33 Fair empl.prac.cas. 195, 32 Empl. Prac. Dec. P 33,856 Paul Lilly, Individually and on Behalf of All Others Similarly Situated, Christopher McKinney John Legrand, Ken Bailey, Frank Sullivan, James Mobley, Jerome Gary, Curtis Jones, Roy Torrence, John Johnson, Willie Hunt, Roosevelt Patterson, Willie Covington, Michael McVay Richard Gregory, Individually and on Behalf of All Others Similarly Situated, Edward Porter, Individually and on Behalf of All Others Similarly Situated, and Phillip Reed, Shirley Gatewood, Woodrow McManus Hazel Fisher, William Carrothers, Tresevant Goodwin, Richard Burch, Mack Ervin, Barbara Anderson, Austin Pharr, Therrell McMoore v. Harris-Teeter Supermarket, a Corporation

720 F.2d 326 33 Fair Empl.Prac.Cas. 195, 32 Empl. Prac. Dec. P 33,856 Paul LILLY, individually and on behalf of all others similarly situated, Christopher McKinney, John LeGrand, Ken Bailey, Frank Sullivan, James Mobley, Jerome Gary, Curtis Jones, Roy Torrence, John Johnson, Willie Hunt, Roosevelt Patterson, Willie Covington, Michael McVay, Richard Gregory, individually and on behalf of all others similarly situated, Edward Porter, individually and on behalf of all others similarly situated, Appellees, and Phillip Reed, Shirley Gatewood, Woodrow McManus, Hazel Fisher, William Carrothers, Tresevant Goodwin, Richard Burch, Mack Ervin, Barbara Anderson, Austin Pharr, Therrell McMoore, Plaintiffs, v. HARRIS-TEETER SUPERMARKET, a corporation, Appellant. No. 82-1831. United States Court of Appeals, Fourth Circuit. Argued April 12, 1983. Decided Oct. 14, 1983. John O. Pollard, Charlotte, N.C. (Richard F. Kane, Blakeney, Alexander & Machen, Charlotte, N.C., on brief), for appellant. Michael A. Sheely, Charlotte, N.C. (Sheely & Blum, Charlotte, N.C., on brief), for appellees. Before WINTER, Chief Judge, SPROUSE, Circuit Judge, and KELLAM,* District Judge. HARRISON L. WINTER, Chief Judge: — 1 — In this class action charging racial discrimination in employment, Harris-Teeter Super Markets, Inc., (Harris-Teeter) appeals from a judgment of the district court entered upon findings that it had engaged in a pattern or practice of racial discrimination in terminations and promotions at its Mecklenburg County, North Carolina warehouse and retail stores, and that it had been guilty of specific instances of racial discrimination against thirteen of the named individual employees. We conclude that the suit was properly certified as a class action with respect to both promotions and terminations, and that various individual plaintiffs were properly permitted to intervene. We affirm the finding of a pattern or practice of racial discrimination in terminations, but reverse the finding of a pattern or practice of discrimination in promotions. We affirm the judgments of discrimination as to three named plaintiffs, but remand for further proceedings as to the individual plaintiffs alleging discrimination in promotions. I. — 2 — Harris-Teeter is a retail grocery chain with offices, retail stores, and a warehouse located in Mecklenburg County. In January 1975, Harris-Teeter discharged Paul Lilly, a black warehouse employee. Lilly filed a charge with the Equal Employment Opportunity Commission (EEOC) alleging that he had been discharged for protesting racially discriminatory treatment on the job and that Harris-Teeter was engaged in systematic discrimination against blacks in its termination practices. Lilly later amended his EEOC charge to include allegations of systematic discrimination against blacks in hiring, promotions, personnel procedures, terminations, and supervisory practices. Lilly was subsequently issued a right-to-sue letter by EEOC and, in June 1976, he commenced the present action. — 3 — Lilly's complaint, brought under both 42 U.S.C. Secs. 1981 and 2000e et seq., alleged that he had been terminated because of his race and in retaliation for protesting racial discrimination. He also claimed to represent a class of black Harris-Teeter employees who, from July 1974 on, had suffered racial discrimination in hiring, promotions, interviewing, terminations, supervision, and discipline. In support of his subsequent motion for class certification, Lilly presented statistical evidence purporting to show disparities in job classification and cited specific examples of alleged discrimination in promotions and terminations. The district court thereupon tentatively certified the class as requested. — 4 — In April 1979, twenty black Harris-Teeter employees filed a joint motion for permission to intervene as plaintiffs, in order to present their individual claims of racial discrimination. The claims of these individuals encompassed alleged discrimination in terminations, promotions, job placement, pay, and transfers. Although none of these individuals had exhausted their administrative (EEOC) remedies, the district court granted the motion for joint intervention. In addition, the district court consolidated the case with those of two other black Harris-Teeter employees, Gregory and Porter, who were alleging racial discrimination in demotions and hiring, respectively. Gregory and Porter had exhausted their EEOC remedies prior to filing their individual suits. Subsequently, the district court certified all twenty intervenors and Gregory as class representatives. — 5 — A bench trial was held from January through March, 1980. In August, 1980, the district court, 503 F.Supp. 29, filed a "Memorandum of Decision," in which the court found a generalized pattern of racial discrimination in hiring, promotions, and terminations. As practices that effectuated this discrimination, the court found that there were no written job descriptions, promotion criteria, job evaluations, or list of employees wanting to change jobs, that job openings were not posted, that the company rule that limited eligibility for promotions to employees in the same department and shift as the opening was strictly enforced as to blacks but not as to whites, that refusals by blacks to take on jobs were carefully remembered, although they were never asked if they had changed their minds, and that claimed previous job