Opinion
Civ. No. B-75-268.
July 14, 1978.
David N. Rosen, New Haven, Conn., Michael P. Koskoff, Bridgeport, Conn., for plaintiffs.
Raymond B. Rubens, Bridgeport, Conn., for defendants.
J. Daniel Sagarin, Bridgeport, Conn., for intervenors.
RULING ON LIABILITY
Plaintiffs, as representatives of the class of Blacks and Hispanics who reside in the City of Bridgeport, Connecticut, are attacking the 1975 Civil Service Exam administered to candidates for Bridgeport's Fire Department. Defendants are the City of Bridgeport, several of its officials, the Bridgeport Civil Service Commission and the Bridgeport Board of Fire Commissioners. A group of Bridgeport's firefighters, Bridgeport Firefighters for Merit Employment, Inc., have been permitted to intervene to assure that the viewpoints of all interested persons are represented.
This opinion is limited to the question of defendants' liability. After extensive discovery and several pretrial conferences, this Court decided to bifurcate the complex factual claims of liability from the question of the appropriate remedy. The liability aspect of the case has been tried to the Court and carefully briefed in thorough post-trial memoranda. On the basis of the credible evidence, the Court finds that Bridgeport's 1975 firefighters exam had a disparate impact and that this exam was not job-related.
Plaintiffs claim that Title VII of the Civil Rights of 1964, 42 U.S.C. § 2000e, et seq., and the anti-discrimination provision of the Revenue Sharing Act, 31 U.S.C. § 1242(a), have been violated. The violation of Title VII is based on the alternative contentions that plaintiffs have been subjected to disparate treatment and that defendants' firefighters exam had a disparate impact. These two theories of discrimination are well explained by the Supreme Court in Int'l Brotherhood of Teamsters v. United States, 431 U.S. 324, 335-36 n. 15, 97 S.Ct. 1843, 1854, 52 L.Ed.2d 396 (1977):
"`Disparate treatment' such as alleged in the present case is the most easily understood type of discrimination. The employer simply treats some people less favorably than others because of their race, color, religion, sex, or national origin. Proof of discriminatory motive is critical, although it can in some situations be inferred from the mere fact of differences in treatment . . .
Claims of disparate treatment may be distinguished from claims that stress `disparate impact.' The latter involve employment practices that are facially neutral in their treatment of different groups but that in fact fall more harshly on one group than another and cannot be justified by business necessity. . . . Proof of discriminatory motive, we have held, is not required under a disparate impact theory" (citations omitted). Id.See Furnco Construction Corp. v. Waters, ___ U.S. ___, ___ — ___, 98 S.Ct, 2943, 57 L.Ed.2d 957 (1978). Because plaintiffs have demonstrated that the firefighters exam violated the disparate impact theory of Title VII, the Court does not reach the question of whether plaintiffs have been subjected to disparate treatment. Moreover, the Court is reserving judgment on plaintiffs' claim that Bridgeport's federal revenue sharing funds should be terminated. As suggested by intervenors, this claim is best considered subsequently when the Court fashions a remedy for plaintiffs. It should be noted, however, that the Court sees little benefit to plaintiffs in the termination of federal funding and views such a step as one of last resort. See generally United States v. City of Chicago, 549 F.2d 415, 449 (7th Cir. 1977).
The legal principles that apply to a claim of disparate impact are well-settled. In Dothard v. Rawlinson, 433 U.S. 321, 329, 97 S.Ct. 2720, 53 L.Ed.2d 786 (1977) the Supreme Court had occasion to summarize the three-step analysis that governs proof of such Title VII claims. The Court explained that to prevail on the basis of a disparate impact theory plaintiffs must first demonstrate that defendants' use of facially neutral hiring requirements has a significantly disparate impact on plaintiffs. That is, plaintiffs need only prove that defendants' seemingly unbiased hiring standards have resulted in a significant pattern of discriminatory employment. If plaintiffs adduce ample proof of such a discriminatory pattern, they have established a prima facie case of a racially disparate impact and the burden shifts to defendants. Id. See 29 C.F.R. § 1607.3 (1977). In the Second Circuit this shifting of the burden means defendants become saddled not only with the burden of going forward, but also with the burden of persuasion. Vulcan Soc. of N.Y. City Fire Dept., Inc. v. Civil Serv. Comm'n, 490 F.2d 387, 393 (2d Cir. 1973). By way of rebuttal, defendants must then demonstrate that their employment requirements are substantially related to job performance. The question whether Bridgeport's firefighters exam was job-related is the critical issue in this case. The third and final step in the analysis is reached only if defendants successfully show that their employment requirements "have a manifest relationship to the employment in question". Griggs v. Duke Power Co., 401 U.S. 424, 432, 91 S.Ct. 849, 854, 28 L.Ed.2d 158 (1971). If defendants prove their requirements are job-related, plaintiffs have an opportunity to rejoin by showing that other selection techniques are available which would advance defendants' interests in maintaining high caliber employees, but which would not have a similar discriminatory effect. Dothard v. Rawlinson, supra, 433 U.S. at 329, 97 S.Ct. 2720; Albemarle Paper Co. v. Moody, 422 U.S. 405, 425, 95 S.Ct. 2362, 45 L.Ed.2d 280 (1975).
