770 F.2d 520
38 Fair Empl.Prac.Cas. 1605,
38 Empl. Prac. Dec. P 35,506
Thomas J. SMITH, Jr., et al., Plaintiffs-Appellants,
E.D. Herron, et al., Intervenors-Appellants,
WESTERN ELECTRIC COMPANY, INCORPORATED, Defendant-Appellee.
United States Court of Appeals,
Sept. 13, 1985.
Law Offices of James C. Barber, James C. Barber, Dallas, Tex., for plaintiffs-appellants.
Thompson & Knight, Stephen F. Fink, Dallas, Tex., for defendant-appellee.
Appeal from the United States District Court for the Northern District of Texas.
Before RUBIN, WILLIAMS and DAVIS, Circuit Judges.
W. EUGENE DAVIS, Circuit Judge:
This is an appeal from a judgment in favor of defendant Western Electric Company in a Title VII and Section 19811 race discrimination suit after a trial on the issue of liability. Because the district court's findings of fact are not clearly erroneous, we affirm its judgment.
The initial plaintiffs were six black installers who claimed discrimination in hiring and promotion by Western Electric's North Texas installation department. A class was certified in August 1981 to include all black applicants and blacks who had been or were installers and who had been subjected to race discrimination in recruitment, hiring, promotion, training, assignment, termination, pay, and any other terms and conditions of employment from March 1973 to the present. At the pretrial conference, plaintiffs dropped their allegations regarding pay, and the opening date of the class was modified to September 30, 1974. On appeal, plaintiffs focus their challenge on discrimination in recruiting, hiring, promotion, and termination.
The defendant maintains twelve base locations in its north Texas area with the central office located in Dallas. Each installer is assigned to a base location, where he routinely does most of his work.
The general duties of the installer include installing and modifying equipment in the telephone company's central office and in large central systems for the telephone company's major industrial, commercial and governmental customers. The work ranges from simple jobs, such as assembling and mounting the electronic framework, to the complex operations of installing, operating and testing complex electronic equipment. This work is performed at a variety of jobsites and the installer is often called upon to travel to jobsites at other locations both in and out of the north Texas area.
The terms and conditions of employment for the installers are governed by a collective bargaining agreement between Western Electric and the Communication Workers of America.
The installer positions at issue in this suit are those at index levels 1 through 5. Supervisory positions above the level of index 5 are not at issue.
For simplicity, we will discuss the particular facts pertinent to each issue during our discussion of that issue.
Plaintiffs contend that job openings were not widely advertised and that potential job applicants learned of openings only through word of mouth. Plaintiffs argue that this recruiting program tended to perpetuate a white work force, by preventing potential black applicants from learning of job openings.
Based upon adequate record evidence, however, the district court concluded that the defendant did not discriminate in its recruiting practices. The evidence was overwhelming that job openings were always listed with the Texas Employment Commission and were occasionally advertised in local newspapers. Additionally, the large number of blacks who applied for the index 1 entry level position belies the plaintiffs' contention that defendant's recruiting practices prevented them from learning of job openings; over thirty percent of the total applicants were black.2 The district court's conclusion that the employer's recruiting practices were not designed with a discriminatory purpose in mind and that these practices had no discriminatory impact on blacks is not clearly erroneous.
Applicants are generally hired by defendant only at index level 1, the entry level rank. These positions usually are filled with applicants without special training or skills. The defendant hired installers on only ten different occasions during the time period at issue to fill positions at base stations in Lubbock, Longview, Midland, Dallas, and Fort Worth.3 This hiring was done by the personnel section of defendant's installation department. During the relevant time period, 103 installers were hired; eighty-one of these were white, and twenty-two were black.
