The case history is long, from 1967, when the first complaint was lodged with the EEOC, to this appeal, the third before this court. In 1976, six present and retired black employees at the Gulf Oil Corporation’s Port Arthur, Texas refinery filed this suit. The complaint alleged a variety of racially discriminatory practices in violation of the Civil Rights Act of 1866, 42 U.S.C. § 1981, and Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e
et seq.
In 1977, the district court dismissed the Title VII claims on procedural grounds, and granted summary judgment on the § 1981 claims
{Bernard I).
This court reversed.
On January 24, 1983, the district court provisionally certified a class of all blacks who had worked or applied for work at the refinery in union covered jobs at any time after December 26, 1966. Pursuant to the order of the district court, written notification of the class action and certification was mailed in September 1983 to the members of the class. In April 1984, the district court modified that certification order to exclude a substantial number of the original class members, and this court upheld that modification.
Bernard IV,
The class appeals from the district court’s ruling in favor of the defendants and from the denial of the motion to notify former class members. The class contends that the Gulf seniority system is not bona fide, that Stipulation 29 was intentionally discriminatory, and that the tests used by Gulf to determine who was eligible for promotion were discriminatory because they had an adverse impact on blacks and were not sufficiently job related. In addition, three individual plaintiffs, Mr. Brown, Mr. Hayes and Mr. Tizeno, appeal from the district court’s ruling that Gulf did not discriminate in not promoting them. We affirm because we are persuaded that the district court was not clearly erroneous in its determination that Stipulation 29 was not purposefully discriminatory and was justified by valid business reasons, the craft tests were sufficiently job related, and there was no discrimination in the failure to promote Mr. Brown, Mr. Hayes and Mr. Tizeno.
I. Factual Background
Gulf’s Port Arthur refinery has had three basic work areas throughout its history: operations and maintenance (“crafts”), which require skilled employees, and until 1954 were staffed exclusively by whites; and labor, which is composed of unskilled employees who assist the craft groups, and until 1956 was exclusively comprised of blacks.
In 1950 the labor department had many subdepartments, each assigned to work with a specific craft department. The twenty integrated craft departments effectively had two separate lines of progression, one for their white employees and the other for the black employees assigned to their departments. Although black laborers effectively worked in many of the same departments as white craft employees, they could not bid into the more desirable white lines of progression. Between 1954 and 1956, the craft and labor departments were assigned to one of two divisions. The craft departments were assigned to the Operat-. ing and Mechanical (“0 & M”) Division, and the labor department, with its various subdepartments, was assigned to the Labor Division. As of 1956, both black and white employees progressed up their separate promotional lines based on plant seniority and ability to perform. Newly hired blacks would be assigned to the Labor Division, while newly hired whites would be assigned to one of the craft departments in the 0 & M Division.
The 1956 contract between Gulf and the OCAW made the “Laborer” classifications in the Labor Division the entry level position for all employees, both black and white. After progressing to the top position in one of the Labor lines of progression, an employee could transfer to an entry level position in one of the 0 & M lines of progression after bidding into the mechanical helper pool, which was a classification apparently created to provide apprentice-type instruction for employees entering an 0 & M line of progression. Bids for entry into the 0 & M lines of progression were selected on the basis of plant seniority. An employee who successfully bid into an 0 & M line of progression was assigned dual seniority, 0 & M Division seniority and plant seniority. White employees generally had more 0 & M seniority because of the earlier practice of hiring whites directly into the craft positions. In 1963, divisional seniority was eliminated, and plant seniority became the basis for all promotions, demotions and lay offs. The effect of this change was that blacks with more plant seniority but less 0 & M seniority than whites in the same job classification could compete for promotions to the next highest classification based upon their longer plant seniority, thereby bypassing white employees with greater 0 & M seniority.
In 1967, Gulf management became convinced that its plant was inefficient. As part of a general restructuring, Gulf and OCAW made a special agreement, Stipulation 29, which reclassified workers in the *738 maintenance departments. 1 As part of the restructuring the Labor Division was eliminated and its employees were assigned to the various crafts with which they had been working. The lines of progression were streamlined and the categories of “mechanical helper” and “craft helper” were eliminated. Those who were mechanical helpers were reclassified as “utility men” (a demotion), while Stipulation 29 reclassified the craft helpers as “craft trainees” (a promotion) if they passed a simple test. 203 of the 204 who took the test passed and were promoted to mechanical trainee. 95.6% of those promoted were white. A large portion of the utility men and mechanical helpers were black.
