Full opinion text
OPINION STAPLETON, District Judge: The Equal Employment Opportunity Commission (“the Government”) brought this action against E. I. duPont de Nemours & Company (“DuPont”), pursuant to Title VII of the 1964 Civil Rights Act, 42 U.S.C. § 2000e et seq., to secure relief against allegedly discriminatory employment practices occurring at DuPont’s Chestnut Run and Christina Laboratory sites. After four and one-half years of litigation and a trial of five weeks duration, the case is now ripe for a decision on the merits. This Opinion constitutes the Court’s findings of fact and conclusions of law. DuPont is engaged in research, development, manufacture and marketing of chemicals and chemical related products which it ships across state lines. Since July 2, 1965, the effective date of Title VII, DuPont has continuously been an employer engaged in an industry affecting commerce within the meaning of Section 701(b), (g) and (h) of Title VII, 42 U.S.C. § 2000e(b), (g) and (h), and has employed more than twenty-five persons. Accordingly, this Court has jurisdiction of the controversy pursuant to Section 706(f)(1) of Title VII, 42 U.S.C. § 2000e-5(f)(l). A detailed account of the administrative proceedings leading up to the filing of this action on November 12, 1972 is set forth in this Court’s prior Opinion and that account need not be repeated here. Suffice it to say that, based upon the facts there recorded, I have concluded that the prerequisites to suit by the Government under Title VII have been satisfied. I. THE FACTUAL CONTEXT. While many of the facts are more appropriately discussed in the context of the various contentions of the parties, an overview of the character of the two sites, certain of DuPont’s personnel practices, and the history of black participation in the work force are helpful at the outset. A. The Character Of The Two Sites. Chestnut Run began operations in 1954 when the Textile Research Laboratory (TRL) of DuPont’s Textile Fibers Department was transferred to the site. Over the next fourteen years, seven other laboratories were established at that location. Each of these laboratories is an arm of a particular operating department within the DuPont Company and works with a given product line or group of products of the parent department. Their purpose is to provide testing facilities and product expertise to existing or potential customers for DuPont products. These laboratories assist customers in overcoming processing difficulties and also engage in the development of new products and new end-uses for existing products. In order to be able to assist customers, each laboratory must be capable of duplicating the manufacturing processes of each customer in a particular line of industry. Thus, these laboratories contain hundreds of pieces of equipment like those used by industrial customers. At any given moment, however, relatively few of those pieces of equipment will be in use. The Christina Laboratory was developed by the Textile Fibers Department in 1965 as a small lot manufacturing plant engaged in the development of new methods and processes for manufacturing new products. As a result, Christina Laboratory is in some respects a research facility engaged in refining the characteristics of products and developing new processing techniques and standards of manufacture. In the course of a product’s development, processing procedures and techniques are modified on an almost daily basis. Pending full development of the facility, all administrative services were to be provided by Chestnut Run personnel on assignment to Christina Laboratory. It was originally intended, however, that, in time, Christina Laboratory would become totally independent of Chestnut Run. Due to unfavorable economic turns, the separation of Christina Laboratory from Chestnut Run never occurred. Because the Textile Fibers Department (i. e., its Textile Research Laboratory) was the first to arrive at the Chestnut Run location, it was assigned the “landlord function”, which included the responsibility to provide all administrative services for itself and the other departmental laboratories that followed. Landlord responsibilities include such general services as safety, protection, transportation, maintenance, power, design and product engineering, purchasing, accounting, medical services, cafeteria, shipping and receiving of materials, plus clerical and secretarial services for the entire site. Another important landlord responsibility, especially in the context of this suit, is the maintenance of a personnel office to oversee recruitment, hiring, training and other personnel services relating to non-exempt employees attached to the departmental laboratories. In addition, the Textile Fibers Department has responsibility for securing, by transfer, promotion or hire, exempt employees to administer the landlord function. All employees at the two sites are assigned to particular “work units”, i. e. functional groupings of related jobs or products. Each departmental laboratory, including Christina Laboratory, is a separate work unit. In addition, there are a number of administrative work units, including the Accounting, Protection, Service, Engineering, Control, Mechanical, Power and Personnel work units. As of August 31, 1976, there were 1,234 non-exempt employees at Chestnut Run and Christina and 98 exempt employees engaged in the landlord function. The nonexempt work force falls into four broad occupational categories. Category Employees % of Total Technical/Operations 766 62.0% Secretarial/Clerical 237 19.2% Craftsmen/Maintenance 186 15.1% Service/Protection 45 3.6% Out of this total non-exempt work force, only approximately 100 employees are in what can be categorized as low-skilled positions, i. e. mail clerk, driver, etc. The remaining jobs all require various types and levels of skills. Combined peak employment at the two sites was reached in 1969, although Christina Laboratory continued to develop until 1974. Since that time, however, economic circumstances have required curtailment df operations, with concurrent reductions of force. B. Certain Of DuPont’s Personnel Policies. Non-exempt jobs on the sites are classified by compensation level. These jobs are also classified as being at, below or above “career level”. A career level job is the highest level job within a particular job family or “progression” that most employees are reasonably expected to reach during their working careers. In the laboratory work units, for example, the laboratory technician job, which was a level 8, was considered the career level job. On the discovery cut-off date, August 31, 1976, the average laboratory technician received approximately $13,000 per year. (PX-56). Testing technicians were level 6, and thus below career level. Technical assistants were level 10, and thus above career level. Throughout the period from 1955 to the present, the passing of a test has been required as a prerequisite to employment in laboratory technician and testing technician jobs, as well as all clerical and office jobs. These positions constituted the bulk of nonexempt jobs available at the sites. Prior to 1969, no testing was required as a prerequisite to being hired into the relatively few processing technician jobs. During this period, people were hired into these jobs on the basis of prior experience in the performance of various textile operations. In 1969, a processing