Opinion for the Court filed by Circuit Judge GINSBURG.
GINSBURG, Circuit Judge:This action, brought under Title VII of the Civil Rights Act of 1964, as amended, 42 U.S.C. § 2000e et seq., centers on the promotion system for upper echelon positions instituted in January 1976 at United States Postal Service (USPS) Headquarters. Plaintiff-appellant Mary P. Valentino charges that the system, in operation, accords women disadvantageous treatment based on their sex. Valentino pursues on appeal an individual charge that, under the promotion system in question, she was discriminatorily denied advancement to the position of Director of the Office of Employee Services, an office within the Employee and Labor Relations (E&LR) Group at USPS Headquarters. She also pursues, on behalf of a class, a charge that women holding upper echelon posts at USPS Headquarters since June 16, 1976, have been denied promotions on the basis of their sex.1
Following a two-week non-jury trial, at which twenty-nine witnesses were heard and hundreds of pages of exhibits were introduced, both sides submitted detailed proposed findings of fact and conclusions of law. The district court substantially accepted, with some modification, USPS’s proposed findings and conclusions2 and *61ruled against Valentino on both her individual and class claims. 511 F.Supp. 917 (D.D.C.1981). Upon close review of the record and consideration of counsels’ arguments, we find no reversible error in the district court’s adjudication and therefore affirm the judgment in favor of USPS.
As to Valentino’s individual claim, we find ample record support for the district court’s conclusions that (1) Valentino established a prima facie case, (2) USPS produced adequate evidence of a legitimate, nondiscriminatory reason for its action — it explained that those involved in the selection process fairly and rationally judged Valentino a well-qualified, but not the best qualified applicant, (3) Valentino failed to carry the ultimate persuasion burden the law placed upon her to demonstrate that the reason USPS proffered was a pretext cloaking sex discrimination. See Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 101 S.Ct. 1089, 67 L.Ed.2d 207 (1981).
Regarding the class claim, both sides tendered statistical evidence heavier in volume than in persuasive content. Valentino’s presentation showed a conceded and familiar pattern — USPS employed a distinctly smaller percentage of women in top posts than in lower level jobs. The higher level promotion positions at issue, however, covered a wide range, including inter alia, confidential secretaries, lawyers, engineers, business managers. Valentino’s data did not offer a sufficiently refined comparison of male and female employees of similar qualifications, her statistics did not hone in on the wide variety of minimum objective qualifications required of applicants for the diverse promotion positions filled at USPS Headquarters during the relevant time period. See Hazelwood School District v. United States, 433 U.S. 299, 308 n.13, 97 S.Ct. 2736, 2742 n.13, 53 L.Ed.2d 768 (1977) (when special qualifications are required to fill professional positions, comparative statistics should take into account those special qualifications); Wilkins v. University of Houston, 654 F.2d 388, 408-10 (5th Cir. 1981) (plaintiff must demonstrate availability of qualified class members for particular professional and administrative occupations); Pack v. Energy Research & Development Administration, 566 F.2d 1111, 1113 (9th Cir. 1977). Moreover, her data did not show marked differences in the advancement rate of men and women after January 28,1976, the effective date of the promotion system operative for the class. See 511 F.Supp. at 940; J.A. 207-09. Although we do not share the district court’s view of the force of USPS’s statistical proof, we disregard as surplusage the findings and conclusions addressed to that data. It sufficed for the district court to determine correctly, as it did, that Valentino failed to prove the claim her complaint stated on behalf of the class.
I. THE INDIVIDUAL CLAIM
A. Facts
Mary P. Valentino has had a long career in the personnel field in government service. Her USPS employment, however, has been relatively short-term. She was initially engaged by the Post Office Department in October 1970 to work as an Employment Specialist (Women). Prior to that assignment, she held the post of Director of Personnel at the Equal Employment Opportunity Commission. The Post Office position increased her annual salary by approximately $2,000. She remained in this position only seven months. The functions of the Post Office Department were being transferred to USPS during this period and Valentino was among a group of employees eligible to retire early with a six-month pay bonus. She opted for early retirement and accepted a bonus of approximately $13,600.
Nine months later, in February 1972, Valentino re-entered federal service first at the Food and Drug Administration and, in December 1972, at the Consumer Product Safety Commission (CPSC), where she became Director of Personnel. In late 1973 *62and early 1974, USPS officials sought out Valentino and she accepted work at USPS Headquarters commencing February 1974 as Director, Office of Career Planning, in the E&LR Group. At CPSC, her annual salary was $32,932. Her starting salary at USPS was $33,500; unlike most early retirees who returned to postal employment, she was not required to repay the six-month bonus paid to her in May 1971. Valentino’s new office at USPS had a total staff of three; it was responsible for supervision of a Women’s Program and career planning for postal employees.
In sum, until 1976, it appears that Valentino’s own Post Office and USPS employment experience entailed no gender-based unfavorable treatment, and we do not understand Valentino to contend otherwise.3 Rather, her individual claim stems from a 1976 efficiency-oriented realignment of the E&LR Group, which eliminated the Office of Career Planning, and her unsuccessful bids for promotion to two newly-created posts.
In September 1975, planning commenced for a realignment of the USPS Headquarters E&LR Group. The objective was to streamline the organization and eliminate functional duplication. All high-level officers, including Valentino, participated in the realignment planning effort. The realignment was announced in April 1976, effective June 4, 1976; it entailed the elimination of approximately 140 positions, including Valentino’s. Several new posts were created (although the retrenchment left approximately 100 fewer positions), and Valentino unsuccessfully applied for two of them: Director, Office of Human Resources (OHR); and Director, Office of Employee Services (OES). Both posts carried ranks higher than the Office of Career Planning position Valentino then occupied.4 In her administrative charge and in the district court Valentino challenged as sex discriminatory USPS’s failure to select her for the OHR as well as the OES directorship. On appeal, she has confined her individual, discriminatory failure to promote argument to the OES post.
