Gilbert Rendon, Cross-Appellant. And Joe Mike Zepeda, Intervenor-Plaintiffs-Appellees, Cross-Appellants v. At & T Technologies, Cross-Appellee

W. EUGENE DAVIS, Circuit Judge: In this pattern and practice disparate treatment 1 Title VII and section 1981 action, AT & T appeals a judgment in favor of a class of Black and Mexican-American telephone installers in AT & T’s San Antonio and Rio Grande operations. The district court found that the class plaintiffs proved AT & T’s “intent to discriminate against class members” and awarded in-junctive relief and damages. We find no error and affirm. The class members cross-appeal the district court’s award of attorney’s fees and expenses. We affirm the attorney’s fee award, but vacate the district court’s denial of expenses and remand for further proceedings consistent with this opinion. I. BACKGROUND In September 1971, Gilbert Rendon and members of a class of Mexican-American telephone installers filed charges with the Equal Employment Opportunity Commission alleging that their employer, Western Electric, Inc., discriminated against its Mexican-American employees. AT & T is now the plaintiffs’ employer and the party defendant. When the Commission’s conciliation efforts failed to produce an amicable resolution to the parties’ dispute, the plaintiffs brought this Title VII and section 1981 action in the district court. In June 1975, Kenneth Cubit and a class of Black installers, also employed by Western Electric, intervened in the Rendon class action charging that Western Electric had also discriminated against members of their class. The district court certified the class in July 1978. The class consisted of all present Mexican-American and Black installers employed by Western Electric in the San Antonio and Rio Grande Valley areas; all past and future Mexican-American and Black applicants for installer positions at Western Electric in the San Antonio and Valley areas; and all former Mexican-American and Black installers at Western Electric in the San Antonio and Valley areas whose claims were not time barred. The district court found that the plaintiffs “failed -to prove discrimination in de *391 fendant’s hiring and termination practices. The plaintiffs do not appeal that ruling. The district judge concluded, however, that there were “gross disparities in promotion times and incidents of overt discrimination” with regard to promotions for installers. AT & T appeals this feature of the district court’s judgment. In order to discuss the merits of AT & T’s appeal of the district court’s judgment granting relief to plaintiffs on their Title VII and section 1981 claim based on discriminatory promotions, we must describe in some detail Western Electric’s promotion system. 2 A. Western Electric’s San Antonio and Rio Grande divisions install, modify and remove telephone company central office equipment and commercial PBX exchanges for Bell System companies and their industrial, commercial and governmental customers. The compensation, and other terms and conditions of employment of the hourly workers in the San Antonio/Rio Grande Valley installation division are governed by a collective bargaining agreement between Western Electric and the installers’ union. Under this agreement, the hourly wage of an installer is mainly dependent upon the installer’s position within an “index plan.” Western Electric’s index plan provides for five index levels — one through five— and each installer is assigned to one of the five indices depending on the installer’s skill level. All new employees in this department begin at index 1 and are promoted to a higher index when they meet the requirements for promotion, outlined below. The higher levels are filled by promotions from the lower indices. An installer’s advancement through the index plan is dependent on two factors: (1) opportunity to perform higher index work, which is provided if a supervisor assigns work in the higher index; and (2) a satisfactory performance evaluation by a supervisor of that higher index work. If an installer is never assigned work at a higher skill level, the installer cannot progress to the next level. Similarly, poor performance evaluations by a supervisor of the higher index work will prevent an installer from moving to the next index. With respect to the first promotion requirement, performance of above index work, the plaintiffs presented evidence that Western Electric has no established procedure for evenhandedly assigning its installers to higher index level work. Indeed, any installer may be assigned duties in any work operation code. In practice, the orbit supervisor selects installers for Western Electric projects and assigns them to sub-orbit and line supervisors. The sub-orbit and line supervisors, in turn, assign work to