experience of blacks was either forgotten or disbelieved, while that of whites was credited and utilized. The district court also found that the statistical evidence broadly supported the court's conclusion from the individualized testimony that blacks had less chance of getting good jobs and promotions than did whites. The district court, in addition, found that individual instances of discrimination had been demonstrated as to fourteen of the named plaintiffs, but not as to the other ten. Finally, the district court instructed plaintiffs' counsel to prepare proposed findings of fact and conclusions of law as to the class-wide issues and as to the prevailing individuals; defense counsel were instructed to do likewise for those individuals found not to be entitled to relief.1 — 6 — Plaintiffs' counsel subsequently submitted lengthy proposed findings of fact and conclusions of law; the district court requested that they be condensed and invited defense counsel to comment on the proposed opinion. Defendant chose not to do so. Plaintiffs then submitted a revised draft. The district court returned this draft to plaintiffs in marked-up form, instructing them to incorporate various editorial changes and to reverse the previous finding of discrimination as to one of the intervenors. Although this was the only major substantive change, notations on the marked-up draft show that the district court checked the proposed findings of fact against the testimony in the record and approved each paragraph individually. The district court, in returning the draft, again requested that defendant submit any comments on or objections to the proposed opinion. Defendant once again chose not to do so. The proposed opinion, as amended, was then adopted by the district court, 545 F.Supp. 686. — 7 — In that final opinion, the district court first made permanent its class certification order. It found that the claims of persons alleging discrimination in promotions and terminations were sufficiently similar to justify single-class treatment.2 The district court excluded from the certified class, however, employees at the company's offices and employees alleging discrimination in hiring for lack of proper class representatives. As to the merits of the terminations claim, the court found that the supervisory force was overwhelmingly white and exercised virtually unlimited discretion, in that the only written statement of Harris-Teeter's termination policy was a list of "don'ts" whose violation might or might not be grounds for an indeterminate amount of discipline. The district court then found that both the statistical data, which showed that the actual number of blacks involuntarily terminated for cause from 1974 through 1978 was almost ten standard deviations above what would be expected from their proportion of the workforce, and the testimony at trial, which revealed several instances in which blacks had been terminated while equally or even more serious misbehavior by whites had been tolerated, demonstrated that this unlimited discretion had been exercised in a racially discriminatory manner. Finding no satisfactory explanation from Harris-Teeter for the statistical disparities and crediting the direct evidence of discriminatory treatment, the district court concluded that Harris-Teeter had engaged in a pattern or practice of racial discrimination in its terminations policy. — 8 — As to the merits of the promotions claim, the district court again found that the overwhelmingly white supervisory force had wielded unlimited discretion, in that there were no written job descriptions or regular system of job performance evaluation, job posting was begun only in 1979, and the promotions criteria (such as "attitude" and "initiative") were vague and subjective. The district court then found that both the statistical evidence, which showed that the black promotion rate was less than 70 percent of the white promotion rate, and the direct testimony revealed that this discretion was discriminatorily exercised. The court rejected Harris-Teeter's explanations, ruling that the company's "same department/same shift" policy had been applied only to blacks, that only the refusal by blacks to take certain jobs was remembered and held against them, although they were never asked if they had changed their minds, and that previous job experience of blacks was ignored or disbelieved while that of whites was utilized. In addition, the district court found, on the basis of statistical data, that the number of blacks hired by Harris-Teeter was some five to seven standard deviations below what would be expected from their proportion of the available pool of applicants, and that this evidence of discrimination in hiring supported the conclusion that Harris-Teeter discriminated in promotions. On the basis of all of this evidence, the district court concluded that Harris-Teeter had engaged in a pattern or practice of racial discrimination in its promotions policy. — 9 — The district court then proceeded to examine the individual discrimination claims of the named plaintiffs. In each case the court applied the analytic framework specified in Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 101 S.Ct. 1089, 67 L.Ed.2d 207 (1981).3 The court found that Harris-Teeter had discriminated on the basis of race as to Porter (hiring), Reed, Mobley, McKinney, Torrence, Patterson, Jones, Sullivan, Bailey, and LeGrand (promotions), Gregory (demotion), Gary (promotion and transfer), and Lilly (termination). The district court then entered judgment, ordering Harris-Teeter to reinstate or offer promotions to the prevailing named plaintiffs and to establish less discretionary systems of promotions and terminations, referring the case to a special master for the determination of back pay for the named plaintiffs and of appropriate relief for the class members, and ruling that plaintiffs' counsel was entitled to an award of attorneys' fees. As to this final point, the district court thereupon entered an award of approximately $86,000 in interim attorneys' fees. — 10 — Harris-Teeter now appeals from the findings