Full opinion text
TATE, Circuit Judge: A class of black and female service/maintenance and clerical employees of the defendant Stephen F. Austin State University (the University) brought suit against their employer, contending that the University engaged in race and sex discrimination in violation of the Civil Rights Act of 1964,42 U.S.C. § 2000e et seq. (Title VII) and the fourteenth amendment to the Constitution. On appeal, both parties contest the findings of and relief granted by the district court. The University contends that the lower court should not have certified such a broad class nor granted injunctive relief to mitigate the effects of the channeling of black and female employees into the lower paid clerical and custodial positions. The plaintiff class asserts that the. district court should have awarded back pay, front pay, and interim attorneys’ fees in addition to injunctive relief and that the court erred in finding no discrimination in terminations of blacks and females and in the administration of the University’s retirement system. We affirm the lower court’s judgment with respect to the issues of channeling, termination, and retirement, but we remand for findings on the discriminatory intent and for monetary relief and interim attorneys’ fees determinations, if Title VII relief is to be afforded. Procedural History The named plaintiffs are two black females, Carpenter and Hunt, terminated from University employment, and a black male, Williams, retired in 1978, all of whom were former hourly employees of the University in its service/maintenance department. They exhausted all administrative prerequisites to filing a lawsuit under the Equal Employment Opportunity Act (the Act or Title VII). The defendant University, as a state government agency which “affects commerce” within the meaning of 42 U.S.C. § 2000e(b), is an employer subject to coverage of the Act from March 24, 1972. On September 2, 1976, the district court certified a class of all “past, present, and prospective black and female employees of the University who have been denied or will be denied employment with the University since November 11, 1971.” On review of a motion for decertification prior to trial, the court limited the. membership of the class to “wage and hourly employees” of the University. This characterization excluded salaried clerical and secretarial employees, about whom evidence relating to sex discrimination was adduced at the liability phase of the .trial. After hearing this evidence, the court redefined the class as “all past, present, and prospective black and female employees who would be subject to job descriptions in Plaintiffs’ Exhibits 29 and 30 [service/maintenance and clerical employees] and/or subject to the Classified Pay Plan of said University who have been or will be denied employment or benefits with the University since November 11, 1971. ” The class claims, contained in the original and three amended complaints, encompassed an across-the-board attack on the University’s alleged racially and sexually discriminatory employment policies and practices. The named plaintiffs claimed that the University unlawfully “channelled” blacks and women to lower paying positions through its hiring and initial assignment procedures, and thereafter maintained discriminatory promotion, transfer, pay, termination and retirement programs. In the liability phase of the trial the plaintiffs introduced statistical data of the high percentages of blacks and women in the lower paying service/maintenance and clerical positions, the absence of these protected groups in higher levels, and the lack of transfer between job ranks and classifications. The plaintiffs presented anecdotal evidence of specific instances of discriminatory policies and practices. The plaintiffs also showed the non-coverage of hourly employees in the University-supported Teachers Retirement System until November 1972, the subsequent exclusion of these classified employees from an Optional Retirement Plan also maintained by the University, and the retirement system’s rules that were alleged to have a disparate impact on the protected classes. The district court entered judgment for the employer University on the retirement, equal pay, and termination issues and granted judgment to the plaintiffs on the channeling and promotion issues. It found that the University workforce was racially stratified and sexually stereotyped, that the higher level and paid employees are predominantly white males, and that there were “dramatic overutilizations” of women and blacks in the lower paying, lower skilled jobs. The court found that several distinct employment practices had a disparate impact on blacks and women: (1) the development, use and subjective implementation of job qualifications for which the University made no showing of job-relatedness; (2) the development and use of a classified pay plan for hourly employees in which blacks and women were in lower-paying job classifications, creating wage differentials for which the University offered no justification; and (3) subjectivity in implementation of job qualifications for hiring and promotion and for placement of classes of employees on the compensation scale. The court then indicated its preference for non-monetary institutional responses to the findings of liability and instructed the parties to make good faith efforts to achieve conciliation. The voluntary attempts to devise relief failed. Following a remedy hearing, the district court entered an order appointing a master to oversee the implementation of injunctive relief that had the goal of changing institutional practices and policies. The court ordered the University to validate and reform job qualification descriptions, placement, and compensation; to implement a preference system for currently-employed class members, advancing them into higher level positions for which they qualify; and to institute appropriate written guidelines and record-keeping procedures for supervisors and others who make hiring and promotional decisions. The court, however, refused to award back or front pay relief, or interim attorneys’ fees, as requested by the class. Employment Practices at the University Stephen F. Austin State University, located in Nacogdoches, Texas, a town with a population of approximately thirty thousand, has a student enrollment of approximately eleven thousand. Personnel are hired by the University in three general categories: academic faculty, administrative executives, and the classified staff. No complaint is made of discriminatory employment practices relating to the academic faculty. The plaintiffs’ complaints relate to the University’s practices regarding the administrative and the classified employees. With regard to the latter two categories, the facts as of 1979 show: There were fewer than 100 employees (two blacks and twenty-one white females) in the administrative category, which includes the President of the University, all department heads, and most of the supervisory positions. The more than 800 classified employees are employed in both salaried and hourly positions. Within the classified ranks a majority of the hourly employees are black and female and are assigned to service/maintenance positions, while a majority of salaried employees are white. Black and female employees each constituted slightly less than one-half the classified staff in 1979, which included both