The record reflects that hiring is conducted in several stages. First, the applicant completes an application form and takes a color vision test. Applicants who are to any extent colorblind are not encouraged to continue in the hiring process. Those who do continue are then interviewed by one of the three interviewers from the personnel section. The three interviewers are guided by Western Electric's interview manual. This manual provides the company's hiring criteria and policies as well as recommended interviewing methods. The manual instructs the interviewers to apply the hiring criteria fairly and without discrimination. At the interview, the applicant is informed of the mobility requirements of the job and is questioned about his willingness to accept this condition of employment. The applicant is also routinely asked about educational background, job experience and criminal record. If the applicant cannot accept the job mobility requirement, he is screened out. Finally, the remaining applicants are given a battery of tests. After test scores are compiled, the chief of personnel and the individual interviewer make the hiring decisions.
The parties dispute the weight given two objective criteria in the hiring process, the test battery and prior education. Plaintiffs contend that the test score was used as a screening device so that those who did not make a minimum score on the test were not considered for employment. The employer concedes that the test was a weighty selection factor, but insists that making a minimum score on it was not an absolute prerequisite to consideration for employment. The district court found that the test battery was not an absolute qualifier, but was a very important factor in the selection process. The record supports this finding: 159 applicants scored at least 173, the recommended minimum, and only ninety of those were hired; yet thirteen applicants who scored less than 173 were hired.
Plaintiffs also argue that a high school diploma was a prerequisite for hiring and a barrier to minority hiring. The employer contends that a high school education was a definite preference, not a requirement. It is unnecessary to consider this argument however because all named class plaintiffs are high school graduates and consequently they lack standing to challenge this alleged requirement. Phillips v. Joint Legislative Committee, 637 F.2d 1014, 1026 n. 20 (5th Cir.1981), cert. denied, 456 U.S. 960, 102 S.Ct. 2035, 72 L.Ed.2d 483 (1982); Payne v. Travenol Laboratories, Inc., 565 F.2d 895, 898 (5th Cir.), cert. denied, 439 U.S. 835, 91 S.Ct. 118, 58 L.Ed.2d 131 (1978). See Fleming v. Travenol Laboratories, Inc., 707 F.2d 829, 832-33 (5th Cir.1983).
A total of 596 applications for index 1 installer was received during the relevant period; 225 of these were black and 371 were white. From this applicant pool, 103 installers were hired; twenty-two of these were black and eighty-one were white.
Plaintiffs' case for discriminatory failure to hire is based almost entirely on statistical evidence which they contend shows gross disparity between black and white hiring. Plaintiffs established that 21.8% of the white applicants were hired, yet only 9.8% of the black applicants were hired. Based on the ratio of black applicants to the total applicant pool, Dr. William Schucany, plaintiffs' expert, calculated that 38.7 blacks should have been hired, and that the number hired, twenty-two, was 3.8 standard deviations below the expected. In Castaneda v. Partida, 430 U.S. 482, 496-97 n. 17, 97 S.Ct. 1272, 1281 n. 17, 51 L.Ed.2d 498 (1977), the Court observed that: "As a general rule for ... large samples, if the difference between the expected value and the observed number is greater than two or three standard deviations, then the hypothesis that [the selection was random] would be suspect to a social scientist."
The principal disputes between the parties as to the validity and the weight that should be given the statistical evidence was in the following areas: (1) What was the appropriate labor pool figure? Plaintiffs urge that the pool of applicants considered for the index one installer position was appropriate (applicant flow). The defendant, on the other hand, contends that the available work force in the general population as reflected by the 1970 census was the most appropriate. (2) To what extent should the raw data be adjusted or controlled for particular factors that operate in the decision to hire, such as the score on the test battery, willingness to comply with the mobility requirements, prior work experience, criminal records, and color vision?