technician’s test was instituted and a substantial number of processing technicians were hired on the Christina site in the period from 1972 to 1974. From the opening of Chestnut Run until 1964, there was no testing of applicants for craftsman positions. During that period, DuPont hired only journeyman craftsmen who had completed an apprenticeship in their respective crafts. In 1964 a test was instituted in order to permit people with prior industrial training and experience to demonstrate that they had knowledge and skills equivalent to those of a journeyman craftsman, but defendant continued to hire only skilled people as craftsmen until 1972. In that year a mechanic’s training program was instituted, admission to which required the passing of an aptitude test but no prior training or skills. The last substantial job category, that of the Service work unit, has never had an entry test. From the beginning, there have been job transfer and seniority systems on the sites. Between 1960 and 1972, under the job transfer system, any non-exempt employee in the Service or Administration work units could file a job transfer request indicating a desire to transfer to a specified type of job, at or below career level, in the laboratory work units. Non-exempts in a lab unit could request transfer to Service and Administrative work units, but not to another lab unit. After 1972, any non-exempt could apply for transfer to a job at or below career level in any other work unit on the site including the lab units. If a non-exempt passed whatever qualifying test was required for the job he sought, he would be placed on the job transfer list based upon his company seniority and the company seniority of the other test qualified employees who had requested transfer to that type of job. Under the seniority system, when an opening occurred, the job was first offered to anyone who had previously been regressed from that job as a result of a reduction in force. If there was no one with “vested rights” by reason of a prior regression, the job was then offered to the employees in the work unit involved, in the order of their work unit seniority. If not filled in this manner, the company went to the job transfer list and offers were made in accordance with company seniority. Only if these three steps produced no willing candidate, did the company “go to the gate” and hire a new employee. Under the seniority system, when it became necessary to lay off employees, the most junior employee on the site in terms of company service was the first to go. Thus, while work unit seniority played a roll in promotions and regressions, it played no part in any layoff decision. Hiring and promotion decisions with respect to above career level jobs and exempt positions were not the responsibility of the personnel office. They were made by work unit and site management. Length of service, except to the extent it was reflected in skills, knowledge, and ability to train, etc., was not the determining factor in these decisions. C. History Of Black Participation In The Work Force. Prior to 1960, the unwritten policy at Chestnut Run was to maintain “white jobs” and “black jobs”. Blacks were hired only into the Service Unit to perform unskilled labor such as janitorial services, building and grounds maintenance, etc. All other jobs were reserved for whites. In the Spring of 1957, Dr. Russell W. Peterson, who then had the responsibility for the development of the Chestnut Run site instructed Harwood Strange, the site’s Personnel Superintendent, to form a committee “on a confidential basis” to propose a strategy for providing wider employment opportunities for blacks. Little tangible progress was made, however, until June of 1960 when Strange, as a result of contacts with an administrator at the predominantly black Walnut Street YMCA, placed a black stenographer in the Textile Research Lab. In 1960, management instructed the personnel office that blacks were to be considered on an equal basis with whites as candidates for all jobs at the site. Not surprisingly, however, this did not result in an immediate increase in black participation in previously white jobs. The message conveyed to black employees and the black community by years of employment discrimination was not to be eradicated by a change of official policy alone. In 1959 or 1960, efforts were begun to find black candidates in the Service work unit for promotion to lab tech jobs. These initial efforts can accurately be described as narrow in scope. All of the personnel files of the 60 to 70 Miscellaneous Services Operators (“MSO”) were reviewed by a first line (immediate) supervisor in an effort to select those most likely to be able to pass the lab tech test. Three black employees were selected by supervision and tutored in math and other subjects related to the demands of the lab tech test. However, none of these three employees passed the test. It was not until the Plans for Progress Program was established as a result of President Kennedy’s Executive Order 10925, signed in the Summer of 1961, which required government contractors to offer equal employment opportunities, that DuPont developed a program for getting the equal employment opportunity message across to its MSO’s and the black community as a whole. In the Summer of 1961, management formulated a set of procedures designed to “communicate its EEO policy to the lowest levels of employment”, and it is clear that the desire of site management to bring this message home was communicated to the supervision of all work units. It is equally clear that the objective of the program was not fully accomplished at this stage. In October of 1961, Samuel Holland, an MSO, passed the Lab Tech test and was assigned to a Lab Tech job in the Plastic Lab. This fact was announced to each working group in the Service work unit. In September of 1963, supervision reviewed the procedures and qualifications necessary for transfer to the other jobs available on site with each MSO employed as of September 1 of that year. A checkoff roster of MSO’s was used to assure complete coverage. Skepticism remained nevertheless. A rumor was circulating about this time in the Service work unit, for example, that there were two tests, one for whites and one for blacks. In November of 1963, all MSO’s were offered the opportunity to take a Beta Intelligence test which had been secured from a local high school. It was designed to examine non-verbal reasoning and it was hoped that the test might identify potential candidates for coaching on a more objective basis. Sixty-three MSO’s took the test. Coaching of interested MSO’s by first line supervision continued thereafter. There was another procedure used during the 1960 to 1965 period which was designed to inform employees of the opportunities for job advancement. The established policy was that the various types of jobs available on the site and the job transfer system would be explained to each new employee by his first line supervisor, as a part of his orientation, during the first two weeks of his employment. The orientation check-off sheets listed this as a subject required to be covered and these sheets were filled out and returned to third line supervision and ultimately to the personnel office. This procedure was reasonably designed to accomplish the desired end. Its practical effectiveness, however, depended, of course, on the attitude and commitment of individual