Full opinion text
FINDINGS OF FACT FULLAM, District Judge. I. Background The Commonwealth of Pennsylvania and several black citizens of Philadelphia, representing a class of black police officers and unsuccessful applicants for positions in the Philadelphia Police Department, brought this action against the City and certain of its officials, alleging that the Police Department’s hiring and promotion practices discriminate unconstitutionally against blacks. The action was commenced in December 1970. After extensive discovery, requiring several applications to the Court, the plaintiffs moved in April 1972 for preliminary relief. Five and one-half days of hearing were held, at the conclusion of which, on May 26, 1972, I entered an interim order temporarily restraining the defendants from hiring any policemen except as required to fill vacancies which existed on March 1, 1972. In hiring to fill such vacancies, the defendants were required to hire at least one black applicant for every two white applicants. 345 F.Supp. 305. Defendants have appealed from this order, and in the interim have decided not to conduct any hiring. On June 8, the Court of Appeals denied defendants’ application for a stay of the order; the appeal is still pending. The May 26 order did not affect promotions, partially in reliance upon the representation of counsel for defendants that only 14 promotions were contemplated. During the week of June 12, defendants promoted 56 policemen. Plaintiffs moved for an injunction against further promotions pending a decision on the motion for preliminary injunction, and for an order requiring defendants to rescind all 56 promotions except those which complied with the one black to two white standard previously set forth for hiring. On June 15, I denied plaintiffs’ motion, except to the extent of enjoining further promotions beyond the additional 16 which were then represented by counsel for defendants to be contemplated and deemed necessary. The principal reason for this result was that, by happenstance, all promotions and proposed promotions after the initial group of 30 (i. e., the additional 26 already promoted and the 16 proposed) included nearly one-third blacks. Applicants for positions in the Police Department must undergo a four-step elimination procedure, consisting of a written examination, a physical and psychiatric examination, and a background investigation and evaluation. Plaintiffs challenge the first and last steps, contending that (a) the written examination eliminates a disproportionate number of blacks and has not been shown to be a valid predictor of performance as a policeman, and (b) the background investigation has a similar discriminatory effect, as a result of the selection of negative factors which are deemed disqualifying, and of the discriminatory application of these factors. With respect to promotions, a three-factor rating (written examination, seniority, supervisor’s performance rating) is employed in selecting among applicants for promotion up to lieutenant; for higher ranks, those three factors are combined with an oral examination. Plaintiffs allege that the written examination operates discriminatorily, and does not predict subsequent job performance. II. Hiring — Overall Statistics The percentage of persons hired as police officers who are black has declined from 1966 to 1970 as follows: YEAR 1966 PERCENTAGE BLACK 27.5% 1967 25.3 1968 15.3 1969 11.2 1970 7.7 The result has been a decline in the proportion of police officers who are black. PERCENTAGE BLACK 20.8% 20.7 20.1 18.6 18.0 Defendants submitted evidence demonstrating that the proportion of blacks in the Philadelphia Police Department compares favorably with those in many other cities. Plaintiffs submitted evidence demonstrating that this proportion compares unfavorably with those in other city agencies in Philadelphia, and that, notwithstanding a substantial increase in the size of the police force, the actual number of black policemen has been declining in recent years. III. Hiring — Written Examination A. Discriminatory Effect Comparison of the passing rates of blacks and whites is the most effective means of determining whether the written examination disqualifies a disproportionate number of black applicants. However, there is no official record from which to derive the statistics necessary to construct such a comparison, and the parties are unable to agree on the figures. Accordingly, plaintiffs submitted evidence from which initial statistical findings could be made. 1. Approximately 35% of all persons who take the written examination for positions in the Police Department are black. The City of Philadelphia is forbidden by state law to maintain records of the race of applicants who take the written examination. Plaintiffs therefore made several studies on which to base an estimate of the actual racial percentages among applicants; all of these studies indicate that approximately 35% to 40% of the applicants are black. One, a study made by William Powe, a graduate student at the University of Pennsylvania, ascertained that 38% of applicants were black. Because Mr. Powe did not himself make the determinations of race, and because the individual who did, presumably, relying on his purported expertise, did not testify, I cannot give any independent weight to this figure; however, since it is close to the other approximations presented by plaintiffs, it tends to corroborate their accuracy. In the course of his study, Mr. Powe interviewed Richard Garlatti, Director of Recruiting for the City of Philadelphia, who advised him that the City “does not have a problem in recruiting blacks to take the police test.” (N.T. 36.) Mr. Garlatti also testified to his “guesstimate” that 40% of all applicants are black. (N.T. 202.) Finally, Dr. Bernard R. Siskin, Assistant Professor of Statistics at Temple University and one of plaintiffs’ experts, conducted a study in which he matched