Davis v. Washington, 352 F. Supp. 187 (D.D.C. 1972)

*188 Richard B. Sobol, New Orleans, La., George Cooper, New York City, Richard Seymour, Washington, D. C., for plaintiffs.

John Salyer, Asst. Corp. Counsel, William H. Schweitzer, Asst. U. S. Atty., for defendants.

GESELL, District Judge.

This phase of this class action involves a claim that promotions by the Metropolitan Police Department (MPD) *189 from patrolman to sergeant discriminate against blacks in violation of the Fifth Amendment and 42 U.S.C. § 1981. A promotion roster is maintained by the MPD, ranking qualified candidates on the basis of their combined score determined by (1) a supervisory report weighted 50 percent; (2) a true-and-false examination weighted 20 percent; and (3) a multiple-choice critical-incident test of judgment and supervisory competence weighted 30 percent. Only the third aspect of the promotion procedure is challenged. Testimony was taken from various testing experts and police officials over a period of two days following extensive discovery resulting in agreement as to much of the pertinent statistical data and formal matters relating to procedures. The parties submitted detailed post-trial briefs, with proposed orders, in support of their positions.

Plaintiffs contend that they are entitled to relief because results on (3), the multiple-choice critical-incident test, reflect a disproportionate number of black failures vis-a-vis white failures and because the test has not been validated as job-related. Defendants claim the tests for the disputed years 1969, 1971 and 1972 have been validated as job-related and that they accordingly met the burden, if any was legally placed on them by the statistical analysis of the test scores.

It should be pointed out at the outset that there is absolutely no proof nor is it even contended that defendants in any way knowingly discriminated. Indeed, it is apparent from the entire record that the MPD has sought to encourage black recruitment and advancement and has one of the best records of effort and success of any police department in the nation. The issue, on the contrary, is highly technical. Plaintiffs have placed into sharp focus the nature of proof required to validate test procedures relied on for promotion where a limited number of specialized supervisory positions is available. Unfortunately the Courts have had little experience with validation and its application to alleged employment discrimination. As this litigation well illustrates, the subject involves an inexact, evolving science where very few premises are yet generally accepted and numerous variables emerge in the context of widely differing employment situations. Before turning to consideration of the few authorities which provide some guidance in this specialized field, it will be appropriate to outline the procedures followed in this particular instance.

The multiple-choice judgment supervisory test consisting of 40 questions is based upon critical incidents developed by a bi-racial Board of five senior police officers who are detailed for several months to the project. These officers have available the advice of a trained, experienced psychological tester and of experts from the United States Civil Service Commission. By unanimous vote they select the 40 questions from a group of some 125 proposed questions submitted from top officers throughout the department in response to a request from the Board designed to elicit questions based on actual experience in the field where correct action depended on judgment. The questions used are all based on actual experience with concrete incidents encountered by sergeants in their regular duty. Care is taken to state the questions in police vernacular, to avoid ambiguity and to present clear-cut alternatives. The incidents proposed and used involve supervisory judgments on significant phases of the work which a sergeant on patrol or field duty has previously encountered. Seventy percent of the sergeants engage in this type of duty and judgment questions in specialized areas such as homicide are avoided. A detailed procedure which has been fully implemented exists for candidates to challenge a particular question and seek its elimination in the marking of all test papers. A new test along these lines is developed for each year the test is given.

Since the questions are based on actual experience, hypotheticals are not used. The incidents are deemed significant *190 for job performance by experienced professionals. Testing procedures and selection of incidents actually used are monitored by qualified, professionally trained testers. There is a substantial body of learning that supports the use of actual incidents to measure supervisory judgment and that indicates such questions are one of the most accurate measures of subsequent job performance. The weighting of 30 percent given this phase is based purely on the judgment of police and testing professionals, supported in general terms by performance studies made in other fields.

The following table illustrates the relative rates of success for blacks and whites applying for promotion overall and on the particular written test in contest.