salaried and hourly. However, blacks held 77.7% of the hourly non-supervisory positions and 50% of the hourly supervisory positions, while women held 60.9% of the hourly non-supervisory positions and 53.1% of the hourly non-supervisory positions and 53.1% of the hourly supervisory positions. In the salaried non-supervisor area, 13.2% were black employees and 75% were female; for salaried supervisor slots, blacks constituted 6.8% and women constituted 38.6%. These percentages had not changed significantly since 1970. The service/maintenanee category — food service, custodial, garbage crew, grounds workers, and supervisors — which includes over one-half of the classified positions, is predominantly black (337 of 405 positions in 1979 or 83.2%, decreasing from 87.8% in 1972) and female (257 of 405 positions in 1979). In 1979, nine of the ten supervisors were white,’ and eight of them were male. The clerical categories are composed almost exclusively of women; in categorizations encompassing nonsupervisory clerks, secretaries and library workers, women held 156 out of 166 jobs in 1979. The separate job classifications are predominantly segregated according to race and sex, but became' decreasingly so during the 1970’s. In 1970, 70 job classifications were all white, 5 all black, and 14 mixed; in 1979, 89 were all white, 5 all black, and 31 mixed. In 1970, 38 job classifications were all male, 37 all female, and 14 mixed; in 1979, 52 were all male, 39 all female, and 39 mixed. This racially and sexually stratified composition of the classified employees in lower paid positions is alleged to result from institutional employment practices that produce these disparities and adversely affect blacks and women. The some 800 classified employees are subject to the Classified Pay Plan, which sets wage rates and salaries generally by job classification and years of tenure. The hourly positions have been traditionally concentrated in job classifications involving low-skilled manual labor, while the salaried positions are filled primarily by white collar office workers or skilled crafts workers (carpenters, electricians, etc.). Salaried positions are funded by the state by each separate job classification, and these employees may be terminated only for cause; all hourly positions are funded in aggregate, and these employees are terminable at will. The system of classifications for hourly and salaried employees was a result of the University’s attempts beginning in 1972 to describe and set forth qualifications necessary for different types of work. Job descriptions were based on questionnaires in which the person then holding a position described his duties and educational and work experience qualifications. These descriptions established criteria for initial work assignments as well as promotions, although movement between classifications is rare. The Classified Pay Plan, a set of numbered grades and lettered steps to which positions are tied, was first used in 1978. Before then, wages were set on an individual basis. The Classified Pay' Plan Committee, composed of the University’s directors and department heads (all white), designated the wage rates and assigned job classifications to an approximate position on the compensation scale. The supervisor who makes the decision to hire within a department has some discretion in determining on what level of the pay plan a starting employee will be placed. Appeals may be made to the Committee for changes in an individual’s or a certain job classification’s assigned pay grade. Black and female workers are concentrated at the lowest wage categories. Applicants for employment and University employees seeking transfers must apply at the University’s central personnel office. Vacancies, with job descriptions and qualifications listed, are posted at this office, sent to Texas employment agencies, and sometimes advertised in local and state newspapers. The personnel office holds applications for 60 days, after which time the person must reapply. The personnel office conducts interviews and at its discretion refers applicants to the supervisors in the various departments who have sent a job request form to the personnel office. These job requests often contained the name of the person to be replaced. The supervisor has discretion to choose among those sent for interviews (although it is unclear whether a group is sent from which a decision will be made, or the personnel department sends other applicants only as the referrals are rejected). The University does not provide the supervisors with objective criteria beyond the job description on which to base their selections. From 1972 to 1977 race indicators for blacks were used in referrals of applications by the personnel office to departments; in 1978, the personnel office began to keep race data on all applicants, but did not compile applicant flow data. Intradepartmental promotions are made upon the recommendation of supervisors; there are no written objective guidelines to assist them in making these recommendations. Retirement Program at the University Until September 1, 1972, all hourly employees at the University were excluded from the Teachers Retirement System, the University’s program for provision of retirement benefits based on employer and employee contributions. Even those employees who requested coverage were excluded. The excluded hourly job classifications were filled predominantly with blacks and women. After Title VII became applicable to the defendant government employer on March 24, 1972, the University decided to include all employees in the retirement system. On September 1, 1972, the beginning of a new fiscal year at the University, all employees were required as a condition of employment to participate in the Teachers Retirement System. Previously excluded employees were given the option of paying off the delinquent contributions that they would have paid during the years of exclusion and then to credit those years of pre-inclusion service toward the ten-year vesting requirement and the years-of-service standard that determined the amount of retirement benefits. Eligibility for retirement benefits depended on the employees’ lump sum payment of all the accrued delinquent years of contribution; the hourly employee hired before inclusion in the retirement system could not choose to accept reduced benefits or wait ten years for vesting in lieu of paying the delinquent amount. The University also maintains an Optional Retirement Program in which academic and executive personnel are eligible to participate if they desire. Employees subject to the Classified Pay Plan may not join the optional program, which has a one-year vesting requirement. At the cessation of employment, participants in the Teachers Retirement System are given an exit interview. The personnel manager conducting the interview follows a check-off form of information for the participant, who is told that he may withdraw all contributions he has made to the retirement fund in a lump sum if he signs a waiver forfeiting all interest in prospective retirement benefits. Issues on Appeal We will discuss the contentions of the parties in the following parts of this opinion: I. Class Certification; II. Channeling; III. Retirement Programs; IV. Terminations; and V. Monetary Relief. I. Class Certification The defendant University contends that the district court clearly erred in certifying a class of all black