of liability on the issues of class-wide discrimination as to both promotions and terminations and individual discrimination as to each named plaintiff and from the amount of the interim attorneys' fees award. II. — 11 — Harris-Teeter first argues that the district court failed to meet its obligation under Federal Rule of Civil Procedure 52(a) to "find the facts specially and state separately its conclusions of law thereon" when it adopted essentially verbatim the proposed opinion prepared by plaintiffs' counsel. We cannot accept this disingenuous argument. It is true, of course, that we repeatedly have cautioned district courts against the practice of simply adopting findings prepared ex parte by one of the litigants. See, e.g., EEOC v. Federal Reserve Bank of Richmond, 698 F.2d 633, 639-41 (4 Cir.1983) (citing cases). As we stated in Federal Reserve, "the findings must be based on something more than a one-sided presentation of the evidence, ... [because] finding facts under Rule 52(a) requires the exercise by an impartial tribunal of its function of weighing and appraising evidence offered, not by one party to the controversy alone, but by both." Id. at 640. In the present case, however, the district court twice specifically requested that defense counsel submit comments on, and objections to, the findings of fact and conclusions of law submitted by the plaintiffs, and defendant never responded. That the presentation was one-sided is therefore attributable not to the district court's procedure, but rather to defendant's own choice. We note, in addition, that the district court reversed its initial finding as to one intervenor during its review of the proposed opinion and that it checked the cited evidence in the proposed findings against the actual trial transcript, approving each paragraph of the findings one by one. Under these circumstances, we do not think that defendant's argument that the district court failed to meet its responsibilities under Rule 52(a) is well taken. — 12 — As a matter of general practice, however, we do not approve of the issuance by the district court of a memorandum of its ultimate findings in sweeping terms followed by a request to the parties to prepare detailed findings as to the issues upon which each side prevailed. The proper practice is for the district court, prior to reaching and announcing any decision, to request proposed findings from both parties as to all of the disputed factual and legal issues, preferably with references to the record supporting the fact requested to be found, and then to prepare its decision based upon its analysis of these proposed findings and the evidence of record. By utilizing this procedure, the district court can be certain of obtaining--and not merely of having requested--both parties' views on each disputed issue and their citation to evidence on both sides of each factual question. In all future cases in which the district court concludes to request proposed findings from counsel, this latter method should be followed. III. — 13 — Harris-Teeter next argues that the district court erred in certifying this case as a class action with respect to both terminations and promotions.4 As to the former, Harris-Teeter contends that although Lilly would have been an adequate representative for a class of discharged black employees who similarly, in connection with their discharges, suffered racial discrimination violative of both Sec. 1981 and Title VII, the existence of such a class was never demonstrated. In particular, Harris-Teeter argues that Lilly never showed any facially neutral termination policy that had a "disparate impact" on blacks and that, with respect to possible "disparate treatment," Lilly alleged only a few isolated instances of discriminatory treatment. — 14 — The prerequisites to class certification under both Title VII and Sec. 1981 are those generally applicable under Federal Rule of Civil Procedure 23(a). See General Telephone Co. v. Falcon, 457 U.S. 147, 102 S.Ct. 2364, 72 L.Ed.2d 740 (1982) (Title VII); Stastny v. Southern Bell Tel. & Tel. Co., 628 F.2d 267 (4 Cir.1980) (Title VII); Gonzalez v. Southern Methodist University, 536 F.2d 1071, 1072-73 (5 Cir.1976) (Sec. 1981), cert. denied, 430 U.S. 987, 97 S.Ct. 1688, 52 L.Ed.2d 383 (1977). Therefore, plaintiffs need not, at the time of the motion for class certification, demonstrate by statistical evidence that blacks have been terminated at a higher rate than have whites, or any other differential. Certification is only concerned with the commonality (not the apparent merit) of the claims and the existence of a sufficiently numerous group of persons who may assert those claims (and, as were conceded here, the adequacy of Lilly and typicality of his claim). As to the commonality requirement, the complaint plainly alleged a practice of disparate treatment in the exercise of unbridled discretion, thus raising questions of law and fact common to all discharged black employees. And as to the numerosity requirement, the statistics at trial5 revealed that 229 black employees were involuntarily terminated from 1974 through 1978, easily enough to demonstrate the existence of a class. We conclude, therefore, that the class with respect to terminations, under both Sec. 1981 and Sec. 2000e, was properly certified. — 15 — As to the propriety of certifying that class to include persons discriminated against with respect to promotions, Harris-Teeter contends that these claims should not have been included for lack of a proper class representative.6 Lilly, although limiting his personal claim to alleged discrimination in his termination, claimed in his complaint to represent all black Harris-Teeter employees who suffered racial discrimination with respect to promotions as well, and the district court certified him as such. — 16 — Determination of whether Lilly's claim is "typical" of the unpromoted employees' claims is guided by the Supreme Court's recent decision in General Telephone Co. v. Falcon, 457 U.S. 147, 102 S.Ct. 2364, 72 L.Ed.2d 740 (1982), announced subsequent to the district court's decision. In Falcon, the