Both before and after Stipulation 29 all others not transferred as a result of Stipulation 29 were required to pass a battery of written tests in order to become craft trainees. These tests had an adverse impact on blacks in that blacks had a significantly lower pass rate than whites. The “Old Tests” were administered prior to 1971, and the “New Tests” after 1971. 2 Both sets of tests had an adverse impact on blacks. 3 No validation study was ever done on the old tests, but Gulf has attempted to show that the new tests are job related through a study to determine the correlation between test scores and job performance.
II. The Seniority System and Stipulation 29
The class first argues that the seniority system is not bona fide because it locks in the effects of pre-Act discrimination. This issue was resolved by this court in
Ber
*739
nard IV,
[W]e question whether a seniority system which is neutral as of the effective date of Title VII, which is based on plant seniority, and which has a single bargaining unit could ever be held unlawful solely because of pre-Act discrimination. That the 1963 changes did not rectify the effects of past discrimination, and in fact operated in some ways to lock those effects in, does not imply, in the absence of purposeful discrimination in connection with the post-Act system, that this system was not bona fide under 703(h). See Teamsters,481 U.S. at 353 ,97 S.Ct. at 1863 (seniority system does not become illegal “simply because it allows the full exercise of the pre-Act seniority rights of employees of a company that discriminates before Title VII was enacted”).
The class really makes two arguments with respect to Stipulation 29: first, that through its operation to make the seniority system non-bona fide; and second, that in and of itself it was a discriminatory act by the employer in violation of Title VII because of its disparate impact on blacks.
The district court found that Stipulation 29 was not purposefully discriminatory and was justified by valid business reasons. The class contests these findings. The standard of review for such a decision is whether, looking at the record as a whole, the district court was clearly erroneous in its determination that there was no purposeful discrimination and that the action resulting in disparate impact was justified by legitimate business reasons. Fed.R. Civ.P. 52(a).
See Anderson v. Bessemer City,
Before Stipulation 29, black utility men could become mechanical helpers and craft helpers, joining whites with less seniority who had started in higher positions than blacks because of the earlier discrimination. When positions opened for craft trainees, those blacks could bypass the whites with more craft seniority if the blacks had longer plant seniority. Stipulation 29 worked to lock blacks into the lower positions by promoting the craft helpers, who at that time were mostly white and often had less plant seniority than blacks, into the trainee positions. Blacks were precluded from opportunities they would have had earlier if it weren’t for Stipulation 29. In
Bernard IV,
this court found that senior blacks were denied the opportunity to bypass junior whites, and the class had, therefore, established a prima facie case of adverse impact.
Section 703(a), 42 U.S.C. § 2000e-2(a), proscribes discriminatory employment practices, including discrimination “with respect *740 to ... compensation, terms, conditions, or privileges of employment” on the basis of race, but section 703(h) insulates bona fide seniority systems from these dictates, providing that
[notwithstanding any other provision ..., it shall not be an unlawful employment practice for an employer to apply different standards of compensation or different terms, conditions, or privileges of employment pursuant to a bona fide seniority or merit system, ... provided that such differences are not the result of an intention to discriminate because of race....
42 U.S.C. § 2000e-2(h).
Whether a seniority system, neutral on its face, is bona fide depends upon “whether there has been purposeful discrimination in connection with the establishment or continuation of [the] seniority system.”
Bernard IV,
(1) whether the seniority system operates to discourage all employees equally from transferring between seniority units;
(2) whether the seniority units are in the same or separate bargaining units and, if separate, whether that structure is rational and industry-wide;
(3) whether the seniority system had its genesis in racial discrimination; and
(4) whether the seniority system was negotiated and maintained free from any illegal purpose.
Id. at 352.
This court found the first three of the factors in favor of defendants, but remanded to the district court to consider whether Stipulation 29 showed that the system had been maintained with a discriminatory purpose.