It is unnecessary for us to examine critically every finding and conclusion of the district court that relates to the above contentions, because plaintiffs clearly cannot prevail if the employer was entitled to consider the test score in making the hiring decision. In other words, no serious argument is made that the district court committed clear error in finding for defendants on this issue if the district court was entitled to conclude from the evidence that these test results were "validated" or assisted the employer in predicting the prospective employee's job performance.4
The district court found, and the record shows, that the test battery was validated under the uniform guidelines for employment tests established by the Equal Employment Opportunity Commission, (29 C.F.R. Sec. 1607). These guidelines are entitled to great deference. See Albemarle Paper Co. v. Moody, 422 U.S. 405, 431, 95 S.Ct. 2362, 2378, 45 L.Ed.2d 280 (1975). Plaintiffs' witness, Dr. Alan Brown, conceded that the tests were validated under the EEOC guidelines, and he offered no reason to disregard them except for his argument that their use was not justified because of the low correlation with job success. Although the correlation was not high, the district court, giving due deference to the guidelines, did not err in finding the tests a useful predictor of job performance.
In sum, the district court's finding for the defendant on this issue under both a disparate impact and disparate treatment theory is not clearly erroneous. The district court was entitled to find that the statistical disparity in hiring demonstrated by plaintiffs was legally insignificant because of plaintiffs' failure to control for the test battery scores.
Plaintiffs next challenge the district court's finding of no disparate impact or treatment in promotion. Under the collective bargaining agreement between Western Electric and the union, wages are dependent on the skill category or index to which the employee is assigned. Western Electric's index plan provides for five index levels (1-5), and each installer is assigned to one of the five indexes depending on his skill level. All new employees in this department begin at index 1 and are promoted up the ladder to a higher index when they meet the requirements for promotion as outlined below. The higher levels are filled by promotions from the lower ranks.
The index plan divides installation work into different coded work operations. To be considered for advancement to the next index, an installer must become proficient in the operation codes required for that higher index and must perform that "above index" work for a specified number of hours.
Each week, installers complete time tickets on which they record the type of work they have done that week, assign each different operation a work code, and record the number of hours spent in each code. Following review and approval, disapproval or possible modification by the job supervisor, the time tickets with notation of hours spent in each work code are forwarded to a central office where they become the basis of an individual installer's experience record. The employer maintains for each installer this experience record on Form 1421 which summarizes by codes and hours his work history. The area supervisor in each base location uses the 1421 Forms to find the personnel with the appropriate skills to perform a particular job.
Every six months, each installer undergoes an index review. At this time management decides whether each installer will advance to the next index level. Although three levels of supervisors review the records of the installer's work for the preceding six months, it is the role of the job installation supervisor, the line supervisor, to specify whether a worker is "qualified" in particular work operations, based on his recollection of the installer's work. The corporate manual provides guidelines to the installation supervisor to aid him in determining whether an installer is sufficiently proficient in a given operation to be considered "qualified." The supervisor's ultimate decision on this question, however, is largely subjective.
Supervisors assign work for a specific job to those who are already in the index level required for that job, and then, as needed, will assign the work to others who are below that particular index level. Although seniority is not a requirement for advancement, long term employees have more opportunity to work in higher code work and this enhances their chances for promotion.
The plaintiffs urge that the index plan operated to the detriment of black employees. They contend that they were at the mercy of primarily white supervisors who could deny them promotions by subjectively concluding that they were not qualified in one or more work operations. They also complain that supervisors channeled more of the available higher level code work to white employees, which had the effect of blocking or delaying eligibility of black employees for promotion.5
B. Plaintiffs' Statistical Evidence
Plaintiffs relied heavily on statistical evidence to establish their claim that blacks were discriminatorily denied promotion. The principal statistical study they relied upon was a "survival analysis" conducted by plaintiff's expert, Dr. Schucany. In this study, plaintiffs' expert followed the progress of installers hired by defendants in north Texas each year from 1965 to 1970, with a particular focus on how soon after hiring each installer was promoted from index 1 to 4 and then from 1 to 5. A similar analysis was made for promotion from index 1 to 3; a study of the progress of installers hired in 1978 was added to this analysis.6 This study is reproduced in the margin.7 For each cohort--the group hired each year the study was made--plaintiffs' expert arrived at a figure which purported to represent the proportional advantage to a faster promotion the white employee enjoyed over the black employee.