the installers. These supervisors control an installer’s advancement because an installer cannot advance to the next higher index if he is not assigned work in codes within that index. The second component of advancement is the process by which an installer is evaluated in the index work codes. It is the role of the job installation supervisor to specify whether a worker is “qualified” in particular work operations, based on the supervisor’s recollection of the installer’s work. If the line supervisor thinks an installer is qualified in a particular code, the supervisor rates the installer as a “9”. If an installer receives rating of “9” in the required codes, the installer advances to the next index level. Although the supervisor’s evaluation controls an installer’s advancement within the index plan, no written criteria are in place to guide the supervisors in their evaluations of an installer’s performance. Job assignments and job ratings by supervisors are based almost entirely upon the supervisor’s subjective judgment, with practically no objective criteria for evaluation. B. The district court found that Western Electric used its highly subjective index *392 plan to discriminate against promotion of minority installers. The district court determined that Western Electric accomplished this discrimination in two ways. First, “class members were assigned to substantially less work which would qualify them to advance to index 3 or higher” indices. White installers received index 3 “level work or higher at a rate of about twice that of the class members.” Second, the district court found that “[c]lass members received dramatically fewer ‘9 ratings’ than whites.” The district court enjoined AT & T from “maintaining practices, policies, customs or usages which discriminate against Mexican-Americans and Blacks because of their race or national origin, with respect to promotion opportunities.” The district judge also awarded attorney’s fees to plaintiffs’ counsel in the amount of $178,305, plus expenses of $36,439.21. Some class members received awards of back-pay. This appeal followed. C. The district court relied heavily on plaintiffs’ statistical evidence in its finding of liability. In this appeal, appellant points primarily to what it perceives as the. shortcomings of plaintiffs’ statistical evidence and why that evidence will not support the inference of discrimination that the district court drew from it. We review the district court’s factual finding of intentional discrimination under the clearly erroneous standard of review. Pullman Standard v. Swint, 456 U.S. 273, 289-90, 102 S.Ct. 1781, 1790-91, 72 L.Ed.2d 66 (1982). If the district court’s account of the evidence is plausible in light of the record viewed in its entirety, the court of appeals may not reverse it even though convinced that had it been sitting as a trier of fact, it would have weighed the evidence differently. When there are two permissible views of the evidence, the factfinder’s choice between them cannot be clearly erroneous. Anderson v. Bessemer City, 470 U.S. 564, 573-74, 105 S.Ct. 1504, 1511, 84 L.Ed.2d 518 (1985). On review, we will reverse a district court’s factual findings only if, on the entire evidence, we are “left with the definite and firm conviction that a mistake has been committed.” United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948). We turn now to the evidence presented to the district court. II. A. Plaintiffs’ Statistical Evidence The class members’ statistical study was conducted by Dr. William R. Schucany, a professor of statistics, who based his analy-ses upon data provided by Western Electric in discovery. Dr. Schucany’s database consisted of seven different cohorts of workers, based upon year of hire in years 1965 through 1971. His database consisted of all employees hired between 1965 and 1971 who entered at index 1 except that: (1) all employees who left Western Electric before they were promoted to index 3 were excluded; and (2) employees whose service with Western Electric had been interrupted for reasons such as military service or continuing education were excluded. Dr. Schucany next grouped installers according to year of hire, primarily to control for economic conditions which would force layoffs based on seniority. Except for individuals excluded for one of the above reasons, the database included all installers identified by Western Electric in discovery. Dr. Schucany’s principal conclusions are summarized in plaintiffs’ exhibits 37-40. Plaintiffs’ Exhibit 37, set out in the margin, 3 indicates that for each of the study *393 cohorts (i.e., installers hired in 1965, 1966, 1967, 1969, 1970 and 1971) the difference between the average time to promotion to index 3 for white installers and minority installers