first line supervisors. I believe that not all blacks hired to the Service work unit after 1960 were informed of the opportunities for advancement. On the other hand, the record suggests that the effectiveness of this procedure improved during the 1960 to 1965 period as management reaffirmed its commitment to equal opportunity and a.s tangible evidence of that policy began to appear on the sites. By 1965, at least 25 of the black MSO’s had filed job transfer requests and had taken the required tests, some of them on a number of occasions. Seven black service workers had been transferred to previously all white jobs at that time. It is a fair inference from these facts and the other evidence of management’s efforts to get the message across, that an MSO who did not know of the job transfer system and the availability to him of testing in June of 1965 was the rare exception, rather than the rule. Besides these internal efforts, there were efforts during the 1960-65 period to bring the equal opportunity message to outsiders. After a somewhat belated start, a substantial campaign was mounted from 1963 to 1965, to recruit black applicants. Openings for lab techs, craftsmen, stenographers, typists and mail clerks were advertised in black newspapers. All hiring ads after 1963 carried the equal opportunity employer notation. Schools and employment agencies were advised by individual letters of the desire for minority applicants. Local schools were visited and guidance counselors, as well as leaders of the black community, were invited to visit the sites to learn of the job opportunities available. It is difficult to find fault with these efforts. While complete records are not available, the evidence does show that by July 2,1965, the effective date of Title VII, at least 23 blacks had been hired or promoted into jobs in work units other than the Service work unit. Nevertheless, on that date blacks constituted approximately 1.2% of the nonexempt, non-Service work unit labor force and approximately 1.4% of the exempt work force. After the effective date of Title VII, efforts to increase black employment beyond the MSO ranks continued. In 1966 a new Services Superintendent met with the MSO’s and reviewed the opportunities and requirements for promotion or transfer to various positions. The tutoring program continued. During the latter part of 1965 and again in May of 1966, presentations were made at meetings of all service workers which reviewed the openings available for stenographers, typists, mechanics and technicians and requested that the employees speak with friends and neighbors about these openings. By 1968, it was concluded that the potential of the tutoring program for MSO’s had been exhausted and a program, later to become known as the “Thirteen Minus Two Program”, was instituted in the Summer of that year. This was an on the job training program in the Textile Research Lab. Thirteen participants were selected from the Service work unit solely on the basis of length of company service and willingness to participate. The participants were assigned to jobs in TRL. In addition to training by supervision in performing their assigned tasks, they participated in bi-weekly training sessions in mathematics and simple geometric and algebraic principles. The goal of the program was for the participants to achieve competence over a five year period in 50% of the tasks ordinarily performed by test qualified lab technicians. Those who achieved such competence were to become laboratory technicians. Participants could qualify at any time by-passing the lab tech test. Two additional programs resulted from a compliance review by the Atomic Energy Commission’s Contract Compliance Office in 1972: a job preference survey and a mechanics trainee program. In July of 1972, a questionnaire was given to all black and female employees on the sites who were below career level. They were asked to indicate any jobs on the sites in which they had an interest and to state any reasons why they did not wish to be considered for transfer or promotion. Where interest in a job was expressed, the employee was asked to file a job transfer request and to take any necessary test. While DuPont had succeeded in hiring one black electrician in 1965, its inability to locate skilled black mechanics continued in the post-Act period. In an effort to alleviate this problem, a maintenance trainee program was established in 1972 which offered on the job training to people who passed an aptitude test. MSO’s were contacted in order of company service to determine interest in this program. DuPont’s efforts to encourage upward mobility for its MSO’s has met with significant success. Of the 98 employees who were hired as MSO’s prior to July 2, 1965 and who remained on roll as of that date, 50 had obtained non-exempt jobs at level 6 or above or exempt jobs by 1975. Seven members of this group (7.1%) had obtained exempt jobs. Five of the seven were in work units other than the Service work unit. The equal employment message was also effectively communicated to the black community. DuPont’s applicant flow logs for the years 1972 to 1976 show that 22.6% of all job applicants during that period were black. While there are a number of specific jobs at the two sites which have never been held by a black, with one exception, there is no work unit on either site which has not had a black employee. The sole exception is the engineering work unit which employs only two people. The racial composition of defendant’s non-exempt work force on August 31, 1976 was as follows: Chestnut Run & Christina Laboratory Non-Exemnt. Workforce hv Job and Race as of 8-31-76 Jobs Total Percent Employees Blacks Black Technical Positions (Sr.) 114 0 0% Laboratory Technicians 402 26 6.5% Process & Testing Tech. 133 33 24.8% Mechanic Positions 153 7 4.6% Clerical Positions (Level 6 and above) 195 15 7.7% Clerical Positions (Level 4 & 5) 29 6 20.7% Clerical Positions (Level 2 & 3) 11 ’ 1 9.1% Garage Mechanic 1 1 100% Service Driver 29 26 89.7% Utility Operator-Prod. 26 9 34.6% All other positions 141 _29 20.6% Source: PX 54 Total 1,234 153 12.4% In summary, DuPont maintained a segregated work force prior to 1960. In initiated planning for desegregation in 1957, but on a “move slowly” basis. In 1960, its policy changed to one of equal employment opportunity. Its initial effort to implement this policy was limited primarily, however, to the placement of a few “hand picked” blacks in previously all white jobs. Prodded by the Executive Order 10925, requiring government contractors to offer equal employment opportunity, and the Plans for Progress implementation of that Order in the early 60’s, plans were developed to effectively communicate the equal employment policy to black employees and the black community. By the effective date of Title VII in 1965, defendant had effectively communicated this change of policy and was actively encouraging greater black participation in previously all white jobs. During the post-Act period, continued encouragement of existing employees, on the job training programs, and increased black employment applications resulted in substantial black participation in those positions. II. THE SUBSTANTIVE STANDARDS, AND THE BURDENS OF PRODUCTION AND PERSUASION. The Government claims that various employment practices of DuPont at these two sites have a disparate impact on blacks which is not justified by any