  Percentage Rates of Success: Black versus White
Year-number in
  top group        1972 (200)  1971 (191)  1969 (233)
FINAL PROMOTIONAL   W:  17.8       20.8       22.4
LIST                B:  13.4       10.8       13.7
WRITTEN TEST        W:  18.2       21.6       22.0
PART II             B:  12.8        9.5       13.3

Plaintiffs' attack focuses primarily on the haphazard manner in which the critical incidents have been selected. A sergeant has a very wide range of duties and may be called on to exercise immediate supervisory judgment in a wide variety of differing contexts. The questions used have not been designed to measure supervisory judgment in all of these differing contexts. Some areas, such as community relations and search warrants, are not covered, while great emphasis is given to riot control and day-to-day relations with business institutions. In short, the selection is based on field duty incidents that come readily to mind among experienced officers furnishing proposed questions and there has been no realistic effort to systematically use supervisory judgment questions that bear on each type of judgment that may be needed or to the panoply of the sergeant's total range of duties where judgment is required. Though not noted by plaintiffs, this same criticism can be made of the Part I true-false test, although the two tests, viewed together, cover a far more representative range of sergeant duties. To underline this asserted deficiency, it is suggested that for ethnic or other reasons black candidates may respond in some judgment situations better or worse than whites, and that the questions themselves are not appropriately drawn since they do not present all the variables in the judgment situation and since it is far from clear that only one of the multiple choices is correct. This series of considerations is said to prejudice blacks, particularly, because blacks are deficient in reading comprehension and generally do less well on written as opposed to oral examinations. While some of these biases are perhaps significant in varying hypothetical degrees, none has been shown to have substantial support when the impact of the test is considered as a whole. But plaintiffs contend the defendants cannot show that these biases are insignificant, as indeed is the case, and that since statistically blacks fare somewhat worse, adverse impact of the testing on blacks has been shown and that the defendants have the burden of validation.

Plaintiffs further insist that the multiple-choice judgment test has not been validated by any acceptable method. First, they argue, there is no content validation, because an element of that process, namely a systematic identification of all types of judgment, supervisory decisions required of rank-and-file sergeants, and emphasis upon those judgments which have been determined to be most important to the job, is missing. Second, it is urged that a showing of job-relatedness can be made adequately only by a criterion validation process that measures scientifically the major judgment elements in the job, and quantitatively correlates the test to predict successful performance in those elements. Defendants admit that no criterion study has been attempted because it is expensive and time-consuming without *191 any substantial guarantee that these job characteristics are susceptible to such a validation method. They contend that the critical-incident studies form of content validation is sufficient to support the use of the judgment test. Plaintiffs' position may well reflect an ideal and a goal, but it is not supported by the state of the law or by the state of the art as reflected in the record compiled by the various experts in this case.

While, strictly speaking, the claims raised here by plaintiffs against these governmental employers were not governed by Title VII of the 1964 Civil Rights Act until it was recently amended,^[*] the Court, as the parties basically agree, looks to the Act and decisions construing it for guidance as to the constitutional constraints on a public employer. See Castro v. Beecher, 459 F.2d 725 (1st Cir. 1972). The Court accepts the fact that confronted even with the poorer statistical showing of blacks present in this case, supra at 3, there is a prima facie case of discrimination which shifts to the defendants the heavy burden of justifying the contested examination by establishing that it is job-related. Griggs v. Duke Power Co., 401 U.S. 424, 91 S. Ct. 849, 28 L. Ed. 2d 158 (1971). In determining whether the challenged test has been shown to be by appropriate methods and professional standards "a reasonable measure of job performance" under Griggs, the Court looks to both the record in this case and the Equal Employment Opportunity Commission (EEOC) Guidelines on Employee Selection Procedures, 29 C.F.R. § 1607.