and female past, present, and prospective employees subject to service/maintenance and clerical categorizations in the pay plan and in permitting this class to conduct an across-the-board attack against the University’s unlawful employment practices. The University argues that the three named plaintiffs — all former custodial workers — could not represent class members in the other job categories constituting the class, particularly the clerical ranks. The named plaintiffs are not adequate representatives or share claims typical with those of their class as required by Fed.R.Civ.P. 23(a), the defendant urges, because they were not qualified for initial placement or transfer to positions other than those to which they were assigned; furthermore, they did not seek, would not have accepted, and were not eligible for promotions. The University also asserts that Carpenter and Hunt were terminated for good cause and they thus did not suffer the claimed injury of racially discriminatory discharge similar to that of the other putative class members. In its latest pronouncement concerning Title VII class actions, the Supreme Court holds that the proposed class representative must satisfy each of the Fed.R.Civ.P. 23 requirements, thereby limiting “the class claims to those fairly encompassed by the named plaintiffs’ claims.” General Telephone Company of the Southwest v. Falcon, 457 U.S. 147, 156, 102 S.Ct. 2364, 2370, 72 L.Ed.2d 740 (1982). In Falcon, the Court found that the plaintiff’s complaint provided an insufficient basis for the trial court to conclude that adjudication of his claim of discrimination in promotion would require the decision of common questions of law or fact regarding hiring and that the plaintiff did not suffer injury from the employer’s alleged unlawful hiring practices. The plaintiff could not represent a class alleging “across-the-board” discrimination because his individual claim to discrimination in promotion was not presumptively typical to claims of rejected applicants to employment. Id. at 2371. In East Texas Motor Freight Company v. Rodriguez, 431 U.S. 395, 404, 97 S.Ct. 1891, 1896, 52 L.Ed.2d 453 (1977), the Court emphasized that a class representative “must be part of the class and ‘possess the same interest and suffer the same injury’ as the class members.” In East Texas Motor Freight, the Court refused to order class certification after the lower court denied certification and dismissed all class allegations, because the named plaintiffs lacked the qualifications to be hired in a superior position and thus did not suffer the injury of those locked into inferior positions because of race. 431 U.S. at 403-04, 97 S.Ct. at 1897. The Court noted, however, that “a different case would be presented if the District Court had certified a class and only later had it appeared that the named plaintiffs were not class members or were otherwise inappropriate class representatives.” Id. at 406 n. 12, 97 S.Ct. at 1898 n. 12. In the present case, the district court did not abuse its discretion by initially certifying the class or by subsequently broadening the class to include clerical employees and all employees subject to the classified pay plan. In their initial and amended complaints, Carpenter and Hunt alleged that they had been channelled into low-paying, sex- and race-stereotyped positions, and subjected to discriminatory transfer and promotion practices, and that they had been wrongly terminated because of their sex and race. As a retiree, the plaintiff Williams alleged discrimination against hourly employees due to their prior exclusion from the retirement fund. Both Carpenter and Hunt alleged that the University acted unlawfully in failing to hire or rehire them into higher-payr ing clerical positions. The defendant University strongly argues that hourly employees were nevertheless not proper class representatives of the salaried clerical workers (allegedly composed predominantly of women similarly channelled into these positions, lower paid than those into which white males were channelled). We agree, however, with the district court’s earlier determination that East Texas Motor Freight does not necessarily require a “congruence” that a named plaintiff possess the employment qualifications of all purported class members. The crucial inquiry is whether adjudication of the class representatives’ claims of channeling into the lower-paying service/maintenance positions involves issues of law or fact common to those of women employees channelled into lower-paying clerical jobs. In Falcon, the Supreme Court stressed that allegations of similar discriminatory employment practices, such as a biased testing procedure to evaluate incumbent employees and applicants, or use of entirely subjective personnel processes that operated to discriminate, would satisfy the commonality and typicality requirements of Rule 23(a). 457 U.S. at 159 n. 15, 102 S.Ct. at 2371 n. 15. The plaintiff custodial workers here, who claimed to have been injured by past discrimination in their assignment, pay, promotion and termination, attack specifically the subjective job placement, qualification, and compensation practices of the defendant that similarly apply to and affect clerical workers. The statistical proof of channeling in both the service/maintenance and clerical job positions and the anecdotal testimony of female clerical employees presented at trial set forth common issues of law and fact with respect to the University’s subjective job assignment procedures that resulted in placement of women, as well as blacks, according to stereotyped views of job qualifications. Thus, unlike the disapproved class representatives in Falcon East Texas Motor Freight, the named plaintiffs here satisfied the Rule 23 commonality and typicality requirements and provide adequate representation of the putative class concerning allegations of the discriminatory effects' of channeling in the initial job assignment, transfer, and promotion practices, and concerning the discriminatory impact of the retirement system. Because we affirm the district court finding that there was discrimination in terminations (see Part IV below), we need not consider Carpenter and Hunt’s individual standing to complain of discriminatory termination. At this stage of the proceedings— where the class claim has been determined on the merits after the class was properly certified — the appropriate remedy is not decertification on appeal. The district court may reevaluate the class and the adequacy of representation in light of evidence adduced at trial. See Guerine v. J & W Investment Company, 544 F.2d 863, 864 (5th Cir.1977). See generally C. Wright & A. Miller, Federal Practice and Procedure § 1785, at 137 (1972). However, if after the class has been certified and its claims heard and the representatives are found to be inadequate for some reason during the course of the class claims or during a bifurcated hearing with respect to individual claims, the appropriate step is appointment of new representatives from the existing class, not decertification. See Satterwhite v. City of Greenville, 634 F.2d 231 (5th Cir.1980) (en banc). II. Channeling A. The District Court’s Findings The district court found that the University had violated Title VII by “channeling” blacks and women into low-level jobs, as a result of “employment practices which discourage entry of protected groups into mid-level and high-level