Court rejected the Fifth Circuit's "across the board" rule that an employee who suffers racial discrimination in any employment practice automatically may maintain a Title VII class action for all company employees as to all discriminatory employment practices. But the Court also rejected those cases that had interpreted its decision in East Texas Motor Freight System, Inc. v. Rodriguez, 431 U.S. 395, 97 S.Ct. 1891, 52 L.Ed.2d 453 (1977), as requiring that the class representative necessarily have suffered discrimination in precisely the same employment practice as did all the other members of the class.7 Instead, the Court reiterated that each case must turn, based upon its particular facts, on whether the individual Title VII claim of the named plaintiff and the claims of the Title VII class he purports to represent raise common questions of law or fact.8 — 17 — When Falcon is applied here,9 it is apparent that Lilly's individual (and class-wide) terminations claim and the promotions claims overlapped on several important pieces of proof: the absence of written, objective criteria for promotions and terminations, the virtual absence of regular job performance evaluations, and the presence of an overwhelmingly white supervisory force.10 The Supreme Court recognized in Falcon that these conditions could justify single class treatment for more than one employment practice: — 18 — Significant proof that an employer operated under a general policy of discrimination conceivably could justify a class of both applicants and employees if the discrimination manifested itself in hiring and promotion practices in the same general fashion, such as through entirely subjective decisionmaking processes. — 19 — 457 U.S. at 159 n. 15, 102 S.Ct. at 2371 n. 15. But plaintiffs' class-wide promotion claim here deviated from a claim of unfettered discretion discriminatorily exercised because of the allegations that defendant put substantial weight on prior work experience and that defendant treated blacks and whites differently in terms of believing and utilizing that prior work experience. There was also an indication that the discretion in promotions was further limited by a "same department/same shift" policy, which was alleged to have been enforced only as to blacks.11 These charges raised significant issues of proof separate from those evidentiary issues implicated by the terminations claim. Thus, although the question is a very close one, we conclude that the requirements of Rule 23(a) were not met and that Lilly was not a proper representative for those persons discriminated against with respect to promotions under either Sec. 1981 or Title VII. — 20 — Plaintiffs argue, however, that even if Lilly could not represent a class including promotions, eleven of the individuals whose petitions to intervene were granted by the district court were proper representatives for a promotions class. These eleven individuals all advanced individual claims of discrimination in promotions and were certified as class representatives by the district court; eight ultimately prevailed on their promotion claims. In support of this position, plaintiffs cite Muskelly v. Warner & Swasey Co., 653 F.2d 112 (4 Cir.1981), which approved the intervention of an individual claiming discrimination in hiring into an across-the-board class action brought by another individual whose personal claim involved the employer's allegedly wrongful failure to promote him, in order for the former to become the proper representative of the "hiring" class and to assert his own hiring claim. In response, Harris-Teeter argues that because these individuals never exhausted their EEOC remedies, the district court never should have let them intervene in the first place.12 — 21 — With respect to this question, we deem it necessary to separate the Title VII claims from the Sec. 1981 claims. As to the promotions claims brought under 42 U.S.C. Sec. 1981, exhaustion of EEOC remedies is not a prerequisite to filing suit. See Johnson v. Railway Express Agency, Inc., 421 U.S. 454, 460-61, 95 S.Ct. 1716, 1720-21, 44 L.Ed.2d 295 (1975). We therefore conclude that the intervenors were properly permitted to intervene with respect to the Sec. 1981 promotions claims, and that they were properly certified as class representatives for that claim.13 — 22 — We turn next to the propriety of the district court's order permitting the individuals to intervene, prior to exhausting their EEOC remedies, in order to present individual claims of, and to represent a class asserting claims of, discrimination in promotions in violation of Title VII. The district court's order permitting such intervention was based on the rule established in Oatis v. Crown Zellerbach Corp., 398 F.2d 496, 498 (5 Cir.1968), that class members may intervene in a Title VII suit without having exhausted their EEOC remedies. But because we have concluded that the class represented by Lilly should not have been certified to include promotions, these individuals were not class members at the time they sought to intervene. Thus, their intervention cannot be premised on the rule from Oatis relied upon by the district court. — 23 — Alternatively, it is suggested that intervention to assert promotion claims under Title VII, without exhaustion, was proper under the rule from Foster v. Gueory, 655 F.2d 1319, 1321-23 (D.C.Cir.1981). Foster held that individuals who are not class members may intervene in a Title VII suit without exhausting EEOC remedies if their claims are so close to the named plaintiff's claim that no further purpose would be served by requiring exhaustion.14 The question, said the Foster court, is whether the — 24 — two claims are so similar that it can fairly be said that no further conciliatory purpose would be served by filing separate EEOC charges [or whether] the two complaints differ to the extent that there is a real possibility that one of the claims might be administratively settled while the other can be resolved only in the courts. Id. at 1322.15 — 25 — In the present case, Lilly's exhausted EEOC charge alleged discrimination by Harris-Teeter