Bernard IV,
The class points to evidence that Stipulation 29 was racially motivated, but there was also evidence of legitimate reasons for the reorganization of employees. In addition, Gulf had various programs to increase mobility for blacks in its plant. It had an affirmative action program and several training programs to try to insure that blacks would be able to rise within the company. The record supports the district court’s determination, and it is not clearly erroneous that Stipulation 29 was not adopted for purposefully discriminatory reasons.
Under a disparate impact theory as set out in
Griggs v. Duke Power Co.,
Under
Ward’s Cove,
if a plaintiff makes out a
prima facie
case by showing that a specific practice has a disparate impact on a protected class, then the burden shifts to the employer to produce evidence, but not to prove, that the “challenged practice serves, in a significant way, the legitimate employment goals of the employer.”
*741 The district court relied in part on the testimony of Dr. Milden J. Fox, an expert in Industrial Relations, who studied four refineries as part of his dissertation, which was entered into evidence. Dr. Fox testified at trial that Stipulation 29 was part of an ongoing effort to improve efficiency at the Refinery. He testified that before Stipulation 29 the method of work assignments contributed to excess maintenance costs, and that management sought to increase efficiency through measures permitting more flexible use of employees, including Stipulation 29. Stipulation 29 was designed to take employees out of specific craft job classifications, and move them into the trainee positions, making them “universal mechanics,” i.e., people who could perform any job associated with a particular craft. Dr. Fox testified that various measures, including Stipulation 29, conformed with industry practices at the time. The testimony of Mr. Charles Draper, who spent 30 years in the Refinery’s personnel department, was consistent with Dr. Fox’s testimony. The district court found that the craft helpers reclassified under Stipulation 29 had at least ten years’ experience in their craft, and as craft helpers they performed a significant amount of the work of the No. 1 (journeyman) craftsmen, so it would take minimal additional training to qualify them as journeyman craftsmen. The district court found that mechanical helpers and laborers did not have comparable experience, so they would require much more training to do the work of journeyman craftsmen efficiently and safely. This is supported by the record, which shows that the mechanical helpers and laborers were shuttled from craft to craft to do whichever labor was necessary to help those in the crafts, but did not receive training at any one specific craft.
Plaintiffs contend that the district court misunderstood the import of Stipulation 29, that it was not a measure designed to change lines of progression, but simply the promotion of 203 mostly white craft helpers. Plaintiffs go too far in simplifying Stipulation 29; it cannot be observed in isolation, but must be considered in the context of Gulf’s reorganization. The district court found that Gulf wanted to simplify the lines of progression through reorganizing, and wanted to increase the number of craft trainees, leading to an increase in the number of persons who were trained to handle all aspects of a craft. In order to do that Gulf had several options. One option was Stipulation 29, which moved workers into trainee positions from the next lowest position. Another option (suggested by the class) would have been to demote all the craft helpers to utility men, then draw craft trainees from the pool of utility men that included senior blacks who had been utility men and junior whites who had been craft helpers. The district court found that Gulf had a legitimate reason for choosing the Stipulation 29 option — it wanted to reclassify as trainees those workers with the most
craft
experience so they would need less training. The district court specifically found that the craft helpers all had at least 10 years of craft experience. The utility men who would have been in the pool under the second option would not have had such experience in
craft
work, even if they would have had greater
plant
seniority. The class has not shown that the proposed alternative would be equally as effective as Stipulation 29 in meeting Gulfs objectives.
See Ward’s Cove Packing Co., Inc. v. Atonio,
- U.S. -,
III. The Craft Tests
The class claims that the tests used to determine which employees were eligible for promotion to journeyman craftsmen were discriminatory. It claims that the tests had an adverse impact on blacks because the pass rate for blacks was significantly lower than the pass rate for whites, and that Gulf has failed to show that the tests were sufficiently job related.
In
Griggs v. Duke Power Co.,
Both the pre-1971 “Old Tests” and the post-1971 “New Tests” had an adverse impact on blacks. On remand the district court was to determine whether “professionally accepted methods” showed that the tests were “significantly correlated” to job performance in each of the relevant crafts.
