Lusardi v. Lechner

OPINION OF THE COURT HUTCHINSON, Circuit Judge. Jules Lusardi and the other named plaintiffs (Lusardi) in this age discrimination action seek to appeal an order of the United States District Court for the District of New Jersey revoking that court’s prior conditional certification of an opt-in class. Recognizing that an action granting or denying class certification is interlocutory and generally not appealable, they ask, in the alternative, that we exercise our jurisdiction under the All Writs Act, 28 U.S.C.A. § 1651 (West 1966), and issue a mandamus to the district court directing it to continue the class certification. On the merits, Lu-sardi asserts a number of errors by the district court in revoking certification, none of which we can reach unless we either have appellate jurisdiction or decide that a writ of mandamus is warranted. For the reasons that follow, we hold that we lack appellate jurisdiction. We also hold that the action of the district court, with one exception, does not present us with a clear error of law and the absence of any alternative remedy, both of which conditions must be met before we have discretion to grant mandamus. Therefore, we will dismiss the appeal for lack of jurisdiction. We will, however, grant mandamus for the limited purpose of directing the district court to vacate its holding on the necessity of timely individual administrative filings of charges of age discrimination by the persons it has held do not constitute a proper opt-in class; to reconsider its order decertifying the class for disparate defenses without relying on the presence, or absence, of individual administrative charges or their timing; and, if the court nevertheless concludes class decertification is required, to modify the form of notice it directed be sent to the members of the putative opt-in class by eliminating any implication that they may be barred from pursuing remedies for asserted age discrimination because they failed individually to file timely administrative claims with the Equal Employment Opportunity Commission (EEOC). I. Jules Lusardi and three other former employees filed a class action against the Xerox Corporation (Xerox) on March 8, 1983, pursuant to § 7 of the Age Discrimination in Employment Act (ADEA), 29 U.S.C.A. §§ 621-634 (West 1985 & Supp.1988). The second amended class action complaint, filed October 6, 1983, alleges that Xerox has a nationwide policy and practice of using age as a determinate factor in carrying out salaried workforce reductions.1 — 1065 — Lusardi sought class certification, injunctions restraining Xerox from continuing to implement age-based corporate policies, an order directing it to implement an affirmative action program, reinstatement of plaintiffs and opt-in members of the putative class to positions comparable to those they held with Xerox before the alleged discriminatory pattern affected them individually, backpay, other pecuniary damages and costs. The class was conditionally certified by the district court on January 30, 1984 and included: All salaried employees in the forty (40)-seventy (70) age group who in the period May 1, 1980 through March 31, 1983 have been terminated or required to retire from employment at an age less than seventy or have been denied equal employment opportunities for promotion at any unit, division or American subsidiary of Xerox Corporation and who contend that such termination, retirement or denial of promotion was caused by age discrimination policies or practices of Xerox Corporation. App. at 180. The conditional certification was made without prejudice to the parties’ rights to move for decertification or enlargement or modification of the class or the order. On February 1, 1984, Xerox appealed the district court’s order2 and petitioned for a writ of mandamus, arguing that the district court erred in (1) ordering Xerox to provide to the court the names and addresses of all potential class members and authorizing Lusardi to send notice to such potential class members and (2) conditionally certifying an across-the-board opt-in class. This court denied Xerox’s petition for mandamus by order dated February 9,1984 and in a published opinion dismissed Xerox’s appeal for lack of jurisdiction. See Lusardi v. Xerox Corp., 747 F.2d 174 (3d Cir.1984) (hereinafter Lusardi I). After concluding that the district court’s order was appealable only if it fell within the “collateral order” exception of Cohen v. Beneficial Indus. Loan Corp., 337 U.S. 541, 69 S.Ct. 1221, 93 L.Ed. 1528 (1949), we analyzed each issue separately in light of Cohen’s three requirements that: (1) the order conclusively determine the disputed question, (2) it resolve an important issue completely separate from the merits of the action, and (3) it be effectively unreviewable on appeal from final judgment. Lusardi I, 747 F.2d at 176 (quoting Coopers & Lybrand v. Livesay, 437 U.S. 463, 98 S.Ct. 2454, 57 L.Ed.2d 351 (1978)). Xerox’s argument regarding conditional class certification failed to meet the first prong of Cohen because the district court’s order was subject to revision. Id. at 177. The district court’s order requiring Xerox to provide a mailing list of potential class plaintiffs and authorizing plaintiffs to send them notice was not appealable because it was “essentially a discovery device” and therefore interlocutory. Id. at 178.3 The parties agreed to randomly select a fifty-one class member “sample group” out of the thirteen hundred persons who responded to the initial notice and opted in. After making the selection, they entered into intensive discovery. It focused on — 1066 — both the sample group and the thirteen plaintiffs named in the second amended complaint. Lusardi’s motion to file a third amended complaint to add a fourteenth named plaintiff and thereby expand the class period to include April 1, 1983 through May 1, 1984, as well as to add class-wide allegations of state contract and age discrimination claims, was denied by Magistrate Ronald Hedges on June 2, 1986 and by the district court on June 17, 1986. Upon the completion of discovery, Xerox moved to decertify the conditional class. Lusardi filed a cross-motion for summary judgment or, in the alternative, for a presumption of class-wide liability. Lusardi also moved to strike Xerox’s expert report. On December 16, 1987, District Judge Alfred J. Lechner, Jr. granted Xerox’s motion to decertify the class and denied Lu-sardi’s motions as moot. In a seventy-five page opinion,4 the district court assessed the class in light of the “similarly situated” requirement of § 16(b) of the Fair Labor Standards Act (FLSA), codified at 29 U.S. C.A. § 216(b) (West Supp.1988) (as amended by the Portal-to-Portal Act of 1947, 29 U.S. C.A. §§ 251-262 (West 1985)).5 Lusardi v. Xerox Corp., 118 F.R.D. 351 (D.N.J.1987) (hereinafter Lusardi II). The district court first found that disparate employment situations6 among the class plaintiffs were “significant not only to plaintiffs' claims but as well to the ability of Xerox to defend against the claims.” Id. at 361. In examining Xerox’s defenses to several of the sample group members’ claims, the court concluded that “the proposed class plaintiffs are not similarly situated and that considerations of fairness, as well as, efficiency are defeated by maintenance of a class action in this case.” Id. at 363. The court reasoned: To proceed without permitting Xerox to raise at the liability stage of trial each and every defense available to it where each potential class member is readily identifiable and must step forward in order to assert and prove an individual claim for liability or at least be the subject of a defense particular to each such — 1067 — plaintiff would deprive defendant of the Fifth Amendment right to due process. App. at 667 (footnote omitted). The court interpreted the FLSA § 16(b)’s “similarly situated” requirement to “presuppose” a common defense, giving the defendant an opportunity to effectively defend against the class allegations. It distinguished International Bhd. of Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977), by explaining that in that case the union did not assert, or seek to assert, individual defenses against each class plaintiff. The district court also cited our decisions in Dillon v. Coles, 746 F.2d 998 (3d Cir.1984) and Gavalik v. Continental Can Co., 812 F.2d 834 (3d Cir.), cert. denied, — U.S. -, 108 S.Ct. 495, 98 L.Ed.2d 492 (1987) for the proposition that the liability phase of a class action is not concluded until the individual class members have