The guidelines recognize the fact that basically there are three methods of validating the job-relatedness of a given test: criterion, content and construct (only the first two are involved here). In accepting content validity in lieu of criterion validity, where a job analysis demonstrates the relevance of the content and the test has suitable samples of the components of the job in question, the EEOC is in accord with the reality of the state of the art as illustrated by this record and confirmed by the American Psychological Association. Either of these recognized processes, if properly used, can establish that an employer's selection device is job-related. See Allen v. City of Mobile, 466 F.2d 122 (5th Cir. 1972), aff'g 331 F. Supp. 1134 (S.D.Ala.1971); Castro v. Beecher, 459 F.2d 725 (1st Cir. 1972), aff'g in pertinent part 334 F. Supp. 930 (D.Mass.1971); Chance v. Board of Examiners, 458 F.2d 1167 (2d Cir. 1972), aff'g 330 F. Supp. 203 (S.D.N.Y.1971). Content validation is preferred where criterion validation would have to depend upon important intangible job aspects that are not suitable to competent measurement, a severe restriction in the test score range or the sample of cases is too small. The guidelines do not suggest and the law does not require blind adherence to a set of absolutes, but rather ideals must be tempered by the reality of factual conditions and the technical feasibility allowed by the state of the art in order to have a workable and meaningful set of standards.

Defendants have demonstrated to the Court's satisfaction that the test has been content validated by current professional standards consistent with the EEOC guidelines. The judgmental questions portrayed by the critical incidents chosen by the most informed and expert members of the Police Department and reviewed by another panel of police and testing experts without doubt comprise a suitable sample of behaviors and skills extremely relevant to the job of police sergeant. See Dr. O'Leary's Study, Plaintiffs' Exhibit 9. Plaintiffs' assertions to the contrary fly in the face of common sense, the record in this case, and the demands of the law.

The testing process and its weighting in the ultimate promotional decision have been carefully and thoroughly developed *192 by these defendants and demonstrate that this use of the test is highly relevant to the sergeant's job. Tests that have been far less carefully prepared in situations where the racial impact was more blatantly obvious have been upheld. Allen v. City of Mobile, supra. Moreover, it is readily acknowledged that there is absolutely no assurance that more technically exact procedures would result in any benefit to black candidates.

This examination's emphasis on reading and analytical skills is consistent with the tasks sergeants perform, and defendants have proceeded with keen awareness of the need to foreclose bias, taking sound, experienced testing technical advice. The Court is satisfied that defendants have met their burden and holds the test is validated for content in the present stage of the law and art of testing.

Defendants have now voluntarily tentatively decided to experiment with their testing approach and will not use a critical-incident judgment test in the forthcoming March exam. Plaintiffs suggest that this is an admission of the invalidity of the test in the past and cannot moot the instant case. They are quite correct that the Court would not allow a change in the rules of the game to moot the question raised by plaintiffs' claims. However, the Court does not find this proposed change probative on the issue of the validity of previous tests which must stand or fall on their own merits in light of the appropriate standards. Defendants are not to be straitjacketed by the filing of this law-suit, but must remain free to exercise their judgment in experimenting with methods for making promotions to sergeant.

At the Court's request, plaintiffs have filed a detailed outline of relief they consider appropriate in this case. They would drop the challenged judgment test and recompile the roster, giving 70 percent to the supervisory rating and 30 percent to the true-false test. Back pay and retirement benefits would be granted solely to blacks who would have qualified under this revision. A decree fashioned along these lines would cause havoc. It would base promotion primarily on the highly questionable subjective supervisory ratings and would grossly discriminate against whites who might also have qualified under the new system. Apart from morale effects, it would stimulate endless litigation and require extensive proceedings after notice to all potentially affected. Even if plaintiffs had prevailed it is extremely doubtful, given the complete discretion of the Court (42 U.S.C. § 2000e-5(g)), that any remedy beyond elimination of the challenged test would be granted.

In accordance with the foregoing findings of fact and conclusions of law here stated, this action is dismissed as to all defendants.

So ordered.

[*] Equal Employment Opportunity Act of 1972, Pub.L. No. 92-261, 86 Stat. 103 (March 24, 1972).