is approximately one year. Thus, white employees enjoyed a faster promotion, on average, than Black and Mexican-American installers. At trial, Dr. Schucany testified that “this difference is a statistically significant difference,” that fell between two and three standard deviations from the mean. Consistently, Plaintiffs’ Exhibit 38, set out in the margin, 4 indicates that for the cohorts which Dr. Schucany analyzed (i.e., installers hired in 1966, 1969 and 1970), white installers enjoyed a faster promotion from index 1 to index 4, on average, than Black and Mexican-American installers. At trial, Dr. Schucany conceded that, with respect to promotion to index 4 there was “too little data, really, on the promotion times to make any kind of firm conclusion. I only observed that it’s consistent with what’s on Exhibit 37 about the movement to Index 3.” Plaintiffs’ Exhibit 39, set out in the margin, 5 demonstrates that for each of the six years covered, white installers received an average of one more qualified rating in above index work than did minority installers. At trial, Dr. Schucany testified that [t]hat, in itself, is statistically significant and ... consistent with promotion times and the work assignments ... this conclusion about ratings is just right in line *394 with the others. But just by itself, it’s significant due to being so consistent in one direction for each of six independent groups of people that were compared. And last, Plaintiffs’ Exhibit 40, set out in the margin, 6 details that white installers received assignments in index 3 level work or higher at consistently higher rates than that of the plaintiffs. At trial, Dr. Schucany testified that his analysis of each of the study cohorts primarily concentrated on the liability period, which began on June 10, 1971. 7 Dr. Schucany’s study continued through the end of 1979. B. Defendant’s Statistical Evidence Defendant’s expert, Dr. Michael Yost, Jr., conducted a different analysis of minority promotion rates and minority representation within the five indices. Dr. Yost conducted three different inquiries: (1) a comparison between the time to promotion of white and minority installers; (2) a comparison between the promotions of minorities and whites in proportion to their respective number eligible for promotion; and (3) a linear regression analysis on the representation of minorities in each of the five indices from 1969 through 1979. In the first inquiry, Dr. Yost conducted twenty-four “T-tests” and found no statistically significant differences between minority and white times-to-promotion. In the second inquiry, Dr. Yost concluded that at the .05 significance level, minorities never received less than the expected number of promotions “given knowledge of the proportion of the pool of people eligible for promotions who were minority members.” And last, Dr. Yost found a trend of increasing minority representation from 1969 through 1979, approaching “a fifty-fifty level” with white installers. Dr. Yost testified that his trend studies did not support a finding of discriminatory treatment against minorities. He emphasized that his results were more reliable than those of Dr. Schu-cany because he had studied the entire population of installers at Western Electric. The district court concluded, however, that Dr. Yost’s testimony was not credible. The district court found first that Dr. Schu-cany’s education and experience were superior to Dr. Yost’s. The district court also determined that Dr. Yost’s conclusions were less reliable than Dr. Schucany’s because Dr. Yost “failed to hold constant for date of hire” by “averaging the advancement times of all white installers, regardless] of date of hire, and then comparing them with minorities.” The district court was entitled to conclude that Dr. Yost’s failure to hold constant for date of hire, i.e., compare white workers to minorities hired the same year, was a serious *395 flaw. Comparing workers hired the same year rules out a number of circumstances other than race that may affect promotion, such as unusually good or bad economic conditions, technical advances, demographics and others. The district court concluded that Dr. Yost did not state how his index-by-index [analysis] would control [a number of] variables, but merely stated in conclusory fashion that his analysis did so. It appears to this court that a year by year analysis is less affected by these changing factors than an index-by-index analysis cutting across many years, and thus, a greater period of time. In sum, defendant’s statistical evidence was subject to serious limitations. The district court did not clearly err by accepting Dr. Schucany’s testimony over that of Dr. Yost. C. Plaintiffs’ Anecdotal Evidence In addition to Dr. Schucany’s statistical