In addition to her Title VII sex discrimination charge, Valentino filed an ultimately unsuccessful administrative charge with the Civil Service Commission in which she attacked the realignment because it was not handled as a reduction in force (RIF), to which the Veterans Preference Act applies. See 39 U.S.C. § 1005(a)(2) (applying 5 U.S.C. § 3502, which governs reductions in force, to USPS preference eligible employees). As the widow of a veteran, Valentino was entitled to preference under that Act. See J.A. 636-37, 639-41. At the trial of the Title VII case now before us, the Senior Assistant Postmaster General who initiated the realignment testified that among the reasons USPS sought to avoid RIF procedures was the serious adverse effects those procedures would have on opportunities for female employees, most of whom hold no veterans’ preference. J.A. 728-29. In its findings and conclusions, the district court noted a certain inconsistency in Valentino’s “advocacy of women’s rights” at the same time she urged that veterans’ preference bumping rights should have attended the retrenchment. 511 F.Supp. at 921, 951.5
*63Valentino’s promotion applications were processed under a system inaugurated in January 1976. Known as the Essential Vacancy System (EVS), the system implemented the Postmaster General’s November 1975 directive that all vacancies at Headquarters be filled from within. The EVS prohibited filling vacant positions not considered essentia], required the posting of vacancies that would be filled, and established review committees to screen applicants. These committees were to include, if possible, either a female or a member of a minority. See 511 F.Supp. at 921. The five-member review committee that screened Valentino’s OHR and OES applications included one woman and one member of a minority group.
Fifty-three persons applied for the OES directorship. Valentino was among the six top candidates interviewed for the post. However, she was not among the three persons the committee referred to the official who made the final selection. Valentino introduced witness testimony tending to show, albeit with less than crystal clarity, that three committee members initially favored forwarding her name to the selecting official. However, the same witnesses also testified that the recommended list, which did not include Valentino’s name, represented the ultimate consensus of the committee. See J.A. 670-71, 680, 844-46.
The person promoted to the OES position, Elmer Weems, previously held a USPS rank and received a salary comparable to Valentino’s. He had an unbroken work record at USPS since 1970. During that period, he held several management posts and had acquired extensive field experience.
Some weeks after the selection of persons other than Valentino for the OHR and OES directorships, the Women’s Program, originally placed in the realignment under OHR, was transferred to OES. See J.A. 717, 841-42. Valentino accepted a lateral reassignment in August 1976. Although she asserted in the district court that the reassignment was a discriminatory “downgrade,” see 511 F.Supp. at 926, 947, she has not pressed that contention on appeal.6 Valentino left the USPS in October 1978 for a higher salaried position with the Community Services Administration.
B. Legal Analysis
To establish a prima facie case of discriminatory refusal to promote, a plaintiff need only
show that she belongs to a protected group, that she was qualified for and applied for a promotion, that she was considered for and denied the promotion, and that other employees of similar qualifications who were not members of the protected group were indeed promoted at the time the plaintiff’s request for promotion was denied.
Bundy v. Jackson, 641 F.2d 934, 951 (D.C.Cir.1981). The burden is “not onerous,” see Texas Department of Community Affairs v. Burdine, supra, 450 U.S. at 253, 101 S.Ct. at 1093-1094, and the district court, departing from USPS’s position on this point, properly determined that Valentino crossed the first threshold.
To meet a prima facie case, a defendant must present with clarity and reasonable specificity a legitimate, nondiscriminatory reason for the action it took. Id. at 254-56, 258, 101 S.Ct. at 1094-1095, 1096. To satisfy this “intermediate burden,” USPS was not required to convince the court that it in fact chose the better applicant. Id. at 258-59, 101 S.Ct. at 1096-1097. However, it was obliged to produce evidence “which would allow the trier of fact rationally to conclude that the employment decision had not been motivated by discriminatory animus.” Id. at 257, 101 S.Ct. at 1095-1096; see St. Peter v. Secretary of the Army, 659 F.2d 1133, 1139 (D.C.Cir.1981) (Mikva, J., concurring in the result).
*64The district court, we believe, correctly held that USPS carried its intermediate production burden by presenting evidence of a substantial, nondiscriminatory reason for the OES directorship choice. USPS pointed to Weems’s longer and more varied postal service experience and the posted announcement that applicants should have “[e]xtensive experience in . .. postal field operations and management.” The announcement also stated that the position would entail work with various programs involving USPS employees. Valentino, although the district court judged her more highly qualified than Weems with regard to the personnel service function of the office, unquestionably had a shorter term, narrower range exposure to USPS programs, and lacked field experience. See 511 F.Supp. at 925.
In view of USPS’s presentation of a cogent, nondiscriminatory explanation for not selecting Valentino for the OES post — her limited experience in postal service operations — it was incumbent upon Valentino to persuade the trial court that “a discriminatory reason more likely motivated [USPS]” or that “[USPS’s] proffered explanation is unworthy of credence.” Texas Department of Community Affairs v. Burdine, supra, 450 U.S. at 256, 101 S.Ct. at 1095. We discern no “clear error” in the district court’s determination that Valentino failed to carry this ultimate burden of establishing intentional sex discrimination.
In her briefs on appeal, Valentino emphasized the transfer of the Women’s Program she once supervised from OHR to OES shortly after both director posts were filled. See J.A. 717, 841-42. She charged deliberate manipulation of the initial job descriptions to block her advancement. Valentino reasons that USPS officials knew her chances for the OHR post were limited and must have considered her a front runner for the OES post if it encompassed responsibility for the Women’s Program. The district court considered but was not persuaded by this argument. See 511 F.Supp. at 925 n.5. We need only point out that the inferences Valentino draws from the change in the Women’s Program location, if plausible, are hardly compelled by the evidence.7
The district court, in sum, could properly conclude that USPS’s explanation for failing to select Valentino for the OES directorship was worthy of credence and was not a pretext to cover up sex discrimination. We must therefore decline to disturb the district court’s decision that Valentino ultimately failed to prove she was the victim of gender-based animus on the part of USPS.
II. THE CLASS CLAIM
After rejecting Valentino’s initial descriptions of the proposed class as overbroad, 511 F.Supp. at 926-27, the district court certified her representation of
all females who since June 16, 1976, have been employed or are now employed by the United States Postal Service (USPS) Headquarters in the Washington, D. C., statistical metropolitan area in the positions compensated according to USPS pay scales at level PES-17 or higher, and who have been accorded disparate treatment because they have been denied advancement on the basis of their sex, excluding, however, any such females who are union members or who participated in the personnel decisions resulting in plaintiff Mary P. Valentino’s individual claims.
511 F.Supp. at 919.8 In its. analysis the district court prescribed the applicable time frame for proof of the class claim, described the legal framework for adjudicating the claim, and appraised both the statistical evidence submitted and the testimony of sev*65eral witnesses concerning experiences, practices, and programs at USPS Headquarters relating to employment opportunities for women in upper echelon posts. We address in turn each of these aspects of the district court’s evaluation.
