proof, 84 N.Y.U. L.Rev. 97, 107 (2009)) (internal quotation marks omitted). By its very nature, such a similarity, whether fatal or not to the merits of the materiality issue, is a common question. Although proof of materiality was “an essential predicate of the fraud-on-the-market theory,” id. at 1195, the court held it “properly addressed at trial or in a ruling on a summary-judgment motion,” id. at 1197.

Notably, Amgen concerned satisfaction of Rule 23(b)(3), which requires not merely a common question but also “find[ing] that the questions of law or fact common to class members predominate over any questions affecting only individual members.” It held proof of a factor that would ultimately be essential to success on the class's claims not merely unnecessary to demonstrate commonality, but also unnecessary to demonstrate that such commonality predominates.Rule 23(b)(3) imposes a “far more demanding” standard than 23(a)(2). Amchem, 521 U.S. at 624, 117 S.Ct. 2231. It was the less demanding standard of Rule 23(a)(2) that was the basis of the district court's denial of certification here.

The principle that courts must consider merits issues only as necessary to determine a pertinent Rule 23 factor, and not otherwise, has special force at the appellate level where, as here, we review a class certification determination under Rule 23(f). Under Rule 23(f), the limitation on consideration of the merits to the relevant class certification questions is of jurisdictional significance. Rule 23(f) permits discretionary, interlocutory appeals “from an order granting or denying class-action certification.” Fed.R.Civ.P. 23(f). Interlocutory appeals are, of course, “the exception rather than the rule.” Chamberlan, 402 F.3d at 959. Absent Rule 23(f) or some other applicable exception, our jurisdiction would be constrained by “ ‘the general rule that a party is entitled to a single appeal, to be deferred until final judgment has been entered.’ ” Mohawk Indus., Inc. v. Carpenter, 558 U.S. 100, 106, 130 S.Ct. 599, 175 L.Ed.2d 458 (2009) (quoting Digital Equip. Corp. v. Desktop Direct, Inc., 511 U.S. 863, 868, 114 S.Ct. 1992, 128 L.Ed.2d 842 (1994)); see also28 U.S.C. § 1291. As the exception to the final judgment rule created by Rule 23(f) applies only to class certification decisions, merits inquiries unrelated to certification exceed our limited Rule 23(f) jurisdiction, as well as the needs of Rule 23(a)-(b). In contrast, were this an appeal from a final judgment after a class was certified (or not certified) and the merits then determined, we would have the option of addressing the merits and, if we decided to hold against the plaintiffs, not addressing the class certification question. We must police the bounds of our jurisdiction vigorously here as elsewhere, see, e.g., Crowley v. Bannister, 734 F.3d 967, 974 (9th Cir.2013), and so may not ourselves venture into merits issues unnecessary to the Rule 23 issue before us.

III.

With that background, we reach the only question properly before us—whether the district court's denial of the officers's motion for class certification was an abuse of discretion. See Ellis, 657 F.3d at 980. An “error of law[,] .... ‘reliance on an improper factor, ... omission of a substantial factor, or ... a clear error of judgment in weighing the correct mix of factors' ” qualifies as an abuse of discretion. Bateman v. Am. Multi–Cinema, Inc., 623 F.3d 708, 712 (9th Cir.2010) (quoting In re Wells Fargo Mortg. Overtime Pay Litig., 571 F.3d 953, 957 (9th Cir.2009)). We conclude that the district court erred in denying class certification because of its legal error of evaluating merits questions, rather than focusing on whether the questions presented, whether meritorious or not, were common to the members of the putative class. By doing so, the district court made an error of law and relied on improper factors, thereby abusing its discretion.