presentation, the class members in their testimony described the dearth of class members in either higher index level jobs or supervisory positions. Class members also testified to specific instances of overt discriminatory conduct. For example, Mr. Rodriguez testified that, on one occasion, he had been assigned to index 4 work on an otherwise all-white analyzation crew. Rodriguez testified that he was relegated to sorting cable tanks while white installers actually performed the project analyzing. One of the white installers performing analyzation was less experienced than himself. Rodriguez also testified to the use of racial slurs by white installers and supervisors. Rodriguez stated that the white installers referred to the Mexican-American installers, as “[w]et-back ... You Mexican.” When Rodriguez complained, his supervisor responded that he “was just [a] rebel rouser ... a trouble maker.” Longoria and Arteaga similarly testified that they had been denied assignments to higher index work and projects commensurate with their technical expertise while white installers, at lower index levels and with less technical expertise, were given higher level, and more challenging, assignments. III. On appeal, AT & T raises seven errors: (a) Dr. Schucany’s analysis focused on conduct that occurred outside the liability period; (b) Dr. Schucany failed to control his study for “basic factors,” such as when an employee was hired, Western Electric’s hard core hire program and credit given by the company for previous work experience; (c) Dr. Schucany’s sample was not representative of the Western Electric installers; (d) Dr. Schucany improperly aggregated data; (e) there was no competent evidence addressing promotions to index levels 4 and 5; (f) the statistical evidence does not show gross disparities between promotion times of white installers and the class members; and (g) the plaintiffs’ anecdotal evidence is insufficient to support a finding of pattern and practice liability. We consider each of these arguments below. A. Appellants first argue that Dr. Schuca-ny’s analysis focused on unactionable conduct that occurred outside the liability period, beginning June 10, 1971. However, the discriminatory conduct extended into the liability period for all but two of Dr. Schu-cany’s cohorts, the 1965 and 1966 hires. In other words, only two cohorts, the 1965 and 1966 hires, were subject to discriminatory promotion practices entirely outside the liability period. AT & T erroneously argues that the district court was precluded from considering any evidence from outside the liability period. For a narrow range of cases, those in which a violation is continuing in nature, this court recognizes that liability may be based on continuing unlawful conduct which preceded the liability period. See Clark v. Olinkraft, Inc., 556 F.2d 1219, 1222 (5th Cir.1977); Crawford v. Western Electric Co., Inc., 614 F.2d 1300, 1309 (5th Cir.1980); Berry v. Board of Supervisors, 715 F.2d 971, 979 (5th Cir.1988). When an employer has a discrimina *396 tory promotion policy or system and the system is maintained into the charge-filing period, a plaintiff may attack the promotion system even though the plaintiff was not denied a job benefit within the charge-filing period. See B. Schlei & P. Grossman, Employment Discrimination Law 1050 (2d ed. 1983). “[Liability of the employer for back pay may be based on acts occurring outside the two-year period if a current violation is shown.” Crawford, 614 F.2d at 1309 (emphasis in original). The district court was entitled to consider AT & T’s index plan as such a continuing discriminatory promotion policy or system. As we stated in Clark, 556 F.2d at 1222, quoting Rich v. Martin-Marietta Corp., 522 F.2d 333, 348 (10th Cir.1975): [Failure to hire] is different from an employee who is seeking promotion. The former takes place on a particular day, whereas in the promotion area it invariably arises during a lengthy period of time. Plaintiffs here challenge the entire promotion system maintaining that it continually operated so as to hold them in lower echelons. Hence, the 90 day [now 180 day] period prior to the filing of the EEOC charges looms inconsequential in this kind of case. Promotion systems, unlike hiring systems, “produee[] effects that may not manifest themselves as individually discriminatory except in cumulation over a period of time.” Glass v. Petro-Tex Chemical Corp., 757 F.2d 1554, 1561 (5th Cir.1985). In summary, once the plaintiffs demonstrated that AT & T used its index plan to discriminatorily deny promotions to plaintiffs within the liability period, plaintiffs were entitled to produce evidence outside the liability period to show the continuing nature of this discriminatory policy. Thus, the district court did