demonstrated their bases for awards. Thus, the court reasoned, the defendant may assert each of its defenses at the liability stage of the trial. Because an ADEA class is formed by “opting-in,” thereby requiring each individual class member to demonstrate liability, “each individual class member must come forward in order to prove causation prior to any finding of liability to that class member and ... in each instance Xerox has the possibility of raising any one of [its] several defenses,” the district court concluded that “plaintiffs cannot be deemed similarly situated.” Lusardi II, 118 F.R.D. at 375. For all of these reasons, the district court determined that the opt-in plaintiffs were not “similarly situated” and ordered the class decertified. The district court also examined the ADEA’s filing requirements, interpreting § 626(d)(2) to require that each individual class plaintiff file an age discrimination charge with the EEOC before joining in a class action. The district court based its reading both on the plain language of the statute and on case law which refused to allow “piggybacking” when the EEOC filings did not allege classwide discrimination. The court read each of the four originally named plaintiffs’ EEOC charges as identical, with the exception of each plaintiff’s personal information. Those EEOC charges alleged a Xerox policy to eliminate persons over forty years old who were too old to be trained in other positions, and an employment practice discriminating against persons over forty as a class. The district court stated that these “references in the filings ... do not suggest the charge of discrimination concerned any person other than the individual named in the filing.” Lusardi II, 118 F.R.D. at 378. It therefore concluded that Lusardi and the other named plaintiffs did not file on behalf of “others similarly situated.” Lusardi appeals the district court’s order decertifying the class and petitions for a writ of mandamus.7 He argues that the order is appealable under the collateral order exception of Cohen. On the merits, Lusardi argues, inter alia, that the district court erred in finding that the opt-in plaintiffs were not “similarly situated” within § 16(b) of the FLSA and in interpreting the ADEA as requiring a timely EEOC filing by each of the potential plaintiffs in order to “opt-in” to the class action. Because we find that we lack appellate jurisdiction and that mandamus is not warranted except with respect to the requirement that each class plaintiff must have filed a timely administrative claim, we do not reach the merits of Lusardi’s other arguments. II. Appealability We will first address the appealability of the district court’s order decertifying the class. “A class action determination, affirmative or negative, is not in this circuit a final order appealable under 28 U.S.C. § 1291.” Katz v. Carte Blanche Corp., 496 F.2d 747, 752 (3d Cir.), cert. denied, 419 U.S. 885, 95 S.Ct. 152, 42 L.Ed.2d 125 — 1068 — (1974);8 see also DeMasi v. Weiss, 669 F.2d 114, 119 (3d Cir.1982). The district court did not certify its order as final for purposes of appeal pursuant to Federal Rule of Civil Procedure 54(b) or as a controlling question of law pursuant to 28 U.S.C.A. § 1292(b) (West Supp.1988). See Saber v. FinanceAmerica Credit Corp., 843 F.2d 697, 701-02 (3d Cir.1988); Katz, 496 F.2d at 752-56. In its order, the district court denied a stay of that order pending appeal.9 Until recently, the district court’s order might have been thought appealable under the Enelow-Ettelson doctrine. See Ettelson v. Metropolitan Life Ins. Co., 317 U.S. 188, 63 S.Ct. 163, 87 L.Ed. 176 (1942); Enelow v. New York Life Ins. Co., 293 U.S. 379, 55 S.Ct. 310, 79 L.Ed. 440 (1935). That avenue of appeal, however, was recently foreclosed by the Supreme Court’s decision in Gulfstream Aerospace Corp. v. Mayacamas Corp., — U.S. -, 108 S.Ct. 1133, 99 L.Ed.2d 296 (1988), overruling the Enelow-Ettelson doctrine. See also Delta Traffic Serv., Inc. v. Occidental Chem. Corp., 846 F.2d 911 (3d Cir.1988). Lusardi’s only other available basis for appellate jurisdiction over the district court’s decertification order is the “collateral order” doctrine of Cohen, set out supra. However, the Supreme Court has held that class action determinations do not fall within the Cohen exception. Coopers & Lybrand v. Livesay, 437 U.S. 463, 98 S.Ct. 2454, 57 L.Ed.2d 351 (1978). In that case, plaintiffs, securities purchasers, brought a Rule 23 class action against an accounting firm for misstatements in a prospectus in violation of the Securities Act of 1933 and the Securities Exchange Act of 1934. The district court first certified and later decertified the class. The Court of Appeals held it had jurisdiction and reversed the order decertifying the class. The Supreme Court granted certiorari and held the order not appealable under 28 U.S.C. § 1291. After setting out the three prongs of Cohen’s “collateral order” exception to § 1291, the Court held that the district court’s order did not fall within that exception: First, such an order is subject to revision in the District Court. Second, the class determination generally involves considerations that are “enmeshed in the factual and legal issues comprising the plaintiff’s cause of action.” Finally, an order denying class certification is subject to effective review after final judgment at the behest of the named plaintiff or intervening class members. For these reasons, ... the collateral-order doctrine is not applicable to the kind of order involved in this case. Coopers & Lybrand, 437 U.S. at 469, 98 S.Ct. at 2458 (footnotes and citations omitted). We dismissed Xerox’s earlier appeal of the district court’s order conditionally certifying the class for similar reasons. See Lusardi I, 747 F.2d at 177-78. Plaintiffs argue that Lusardi I is distinguishable because it dealt with the conditional certification of a class rather than a decertification, which plaintiffs term a “final order.” A decertification order is reviewable upon appeal from a final judgment. Samuel v. University of Pittsburgh, 506 F.2d 355, 360 (3d Cir.1974). Therefore, such an or — 1069 — der, whether based on a Rule 23 or an FLSA § 16(b) class action, fails to fall within the Cohen “collateral order” exception. The wisdom of denying appellate review of class determinations is well illustrated by this case. In opt-in class actions, with their similarities to bills of peace, permissive joinders and intervention and their accompanying tolling problems, it is of the utmost importance that the decision on class status be made promptly so that class members know whether to accept the “invitation to come in” or look out for themselves. See Z. Chafee, Some Problems of Equity, 259-61 (1950) (hereinafter Chafee). Lusardi filed his first complaint on March 8, 1983. On January 30, 1984 the district court decided to conditionally certify a class. There was an appeal and accompanying petition for mandamus. The case was transferred to a different district judge. Discovery dragged on. On December 16, 1987, the district judge decertified the class. As noted, we stayed that order. The case will now be remanded with the class status to be redetermined. III. Mandamus In their petition for mandamus, Lusardi and the other individual plaintiffs assert five reasons why the writ should be granted. Before considering those reasons separately, we believe it is useful to set forth generally the principles applied in this jurisdiction in determining whether this extraordinary writ should issue, as well as those relating to the certification and maintenance of actions under FLSA § 16(b). A. Authority to issue mandamus is granted by 28 U.S.C.A. § 1651. It provides: (a) The Supreme Court and all courts established by Act of Congress may issue all writs necessary or appropriate in aid of their respective jurisdictions and agreeable to the usages and principles of law. 28 U.S.C.A. § 1651(a) (West 1966). The writ is an extraordinary use of power by an appellate court. Its use is contrary to the common law policy of avoiding piecemeal appellate review of cases pending in courts of original jurisdiction. The policy is of great importance in preserving judicial efficiency and orderly procedure. Cases which go through a trial court by fits and starts are not likely to have a reasonably prompt disposition. Halting their flow through the judicial system while awaiting appellate consideration of interim orders or decrees is, at best, disruptive. Also, use of the writ is likely to require resolution of issues which disappear or shrink