the addresses of a random sample of one-fourth of all applicants in 1969 and 1970 with the appropriate census block. Assuming that the likelihood that a particular applicant was black corresponded to the proportion of blacks in the census block (utilizing 1970 figures), Dr. Sis-kin concluded that 34.7% of all applicants were black. Defendants challenge the accuracy of the assumption underlying the use of the technique, arguing that blacks do not necessarily apply to become policemen in the same proportions as their number in the population, and Dr. Siskin admitted that the “non-statistical bias” could detract from the accuracy of his result. (N.T. 245.) But the defendants have made no effort to demonstrate that the census-block approach cannot validly be applied. If this contention were true, a survey of a sample of Dr. Siskin’s sample for which race is known or ascertainable might demonstrate the invalidity of his method. One of defendants’ experts, Dr. Erwin Taylor, testified: “I would say that in the so-called black ghetto where you would find very few whites, the likelihood of black applicants would be much lower than their representation in the population. I would say in some middle-class mixed neighborhoods it might be higher.” (N.T. 539.) This contention (unsupported by evidence) tends to support the use of a census-block technique. In the ghettos, where black populations may approach 100%, use of this method almost surely provides accurate results: if a block is 100% black, any applicant from the block will be black. In the integrated blocks, where there is more room for error, Dr. Taylor’s statement, if true, allows the plaintiffs to place somewhat greater confidence than otherwise possible in the assumption that blacks apply in proportion to their proportions in the population of a block. Moreover, other statistics in evidence tend to support the accuracy of 34.7% as a rough figure, perhaps even conservative in its estimate of black percentage. In 1969 and 1970, 2,027 whites and 586 blacks underwent background investigations; 22.4% of the total was black. (P-4, p. A-2.) Assuming that equal percentages of blacks and whites failed the intervening physical and psychiatric examinations, or dropped out before the background stage, 22.4% of those who passed the written exam were black. Assuming for the moment the accuracy of the estimate that 34.7% of the written test applicants were black, the passing rate among blacks is seen to be approximately 20.5%, rather than the 16.5% that Dr. Siskin estimated. Therefore, it is likely that Dr. Siskin’s technique tended to undercount the blacks in the population studied; at minimum, allowing for statistical error, the foregoing supports Dr. Siskin’s figure as reasonable. Finally, and very significantly, defendants have disclosed no attempt to develop their own estimate, despite adequate opportunity to do so. Dr. Taylor testified to his “best estimate” that 20% to 25% of all applicants were black, but on examination it was apparent that he was engaging in speculation. (N.T. 623-24.) His recognition that the proportion of blacks passing the written examination would be equal to the proportion of whites passing if 22.6% of applicants were black (D-8, p. 18) suggests that the determination of his “best estimate” may have been motivated by other than purely factual considerations. 2. The percentage of whites who pass the written examination is approximately 1.82', times as great as the percentage of blacks who pass. As demonstrated earlier, it is possible to ascertain from the statistics developed in connection with the background reports the approximate ratio of white passing percentage to black passing percentage. Assuming that the percentage of applicants who are black is 34.7%, the ratio is approximately 1.82 to 1. The far lower, and probably too conservative estimate of 30% would provide a ratio of 1.46 to 1. B. Validity of Examination Plaintiffs’ testing expert, Dr. Richard S. Barrett, testified to his opinion that the written examination has not been devised in a professionally acceptable manner or in a manner consistent with guidelines promulgated by the Equal Employment Opportunities Commission; that the result is a poorly-constructed test which discriminates unfairly against blacks and which may not adequately predict job performance; and that it would be possible to fashion an appropriate examination. Defendants’ evidence was to the effect that the examination is a valid predictor of performance at the Police Academy for blacks and whites; that performance on the job cannot be predicted; and that mere inspection of a test is insufficient to permit an evaluation of its appropriateness. 3. The written examination has not been demonstrated to be job-related. Defendants’ experts, Drs. Joseph Mamelak and James W. Gaither, presented a report (D-13) demonstrating a statistically significant correlation between entrance examination grades, on one hand, and Police Academy final written examination grades and final Academy average, on the other. Dr. Taylor made a similar analysis, including also correlations between entrance score and grade on the three particular Academy courses for which grades are available (firing range, driver instruction, first-aid) and correlations by race. Almost all correlations developed were highly significant. On the basis of these studies, defendants contend that the entrance examination has been shown to be a valid predictor of performance. For several reasons, I cannot accept this conclusion. Assuming the accuracy of defendants’ study, there has been no showing of any' correlation between success at the Police Academy and effective performance on the job. Defendants contend that it is appropriate to develop entrance examinations to predict performance in a training program, but absent evidence of the program’s success in achieving its objectives, evidence of the examination’s competence to