The opinion of the court was delivered by: LASKER
This suit is another in an ever-extending series of challenges to civil service examinations. Plaintiffs, who are Correction Officers,
provisionally appointed to the rank of Correction Sergeant (Male), contend that the test for promotion and permanent appointment to that position discriminated against them on the basis of race. They seek to represent all Black and Hispanic Correction Officers or provisional Correction Sergeants who failed the examination, who passed it but ranked too low to be appointed or who were deterred by the appointment system from seeking promotion. Defendants are the New York State Department of Correctional Services, its Commissioner, and the New York State Civil Service Commission and its Commissioners.
The action is brought under the Fifth and Fourteenth Amendments to the Constitution and under the Civil Rights Act (42 U.S.C. §§ 1981 and 1983) and its jurisdictional counterpart (28 U.S.C. §§ 1343(3) and (4)). Plaintiffs make no claim under Title VII of the Civil Rights Act of 1964 (42 U.S.C. §§ 2000e-2000e-17), despite the availability, by recent amendment, of remedies under it against states and municipalities (id. at § 2000e(a)).
In spring, 1972, the 1970 eligible list for Sergeant appointments was exhausted. To fill needed positions pending establishment of a new list, the Department of Corrections appointed provisional Correction Sergeants, in August, 1972, to hold their posts until permanent appointments could be made. Both named plaintiffs were appointed at that time.
Upon request of the Department of Corrections, the Civil Service Commission prepared a promotional examination which was administered on October 14, 1972. That examination, 34-944, was taken and failed by plaintiffs and is the subject of this action.
34-944 was taken by 1,383 persons,
including 1,264 whites, 103 Blacks and 16 Hispanics. The candidates examinations were graded and the passing grade was established at 70%. After adjustment for veteran's preference and seniority, those who passed were ranked by grade and an eligible list was promulgated on March 15, 1973. On April 10, 1973, this suit was filed and a temporary restraining order entered preventing defendants from making appointments from the list and from terminating the provisional appointments of plaintiffs or members of the class. By modification and stipulation, the restraining order was extended to maintain the status quo until a decision on the merits.
The ground rules for cases such as this have been thoroughly elucidated by recent decisions of the Court of Appeals for this Circuit. We note in particular Vulcan Society of the New York City Fire Department, Inc. v. Civil Service Commission (" Vulcan "), 490 F.2d 387 (2 Cir. 1973), aff'g, 360 F. Supp. 1265 (S.D.N.Y. 1973); Bridgeport Guardians, Inc. v. Bridgeport Civil Service Commission (" Guardians "), 482 F.2d 1333 (2d Cir.), aff'g in part and rev'g in part, 354 F. Supp. 778 (D. Conn. 1973); and Chance v. Board of Examiners (" Chance "), 458 F.2d 1167 (2d Cir. 1972), aff'g, 330 F. Supp. 203 (S.D.N.Y. 1971). To summarize the approach adopted by the cases, plaintiffs must first establish a prima facie case showing that the examination has had "a racially disproportionate impact." Vulcan, at 391; Castro v. Beecher (" Castro "), 459 F.2d 725, 732 (1st Cir. 1972). If they succeed, it then becomes defendants' burden to justify the examination's use despite its differential impact by proving that it is job-related (Vulcan, at 391) and that any disparity of performance results solely from variance in qualification and not from race (Griggs v. Duke Power Co., 401 U.S. 424, 430-31, 28 L. Ed. 2d 158, 91 S. Ct. 849 (1971); Chance, 330 F. Supp. at 214). Discharging this burden would entitle defendants to judgment; failure would, of course, require the court to take the third step of determining what remedy would be appropriate.
As is typical in cases of this type, plaintiffs do not allege that defendants have intentionally discriminated against their class. Such an allegation is not a necessary part of their case. Chance, 458 F.2d at 1175-76. As the Supreme Court stated in Griggs.4
"[Good] intent or absence of discriminatory intent does not redeem employment procedures or testing mechanisms that operate as 'built-in headwinds' for minority groups and are unrelated to measuring job capability." 401 U.S. at 432.