to insignificance in the context of a final judgment. As stated, this case is a good example of the problems its use or solicitation creates. Modern appellate practice includes various devices unknown to the common law when the writ was developed. They both serve to oil the somewhat arthritic joints of the strict common law rule of finality and to decrease the need for parties to rely exclusively upon the writ for relief from a court order. The lubricants include the certified interlocutory appeal, 28 U.S.C.A. § 1292(b), and the judicially developed collateral order exception of Cohen. Accordingly, the writ is seldom issued and its use is discouraged. Nevertheless, when no other avenue is open, i.e., when a party seeking the writ has no other adequate means to attain the relief he desires and the court below has committed a clear error of law, it may issue. Cipollone v. Liggett Group, Inc., 785 F.2d 1108, 1118 (3d Cir.1986) (quoting Sporck v. Peil, 759 F.2d 312, 314 (3d Cir.1985)), cert. denied, — U.S. -, 108 S.Ct. 487, 98 L.Ed.2d 485 (1987). The clear error should at least approach the magnitude of an unauthorized exercise of judicial power, or a failure to use that power when there is a duty to do so. Will v. Calvert Fire Ins. Co., 437 U.S. 655, 661, 98 S.Ct. 2552, 2556, 57 L.Ed.2d 504 (1978); Citibank, N.A. v. Fullam, 580 F.2d 82, 86 (3d Cir.1978). These requirements are in accord with § 1651(a). When they are met, issuance of mandamus may be necessary to preserve appellate jurisdiction. Otherwise, our abili — 1070 — ty to review an issue likely to control the availability of the relief sought would be lost. It is also “agreeable to the usages and principles of law:” “[Traditional use of the writ in aid of appellate jurisdiction both at common law and in the federal courts has been to confine an inferior court to a lawful exercise of its prescribed jurisdiction or to compel it to exercise its authority when it is its duty to do so.” Calvert Fire Ins. Co., 437 U.S. at 661, 98 S.Ct. at 2556 (quoting Roche v. Evaporated Milk Ass’n, 319 U.S. 21, 26, 63 S.Ct. 938, 941, 87 L.Ed. 1185 (1943)). Still, the presence of power does not compel its exercise. Even in the presence of a clear error of law which would otherwise escape review and a showing by a party that his right to this relief is “clear and indisputable,” issuance of a writ is within our discretion. Kerr v. United States District Court, 426 U.S. 394, 403, 96 S.Ct. 2119, 2124, 48 L.Ed.2d 725 (1976). In using that discretion, judges should proceed both carefully and courageously. B. Special problems are present in determining whether mandamus is warranted in the context of the kind of class action before us. The availability of the class action device to Lusardi is not controlled by Federal Rule of Civil Procedure 23, but by § 16(b) of the FLSA, incorporated into the ADEA by § 626(b) of that statute. The current version of FLSA § 16(b) reads, in relevant part: (b) ... An action to recover the liability prescribed in [this subsection] may be maintained against any employer (including a public agency) in any Federal or State court of competent jurisdiction by any one or more employees for and in behalf of himself or themselves and other employees similarly situated. No employee shall be a party plaintiff to any such action unless he gives his consent in writing to become such a party and such consent is filed in the court in which such action is brought. 29 U.S.C.A. § 216(b) (West 1988). On June 25, 1938, the date of § 16(b)’s enactment, the Federal Rules of Civil Procedure had been proposed and were pending before Congress, but had not yet become effective. One problem bedeviling the common law of class actions in this country in 1938 was the binding effect of the class action decree on absent class members. Chafee, supra, at 224-25. In the United States the application of res judicata to absent parties in class suits raised constitutional issues of due process. Compare Supreme Tribe of Ben Hur v. Cauble, 255 U.S. 356, 41 S.Ct. 338, 65 L.Ed. 673 (1921) (holding all members of beneficial society bound by state decree in class action, including citizens of the state in which federal action seeking to enforce state decree was brought, who were not parties to the state suit) and Restatement of Judgments § 86 (1942) (all members of class bound) with Hansberry v. Lee, 311 U.S. 32, 61 S.Ct. 115, 85 L.Ed. 22 (1940) (reversing Illinois Supreme Court order that decree in class action enjoining violation of racially restrictive covenants bound absent class members).10 Two efforts were made in 1938 to deal with the due process problem in applying res judicata to class actions. Those efforts formalized the opt-in class. First, Congress enacted § 16(b) for the special purpose of facilitating enforcement of the FLSA. Second, and more generally, the Advisory Committee on the Federal Rules provided for notice to absent class members and included in old Rule 23, 23(a)(3)’s “spurious class.” Under both the statute and the rule, class actions in which absent members were not bound by the judgment were recognized. See Pentland v. Dravo Corp., 152 F.2d 851, 853 (3d Cir.1945). Difficulties in assigning a particular class action to one of the three categories of old Rule 23(a) led to the abandonment of the spurious class in the 1966 amendment to the rule, but the . opt-in class remains with — 1071 — us under § 16(b). See Fed.R.Civ.P. 23 advisory committee’s note (1966 amendment). Because § 16(b)’s coverage has been extended to age and sex discrimination laws to facilitate their enforcement, the class action in which only members who agree to participate are bound remains an important tool for their enforcement in cases of systemic discrimination. Used properly, it reduces multiplicity of suits and offers a convenient means of settling issues common to a large number of persons whose interest is sufficiently similar. To accomplish the Congressional purpose of encouraging private enforcement of these statutes by making it convenient and inexpensive for the persons they are meant to protect, the members of the class who opt-in receive the benefit of final judgment despite the action’s non-binding effect on absent members. Considering the historical development of opt-in class actions, and the analogies drawn to bills of peace, permissive joinder and intervention, the question arises whether each member of the opt-in class must individually satisfy all timeliness requirements. The resolution of that issue depends upon the intent of Congress in enacting § 16(b) and the incorporation of that section into the ADEA. Those issues are discussed infra section IY(D). IV. With these general principles and background in mind, we examine Lusardi’s arguments for granting mandamus. A. Lusardi first argues that the district court’s December 16, 1987 order violates the doctrine of law of the case because it conflicts with the January 31, 1984 order of another judge of that court conditionally certifying the class. In support, he cites Hayman Cash Register Co. v. Sarokin, 669 F.2d 162 (3d Cir.1982), for the proposition that mandamus lies under 28 U.S.C.A. § 1651 when one district judge re-examines and reverses an order by another directing a transfer of a case. In Hayman, this court quoted with approval a statement by Justice Frankfurter that such conduct promotes “judicial unseemliness.” 