A. The Time Dispute
The trial court ruled untimely Valentino’s claims of sex discrimination between June 1974 and January 1976. 511 F.Supp. at 949-51. We agree and need not rehearse in detail the reasons supporting that ruling. Relief under Title VII is ordinarily dependent upon the filing of a timely administrative charge. United Airlines, Inc. v. Evans, 431 U.S. 553, 558, 97 S.Ct. 1885, 1889, 52 L.Ed.2d 571 (1977); Alexander v. Gardner-Denver Co., 415 U.S. 36, 47, 94 S.Ct. 1011, 1019, 39 L.Ed.2d 147 (1974); see Zipes v. Trans World Airlines, Inc., - U.S. -, -, 102 S.Ct. 1127, 1132, 71 L.Ed.2d 234 (1982) (timely charge filing requirement is not a jurisdictional prerequisite to suit in district court; it operates “like a statute of limitations, [which] is subject to waiver, estoppel, and equitable tolling”).9 Valentino, subject to the requirement that she bring a discrimination claim to the attention of her agency’s Equal Employment Opportunity Counselor within 30 days of the alleged violation, 5 C.F.R. § 713.214(a)(1) (1978) (redesignated in 1979 as 29 C.F.R. § 1613.214(a)(1)), did so in June 1976. Complaint at 6, J.A. 47. The timing of her informal administrative complaint, therefore, would indicate May 1976 as the opening point for her proof.
Valentino would reach back to June 1974 as the starting place for proof of her class claim. To avoid the time bar, she invokes the “continuing violation” theory. See Evans, 431 U.S. at 558, 97 S.Ct. at 1889; Milton v. Weinberger, 645 F.2d 1070, 1074-77 (D.C.Cir.1981); Laffey v. Northwest Airlines, Inc., 567 F.2d 429, 473 (D.C.Cir.1976), cert. denied, 434 U.S. 1086, 98 S.Ct. 1281, 55 L.Ed.2d 792 (1978). To establish a continuing violation, however, Valentino would have to show “a series of related acts, one or more of which falls within the limitations period, or the maintenance of a discriminatory system both before and during the [limitations] period.” B. Schlei & P. Gross-man, Employment Discrimination Law 232 (Supp.1979). She is in no position to do so. As we earlier observed, see supra p. 63, USPS inaugurated a new promotion system, the EVS, in January 1976. The district court considered the operative features of the new system and fairly concluded that “promotion claims which may have arisen in 1974 and 1975 involved practices which ceased” upon implementation of the EVS. 511 F.Supp. at 950.10 Moreover, in view of her own pre-1976 employment history with USPS, see supra pp. 61-62, Valentino is a questionable champion of those who may have encountered sex-based discrimination prior to the advent of the EVS.
Valentino has not successfully identified any feature of the EVS that projects into 1976 and beyond a specific past discriminatory device or practice. See Evans, 431 U.S. at 558, 97 S.Ct. at 1889 (critical question is not whether past practices have current consequences, but whether “any present violation exists”) (emphasis in original). The position advertising requirement of the EVS has made women aware of promotion opportunities, 511 F.Supp. at 933, promotion application screening committees under the EVS are composed with a view toward inclusion of women and minority groups members, id. at 921-22, and Val enti-*66no demonstrated no built-in bias or unfair procedure tainting the system, id. at 951. Cf. Kohn v. Royall, Koegel & Wells, 59 F.R.D. 515, 521 (S.D.N.Y.1973). She attacks the EVS because it does not establish objective criteria for each position, for its flexibility — for example, it does not specify the precise number of individuals to be interviewed for each vacancy — and for inadequate record maintenance. But as the district court observed, it is difficult to imagine an effective selection process for high-level management positions that operates with the objectivity and rigidity Valentino’s criticisms suggest. 511 F.Supp. at 933, 952. See generally Waintroob, The Developing Law of Equal Employment Opportunity at the White Collar and Professional Level, 21 Wm. & Mary L.Rev. 45, 48-62 (1979) (at higher levels, test of legality of white collar employment practices is not whether they are subjective, as inevitably they are, but whether they are applied fairly).
In sum, the district court correctly determined that Valentino was unable to identify a continuing pattern or scheme of the kind that would justify avoidance of the general applicable Title VII time frame.11
B. The Character of the Class and the Legal Framework for the Representative’s Disparate Treatment Case
Position titles in the sector of USPS to which Valentino’s class claim relates include national medical director, mail classification specialist, confidential secretary, real estate specialist, electronic engineer, psychologist, computer systems analyst, accountant, editor, principal economist, occupational safety and health specialist, senior attorney, philatelic program specialist, mechanical engineer, architect, labor relations executive, market specialist, news information officer, consumers affairs associate, government relations representative, and photo-journalist. J.A. 374-512. In view of this far-ranging diversity of professional and administrative occupations, and the distinct qualifications necessary for entry and advancement in each occupational category, Valentino’s burden to show that USPS promotion decisions were based on gender — rather than on factors other than the promotion-seeker’s sex — was especially difficult to meet.12 The evidence she presented failed to take into account mini*67mum objective qualifications (specialized education and experience) prerequisite to consideration for the variety of upper echelon promotion positions filled at USPS Headquarters under the EVS. This failure to grapple with the qualifications represented among class members in relation to the qualifications necessary for the promotion-positions at issue proved fatal to her case.13
“[A]ny Title VII plaintiff must carry the initial burden of offering evidence adequate to create an inference that an employment decision was based on a discriminatory criterion illegal under the Act.” International Brotherhood of Teamsters v. United States, 431 U.S. 324, 358, 97 S.Ct. 1843, 1866, 52 L.Ed.2d 396 (1977). Because Valentino alleged systemwide discrimination,
[she] ultimately had to prove more than the mere occurrence of isolated or “accidental” or sporadic discriminatory acts. [She] had to establish by a preponderance of the evidence that [sex] discrimination was the [employer’s] standard operating procedure — the regular rather than the unusual practice.