not err in considering evidence of pre-liability period discrimination. B. Next AT & T contends that Dr. Schucany failed to control for other nonracial factors it terms “basic factors,” that make his conclusions unreliable. The defendant alleges that the time within each year of hire strongly affects the number of index reviews an installer undergoes “and thus his opportunities for promotion.” Although control for this factor would be desirable, we are persuaded that the district court was entitled to determine that it was not a critical, methodological flaw. Relatedly, AT & T contends that Dr. Schu-cany failed to consider the effect of the company’s “Hard Core Employment Program,” in which they hired a number of minorities who did not measure up to their usual standards. The number of these “hard core” workers hired was small and the district court did not err in determining that the expert’s failure to control for this factor was not a critical flaw in his study. Moreover, AT & T did not present evidence at trial detailing the effects of these variables on AT & T’s practices. In fact the record supports the view that these variables were inconsequential. For example, the named plaintiffs testified that the hard core hires did not remain very long at Western Electric. Because Dr. Schucany factored out all installers who did not remain long enough at Western Electric to be promoted to index level 3, the hard core hires probably had no effect on Dr. Schuca-ny’s analyses. While Dr. Schucany’s conclusions may have been more reliable if he had controlled for these variables, his failure to do so goes to the weight of his testimony. Even if Dr. Schucany’s model was not perfect, it was not so flawed that the district court was required to reject it. C. The defendant next challenges the validity of Dr. Schucany’s analyses because he allegedly employed an unrepresentative sample. Specifically, the defendant challenges Dr. Schucany’s decision to exclude from analysis white installers hired before 1965 and employees whose service with Western Electric had been interrupted for reasons such as military service or continuing education. AT & T argues that when Schucany excluded these installers, the sample was no longer representative and *397 inferences drawn from unrepresentative samples are inherently flawed. Dr. Schu-cany testified that his database excluded these installers so that he could compare the treatment of similarly situated installers who differed only in ethnicity. At trial, when pressed on this same point, Dr. Schucany stated, “my dataset is very nearly the dataset of interest. And not a question of whether it’s small or representative. It’s almost the entire set that we’re talking about.” The district court did not err in crediting Dr. Schucany’s testimony on this point that excluding these installers improved his database by removing nonracial factors from consideration. D. AT & T next asserts that the district court erroneously accepted Dr. Schucany’s aggregation of the data summarized in Plaintiffs’ Exhibit 37. This objection is based principally on the testimony of AT & T’s own expert, Dr. Yost, who stated that aggregation can spuriously suggest long term discriminatory practices by giving too much weight to one atypical group. The district court apparently rejected Dr. Yost’s conclusion and this finding is not clearly erroneous. Aggregation is not impermissible as a matter of law. Indeed, other circuits have held that “the pooling issue should be decided on the facts of the particular case.” Coates v. Johnson & Johnson, 756 F.2d 524, 542 (7th Cir.1985). See also Adams v. Gaudet, 515 F.Supp. 1086, 1145 (W.D.La.1981); Brown v. Delta Air Lines, 522 F.Supp. 1218, 1228 (S.D.Tex.1980). Here, the record contains evidence from which the district judge could conclude that aggregation was appropriate. At the trial Dr. Schucany stated, “it’s certainly a proper statistical approach to break an analysis into groups. But it is equally important to recombine the results of those separate tests to get the overall picture.” We agree with the Seventh Circuit that we do not have the statistical expertise to declare a particular statistical technique inappropriate where a qualified expert expresses a contrary view and the district court credits the testimony. Coates, 756 F.2d at 542. Thus, we cannot say that the district court’s acceptance of pooled data was clearly erroneous. E. AT & T also argues that the evidence was insufficient to support the district court’s finding of discrimination in promotions to index 4 and 5. We need not consider this argument. The district court primarily based its finding of liability on defendant’s discrimination in promotion of plaintiffs to index 3, which is amply supported by the evidence. The district