in promotions as well as in terminations, hiring, personnel procedures, and supervisory practices. Gregory and Porter, whose cases were consolidated with the class action, had exhausted their EEOC remedies with respect to their claims of discrimination in demotions and hiring. Under these circumstances--with EEOC remedies as to termination, demotions, and hiring claims exhausted, with Lilly's EEOC charge including promotions, and with no indication that any discrimination claims against the company could be settled out of court--we think that exhaustion by the intervenors of EEOC remedies with respect to promotions would have been futile. Although the claims are not precisely the same and cannot, as we have indicated, comprise a single class, we think that they are sufficiently similar so that it fairly can be said that the failure successfully to conciliate the three exhausted claims foreshadows an inevitable lack of success in conciliating the promotions claim. We therefore conclude that exhaustion of EEOC remedies was excused for the intervenors and that their intervention was proper. Since eleven of these intervenors were proper Title VII class representatives for a class of persons discriminated against in regard to promotions, we conclude that the Title VII promotions claim ultimately was properly certified as a class action.16 IV. — 26 — Harris-Teeter's next challenge is to the sufficiency of the evidence of a pattern or practice of racial discrimination with regard to both terminations and promotions. As to the former, plaintiffs' statistical evidence showed that blacks comprised, on average, 15.7 percent of the Harris-Teeter workforce from 1974 through 1978, but represented over 28 percent of the involuntary terminations during that period. Statistical analysis of the data underlying this disparity revealed that the number of involuntary "for cause" terminations of blacks was, over this period,17 9.71 standard deviations (by the binomial model18) greater than would be expected on the basis of chance.19 Under the test discussed in EEOC v. American National Bank, 652 F.2d 1176, 1190-93 (4 Cir.1981) (citing Hazelwood School District v. United States, 433 U.S. 299, 311 n. 17, 97 S.Ct. 2736, 2743 n. 17, 53 L.Ed.2d 768 (1977), and Castaneda v. Partida, 430 U.S. 482, 496 n. 17, 97 S.Ct. 1272, 1281 n. 17, 51 L.Ed.2d 498 (1977)), cert. denied, --- U.S. ----, 103 S.Ct. 235, 74 L.Ed.2d 186 (1982), because the number of standard deviations was more than two or three, this statistical evidence conclusively ruled out chance as the cause of the disparity in the termination rates.20 — 27 — NOTE: OPINION CONTAINS TABLE OR OTHER DATA THAT IS NOT VIEWABLE — 28 — In addition to this statistical evidence, the district court found that Harris-Teeter's terminations policy was highly discretionary21 and had in specific instances been applied unevenly to blacks as against whites. The district court credited testimony that whites received disciplinary warnings, if at all, only after extensive absences or tardiness, while blacks, including Lilly, were terminated for insubstantial reasons. We think that the strong statistical evidence, when combined with the specific instances in which the discretion was unevenly exercised, is sufficient to support the district court's finding that Harris-Teeter engaged in a pattern or practice of racially discriminatory treatment of blacks in connection with its terminations policy.22 — 29 — We turn next to the sufficiency of the proof of class-wide racial discrimination in promotions. The district court found that in 1976 the combined black promotion rate in the stores and warehouse was 69 percent of the white promotion rate, and that in the stores alone the black promotion rates for 1975 and 1976 were 70 percent and 67 percent, respectively, of the white promotion rates. The first problem with this data, however, is that its scope--covering the stores and warehouse for only 1976 and only the stores for 1975--is insufficient to prove discrimination from 1974 through 1978. Second, when additional, uncontested data from the record is added into the overall numbers, the black promotion rate rises to 86% of the white rate.23 Moreover, a statistical disparity cannot, standing alone, make out a prima facie case of class-wide discrimination unless a standard deviation analysis has ruled out chance as the cause of the disparity. When such an analysis is performed on the promotions data, it appears that the number of blacks promoted is fewer than two standard deviations below what would be expected from their proportion of the workforce; thus chance is not excluded as the cause of the disparity.24 — 30 — In an attempt to demonstrate that discriminatory intent rather than chance was the cause of the disparity in the promotion rates, plaintiffs submitted statistical evidence demonstrating that fewer blacks were hired by Harris-Teeter than would have been expected from their proportion of the applicant pool. Standard deviation analysis of this data conclusively ruled out chance as the cause of the disparity in the hiring rates, thus raising a presumption that discriminatory intent was the explanation for the hiring disparity. The district court accepted this as being probative of a similar intent behind the disparity in the promotions data. On the present record, however, we cannot accept this analysis. We agree that evidence of discriminatory intent in one employment context (e.g., hiring) may be probative of discriminatory intent in a different context (e.g., promotions) where it has been demonstrated that the same company managerial personnel were responsible for decisionmaking in both contexts. In the present case, however, the district court made no finding that the store supervisors responsible for promotions were also responsible for hiring.25 We therefore conclude that the inference of discriminatory intent with respect to hiring should not have been considered in determining whether discriminatory intent was the cause of the disparity in the promotions data. — 31 — The other factor relied upon by the district court to