Id. at 336, 97 S.Ct. at 1855 (footnote omitted). Moreover, because her case concerned high level professional, administrative, and managerial positions rather than jobs involving skills “many persons possess or can fairly readily acquire,” she could not succeed simply by submitting proof that might suffice where entry level or blue collar promotion posts are at stake. See Hazelwood, 433 U.S. at 308 n.13, 97 S.Ct. at 2742 n.13.14
We set out above, quoting from Bundy v. Jackson, the character of proof sufficient to establish a prima facie case of discriminatory refusal to promote an individual. See supra p. 63. The Bundy prescription is an adjustment of the formula established for discriminatory refusal to hire an individual in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802, 93 S.Ct. 1817, 1824, 36 L.Ed.2d 668 (1973).15 McDonnell Douglas recognized that “Congress did not intend by Title VII ... to guarantee a job to every person regardless of qualifications.” Id. at 800, 93 S.Ct. at 1823 (quoting Griggs v. Duke Power Co., 401 U.S. 424, 430, 91 S.Ct. 849, 853, 28 L.Ed.2d 158 (1971)). Hence the Supreme Court identified as an essential ingredient of the plaintiff’s prima facie case a showing that the job seeker met minimum qualifications for the position in question.
Qualifications also figure importantly in the representative’s threshold case on behalf of a class when professional and management positions are at issue. “When special qualifications are required to fill partic*68ular jobs,” proof that does not center on those “who possess the necessary qualifications” falls short. Hazelwood, 433 U.S. at 308 n.13, 97 S.Ct. at 2742 n.13.16
In Title VII class actions, statistical proof is a prominent part of the prima facie case. Indeed, such proof, when it is compelling, may even suffice to carry the class over the threshold. Teamsters, 431 U.S. at 339, 97 S.Ct. at 1856. But statistics “come in infinite variety,” and “their usefulness depends on all of the surrounding facts and circumstances.” Id. at 340, 97 S.Ct. at 1856-1857; Hazelwood, 433 U.S. at 312, 97 S.Ct. at 2744. When the job qualifications involved are ones that relatively few possess or can acquire, statistical presentations that fail to focus on those qualifications will not have large probative value. See Hazelwood, 433 U.S. at 308 n.13, 97 S.Ct. at 2742 n.13; Ste. Marie v. Eastern Railroad Association, 650 F.2d 395, 400-01 (2d Cir. 1981).17
“[N]ot every conceivable factor relevant to a [high level] promotion decision” must be considered in the statistical data presented initially by a plaintiff, but “the minimum objective qualifications necessary for one to be eligible for promotion must be considered.” Davis v. Califano, 613 F.2d 957, 964 (D.C.Cir.1980) (emphasis in original). The burden of comparing appropriate groups in terms of minimum objective qualifications, onerous here because of the disparate occupational categories involved, is far more tractable when all members of the class are professiorial, administrative or technical employees with generally similar job skills and seek advancement to positions involving those same skills. See Trout v. Hidalgo, 517 F.Supp. 873, 883 (D.D.C. 1981).18
Generally, as part of their prima facie case, class action plaintiffs offer a combination of statistical proof and individual testimony of special instances of discrimination. Even when the statistical proof is so compelling that it might, in itself, satisfy the plaintiff’s initial burden, the prima facie case is bolstered and the court’s evaluation is aided by testimony recounting personal experiences of class members. Such testimony may “[bring] the cold numbers convincingly to life.” Teamsters, 431 U.S. at 338-39, 97 S.Ct. at 1855-1856 (statistics bolstered by testimony of over 40 specific instances of discrimination); see Hazelwood, 433 U.S. at 303, 305, 97 S.Ct. at 2739, 2740 (prominence of specific instances in race discrimination “pattern and prac*69tice” attack on school district’s hiring of teachers); Chisholm v. United States Postal Service, 665 F.2d 482 (4th Cir. 1981) (statistics showed gross disproportion between blacks in supervisory posts and in total work force; 20 former and present black employees testified about their educational and employment backgrounds and attainments, and their unsuccessful efforts to obtain promotions). And when the statistical evidence does not adequately account for “the diverse and specialized qualifications necessary for [the positions in question],” strong evidence of individual instances of discrimination becomes vital to the plaintiff’s case. Wilkins, 654 F.2d at 410; see also Ste. Marie, 650 F.2d at 406-07.
Valentino’s ease on behalf of the class included both statistical presentations and individual testimony. The district court determined that the quality of both modes of proof fell below the threshold necessary to establish a prima facie case. We agree, but add this cautionary note. Because the fate of disparate treatment claims is heavily dependent on the setting, facts, and circumstances of the particular case, decisions in such cases “have limited precedential value.” Ste. Marie, 650 F.2d at 397. “[I]t is neither practical nor useful to write appellate opinions dealing in detail with every facet of each case.” Id. Accordingly, we do not “discuss every argument made by the parties,” id., or retread, step by step, the territory covered in the district court’s findings and conclusions. Instead, we confine our discussion to the principal infirmities in the quality of the proof presented to and described at length by the district court.
C. The Statistical Evidence
Valentino presented three principal categories of statistical evidence: exhibits showing that as grade level increases, the percentage of women in the USPS work force significantly decreases; exhibits comparing the grade increases received by men and women at USPS Headquarters; and the “centerpiece” of her case, 511 F.Supp. at 943, regression analyses attempting to measure the influence of gender on the salary of USPS employees. This evidence unquestionably demonstrated that men dominate the top-salary positions at USPS. But that fact is not in dispute, and Valentino’s statistical proof failed to establish much more. Because Valentino did not control for occupational classifications, the district court could not make the “quantum leap” from her showing of few women at the top to a conclusion that sex discrimination by the employer is the explanation. See Ste. Marie, 650 F.2d at 400. As the district judge summarized: “[I]t is impossible to determine from [Valentino’s] statistics whether similarly situated and similarly available men and women have been treated differently from each other.” 511 F.Supp. at 940 (emphasis in original).
1. Grade level and grade increase statistics
Valentino presented a series of exhibits19 which “generally repeat the same theme: in USPS Headquarters at the 17 and above levels, there are a larger percentage of women in the lower levels than in the upper levels.” 511 F.Supp. at 941. The exhibits convincingly demonstrated that USPS did not fill top jobs “random fashion,” but Valentino does not contend it should. This statistical presentation would be meaningful only if, in the absence of any sex bias on the part of USPS, one would expect to find the same distribution of men and women at each level, 17 and above. Such an expectation is hardly realistic given a national work force in which “the majority of secretaries are women [and] the majority of lawyers and engineers are men.” Id. at 941; see infra note 22; J.A. 617-20. In view of the *70highly specialized education and training necessary for the diverse occupational categories this case involves, and the uneven distribution of the sexes in those occupational categories, no inference of discrimination can be drawn from the bare fact that “USPS promotion policy [does not yield] an equal distribution of men and women at each level.” 511 F.Supp. at 941-42.
