Full opinion text
MEMORANDUM NEWCOMER, District Judge. This employment discrimination class action was tried to the Court in December, 1982. The following shall constitute my Findings of Fact and Conclusions of Law. Findings of Fact 1. This is a Title VII and § 1981 class action brought on behalf of all black persons who unsuccessfully sought employment at the Fairless Hills, Pennsylvania plant of United States Steel Corporation (“USS”) between July 11,1972 and the present. The class described above was certified by Order dated August 26, 1980. 2. The defendant, USS, is a corporation doing business in the Commonwealth of Pennsylvania, in many other states, and in foreign countries. Throughout the class period, USS owned and operated steel manufacturing and other facilities located in Fairless Hills, Pennsylvania and employed from 7,000 to 10,000 persons there. USS is an “employer” within the meaning of Title VII. (Unless otherwise specified, references hereinafter to USS shall be understood to refer specifically to the Fairless Hills facilities of USS.) 3. Named plaintiff Elbert Green is a black male who unsuccessfully sought employment as a laborer in USS’s Production & Maintenance department (“P & M”) by application dated April 11,1973. On May 7, 1973, Green filed with the Equal Employment Opportunity Commission (“EEOC”) a timely charge of race discrimination in employment against USS, and subsequently he received a notice of his right to sue in federal court. 4. Green was an unnamed member of the plaintiff class of unsuccessful black job applicants at USS in Dickerson v. United States Steel Corp., C.A. No. 73-1292, a Title VII and § 1981 race discrimination suit. 5. On November 29, 1976, following the severance of the applicant class from the Dickerson lawsuit, Green filed this action on his own behalf and on behalf of the applicant class. 6. Named plaintiff Robert Danley is a black male who unsuccessfully sought a P & M job by application dated June 4, 1976. 7. Danley was also initially an unnamed class member in Dickerson. Following his receipt of notice of the severance of the applicant class from that action, Danley filed with the EEOC a timely charge of race discrimination in employment against USS and received a notice of his right to Sue in federal court. 8. When Green filed this action on behalf of the applicant class, Danley became an unnamed class member. On January 10, 1979, Danley filed a motion to intervene as a named plaintiff. Prior to a ruling by the Court, the parties stipulated that Danley could serve as Green’s co-representative of the applicant class. 9. By a Memorandum and Order dated August 29,1979,1 entered partial summary judgment in favor of USS on Green’s individual claim. Discovery had established that Green had falsified his employment history on his job application. I held that this evidence satisfied USS’s burden of articulating a legitimate, nondiscriminatory reason for its failure to hire Green. I reserved for trial the questions whether plaintiff was qualified for the job for which he applied; whether USS was hiring anyone for those jobs at the time he applied; whether it was hiring for those jobs immediately after his rejection; and whether, if plaintiff could establish a prima facie case, plaintiff could establish that USS’s articulated reason for rejecting him was a standard applied equally to all races or was, rather, a mere pretext for racial discrimination. 10. In my August 26, 1980 class certification order, I ruled that Green could not serve as a class representative because his claim, presenting the narrow and atypical issue whether USS had uniformly and nondiscriminatorily rejected applicants who falsified their job applications, was not appropriate for class treatment. Danley thus became the sole class representative, and Green remained a named plaintiff on his individual claim only. 11. At trial, plaintiffs presented no evidence relevant to Green’s individual claim. 12. On the first day of trial, plaintiffs narrowed their case to include only those blacks who unsuccessfully sought employment in P & M jobs during the class period. Plaintiffs also indicated that they were not pressing their case as to the years 1975, 1980, and 1981, when very little hiring was done for P & M jobs at Fairless Hills, or as to the years 1976 and 1977, when plaintiff’s evidence showed that a moderate amount of hiring was done but that blacks were hired at levels close to or exceeding their representation in the applicant pool. Plaintiff’s case therefore focussed on the question whether USS unlawfully discriminated against black P & M applicants at Fairless Hills in the years 1972-74, 1978, and 1979. 13. Plaintiff Danley did not testify at trial, and plaintiffs presented no proof directly relating to him as an individual, other than his application for employment and a stipulation relating to his satisfaction of the jurisdictional prerequisites. His application date of June, 1976 places him among those class members who applied to USS at a time when plaintiff’s statistical evidence shows that USS was hiring blacks at rates equalling or exceeding their representation in the applicant pool. 14. Plaintiffs presented the testimony of two unnamed class members, Christine Davidson and Frederick Wright. Ms. Davidson, a black woman, submitted an application for a P & M job on August 24, 1976. She had a GED degree, vocational training, and a record of steady employment. Mr. Wright, a black man, submitted an application for a P & M job on June 8, 1977. He had several years of college education, prior work experience as a laborer, and a record of steady employment before, during, and after college. Although both Ms. Davidson and Mr. Wright possessed the qualifications necessary to be considered for employment in an entry-level P & M job, neither was hired by USS. Thereafter, USS continued to hire persons for entry-level P & M jobs. 15. The P & M job category is the largest at USS. Approximately 95% of new hires (that is, people hired from outside the company rather than promoted or transferred from within) in P & M during the class period were hired for the position of laborer, an entry-level unskilled job. About 4% of new hires were for skilled jobs in the trade and craft category, although USS usually filled these positions through its internal apprenticeship programs. P & M jobs other than those in the laborer or trade and craft categories were customarily filled from within the plant, by a posting and bidding system, although in rare instances individuals were hired from outside to fill such jobs. 16. Approximately 15% of new P & M hires during the class period were hired for summer jobs. Virtually all of these jobs were in the laborer category. 17. USS’s minimum hiring criteria for all P & M positions remained constant throughout the class period. Anyone hired had to be at least 18 years old, pass a physical examination, and be able to read safety signs (a requirement so minimal that defendant’s witnesses could recall no one ever having failed to meet it). Additional minimum requirements applied to those seeking skilled trade and craft jobs. However, as long as an applicant met the three minimum requirements described above, there was no single quality or trait which would make an applicant ineligible per se for a laborer job. Cf. P-55 (USS policy restricting employment of relatives within a single supervisory unit); P-56 (USS policy stating that USS would “usually” not consider applicants who had been convicted of arson, malicious destruction of property, or acts of sexual perversion); P-57 (USS policy stating that alien applicants should be required to produce appropriate documentation). 