To assess whether the putative class members share a common question, the answer to which “will resolve an issue that is central to the validity of each one of the [class members's] claims,” we must identify the elements of the class members's case-in-chief. Wal–Mart, 131 S.Ct. at 2551. The requisite prima facie case for a disparate-impact age discrimination claim under California's FEHA is parallel to that under the ADEA. See Katz v. Regents of the Univ. of Cal., 229 F.3d 831, 835 (9th Cir.2000); see also Clark v. Claremont Univ. Ctr., 6 Cal.App.4th 639, 666, 8 Cal.Rptr.2d 151 (1992). Under both statutes, “a plaintiff must demonstrate ‘(1) the occurrence of certain outwardly neutral employment practices, and (2) a significantly adverse or disproportionate impact on persons of a particular [age] produced by the employer's facially neutral acts or practices.’ ” Katz, 229 F.3d at 835 (alteration in original) (quoting Palmer v. United States, 794 F.2d 534, 538 (9th Cir.1986)). This formulation requires a plaintiff to “ ‘isolat[e] and identify[ ] the specific employment practices that are allegedly responsible for any observed statistical disparities.’ ” Smith v. City of Jackson, 544 U.S. 228, 241, 125 S.Ct. 1536, 161 L.Ed.2d 410 (2005) (emphasis in original) (quoting Wards Cove Packing Co. v. Atonio, 490 U.S. 642, 656, 109 S.Ct. 2115, 104 L.Ed.2d 733 (1989)). The plaintiff must also demonstrate a causal connection between those specific employment practices and the asserted impact on those of a particular age. See, e.g., Lewis v. City of Chicago, 560 U.S. 205, 212, 130 S.Ct. 2191, 176 L.Ed.2d 967 (2010); Katz, 229 F.3d at 836.

Here, the officers have identified a single, well-enunciated, uniform policy that, allegedly, generated all the disparate impact of which they complain: the SFPD's decision to make investigative assignments using the Q–50 List instead of the Q–35 List. Each member of the putative class was on the Q–35 List. Each suffered the effects of its elimination, whatever those were.

“Identifying a specific practice is not a trivial burden” in age discrimination cases alleging disparate impact. Meacham, 554 U.S. at 101, 128 S.Ct. 2395. Indeed, the first Supreme Court case to recognize the viability of a disparate-impact theory under the ADEA rejected the claim on the merits for “failure to identify the specific practice being challenged,” among other shortcomings. Smith, 544 U.S. at 241, 125 S.Ct. 1536. Requiring plaintiffs to name a specific employment practice “has bite,” Meacham, 554 U.S. at 100, 128 S.Ct. 2395,both on the merits and for purposes of determining whether there is a common question in a disparate impact case. Here, the putative class is challenging a single employment practice: making investigative assignments from the Q–50 List instead of the Q–35 List.

Once a specific practice is identified in a disparate impact case, the next—although not the only—question becomes whether that practice had a disproportionate adverse impact on otherwise eligible officers over forty. “Generally disparate impact analysis is used in a class action, but it may also form the basis of an individual claim.” Bacon v. Honda of Am. Mfg., Inc., 370 F.3d 565, 576 (6th Cir.2004); see also Watson v. Fort Worth Bank & Trust, 487 U.S. 977, 983, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988) (addressing an individual disparate impact claim); Pottenger v. Potlatch Corp., 329 F.3d 740, 749–50 (9th Cir.2003) (same). In whatever procedural guise a disparate impact claim appears, the party asserting it must demonstrate a statistical disparity affecting members of the protected group. Absent such a group-based disparity, the claim fails, whether it is articulated by an individual or a class.

The officers produced a statistical study purportedly showing a disparate impact. The district court, and the City, critiqued that study as inadequate for—among other reasons—failing to conduct a regression analysis to take account of alternative explanations, unrelated to age, for any statistical imbalance. But whatever the failings of the class's statistical analysis, they affect every class member's claims uniformly, just as the materiality issue in Amgen affected every class member uniformly. Each member of the putative class suffered the effects of eliminating the Q–35 List. If those effects amount to a disparate impact on account of age, it will be so for all class members or for none; their claims rise and fall together.

That the City primarily relies on Katz for its contrary argument reveals the fallacy of its position. Katz considered not class certification but an already certified class's appeal of an adverse judgment on the merits. 229 F.3d at 833.