legitimate considerations relating to the business there transacted. It has offered extensive statistical data in support of its claims as well as the testimony of eight DuPont employees which is intended to “bolster” the statistical evidence by demonstrating the effect of the challenged policies and procedure as actually practiced. DuPont has countered with statistical data of its own and with testimony from its managerial personnel describing the nature of the business conducted, the jobs available at the sites, and the way in which its personnel policies have been established and carried out. The legal principles to be applied in analyzing a record of this kind are relatively well established. Section 703(a) of the Act, 42 U.S.C. § 2000e-2(a), defines an unlawful employment practice as follows: (1) to fail or refuse to hire or to discharge any individual, or otherwise to discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual’s race, color, religion, sex, or national origin; or (2) to limit, segregate, or classify his employees or applicants for employment in any way which would deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect his status as an employee, because of such individual’s race, color, religion, sex, or national origin. The objective of Title VII is the elimination of discrimination in employment. As the Supreme Court observed in Griggs v. Duke Power Co., 401 U.S. 424, 429-31, 91 S.Ct. 849, 853, 28 L.Ed.2d 158 (1971): The objective of Congress in the enactment of Title VII is plain from the language of the statute. It was to achieve equality of employment opportunities and remove barriers that have operated in the past to favor an identifiable group of white employees over other employees. ****** Congress did not intend by Title VII, however, to guarantee a job to every person regardless of qualifications. In short, the Act does not command that any person be hired simply because he was formerly the subject of discrimination, or because he is a member of a minority group. Discriminatory preference for any group, minority or majority, is precisely and only what Congress has proscribed. See also Tram World Airlines, Inc. v. Hardison, 432 U.S. 63, 97 S.Ct. 2264, 53 L.Ed.2d 113 (1977). The plaintiff in a case of this character can establish a prima facie case with statistical evidence showing a significant disparity between the black population in the defendant’s work force, or any relevant portion thereof, and the black population of the labor pool from which that work force, or relevant portion thereof, comes. Hazelwood School District v. United States, 433 U.S. 299, 97 S.Ct. 2736, 53 L.Ed.2d 768 (1977). Other evidence may be submitted tending to support the inference of disparate impact suggested by the statistical data. If the plaintiff makes out a prima facie case, the defendant then has the burden of coming forward with evidence which tends to negate the inference of disparate impact, to explain any such impact by legitimate job related considerations or to show that the plaintiff’s statistics are a product of pre-Act policies only Hazelwood School District v. United States, supra. If such evidence is introduced, the relevant inquiry for the Court is whether, considering all the evidence, the plaintiff has proved by a preponderance of the evidence that the challenged practices have a disparate impact which is not dictated by the legitimate needs of the employer. If it appears that any disparate impact is justified by the nature of the employer’s business, the plaintiff may nevertheless prevail if the preponderance of evidence indicates that policies and practices having a lesser adverse impact on blacks would serve the employer’s legitimate interests substantially as well. With one exception, the presence or absence of discriminatory intent is not relevant to the liability issues posed by a case of this character. III. MATTERS NOT IN ISSUE. Before addressing the issues for resolution by this Court, it is important to understand that there are several key areas in which there is no dispute between the parties. With the exceptions heretofore noted, all of the non-exempt jobs at the two sites have required qualification by test throughout the post-Act period. In addition, there has been testing in connection with promotions to some exempt positions. The Government in its original and restated complaint challenged these tests as having an unjustified disparate impact upon blacks. After examining the materials produced in discovery and the documentary exhibits prepared by DuPont for use at trial, however, the Government stipulated that all of these tests had been validated in accordance with the EEOC Guidelines, 29 C.F.R. § 1607, et seq., and are in compliance with Title VII. Moreover, the Government, with two exceptions, concedes that defendant’s use of these validated tests is in accordance with Title VII. It does maintain, however, that blacks were not given equal access to testing and that the Thirteen Minus Two Program does not have as great an adverse impact on blacks as DuPont’s Laboratory Technician test and “would also serve its ‘legitimate interest in efficient and trustworthy workmanship’ ”. The second area which is not the subject of dispute is the area of discrimination in recruitment. The Government pointed out at the end of its case that it did not contend that DuPont had discriminated in the recruitment of applicants for non-exempt employment and there is no evidence to suggest that it has. To the contrary, the record shows substantial special efforts by DuPont in the post-Act period, as well as before, to recruit black employees for nonexempt jobs and its applicant flow log shows that, from 1972 to 1976, 22.6% of all applicants were black. (DX-13a). The highest figure which the Government has suggested for the black population of the appropriate labor pool is 16.9%. Finally, the Government concedes that the data relating to the period from 1972 to 1976 does not suggest discrimination in overall hiring by DuPont. Plaintiff’s Exhibit 85, for example, shows that of the 515 people hired during this period 21.9% were black. The Government does strenuously assert, however, that there was discrimination in overall hiring during the period from 1965 to 1969, and that there was discrimination in initial job assignment throughout the entire post-Act period. IV. GENERAL PATTERNS AFFECTING EVALUATION OF THE STATISTICAL EVIDENCE. The Government relies in part on statistical data showing the racial composition of DuPont’s work force at various points in time after July 2,1965. As earlier noted, such data, together with evidence of the racial composition of an appropriate labor pool, can establish a prima facie case. The Government’s work force statistics, however, reflect pre-Act hirings, promotions and other personnel decisions as well as post-Act ones. This is of great significance in a case of this kind where the evidence shows that an employer practiced discrimination in the pre-Act period. While evidence of pre-Act discrimination may be relevant and helpful in understanding what has happened after July 2, 1965, liability cannot be predicated on pre-Act practices. International Brotherhood of Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977). As the Supreme Court noted in the Hazelwood case: [an] . . employer who from [the effective] date forward made all of its employment decisions in a wholly nondiscriminatory way would not violate Title VII even if it had formerly maintained an all-white work force by purposefully excluding Negroes. For this reason, the Court cautioned in the Teamsters opinion that once a prima facie case has been established by statistical work force disparities, the employer must be given an opportunity to show “that the claimed discriminatory pattern is a product of pre-Act hiring rather than unlawful post-Act discrimination”. 