demonstrate an intentionally discriminatory pattern of racial discrimination was the testimony of the nine intervenors who prevailed on their promotion claims. But almost none of this testimony related to instances of direct racial discrimination, and there certainly were too few instances of direct discrimination from which any pattern or practice of such discrimination could be inferred. Instead, most of the testimony dealt with the absence of written, objective criteria for promotion decisions. As we have noted, supra note 21, the possession by management of unbridled discretion will tend to confirm implications of racial discrimination drawn from statistical disparities. But here the promotions data did not reveal any statistically significant disparity between the black and white promotion rates. Thus, because Title VII and Sec. 1981 prohibit racial discrimination, rather than simple arbitrariness or caprice, the mere existence of vague and subjective criteria is not alone proof of unlawful discrimination. The district court's finding of class-wide discrimination with respect to promotions therefore cannot stand. V. — 32 — Harris-Teeter next challenges the district court's findings as to each of the thirteen named plaintiffs who prevailed. We turn first to the ten plaintiffs--Mobley, Reed, Gary,26 McKinney, Torrence, Patterson, Jones, Sullivan, Bailey, and LeGrand--who alleged that they had been denied promotions because of racial discrimination. We conclude that, in light of our reversal of the district court's finding of class-wide discrimination in promotions, these cases must for two reasons be returned to the district court for further proceedings. First, in each instance, the district court appears to have grounded its conclusion, at least in part, on its earlier finding that the promotions system was, as a whole, discriminatorily applied as to blacks. Because the district court did not indicate explicitly whether this finding was a necessary, or merely a cumulative, piece of evidence, we must remand to the district court for a determination of whether any or all of the individual claims are made out in the absence of proof of class-wide discrimination. Second, the district court made no case-by-case findings as to the "subjective" criteria advanced by Harris-Teeter as the basis for its promotion decisions. As noted above, the district court may, where the statistical data reveals a disparity between the white and black promotion rates of sufficient magnitude to warrant an inference of discrimination, conclude that the subjective criteria do not overcome that inference. But the premise that the subjective criteria were mere pretexts for racial discrimination cannot stand as an across the board conclusion where, as here, the existence of class-wide discrimination is not demonstrated. Rather, the district court must, on a case-by-case basis, determine whether the subjective criteria advanced by Harris-Teeter were in fact utilized in making the individual decision, or whether the individual decision was based upon unlawful discrimination. — 33 — The district court next found that plaintiff Lilly had proven that his discharge from Harris-Teeter was based on racial discrimination. We agree. The evidence revealed that Lilly was discharged only two days after complaining about racial discrimination to Harris-Teeter's personnel director. Further, although Harris-Teeter claimed that Lilly's job performance in quantity checking had been deficient, it never introduced any written substantiation of these claims, and Lilly's supervisor admitted that he never checked as to whether the errors in counting might have been at the receiving end. Under these circumstances, the district court did not clearly err in rejecting Harris-Teeter's purported justification and in finding that Lilly had been the victim of racial discrimination.27 — 34 — Next, the district court found that Richard Gregory's demotion and subsequent constructive discharge were based upon racial discrimination. The district court credited testimony that Gregory was assigned duties not given to whites at his level, that his termination of a white employee who had called him a "nigger" was reversed by his superiors, and that his job performance had been praised. Further, the district court noted that although Gregory had been discharged by a black, that was done on orders by, and in the presence of, a white territorial supervisor. We thus conclude that the district court's finding was not clearly erroneous. — 35 — The final individual claim decided by the district court was that of Edward Porter, who challenged Harris-Teeter's decision not to hire him as a tractor-trailer driver. The evidence showed that whites with less experience than Porter were hired by Harris-Teeter after he applied for the position. Harris-Teeter contended that they were hired instead of Porter because they came to the company when openings were available, but the district court found that, in at least one instance, Harris-Teeter had filled a subsequent position with a white who had applied before Porter, thus revealing that Harris-Teeter maintained, and utilized, its list of applicants on file. Moreover, the company's asserted statistics as to the percentage of black drivers are meaningless in the absence of data as to the pool of applicants for the positions. The district court's conclusion as to Porter is therefore also not clearly erroneous and will be sustained. VI. — 36 — Harris-Teeter's final contention challenges the amount of the attorneys' fees awarded to plaintiffs' counsel. Under our opinion in Anderson v. Morris, 658 F.2d 246 (4 Cir.1981), one of the factors to be considered in determining the amount of the award is the result ultimately obtained. Because we have concluded that the finding of class-wide discrimination in promotions must be reversed, and that the individual promotions claims must be considered further, we vacate the award of attorneys' fees and remand this issue to the district court for a determination of what effect, if any, these changes in the ultimate result of the case should have