For a list of cases in which written employment examinations are challenged on the basis of a racially disparate impact, see Bridgeport Guardians v. Bridgeport Police Dept., 431 F. Supp. 931, 933 (D.Conn. 1977).
Intervenors argue that the analysis outlined in Dothand v. Rawlinson, supra, is no longer appropriate and that plaintiffs must demonstrate an intent to discriminate in addition to a disparate impact. This contention is meritless. The case upon which intervenors principally rely, Washington v. Davis, 426 U.S. 229, 96 S.Ct. 2040, 48 L.Ed.2d 597 (1976), states that a showing of discriminatory animus is an essential element of discriminatory claims based on equal protection. Accord, Regents of the University of Calif. v. Bakke, ___ U.S. ___, ___ n. 27, 98 S.Ct. 2733, 57 L.Ed.2d 750 (1978); Chance v. Board of Examiners, 561 F.2d 1079, 1087 (2d Cir. 1977). Plaintiffs' claim, however, is based on Title VII and not equal protection. Therefore, Washington v. Davis, supra, and its progeny are inapposite.
The statistical data presented at trial by plaintiffs leave no doubt that the firefighters exam had a disparate impact on the named plaintiffs and the class of persons they represent. Between the taking of the 1970 census and the 1974 census the minority population in Bridgeport increased from 25.3% to 41%. Of the 153,500 residents of Bridgeport in 1974, 27% were Black and 14% were Hispanic. In marked contrast to the racial composition of the City's population, there was but one minority firefighter in 1975 out of 428 members of the Bridgeport Fire Department. Thus, at the time the firefighters exam was given, minorities constituted 41% of Bridgeport's population, but only 0.2% of its Fire Department.
The firefighters exam did little to alter this imbalance. Only three of the eighty-four candidates who successfully passed the exam were memers of minority groups. In terms of the pass rate, minority candidates passed with one-third the frequency of nonminority candidates; a 9% minority pass rate as compared with a 27% nonminority pass rate. In addition, a statistical test performed by plaintiffs' expert revealed that the difference between the mean exam scores of minority candidates and nonminority candidates could have been the result of chance in less than one in 10,000 samples. Plaintiffs' expert also introduced statistical evidence that demonstrated that the disparity between minority and nonminority performance on the exam was not the result of differences in a candidate's formal education. Assuming that every candidate who passed the firefighters exam is ultimately hired, minorities will constitute something less than 1% of Bridgeport's Fire Department. Therefore, if the results of the exam remain unaltered, the operative effect of the exam will be to perpetuate the composition of an already racially imbalanced fire department. The disparate impact of the firefighters exam, especially when viewed in the light of the Bridgeport Fire Department's prior racial imbalance, compels the conclusion that the exam should be upheld only upon a persuasive showing that it was substantially related to the performance of firefighters.
In fact, plaintiffs' expert testified that there was an inverse relationship between a candidate's exam score and his formal education. That is, candidates of the same race and comparable age did more poorly on the firefighters exam if they had received some college education. This seemingly anomalous result is most probably the result of the weighing of scores on the PRF Understanding test. See the discussion of this test in the text at 755 infra.
The most difficult question raised by plaintiffs' claim is whether the 1975 exam is job-related. In search of an answer, the attorneys and the Court have been led by the conflicting testimony of two experts, who spoke in an alien tongue, through a labyrinth of statistical analysis that might well have intimidated Odysseus. While the Court does not concur with plaintiffs in all of their criticisms of the exam, it is apparent that the exam suffers from several serious defects.
Before discussing the specific defects in the exam, a brief digression is necessary to explain how the exam was developed. The firefighters exam was the result of a concurrent criteria-related validity study. Such studies begin with the collection of data on present employees. These employees are ranked by their superiors and then given a battery of tests that may or may not concern the employee's knowledge of his field. After these tests are scored, a statistical analysis is performed to determine whether the highly ranked employees have performed similarly on any of the tests. If the highly ranked employees uniformly have performed either better or worse than employees with low rankings on any of the tests, then these tests are presumed to identify desirable employees.