433 U.S. at 309, 97 S.Ct. at 2742. As earlier noted, DuPont began the post-Act period with a highly segregated work force and the statistical data evidencing the black population in DuPont’s work force and various segments thereof at given points in time during the post-Act period must be evaluated with the recognition that racially neutral post-Act employment practices would not immediately produce a black work force at parity with the black population of the available labor pool. In this connection, it is also important to recognize several facts relating to the stability of DuPont’s work force at Chestnut Run and the Christina Laboratory. Of those on roll in the relevant work force on August 31, 1976, 55.1% were on roll before July 2, 1965. The relatively low turn over figures which in part are responsible for this statistic show a somewhat higher separation rate for black employees than for whites. Also, as one would expect, there was greater stability in the post-Act period among those with the more skilled, higher paying jobs. The highly segregated character of DuPont’s work force on the effective date of the Act is also a relevant consideration in evaluating the data relied upon by the Government in support of its claim of discrimination in promotions to above career level and exempt positions. Candidates for above career level and supervisory positions come largely from the ranks of those having substantial experience in career level jobs. Accordingly, it is to be expected that racially neutral promotion procedures would not significantly increase black representation ifi these higher positions until sometime after significant black participation at the career job level is reached. While the Government stresses hiring and work force population statistics during the period from 1965 to 1969, DuPont puts major emphasis on statistical data showing the racial composition of its work force on August 31, 1976, and on the number of blacks hired and promoted over the entire decade from 1966 to 1975. This is undoubtedly because the hiring and promotion statistics for the period from 1970 to 1975 are somewhat more favorable to DuPont than are those relating to the 1965 to 1969 period. While DuPont’s performance during the later period is, of course, relevant, it cannot serve as a complete response to the Government’s claims regarding the earlier period. If the record shows that DuPont’s personnel practices during the 1965-1969 period had an unjustified disparate impact upon blacks, the Government has established liability regardless of DuPont’s subsequent performance. Finally, one further comment is in order regarding the statistical data. The parties recognize, as they must after the Hazelwood case, that absent evidence of discriminatory recruitment, the most significant statistical data relevant to the question of the available labor pool is applicant flow data. DuPont currently has applicant flow data by race for the period from 1972 to 1976. It does not, however, have such data for any year prior to 1972. The parties agree that it is, therefore, necessary to look to other data sources for the relevant pool in the earlier years, but the Government maintains that the Court should draw an inference from DuPont’s failure to produce earlier applicant flow data by race that, if available, it would show disparate impact upon blacks in the area of hiring. I decline to draw such an inference. Both the trial testimony and contemporary documentation show that DuPont first started keeping applicant flow data by race in May of 1972 as a result of a suggestion from the Atomic Energy Commission’s Contract Compliance Office. The failure to maintain an applicant flow log indicating race prior to that time was attributable to an understanding that an employer was not supposed to keep applicant records by race. V. HIRING AND INITIAL . ASSIGNMENT. A. Non-Exempts. 1. The appropriate labor market. The initial issue to be determined in connection with the charge that DuPont has discriminated against blacks in the hiring and job assignment of non-exempts is the relevant labor market area. The Government argues for a market defined by its expert, Dr. Haber, after a pre-trial study of the distribution of the residences of those employed at the two sites. The black population in this market, according to census figures is 16.9%. DuPont, on the other hand, urges the Court to select the work force of the Wilmington Standard Metropolitan Statistical Area (“Wilmington SMSA”) as the appropriate labor market. The black population of the employed labor force in this area is 11.2%. Dr. Haber, a labor economist, analyzed data supplied by DuPont which identified the residence at the time of hire of all individuals employed at the two sites in non-exempt jobs as of March 1973. He determined that greater than 85% of all employees resided within a fifteen mile radius of each site, and that over 95% resided within a thirty mile radius of each site. He then drew 360 degree concentric circles around each site at distances of 5 miles, 10 miles, 15 miles, and 30 miles, respectively. After ascertaining from census tract data the work force population of each area thus defined, he weighted the population of each area in proportion to the number of DuPont employees residing in that area. The result of this weighing process was the conclusion that 16.9% of the total population in the relevant market was black. There are several problems with Dr. Haber’s approach. First, his 16.9% relates to total population in the market area. Dr. Haber made no effort to isolate the labor force. His population thus included people above and below the working age, those in colleges, prisons, mental hospitals, other long term health facilities, and so forth (Francese J — 10—11; Haber B-81). The labor force is substantially smaller than the total population. Dr. Haber’s figures, for example, show that the labor force in the State of Delaware is only approximately 40% of the total population. On cross-examination he acknowledged that one would expect the percentage of blacks in the total population to exceed that in the labor force. While Dr. Haber’s report suggests that there is a correlation between black representation in the total population and black representation in a population of employed persons with less than four years of college and that, therefore, the total population is a meaningful standard in the context of nonexempt jobs, I am unpersuaded by his analysis. More importantly, while Dr. Haber’s approach takes recognition of the total distance by air between the residences of those employed at the two sites and their jobs, it did not involve any analysis of actual commuting patterns which are a function of time, cost, and intervening employment opportunities. (Siskin T-32-36, 44-47, Q-33-35; Haber C-74-76). The concentric circle approach, however, makes a very basic assumption about these matters. While it incorporates a consideration relating to