jobs and encourage a failure to move the groups up from a low-level entry.” The court concluded that the employer had created a racially stratified and sexually stereotyped work force in which blacks are primarily employed as janitors, cooks, and groundskeepers and women hold most of the clerical positions. Overuti-lization of the protected groups in initial job assignments to the lower echelons, aggregate pay comparisons in which blacks and women earned far less than white males, and the continued stratification of the workforce over a nine-year period (indicating lack of opportunities for promotion) led the court to assume that the University had discriminated in its initial placement of blacks and women. The district court attributed the existence of the channeling cause of action to “the isolation of certain employment practices that disparately impact on protected groups.” According to the court, three practices individually and collectively violated Title VII: educational qualifications for job assignment and promotion that were not proved to be related to job performance; systematic lower compensation of the class by rates set in the Classified Pay Plan; and the University’s subjectivity in initially assigning and promoting employees and placing them on the compensation scale; These practices resulted in overrepresentation of blacks and women in the lower echelons of the employment force, and in stereotyped positions (women as secretaries, blacks as custodians); however, the district court found that, despite segregation of blacks and women within certain job classifications, there was no underutilization of these protected groups in the higher employment levels. The district court found that the job descriptions incorporated the attributes and qualifications of the persons then holding the position at the time the job descriptions were formulated in 1972, instead of criteria actually related to job performance. For instance, the educational requirements set forth in the job descriptions (in particular, the high school requirement) represented “qualifications” that were much more predominantly possessed by white rather than black employees, while in the district court’s determination, moreover, the University did not offer any evidence to sustain its burden of proof that these educational qualifications were job related. That is, the largely predominating incumbent white employees possessed these educational characteristics in 1972 when the job descriptions were formulated — the white employees predominating because of the prior channeling of initial assignments of employees that resulted in minimal or no black representation in the particular higher-paid, classifications in question. The 1972 job-description elevation of these educational characteristics of the incumbent whites into a requirement (not shown to be job-related) for employment in that classification subsequently disparately barred blacks from employment in these higher-paid classifications. With respect to the University’s pay plan, the district court determined that blacks and women are disproportionately distributed in the lowest wage categories (reflecting the fact that they hold the lower level positions), although within individual job classifications no discernible difference in compensation is shown as between the blacks and whites or the males and females. The court also found that, in general, the Classified Pay Plan Committee’s placement of jobs on the pay scale was subjective, without any meaningful correlation made to the wages paid for similar jobs, that there are no objective standards which prescribe an employee’s entry level on the plan, and that the Classified Pay Plan Committee granted a disproportionate number of requested pay grade changes to white males. Finally, in the court’s view, the subjective discretion given to University administrators and supervisors, primarily white males, in implementing and enforcing personnel practices, in combination with the lack of objective criteria for job placement and pay, effected and froze the discriminatory stratification of the workforce. In addition to the subjectivity built into the job classifications and operation of the Classified Pay Plan, the district court emphasized that application procedures whereby the personnel office classified and referred applicants permitted racial considerations to motivate employment decisions. Furthermore, individual supervisors, mainly white males, had discretion to promote without reference to any objective guidelines. In support of its/conclusions, the district court also relied on the shown absence of mobility between salaried and non-salaried classifications, as well as between supervisory and non-supervisory ranks. The district court’s injunctive remedy sought to change the institutional barriers that channelled blacks and women into these low-level positions with no opportunity for advancement. It appointed a special master to structure the necessary relief, at University expense, by achieving four goals: (1) validation and reform of the existing job classifications, so that prescribed qualifications would be job-related; (2) replacement of the University’s present classified employee’s pay plan with a compensation system based on objective, job-related criteria both for compensating similar work at similar rates of pay and also for the initial assignment of employees within the scheme; (3) development of a preference system for currently employed class members (those concentrated in service/maintenance and clerical jobs) to be first in line for promotions or transfers to higher level classification or to new positions for which they are qualified under the new system of job classification; and (4) institution of appropriate written, objective guidelines for supervisors and other University employees who make hiring and promotion decisions, and a record-keeping requirement for personnel decisions relating to the protected class in order to monitor University compliance and the effectiveness of the remedy. B. Analysis: Disparate Impact or Disparate Treatment? A Title YII class action may proceed under the theory of disparate treatment or of disparate impact. Either theo.ry may be applicable to the same set of facts. Wheeler v. City of Columbus, 686 F.2d 1144, 1150 (5th Cir.1982). Generally, the class must prove in a disparate treatment case the employer’s “pattern and practice” of race or sex discrimination, see, e.g., International Brotherhood 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), and establish discriminatory motivation or intent by direct or circumstantial evidence. Id. Under the disparate impact theory, the plaintiff class must set forth discrete, facially neutral practices that have a more severe impact on the protected group than on the unprotected group. See, e.g., Griggs v. Duke Power Company, 401 U.S. 424, 431-33, 91 S.Ct. 849, 854, 28 L.Ed.2d 158 (1971) (high school diploma requirement); Dothard v. Rawlinson, 433 U.S. 321, 329-30, 97 S.Ct. 2720, 2727, 53 L.Ed.2d 786 (1977) (height and weight requirements). The plaintiff is not required to prove the employer’s discriminatory intent in a disparate impact case. The trial court analyzed the evidence in this case under the disparate impact model. It found that three employment practices had adverse impact on the class and individually and collectively violated Title VII. The first allegedly unlawful