court was certainly entitled to consider Dr. Schu-cany's testimony on promotions to index 4 as corroborative of Dr. Schucany’s earlier studies that revealed discrimination in promotions of minorities to index 3. F. AT & T next contends that, as a matter of law, the plaintiffs’ statistical evidence did not justify an inference of discrimination because the disparity between promotions of white installers and the plaintiffs was less than three standard deviations from the mean. According to AT & T, the combined standard deviations in Dr. Schucany’s promotion times analysis (regarding promotions to index 3) is only 2.9. Assuming without deciding that AT & T correctly translated the data in Exhibit 37 to standard deviations, we nonetheless conclude that the district court was entitled to find that Dr. Schucany’s statistical analyses demonstrated statistically significant disparities between promotion times of minority and white installers. We reject AT & T’s claim that there is a strict legal benchmark that requires three standard deviations to demonstrate that data has statistical significance. AT & T’s reliance on cases in which we affirmed as not clearly erroneous liability determinations based on statistical evidence showing three standard deviations is misplaced. See Williams v. New Orleans Steamship Association, 673 F.2d 742, 749 n. 13 (5th Cir.1982), cert. denied, 460 U.S. 1038, 103 S.Ct. 1428, 75 L.Ed.2d 789 (1983) (“A deviation greater than three times the standard *398 deviation is prima facie proof that the selection system is not random ... ”). AT & T cites no case which “requires” a finding of three standard deviations for liability. Indeed, we have affirmed liability determinations based on the difference between expected and observed values in “excess of two or three standard deviations” at the five percent significance level. Pegues v. Mississippi State Employment Service, 699 F.2d 760, 768 n. 9 (5th Cir.1983), cert. denied, 464 U.S. 991, 104 S.Ct. 482, 78 L.Ed.2d 679 (1983). See also Hazelwood School District v. United States, 433 U.S. 299, 308 n. 14, 97 S.Ct. 2736, 2741-42 n. 14, 53 L.Ed.2d 768 (1977); Castaneda v. Partida, 430 U.S. 482, 497 n. 17, 97 S.Ct. 1272, 1281 n. 17, 51 L.Ed.2d 498 (1977). Two standard deviations are also sufficient to establish liability in our sister circuits. See, e.g., Palmer v. Shultz, 815 F.2d 84, 92 (D.C.Cir.1987); Eastland v. TVA, 704 F.2d 613, 622 n. 12 (11th Cir.1983), cert. denied, 465 U.S. 1066, 104 S.Ct. 1415, 79 L.Ed.2d 741 (1984). Justice O’Connor’s plurality opinion in Watson v. Fort Worth Bank and Trust, — U.S. —, 108 S.Ct. 2777, 2789 n. 3, 101 L.Ed.2d 827 (1988), quoting, Teamsters v. United States, 431 U.S. 324, 340, 97 S.Ct. 1843, 1856-57, 52 L.Ed.2d 396 (1977), is consistent with this view: We have emphasized the useful role that statistical methods can have in Title VII cases, but we have not suggested that any particular number of “standard deviations” can determine whether a plaintiff has made out a prima facie case in the complex area of employment discrimination. ... At least at this stage of the law’s development, we believe that ... a case-by-case approach properly reflects our recognition that statistics “come in infinite variety and ... their usefulness depends on all the surrounding facts and circumstances.” Here, by AT & T’s own calculation, we have a disparity equivalent to 2.9 standard deviations from the mean. Dr. Schucany testified that there was less than a five percent chance that the difference between the promotion times of minority installers and white installers could occur by chance. By both standards, the class members met their burden of proving “gross disparities” between the promotional treatment of white and minority installers. G. Last, AT & T contends that the class members’ anecdotal testimony is not sufficient to establish a prima facie case of pattern and practice disparate treatment. We need not consider this argument. The relevant inquiry is whether all the evidence, the statistical and the anecdotal, is sufficient to support the judgment. We answer this question in the affirmative and AFFIRM the district court’s findings of liability. IV. A. The plaintiffs contend in their cross-appeal that the district court’s attorney’s fee award was inadequate. 