18. In its newspaper advertisements for laborers, USS stated that there was “no experience necessary,” and that USS required only “common sense and a desire to work” (P-114). In fact, many of those hired into P & M laborer jobs during the class period had no prior work experience. USS expected new hires for P & M jobs other than those in the trade and craft category to learn any skills necessary to perform their jobs, or other jobs they might be promoted into, through on-the-job training or classroom instruction provided by USS. 19. Throughout the class period, USS’s official policies relative to hiring new employees were set forth in the following documents: Exhibits P-43, P-53, P-54, P-55, P-56, P-57, P-58, P-59, and P-64. Exhibit P-43, entitled Procedure for Uniform Employment Practices, set forth USS’s official policy of nondiscriminatory hiring of the best qualified candidates. 20. USS did not conduct any training sessions to indoctrinate the many individuals responsible for implementing its hiring policies as to their meaning and application. 21. During the class period, USS received written employment applications on a variety of different forms, all of which requested the applicant to list basic background information concerning schooling, prior employment, military record, age, physical condition, and whether the applicant had any relatives in USS’s current or past work force. Such applications were received by mail or hand-delivery, or might be filled out in USS’s personnel office by individuals who stopped by to inquire about employment. USS accepted new applications from persons who had previously applied but had not been hired. 22. Upon receiving a job application, USS’s practice was to record it in a business record called the Job Applicant Log (“JAL”). The following information was entered in the JAL from each application: date of application; name of applicant; EEOC code (indicating race and sex); job applied for; and comments. 23. Separate JALs were prepared for summer applicants for 1972,1973, and 1974, the only years during the class period in which USS made summer hires. 24. Larry Edwards, who was USS’s Supervisor of Employment until 1975 and General Supervisor of Personnel thereafter, and Mabel Williams, who became Supervisor of Employment when Edwards was promoted in 1975, both instructed personnel office employees repeatedly to maintain the JALs as promptly and accurately as possible. 25. On occasion, a backlog of one or more weeks developed in the entry of application information into the JAL. All such information was, however, ultimately entered into the JALs. Any errors in, or omissions from, the JALs were due merely to normal clerical mistakes. From their personal experience with the JALs, Mr. Edwards and Mrs. Williams believed that no errors or omissions in the JALs occurred in a racially disproportionate manner. That is, they understood that blacks were included in the JALs with the same frequency as non-blacks. 26. In interrogatory answers, USS stated under oath that information concerning applicant flow should be obtained from the JALs (USS’s Response to Interrogatory 4 of Plaintiffs' Third Set of Interrogatories). At trial, Mr. Edwards testified that all existing information on applicant flow at USS during the class period is in the JALs. During the class period USS used applicant flow data from the JALs to defend against 43 EEOC charges of race discrimination in hiring. USS also used applicant flow data derived from its JALs to prepare its Affirmative Action Quarterly Reports (AAQRS), which it filed with the United States government pursuant to certain consent decrees entered into in the course of prior litigation. 27. Mr. Edwards and Mrs. Williams reviewed newly submitted applications from time to time and would occasionally single one out for special consideration. One reason why an application might receive special consideration would be that the applicant had a relative employed at USS. Such applications were moved ahead of other applications and placed immediately in the “ready” file—that is, the file of processed applications which had been cleared as worthy of consideration and which were ripe for the scheduling of an interview. 28. The “comments” section of the JALs contain occasional entries indicating that Edwards had given the applicant special consideration. 29. Those applications which were not placed immediately into the “ready” file were placed into general application files which were segregated according to the EEOC category and job preference of the applicant. An inexact system of prioritizing was employed with respect to applicants who fell into more than one category. For instance, a black male veteran’s application might be filed in the “black males” category or the “veterans” category or might be duplicated and filed in both. USS also often duplicated and cross-filed applications of individuals who applied for more than one type of job, for example, Production & Maintenance or Clerical & Technical. 30. Prior to about 1977, USS followed a general policy of considering for employment persons whose applications had been submitted, to USS within the previous twelve months. Therefore, the general application files were reviewed periodically to remove those applications which were more than a year old. In about 1977, USS adopted a policy in which applications were supposed to be considered for employment for only 120 days, unless renewed by the applicant by a telephone call or visit to the plant. That 120-day policy was not consistently followed, however. Even after 1977, applications were considered for hiring purposes beyond the 120-day period, and they were kept in the general application file for one year. 31. USS personnel clerks and interviewers, with the supervision of Mr. Edwards and Mrs. Williams, were responsible for pulling applications from the general applications files and placing them in the “ready” file. The clerks were instructed to maintain a “good mix” of applications in the “ready” file—-that is, the “ready” file was to contain representative applications from each of the EEOC categories so that those called in for interviews would be a racially balanced group. Mr. Edwards testified that he instructed the personnel department employees to see that 20-30% of those interviewed were minorities. 32. USS hired applicants in response to personnel requisitions submitted to the personnel department by the various operating departments. Upon receiving a personnel requisition, USS’s personnel clerks or interviewers would select the approximate number of applications from the “ready” file which experience told them was necessary to yield the desired number of hires. These applicants were then contacted and invited to schedule interviews. 33. USS personnel department employees compiled a business record called the Daily Call-In List reflecting the names of applicants scheduled for interviews and the dates of their interviews. After 1974, the Daily Call-In List included a notation of the race of each applicant interviewed. 34. Some applicants scheduled for interviews did not report for the interview and were eliminated from consideration. A record was kept reflecting which applicants did not report for interviews. 