However, the fact that the alleged discrimination is not claimed to be deliberate modifies the burden placed on the state to justify its actions. Intentional racial discrimination would require the state to demonstrate a compelling necessity for its selection methods. Cf. Loving v. Virginia, 388 U.S. 1, 18 L. Ed. 2d 1010, 87 S. Ct. 1817 (1967); Yick Wo v. Hopkins, 118 U.S. 356, 30 L. Ed. 220, 6 S. Ct. 1064 (1886). However, "the Supreme Court has yet to apply that stringent test to a case such as this, in which the allegedly unconstitutional action unintentionally resulted in discriminatory effects." Chance, 458 F.2d at 1177. Agonizing over whether the state can discharge its constitutional obligations merely by suggesting a rational basis for the examination's use or whether it must satisfy a more demanding standard, short of the compelling interest test, is unnecessary. The guidelines have been so refined by the cases that no ambiguity obscures the road to determination regardless of the difficulties of classification which may remain to plague the theorists. Guardians, 482 F.2d at 1337. The decisions impose on the state "a heavy burden of justifying its contested examinations by at least demonstrating that they were job-related." Chance, 458 F.2d at 1176; see also Guardians, 482 F.2d at 1337. This "heavy burden" is discharged if the state "[comes] forward with convincing facts establishing a fit between the qualification and the job." Vulcan, at 393, quoting Castro, 459 F.2d at 732. Once the state proves its case to that extent, it need not establish, as would be required under the compelling interest approach, that no alternative means of selection are open to it. Castro, 459 F.2d at 733; see also Vulcan, at 393.
However clearly the issues are delineated by well-established precedent, nothing can make easy the task of deciding a case such as this. The competing interests are vital to the named parties, to other individuals who may be affected by the outcome and to the public at large. Plaintiffs strive to insure for themselves and the minorities they seek to represent the fair treatment in the public employment sphere which the Constitution guarantees. Their efforts bring them into conflict with those individuals who passed the challenged examination and have a vested interest in securing the promotions which are rightfully theirs if the examination is upheld. For both groups, the outcome is critical since it affects their ability to earn a living by advancing in the profession of their choice. Last and perhaps most important is the public's stake in establishing and maintaining a system of prison administration which is both competent and representative of the population. As members of the public, we include, of course, the inmates of the prison system who, more than anyone else in the community, are directly affected by the quality of correctional supervision. The delicacy of the decision is further compounded by the potential for heightened tension which attends any direct conflict along racial and cultural lines.
Bearing these factors in mind, we proceed, with caution but without more ado, to a consideration of plaintiffs' prima facie case.
I. Disproportionate Impact
Plaintiffs rest their case on the following uncontested statistics. The figures computed by defendants indicate that White candidates passed 34-944 at a rate of 30.9%, while only 7.7% of Black candidates and 12.5% of Hispanic candidates achieved a passing score. (Transcript of 500). That is, Whites passed at a rate approximately four times that of Blacks and 2.5 times that of Hispanics. Defendants concede the statistical significance of these differences. (Post-trial Memorandum at I-4.)
Plaintiffs' evidence reveals an even more startling disparity among those who ranked high enough to be appointed. The Department of Corrections intends to appoint a maximum of 147 persons from the present eligible list.
A computer display of the results of 34-944 (PX-12) reveals that, of 159 persons who scored 57 or above (a group large enough to satisfy the Department's projected needs), 157 were White, two were Black and none was Hispanic. Thus, 12.5% of the Whites who took 34-944 are likely to be appointed, while only 1.9% of Black candidates and no Hispanics have a chance at appointment. These results would lead to the appointment of Whites at 6.5 times the rate of Blacks and would bar completely the appointment of Hispanics.
The statistical significance of these figures is established beyond dispute by the earlier cases. In Chance, Guardians and Vulcan, the impact was less drastically disproportionate among the races. In Chance, the passing rate for Whites was 1.5 times that of Blacks and Hispanics (330 F. Supp. at 210); in Guardians, Whites passed at 3.5 times the rate for Blacks and Hispanics (354 F. at 784); and in Vulcan, Whites scored high enough to have a chance at appointment at 2.8 times the rate for Blacks and Hispanics (360 F. Supp. at 1269).