8 In May 1986, the district court ordered payment of the plaintiffs’ interim attorney’s fees. Initially, the district court computed a lodestar in the amount of $237,740 in fees. The district court next found “that a multiplier of .5 shall be applied to the lodestar,” because the plaintiffs' attorneys had spent eleven years on the case and incurred large out-of-pocket expenses; thus, the multiplier raised the fee award to $356,610. Then the district judge reduced the entire award by *399 50%, to $178,305, because the plaintiffs “failed to prevail on some issues and because of the plaintiffs’ very limited success.” As examples of limited success, the district court pointed to the plaintiffs’ failure to prove discrimination in defendant’s hiring and termination practices, and the fact that only 26 of 150 class claimants recovered back pay and compensatory damages. On appeal, plaintiffs argue that under the “results obtained” test of Hensley v. Eckerhart, 461 U.S. 424, 435 n. 11, 103 S.Ct. 1933, 1940 n. 11, 76 L.Ed.2d 40 (1983), it is not necessary that plaintiffs receive all of the relief requested in order to receive full attorney’s fees. We agree in principle that the plaintiffs are entitled to a fee award when they prevail on “any significant issue in litigation which achieve[d] some of the benefit the parties sought in bringing the suit.” Texas State Teachers Assn. v. Garland Indep. School Dist., — U.S. —, 109 S.Ct. 1486, 1491, 103 L.Ed.2d 866 (1989), quoting Nadeau v. Helgemoe, 581 F.2d 275, 278-79 (1st Cir.1978). While plaintiffs may be entitled to a fee, however, they are not automatically entitled to a fee for all services, necessary or not, that counsel performed. As the court stated in Hensley: If, on the other hand, a plaintiff has achieved only partial or limited success, the product of hours reasonably expended on the litigation as a whole times a reasonable hourly rate may be an excessive amount. This will be true even where the plaintiff’s claims were interrelated, nonfrivolous, and raised in good faith. Congress has not authorized an award of fees whenever it was reasonable for a plaintiff to bring a lawsuit or whenever conscientious counsel tried the case with devotion and skill. Again, the most critical factor is the degree of success obtained. Hensley, 461 U.S. at 436, 103 S.Ct. at 1941. Here, plaintiffs contend that, although many of the plaintiffs were denied actual damages in a class action case, an attorney is required to represent all of the individual class members: thus, they were not free to select those plaintiffs most likely to prevail in the damages phase of the case. They argue that by reducing the attorney’s fee award, the court penalized their counsel for representing all class members in the damages feature of the case. Plaintiffs also contend that the district court failed to adequately consider the benefit they all received from the injunctive relief. We review a district court’s determination of a fee award under the abuse of discretion standard. Hensley v. Eckerhart, 461 U.S. 424, 103 S.Ct. 1933, 76 L.Ed.2d 40 (1983); Cruz v. Hauck, 762 F.2d 1230, 1233 (5th Cir.1985). Contrary to plaintiff’s argument, the district court did not reduce the attorney’s fees solely because of counsel’s unsuccessful efforts to recover damages for individual class members. While the limited success of the individual class members at the damages phase affected the district court’s final award, it expressly stated that “discrimination in defendant’s hiring and termination practices,” on which the plaintiffs did not prevail, “were significant aspects of the plaintiffs’ case.... [T]he lack of success on these vital issues cannot be ignored.” Our review of the record reflects that counsel expended considerable time and effort on the plaintiffs’ claims of discriminatory hiring and termination practices. The district court did not abuse its discretion in arriving at the fee award. B. Next, the plaintiffs argue that the district court erroneously failed to award them $7,376.81 in expenses incurred during the damage phase of the litigation. The district court denied the plaintiffs’ request for expenses because “[plaintiffs’ updated application for attorney’s fees was filed August 30, 1985 and should have included all hours sought to that point. Additional attorneys’ fees evidence should have been timely submitted.” The record indicates, however, that some of the requested expenses were incurred after the August 30, 1985 deadline. While the district court did not abuse its discre *400 tion by refusing to consider expenses from December 1984 through August 30, 1985, the date the plaintiffs submitted their application for fees and expenses, it should not have refused to consider plaintiffs’ request for expenses incurred after August 30, 1985. The district court may not deny a motion for expenses or fees as untimely unless the plaintiffs have violated a clear deadline established by the court, or the district court finds that a post-judgment motion “unfairly surprises or prejudices the affected party.” Cruz, 762 F.2d at 1237. Here, the district court made no such finding. Accordingly, we VACATE the order denying $7,376.81 in expenses and REMAND to the district judge for further consideration consistent with this