669 F.2d at 166-68 (quoting Hoffman v. Blaski, 363 U.S. 335, 346-49, 80 S.Ct. 1084, 1091-93, 4 L.Ed.2d 1254 (1960) (Frankfurter, J. dissenting)). An examination of the prior order in this case disposes of that argument. It states: 7. Conditional certification of this action as a class action is without prejudice to the respective parties’ rights to move for decertification of the action as a class action or for enlargement or modification of the class definition or for modification of this Order. App. at 182. At a hearing on December 22, 1983 on the question of class certification, the district judge who entered the order of January 30, 1984 asked: “Can I do this conditionally?” Counsel for Lusardi responded: “Yes, your Honor. I expect this to be conditionally. Totally.” App. at 105. Thereafter, the following colloquy took place: THE COURT: And then we’ll — you know, what that means really is conditional. If he can’t establish a broad base centralized policy, then I’m probably not going to go ahead with the case in this form. I’m not interested in creating a monster that no one can deal with, made up with a lot of individual people with specific grievances. MR. FREUND: That’s right, your Honor. THE COURT: No one is going to kill it any faster than I am. I mean in terms of not sticking — I’m not going to stick a thousand or 5,000 or 10,000 eases together that shouldn’t be. I wouldn’t be able to handle it. If there wasn’t a broad-base policy here, then it is hard to believe this thing should go forward. But it is that very allegation which interests me in continuing this investigation. App. at 108-09. We dismissed Xerox’s appeal from this order conditionally certifying the class for lack of jurisdiction. Lusardi I, 747 F.2d at 178. — 1072 — Lusardi’s law of the case argument poses a conflict between the rationale of the order conditionally certifying the class and the rationale of the order decertifying it. Arguably, the earlier order made decertifi-cation dependent only on Lusardi’s failure to produce evidence of a company-wide pattern or practice of age discrimination. If so, once evidence of that pattern was produced, Xerox’s evidence of, or reliance on, disparate individual defenses would be irrelevant on the issue of whether the persons electing to opt-into the litigation were similarly situated. Accordingly, Lusardi contends that the order decertifying the class because of disparate defenses, as evidenced by the opinion in support of decerti-fication, conflicts with the order conditionally certifying the class. The order conditionally certifying the class expressly contemplates the possibility of later decertification. When the recognized possibility of decertification is made a reality by the later order, there is no clear conflict or unwarranted re-examination. A writ of mandamus based on conflicting action by two coordinate judges should issue only on a plain conflict in their orders. Appeals do not lie and writs do not issue to reconcile ratio decidendi. They are used to vacate, reverse, affirm or modify orders. The asserted conflict between the order of conditional certification and the order of decertification does not warrant mandamus.11 B. Lusardi’s second contention is that “[t]he district court committed a clear abuse of discretion by requiring direct proof of discriminatory intent by Xerox in order for plaintiffs Lusardi to establish a prima facie case of a pattern and practice of age discrimination in this FLSA section 16(b) class action.”12 Plaintiffs’ Petition for Mandamus at 37. As he develops this argument, Lusardi shifts to a contention that statistical proof of the disparate impact of a reduction in force creates a pri-ma facie case entitling each member of the protected class who opts in to jury consideration of his discrimination claim.13 This — 1073 — presence of statistical proof, he contends, is a common issue of fact that makes all the opt-in claimants similarly situated within the meaning of § 16(b) of FLSA. See Distelhorst v. Day & Zimmerman, 58 F.Supp. 334, 335 (S.D.Iowa 1944) (holding any common issue of fact meets the similarly situated test of FLSA § 16(b)), appeal dismissed, 150 F.2d 541 (8th Cir.1945). In making that argument, Lusardi relies on United States Postal Serv. Bd. of Governors v. Aikens, 460 U.S. 711, 103 S.Ct. 1478, 75 L.Ed.2d 403 (1983), as well as our recent in banc decision in Chipollini v. Spencer Gifts, Inc., 814 F.2d 893, (3d Cir.), cert. dismissed, — U.S. -, 108 S.Ct. 26, 97 L.Ed.2d 815 (1987). Those cases stand for the proposition that direct evidence of discriminatory intent is not necessary to establish a prima facie case or prove unlawful discrimination.14 We will assume, ar-guendo, that statistical evidence may create a prima facie case of discrimination. See Teamsters, 431 U.S. at 339, 97 S.Ct. at 1856; EEOC v. American Nat’l Bank, 652 F.2d 1176, 1188 (4th Cir.1981), cert. denied, 459 U.S. 923, 103 S.Ct. 235, 74 L.Ed.2d 186 (1982). However, the law is not clear, in this circuit or elsewhere, that the presence of statistical evidence of a disparate impact or a pattern of discrimination deprives a district court of its discretion to decide whether to certify, or continue to certify, an opt-in class under FLSA § 16(b). To say a district court has no discretion to refuse or revoke class certification in a § 16(b) systemic discrimination case whenever there is statistical evidence of disparate impact, for example, is equivalent to saying that all individuals in a protected class are “similarly situated” under FLSA § 16(b) when statistical evidence indicates a personnel policy has a disparate impact on the protected class as a whole. We do not think that legal proposition is as yet established, and is certainly not yet established as a matter of clear law. If the district court does, in fact, require Lusardi to produce both direct proof of discrimination and appropriate statistical evidence of a disparate impact or a practice of discrimination in order to get his claim to a jury, that issue will be subject to review on appeal from a final order. We then can decide whether the conflicting statistical evidence required the district court to maintain the case as an opt-in class action simply because of the common issue which exists in all systemic discrimination cases. If those questions are properly presented on appeal from a final order and we determine that class certification is required, all opt-in members who have not chosen to proceed individually will be fully protected by reinstatement óf the action as a class action. The case would then necessarily be remanded to the district court for continuation as a class action pursuant to the provisions of FLSA § 16(b). The reinstatement of the class will at least relate back to their acceptance of what Chafee calls the “invitation to come in,” that is, to when they opted in. See Gelman v. Westinghouse Elec. Corp., 556 F.2d 699, 701 (3d Cir.1977). Accordingly, the remedy of appeal is an adequate means of relief with respect to these issues and mandamus does not lie on this ground. See Cipollone v. Liggett Group, Inc., 785 F.2d 1108, 1118 (3d Cir.1986), cert. denied, — U.S. -, 108 S.Ct. 487, 98 L.Ed.2d 485 (1987). C. Lusardi’s key attack on the decertification order is his contention that the district court “committed a clear abuse of discretion” in holding that Xerox’s assertion of disparate defenses to individual claims precludes an opt-in class action under FLSA § 16(b). EEOC, in its amicus brief, joins — 1074 — Lusardi in this contention. Translated into Cipollone terms, Lusardi and EEOC argue that it is clear legal error for a district court to consider disparate individual defenses in deciding whether an FLSA § 16(b) action can be maintained as a class action. If they are correct, a question on which we express no opinion, it is not necessary to resort to the extraordinary writ of mandamus in order to preserve our “potential appellate jurisdiction.” Calvert Fire Ins. Co., 437 U.S. at 661-62, 98 S.Ct. at 2556-57. The question of whether the statistical issue common to all persons claiming systemic discrimination makes them all “similarly situated” under FLSA § 16(b) as a matter of law, and so requires class certification, will be reviewable on appeal from a final order on Lusardi’s individual claim. Coopers & Lybrand, 439 U.S. at 469, 98 S.Ct. at 2458. If the presence of that common issue of statistical fact does not make all the opt-ins similarly situated as a matter of law, the issue of whether the district court abused its discretion in denying certification on the facts of this case can be addressed in an appeal on behalf of the class. In the face of Lusardi’s assertion and the district court’s finding that he and the other named plaintiffs are adequate class representatives, we cannot assume they will abandon the cause of the class simply because they win individual relief. To issue mandamus because a trial judge has denied class certification based on the assertion of individualized defenses by an employer would remove the judge’s discretion over manageability of the litigation. Cf. In re School Asbestos Litigation, 789 F.2d 996, 1011 (3d Cir.) (“The potential for individualized defenses does not detract from the commonality of the questions as viewed from the standpoint of the class members, but the problem clearly poses significant case management concerns.”), cert. denied, — U.S. -, 107 S.Ct. 182, 93 L.Ed.2d 117, cert. denied, — U.S. -, 107 S.Ct. 318, 93 L.Ed.2d 291 (1986). It would also violate this circuit’s rule that the writ is not available to review a class determination, affirmative or negative, which the district court has not certified for interlocutory appeal under Rule 54(b), unless the district court acts “outside its