Valentino did factor years of education into some of her grade level exhibits. J.A. 149-69. The district court properly evaluated those exhibits as “of little if any value.” The educational data were manifestly inaccurate. Moreover, the exhibits were characteristic of the problem that pervades Valentino’s proof. They did not group employees by job category, nor did they take into account type of education or experience. 511 F.Supp. at 943.
Valentino’s grade increase exhibits attempted to compare promotions of men and women at USPS Headquarters, but again the comparisons were between “men, in general [and] women, in general.” Id. at 939. The exhibits provide no basis for comparing men and women similarly educated and equipped to pursue the same occupations. Moreover, the grade increase statistics failed to show any pattern operating to the disadvantage of women; they did not demonstrate that women received grade increases less often than men as regular occurrence.20
2. Regression Analyses
Multiple regression analysis is a statistical method for making “quantitative estimates of the effects of different factors on some variable of interest.” Fisher, Multiple Regression in Legal Proceedings, 80 Colum.L.Rev. 702, 702 (1980). The regression technique has come into vogue in employment discrimination cases as a means of estimating the discrete influence factors such as sex, experience, and education have had on determining salary level. See, e.g., Finkelstein, The Judicial Reception of Multiple Regression Studies in Race and Sex Discrimination Cases, 80 Colum.L.Rev. 737 (1980). Valentino’s regression analyses attempted to estimate how much influence sex, length of government service, and years of education had on the salaries of USPS employees.21 Her analyses might have been impressive were this a case in which the women and men in the employer’s work force held similar jobs. See Trout, 517 F.Supp. at 883 (“all members of the class [were] professional technical employees with generally similar job skills”). But Valentino utterly failed to control for type of education and job classification, see Fisher, supra, 80 Colum.L.Rev. at 721 n.32. The array of positions at issue and the difference in the availability of men and women in the nation and at USPS for the varied high level professional and administrative positions filled at Headquarters stripped her analyses of the significance she *71would attribute to them. Cf. Wilkins, 654 F.2d at 409 n.37.22
When it is clear that qualification, e.g., as an economist, engineer, lawyer, computer expert, statistician, accountant, business manager, secretary, is a prime factor in the selection process, a Title VII plaintiff cannot shy away from that factor in developing her prima facie case.23 In the setting we have here it would be irrational to assume “equal qualifications” to fill engineering or secretarial vacancies among persons educated the same number of years24 and employed by the government for the same length of time.25 Because Valentino’s regression model ignores information central to understanding the causal relationships at issue,26 the district court could not accept her “centerpiece” proof as adequate to raise an inference that sex discrimination accounts for salary differences in the upper echelons at USPS Headquarters. See D. Baldus & J. Cole, Statistical Proof of Discrimination 190-91, 243-44, 274 (1980).27
D. Individual Testimony
Valentino presented testimony of five class members. Each testified to having been denied a promotion and to her belief that her sex was the operative reason. The district court characterized this testimony as “conclusory” and, in any event, insufficient “to show that sex discrimination is the ‘standard operating procedure’ of the USPS Headquarters or of the Essential Vacancy System.” 511 F.Supp. at *72951 (quoting Teamsters, 431 U.S. at 336, 97 S.Ct. at 1855). We cannot gainsay those determinations.
Four of the five witnesses Valentino presented had started with the USPS as clerical or blue collar workers and had advanced to professional positions. One complained of an unsuccessful bid for a position in the Northeastern Region. This case, however, is confined to employment practices at USPS Headquarters. Two asserted that gender discrimination was responsible for rejection of their competing applications for a Marketing Specialist position. Whether their claims were well-founded depended on their credibility versus the credibility of the selecting officer. The district court credited the selecting officer’s testimony, 511 F.Supp. at 928, and we are in no position to reject that evaluation as “clearly erroneous.” See Fed.R.Civ.P. 52(a).
A fourth witness said the official who evaluated her performance was motivated by a general bias against women, but the woman who chaired the review committee for the promotion in question gave a different explanation. She testified that the low committee rating was based on the interview and on the applicant’s unimaginative approach to problem solving. The fifth witness’ testimony was similarly countered by another woman review committee head. In both cases, the district court credited the women who chaired the review committees. 511 F.Supp. at 930-32.
Finally, Valentino introduced evidence concerning the efforts of an official to block the promotions of two class members. Both achieved the promotions and neither testified that she considered herself the victim of sex discrimination.
In rebuttal, USPS called three women who occupied high level posts and had served on or chaired several review committees. They testified to the advancement opportunities they and other women had been afforded through USPS education and training programs. 511 F.Supp. at 930-33.28
Our review of the individual anecdotal testimony fully accords with the district court’s evaluation. The testimony falls far short of indicating a systematic policy of gender-based discrimination at USPS Headquarters.29
E. A Note in Summation
We have indicated some of the shoals in the way of plaintiffs who would prove discrimination against a class of employees seeking access or promotion to diverse, highly-skilled positions. Indeed, the breakdown of highly specialized workplaces into occupational categories for the purpose of examining the treatment of similarly qualified employees may yield numbers too small to conduct certain types of statistical analyses relied upon to show discrimination in workplaces less specialized. See supra p. 68; Wilkins, 654 F.2d at 409 n.37. Nonetheless, small numbers are not per se useless, especially if the disparity shown is egregious.30 See D. Baldus & J. Cole, su*73pra, at 300, 309-10. The “inexorable zero,” Teamsters, 431 U.S. at 342 n.23, 97 S.Ct. at 1858 n.23, can raise an inference of discrimination even if the subgroup analyzed is relatively small.
Statistics are not the only way to show discrimination:
The great debates concerning the nature and use of statistical proof in employment discrimination cases should not obscure the fact that nonstatistical proof also plays an important role in the determination of the prima facie case in many class actions, particularly where ... the statistical sample is small or the disparity not egregious.
B. Schlei & P. Grossman, supra, 1976 main text at 1193. When the capability of statistics to show discrimination is limited, see, e.g., id. at 1190, 1979 Supp. at 329-30, plaintiffs can press into better service nonstatistical proof of disparate treatment. See supra pp. 68-69; F. Morris, Current Trends in the Use (and Misuse) of Statistics in Employment Discrimination Litigation 49 (1977).