35. When an applicant showed up for a scheduled interview, his or her preliminary interview was conducted by a USS personnel department interviewer. The applicant was then interviewed by the foreman or general foreman of the operating department which had requisitioned employees. The foreman or general foreman had the authority to reject the applicant, or to accept him or her, subject to the applicant’s successfully completing a physical examination. When a foreman or general foreman turned an applicant down, Mr. Edwards or Mrs. Williams would sometimes review the reasons for the rejection and possibly refer the same applicant to another operating department which needed new employees for another interview. 36. Over the class period, between 50 and 100 USS foremen or general foremen and some 12-15 USS personnel department interviewers interviewed applicants for work at USS. During that period, for example, one foreman conducted some 600 interviews. 37. Although the USS Procedure for Uniform Employment Practices required all interviewers to be trained, USS did not train its interviewers other than through on-the-job exposure to other interviewers’ techniques. Nor did USS have any minimum education or work experience requirements for persons hired as interviewers. 38. Despite the requirements in the Procedure for Uniform Employment Practices that “an investigation is to be made of each individual being considered for employment” (P—43, ¶ I), USS made no such investigations. 39. When making hiring decisions, USS’s evidence was that its interviewers evaluated applicants using the following hiring criteria: education; work experience; attitude; initiative; personality; ability to take directions; alertness; intelligence; physical fitness (that is, physical restrictions); ability to communicate; military experience or training; vocational training; ability to meet work schedules; prior criminal record; familiarity with industrial or factory work; interest; prior employment history at Fairless Works; personal references; citizenship or alien status; having a relative in the USS work force; and age. Most of these criteria are wholly subjective. As to those criteria which can be objectively measured, such as education and work experience, USS employed no uniform system in assessing one applicant’s qualifications vis-a-vis another’s. Nor were the various interviewers and foremen trained in any way as to what the different criteria meant or how to implement them uniformly, although many of them are vague and all are subject to differing interpretations. 40. USS took no steps to validate its hiring criteria or to ascertain whether they were job-related. 41. USS’s own Procedure for Uniform Employment Practices cautioned against testing applicants for “mental alertness and general ability” unless those factors were demonstrated to be work-related. (P—43, ¶ H.3.) 42. Under consent decrees entered into in the course of prior litigation, USS was bound not to use employee selection criteria “unless such procedures have been validated in accordance with the Equal Employment Opportunity Commission’s ‘Guidelines on Employee Selection Procedures’ (29 C.F.R. § 1607).” (P-112, pp. 431:139 and 431:151.) USS made no attempt to validate its employee selection criteria in accordance with EEOC regulations. 43. USS’s Procedure for Uniform Employment Practices requires written documentation of the reasons why any applicant is rejected. (P-43, ¶¶ D.8, G.3, J.3.) The policy requires such documentation “to substantiate conformity with the code of nondiscriminatory treatment ... of applicants,” (P—43, ¶ D.3), to provide “a later justification of selection should it be questioned,” (P—43, ¶6.4), and to serve as “check points to insure that the processing described in (P-43) is completed for all new employees” (P-43, ¶ J.3.a). Despite the Procedure’s requirement for documenting the reasons an applicant was not hired, USS only unevenly and imperfectly recorded those reasons. Nor did USS’s personnel department regularly review the reasons for rejection that were occasionally recorded. Finally, most of those documents which did contain interview comments were destroyed by USS. 44. Contrary to the requirements of its own personnel policies, USS maintained no system to determine whether its subjective hiring criteria were being fairly and nondiscriminatorily applied to all applicants. 45. Applicants who successfully completed the interview procedure were extended job offers conditional upon passing a physical examination. 46. Besides those applicants who were scheduled but did not report for interviews, some applicants voluntarily withdrew at some stage of the interview/physical-examination process. Others would decline job offers when they were made, or would accept offers but fail to report for work when scheduled. 47. When an individual was hired and reported for work, he or she would be recorded in a business record called the Daily Employment Report (DER). Those who failed to report as scheduled would be logged in with notation “DNS” (for “did not start”). The DER recorded the following information about those hired: pay number; name; starting date; employee status; department; race code; remarks; supervisor; and social security number. 48. USS prepared AAQRs for P & M applicants and hires for the periods January 1, 1972—December 31, 1973 and July 1, 1974—April 30, 1982. The JALs and DERs were key data sources for these reports, which USS filed with the United States Government. 49. USS destroyed thousands of job applications submitted during the portion of the class period prior to 1975, although EEOC regulations require the maintenance, during the pendency of a charge of discrimination, of application forms completed by the unsuccessful applicant and by all other candidates for the same position for which the aggrieved person applied and was rejected. Prior to some point early in 1975, applications which did not result in employment were not retained. USS retained most applications filed thereafter, although some were lost due to clerical error. 50. USS interviewers sometimes used interview sheets to take notes of information gathered during applicant interviews. USS preserved certain interview sheets for successful applicants during the class period, but destroyed all interview sheets for unsuccessful applicants. 51. During the class period, USS conceived of its obligation not to discriminate against racial minorities in hiring as an obligation to hire minorities in proportion to their representation in the work force in Bucks (Pennsylvania), Burlington (New Jersey) and Mercer (New Jersey) Counties. USS’s corporate headquarters approved the practice of setting minority hiring goals by reference to the percentage of minorities believed to be in the geographically defined labor market. Over the class period, USS’s goals for minority hiring, derived from labor market data relating to the three-county area, varied between 9.5% and 11.5%. 