employment practice, the development of employment and promotion qualifications that incorporated educational requirements which predominated in whites (high school diploma or equivalency), without any showing of job relatedness, is appropriately assessed under the disparate impact theory. See Griggs, supra, 701 U.S. at 432-33, 91 S.Ct. at 854. Were this a case of first impression in this court, we would likewise have concluded that the other channeling practices likewise fell clearly under the disparate impact model, under the literal terms of section 703(a) of Title VII, 42 U.S.C. § 2000e-2(a)(2), since they “limit, segregate, or classify” employees in a manner that “would deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect his status as an employee” because of race or sex. See note 7 supra. Nevertheless, subsequent to the district court’s decision and the lodging of this appeal, a decision of this court, Pouncy v. Prudential Insurance Company, 668 F.2d 795 (5th Cir.1982), to be discussed below, now requires that the discriminatory treatment model, requiring proof of discriminatory intent, be applied in determining whether the obviously disparate effects of the other two channeling practices — the systematic assignment of lower compensated employment to blacks and women in the Classified Pay Plan, and the use of subjectivity in implementation of job qualifications for initial job assignment and promotion and the placement of employees on the compensation scale — nevertheless reflected race and gender discrimination prohibited by Title VII. In Pouncy, we held the disparate impact model of proof is not to be used to challenge multiple employment practices simultaneously, and is not “the appropriate vehicle from which to launch a wide ranging attack on the cumulative effect of a company’s employment practices. Nor may just any employment practice be challenged under this model simply because an uneven racial balance exists in an employer’s work force.” Pouncy v. Prudential Insurance Company, 668 F.2d 795, 800 (5th Cir.1982). See also Pegues v. Mississippi State Employment Service, 699 F.2d 760 (5th Cir.1983). According to Pouncy, a subjective classification practice that depends on the employer’s discretionary decisions is not included within the category of facially neutral procedures — such as the high school educational requirement in this case — whose discriminatory impact may be isolated and thus specifically shown to have a causal connection to a class-based imbalance in the work force so as to require no further proof of discriminatory motivation or intent. In Pouncy, the court determined that employment practices (similar to those complained of in the present case) — the failure to post job openings, use of a level system, and evaluations of employees with subjective criteria — that resulted in the overrepresentation of the protected group in the lower levels of the work force, did not constitute “proof of a causal connection” under the disparate impact model sufficient to establish employment discrimination prohibited by Title VII. 668 F.2d at 801. The difficulty we face with regard to the district court’s findings as to the discriminatory Title VII violations through the latter two channeling practices is that, measuring Title VII violation under the disparate impact model, it made no express findings as to discriminatory intent motivating them. Since under Pouncy the Title VII discrimination must be measured under the disparate treatment model, “[p]roof of discriminatory motive is critical”, although, however, “in some situations it can be inferred from the mere fact of differences in treatment.” Teamsters, 431 U.S. at 333 n. 15, 97 S.Ct. at 1854 n. 15. From the record before us, discriminatory intent might be inferred from the gross race and gender stratification disparities that resulted from entrustment of assignment and promotion determinations to the purely subjective decision of the University’s white supervisors who made such determinations on behalf of the University. However, for reasons to be stated, we cannot find that the district court even by implication made a finding of discriminatory intent — perhaps solely because, under its view of the decisional law then in effect, it was unnecessary for it to do so. Since discriminatory intent is a factual finding even though made by inference, see Pullman-Standard v. Swint, 456 U.S. 273, 102 S.Ct. 1781, 72 L.Ed.2d 66 (1982), we believe that it should in this case be made in the first instance by the trial court, subject to review under the clearly erroneous standard, id., rather than made independently by the appellate court on review, id. Much as we may regret to prolong this litigation by doing so, we ultimately conclude that we should remand to the district court for a finding as to the requisite discriminatory intent to cause the latter two channeling practices to constitute violations of Title VII. In summary, we conclude that, in determining whether there is Title VII liability for a discriminatory employment practice, we will review the district court’s finding with respect to the defendant University’s educational requirements under the disparate impact theory, but we must review under the disparate treatment model the sufficiency of the other statistical and anecdotal evidence that supports the findings of discriminatory channeling to lesser employment opportunities. The plaintiff class argues, we must note, that its proof supports findings of Title VII employment discrimination under either discriminatory impact or discriminatory treatment. Although we ultimately conclude that the discriminatory treatment standard, requiring proof of discriminatory intent, is not met with respect to the channeling practices other than that pertaining to non-job-related educational requirements — and therefore remand is required for finding as to discriminatory intent as to them — we will review all channeling issues argued on appeal, in an effort to provide guidance to the district court on remand and, possibly, to avoid yet a further appeal as to those issues. C. Disparate Impact of the Educational Requirements The plaintiff class attacked at trial the high school educational requirements for many of the jobs at the University because of their disparate impact on blacks in the work force. To establish a prima facie case of discrimination, the plaintiff need only show that the facially neutral employment standards operate more harshly on one group than another. See Dothard v. Rawlinson, supra, 433 U.S. at 329, 97 S.Ct. at 2726-27; Wheeler v. City of Columbus, supra, 686 F.2d at 1150. The burden then shifts to the employer to show that the specific requirement has “a manifest relationship to the employment in question,” Griggs v. Duke Power Co., supra, 433 U.S. at 432, 91 S.Ct. at 854, for if “an employment practice which operates to exclude [the protected class] cannot be shown to be related to job performance, the practice is prohibited.” Id. Besides providing evidence of the business necessity of the qualification, thereby validating it for purposes of the disparate impact theory, the employer may also attack the plaintiff’s case by showing “the total unacceptability of the plaintiff’s statistical proof.” Johnson v. Uncle Ben’s Inc., 628 F.2d 419, 424 (5th Cir.1980), vacated on other grounds, 451 U.S. 902, 101 S.Ct. 1967, 68 L.Ed.2d 290 (1981). The district court based its conclusion that the educational requirement violated Title VII on its findings that, in the relevant market, more whites graduate from high school than do blacks, and more whites and males are college graduates than are blacks and women; the job qualifications were prescribed solely by describing the educational status of the current holder of the position at the time the job descriptions were formulated (and when the higher paid and more favorable positions were almost exclusively held by white males); and the University produced no evidence showing the job relatedness or business necessity of the educational requirement. At trial, the plaintiff class presented data showing that 47% of whites and 24.8% of the blacks over the age of 25 in the population of the county surrounding the University have high school degrees. In Griggs, a diploma requirement was found to have disparate impact because 34% of the white males in the state had completed high school, while only 12% of black males had graduated. 