Defendants do not challenge the accuracy of plaintiffs' figures (for which they are the source) nor do they deny the statistical significance of the differential impact indicated by them. They contend, however, that the approach taken by plaintiffs, that is, consideration of the statistics as to the statewide impact of the entire exam, does not accurately reflect the performance of the groups in relation to each other. They urge us, rather, to base our determination of racial impact on the candidates' performances facility by facility rather than throughout the state. They contend that otherwise it is impossible to determine whether minority candidates are succeeding less well as a group because of their racial and cultural backgrounds or because they are located at facilities which, for reasons unspecified, prepare their officers less well for the promotional exam. In fact, the great majority of minority candidates are located at Ossining (82 Blacks out of a total of 104, 9 Hispanics out of a total of 16) with the second largest concentration of Blacks at Greenhaven (8). (PX-12, codes 1007 and 1008.) Defendants argue that if both Whites and minority candidates at Ossining perform less well than persons -- White, Black or Hispanic -- employed at other facilities, then 34-944 has not been shown to differentiate on the basis of race. Second, defendants contend that, since 34-944 is composed of five subtests, comparative performance on each subtest should be determinative rather than performance on the test as a whole. If these approaches are adopted, they claim, the three groups of candidates will be shown not to have performed sufficiently differently to make out a prima facie case of disproportionate impact.
To support their argument that the results of 34-944 are relevant only if separated by facility, defendants rely on an analysis of the computer display of examination results (PX-12) drawn up by Kenneth Siegel, the Associate Personnel Examiner who was responsible for the preparation of 34-944. He analyzed the performances of the groups in terms of mean scores on the total exam and on each of the five subtests at Ossining, Green Haven, all the other facilities and all the facilities taken together (DX-DD). The reason for selecting Ossining and Green Haven for special attention was the concentration of minority candidates at those facilities. Siegel's written analysis (DX-DD) does not indicate passing rates, but only mean scores. However, Siegel testified that the difference in passing rates between Whites and Blacks at Green Haven (Transcript at 511) and all other facilities except Ossining is not statistically significant (Transcript at 509, 515). Based on Siegel's testimony, defendants argue that as a result plaintiffs' prima facie case fails with respect to all facilities except Ossining.
The principal obstacle to accepting defendants' analysis is that it is premised on assumptions which are factually erroneous. Their own statistics bely their theory. Siegel's analysis (DX-DD) of the computer display (PX-12) reveals not only that the mean score for Whites state-wide (48.9) is superior to that of Blacks (43.2) and Hispanics (44.2), but also that the mean scores at Ossining, Green Haven and other facilities considered separately reflect the same pattern. Whites at Ossining achieved a mean score of 47.32, compared with 42.96 for Blacks and 41.56 for Hispanics. The disparity at Ossining is virtually identical to that derived from a comparison of statewide figures for Whites and Blacks (48.9 to 43.2) and is greater than the state-wide difference between Whites and Hispanics (48.9 to 44.2). This effectively refutes defendants' theory that minority candidates generally performed less well than Whites solely because they were concentrated at Ossining where candidates as a whole did less well. The range at Green Haven is almost as striking and indicates again a greater variance than is found statewide between Whites and Blacks and an almost identical disparity as that found state-wide between Whites and Hispanics: Whites, 48.68; Blacks 42.00; Hispanics, 44.00. A comparison of results at facilities other than Ossining and Green Haven bears out the trend: Whites, 49.00; Blacks, 45.21; Hispanics, 48.17. It is true that Hispanics at these facilities fared better than at Ossining and Green Haven and their scores more closely approximate the performance of Whites. However, the importance of this discovery is somewhat discounted by the small size of the sample (6 Hispanic candidates) which decreases the possibility of statistical accuracy (Transcript at 936-37). Furthermore, Siegel's analysis indicates that the standard deviation in mean scores between Whites and Blacks was statistically significant at Ossining, Green Haven and all other facilities as well as state-wide, and the same is true of Whites and Hispanics at Ossining where the largest concentration of Hispanics is found. (DX-DD).
An analysis of passing rates, which is more appropriate since it is the passing score which determines a candidate's eligibility for appointment, is even more illuminating. Siegel testified that there was a significant difference between the passing rates of Whites and Blacks at Ossining (Transcript at 509), but that no such difference existed between Whites and Blacks at Green Haven and facilities other than Ossining and Green Haven and none between Whites and Hispanics at Ossining, or other facilities. (Transcript at 509-515.) He did not compare the passing rates of Whites and Hispanics at Green Haven because there was only one Hispanic candidate at that facility. (Transcript at 511.) Nor did he testify as to the difference between the passing rates of Whites and Hispanics at facilities other than Ossining and Green Haven. Siegel is correct that the disparity in passing rates between Whites and Blacks at Ossining is significant: Whites passed at a rate of 23.5% and Blacks at a rate of 4.9%. (PX-33.) However, his testimony as to Blacks at Green Haven and at other facilities and as to Hispanics at Ossining flies in the face of the figures in evidence. To the contrary, comparison of the groupings mentioned above indicates in each instance a significant disparity between the passing rate of White and minority candidates. Whites at Green Haven passed at a rate of 31.6%, while Blacks and Hispanics achieved rates of only 12.3% and 0%
respectively. 30.7% of Whites at facilities other than Ossining and Green Haven
passed 34-944, while only 14.3% of Blacks passed. Although Hispanics at facilities other than Ossining and Green Haven passed at a higher rate than Whites (33.3% compared to 30.7%), the reliability of this computation is put in doubt by the smallness of the sample. Hispanics at Ossining, on the other hand, passed at a rate of 0% compared to a White passing rate of 23.5%. Accordingly, contrary to Siegel's conclusion, the disparity between White and minority candidates was significant with regard to Blacks at Ossining, Green Haven and all other facilities, as well as statewide, and was significant with regard to Hispanics at Ossining, where the largest number of Hispanics are located.