52. USS’s personnel department employees reviewed the DER periodically to see what percentage of minorities was being hired. When USS wanted to change the percentage of minorities hired, it would simply change the percentage of minorities scheduled for interviews. For example, Mr. Edwards testified that in 1976, when USS was defending against many charges of race discrimination, and when it was under pressure to increase the number of blacks in its apprenticeship programs in order to meet its obligations under the consent decrees referred to earlier, USS consciously increased the percentage of minorities scheduled for interviews in order to achieve a higher percentage of black hires. It was necessary to increase the number of black laborers in order to increase black representation in the apprenticeship programs because the latter were stocked from the ranks of laborers and other P & M workers. USS in fact achieved a dramatic increase in its percentage of black hires in 1976, as was shown by the statistical evidence discussed below. Mr. Edwards admitted that the increase in black hires in 1976 was not due to USS’s receipt of applications from blacks with better qualifications. 53. In setting its hiring goals for minorities, USS paid no attention to the percentage of minorities in the applicant pool. 54. In support of their claim, plaintiffs presented the testimony of a statistical expert, Dr. Samuel Litwin, who analyzed USS’s hiring practices. Dr. Litwin’s report was plaintiffs’ Exhibit 122. Dr. Litwin’s studies addressed the question whether the hiring practices and policies of USS during the class period had a statistically significant disparate and unfavorable impact on black applicants. 55. Dr. Litwin’s studies were based on the most accurate USS data available. Dr. Litwin received data derived from USS’s JALs and DERs in computer-readable form. He and those working under him then verified the accuracy of the data supplied by USS, line-for-line, against the actual JALs and DERs. Where the JALs and DERs did not provide complete information because records had been destroyed, Dr. Litwin used data derived from USS’s Affirmative Action Quarterly Reports (“AAQRs”) and a document prepared by USS Employee Relations staff assistant Kevin Goodwin in 1978 (“the Goodwin Report”). Finally, Dr. Lit-win obtained random samplings of actual job applications from the years after February, 1975 (prior applications having been destroyed), the Daily Call-In Logs, and the “DNS-DNR” file, which listed those applicants who either did not report for a scheduled job interview or did not start work as scheduled after accepting a job offer. These sources were used to supplement and check the reliability of data obtained from the JALs, DERs, AAQRs, and the Goodwin Report. 56. Dr. Litwin included in his studies a correction factor to correct for clerical errors in recording the race code of applicants. 57. Using available USS employment records, plaintiffs established that 49,585 applications for P & M positions were submitted to USS in the period from July 11, 1971 through 1982. Of this total, 12,857 P & M applications, or 25.9%, were submitted by blacks. Information about applications submitted during the years 1973 through 1982 was obtained directly from the JALs prepared by USS. USS destroyed the JAL for 1972. Information about applications submitted during 1972 was therefore obtained from the 1972 AAQRs, which USS prepared from the 1972 JAL and DER for submission to the U.S. Government. Because USS considered applications to be eligible for employment for one year and because the class period for hiring begins on July 11,1972, the appropriate applicant pool in this case commences with applications filed after July 11, 1971. Information about applications submitted during the last half of 1971 was obtained from the Goodwin Report, a group of documents USS prepared in 1978 from the JALs and other USS records for submission to the EEOC in defense of 43 charges of race discrimination. The Goodwin Report states the number of P & M applications submitted to USS in 1971. 58. The applications data plaintiffs developed from USS business records show, by year, the following number of applications (total and black) submitted for P & M positions: YEAR TOTAL APPLICATIONS BLACK APPLICATIONS %BLACK 1971 (July-Dee.) 1,210 280 23.1% 1972 3,953 801 20.3% 1973 5,059 1,222 24.2% 1974 8,985 1,925 21.4% 1975 1,328 267 20.1% 1976 7,022 2,233 31.8% 1977 5,119 1,389 27.1% 1978 10,191 2,969 29.1% 1979 6,665 1,762 26.4% 1980 23 6 26.1% 1981 26 26.1% 1982 4 0 All Years 49,585 12,875 25.9% 59. Dr. Litwin’s data show that, during the class period, USS hired 5,516 persons for P & M positions at its Fairless Works. Of these 5,516 P & M hires, 972—or 17.6%—were black. These figures were obtained directly from the DERs, contemporaneously prepared corporate records of USS hiring activity, 60. The DERs show the number of P & M hires (total and black) at USS for each year in the class period: YEAR TOTAL BLACK HIRES HIRES % BLACK % BLACK HIRES APPLICANTS 1972 (July-Dee.) 609 90 14.8% 20.3% 1973 1,781 287 16.1% 24.2% 1974 1,249 164 13.1% 21.4% 1975 6 2 33.3% 20.1% 1976 253 107 42.3% 31.8% 1977 240 63 26.2% 27.1% 1978 769 147 19.1% 29.1% 1979 582 111 19.1% 26.4% 1980 1 0 0 26.1% 1981 26 1 3.8% 11.5% 1982 0 0 0 0 All Years 5,516 972 17.6* 25.9% (inc. 1971) 61. The greatest P & M hiring at USS occurred in the years 1972, 1973, 1974, 1978, and 1979. Two-thirds of the class period hiring took place in the years 1972, 1973, and 1974 (3,639 hires); another one-fourth of the class period hiring took place in 1978 and 1979 (1,351 hires). More than 90% of all P & M hiring during the class period took place in 1972-74, 1978, and 1979. Virtually no hiring took place in 1975 (6 hires), 1980 (1 hire), 1981 (26 hires), or 1982 (0 hires). Only moderate hiring took place in 1976 (253 hires) and 1977 (240 hires). 62. Dr. Litwin’s data show that, of the 5,516 persons hired into P & M jobs, 5,245— or 95%—were placed in entry-level unskilled laborer positions. Only 258 of the P & M hires—about 4%—were for skilled trade and craft positions. The remaining 13 P & M hires (only 1%) were for positions other than laborer or trade and craft. 63. Dr. Litwin analyzed the P & M applications and hires data by four different methods. Under each method of analysis, Dr. Litwin considered whether USS’s hiring practices had a statistically significant disparate and negative impact upon black P & M applicants. Because virtually all applicants possessed the minimum qualifications USS had established for consideration for hire into entry-level P & M jobs (see Finding of Fact No. 17, supra), Dr. Litwin’s hypothesis was that blacks would be hired into the P & M jobs in proportion to their representation in the applicant pool. However, Dr. Litwin’s statistical studies revealed that USS hired blacks into P & M jobs at a rate significantly below their representation in the applicant pool. 64. In his analyses, Dr. Litwin employed 1.64 standard deviations as the measure of statistical significance. 65. Dr. Litwin’s first method of analysis (Method “A”) examined the fraction of applications submitted by blacks which led to employment and compared it to the fraction of applications from non-blacks leading to employment. Measuring applicants and hires whose race was known, he found that during the class period only 7.6% of the black applications led to employment, whereas 12.9% of the non-black applications led to employment. This meant that at USS applications submitted by non-blacks had a 71% employment advantage over applications submitted by blacks. 