401 U.S. at 430 n. 6, 91 S.Ct. at 853 n. 6. See also Johnson v. Goodyear Tire & Rubber Company, 491 F.2d 1364, 1371 (5th Cir.1974) (high school requirement enjoined where- 25.3% of black and 45.8% of white population of city area had diplomas). The racial composition of classified positions requiring a high school diploma is primarily white — in 1979, 258 whites and 54 blacks held these positions (in 1972, 204 whites and 13 blacks). The showings that over a third of the classified positions have an educational requirement and that over 80% of the jobs requiring high school diplomas were held by whites, also exhibit the harshly disproportionate impact of the requirement on blacks. The University argues that the plaintiff’s data did not support a prima facie case of adverse impact, that no “otherwise qualified” employee was denied employment in or promotion to a certain position because of a degree requirement, and that the education and experience requirements were “validated” by the personnel office. As we have stated, the plaintiffs clearly showed the disparate impact of the educational qualifications. Indeed, the significant disparity shows in itself that “otherwise qualified” black employees may have been excluded; it is the defendant’s burden to show that the requirements are necessary to job performance. The defendant maintains that adverse impact of the requirement is not present because there is no underrepre-sentation of the protected group; however, in Connecticut v. Teal, 457 U.S. 440, —, 102 S.Ct. 2525, 2533, 73 L.Ed.2d 130 (1982), the Supreme Court held that “bottom line” adequate minority representation did not validate selection requirements that disproportionately excluded individual members of the class. The University did not present any specific evidence of the job relatedness of the educational requirement — that it bears “a demonstrable relationship to successful performance of the jobs for which it was used,” Griggs, 401 U.S. at 431, 91 S.Ct. at 853 — except to assert that a supervisory employee benefits from formal education. According to the University’s business manager, the University had never attempted to validate its educational requirement in accordance with Equal Employment Opportunity Commission regulations, see 29 C.F.R. § 1607 et seq. (1978), which, in Johnson, supra, compelled the court’s conclusion that the requirement was invalid. 491 F.2d at 1371. The defendant’s additional assertion that some of the positions would permit a high school diploma or its equivalent in experience does not rebut the plaintiffs’ showing, because again the University presented no data as to number of such positions of the number of employees for whom the formal degree requirement was waived. We thus find that the district court was not clearly erroneous in determining that the plaintiff class established a prima facie case of the adverse impact of the educational requirement that the defendant did not rebut. Nor did the court abuse its discretion in fashioning relief to require the University to validate the job qualifications and provide preferential treatment to those employees who had wrongfully been denied more advanced job assignments. D. Disparate Treatment and the Channeling Claim We review the remaining allegedly unlawful employment practices of the defendants under the disparate treatment theory, which requires proof of the employer’s discriminatory intent. The plaintiffs must demonstrate more than “the mere occurrence of isolated or ‘accidental’ or sporadic discriminatory acts.” Teamsters, supra, 431 U.S. at 336, 97 S.Ct. at 1855. The Title VII violation is founded on a “pattern or practice” of discrimination, and the plaintiffs must establish “by a preponderance of the evidence that racial discrimination was [the employer’s] standard operating procedure— the regular rather than the unusual practice.” Id., 431 U.S. at 436-37, 97 S.Ct. at 1854-55. In some cases, statistical proof of a substantial or “gross” disparity in treatment of protected and unprotected groups may make out the plaintiffs’ prima facie case and justify the inference of discriminatory animus. Hazelwood School District v. United States, 433 U.S. 299, 307-08, 97 S.Ct. 2736, 2741, 53 L.Ed.2d 768 (1977); Payne v. Travenol Laboratories, Inc., 673 F.2d 798, 817 (5th Cir.), cert. denied,-U.S.-, 103 S.Ct. 451-52, 74 L.Ed.2d 605 (1982). If the statistical disparity is deficient, the class may “buttress” its case with evidence of “a history of discrimination practiced by the employer, individual instances of discrimination, and opportunities to discriminate that exist in the employer’s decision-making processes.” Payne, supra, 673 F.2d at 817. The defendant employer may rebut the plaintiffs’ prima facie case by either discrediting the statistical evidence as being “inaccurate or insignificant” or by advancing a non-discriminatory rationale for the presumptively discriminatory treatment of the employee. See, e.g. Teamsters, supra, 431 U.S. at 360 n. 46, 97 S.Ct. at 1867 n. 46; Wheeler, supra, 686 F.2d at 1152; Pegues, supra, 699 F.2d at 766. In the present case, the defendant does not defend the specific job practices (except job qualifications, discussed above) on which the district court focused; rather, it asserts that there can be no Title VII violation where there is over-representation of protected groups in the lower employment echelons, so long as these groups are not underutilized in the higher levels. On the other hand, the plaintiff class on appeal attacks the specific employment practices that discriminatorily isolate and concentrate the class in these lower levels with little opportunity for advancement. The University bases its arguments on the qualified labor force comparisons about which both parties testified at trial, as summarized by the district court. The University utilized the standard Equal Employment Opportunity (EEO) broad job occupational qualifications that describe availability for six standard occupational and professional job classifications: (1) executive, administrative, and managerial; (2) other professionals not within EEO(l); (3) technical and para-professional skilled employees; (4) secretarial and clerical; (5) skilled crafts workers; and (6) service and maintenance personnel. It then compared the representation of currently-employed blacks and women at the University in these categories, to the availability of blacks and women qualified for