In Katz, a class of laboratory employees claimed the decision to offer certain early retirement incentives to participants in the University of California Retirement Plan, whose average age was 55, and not to participants in the Public Employee Retirement System, whose average age was 60, generated a disparate impact on the basis of age. Id. at 833–34. We affirmed the district court's judgment as a matter of law against the plaintiffs, reasoning they had “failed to demonstrate causation,” insofar as their statistical evidence was “insufficient to raise an inference that the disparate impact fell upon employees by virtue of their membership in a protected age group.” Id. at 836. The failure to produce statistics sufficient to demonstrate a causal relationship between employee age and employer practice defeated the claims of the entire class.

The same will be true here—the statistical showing of disparate impact due to the challenged policy will either succeed as to the class as a whole or, as the City argues, fail—again, with respect to the class as a whole. In highlighting the questions of statistical proof of disparate impact and causation, the City has strengthened, not weakened, the case for certification, as it has identified a common question, the resolution of which will uniformly affect all members of the class. And whether there is a disparate impact on the putative class, to be established through statistical proof, is “a single significant question of ... fact,” Abdullah, 731 F.3d at 957 (emphasis in original) (internal quotation marks omitted), the resolution of which is “central to the validity” of each of the class members's claims, Wal–Mart, 131 S.Ct. at 2551.

The City offers various reasons the putative class, or various members of it, may not prevail, including: the decision not to use the Q–35 List affected all officers on the list equally, without regard to age; all the officers on the Q–35 List could have taken the Q–50 examination, so there was no detrimental impact on them from the policy change; there were no appointments of any Assistant Inspectors from 2007 through 2009, so the Q–35 List was inapplicable during that period; and many of the class members would not have been promoted even if the Q–35 List were used, as there would not have been enough positions for all of them. Any and all of these considerations may prove pertinent to the merits of the case, and possibly to whether common issues predominate under Rule 23(b)(3). But they do not eliminate the significant common question we have identified, however that question is ultimately answered and whether or not other, individual questions are later determined to predominate.

In short, the officers are all challenging a single policy they contend has adversely affected them. The question whether the policy has an impermissible disparate impact on the basis of age necessarily has a single answer. To so recognize is in no way to approve of the statistical showing the officers have made as adequate to make out their merits case. Nor does identifying a common question sufficient for Rule 23(a)(2) purposes entail any disagreement with the City's contention that proper statistical analysis would negate any disparate impact traceable to age—or, for that matter, to prejudge any other argument or defense the City may offer. The defects the City has identified may well exist, but they go to the merits of this case, or to the predominance question, see infra Part IV. Given the interlocutory nature of this appeal, and its consequent limitation to class certification factors only, we may not consider the merits questions, even as an alternative ground for affirmance.

IV.

Having disposed of the officers's certification motion for want of commonality, the district court expressly declined to evaluate the putative class's argument that it satisfied the predominance and superiority requirements of Rule 23(b)(3). The City now contends that “individual questions will predominate with respect to [the officers's] claims including: (a) whether an officer took the 2006 Sergeantsexam or not; (b) comparisons of individual rank on and between the 1998 Assistant Inspector list and the 2007 Sergeant list; (c) the relative qualifications of each officer on the list; and, (d) other factors affecting an assignment to the Investigations Bureau.” The City may or may not be correct. But those questions, as they affect predominance, have been inadequately briefed here. They have no effect on commonality, because there is a single, logically prior, common question: whether the cancellation of the Q–35 List generated a disparate impact on older officers.

The effect of the questions the City raises on the predominance inquiry are best addressed by the district court, which is “ ‘in the best position to consider the most fair and efficient procedure for conducting any given litigation,’ Doninger v. Pac. N.W. Bell, Inc., 564 F.2d 1304, 1309 (9th Cir.1977), and so must be given ‘wide discretion’ to evaluate superiority [under Rule 23(b)(3) ], Lerwill v. Inflight Motion Pictures, Inc., 582 F.2d 507, 512 (9th Cir.1978).” Bateman, 623 F.3d at 712 (citation formatting in original). We thus REMAND to the district court to consider in the first instance whether the putative class satisfies the strictures of Rule 23(b)(3), as well as the other prerequisites for class certification.

V.

We REVERSE the district court's denial of certification for want of commonality. That determination was an abuse of discretion, as it disregarded the existence of common questions of law and fact and impermissibly addressed the merits of the class's claims.

REVERSED and REMANDED.