employee distance from work, it in effect substitutes a geometric figure for an analysis based on the social and economic integration of the areas surrounding the employer. This necessarily injects an arbitrary factor which makes the analysis a less than satisfactory guide. This factor is illustrated the fact that Dr. Haber’s 30 mile circle includes large parts of Philadelphia and Camden while the basic data shows only one DuPont hire from Philadelphia and none from Camden. (PX-1). While the effect of including these areas is substantially reduced by Dr. Haber’s weighting process, it appears to the Court that the black population of these areas is nevertheless given undue weight. For example, even after weighting, Pennsylvania accounted for over 21% of the labor market for Christina Laboratory, while only 7.2% of those at the site resided in Pennsylvania at the time of hire. by Finally, there is a pragmatic problem with adopting the Government’s approach. It is apparent from the record and from the Court's on-site inspection that not all people in the labor force have the necessary skills and knowledge to perform the majority of defendant’s non-exempt jobs. It thus becomes relevant to try to isolate those in the relevant area with the requisite skills. The census tract data with which Dr. Haber worked was not sufficiently detailed to enable him to "break out” those with particular skills. He had no idea, for example, how many of the blacks included in his 16.9% were qualified to be electricians or mechanics. (Haber T-82-83). For this reason, the Government’s suggestion with respect to market area is not a helpful one. DuPont, in addition to criticizing Dr, Haber’s methodology, exhibits some understandable frustration with what it perceives to be conflicting positions on the part of the Government. The EEOC’s “Affirmative Action and Equal Employment, a Guidebook for Employers” (DX-73) advises employers that: Your labor area should generally be the Standard Metropolitan Statistical Area (SMSA) for which the Census Bureau and other employment data is available. In addition, the work force data cited by the Atomic Energy Commission’s Office of Compliance in investigating and working with DuPont to develop an affirmative action program was apparently Wilmington SMSA data. While DuPont acknowledges that the Government has not committed itself to the position that SMSA data is the most appropriate for all employers in all situations, it nevertheless suggests that, given the Government’s position, SMSA data should be acceptable to the Government unless there are factors suggesting it is an unreliable indicator of a particular employer’s available labor pool. While- this contention has equitable appeal, it is unnecessary to pass upon it in this case. The record convinces me that, of the data available, the Wilmington SMSA census data provides the most reliable indication of the labor pool available to these two sites. SMSA's are established by the Office of Management and Budget, with the assistance of the Bureau of the Census, as reflections of established areas of social and economic integration primarily on the basis of existing commuting patterns. United States v. Connecticut National Bank, 418 U.S. 656, 670, 94 S.Ct. 2788, 41 L.Ed.2d 1016 (1974). (Francese J-4-9). While they are undoubtedly imperfect indicators of human behavior, they are an attempt to apply the criteria that are of particular relevance in Title YII cases, and the Bureau of the Census has collected empirical data suggesting that this attempt is relatively successful. In the Philadelphia SMSA, for example, it has been ascertained that only 5% of the workers leave the SMSA to work. (Francese J-7-8). For these reasons, courts have frequently used SMSA data in Title VII cases to determine the appropriate labor pool. See, e. g., Thompson v. McDonnell Douglas Corp., 416 F.Supp. 972, 980 (E.D. Mo.1976), aff’d. 552 F.2d 220 (8th Cir. 1977); Louis v. Pennsylvania Industrial Dev. Auth., 371 F.Supp. 877 (E.D.Pa.1974), aff’d. 505 F.2d 730 (3rd Cir. 1974), cert. denied 420 U.S. 993, 95 S.Ct. 1430, 43 L.Ed.2d 674 (1975); Patterson v. American Tobacco Co., 535 F.2d 257 (4th Cir.), cert. denied 429 U.S. 920, 97 S.Ct. 314, 50 L.Ed.2d 286 (1976). Beyond this authority vouching for the general reliability of SMSA data for this purpose, the record indicates that it is an appropriate measure for these two sites. Neither site is on the periphery of the SMSA. While it is true that they are not located at the geographic center of the SMSA, they do lie at or near its social and economic focal point, the City of Wilmington. (Francese J-4). Moreover, based on the applicant flow log, we know that between 1972 and 1976, just over 90% of the applicants and hires resided in the Wilmington SMSA. Dr. Haber indicated that whenever one could locate the area from which 85 to 90% of the applicants or hires come, it is a good measure of the geographic hiring area. (C-105). Nor am I persuaded that in this instance an SMSA approach gives undue emphasis to relatively remote areas of the SMSA and too little weight to the black population of Wilmington which is relatively close by. The record indicates that the jobs at Chestnut Run and Christina Lab were well paid, desirable ones for which people who do not have more accessible, comparable opportunities would be willing to commute substantial distances. (Haber D-58-60; Siskin T— 32-35, 44 — 45). The black population of the Wilmington SMSA in 1970 was 12.2%. The black work force in the SMSA at that time was 11.2%. The Government correctly points out that this latter pool does not include those who have been unemployed but seeking work during the relevant period and that this group may have a disproportionate percentage of blacks. For this reason, the 11.2% figure may be somewhat low, particularly in the context of the mid-70’s. It is the best figure in the record, however, and, given DuPont’s overall hiring statistics, it is unnecessary to find more than that the available labor force in 1970 was in the neighborhood of 11.2%. 2. The overall hiring of non-exempts. Of all new hires at the two sites between 1966 and 1975, 17.5% were black. (DX-14). While it is true that the relatively large number of black hires during the 1972 to 1975 period makes this an unreliable indication of the experience of blacks during the 1966 to 1969 period, a yearly breakdown of the racial composition of new hires shows that 15.5% of all hires during the four years from 1966 to 1969 were black. (DX-14). The percentage of new hires thus substantially exceeded the percentage of blacks in the available labor pool, without taking into account job qualification. These figures demonstrate that DuPont’s personnel procedures in the post-Act period have not had a disparate impact upon blacks in terms of overall hiring. The Government’s case with respect to new hires thus turns on whether those procedures discriminated with respect to initial job assignment. 