jurisdiction” or “in disregard of appropriate procedural safeguards.” Katz, 496 F.2d at 752.15 Just as the drafters of Rule 23 in both its present and former versions were familiar with the common law of class actions from which the principles set forth in Rule 23 are derived, we must assume that Congress was also aware of the common law background in 1938 when it provided that class actions for FLSA violations could be maintained on behalf of all persons “similarly situated” who consent to join the class. While it would be wrong to freeze a device as important to the vindication of civil rights as this kind of class action into the mold of ancient law, we are nevertheless not prepared to say that the law has developed to a point where a district court is required to grant class certification to persons opting in under FLSA § 16(b) in a systemic discrimination case without regard to the management difficulties created by numerous discrete defenses.16 Whether a class action is inappropriate in this particular case because of the disparate individual defenses, whether that problem can be solved by the creation of subclasses and, if so, the extent to which an outcome favorable to the named plaintiffs — 1075 — on the discrimination issue inures to the benefit of the members of the putative class, are issues that can be brought before us on appeal from a final order. In the first instance, they are entrusted to the district court’s sound discretion. On an appeal from a final order asserting an abuse of that discretion, we can consider the threshold question of whether FLSA § 16(b) requires class treatment in a systemic discrimination case despite the necessity, in order to give Xerox an opportunity to present its defenses, of a second tier division of the case into individual actions for relief. If class treatment is not mandatory as a matter of law, we can also decide whether the district court abused its discretion in decertifying the opt-in class.17 D. Finally, Lusardi argues that the district court committed a clear abuse of discretion in ruling that the decertified opt-in claimants would be barred from filing individual ADEA suits if they had not each filed timely administrative charges with EEOC. The problem is compounded by the inclusion of that ruling in the notice the court ordered the parties to send to the opt-in claimants. The order decertifying the class reads: 3. Within fourteen (14) days of the entry of this order on the docket but no later than January 1, 1988, counsel for the named plaintiffs shall mail, at their expense, by first class mail, a copy of the attached Notice of Decertification to Conditional Class Members to all persons who filed a written consent to join this suit as party plaintiffs; the Honorable William T. Walsh, Clerk of the court is hereby directed to sign the Notice of Decertification to Conditional Class Members in the form attached. App. at 750. Paragraph 3 of the Notice itself reads: 3. DECISION OF THE COURT: In an opinion dated November 5,1987, United States District Judge Alfred J. Lech-ner, Jr. determined this case should be decertified as an ADEA class action. This means that your consent to be a party plaintiff has been revoked so that you are no longer a party to this action. If you did not timely file an administrative charge of age discrimination with the United States Equal Employment Opportunity Commission (“EEOC”) or an appropriate state agency that handles discrimination complaints, your claim could be barred under statutes of limitations provided for in ADEA. Id. at 753-54 (emphasis supplied). Paragraph 4 of that Notice reads, in part: 4. EFFECT OF COURT’S DECISION: You may still have the right to pursue in federal or state court or before an appropriate state agency your individual claim of age discrimination against Xerox as well as any other legal claim you may have. Statutes of limitations made part of [the] ADEA limit the time during which claims may be made by the complaining party against an employer. The statutes of limitations applicable to your federal age discrimination claim against Xerox have been tolled, or stayed, from the time you filed your consent to join this lawsuit. As a result of the court’s decision to decertify this case as a class action, that stay will expire as of midnight January 12, 1988. Thereafter, the period of time in which you must assert a claim of age discrimination against Xerox will begin to run again. Id. at 754. The notice is, at the least, confusing in its mingling of the ADEA statute of limitations with its requirement that a charge of age discrimination be filed with the EEOC as a prerequisite to suit. It thus raises questions of procedural fairness. The problem is not cured by the concluding sentence of the notice, which reads: — 1076 — If you wish to assert an individual claim of age discrimination against Xerox, as well as any other claim, you should seek legal advice IMMEDIATELY. Id. The decertification order can be read to require each member of the decertified opt-in class to file a discrimination charge with the EEOC within the time limits of 29 U.S.C.A. § 626(d) (West 1985). This is made clear to the opt-ins by the use of the past tense “did” in paragraph 3 of the notice itself. It is also apparent from the notice’s provisions with respect to equitable tolling. The order and notice provide for equitable tolling of the statute of limitations. They make no similar provision with respect to the time limits the statute sets out for filing an administrative charge of discrimination. The problem is aggravated on this record by the notice’s implied limitation of the available administrative forums. The record shows that some of the opt-ins may have chosen to file administrative claims with state agencies, an alternate procedure allowed by 29 U.S.C.A. § 633(b).18 This notice is not only confusing but, in stating that an individual claim could be barred if the individual did not file a timely administrative charge, also plainly wrong as a matter of law. In its rationale that a class action was inappropriate given Xerox’s disparate defenses to individual class member claims, the district court at least partially relied on when or whether individual class members filed administrative claims. In this respect, it held: The ADEA requires that all plaintiffs file timely charges of discrimination with the ... (“EEOC”) before commencing an action_ Accordingly, ... all individuals in this matter who did not file [administrative] charges ... within, at most, 300 days of the occurrence of their termination (the alleged discriminatory practice) are time barred by the ADEA from proceeding with this complaint. Lusardi II, 118 F.R.D. at 376-77. This holding, and the inclusion of this factor in the district court’s conclusion that the class plaintiffs are not “similarly situated,” is a clear error of law.19 In Albemarle Paper Co. v. Moody, 422 U.S. 405, 95 S.Ct. 2362, 45 L.Ed.2d 280 (1975), Justice Stewart, writing for the Court, said: The petitioners also contend that no backpay can be awarded to those unnamed parties in the plaintiff class who have not themselves filed charges with the EEOC. We reject this contention. The Courts of Appeals that have confronted the issue are unanimous in recognizing that backpay may be awarded on a class basis under Title VII without exhaustion of administrative procedures by the unnamed class members. The Congress plainly ratified this construction of the Act in the course of enacting the Equal Employment Opportunity Act of 1972, Pub.L. 92-261, 86 Stat. 103. The House of Representatives passed a bill, H.R. 1746, 92d Cong., 1st Sess., that — 1077 — would have barred, in § 3(e), an award of backpay to any individual who “neither filed a charge [with the EEOC] nor was named in a charge or amendment thereto.” But the Senate Committee on Labor and Public Welfare recommended, instead, the re-enactment of the backpay provision without such a limitation, and cited with approval several cases holding that backpay was awardable to class members who had not personally filed, nor been named in, charges to the EEOC, S.Rep. No. 92-415, p. 27 (1971). See also 118 Cong.Rec. 4942 (1972). The Senate passed a bill without the House’s limitation, id., at 4944, and the Conference Committee adopted the Senate position. A Section-by-Section Analysis of the Conference Committee’s resolution notes that “[a] provision limiting class actions was contained in the House bill and specifically rejected by the Conference Committee,” id., at 7168, 7565. The Conference Committee bill was accepted by both Chambers. Id., at 7170, 7573. 