In the case of Bridgeport's firefighters exam, data on firefighters in several Connecticut cities were gathered. Statistical analysis of these data revealed that there was a correlation in Bridgeport and Hartford between highly ranked firefighters and their test scores on three of the tests in the test battery. In the cases of the FIT Mechanics and the PRF Social Recognition Tests, there was a direct correlation between highly ranked firefighters and high test scores, whereas in the case of the PRF Understanding Test, highly ranked firefighters tended to receive low test scores. These three tests were then administered to the 1975 candidates, and the results were evaluated by weighing positively high scores on the first two tests and weighing negatively high scores on the PRF Understanding Test. Plaintiffs contend that the development of this exam was "shoddy." Although the Court does not agree with this assessment, there were deficiencies in the preparation of the exam and in the selection of a passing score.
The management consulting firm that developed the exam took the additional step of attempting to eliminate those tests in which there was a significant difference between the scores of Blacks and Whites. While this was a commendable measure, the Court is not persuaded that this additional step made the test unbiased. Of necessity the consulting firm's analysis was based on the evaluation of a small number of Blacks, most if not all of whom had already demonstrated an ability to perform well on civil service exams. In addition, the firm was never told to eliminate tests that had a discriminatory impact on Hispanics.
PREPARATION OF EXAM
Chief among the deficiencies in the exam was its preparation. It would seem that a thorough job analysis would be fundamental to the development of a job-related exam. However, the testimony introduced at trial indicated that the firm which had developed the exam, Hay Associates (Hay), relied on a perfunctory analysis of a firefighter's job. Plaintiffs quite properly note that the insufficient data gathered by Hay concerning the job characteristics of a firefighter tainted all subsequent efforts to develop a job-related exam. For example, because a detailed list of critical work behaviors was not developed, the supervisory personnel who ranked the firefighters used only a very generalized list of job qualities. Not only did this list lack specificity, but there was no differentiation between attributes on the basis of their importance. As a result, a supervisor's high ranking of a firefighter for having a good sense of "public relations" had the same effect on a firefighter's score as a high ranking for "composure under pressure." While both attributes are commendable in a firefighter, Hay's methodology provided no means by which their relative importance could be considered.
The Court also finds that the ranking procedure used to distinguish among firefighters was inadequate. Hay had supervisory personnel rank firefighters over whom they had authority in a linear order from best to worst. This technique has several defects when used to evaluate small groups that vary in size. First, it assumes that the best firefighter under one supervisor is better than the second best firefighter under another supervisor. This criticism is particularly important here because all the supervisors were ranking members of small groups; almost one-half of the firefighters were in groups of no more than five and one-quarter were in groups of two or three. The consequence of the ranking procedure is that a firefighter who was the second best out of a group of two was equated with a firefighter who was the second best out of a group of six. The possibility of obtaining valuable data from this ranking procedure is further diminished by the fact that the supervisory personnel who performed the rankings were given insufficient training.
The final deficiency in the exam's preparation concerns the exclusive use of volunteers as a sample group. Plaintiffs' expert testified that complete reliance upon volunteers in a concurrent validity study often results in skewed data. He stated that within his field volunteers are considered a notoriously unrepresentative sample because their motivation tends to differ from the motivation of their peers.
SELECTION OF PASSING EXAM SCORE
Not only is the exam's preparation suspect, but the manner in which it was used is improper. Specifically, plaintiffs and intervenors criticize the passing score as being arbitrary. Although the ultimate effect of the exam turns upon the score used to differentiate between passing and failing scores, Hay did not recommend a passing score. Apparently, this failure was due to the fact that the ranking procedure used in the preparation of the exam provided no information upon which to base such a recommendation. The critical choice of a passing score must be scrutinized with particular care because this decision was in large part responsible for the disparate impact of the exam.
The discriminatory impact of the exam would have been significantly alleviated if the passing score had been designated as 6. If this score had been used, 55.2% of the Blacks and 68.6% of the Whites would have passed. Plaintiffs contend that even with such an alteration of the passing score, the exam would be discriminatory because only 9.2% of the total number of candidates to pass would be members of minority groups. The force of this contention is weakened by the fact that only 10.9% of those who applied to take the firefighters exam were members of minority groups.