Bernard IV,
The issue before us is whether Gulf’s validation methods were properly accepted by the district court. Gulf had a validation study done which compared New Test scores with performance in five crafts. Three of the studies were inconclusive regarding correlation between test scores and performance, while two studies yielded correlation coefficients (measurements of the frequency with which higher test scores correlated with better job performance) for the pipefitter and boilermaker crafts. 4 The study then computed adjustments for various factors, resulting in seven different adjusted correlation coefficients for each craft. The class contends that the district court should have specifically stated which correlation coefficient it found a statistically significant indication of job-relatedness, and why a particular adjustment was chosen. Plaintiffs urge that a correlation coefficient in the .30-40 range be established as the minimum for proof of a job related test. We decline to establish a bright line cut-off point for the establishment of job-relatedness in testing.
In Watson v. Fort Worth Bank & Trust, the Supreme Court stated:
[Ejmployers are not required, even when defending standardized or objective tests, to introduce formal ‘validation studies’ showing that particular criteria predict actual on-the-job performance.
This court held in its earlier opinion that:
To establish the job relatedness of the tests, the degree of correlation between test scores and job performance ratings must be examined. The district court upheld the validity of the tests without making any findings concerning the sufficiency of correlation. Because a finding of significant correlation between test results and job performance is a prerequisite to a holding that the tests are job related, we assume that the court sub silento made this finding.
1. The study showed that performance on the New Tests correlated .32 with job performance as a Boilermaker, and .22 with performance as a Pipefitter, and that these correlations are both statistically significant.
2. Although the unadjusted correlations are statistically significant, the adjusted figures, which are even higher than unadjusted, are better estimates of validity, and even they underestimate the true validity of the New Tests.
3. The testimony of Gulfs expert witnesses was convincing, and they adequately explained, based upon research and their past experience, why the tests were job related, why the correlation coefficients resulting from the study needed to be adjusted for various factors, and why these adjusted figures were more accurate.
The district court was not clearly erroneous in its findings, which are supported by the expert testimony in the record. “The question of job relatedness must be viewed in the context of the plant’s operation and the history of the testing program.”
Albe-marle,
In its prior opinion, this court invited the district court to consider whether the similarity between pairs of jobs could be used to generalize the validity of the new tests from boilermakers and pipefitters to craft jobs involving similar abilities.
Bernard IV,
Having found that Gulf had shown the new tests were significantly correlated with performance as a boilermaker and pi-pefitter, the district court considered and compared the most important abilities in those two crafts with the important abilities in all the other crafts. If the most important abilities required for the other crafts are closely related to the most important abilities of the boilermaker or pipe-fitter, it may be concluded that the new tests, which predict significant aspects of job performance for the boilermaker or pi-pefitter, also predict important abilities related to the performance of all other craft jobs as well. The district court found that the most important abilities for boilermakers or pipefitters were closely related to the important abilities for all of the other *744 crafts, and extrapolated from that to find that the tests were job related for all of the crafts.
The district court accepted similar methods in
Cormier v. P.P. G. Industries, Inc.,
where “cluster analysis” (looking for similar job traits, as was done here) was approved as a way to find correlations between test scores and performance.
Finally, the class seeks to refute Gulfs showing that the tests were job related by showing an alternate method was available that would have had less adverse impact yet met the same business purposes as the Old and New Tests.
See Albemarle Paper Co. v. Moody,
The district court rejected this alternative, making no findings as to its adverse impact, and finding it unlikely to suit Gulfs business purposes as well as the tests actually used. The simple test was used once, as part of a reorganization. The only people taking the test already had ten years craft experience. Its purpose was to screen out total incompetence, not determine who would do best at a job. In order to require an alternative with less disparate impact, the disparate-impact plaintiff must prove that the proposed alternative is “equally effective” as the employer’s procedure.
Ward’s Cove Packing Co., Inc. v. Atonio,
- U.S.-,
The Supreme Court has cautioned that the “judiciary should proceed with care before mandating that an employer must adopt a plaintiff’s alternate selection or hiring practice in response to a Title VII suit.”