3. Initial job assignment. The statistical data relied upon by the Government in support of its claim of discrimination in initial job assignment is summarized in the following table which shows the total number of employees hired into various categories of jobs during the period from 1960-1971 and the percentage of those assigned who were black. NON-EXEMPTS INITIAL ASSIGNMENT UPON HIRE BY RACE AND JOB 1960-1971 1960-1964 1965 1966 1967 Job Tot. Blk. % Tot. Blk. % Tot. Blk. % Tot. Blk. % Lab Tech 128 0 0% 129 3 2.3% 83 1 1.2% 73 3 4.1% Test & Proc Tech 58 3 5.2% 36 1 2.8% 25 0 0 20 1 5.0% Mechanic 13 0 0% 25 1 4.0% 18 0 0% 4 0 0% Clerical (2, 3) 104 2 1.9% 20 0 0% 23 1 4.3% 16 1 6.2% Clerical (4, 5) 67 1 1.5% 33 1 3.0% 27 1 3.5% 25 0 0% Clerical (6 & up) 26 0 0% 19 0 0% 19 0 0% 12 1 8.3% Misc. Serv Oper 12 11 91.7% 36 26 72.2% 30 25 83.3% 17 16 94.1% Other 32 2 6.2% 16 0 0% 9 0 0% 3 0 0% 1968 1969 1970 Job Tot. Blk. % Tot. Blk. % Tot. Blk. % Lab Tech 123 8 6.5% 91 5 5.5% 0 0 0% Test & Proc Tech 41 8 19.5% 39 6 15.4% 0 0 0% Mechanic 4 0 0% 10 0 0% 0 0 0% Clerical (2, 3) 26 5 19.2% 24 4 16.7% 5 1 20.0% Clerical (4, 5) 25 1 4.0% 21 1 4.8% 9 1 11.1% Clerical (6 & up) 8 0 0% 4 0 0% 3 0 0% Misc. Serv Oper 30 29 96.7% 37 36 97.3% 0 0 0% Other 1 0 0% 17 4 23.5% 6 1 16.7% 1971 Job Tot.. Blk. % Lab Tech 0 0 0% Test & Proc Tech 4 2 50.0% Mechanic 0 0 0% Clerical (2, 3) 0 0 0% Clerical (4, 5) 2 0 0% Clerical (6 & up) 0 0 0% Util Oper Prod 21 9 42.9% Other 2 0 0% The Government points out that from 1965 through 1969 the overwhelming majority of blacks was still being assigned to MSO positions. Thus, from 1965 to 1969, of the 150 employees assigned to Miscellaneous Service Operator 132, or 88%, were black. Moreover, the Government stresses, those blacks who weren’t assigned to the Service work unit were assigned to the lower level lesser paying jobs among the technician and clerical positions. For example, of the 161 employees assigned to Processing and Testing Technician jobs (level 6) during this period 9.9% (16) were black; in contrast, of the 499 employees assigned upon hire to Laboratory Technician jobs (level 8) only 4% (20) were black. The Government also perceives a similar pattern in the clerical job area. Of the 109 employees assigned to level 2 and 3 clerical jobs, 11 or 10.1% were black, whereas of 131 employees assigned to level 4 and 5 clerical jobs, 4 (3%) were black and of 62 employees assigned to clerical level 6 and above, only 1 (1.6%) was black. Finally, the Government stresses that from 1965-1969 only 1 black was assigned to a mechanic’s job while 60 whites were being assigned to this job. DuPont’s principal response to the Government’s argument based on these statistics is that it erroneously assumes that all blacks in the available work force are qualified to perform each category of job. It maintains that the disparity between the percentages of blacks assigned on initial hire to higher and lower level jobs is attributable to the requirements of those jobs and the numbers of black applicants qualified to perform them. Primary reliance is placed upon the fact that the qualification tests administered to applicants have been conceded to be accurate predictions of the ability to adequately perform the jobs. This fact combined with other evidence tending to show equal access to testing for blacks and equal, if not preferential, treatment of blacks by employment personnel is said to demonstrate that DuPont’s employment practices have no unjustified disparate impact. There can be no argument with DuPont’s basic premise. If validated tests are the sole criterion of employment in the higher level jobs and there is equal access to testing, it follows that any statistical disparity which results is justified by the employer’s requirements. Or stated somewhat differently, in such circumstances the available pool is the pool of those who pass the test and if those who pass the test are hired to the jobs for which they have qualified, there will be no disparity in racial composition between those hired for a particular category of job and the available qualified labor pool. DuPont’s argument rests upon a factual assumption of equal access, however, an issue as to which there is conflicting testimony. Before addressing this credibility question, I will analyze the statistical evidence to determine whether it is of aid in answering that question and in resolving the overall issue of disparate impact. (a) The statistical evidence. The analysis of the statistical evidence must commence with the realization that labor market and hiring data cannot be evaluated without reference to the job skills involved. Hazelwood, supra. While this seems obvious, it is ignored by the Government’s argument that higher percentages of blacks assigned to lower level jobs than to higher level ones demonstrates disparate impact. Statistics showing a higher percentage of blacks assigned to lower level jobs does not, in and of itself, establish a disparate impact upon blacks. If the percentage of blacks assigned to the higher skilled jobs is reflective of those blacks in the labor pool who are qualified to perform them, for example, the fact that a higher percentage of jobs requiring lesser skills went to blacks does not indicate disparate impact. Thus, if 4% reflects the black participation in the available labor pool of those qualified to perform level 8 lab tech jobs, and 9.9% reflects the black participation in the comparable pool for level 6 jobs, the fact that 88% of the MSO jobs went to blacks would not alone suggest discrimination against blacks. The more reasonable inference would be that substantially more blacks than whites were willing to take the service worker jobs. During the period which the Government picks as the “worst case” of discrimination in initial job assignment, 57.5% of all jobs filled were technician jobs. If DuPont’s facially neutral hiring and assignment policies were being applied by its personnel in a discriminatory fashion, one would expect the statistics relating to this segment of new hires to reflect that fact. While the data is far from satisfactory in the lab tech area, I conclude that the statistics do not show disparate impact. As earlier noted, blacks constituted 9.9% of those hired to testing and processing technician jobs. Most of these 161 hires were to testing technician jobs. Testing technicians set up and operate testing equipment to carry out a variety of physical, chemical and optical tests. They tabulate data and calculate tests results. At the time of trial, testing technicians were receiving approximately $11,000 per year. The testing technician test seeks to ascertain whether an applicant has the basic math and measuring skills required for the job. While the skill level required is not high, it is safe to say that not all of the work force is qualified. Even if an 11.2% figure is. utilized as the black portion of the available labor pool, however, the disparity between the expected value of 18.032 black hires and the observed 16 blacks actually hired as testing and processing technicians is not statistically significant. While laboratory technician jobs differ in character depending upon the laboratory unit involved, their basic function is to carry out various chemical and textile processes for professional employees and to report the results to them. The work is not confined to the repetition of a single process. An ability to read and apply technical instructions and to work with formulas, percentages, ratios, fractions and the like is required. Based upon the testimony and the inspection of the sites, one can say beyond peradventure that a substantial segment of the labor force does not have the requisite skills to successfully perform these jobs. Ascertaining with any degree of precision what segment of the labor force is qualified, however, is not possible on the