422 U.S. at 414 n. 8, 95 S.Ct. at 2370 n. 8 (case citations omitted). This rule was reaffirmed in an ADEA ease, Oscar Mayer & Co. v. Evans, 441 U.S. 750, 99 S.Ct. 2066, 60 L.Ed.2d 609 (1979). In ruling that ADEA claimants must first resort to administrative remedies in deferral states, pursuant to § 633(b), the Court cautioned that “[t]his rule, of course, governs only claims for individual relief, such as the present case. Nothing in our decision in anywise disturbs the rule of Albemarle ... concerning the rights of unnamed parties in plaintiff class actions.” 441 U.S. at 758 n. 6, 99 S.Ct. at 2073 n. 6. In its amicus brief, the EEOC supports Lusardi’s position that individual filings are not a necessary prerequisite to “opting into” a class action, so long as a representative complaint was filed with the EEOC within the 180/300 day time limit provided for in 29 U.S.C.A. § 626(d). It contends that a representative complaint meets the purpose of the EEOC filing, which is to provide notice to the employer and an opportunity for conciliation. See H.R.Conf.Rep. No. 950, 95th Cong., 2d Sess. 12, reprinted in 1978 U.S.Code Cong. & Admin.News, 504, 534. If there were any doubt about the continuing vitality of Justice Stewart’s rejection of the position taken by the district court in this case, it was settled by Congress’s 1978 amendment of § 7(d) of the ADEA. At the time Albemarle was decided, that section read:, “no civil action may be commenced by an individual under ... this section until the individual has given the Secretary ... notice_” Pub.L. No. 90-202, 81 Stat. 602, 604 (1967). On April 6, 1978, Congress amended it to read that “[n]o civil action may be commenced by an individual under this section until 60 days after a charge alleging unlawful discrimination has been filed with the Secretary.” Pub.L. No. 95-256, 92 Stat. 189, 190.20 See also Spahn, supra note 1, at 133-39; Ami-cus Brief of EEOC at 14-15. Oscar Mayer, supra, was decided subsequent to the amendment. The majority of courts that have addressed the issue since the 1978 amendment have held that individuals can rely on the named plaintiffs’ timely charge if it gives notice of class discrimination. See Kloos v. Carter-Day Co., 799 F.2d 397, 400 (8th Cir.1986) (charge must allege class-wide age discrimination or claim to represent class in order to serve as basis for ADEA class action under § 216(b)); Naton v. Bank of California, 649 F.2d 691, 697 (9th Cir.1981) (notice inadequate because complainant expressed no intention to sue on behalf of anyone other than himself); Mistretta v. Sandia Corp., 639 F.2d 588, 593-94 (10th Cir.1980) (charges alleging employer’s “arbitrary action constitutes age discrimination against workers over 40” and notice to agencies that suit was intended to be class action gave agencies opportunity to investigate and act within statute of limitations); Bean v. Crocker Nat’l Bank, 600 F.2d 754, 760 (9th Cir.1979) (notice filed on behalf of named — 1078 — plaintiffs and others similarly situated put Secretary and employer on notice that “discrimination charges encompassed a pattern of unlawful conduct transcending an isolated individual claim and that they should act accordingly”); see also Anderson v. Montgomery Ward & Co., 631 F.Supp. 1546, 1549 (N.D.Ill.1986) (allowing “piggybacking” of opt-in ADEA claimants although EEOC charge was not expressly representative). But see McCorstin v. U.S. Steel Corp., 621 F.2d 749, 755 (5th Cir.1980) (refusing to decide issue of whether notice embracing claims of similarly situated employees satisfied § 626(d) since district court’s finding that notice did not purport to include others similarly situateed was not clearly erroneous); Price v. Maryland Casualty Co., 561 F.2d 609, 610-11 (5th Cir.1977) (affirming district court’s order that “opt-in” procedure be used for FLSA § 16(b) class actions, thereby implicitly requiring both filing of notice of intent to sue and written consent from opt-in plaintiffs to be included in class). The district court found that it “presently appears” none of the four originally named plaintiffs’ EEOC charges were filed on behalf of “others similarly situated.” App. at 688. Lusardi’s EEOC charge alleged: Xerox has engaged and is continuing to engage in employment practices related to hiring, training, promotion and termination of past, present and future employees which discriminate against persons such as myself over forty as a class particularly with respect to the implementation of a reduction in salaried work force. Lusardi II, 118 F.R.D. at 378. Although the EEOC charge does not use the words “similarly situated” or allege specifically that a class action is going to be brought, we fail to see how Xerox could claim prejudice by the bringing of a class action. The charge clearly notifies Xerox that it allegedly discriminates against persons over forty years old as a class. Accordingly, the charge provides sufficient notice to the parties to encourage meaningful conciliation, the purpose of requiring it. See Bean, 600 F.2d at 760. So long as class issues are alleged, a timely charge may serve as the basis for a class action. This is the position recently taken by the district court in Nowicki v. USX Corp., 672 F.Supp. 854 (W.D.Pa.1987). It is also the position of the Eighth, Ninth and Tenth Circuits. See cases cited supra. We believe the reasoning of those courts is sound. We recognize that in other contexts the opt-in class action has been, analogized to permissive joinder and intervention. See, e.g., Zahn v. International Paper Co., 414 U.S. 291, 296, 94 S.Ct. 505, 509, 38 L.Ed.2d 511 (1973) (permissive joinder); Pentland v. Dravo Corp., 152 F.2d at 856 (intervention).21 To view opt-in suits under § 16(b) as either permissive joinders or efforts to intervene would necessarily require that the plaintiff individually fulfill all of the prerequisites to suit. However, it makes little sense to permit the opt-in class action to be started by a few and later joined by many if individual requirements of exhaustion of remedies can defeat the individual claims of those who opt-in, but have not individually met administrative filing requirements. This could be especially unfair where the time in which claimants must meet filing requirements expires while the court is deciding whether the predominance of a common issue in the claims of class members will realistically prevent or reduce a multiplicity of suits and that individual differences are not so great as to make the action impossibly complex, i.e., determining whether the class action is convenient under the old cases or, in modern terms, manageable. The resolution of these questions of efficiency, convenience and manageability is necessarily left to the discretion of the trial judge who controls the litigation. The exercise of that discretion requires time— time to identify the class members, time to — 1079 — determine the predominance of the common question and time to assess individual differences. If the clock continues to run against those who opt-in after the action is filed, or the clock is turned back when the class is decertified to the date the action commenced, sensible exercise of judicial discretion in determining the relative convenience and efficiency of an opt-in class action will be hampered, and the Congressional intent to allow class actions in cases involving systemic age discrimination will be frustrated. Accordingly, although we recognize the historical analogy of the opt-in class action to permissive joinder and intervention, we conclude that the intent of Congress in enacting § 16(b), as expressed in its language and legislative history, is unlikely to be adequately realized by considering whether each member has timely filed an administrative charge pursuant to 29 U.S.C.A. § 626(d), in deciding whether to continue class certification. The second prerequisite for mandamus is also present with respect to the notice, in that no other adequate means of obtaining relief is available. If the district court’s decertification of the class is upheld on appeal from a final order in Lusardi’s case, claimants who did not file timely charges with the EEOC and were deterred from filing individual actions by the statement in the notice or the holding in the opinion that timely individual filing was required are likely to be barred by the statute of limitations, even if the conditional certification equitably tolled the time for filing a charge. Likewise, those who did not file individual charges and nevertheless forge ahead with