In this case, the combination of unrefined statistics and thin proof of individual instances of discrimination leaves the adjudicator without any basis for concluding that gender impeded Valentino and the class she would represent from moving into the highly skilled, diverse positions in the upper ranks at USPS Headquarters.31 If race or sex bias in fact infects selection across-the-board for jobs at the top, however, it should not be impossible to assemble the evidence from which a pattern of discrimination can be inferred. “[Pjractition-ers should not lose sight of the fact that employment discrimination cases require the same ingenuity regarding methods of proof as any other court litigation.” B. Schlei & P. Grossman, supra, 1976 main text at 1193 (footnote omitted).
CONCLUSION
Our review of the record and arguments in this case persuades us that the district court correctly concluded that USPS presented a legitimate, nondiscriminatory reason, unrebutted by Valentino, for not selecting Valentino for the OES post, and that Valentino’s unrefined statistics coupled with her nonstatistical evidence failed to establish a prima facie case of discrimination against the certified class. Accordingly, the judgment of the district court is
Affirmed.
. “Disparate treatment” claims, such as Valentino’s individual and class claims here, rest on charges that “[t]he employer simply treats some people less favorably than others because of their race, color, religion, sex, or national origin.” International Bhd. of Teamsters v. United States, 431 U.S. 324, 335 n.15, 97 S.Ct. 1843, 1854 n.15, 52 L.Ed.2d 396 (1977). Disparate treatment claims may be distinguished from “disparate impact” claims. “The latter involve employment practices that are facially neutral in their treatment of different groups but that in fact fall more harshly on one group than another and cannot be justified by business necessity.” Id. Discriminatory motive is ultimately an essential element in a disparate treatment case; such a motive is not a necessary ingredient of a disparate impact claim. Id. Although “[ejither theory may ... be applied to a particular set of facts,” id., Valentino’s claims have proceeded solely on a disparate treatment theory. See Valentino v. United States Postal Service, 511 F.Supp. 917, 945, 948 (D.D.C.1981). There is accordingly no issue before us as to the validity of any facially neutral promotion test or standard USPS uses. See McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802 n.14, 93 S.Ct. 1817, 1824 n.14, 36 L.Ed.2d 668 (1973).
. While we do not believe the district court is fairly criticized for “mechanical[ly]” adopting USPS’s submission, see Brief for Appellants at 7 n.4, we have examined the decision with special care in light of Valentino’s point that the findings were not initially penned by the district judge. We reject the suggestion, however, that the district court’s substantial acceptance of USPS’s proposed findings warrants *61our departure from the “clearly erroneous” standard of review. See Hagans v. Andrus, 651 F.2d 622, 626 (9th Cir.), cert. denied, - U.S. -, 102 S.Ct. 313, 70 L.Ed.2d 157 (1981).
. Valentino does assert, however, that the effort “to induce [her] to rejoin the USPS and establish the Women’s Program,” was made in response to the complaints of “[fjemale employees at USPS Headquarters . . . that they were the subject of discriminatory practices, particularly the lack of proper advancement.” Brief for Appellants at 20-21.
. Valentino’s Career Planning position ranked at level PES (Postal Executive Schedule) -28 on the USPS pay scale then effective. The OHR post was scheduled at PES-30, and the OES directorship at PES-29. See 511 F.Supp. at 919-21.
. Brief for Appellants at 40 cites one of the findings on this point as evidencing the trial court’s “extraordinary and inexplicable hostility” toward Valentino. However, Valentino does not dispute that the RIF procedures she advocated “disproportionately favor male employees,” and her brief does not address, specifically, the district court’s comment that “[her] advocacy of RIF procedures which might help her but not women generally impairs the efficacy of her representative position.” 511 F.Supp. at 921.
. Nor has she reasserted here her contention in the district court that the realignment improperly lowered the Women’s Program in the USPS hierarchy. She does maintain, however, that the transfer of the program from OHR to OES after the realignment and after OHR and OES directors were selected was pre-designed to prevent her from gaining the OES post. See infra p. 64.
. We note in this regard the district court’s finding, with record support, that “[d]uring th[e] planning phase, a conscientious effort was made to avoid considering what specific people would fill what specific positions in the new organization.” 511 F.Supp. at 920.
. Valentino does not challenge on appeal this delineation of the class.
Subsequent to the class certification, new designations replaced the PES pay scale. See 511 F.Supp. at 919. The changes in the pay scale system are not material to the issues before us.
. If a class action succeeds, however, unnamed class members may qualify for relief even if they failed to file administrative charges. Zipes, - U.S. at -, 102 S.Ct. at 1135; Franks v. Bowman Transp. Co., 424 U.S. 747, 771, 96 S.Ct. 1251, 1267, 47 L.Ed.2d 444 (1976); Albemarle Paper Co. v. Moody, 422 U.S. 405, 414 n.8, 95 S.Ct. 2362, 2370 n.8, 45 L.Ed.2d 280 (1976).
. The district court treated as fully relevant all proof relating to the EVS from its inception, even though Valentino did not complain to the Equal Employment Opportunity Counselor until the system had been in operation some four months. This was appropriate because, had Valentino proved discrimination during the charge-filing period, a continuing violation back to the inauguration of the EVS might well have been shown.
. While employment practices in the time-barred period may serve as relevant background evidence, see Hazelwood, 433 U.S. at 309 n.15, 97 S.Ct. at 2742 n.15; Evans, 431 U.S. at 558, 97 S.Ct. at 1889, given the changes wrought by the EVS, and plaintiffs uncertain qualification to represent those who may have suffered the consequences of discrimination prior to 1976, we do not believe the district court erred in concentrating on the EVS evidence.
. The district court indicated some hesitancy concerning Valentino’s qualification to represent the class, see supra note 5 and accompanying text, and with respect to certification of a class embracing positions that range so widely. In East Texas Motor Freight Sys. Inc. v. Rodriguez, 431 U.S. 395, 405-06, 97 S.Ct. 1891, 1897-1898, 52 L.Ed.2d 453 (1977), the Supreme Court cautioned:
We are not unaware that suits alleging racial or ethnic discrimination are often by their very nature class suits, involving class-wide wrongs. Common questions of law or fact are typically present. But careful attention to the requirements of Fed.Rule Civ. Proc. 23 remains nonetheless indispensable. The mere fact that a complaint alleges racial or ethnic discrimination does not in itself ensure that the party who has brought the lawsuit will be an adequate representative of those who may have been the real victims of that discrimination.