With respect to the analysis of promotions from index 1 to index 3, Dr. Schucany concluded that significantly fewer blacks were promoted with the same rapidity as white employees; this disparity was expressed at a level of 2.3 standard deviations. On his studies of promotions from index 1 to 4 and index 1 to 5, he found much less significant disparity.
The defendant raised serious questions about the validity of Dr. Schucany's conclusions from his survival analysis. The defendant points out that the plaintiffs' expert commingled data from before 1974--the beginning of the class period--with data after that date. In fact the only data that was generated exclusively from liability period facts was that which related to employees hired in 1978 in the analysis for promotions from index 1 to index 3. In short, the value of the conclusion Dr. Schucany draws in this study is limited because of his inability to document which conduct occurred during the liability period. As indicated above, the only study in this survival analysis that showed statistically significant disparity was the one that considered promotions from index 1 to index 3 for installers hired from 1965 to 1970 and 1978. Because advancement from index 1 to index 3 ordinarily takes four to five years, it is reasonable to infer that most of the promotions considered in this study occurred outside the liability period.
Defendant's expert, Dr. Myrick, also criticized Dr. Schucany for failing to adjust or control the raw data for the individual installer's base job location. The defendant produced evidence which indicated that the rate at which an employee could expect promotion depended to some extent on his job location. This was attributed to the fact that some job locations had more work coded in the higher indexes and employees stationed at such locations obviously received more opportunity to be credited with above-index work. Also, the defendant produced evidence that tended to show that at least in indexes 4 and 5, where more specialized knowledge and higher skills were required, Dr. Schucany's analysis was flawed because he did not control for special skills acquired as installers became specialists in certain systems as they advanced to the higher levels.
C. Defendant's Statistical Case
Defendant conducted a completely different analysis of promotion rates. Defendant's expert, Dr. Myrick, conducted "T" tests in which he computed and compared average time to promotion for black and white installers during various time periods from 1964 through 1981. He found no significant difference in the time to promotion between black and white installers hired between 1964 and 1981. In approximately 50 percent of the instances where Dr. Myrick discovered disparities, black installers were promoted quicker than whites. This study also had significant limitations. It only considered installers who were actually promoted and did not consider those employees who never received a promotion. Also, if in a particular time period promotions were given to all black installers or all white installers so that no comparison was possible, this data was not considered.
Dr. Myrick also performed a chi square test designed to determine whether black and white installers were promoted in proportion to their representation in the installation department of defendant's north Texas operations. This study first considered all actual promotions received by black and white installers for each year of the liability period. That figure for each year was then compared to the number of promotions expected for installers of each race based on the racial composition of the north Texas installation department. This study reflected that actual promotions of black installers exceeded expected promotions in six out of the eight years studied and by statistically significant numbers in two of those years. The disparities during the two years in which the white promotions were disproportionate to black promotions were statistically insignificant. As a part of the same study the defendant also isolated each index from 2-5 and, based on the ratio of black installers to the total installers at the index level under consideration, tested for expected black promotions to each rank during the entire class period. He then compared the expected black and white promotions to actual promotions in each race. Dr. Myrick found no statistically significant disparity from this study. Plaintiffs' expert criticized this study because it does not consider the length of time each installer remains at a particular index level before promotion; for example, these figures do not distinguish between the promotion of an installer who is promoted to index 3 after remaining in index 2 for five years from another installer who receives a similar promotion after serving in index 2 for only two years.
The statistical evidence from neither side is persuasive. Plaintiffs' statistical studies are based primarily on facts that occurred outside the liability period. The failure to control for job location and special skills in analyzing disparity in promotion to indexes 4 and 5 also limits the value of plaintiffs' conclusion. As indicated above, the defendant's statistical studies also had serious limitations. Defendant's "T" tests excluded from consideration installers never promoted during the time period under consideration and promotions to all white installers or all black installers during a particular time period. Also, the chi square test considered all promotions without regard to length of time required for a promotion.