66. Conducting the Method “A” analysis for each separate year within the class period, Dr. Litwin found that in the years 1972, 1973, 1974, 1978, and 1979 the fraction of non-black applications which led to employment was considerably larger than the fraction of black applications which led to employment. Dr. Litwin found these disparities to be statistically significant. In 1972, applications submitted by non-blacks had a 47% greater chance of hire than black applications. This advantage to non-blacks was -3.6 standard deviations from its expected value, meaning that the probability that such an advantage to non-blacks occurred by chance was less than 1 in 1,000. In 1973, blacks suffered a 77% disadvantage; in 1974, an 85% disadvantage; in 1978, a 76% disadvantage; and in 1979, a 68% disadvantage. The standard deviations for the years 1973, 1974, 1978, and 1979 ranged from -5 to -11 below the expected values. 67. When Dr. Litwin biased his Method “A” analysis in USS’s favor by treating all applicants of unknown race as non-black and all hires of unknown race as black, non-blacks were still found to have statistically significant employment advantages in the years 1972, 1973, 1974, 1978, and 1979. 68. Over all years in the class period, using a weighted average fraction to measure the chances for hire by race, Dr. Litwin concluded that 12.6% of the black applications resulted in hire, whereas 21.6% of the non-black applications resulted in hire. Calculated in this fashion, black applications suffered a 72% overall disadvantage compared to non-black applications. This disadvantage was some 16 standard deviations below its expected value. 69. Dr. Litwin, using a second method of analysis (Method “B”), computed the expected number of black hires (if blacks were hired in proportion to their representation in the applicant pool), and compared it with actual black hires at USS during the class period. For the period from July, 1972, through 1980 (1981 and 1982 were excluded because only 1 black was hired during those years), the number of expected black hires is 1,391. During that time only 971 blacks were actually hired by USS, yielding a shortfall of 420 blacks. This shortfall is more than 14 standard deviations below the expected value. 70. Dr. Litwin employed a third method of analysis (Method “C”) to refine his calculations by taking into account when hiring decisions at USS were actually made and when applications were actually submitted. Considering the different dates when applications were actually filed, and using 18 different test assumptions as to the period of time USS considered applications to be active (beginning with the day of submission and ranging up to one year after they were first filed), the computer analyzed daily the percentages of blacks in various applicant pools. Referring to each such applicant pool, the computer then measured the difference between the expected and actual black hires for each of the 2,556 days in the period from January 1, 1973, through December 31, 1979. (This analysis could not be performed for hiring decisions made in 1972, because USS had destroyed the 1972 Job Applicant Log. Nor was it performed for the period 1980-1982, due to the absence of significant hiring activity.) Under all 18 test assumptions, the 18 applicant pools analyzed under Method “C” revealed statistically significant levels of black hiring shortfalls, ranging from shortfalls of 121 blacks to shortfalls of 241 blacks in the periods studied. These shortfalls ranged from 4 to 8 standard deviations below the expected values. 71. When the results of the Method “C” analysis are broken down by year (for the period 1973-1979 covered by this particular study), it appears that blacks suffered statistically significant hiring disadvantages in 1973, 1974, 1978, and 1979. The results of the Method “C” study, therefore, correspond to those obtained through Methods “A” and “B." 72. In a fourth method of analysis (Method “D”), Dr. Litwin further refined his calculations to ascertain for each P & M hire the precise period of delay between the date of application and date of hire. The computer was instructed to recreate exactly who was in the applicant pool during each hiring day when the persons hired submitted their applications. In this way, the timing and racial composition of each day’s applicant pool was accurately measured. Once again, this analysis could not be performed for 1972 or the first half of 1973 because USS’s destruction of the 1972 applicant log made it impossible to determine the applicant pools for persons hired in 1972 and early 1973. 73. Under Method “D”, if applications are deemed to have a lifespan of 26 weeks, the overall black shortfall in P & M laborer hires was 192. This figure represents substantial shortfalls-in 1973 (81 blacks), 1974 (82 blacks), 1978 (62 blacks), and 1979 (23 blacks). These shortfalls are statistically significant under Dr. Litwin’s assumptions, showing -6.4 standard deviations for 1973; -6.0 standard deviations for 1974; -5.1 standard deviations for 1978; and -2.3 standard deviations for 1979. 74. Using a 52-week lifespan for applications, over the shorter period of January 1, 1974 through the end of 1979, the results show a P & M laborer hiring shortfall of 137 blacks. This is almost 6 standard deviations below the expected value. 75. Dr. Litwin checkeid the reliability of the data he used for his four principal studies, and tested the soundness of his conclusions, by reference to other data supplied by USS. Specifically, Dr. Litwin used a computer printout prepared by USS listing job applications and other information from 1973 to 1979, the Goodwin Report, and the AAQRs concerning P & M applications and hires. These alternative data sources supported the accuracy of the data used for, and the conclusions reached by, Dr. Litwin’s principal studies. 76. During the class period, USS hired 258 persons for trade and craft, as opposed to entry-level, P & M jobs. Of these, 19— or 7.4%—were black. 77. Dr. Litwin did not perform analyses for trade and craft hires of comparable sophistication to those he performed for P & M hires overall because, prior to 1977, the JALs did not indicate whether an individual applied for a craft job or an entry-level P & M job. Instead, Dr. Litwin simply compared the percentage of black hires for craft jobs with the percentages of blacks in the overall applicant pools for the years within the class period. Although he found statistically significant shortfalls in black craft hiring for 1973 (shortfall of 9 hires), 1974 (shortfall of 10 hires), 1978 (shortfall of 3 hires), and 1979 (shortfall of 5 hires), his analysis does not address the possibility that the proportion of blacks who applied for craft jobs did not correspond to the proportion of blacks in the overall P & M applicant pool. 78. One of defendants experts, Dr. Bernard Siskin, performed a study similar to Dr. Litwin’s for craft hires but restricted his study to the years 1977-1979 in order that he might compare black craft hires directly to black craft applications. Dr. Sis-kin found a shortfall of zero for that three-year period. 79. Dr. Siskin testified that an assumption that proportionately as many black as white applicants were qualified as craftsmen would be unreasonable based on available census data and other information, including USS’s JALs for 1977-1979. 80. Mr. Edwards testified that applications from minority craftsmen were rarely received, although actively solicited. 