individual actions despite the district court’s erroneous holding may find their actions defeated on a law of the case theory. In either event, they would be deprived of their right to appeal a clear legal error. The situation is no different for those who file individual suits if we reverse the decertification order and reinstate the class. In that case, those who accepted the district court’s ruling and did not file individual actions may well be protected. Those who filed individual actions and suffered an adverse final judgment would be barred from rejoining the class. Finally, the holding that timely filing of individual charges was necessary to the maintenance of individual actions by opt-in class members when the class is decertified meets the most stringent test for mandamus. It is beyond the district court’s jurisdiction. A court which rejects a class as improper has no power to bind class members not properly before it. The power to judge the merits of their individual claims lies with the forum in which those claims are presented. Therefore, with respect to the district court’s holding that an individual’s failure to timely file an administrative charge is properly considered in deciding whether opt-in class members are “similarly situated” under 29 U.S.C.A. § 216(b), all the requirements for mandamus are met. A district court has no power or jurisdiction to rule on the merits of the claims of individual members of a putative opt-in class when it denies class certification. This is clear legal error. There is no other adequate remedy. Appellate jurisdiction over the issue may be lost. The language of the notice concerning the necessity for administrative filing and the mingling of that issue with the tolling of the statute of limitations is confusing and has elements of procedural unfairness. We therefore have power under 28 U.S.C.A. § 1651 to issue a writ of mandamus to the district court directing it to correct this particular error. We also believe it is appropriate for us to exercise our discretion and use that power in this case. The issue is important not only to the administration of the ADEA, but also to Title VII. Compare ADEA § 7(d), 29 U.S.C.A. § 626(d) with § 706(e) of Title VII of the Civil Rights Act of 1964, 42 U.S.C.A. § 2000e-5(e) (West 1981); cf. Bean, 600 F.2d at 759 (law developed under Title VII provides useful analogy in interpreting ADEA). It affects the vindication of rights Congress has deemed important to a just society. If the district court’s error is left uncorrected, the utility of the class remedy Congress prescribed in ADEA by incorporating FLSA § 16(b) will be unduly circumscribed. — 1080 — Indeed, little will be left of it in cases alleging systemic discrimination by a large employer. See Spahn, supra note 1, at 151-53. The effect will be felt with particular force in cases of disparate impact when the discrimination is apparent only from statistical evidence. The importance of that theory in the elimination of discrimination has been recently underlined by the Supreme Court. Watson v. Fort Worth Bank & Trust, —- U.S. -, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988); see also Atonio v. Wards Cove Packing Co., 810 F.2d 1477 (9th Cir.) (en banc), on return to panel, 827 F.2d 439 (1987), cert. denied, — U.S. -, 108 S.Ct. 1293, 99 L.Ed.2d 503, cert. granted in part, — U.S. -, 108 S.Ct. 2896, 101 L.Ed.2d 930 (1988). Accordingly, we will grant mandamus and the district court will be directed to vacate its holding on the necessity of timely administrative filings of charges of age discrimination by persons it has held do not constitute a proper opt-in class; to reconsider its order decertifying the class on the basis of disparate defenses without relying on the presence, or absence, of individual administrative charges, or their timing; and if it, nevertheless, concludes class decertification is required, to modify the proposed notice to the persons who opted in to the putative class in conformity with this opinion by eliminating the last sentence of paragraph 3 and conforming the date in paragraph 4 for expiration of the stay or tolling of the statute of limitations on the opt-ins’ individual claims to a date at least thirty days after entry of its order correcting the notice. The stay of the decertification order heretofore granted by this court shall continue until that order is entered. In all other respects, the petition for a writ of mandamus is denied and the appeal dismissed for lack of jurisdiction. Each party shall bear its own costs. Footnotes [1] The district court classified the action as "a disparate treatment claim in which plaintiffs allege they were adversely affected by a pattern or practice of discriminatory treatment.” Plaintiffs’ Appendix (App.) at 645. The district court based its classification on statements made by plaintiffs’ counsel at oral argument. Id. Lusardi’s EEOC and district court complaints allege a pattern of employment discrimination in Xerox’s employment practices — 1065 — that plaintiffs claim could fall within the category of disparate treatment or disparate impact. App. at 22, 72-73. Although the district court also set out the disparate impact theory, identifying it as "facially neutral policies which disproportionately burden those protected,” the court did not discuss this theory in terms of the potential plaintiffs involved in this action. Plaintiffs do not have to choose, of course, under which theory they intend to proceed until discovery has been completed. For purposes of this opinion, we will simply refer to both theories, as applied to these plaintiffs, as "systemic discrimination." For a discussion of the various substantive theories of discrimination, see Spahn, Resurrecting the Spurious Class: Opting-In to the Age Discrimination in Employment Act and the Equal Pay Act through the Fair Labor Standards Act, 71 Geo.L.J. 119, 150-52 (1982) (hereinafter "Spahn”). See also International Bhd. of Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977) (disparate treatment); Griggs v. Duke Power Co., 401 U.S. 424, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971) (disparate impact). [2] The district court declined to certify the appeal. [3] The district court refused to stay the order pending what it correctly saw as an interlocutory appeal. [4] The opinion was filed November 5, 1987. [5] ADEA § 7(b), 29 U.S.C.A. § 626(b) (West 1985), incorporates the enforcement "powers, remedies and procedures" of FLSA § 16, 29 U.S.C.A. § 216(b). [6] Xerox’s reduction in force (RIF) programs included voluntary (VRIF) and involuntary (IRIF) employee terminations. The district court found the individual employment situations varied: The members of the sample group were employed by various Xerox organizations at numerous geographic locations. For example, without regard to departmental differences, members of the sample group were employed by seventeen different Xerox groups or organizations in thirty-four cities or towns in sixteen different states. The claims of the sample group also arise from a variety of employment actions. In this regard, of the sixty-four people in the sample group, twenty-six were terminated as the result of an IRIF, ten allege the VRIFs they accepted were not voluntary, eight employees who were active during the class period allege denial of promotion, five employees who were terminated for unsatisfactory performance allege they were terminated as a result of their age, six employees who left Xerox to accept or seek other employment now allege they were denied promotions or forced to resign, four employees who were permitted resignation now allege their resignations were not voluntary, two employees who accepted normal retirement allege they were forced to do so, two employees who voluntarily separated from Xerox due to job dissatisfaction now allege they were forced to retire; and one employee who resigned before probable discharge alleges the resignation was forced. The discovery also reveals a dramatic lack of similarity in age, salary, organization employment, and geographic location by state and city. With regard to age, twenty-four of the sample group are in the range of forty to forty-five years, seventeen are in the range of forty-six to fifty years, fifteen are in the range of fifty-one to fifty-five years, six are in the range of fifty-six to sixty years, and two are in the range of sixty-one to sixty-five years. With regard to the total compensation received by each of the individuals in the sample group, six received between $10,000 and $20,000 per annum, twenty-six