upon the amount of the award.28 — 37 — AFFIRMED IN PART; REVERSED IN PART; AND REMANDED. Footnotes [-s] * Honorable Richard B. Kellam, Senior United States District Judge for the Eastern District of Virginia, sitting by designation [1] Defendant's proposed opinion as to the ten unsuccessful plaintiffs was subsequently adopted by the district court and is not contested on appeal [2] In its judgment entered July 12, 1982, the district court certified the class as: Plaintiffs Lilly, Gregory, Reed, Mobley, Gary, McKinney, Torrence, Patterson, Jones, Johnson, Gatewood, Sullivan, Bailey and LeGrand; and all blacks currently employed by Harris-Teeter at its warehouse and store numbers 1, 2, 5, 8, 9, 22, 39, 52, 53, 55, 59, 62, 66, and 83, located in Mecklenburg County; and all blacks employed by Harris-Teeter at [said warehouse and stores] at any time since July 20, 1974 who are not currently employed, all of whom may have been or may be subjected to racial discrimination by Harris-Teeter in its employment policies and practices in reference to terminations and promotions. [3] Under Burdine, plaintiff must first establish a prima facie case of discriminatory treatment, defendant must then articulate a legitimate, non-racial reason for its practices, next the plaintiff must be given the opportunity to prove that the defendant's explanation is pretextual, and then the court must decide, based on all the evidence and with the ultimate burden of persuasion on plaintiff, whether the challenged treatment was based on intentional racial discrimination [4] Harris-Teeter concedes in both cases that if the prerequisites of Rule 23(a) were met, these claims were properly maintainable in class action form under Rule 23(b)(2) [5] It is true that the numerical evidence presented at the time of tentative class certification did not reveal how many blacks had been discharged by Harris-Teeter, and therefore the tentative certification may have been improper, but the numerical evidence at trial, as discussed in the text, satisfied the numerosity requirement [6] Here too the commonality and numerosity requirements were satisfied. The complaint alleged disparate treatment in the promotions decisions, thus raising common questions of law and fact. The numerosity requirement was met by the fact that Harris-Teeter employed some two hundred blacks and promoted over one hundred employees annually [7] This would seem to cast doubt on the continued vitality or sweep of the language of some of our recent cases. See, e.g., Abron v. Black & Decker, Inc., 654 F.2d 951 (4 Cir.1981); Hill v. Western Elec. Co., 596 F.2d 99 (4 Cir.), cert. denied, 444 U.S. 929, 100 S.Ct. 271, 62 L.Ed.2d 186 (1979) [8] Such a focus serves two important purposes. First, satisfaction of the commonality requirement ensures that a primary justification for the class action--efficiency in the adjudication of duplicative claims--will be present. Hence the emphasis on common issues of proof. Second, commonality of proof serves to assure adequacy of the representation that the absent class members will receive, in that the named representative will have the greatest incentive aggressively to discover and present evidence needed for his own personal claims [9] The appropriateness of class certification of a Sec. 1981 claim also depends upon the Rule 23 criteria. See, e.g., Gonzalez v. Southern Methodist University, 536 F.2d 1071, 1072-73 (5 Cir.1976), cert. denied, 430 U.S. 987, 97 S.Ct. 1688, 52 L.Ed.2d 383 (1977) [10] Unfortunately, it is not wholly clear from the record whether the supervisors who made the decisions regarding promotions were the same people who made the decisions regarding terminations [11] We of course give this factor less weight because it was largely raised by Harris-Teeter as a justification for certain instances of refusals to promote blacks, rather than by plaintiffs in their certification motion, and because it was regarded by the district court as, arguably, a pretext developed after the fact [12] Through this argument, Harris-Teeter of course also contends that the individual claims of the intervenors never should have been presented and thus should be vacated on appeal [13] Although the Sec. 1981 promotions claims of the intervenors and the class they represented were not, therefore, proper until April of 1979, the statutory time limit for filing these claims was tolled by the existence of Lilly's lawsuit in which promotions were contained in the putative class. See American Pipe & Construction Co. v. Utah, 414 U.S. 538, 94 S.Ct. 756, 38 L.Ed.2d 713 (1974); Crown, Cork & Seal Co. v. Parker, --- U.S. ----, 103 S.Ct. 2392, 76 L.Ed.2d 628 (1983) [14] See also Spirt v. Teachers Ins. & Annuity Ass'n, 93 F.R.D. 627, 641-42 (S.D.N.Y.1982) [15] Cf. Hill v. Western Elec. Co., 672 F.2d 381, 390 n. 6 (4 Cir.) (suggesting that intervention by persons claiming discrimination in hiring, who had not exhausted EEOC remedies at the time the class action was filed by an individual claiming discrimination in promotions, was proper if the employer had, by the time the suit was filed, been given fair notice of, and opportunity to resolve through conciliation, the hiring claim), cert. denied, --- U.S. ----, 103 S.Ct. 318, 74 L.Ed.2d 294 (1982) [16] Because Lilly's filing suit of a class purporting to include promotions tolled the limitations period, these claims were timely filed. See supra note 13 [17] If possible, it is highly preferable to examine the statistical data for the time period in combined form, rather than year by year. Combined data is more likely to demonstrate the "pattern or practice" of defendant's policies, whether discriminatory or not. Moreover, by increasing the absolute numbers in the data, chance will more readily be excluded as a cause of any disparities found. For example, if a coin were tossed ten times in the first day and came up heads four times, no one would think the coin was biased (0.632 standard deviations), but if this same ratio occurred for a total of 10,000 tosses, of which 4,000 were heads, the result could not be attributed to chance (20 standard deviations) [18] As