In sum, we conclude that the district court was justified in rejecting as unconvincing the plaintiffs' statistical evidence. This rejection is supportable more on the limitations of plaintiffs' studies than on the strengths of the defendant's studies.
D. Non-Statistical Evidence
Plaintiffs offered the testimony of several black installers who complained that they were treated in a discriminatory fashion by the defendant. More particularly, they complained that white installers received better assignments, had more access to training programs, and were permitted by white supervisors as a matter of course to record their work under higher index code numbers than black installers performing the same work. They also related instances when supervisors addressed them with racial slurs. Plaintiffs also bolstered their statistical proof by showing that the assignment and index review process is subjective. The job supervisors who assign specific jobs to particular installers are, with two exceptions, white. Assignment of above index work is discretionary, and is based in part on the supervisors' subjective assessment of the individual installer's skills. These job supervisors also approve or disapprove an installer's weekly assessment of which code includes the particular work just performed by the installer. When supervisors consider whether an installer is "qualified" in a code, they must rely to a large extent on their recollection of the quality of his work during the past six months because few helpful records are maintained for this purpose. The defendant contends that such record keeping is too burdensome on supervisors and that this deficiency is offset by other checks. The employer points to the manual that provides detailed instructions for the index plan's administration and a complete description of work codes, the weekly records of installer time in different work codes, and a four-stage process to hear grievances from installers who feel they have been unfairly denied promotions. The defendant also made the point that installers work on many different jobs under the supervision of several supervisors so that the biased subjective judgment of several supervisors would be required to block the advancement of a deserving black installer.8
In conclusion, the plaintiffs' statistical evidence was not convincing and the district court's rejection of it is not clearly erroneous. As the Supreme Court has cautioned, "statistics are not irrefutable; they come in infinite variety and, like any other kind of evidence, they may be rebutted. In short, their usefulness depends on all of the surrounding facts and circumstances." Teamsters v. United States, 431 U.S. 324, 340, 97 S.Ct. 1843, 1856, 52 L.Ed.2d 396 (1977).
The record fairly supports a conclusion that many of the decisions made with respect to job assignment and promotion were subjective in nature and if the defendant was inclined to discriminate, it had ample opportunity to do so. An opportunity to discriminate, standing alone, will not establish discriminatory impact or treatment and the record does not contain sufficient evidence, statistical or otherwise, to undermine the district court's findings.
Plaintiffs' least substantial claim is that the defendants discriminatorily terminated blacks. The plaintiffs' case was based almost exclusively on one statistical study of black and white installers terminated by defendants from July 2, 1965 until December 1981. Plaintiffs' expert developed from defendants' records that during this period a total of seven black and seven white installers were terminated from a work force of 463 whites and eighty-five blacks, or a total of 548 employees. The termination rate for white employees was 1.5 percent, and that for blacks was 8.2 percent.
This study was seriously flawed and therefore unconvincing. Only four of the fourteen terminations occurred between September 1974 and the trial, which, of course, was the liability period. Plaintiffs' expert did not isolate and evaluate the terminations that occurred during the liability period and the record is therefore silent on the statistical significance of the four terminations. In any event, the four terminations represented such a small sample size, it is doubtful that any statistical study of them would have been helpful. See Mayor v. Educational Equality League, 415 U.S. 605, 621, 94 S.Ct. 1323, 1333, 39 L.Ed.2d 630 (1974).