employment in these categories in Angelina and Nacogdoches counties in Texas, the relevant geographical labor market. The county-wide availability data was based on United States census data gathered and reported in the 1970 census. Overall, according to the 1970 census, 18.77% of the population and 18.14% of the entire labor force was black in these counties. Women constituted 51.51% of the population but only 36.54% of the labor force. The EEO data from the counties showed low availability of qualified blacks and women in the upper categories (e.g., in EEO(l), supra, only 3.33% blacks were available for or employed in executive, administrative and managerial positions), and high availability in the relatively low-skill categories (e.g., in EEO(4), secretarial/clerical, 3.02% available or employed persons were black and 72.12% were women; in EEO(6), service/maintenance, 34.01% black and 67.2% women). The district court compared these availability figures within the counties in 1970 to the similarly-categorized workforce at the University and determined the actual and expected (the EEO census percentage in each category times the total University workforce) percentages of blacks and women employees hired into these categories in the years 1972-79. It concluded that the data did not reveal any “consistent failure to have black persons in the upper job categories”; nor was there observed underutili-zation of blacks or women in any of the EEO groups. The district court noted, however, consistent overutilization of blacks and women in certain broad EEO categories. Blacks were overrepresented in hiring in the EEO(6) service/maintenance and the EEO(2) professional, non-executive categories. Women were overrepresented in hiring in the EEO(4) secretarial/clerical category and slightly overrepresented in EEO(l) executive-administrative and EEO(S) skilled-crafts hires in a couple of the years, as well as slightly underutilized in the EEO(3) technical-skilled and EEO(6) service-maintenance categories. The statistical significance of oVerrepre-sentation of blacks and women in certain of these EEO categories was evaluated by the court. The court looked at the standard deviation between actual and expected hires. Standard deviation analysis is a quantification (a binomial distribution) applied to statistical disparities to eliminate chance as a likely explanation for the revealed difference between an expected outcome and the observed outcome (e.g., here, the overrepresentation of blacks and women in certain job categories, as compared to their representation in the overall geographical availability pool). See, e.g., Castaneda v. Partida, 430 U.S. 482, 496, 97 S.Ct. 1272, 1281, 51 L.Ed.2d 498 (1977); Pegues v. Mississippi State Employment Service, supra, 699 F.2d at 768-769 n. 9. See generally D. Baldus & J. Cole, Statistical Proof of Discrimination § 9.03, at 293-96 (1980 & Supp.1982). As the number of standard deviations increases, the probability decreases that mere chance has caused the difference between the expected and observed outcome. See id. Generally, a difference of 2 or 3 standard deviations supports the inference that unlawful discrimination-was a reason for the discrepancy. See Castaneda, supra, 430 U.S. at 496 n. 17, 97 S.Ct. at 1281 n. 17; Pegues, supra, 699 F.2d at 768-69, & n. 9. Here, the district court found that the standard deviation for hiring blacks to service/maintenance positions for the years 1972 to 1979 always exceeded 2, and more frequently approximated 6. The court did not state the standard deviation for placement of women hires in secretarial/clerical posts (although the trial exhibits on which the court adduced this data — while showing overrepresentation — set forth a standard deviation of less than 2). Stratification through overrepresentation of protected groups in the lower positions of the workforce has routinely been held actionable under Title VII, if resulting in decreased employment opportunities. See, e.g., Johnson v. Uncle Ben’s, Inc., 628 F.2d 419, 423 (5th Cir.1980); Fisher v. Procter & Gamble Manufacturing Company, 613 F.2d 527, 543-44 (5th Cir.1980); James v. Stockham Valves & Fittings Company, 559 F.2d 310, 328 (5th Cir.1977), cert. denied, 434 U.S. 1034, 98 S.Ct. 767, 54 L.Ed.2d 781 (1978); Pettway v. American Cast Iron Pipe Company, 494 F.2d 211, 226 (5th Cir.1974), cert. denied, 439 U.S. 1115, 99 S.Ct. 1020, 59 L.Ed.2d 74 (1979). The defendant maintains that these cases have also involved underrepresentation in the upper levels of the same groups that were overrepresented in the lower levels. The University makes the facially sensible argument that it is guilty merely of providing extra job opportunities to blacks and women. What the University’s argument overlooks is that, as found by the district court, the specific employment practices that channelled women and blacks into stereotyped lower-level classifications likewise deprived them of the opportunity to be employed or to be promoted to the higher level positions, regardless of their job-related qualifications for the higher level position. Title VII forbids an employer to “limit, segregate, or classify” employees “in any way which would deprive or tend to deprive any individual employment opportunities or otherwise adversely affect his status as an employee” because of race or sex. Section 703(a)(2), 42 U.S.C. § 2000e-2(a)(2). In this case, the University’s placement and promotion practices are similar to the employer’s personnel policies in James, supra, 559 F.2d at 310, where Title VII liability was found in the imbalance of black concentration in lower skill and pay levels. In James the plaintiff class made a prima facie showing of discrimination based on “a definite pattern of intentioned racial staffing . .. revealed by statistical evidence in the disparities in black and white representation in seniority departments, on the relatively few integrated jobs in 1973, on the concentration of blacks in the lower job classes, and on the wage disparities between black and white.” 559 F.2d at 328. Thus, on the statistical level, the segregation of job classifications (which is the case here), with blacks being assigned to lower-paying positions, bolstered the finding of discrimination in assignment of blacks to these positions. In addition, the James court considered persuasive the imbalance between the high percentage of black workers in the low skill maintenance department and the totality of white supervisors in that department. 559 F.2d at 345. Personnel policies in James were tainted by the subjectivity of employment decisions for initial assignment and promotion. As in the present case, a central personnel office screened applications and made referrals. The educational and experience qualifications for certain positions were not validated or were waived for some white applicants, as is the case here. The supervisors had discretion to select employees without the benefit of formal guidelines; promotion and appointment of the supervisory staff similarly depended upon discretionary determinations by white supervisors and administrators, providing a “ ‘ready mechanism’ for discrimination.’ ” 559 F.2d at 349. See also Payne, supra, 673 F.2d at 815; Fisher v. Procter & Gamble Mfg. Co. supra, 613 F.2d at 549; Rowe v. General Motors Corporation, 457 F.2d 348, 356 (5th Cir. 1972). The evidence in the present case that the personnel office noted race on referral applications, and the testimony of class members regarding the higher level jobs for which they applied, as compared to those lower-level positions to which alone (despite any qualifications for the high level employment) they were referred and in which they were ultimately placed, indicates to us that the'lack of objective criteria in the system encouraged a pattern and practice of discriminatory employee assignments stereotyping black and female applicants for consideration only for the stratified lower level positions predominantly filled by members of their race or gender. The University’s Classified Pay Plan seems also to be marred by discretionary decisions on job ranking (level of pay, primarily) as determined by the all white, predominantly male, committee administering the plan. Moreover, the system of subjective initial placement of a new employee on the pay scale and the evidence that white males were most often granted appeals of their pay grade, provides the explanation for instances of discrimination, especially in light of the stratified workforce. Although no problem is shown with regard to discriminatory compensation within job classifications, nevertheless the placement of the primarily race and sex segregated classifications on the lower end of the pay scale was not made with regard to any articulated standards. We do not, however, rely primarily on the district court’s determination that on the whole, blacks and women received significantly less compensation than white males, for this data made no comparison of employee pay based on the level of skill, education and training, see Pouncy supra, 668 F.2d at 802. We do find, however, that the district court acted within its discretion in fashioning injunctive relief requiring the employer to modify its subjective practices and with regard to a pay plan upon compensating similar work at similar rates of pay in accordance with objective criteria — to replace the purely subjective characterizations that resulted in disparate effect upon the classifications in which the protected groups predominate because of the University’s channeling practices. The district court’s findings regarding the University’s promotion practices are unclear. The evidence adduced at trial is that there is little mobility within the job ranks or classifications, so that employees generally stay in their initial assignment into the workforce, and that educational requirements for certain job classifications (presumably, the higher levels into which employees would desire promotion) has a disparate impact on class members. The plaintiff class did not present statistical evidence of the required comparison to a qualified pool of employees presumptively eligible for promotion. See Pouncy, supra, 668 F.2d at 803; cf., Wilkins v. University of Houston, 654 F.2d 388, 401 modified on other grounds, 662 F.2d 1156 (5th Cir.1981), cert. denied,-U.S.-, 103 S.Ct. 51, 52, 74 L.Ed.2d 57 (1982). Since the plaintiffs did not present evidence that the University obtains its higher-level employees from promotions within the workforce, we cannot use the disproportionately black and white-composed lower level job classifications as the qualified labor pool to determine if promotion discrimination exists. Payne, supra, 673 F.2d at 825; see, e.g., Johnson, supra, 628 F.2d at 423; Fisher, supra, 613 F.2d at 542. Nevertheless, despite the EEO comparisons upon which the district court based its conclusion of no underutilization of blacks or women in higher job categories, a marked absence of blacks and women in the University-designated higher executive levels indicates a lack of promotion opportunities for those in the “dead end” jobs in the lower echelon race- and sex-segregated classifications. The absence of mobility, caused by subjective promotional standards and adversely-impacting nonemployment-related job qualifications, exacerbates the injury inflicted by the discriminatory channeling that results from the subjective initial job assignments. We therefore do not disturb the district court’s characterization of the practice of channeling, and find no abuse of its discretion in fashioning an injunctive remedy that requires, inter alia, adaptation of the University’s promotion policies to address the adverse effects of the channeling practices. III. Discrimination in the Retirement Programs The plaintiffs allege that the structure and implementation of the University’s retirement programs violate Title VII and the Equal Protection and Due Process Clauses of the fourteenth amendment to the Constitution. They contend that the pre-Title VII exclusion of hourly employees was applied in a discriminatory manner because at least one hourly white employee was included at that time and that white employees’ positions were reclassified as salaried (in particular, white supervisors were made salaried, while black supervisors remained hourly employees) in order to bring about the preferential coverage of whites. They protest the exclusion of hourly employees from March, 1972 (the date the University became subject to Title VII) to September, 1972 (the date coverage began). The plaintiffs argüe that current operation of the University’s retirement plan constitutes a “continuing violation” of Title VII because the pre-Title VII exclusion of hourly employees — predominantly blacks and women — has resulted in retirement plan procedures that have an adverse impact against them in comparison to those always included in plan coverage or those hired after Title VII. The retirement plan’s regulations which require the previously excluded employee to deposit all delinquent contributions for the years of exclusion before he may become eligible for any benefits cause a hardship that has an adverse impact on protected class members, the plaintiffs argue, for in the two months following plan coverage of hourly employees, all but 6% of those joining the Teachers Retirement System (presumably, in most part, those who were previously excluded) were black or female. The plaintiffs urge also that post-Title VII facially neutral aspects of the University’s retirement practices have an unlawful disparate impact on blacks and women. Since the executive and faculty levels of the University are covered by the Optional Retirement Program — and these groups predominantly are composed of white males, — blacks and women are disproportionately excluded from this more attractive benefit plan. The procedure whereby the exiting employee may recover all contributions to the fund in exchange for executing a waiver of potential retirement benefits has a similarly disparate impact, the class contends, for between 1972 and 1979, 75% of black hourly and 50% of the white female hourly employees signed the waiver, while only 25% of the white male hourly and 27% of the white male salaried employees (there were no black salaried retirees) executed the waiver. Members of the class testified at trial that during their termination interviews they received no explanation of the potential detrimental effects of signing the waiver of future benefits; the plaintiffs contend that this failure to advise constitutes deprivation of property without due process in violation of the fourteenth amendment. In rejecting these contentions, the district court found the University’s operation of its retirement program to be free of discriminatory intent, and that it did not violate Title VII (the court did