Cf. Wilkins, 654 F.2d at 409 n.37 (“[W]hile we ... do not decide the question, it may be that claims of [discrimination] with respect to ... positions requiring diverse and specialized qualifications relating to education and work experience simply are inappropriate claims for class actions that rely on statistical proof.”).
We note that class certification ordered at the initial stage of litigation may be withdrawn, altered or amended when the merits of the case unfold. See Fed.R.Civ.P. 23(c); Vuyanich v. Republic Nat’l Bank, 505 F.Supp. 224, 233 (N.D.Tex.1980) (court has continuing responsibility to reevaluate class status as case progresses). Accord Social Services Union, Local 535 v. County of Santa Clara, 609 F.2d 944, 948-49 (9th Cir. 1979); Guerine v. J. & W. Investment, Inc., 544 F.2d 863, 864 (5th Cir. 1977). When the nature of the proof and the number and diversity of the occupations involved in this case became apparent, the district court might appropriately have revisited the certification to determine whether further trimming, refinement by subdivision, or even retraction was warranted.
. The discussion that follows regarding the respective qualifications of the groups compared at the liability stage of a Title VII class action should not be confused with the question of each class member’s individual qualifications, an inquiry relevant only at the relief stage once a pattern and practice of class discrimination has been established. At that juncture, individual class members need show only that they applied for or, absent discrimination, would have sought the job or promotion at issue. The burden then rests on the employer to demonstrate that the individual was not qualified for the job or, for other lawful reasons, is not entitled to relief. See International Bhd. of Teamsters v. United States, 431 U.S. 324, 357-62, 97 S.Ct. 1843, 1865-1868, 52 L.Ed.2d 396 (1977); Franks v. Bowman Transp. Co., 424 U.S. 747, 772-73, 96 S.Ct. 1251, 1267-1268, 47 L.Ed.2d 444 (1976).
. As stated in recent commentary:
To a large degree, the law of equal employment opportunity has developed in a blue collar context. Case law fashioned to deal with the problems of providing equal employment opportunity for employees who work with their hands rather than with people, paper, or ideas cannot be applied without alteration or adjustment to employment practices at the white collar and professional levels.
Waintroob, supra, 21 Wm. & Mary L.Rev. at 46.
. Under the McDonnell Douglas formula a plaintiff in a Title VII sex discrimination case would meet the initial burden of establishing a prima facie case if she showed that she “applied and was qualified for a job for which the employer was seeking applicants,” that “despite [her] qualifications, [she] was rejected,” and that, “after [her] rejection, the position remained open and the employer continued to seek applicants from persons of [her] qualifications.” 411 U.S. at 802, 93 S.Ct. at 1824 (footnote omitted).
. We confine our analysis to high level professional, administrative, and managerial employment. In other settings, plant jobs, for example, comparison of the race/sex/ethnic composition of the employer’s entry level work force with the composition of its craft and production supervisor corps may well be appropriate. See B. Schlei & P. Grossman, supra, at 324 & nn. 92, 93 (citing cases).
. See also Wilkins, 654 F.2d at 408 (while class action plaintiffs convincingly demonstrated women were absent from top level jobs in university’s professional and administrative hierarchy, their statistics were not probative of their claim that the university discriminated against women by not hiring them for the higher level positions — “where special qualifications are required for a particular job, plaintiffs asserting a [discrimination] claim must demonstrate an availability of qualified class members to give their statistical showing meaning”); Pack, 566 F.2d at 1113 (statistical evidence in individual disparate treatment claim failed to establish a prima facie case of sex discrimination, despite showing that “approximately 98% of the employees at grades GS-11 and above were males, whereas 95% of the lower-grade professionals were females,” where evidence did not “demonstrate that the lower-grade professional women were qualified to occupy the higher positions”).
. Compare Hazelwood (Government “pattern or practice” suit alleging school district discrimination in teacher hiring), and Davis (individual sex discrimination complaint concerning promotion opportunities for research chemists), with Wilkins, 654 F.2d at 398 n.13 (sex discrimination class complaint against university involving diverse professional and administrative job categories), and Ste. Marie, 650 F.2d at 400-01 (secretarial jobs did not provide training that could lead to advancement in other work; class action plaintiff’s statistics were “hopelessly flawed” by lumping secretarial jobs together with positions that offered incumbents an opportunity to acquire the skills and experience that would enable them to qualify for promotion to technical and managerial posts).
. These exhibits show that the average wage of a female employee is lower than that of a male, J.A. 115-16, that women are disproportionately clustered at lower levels, J.A. 118-24, 126-34a, that the probability is low that a woman will have as high a grade as a man, J.A. 125, 134, that women are disproportionately hired into lower levels, J.A. 142 — 43, 145-46, and that the probability is high that a woman will be hired at a grade level lower than the level at which a man is hired. J.A. 147.
. The comparisons for January 1976 to January 1977, January 1977 to January 1978, and January 1978 to January 1979 show: (1) for all three years, in grades 23 to 25 and 26 to 28, a greater percentage of the women than of the men received increases; (2) the percentage of women at levels 17 to 19 who received increases was lower than the percentage of men in 1976 and 1978 but the difference in 1977 was negligible; (3) the difference in levels 20 to 22 was negligible in 1976 but the percentage of women who received increases in 1977 and 1978 was lower than that of men; and (4) at levels 29 to 31, in 1976 and 1977 women received no increases, whereas a small percentage of men did, while in 1978, one woman out of three received an increase, which made the percentage of women receiving increases higher than the percentage of men. J.A. 207-09.
. Valentino also included “special degree” as a variable, but this factor did not necessarily indicate, as its appellation might suggest, whether the degree was in business, law, engineering, or other specialized field. Rather, the “special degree” factor depended solely on whether an employee had at some time indicated his or her specialty on a personnel form. A college history major who put “history” on the form would be recorded as holding a “special degree” while one who did not fill in his or her major would not be recorded as a “special degree” holder. The district court properly dismissed the “special degree” factor as “nonsense.” 511 F.Supp. at 943 n.20.