The company rebutted plaintiffs' proof with evidence that written procedures and standards guided terminations. Further, it showed that discharge was subject to the safeguard of a grievance and arbitration procedure. In addition, Western Electric advanced evidence of legitimate reasons why twelve of the fourteen employees were fired. Although in a class action an employer cannot attack a prima facie case that is based on statistical evidence with proof concerning a few individual cases, Boykin v. Georgia-Pacific Corp., 706 F.2d 1384, 1393 (5th Cir.1983), cert. denied, --- U.S. ----, 104 S.Ct. 999, 79 L.Ed.2d 231 (1984), Western Electric, in this case, provided explanations for most of the terminations in the sample, in addition to challenging the validity of plaintiffs' proof. See Lewis v. NLRB, 750 F.2d 1266, 1274 n. 11 (5th Cir.1985).
The district court did not err when it found no discrimination in terminations.
The district court's conclusion in this case that defendant was guilty of no classwide violation of Title VII is not clearly erroneous and is therefore
Because the same analyses apply to claims under section 1981 as under Title VII, Rivera v. City of Wichita Falls, 665 F.2d 531, 534 n. 4 (5th Cir.1982), we shall consider this claim together with the Title VII claim. The district court made no separate finding on this issue
As a reference point, blacks made up approximately twelve percent of the available work force in north Texas in 1970
This hiring all occurred between 1977 and 1979
Black and white applicants alike who scored between 150 and 173 were hired in roughly equivalent percentages. Plaintiffs' expert therefore frankly conceded that if the employer was entitled to consider the results of the test battery in making the decision to hire, black and white applicants were treated equally in the hiring process
We reject appellants' assertion that Crawford v. Western Electric Co., 614 F.2d 1300 (5th Cir.1980) on second appeal, 745 F.2d 1373 (11th Cir.1984), precludes relitigation of the issue of whether appellants established a prima facie case of discrimination in promotion. The court in Crawford simply determined that the index plan was discriminatorily applied to four installers in Western Electric's Jacksonville, Florida installation division. The issues in Crawford thus turn on different actions by different supervisors and over a different period of time. Because the relevant facts in the two cases are almost entirely different, neither res judicata nor collateral estoppel precludes a relitigation of the issue presented in this case. See generally Wright, Miller & Cooper: Federal Practice and Procedure Sec. 4417 (1977); New Orleans Steamship Association v. EEOC, 680 F.2d 23, 25-26 (5th Cir.1982)
The only year between 1974 and 1978 that installers were hired in north Texas was 1978
ANALYSIS 1 OF PROMOTION TIMES TO INDEX 3 TOTAL PROMOTEES GEHAN 3 PROPORTIONAL 4 INDEX 1 ----- --------- ------- ------------- YEAR W k B B W k B B STATISTIC ADVANTAGE ----------- --------- ------- --------- -------- --------- -------------- 1965 2 31 5 30 4 29 .61 1966 57 2 57 2 38 .67 1967 68 5 67 5 -113 .32 1968 20 2 20 2 18 .75 1969 118 32 113 29 645 .62 1970 108 12 93 9 241 .60 1978 28 8 15 3 33 .60 --- -- --- -- --- -- TOTAL 430 66 395 54 891 .60
Combining across cohorts yields an equivalent Z-value of 2.30. In a racially
neutral environment, the probabiltiy of a Z-value as large as this is .011.
Only time to first service interruption is analyzed; excludes installers
definiteely hired outside North Texas.
1965 cohort consists of installers with Index 1 date between July 2, 1965
and December 31, 1965.
Number of instances in which the time to promotion for a black is greater
than the time to promotion for a white minus the number of instances of white
greater than black.
Estimated probability that a randomly selected white installer would be
promotedd to Index 3 before a similarly situated randomly selected black
Plaintiffs conducted a statistical study which isolated a six-month period in 1974 and another similar period in 1982 when black installers were assigned less above code work than white installers. For these two six-month periods the statistical study reflected a disparity that was significant. The record does not reflect any basis to conclude that the two six-month periods selected were representative; in fact the record does not establish any particular reason why these periods were selected for the study. We do not find the district court's rejection of this statistical study clearly erroneous