The only explanation given for the passing score used was the arbitrary requirement of Bridgeport's City Charter. See Bridgeport Guardians, Inc. v. Bridgeport Civil Serv. Comm'n, 482 F.2d 1333, 1338 (2d Cir. 1973). Unlike the passing score used in the detective exam that was upheld in Bridgeport Guardians v. Bridgeport Police Dept., 431 F. Supp. 931 (D.Conn. 1977) (Guardians II), the passing score for the firefighters exam bore no relation to "normal expectations of proficiency." 29 C.F.R. § 1607.6 (1977). In Guardian II, the Court found that the test had been designed so that a successful candidate would have had to obtain a higher score on the exam than had been obtained by most other candidates who had previously taken the same test. In contrast to the firefighters exam, the detective exam challenged in Guardians II had been planned so that the passing score mandated by Bridgeport's Charter would reflect a determination of desired proficiency. The fact that the passing score was not "based on sufficiently sound approach," Guardians II, 431 F. Supp. at 490, is itself ample reason in this case to conclude that the exam was not job-related.
The uncontroverted testimony at trial revealed that the passing score used by Bridgeport was selected because the City Charter requires candidates to answer correctly at least 75% of all questions on civil service exams. Since such a requirement was meaningless on the test battery developed by Hay, Bridgeport used the seventy-five percentile marks as the passing score. The selection of this passing score automatically meant that 75% of the candidates who took the exam would fail. In its brief, Bridgeport quite properly concedes that there is no statistical relation between the passing score required by its Charter and the passing score it used in the firefighters exam.
Having determined that the exam was not sufficiently job-related, the inquiry could end without considering whether a better examination is available. However, the Court feels that it should address defendants' repeated protestations that it is being required to do the impossible — to devise a perfect exam. While the Court finds that Bridgeport is making advances in its employment practices, it cannot be said that the City has done all that is within its power to obtain highly qualified minority firefighters.
The firefighters exam represented a decided improvement over some of the earlier civil service exams employed by Bridgeport. See, e.g., Bridgeport Guardians, Inc. v. Bridgeport Civil Serv. Comm'n, supra. Furthermore, based on the evidence before the Court, there appears to be no bad faith on the part of defendants in using the exam or in selecting the respected management consulting firm of Hay to develop the exam. Defendants made a substantial effort to obtain an exam that would withstand judicial scrutiny. Over a period of one and one-half years some 2745 man-hours were spent in the development of the exam at a cost of approximately $100,000 in federal funds. It should also be noted that the use of written civil service exams as the critical component of the hiring process is itself an important improvement over the political spoils system that had long been employed in many municipalities.
However, the Court finds that defendants' recruiting efforts were unsatisfactory. Until quite recently Bridgeport's Fire Department had only one black officer and two black firefighters. Not surprisingly, the City's reputation for hiring minority firemen was deplorable. Little was done to overcome this reputation. The credible evidence indicated that Bridgeport used only one of three local radio stations to advertise the firefighters exam. And, although some recruiting of minority candidates was done, these efforts were neither initiated nor sponsored by Bridgeport. Thus, only 10.9% of those who applied to take the firefighters exam were members of minority groups while 41% of the total population were members of minority groups.
It bears mentioning that those who had Worked conscientiously to recruit minority candidates felt that their credibility was all but destroyed by the firefighters exam.
The Court also believes that defendants have not considered all plausible alternatives for distinguishing qualified from unqualified firefighter candidates. The underlying issues in all cases involving employment examinations is what constitutes a "qualified" employee, and how can he or she be identified in an unbiased and culturally neutral manner. The three-step analysis used to evaluate plaintiffs' disparate impact claim is merely an orderly method of assuring that potential employees are afforded an equal opportunity and that employers are permitted to maintain the caliber of employee that they desire. The firefighters exam represents but one unsuccessful attempt at developing a workable method of accomplishing these objectives. The possible solutions to this problem are subject to more hypotheses than either party has suggested.
For example, it does not seem axiomatic to the Court that multiple-choice tests must be the critical component of a process aimed at identifying qualified firefighters. While this type of test has demonstrated some ability to "predict" subsequent performance of individuals on other written exams, it does not follow that multiple-choice exams are adequate to predict all types of human behavior. In fact, the repeated failure of civil service exams to qualify as job-related, see note 1, suggests that these tests are inadequate to the task of predicting complex patterns of behavior (e.g., those behaviors which would distinguish an average firefighter from an exceptional firefighter). The Court also finds it peculiar that exams, such as the firefighters exam, tend to place so little emphasis on physical ability. It would seem that the physical ability necessary for a qualified firefighter is neither more difficult to identify nor more complicated to measure than the intelligence level of a qualified firefighter.