present record. All one can do is look to the best available evidence and try to get an idea of range in which the black population of the qualified labor pool might fall. The Government did not introduce any occupational breakdown of the labor force. While DuPont introduced occupational figures for the Wilmington SMSA from the 1970 census, the evidence concerning the composition of the possibly relevant categories is scant and the two categories which would appear to be the most relevant show substantially different black population. The “Chemical Technician” category (total population 2,166) shows a black population of 3.7%, while the substantially smaller “Clinical Lab Technologist and Technician” category (total population 288) shows a black population of 9.4%. If these categories are combined as a rough guide, the black population would be 4.4%. In addition to the SMSA figures, DuPont offers two other suggestions. Each is flawed but of some help. First, DuPont suggests that the kinds of skills required of a laboratory technician are ones ordinarily learned in high school and that there is a “high correlation” between the ability to perform these jobs and the level of skill of high school graduates. While the record cited in support of this “high correlation” is not strong, there is reason to believe that some correlation does exist. For example, 89% of those employed nationally in jobs which the census categorizes as science and engineering technicians have a high school education. If one makes the assumption that a high school education is some indicia of the qualifications required for the lab tech job and applies the best available 1970 Wilmington SMSA figures showing 33.3% of all blacks and 90.9% of all whites having completed high school, one would expect blacks to represent 3.7% of the qualified labor pool. DuPont’s second suggestion is based upon evidence introduced by the Government showing the pass-fail rates among 382 whites and 75 blacks who took the validated laboratory technician test prior to 1972. The pass rate for whites was 80% and for blacks was 25%. If one assumes a black labor force population of 11.2% and a 80%-25% pass rate for whites and blacks, respectively, one would expect 3.8% of the qualified labor force to be black. As the Government points out, the 25% pass rate is not a wholly reliable indicator of how the available black work force might fare when taking the test. While this testing data was put into evidence without explanatory comment, there appears to be no dispute that the people represented in the sample were people who were at some point actually employed by DuPont (and, therefore, had their test scores retained in their personnel folders). The Government, therefore, suggests it is likely that employees assigned to the Service work unit represented a substantial segment of the black sample and, accordingly, that a sample consisting of all black applicants might have fared better. While this is true, PX-61A suggests that the sample contained employees from a fairly broad spread of departments, many of which are not in the Service work unit. These available qualified labor force percentages are each subject to some question. They may be conservative, as the Government strenuously maintains. In the Court’s judgment, however, they represent the best indicia in the record of the qualified pool. Moreover, even if one assumes that the qualified labor pool during the 1965 to 1969 period was 5% black, the expected value would be 24.95 and the observed value in DuPont’s hires to lab tech jobs was 20. The difference is not statistically significant. Turning to the clerical area, the 10.1% black participation in the 109 hires to the unskilled and lesser skilled level 2 and 3 jobs does not represent a statistically significant deviation from the black percentage in the available general labor force. The record contains insufficient information to relate most of the data on hires to the level 4 and 5, and level 6 and above data with an available qualified labor pool. One can say that there is a disparity between the black participation in hires during the 1965-1969 period to Stenographer B positions and the participation of blacks in the “Stenographers” census category for the Wilmington SMSA in 1970. Of 98 hires to that position during this period, only 2 or slightly more than 2% were black. Of those employed as stenographers in the SMSA in 1970, 6.2% were black. This figure must be viewed, however, in light of the fact that the black stenographer population in the same area in 1960 was only .3%. While it is concededly a guess, if one. assumes a 5% average availability pool throughout the 1965 to 1969 period, the expected value would be 4.9, as compared with the observed value of 2. While I consider this evidence regarding the hiring of stenographers to be relevant on the issue of disparate impact, the disparity is not sufficient to make the inference of discrimination compelling. The expected value of 4.9 is 1.35 standard deviations greater than the observed value of 2. Under the standards suggested in Castaneda v. Partida, supra, for substantially larger pools, a disparity of this size does not require a conclusion that race was a factor in the selection process. The Government’s strongest statistical case is in the area of hires to skilled craft positions, i. e., electricians, instrument mechanics, sheet metal workers, pipefitters, machinists, millwrights, carpenters and welders. During 1965 to 1969 only 1.4% of new hires were black. The 1960 and 1970 SMSA figures for black craft positions were as follows: Black Availability — Wilmington SMSA Job Category 1970 fDx271 1960 (Table 122) Carpenters 2.3% (51/2214) 3.6% (57/1591) Electricians 1.7% (31/1816) 1.0% (13/1277) Machinists 4.3% (48/1128) 2.2% (25/1124) Millwrights 2.0% (12/604) 0% (0/433) Sheetmetal Workers 3.1% (15/490) 2.0% (7/357) Welders and Framecutters 9.2% (130/1412) 5.0% (50/993) While the record does not contain a breakdown of hires which would permit a detailed analysis, it appears that there is a significant disparity in this area. In summary, I find the statistical data concerning initial hires as a whole to be inconclusive. While some portions of it standing alone might support a finding of disparate impact, the total statistical data do not require such a conclusion. Accordingly, I turn to the non-statistical evidence. (b) The non-statistical evidence. Throughout the post-Act period it has been the stated policy of management to hire qualified blacks at all levels. There have been standing instructions throughout the period that no qualified black was to be turned down if there was an opening. The personnel people understood the significance of these instructions. Each year forecasts had to be made, goals had to be set for hiring minorities, and the results had to be reported to the AEC (later ERDA) and the EEOC. Compliance reviews were conducted by the AEC in January of 1964, September of 1969, and May of 1972. There is evidence, which I credit, indicating that management’s instructions were carried out. At no time was a black who passed the laboratory technician test not hired as a laboratory technician when a position was open. The same can be said for any black who qualified on the clerical test. Prior to 1969, only one black who qualified by test for a testing technician job was denied placement in an opening in that job category. Afte