81. Dr. Litwin performed an additional study using the randomly selected applications of 1,566 successful applicants and 4,412 unsuccessful applicants. He found that 21% of successful applicants, but only 9% of unsuccessful applicants, had fathers who were USS employees. This supports an inference that having a father in the USS work force gives an applicant an advantage in the hiring process. Dr. Litwin also found that only 9.7% of successful black applicants had fathers employed at USS, as opposed to more than 26% of successful white applicants. This supports a finding that the practice of giving a hiring advantage to applicants with fathers in the USS work force had a disparate and adverse impact on blacks. 82. USS’s work force from 1967 to 1978 was approximately 91.5% non-black. 83. Based on the results of his four distinct methods of analysis and the several data sources he considered, Dr. Litwin reached the firm conclusion that USS’s hiring had a statistically significant negative and disparate impact upon blacks during the overall class period, and during each of the years 1972, 1973, 1974, 1978, and 1979. His conclusions were the same whether summer hires were included or excluded, and whether persons of unknown race were included or excluded. His conclusions were the same after adjusting for clerical errors in recording race codes, after adjusting for the possibility that more black applications were recorded in the JALs than non-black applications, and after adjusting for the fact that some persons filed more than one application. 84. Dr. Litwin is a highly qualified statistician, and his presentation and conclusions were credible. The two defense experts who responded to Dr. Litwin—Dr. Siskin and Dr. Finis Welch—confirmed that the analyses Dr. Litwin performed in relation to the overall P & M applicant pool were appropriate, and that the results he obtained were reliable, given his data and assumptions. 85. Defendant attempted to impeach Dr. Litwin’s conclusions by attacking the reliability of the data he used. 86. USS cannot in good faith attack the reliability of the JALs, which its own employees prepared as business records of the company, and which the company relied on when it was to its advantage to do so. USS used applicant flow data from the JALs to defend against numerous EEOC charges of race discrimination in hiring. USS also used applicant flow data from the JALs to prepare its AAQRs for submission to the United States government. 87. The fact that, some persons submitted more than one application does not make the JALs unreliable, nor does it affect Dr. Litwin’s conclusions. Defendant’s expert, Dr. Welch found that a large majority, about 80%, of the applicants did not submit more than one application. With respect to those persons who did submit multiple applications, since USS did not consider applications that were more than one year old, applications from a single person filed more than one year apart over the 10 year class period should be counted separately because each application triggered a separate USS hiring decision. In any event, the submission of multiple applications would be significant only if blacks submitted substantially more multiple applications than non-blacks. Dr. Litwin studied the number of applications filed by each successful black applicant and each successful white applicant. He found no significant difference in the frequency with which blacks and non-blacks submitted multiple applications. 88. Nonetheless, Dr. Litwin’s analysis made adjustments for multiple applications. When an applicant was hired, he instructed the computer to delete from the applicant pool all applications submitted by the hired employee. Thus, these multiple applications were not considered when evaluating the racial composition of applicant pools for future hires. In making this correction, Dr. Litwin biased the adjustment in the defendant’s favor by assuming that blacks submitted multiple applications more often than non-blacks did. Even with this built-in bias in favor of the defendant Dr. Litwin obtained results consistent with his other studies. Blacks would have to have filed multiple applications with a frequency almost twice that of non-blacks in order for Dr. Litwin’s conclusions to begin to be affected. Even at that frequency, the results remain statistically significant overall, and for all individual years in which disparity was found except 1979. There is no evidence to suggest that blacks submit twice as many multiple applications per person as non-blacks. 89. Dr. Welch’s method of attempting to determine how many repeated applications were in the JALs, treating as “identical” all persons having the same last name and first initial, is not realistic or accurate. His approach incorrectly counted separate applications filed, for example, by Bob, Bert, Bruce and Bernie Jones as having been filed by one person. His approach improperly treated two applications, submitted by the same person several years apart during the class period, as one application. Dr. Welch admitted that his method overstated the number of multiple applications, improperly treating different individuals as the same, and that it could eliminate more blacks than whites due to fewer spelling errors of black applicants’ names in the data. 90. Although Dr. Welch did find a slightly greater frequency of black multiple applications in the JALs when he incorrectly used just the last name and first initial, he also conducted an entirely separate study of multiplicity in the “applications file.” There he identified multiple applications by social security number. This study gave a different result, showing that non-blacks filed multiple applications more often than blacks. Dr. Welch conceded that the use of social security numbers was a preferred method of testing for multiple applications. His second study had the effect of actually increasing the percentage of blacks in the applicant pool, and thus the racial disparity, because deletion of multiple applications would eliminate relatively more white applications. 91. The possibility that some applications might have been omitted from the JAL does not make the plaintiffs' employment data unreliable, nor does it affect Dr. Litwin’s conclusions. The importance of the JALs is that they establish the percentage of black applications submitted. Missing applications would be pertinent only if proportionately more white applications were missing than black applications. 92. To test the accuracy of the JALs, Dr. Litwin compared actual applications to the JALs to see if each application had a matching JAL entry. Among the applications of persons hired into P & M jobs during the class period, Dr. Litwin found matching entries in the JALs for a very large majority, approximately 85%. Also, using a sample of more than 4,000 unsuccessful applications submitted during the class period, he found matching entries in the JALs for approximately the same percentage. 93. Dr. Welch studied some 30,000 retained applications and found matching entries in the JALs for at least 90-94%. 94. Dr. Litwin’s analyses compensated for missing applications for persons who were hired into P & M positions. If he could not find an entry in the JAL for a person hired into a P & M position, he instructed the computer to add to the applicant pool a person of the same race as the person hired. In this way he completely corrected for the missing application. With or without this adjustment, he achieved the same results. 