received between $20,000 and $30,000 per annum, fourteen received between $30,000 and $40,000 per annum, seven received between $40,000 and $50,000, eight received between $50,000 and $60,000, two received between $60,000 and $70,000, and one received between $70,-000 and $80,000. Lusardi II, 118 F.R.D. at 357-58 (footnotes omitted). [7] The EEOC, American Association of Retired Persons and Senator John Melcher of Montana, Chairman of the United States Senate Special Committee on Aging, filed motions before this court and were granted leave to file briefs amici curiae in support of the plaintiffs. [8] Although Katz dealt with Rule 23 class actions, we believe the jurisdictional posture of an order granting or denying certification to a class under FLSA § 16(b) is not materially different for purposes of appealability. The essential difference between a Rule 23 and a FLSA class action is that the former includes all class members, present or absent, who do not “opt-out," while the latter requires class members to affirmatively "opt-in." See Lusardi I, 747 F.2d at 177 n. 3. Courts have generally recognized that Rule 23 class actions may not be used under FLSA § 16(b). See, e.g., LaChapelle v. Owens-Illinois, Inc., 513 F.2d 286, 288 (5th Cir.1975) (per curiam) ("fundamental, irreconcilable difference" between Rule 23 and FLSA § 16(b) class actions). While we recognize these substantive differences, we will nevertheless look to Rule 23 decisions by analogy in determining appealability. See generally Spahn, supra note 1, at 144-50. Indeed, the rule denying appealability for class certifications or decertifications may be even more salutary in FLSA § 16(b) opt-in actions than in Rule 23(c) opt-out cases. See infra at 1068. [9] By order dated January 10, 1988, a motion panel of this court stayed the district court’s order pending action on the merits. [10] Such covenants were, of course, later denied enforceability on substantive grounds. Shelley v. Kraemer, 334 U.S. 1, 68 S.Ct. 836, 92 L.Ed. 1161 (1948). [11] In denying the writ on this ground, we express no opinion on whether, either generally or in the context of this case, the presence of conflicting statistical evidence in a systemic discrimination case requires the presentation of that conflicting evidence to a jury for determination of whether a case of discrimination has been made out for all the members of the opt-in class, without regard to their numbers, the disparity of asserted defenses, the utility of a division into sub-classes, or any other factor that may make a class action no more convenient or no more efficient than individual actions in determining the rights of the persons seeking relief by opting into the class. [12] The district court observed: In Gavalik [v. Continental Can Co., 812 F.2d 834 (3d Cir.1987)] the class did in fact have direct, smoking gun evidence of the employer’s discriminatory motivation in certain liability avoidance programs. By contrast, after twenty months of extensive discovery, plaintiffs have not found the smoking gun, the direct evidence of discriminatory motive in this case. The present lack of direct evidence is more than academically significant. If direct evidence of discriminatory motivation existed, Xerox would not be permitted to merely articulate legitimate nondiscriminatory reasons for its actions in response to a prima facie case established by plaintiffs. However, Xerox is not so limited in this case because of the lack of direct evidence to support plaintiffs’ claims. Lusardi II, 118 F.R.D. at 374 (footnote and citations omitted). In a letter addressed to counsel for Lusardi, the district court identified this argument in Lusardi's petition tor mandamus as being based on counsel’s "deliberate summarization” of the opinion: My comment was neither a finding with regard to the evidence nor was it a requirement, as you assert in your Petition, that plaintiffs produce direct proof of discriminatory intent to establish a prima facie. Significantly, I explicitly point out ... "plaintiffs may satisfy [their] burden through use of direct evidence or circumstantial evidence, including statistical evidence.” District Court Letter of January 29, 1988 at 3. [13] Whether this is a disparate impact case or a case involving a pattern of disparate treatment is yet to be determined. See supra note 1. Statistical evidence is an element of proof in both. They differ in other elements. Heretofore, a disparate impact case seems to have required proof of an identifiable facially neutral practice or procedure which bears a disproportionate adverse impact on a protected class. See, e.g., Watson v. Fort Worth Bank & Trust, - U.S. -, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988); Griggs v. Duke Power Co., 401 U.S. 424, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971). A pattern — 1073 — or practice of disparate treatment may be established through statistical and anecdotal evidence of discriminatory intent. International Bhd. of Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977). The category into which this case falls is not yet clear ana Lusardi may rclv on either, or both, as the case develops. [14] Aikens established that direct evidence of discriminatory intent is not required to establish a prima facie case of discrimination under Title VII. Chipollini applied that holding in an action brought under the ADEA. [15] Because Katz is a Rule 23 case, we again acknowledge Lusardi’s correct statement that FLSA § 16(b) class actions alleging a pattern or practice of discrimination are not governed by Fed.R.Civ.P. 23. Nevertheless, as stated, Rule 23 cases can be examined by analogy. See note 8, supra. For the reasons set forth in Part III B. of this opinion, we think it would be inappropriate to treat § 16(b) as merely a permissive join-der. We are instead inclined to the view set forth by Professor Spahn analogizing actions under § 16(b) to the spurious class action “treated in the former version of Rule 23.” See Spahn, supra note 1, at 139. [16] Although Professor Spahn suggests that systemic cases are generally appropriate for class treatment because the alleged discrimination creates a common issue of fact coupled with the assertion of an important right, she does not argue that class treatment is warranted in all such cases. See Spahn, supra note 1, at 149-50. [17] Lusardi’s fourth argument for mandamus, that the district court exceeded its lawful power by including among the defenses available to Xerox the releases it obtained from members of the opt-in class who accepted "voluntary" terminations, is also unavailing because that issue will be reviewable on appeal from a final order. [18] (b) Limitation of Federal action upon commencement of State proceedings In the case of an alleged unlawful practice occurring in a State which has a law prohibiting discrimination in employment because of age and establishing or authorizing a State authority to grant or seek relief from such discriminatory practice, no suit may be brought under section 626 of this title before the expiration of sixty days after proceedings have been commenced under the State law, unless such proceedings have been earlier terminated: Provided, That such sixty-day period shall be extended to one hundred and twenty days during the first year after the effective date of such State law. If any requirement for the commencement of such proceedings is imposed by a State authority other than a requirement of the filing of a written and signed statement of the facts upon which the proceeding is based, the proceeding shall be deemed to have been commenced for the purposes of this subsection at the time such statement is sent by registered mail to the appropriate State authority. 29 U.S.C.A. § 633(b) (West 1985). [19] We take no position on whether the ADEA filing time, as well as the statute of limitations, was equitably tolled for the opt-in plaintiffs during the period of conditional class certification. That issue must be addressed by the forum in which a future complaint is filed. See Dempsey v. Pacific Bell Co., 789 F.2d 1451, 1453 (9th Cir.1986) (giving plaintiff 90 days to refile complaint and leaving question of statute of limitations to district court); Naton v. Bank of California, 649 F.2d 691, 695-96 (9th Cir.1981) (leaving issue of equitable tolling to district court). [20] Section 7(d) currently provides that the charge is to be filed with the EEOC. See 29 U.S.C.A. § 626(d). [21] See abo Spahn, supra note 1, at 132-33 (discussing improper analogy between § 16(b) actions and joinder devices).