opinion. The judgment of the district court is AFFIRMED in all respects except that the order denying $7,376.81 in expenses is VACATED and REMANDED to the district court for further consideration consistent with this opinion. Footnotes [1] After the argument in this appeal, the Supreme Court decided Wards Cove Packing Co., Inc. v. Atonio, — U.S. —, 109 S.Ct. 2115, 104 L.Ed.2d 733 (1989). In Wards Cove, the Supreme Court clarified the burdens of production and persuasion in disparate impact cases. This is a disparate treatment case, not affected by the Wards Cove decision. Although the district court used some language from disparate impact cases, the record in its entirety demonstrates that the parties tried this case as a disparate treatment action and the parties on appeal briefed this case according to the relevant disparate treatment precedents. [2] This system has been described in two previous decisions of this court. See Crawford v. Western Electric Co., Inc., 614 F.2d 1300 (5th Cir.1980); Smith v. Western Electric Co., Inc., 770 F.2d 520 (5th Cir.1985). [3] Summary of the Analysis Time-to-Promotion to Index 3 Year of Hire No. of White a No. of Class Average Time b to Index 3 (Years) Members a White Class Difference Significance Probability c 1965 11 4.61 6.36 1.75 .077 1966 10 4.01 5.01 1.00 .234 1967 4 3.53 4.81 1.28 .200 *393 1968 1 No comparison possible 1969 15 9 2.93 3.81 .88 .048 1970 22 6 3.42 4.85 1.43 .109 1971 33 5 2.75 3.77 1.02 .009 Combined Significance Probability = .0019 d Sources: Appendix G, Defendants Answers to Interrogatories Supplemental Answers for Valley Employees, Exhibit 1 Computer History Tape [a] - Excludes employees hired outside South Texas or above Index 1 or with interruption. [b] - Conservative assignment of promotion times for early terminatees in Years 1965-1970. Promotion times only for 1971. (Of the 14 not promoted before termination 4 were Class members and 10 were Whites), [c] - Mann-Whitney-Wilcoxon rank-sum test. [d] - Fisher’s method for combining statistically independent tests. [4] Summary of Time-to-Promotion to Index 4 Average Time (Years) to Index 4 White Class for those promoted Year of Hire White Total Number Class Number Promoted (%2 White Class (%1 1965 10 (17) 9 0 (0) 6.45 1966 10 (23) 9 2 (18) 1967 4 (20) 4 0 (0) 1969 14 (39) 10 (29) 5.80 0\ 1970 20 (23) 13 (13) 6.32 (O 1971 26 (13) 8 (0) 4.13 Sources: Appendix G, Defendant's Answers to Interrogatories. Exhibit 1, Supplemental Valley Employees. Computer History Tape. > [5] Summary of Ratings as Qualified on Index 3 Tasks Year of Hire Second Semiannual Average Number of 9's White Class Report for Year Difference 1965 70 3.00 1.50 1.50 1966 70 2.67 2.00 .67 1967 70 3.00 1.00 2.00 1969 71 1.40 .25 1.15 1970 72 1.81 1.17 .64 1971 73 2.41 1.40 1.01 Source: Copies of Western Electric Forms SD-4-1421 "Installer Experienc Record” [6] Summary of Numbers of Hours Assigned to Work Operation Codes in Index 3 or Higher Year of Second Semiannual Average Number of Hours Hire Report for Year White (Proportion of Total) Class 1965 70 4370 (.48) 2700 (.28) 1966 70 3060 (.44) 3183 (.45) 1967 70 2433 (.43) 1064 (.18) 1969 71 1540 (.41) 697 (.20) 1970 72 1783 (.41) 1077 (.29) 1971 73 1857 (.42) 1623 (.42) Source: Copies of Western Electric 1421 Forms. [7] In its amicus brief, the EEOC argues that the liability period began on March 12, 1971 because the 180-day filing period, measured back from the date of the charge, applies to this action. Although we recognize the rule is different in other circuits, in this circuit, the 90-day liability period applies to all cases pending before the EEOC before the 1972 amendments to Title VII. Payne v. Travenol Laboratories, Inc., 673 F.2d 798, 815 & n. 20 (5th Cir.1982), cert. denied, 459 U.S. 1038, 103 S.Ct. 451, 452, 74 L.Ed.2d 605 (1982). Here, the underlying charge was filed on September 8, 1971, six months before the March 24, 1972 amendments. Thus, the 90-day charge-filing period applies and the liability period opened on June 10, 1971. [8] The plaintiffs’ notice of appeal designated “Gilbert Rendon, et al." as appellant. AT & T contends that we have no jurisdiction to consider the cross-appeal because under Fed.R.App.P. 3(c) the phrase "et al.” gives no notice of the party seeking an appeal. Although AT & T’s argument may have merit where there is no class certification, see Torres v. Oakland Scavenger Co., — U.S. —, 108 S.Ct. 2405, 101 L.Ed.2d 285 (1988), we reject this argument when a class has been certified. Rendon was designated as the class representative below and, as a result, his actions bound the entire class. Liberally construing the specificity requirement of Rule 3(c), we hold that “Gilbert Rendon, et al.” was sufficient to designate the certified class of plaintiffs as appellants.