These computations destroy the factual premise of defendants' argument that minority performance reflects the facilities in which they concentrated rather than their minority characteristics. We would in any event be forced to reject defendants' theory as a matter of law, even if it could be factually substantiated. Attempts to correlate racial performance to such non-racial characteristics as quality of schooling or educational and cultural deprivation have been rejected as irrelevant to rebut a statistical prima facie case. As the district court opinion in Guardians stated:
"More fundamentally, this data fails to remove the prima facie showing of discrimination because it does not alter but only tries to explain the difference in passing rates." 354 F. Supp. at 785; see also Vulcan, 360 F. Supp. at 1272. Cf. Castro, supra.
The controlling decisions clearly posit that, in order to shift to defendants the burden of showing that performance on the examination correlates to performance on the job, plaintiffs are required to do no more than demonstrate that minority candidates as a whole fared significantly less well than White candidates, regardless of possible explanations for their poorer performance. To quote Guardians once more:
"The point is that a discriminatory test result cannot be rebutted by showing that other factors led to the racial or ethnic classification. The classification itself is sufficient to require some adequate justification for the test." Id. at 786.
Finally, we fail to understand the relevance of defendants' attack on plaintiffs' prima facie case. Defendants appear to concede that, at the very least, Blacks at Ossining who failed 34-944 have established their right to challenge its job relatedness. (Post-trial Memorandum at I-11.) This group constitutes two-thirds of the proposed plaintiff class (77 out of 117 Blacks and Hispanics combined), but if even a far smaller number had succeeded in proving disproportionate impact determined to themselves, defendants would be obliged, as they themselves concede, to prove job relatedness.
We turn to defendants' second challenge to plaintiffs' case. Siegel's analysis of the computer display indicates that although there is a statistically significant difference in the total mean scores of Whites and Blacks and Whites and Hispanics statewide and at Ossining, and, as to Blacks, at Green Haven and facilities other than Ossining and Green Haven, not every subtest indicates such a disparity. (DX-DD.) It is unnecessary to detail the permutations sub-test by sub-test and facility by facility, since the suggested approach itself is invalid as a matter of law. The cases indicate that a showing that the over-all examination procedure produced disparate results cannot be rebutted by fragmenting the process and demonstrating that separately the parts did not differentiate along racial or cultural lines. In Chance, for example, the fact that minority candidates had a higher passing rate than White candidates on seven out of fifty examinations did not vitiate plaintiffs' proof that the series of examinations as a whole discriminated against them and their class. 330 F. Supp. at 211; see also Guardians, 354 F. Supp. at 786. In Vulcan, the very question whether a single examination procedure can properly be subdivided and the parts considered separately, was raised and Judge Weinfeld rejected the proposition:
"Moreover, the examination may not be truncated; whether or not it has an adverse discriminatory impact on minority groups should be considered in terms of the total examination procedure. Here there can be no doubt, whatever the relative impact of component parts, that in end result there was a significant and substantial discriminatory impact upon minorities. . . ." 360 F. Supp. at 1272.
Any other approach conflicts with the dictates of common sense. Achieving at least a passing score on the examination in its entirety determines eligibility for appointment, regardless of performance on individual sub-tests. Accordingly, plaintiffs' case stands or falls on comparative passing rates alone. Thus, in law and in logic, we find defendants' approach unwarranted.
Rejection of defendants' dual attack on plaintiffs' showing of differential impact leaves no doubt that plaintiffs' prima facie case has been amply established. Accordingly, the burden of proof swings to defendants to demonstrate that 34-944 ...