Id.
at 2127. In the face of statistically significant correlations, professional test construction, commonality of abilities associated with the various jobs at issue, the “path analysis” correlations of the Old and New Tests, and the supporting expert opinion based upon massive validation studies of similar tests in the industry, this court will refrain from making business decisions for an employer when it is not clearly erroneous that its promotion practice is validly job related. “[I]t must be borne in mind that ‘[cjourts are generally less competent than employers to restructure business practices, and unless mandated to do so by Congress they should not attempt it.”
Watson,
487 U.S. at-,
IV. The Individual Claims
Three individual plaintiffs claim that Gulf discriminated against them in making promotions. Mr. Hayes, Mr. Brown and Mr. Tizeno all claim that they were qualified to become supervisors, but were not promoted, while whites were.
A. Hayes and Brown
The district court made specific findings of fact that Mr. Hayes and Mr. Brown, while complaining that they had *745 not been promoted to permanent supervisory jobs, had testified that they had not made known to anyone in authority their interest in becoming supervisors. The district court further found that employees generally knew when they were eligible and that they were invited to inform supervisors of their interest in promotion. The district court also found several reasons why some employees would not want to be supervisors, and why it could not be assumed that all employees should automatically have been considered.
Under
Texas Dep’t. of Community Affairs v. Burdine,
in a failure-to-promote case the plaintiff must prove that he
“applied
for an available position for which [he] was qualified, but was rejected under circumstances which give rise to an inference of unlawful discrimination.”
The district court found that Hayes and Brown had not established a prima facie case because their failure to express interest made it impossible to tell which promotion they did not get and thus whether a white or a black was selected in preference to them. Blacks as well as whites were made supervisors during the relevant period in numbers proportional to those in the eligible pool.
Even assuming arguendo that Hayes & Brown had established a prima facie case, the district court found that Gulf had a sufficient business reason to rebut the inference of discrimination.
If plaintiff makes out a prima facie case, the employer must produce “evidence that the plaintiff was rejected, or someone else was preferred, for legitimate, non-discriminatory reasons_” [Burdine 450 U.S.] at 253-254 [101 S.Ct. at 1094 ], If the employer .meets this burden, “the presumption of discrimination ‘drops from the case’ and the District Court is in a position to decide whether the particular employment decision at issue was made on the basis of race.”
McDonnell-Douglas,
The district court held that the reason Hayes and Brown were not promoted was their failure to inform the company that they wished to be promoted. Plaintiffs argue that this is a clearly erroneous finding because it is supported by insufficient evidence. Plaintiffs maintain that Gulf should have shown that only those who requested promotion were promoted. Plaintiffs mistake the allocation of burdens here. Once Gulf produced evidence to show what its business reasons were for not promoting Mr. Hayes or Mr. Brown, it was up to the plaintiffs to produce evidence to show that the reason given was a sham.
Burdine,
1. Mr. Hayes and Mr. Brown each testified that they had told no one in authority of their interest in becoming supervisors;
*746 2. Mr. Draper testified that employees were invited to inform supervisors of their interest in promotion;
3. There was no evidence that all employees wanted to become supervisors; and
4. Mr. Brown’s testimony indicated he knew how to express interest, but he chose to remain silent.
The district court’s finding that Mr. Hayes and Mr. Brown were not promoted for nondiscriminatory reasons is not clearly erroneous.
B. Tizeno
Mr. Tizeno did apply twice for a supervisory position (once for a position as “planner” and once for a position as “supervisor”). On remand the district court found that Mr. Tizeno had established a prima facie case, “although a weak one.” The district court found that the record as a whole did not support an inference that racial discrimination was likelier than not the reason that Tizeno was not promoted. Specifically, the district court found that:
(1) Although Mr. Chesser, a white who was promoted to planner, had less seniority than Mr. Tizeno, Mr. Chesser also had less seniority than most other pipefitters, including 108 who were white, 80 of whom had greater seniority than Mr. Tiz-eno as well.
(2) There was no evidence that Mr. Tiz-eno was otherwise more qualified than Mr. Chesser.
(3) Several persons, including 4 blacks were appointed to supervisory positions in 1979, making it hard to tell which position Mr. Tizeno did not get. At that time Mr. Tizeno was one of the most junior pipefitters.
The district court was not clearly erroneous in its determination that the reason Mr. Tizeno was not promoted was not discriminatory.