. Although significant change Is occurring, men continue to receive a large majority of degrees in law, medicine, engineering, mathematics, and sciences, while women receive most of the degrees in education, the humanities, sociology, nursing, and library sciences. In 1978, men received the following percentages of bachelor’s degrees conferred in the United States with specialization in the following subjects: engineering, 93%; physical sciences, 78.7%; economics, 75.1%; computer and information sciences, 73.6%; business and management, 72.8%; biological sciences, 61.6%; mathematical subjects, 58.7%. On the other hand, the following percentages of bachelor’s degrees with specialization in the following subjects were conferred on women: nursing, 94.7%; library sciences, 85.7%; foreign languages, 75.6%; education, 72.4%; sociology, 63.6%; fine and applied arts, 62%; letters, 57.1%. At the graduate level, in 1978, men received 94.8% of all master’s degrees in engineering, 78.5% of all medical degrees, and 74% of all law degrees. Bureau of the Census, U. S. Dep’t of Commerce, Statistical Abstract of the United States 174-75 (1980) (percentages presented above are approximations from raw figures).
. Valentino’s statistical analyst had the occupational codes for USPS Headquarters employees at level 17 and above; he did not explain why he did not pursue analyses utilizing them. J.A. 827.
. We reiterate that the qualifications a Title VII plaintiff must grapple with in the first instance are threshold or “minimum objective” qualifications. Davis, 613 F.2d at 964. Subjective qualifications which, in any event, may be impossible to quantify, are appropriately introduced on rebuttal. See Wilkins, 654 F.2d at 401. Threshold qualifications are those prerequisite to consideration for a position (e.g., a law degree), not attributes of relative advantage or disadvantage (e.g., reputation of law school attended, service on a law journal). See D. Bal-dus & J. Cole, Statistical Proof of Discrimination 182-83 (1980).
. Valentino’s regressions also ignore the possible relevance of nongovernment experience to an employee’s salary. Not until 1975, when the Postmaster General directed that all vacancies be filled from within, see supra p. 63, did outside hiring by the USPS stop. 511 F.Supp. at 921. The regressions, in essence, treat as similarly qualified an engineer with 20 years of experience, two of which were spent in the government, and an engineer with only two years total experience, both spent in the government.
. We have strong reservations concerning some of the other views expressed by the district court on the infirmities of Valentino’s regressions. We do not consider her regressions flawed by failure to include grade level as an independent variable. See Trout, 517 F.Supp. at 886 n.47; cf. infra note 30. Nor are we prepared to say that failure to factor out time-barred discrimination discredited the analyses. Statistics tuned to the proper time period are more probative than statistics not so tuned, but categorical rejection of the latter is not warranted. See EEOC v. American Nat’l Bank, 652 F.2d 1176, 1188-89 (4th Cir. 1981); Movement for Opportunity and Equality v. General Motors Corp., 622 F.2d 1235, 1244-45 (7th Cir. 1980); Vuyanich, 505 F.Supp. at 360; B. Schlei & P. Grossman, supra, at 326-27. See also Hazelwood, 433 U.S. at 308 n.13, 97 S.Ct. at 2742 n.13.
. Proof that discrimination exists within occupational categories may well support an inference of workplace-wide discrimination. Our sole point is that, in the first instance, comparisons must hone in on similarly qualified employees.
. The district court noted too that “[t]he USPS refused to comply with plaintiffs demand that it structure a major realignment so as to bring it within the reach of the Veterans’ Preference Act,” and that “[i]t did so, at least in part, to protect its female employees from the adverse impact which Veterans’ Preference would have had on their employment status.” 511 F.Supp. at 951; see J.A. 728-29.
. Valentino also cites as evidence of discrimination critiques of the USPS’s treatment of female employees written by the Civil Service Commission and by the USPS itself. J.A. 265-366. We have examined these reports, which are not discussed in the district court’s opinion. The most severely critical report, the July 1979 Civil Service Commission paper on the Postal Service’s equal employment opportunity program was not confined to Headquarters; it reviewed the entire Postal Service. None of the reports focused on level 17 and above. The reports reveal USPS reluctance to implement an affirmative action program for women but do not serve as reliable indicators of discrimination proscribed by Title VII in upper echelon jobs at USPS Headquarters. The acknowledgement that women are underrepresented in middle and upper management levels cannot be translated into a confession that qualified women have been denied the opportunities to advance accorded men.
. While we do not reach the question whether, had Valentino proved a prima facie case, USPS would nonetheless prevail based on its rebuttal, we note here our concerns about the district court’s hardy endorsement of USPS’s statistical evidence, specifically, the USPS EVS data, regression and cohort analyses.
The EVS data showed that when men and women apply for the same position, women are statistically more likely to be certified than men. 511 F.Supp. at 937-38. But the male applicants who were compared to the female applicants may have submitted multiple applications, they may have been selected for a job for which no women applied.
*73The district court considered it appropriate to regress salary on grade level in a suit alleging discrimination in promotion. We disagree. Grade level, in a case such as this one, is an “inappropriate variable.” See Finkelstein, supra, 80 Colum.L.Rev. at 738-2. Absent clear, affirmative evidence that promotions were made in accordance with neutral, objective standards consistently applied, there is no assurance that level or rank is an appropriate explanatory variable, untainted by discrimination. Id at 742.
Finally, the cohort analysis, which included several groups of less than ten people, had many of the weaknesses cogently identified in Trout, 517 F.Supp. at 884-86.
. From January 1976 to October 1979, 831 vacancy announcements resulted in promotions to jobs at level 17 or higher. 579 of the files in question were complete, 86 were incomplete, and 166 had been destroyed pursuant to routine USPS record destruction procedures. Valentino contends the “spoliation” doctrine should have been invoked against USPS for its destruction of relevant EVS files. See Boyd v. Ozark Air Lines, Inc., 568 F.2d 50, 53 (8th Cir. 1977); Vick v. Texas Employment Comm’n, 514 F.2d 734, 737 (5th Cir. 1975); United States v. Roelof Constr. Co., 418 F.2d 1328, 1331 (9th Cir. 1969). However, the circumstances of the destruction here provide no basis for attributing bad faith to USPS. See Vick, 514 F.2d at 737. Until the June 13, 1978, certification of a class the district court considered appropriate and manageable, USPS had no clear indication of its obligation regarding record preservation. (Valentino’s initial delineation of the class to encompass all past, present, and future female USPS employees and applicants for employment addressed a work force of over 650,000.) Upon certification of the class, USPS immediately acted to preserve records relating to the class claim. We find no reason to disagree with, the district court’s disposition of this issue. See 511 F.Supp. at 938-39, 953-54. EEOC v. American Nat’l Bank, 652 F.2d at 1195-96, cited by Valentino, holds that an employer who destroys records cannot use the lack of records to lessen its burden on rebuttal; it does not address the relationship between an employer’s routine destruction of records no longer in use and the plaintiff’s burden.