95. Dr. Welch’s methods of matching to determine the extent of missing applications among the P & M hires were flawed. Dr. Welch tried to match using computer printouts typed by USS personnel, and not with the raw data itself, that is, the JALs and DERs. On a number of occasions his inability to find perfect matches was due to USS typing errors in the preparation of the printouts, and not due to an omission in the raw data. Moreover, by limiting the lifespan of the applications he wanted to match to only 270 days, he automatically prevented 10% of the hires from being matched because approximately 10% of persons hired had applications more than 270 days old. He obtained an artificially low match, rate for 1973 hires by starting to seek matches on February 1,1973, despite the destruction of the 1972 JAL. This meant that anyone who was hired in 1973, but who applied in 1972, could not be matched. Finally, because Dr. Welch used a computer to do his matching, his results were depressed. A computer, being totally literal, would not match if there was a spelling error, even if the same person was actually in both data sources. 96. Even when Dr. Welch made adjustments to the data for missing applications he still found a statistically significant shortfall in black hiring, according to his report (D-335, p. 52). 97. Most important, none of defendant’s experts did studies which found that more unsuccessful white applications than black applications were missing from the JALs. In the absence of evidence of a racial disparity in match rates, missing applications are really a non-issue. 98. Dr. Litwin was the only expert who actually studied whether there was a racial disparity in match rates. He found there was no significant difference in the frequency with which entries for unsuccessful blacks and non-blacks were missing from the JALs. 99. Dr. Welch attempted to show that the employment data Dr. Litwin used were so sensitive that several minor “adjustments” could easily convert Dr. Litwin’s demonstrated black shortfall into an excess of black hires. Dr. Welch’s proposed adjustments do not undermine the validity of Dr. Litwin’s analyses. 100. To support his adjustment theory, Dr. Welch chose only one of Dr. Litwin’s 18 applicant pool combinations (D = 0, T = 52). The one he selected yielded a shortfall of 137 blacks, but ignored all of 1972 and 1973. Selection of D = 0, T = 26 would have been more representative, with an overall gross shortfall of 248 blacks during the years 1973, 1974,1978, and 1979. Thus, Dr. Welch’s analysis began by understating the black shortfall and by ignoring two heavy hiring years. Moreover, although Dr. Litwin’s Method “D” analysis was the sum of thousands of daily hiring calculations, the adjustments made by Dr. Welch were not. He incorrectly assumed that everyone in the class period was hired at one time. 101. In addition, Dr. Welch’s adjustments disregarded the fact that Dr. Litwin had already adjusted for hires whose applications could not be found in the JALs. Dr. Welch’s “minor” adjustment for missing applications in effect added some 30,000 more applications to the JALs, on the unsupported assumption that a high number of unsuccessful applications were missing from the JALs. This adjustment was inconsistent with Dr. Litwin’s finding that there are not large numbers of unsuccessful applications missing from the JALs. 102. Dr. Welch admitted that he was not confident of his adjustment for race code errors. Even if he had been, Dr. Welch conceded it would have only a very minor effect. In any event, Dr. Litwin had already accounted for race code errors in his analyses. 103. Dr. Welch conceded that no hard conclusion could be drawn from his adjustments, and he was not willing to vouch for the numbers that they produced. 104. Based on the above, I find that plaintiffs have proved that disproportionately few black applicants were hired for entry-level P & M jobs, overall during the class period and during 1972, 1973, 1974, 1978, and 1979 in particular, in relation to the number of blacks in the applicant pool for these jobs. Based on Dr. Litwin’s presentation, I find that the shortfall in black entry-level hiring was statistically significant. (The legal significance of this fact will be discussed in my Conclusions of Law, below). 105. I further find that plaintiffs have not proved that disproportionately few black applicants were hired for trade and craft or other non-entry-level P & M jobs during the class period, in relation to the number of black applicants for these jobs. 106. Besides attacking the reliability of Dr. Litwin’s data, USS asserted: (a) that it hired blacks in proportion to their representation in the labor market surrounding the plant; (b) that it did not hire at random to fill entry-level P & M jobs from among all those who met the minimum age, physical condition, and basic literacy requirements but rather sought at all times the “best qualified” candidate for any given opening; and (c) that any disparity between the percentage of blacks in the applicant pool and the percentage of blacks among those hired was due to the fact that a higher proportion of blacks than whites dropped out during the interviewing process, for one reason or another. 107. Mr. Edwards testified on behalf of USS that throughout the class period, USS sought to hire minorities in percentages sufficient to reflect at least the minority representation in the local labor market. Mr. Edwards and other USS employees decided that the relevant geographic labor market consisted of Lower Bucks County, Pennsylvania, and Burlington and Mercer Counties in New Jersey. Data USS obtained from Pennsylvania’s and New Jersey’s State Employment Services showed that minority representation in this geographical area was between 9.5 and 11.5 percent during the class period. However, Dr. Welch, defendant’s expert, testified that the local work force had 13.5% minorities. 108. The record is clear that, throughout the class period, USS hired minorities for entry-level P & M jobs at rates equalling or exceeding their representation in the geographical labor market defined by USS. (The legal significance of this fact will be discussed in my Conclusions of Law, below.) It is also clear that USS personnel believed that this fact discharged USS’s legal responsibility to black applicants under the anti-discrimination laws and insulated USS from liability for discriminating against minorities in hiring. 109. By asserting that it hired only the “best qualified” applicants for entry-level P & M jobs, USS sought to undermine the basic premise of Dr. Litwin’s statistical analyses: that, absent discrimination on the basis of race, one would expect black applicants to be hired in roughly the same proportion as their representation in the applicant pool. That premise would be correct, USS asserts, only if USS engaged in random hiring from among all persons who were 18 years of age or older and could pass a physical examination and read safety signs. On the contrary, USS states, the company puts all applicants through an elaborate application and interviewing procedure designed to determine which are best qualified to perform the physically and mentally demanding work of a laborer in a modem steel mill. According to USS, any discrepancy between the racial make-up of the applicant pool and the pool of those hired may therefore be assumed to be the result of real differences in qualifications between black and white applicants. The list of hiring criteria set forth in Finding of Fact No. 39, supra, represents USS’s conception of the standards it applied to all applicants in order to determine which were best qualified. USS not only conceded but underlined the subjectivity of the individual criteria as well as of the overall “best qualified” concept. USS emphasized that no single qualification or combination of qualifications would ensure either that a given applicant would, or would not, be hired, and that there was no set formula for balancing the various criteria in making any individual hiring decision. 