we have touched upon previously, there are two common models of statistical analysis: the binomial model and the hypergeometric model. See EEOC v. Federal Reserve Bank of Richmond, 698 F.2d 633, 650 (4 Cir.1983). The binomial model is appropriate where the "selection" (i.e., hiring, promotion, or termination) of the individuals does not significantly alter the racial composition of the pool from which future selections are to be made. This would be true, for instance, in selecting 50 people to be hired for unskilled work from a large city, because making the selection will not, even if all those chosen as the process goes along are white, significantly affect the racial composition of the applicant pool. The hypergeometric model, on the other hand, is appropriate where the selections could significantly alter the racial composition of the pool from which future selections are to be made (that is, a finite pool). An example of this would be a situation in which 10 managers are to be promoted from a group of 30 foremen, of whom half are black and the other half white. The key distinction here is that the selection of a white for the first position will significantly alter the racial composition of the selection pool, from 15 of 30 (.500) to 15 of 29 (.517) black. Thus, if another white is chosen, the inference of discrimination will properly be heightened. See Peterson, Binomial v. Hypergeometric Employee Selection Models, 2 Personnel Research Report 1 (April 1983) The binomial process is probably appropriate here because although the pool of persons who can be terminated is fixed at any given time, it is an open pool when examined over a five year period, as here, in that new employees will be hired and then themselves be subject to termination. [19] Under the binomial model, n = total employees terminated p = black percentage of the workforce 1-p = white percentage of the workforce Q = actual number of black terminations E = expected number of black terminations = np § = standard deviation = number of standard deviations = (Q-E)/s Here, n = 815, p = .1572, 1-p = .8428, Q = 229, E = np = 128.12 § = 10.39, and the number of standard deviations = 9.71 [20] By conclusively ruling out chance as the cause of the disparity in the termination rates, plaintiffs thereby made out a prima facie case of class-wide discrimination, thus shifting the burden to Harris-Teeter to come forward with a credible lawful explanation for the disparity [21] The fact of this discretion is itself relevant. See Sledge v. J.P. Stevens & Co., 585 F.2d 625, 635 (4 Cir.1978) ("where [strong statistical] proof is coupled with evidence that the defendant based hiring and other employment decisions upon the subjective opinions of white supervisors, the trial court is entitled to infer, as it did here, that the defendant illegally discriminated"), cert. denied, 440 U.S. 981, 99 S.Ct. 1789, 60 L.Ed.2d 241 (1979) [22] Harris-Teeter advanced no explanation for the disparity between the black and white termination rates, instead arguing that the plaintiffs' statistics "showed only that both blacks and whites had committed dischargeable offenses and had been discharged." That assertion is plainly incorrect, for the standard deviation analysis demonstrates that the significant disparity between the termination rates was not due to chance, but instead to some other cause. That cause is presumed to be unlawful discrimination until the defendant shows otherwise, see supra note 20, and Harris-Teeter failed to do so here [23] The promotions data for the warehouse in 1975, available in the record, showed that more blacks were promoted than would be suggested by their proportion of the workforce. See infra note 24 Of course, we do not suggest that a 14% disparity between black and white promotion rates is, if proved to be attributable to racial discrimination, any less illegal than a 30% disparity. However, it is less persuasive as proof that such discrimination did, in fact, exist. [24] The promotions data, in the form found by the district court and with standard deviations by the binomial method provided, is: Employees Promotions # Standard Unit Year White Black White Black Deviations ------ ---- ----- ----- ----- ----- ---------- Ware.k Stores 1976 1603 284 171 21 1.59 Stores 1975 1191 99 104 6 0.87 Stores 1976 1192 111 129 8 1.12 When adding the additional data mentioned in footnote 23, supra, the promotion figures are: Employees Promotions # Standard Unit Year White Black White Black Deviations ------ ----- ----- ----- ----- ----- ---------- Ware. 1975 223 192 18 21 -0.96 Ware.k 1975k Stores 1976 3017 575 293 48 0.97 [25] Although ultimate formal responsibility for hiring, promotions, and terminations all rested in the district managers, the district court found that these managers adopted almost automatically the recommendations of the supervisors at the stores. There was no evidence that the district managers instructed or encouraged the store supervisors to make recommendations that were influenced by racial discrimination [26] We also include here the "transfer" claimed by Gary, because the district court appears to have treated this claim along with Gary's promotion claim, and because Gary testified that this "transfer" would have included a pay raise, indicating that it too was actually a claimed promotion [27] This conclusion is buttressed by the overwhelming statistical disparity between the black and white discharge rates, which proves class-wide racial discrimination in terminations [28] We reject, however several of Harris-Teeter's other objections to the attorneys' fees award. First, the results obtained are merely one factor to be considered by the district court, and the award is not necessarily limited to time spent on issues upon which plaintiffs ultimately prevailed. Second, it is plain that paralegals can be billed at a rate that includes such overhead as rent and secretarial services. Finally, the district court did not err in awarding attorneys' fees for time spent preparing proposed findings. Such a request is perfectly proper and chargeable to the defendant