V. Notification of Exclusion from the Redefined Class
Rule 23 of the Federal Rules of Civil Procedure requires a district court to give notice to absent class members of developments in the suit in only two situations:
The first is when the court certifies a class under Rule 23(b)(3) because of common questions of law or fact that predominate over other aspects of the suit and render a class action the appropriate vehicle to resolve claims, (citation omitted). The second is when a class action is to be dismissed or compromised. Fed. R.Civ.P. 23(e).
Payne v. Travenol Laboratories, Inc.,
Since this case involves the district court’s modification of the class definition to exclude certain class members, it does not fall within either of the foregoing categories. Therefore, notification to those excluded members was not mandated, but instead was left within the discretion of the district court. Id. at 812. Although discretionary notification is encouraged in those instances where a plaintiff has shown that the excluded members: (1) received notice of their initial inclusion in the class; (2) relied on the class suit to protect their interests; and (3) would be prejudiced as a practical matter by their exclusion, Id. at 813, no such showing has been made by the class herein. Therefore, we find that the district court did not abuse its discretion in failing to give the excluded class members notice of the redefinition of the class.
VI. Conclusion
After reviewing the massive amounts of evidence in the record we find that the district court was not clearly erroneous in ruling that Stipulation 29 was not purposefully discriminatory, and that even if it had an adverse impact it was justified by legitimate business purposes. Nor was the district court erroneous in finding that the tests Gulf used to determine which employees were eligible for promotion were job related, based upon the validation studies and expert testimony. The record supports the district court’s determination that Mr. Brown and Mr. Hayes did not show a pri-ma facie case of discrimination in Gulf’s failure to promote them, and that even if *747 they made out a 'prima facie case, Gulf had shown legitimate reasons for not promoting them. Likewise, the district court was not clearly erroneous in holding that Mr. Tizeno had not proved that Gulfs failure to promote him was discriminatory. Accordingly, we AFFIRM.
Notes
. BEFORE REORGANIZATION & STIPULATION 29 AFTER REORGANIZATION & STIPULATION 29
0 & M DIVISION LABOR DIVISION MAINTENANCE DIVISION
Instrument Depart. Labor Subdepart. Instrument
Instrument Man No. 1 Instrument Man
Mechanical Trainee Mechanical Trainee
Instrument Man No. 2 (Craft Helper) Utility Man
Instrument Man No. 3 (Craft Helper) Laborer
Mechanical Helper Pool
Utility Man No. 1
(Intermediate Job Classifications)
Laborer
. The Old Tests consisted of six separate tests: (A) Test of Reading Comprehension; (B) Test of Arithmetic Fundamentals; (C) Wonderlic Personnel Test; (D) Mechanical Aptitudes Test; (E) Mechanical Insight Test; (F) A Lee-Clark Arithmetic Test.
The New Tests consisted of four parts: (A) Den-nett Mechanical Comprehension Test; (B) Test of Chemical Comprehension; (C) Arithmetic Test; (D) Test Learning Ability. After 1971, these tests were also used for the hiring of new employees. Thus, any employee hired after 1971 was not required to take any additional tests to enter a craft.
. 82.5% of whites who took the old tests between January 1969 and March 1971 ultimately passed. Only 42.8% of blacks who took the same tests during that period passed. Between 1971 and 1980, 97.7% of the whites who took the new tests passed, while only 66% of the blacks passed.
. Test scores were compared to job performance for a number of job classifications. The study developed by Richardson, Bellows and Henry (''RBH") attempted to establish the validity of the new tests for five of the OCAW crafts: boilermaking, pipefitting, welding carpenter, and instrument. However, RBH was unable to validate the tests for the latter three crafts. The job performance ratings for the welders were too similar to establish a significant correlation between test and job performance. In the case of carpenters and instrument mechanics, the ratings given by the two raters used to evaluate job performance (each employee was evaluated by two supervisors and their ratings were compared to derive a single job performance rating) differed to such a degree that RBH concluded that neither rating was a reliable measure of the employees' job performance. Therefore, the 1985 study introduced into evidence contained data about only two of the craft departments, boilermaking and pipefitting.