110. An issue arose at trial as to whether USS would be permitted to supplement the list of its subjective criteria with an additional one, “promotability.” For reasons which were adequately set forth in the record at trial, I concluded that defendant had foreclosed plaintiffs from inquiring into the issue of promotability as a hiring criterion during discovery, and I therefore excluded evidence concerning promotability at trial. Because defendant’s motion for reconsideration of this evidentiary ruling was left pending at the conclusion of trial, I will take this opportunity to reaffirm it. Although I stand by my earlier determination that the proffered evidence should be excluded because it would be unfair to allow defendant to assert at trial defenses it had deliberately foreclosed plaintiffs from inquiring into prior to trial, I would also emphasize that I do not believe the excluded evidence would have any effect, one way or another, on my analysis of this case. What is at issue here is the process by which USS hires individuals into entry-level P & M jobs. It cannot be seriously disputed that USS, like any other business entity, has a right, and perhaps a duty to its stockholders and the national economy, to hire the best qualified personnel it can. It is abundantly clear that ability to absorb on-the-job industrial training (which may be assumed to relate to native intelligence, alertness, and judgment) is an important criterion in determining who is best qualified, even for entry-level jobs. It is also clear that the process by which USS divined who, among its applicants, were the best qualified was wholly subjective, consisting essentially of combining the gut reactions to the applicant of employees in the personnel office and one or more foremen in the plant. No evidence was presented or offered to show that applicants were tested in any way for ability to do specific tasks required of workers above the laborer level. Rather, the evidence was that those hired were expected to learn their initial jobs and any subsequent jobs they might be promoted into through on-the-job training. Given, then, that plaintiffs’ statistical evidence showed that blacks were hired at levels significantly disproportionate to their representation in the applicant pool, this question remains: why did USS’s wholly subjective process of culling out the “best qualified” applicants select whites for hire significantly more often than blacks, in relation to the racial make-up of the applicant pool? It is this question which I must answer as fact-finder. It is of no help to me to know that entry-level hires are frequently promoted unless there is also evidence that USS’s hiring process was designed to, or did in fact, identify which applicants had the greatest ability to absorb on-the-job training, and that black applicants tended to possess this ability to a lesser degree than white applicants. 111. USS presented the testimony of Dr. Joan Haworth, an expert qualified in economics, econometrics, and employment discrimination matters, in support of its contention that it was reasonable for USS to attempt to hire the best qualified applicants. Not surprisingly, Dr. Haworth testified that it was not only reasonable but a necessary business practice for USS to attempt to hire the best qualified applicants. As I have indicated above, I do not believe Dr. Haworth’s testimony in this regard was controversial. 112. USS presented the testimony of Dr. Finis Welch, an expert labor economist, in support of its contention that any disparity between the racial composition of the applicant pool and the pool of those hired reflects actual differences in applicants’ qualifications rather than racial bias in the selection process. Dr. Welch testified concerning a weighted probit analysis which he performed (the “Winner/Loser study”), statistically analyzing application information about selected successful and unsuccessful applicants to determine whether certain characteristics tended to give applicants an advantage or a disadvantage in hiring. This study purportedly demonstrated that “qualifications count” in hiring, but the study was inconclusive at best. 113.. Although two-thirds of USS’s hiring during the class period occurred in 1972, 1973, and 1974, Dr. Welch’s study did not evaluate applications and hires at any time prior to 1976. 114. Dr. Welch constructed his Winner/Loser study to measure only a handful of the characteristics USS used to make hiring decisions. The study failed to take into account most of the factors USS used in its hiring decisions, such as attitude, initiative, personality, ability to communicate, interest, ability to meet work schedules, prior USS employment, personal references, etc. 115. Dr. Welch’s study was based on non-random samples of successful and unsuccessful applications selected from an applications file maintained by USS. Dr. Welch admitted that USS’s applications file was incomplete and that the weighting and validity of his entire probit analysis were affected by the incompleteness of his data source. 116. Dr. Welch admitted that he was unable, with the data he possessed, to measure the quality of the applicant’s prior work experience, but only the length of employment. 117. Dr. Welch’s groupings and weightings of occupational experience were admittedly subjective. 118. The study ignored the work experience of persons employed less than one year, and equated the work experience of all others, whether they had worked one year, or ten years, or twenty years. 119. If the Winner/Loser study revealed anything about what USS looked for in its applicants, it showed that USS’s conduct was bizarre. Dr. Welch found that an applicant who had spent 7 of the past 10 years working as a manufacturing operative had no better chance of hire at USS than a person who had never worked a day in his life. He found that an applicant with one year of experience as an assembly line worker in an ice cream plant was more likely to be hired by USS than an applicant with 10 years of experience as a steel mill laborer. He found that applicants who had a college degree or some college training had a reduced chance of hire at USS, as did applicants who had clerical experience. Ironically, he found that being white gave an applicant an “edge,” that is, an increased probability of being hired. 120. Of the ten variables Dr. Welch analyzed, only one proved statistically significant under all three of the weighting systems he used. That variable was being a “new entrant” to the labor market, which meant that applicants who had never worked before had the most statistically significant advantage in USS’s hiring process. He found only two other variables to be statistically significant under any weighting system: having some college education (a disadvantage in hiring), and having manufacturing experience (an advantage). The advantage of manufacturing experience was less s