predict ability to complete training is of little assistance in determining whether the examination is sufficiently job-related to permit its use despite its discriminatory effect. Moreover, unless there is perfect correlation between Academy grades and on-the-job performance, a virtual impossibility, an examination which predicts the latter would be preferable to one which predicts success in training. Defendants’ experts, Drs. Mamelak (N.T. 698-99) and Gaither (N.T. 782-85) assert that it would be impossible to develop a predictor of on-the-job performance, primarily because a policeman’s functions are extremely varied and it might be difficult to define success. Dr. Mamelak has little or no expertise in the area of testing, however (his description of work done in the testing field, N.T. 669, indicates that his function has been as mathematician), and Dr. Gaither has none in the area of employment examinations. Dr. Taylor, who does have such expertise, agreed with the Mamelak-Gaither opinions (N. T. 555), without stating reasons. It appears that this conclusion is based on a misapprehension as to the state of the art. Dr. Barrett’s testimony to the effect that creation of a valid job-related examination is possible (N.T. 329-30; 853-58) seems to be based upon an appreciation that a policeman’s job is a complex one, but also on a belief that other positions for which valid tests have been developed are no less complex. The existence of validation studies in Detroit and Chicago indicates that the job is not so hopeless that experts are not willing to attempt to create a valid police examination, although the examinations that have been developed have not been put into effect. Since the City has apparently never attempted to develop a test to predict performance on the job, its assertion that any attempts would be futile is premature. Moreover, the correlation made by Drs. Mamelak and Gaither may be misleading. There is no evidence as to the type of examination administered at the Academy. If in fact tests of the type contained on the entrance examination were used, a high correlation could be anticipated even if an individual had not mastered the training material. Finally, the Mamelak-Gaither study did not, and of course could not, consider the effect that the designation of a passing grade for the entrance examination could have had (N.T. 739). It is possible that candidates with lower scores could have successfully completed the training. That no recruit has been dismissed from the Academy in the past five years for academic reasons (P-19) suggests that a lower passing grade might well have been appropriate. Dr. Taylor testified (N.T. 547) that in theory, the passing grade should be determined by means of a validity analysis, but he believed that it would be impractical, because “it would mean admitting . into the academy three or four or five or ten times the number of people that we plan to graduate and letting them go through . . . then throwing out say 90% of them. . . . ” (because a validation study requires acceptance of a large number of applicants, including some who might not otherwise be accepted, and correlation of their examination results with subsequent job performance). The assumption that 90% would fail out of the Academy assumes the conclusion that a lower passing rate would be inappropriate. Moreover, it would not necessarily be appropriate or mandatory to admit into the Police Academy all persons who take the written examination (and who are not otherwise disqualified) in order to ascertain a valid passing grade. It is very likely that some of the present members of the Police Department failed the test before ultimately passing it; their job performance would be informative. Additionally, testing experts can reach judgments as to whether some applicants may be so obviously unqualified on the basis of their performance on the examination that their acceptance on the police force would not assist in the validation of a passing grade and/or would constitute an unacceptable risk. It should be remembered that the passing grade that must be validated is a score on an examination developed by testing experts and therefore more likely to be valid, according to Dr. Barrett’s testimony, than the present test; a refusal to give further consideration to applicants who score below a given low grade on such a test, which is likely to be found valid, would therefore be justifiable. In view of the finding that the examination has not been validated as job-related, it is unnecessary to consider plaintiffs’ assertion (which is, of course, bound up with the contention that the test is not job-related) that the test is a poor one in that it is too difficult, it rewards test-taking ability, and it examines applicants in inappropriate subject-matters. If the examination were proved to be job-related, these contentions might be relevant if it were thought appropriate to require defendants to devise the least discriminatory test possible. See Chance v. Board of Examiners, 458 F.2d 1167, 1177, 1178 (2d Cir., 1972) (unnecessary to decide whether least discriminatory means standard is appropriate); Castro v. Beecher, 459 F.2d 725, 733 (1st Cir., 1972) (least discriminatory means standard inappropriate). Nor is it necessary to consider at this point plaintiffs’ contention that defendants have not employed the proper means to create a valid entrance examination. IV. Hiring — Background Investigation A. Accuracy of Plaintiffs’ Data 4. Plaintiffs’ statistical evidence with respect to the background investigations is substantially accurate, and sufficiently probative to warrant its use at this preliminary stage of these proceedings. The data submitted by plaintiffs in support of their contentions that the background investigation process is discriminatory is based in large measure on a study of the police files which contain the investigative reports of all applicants accepted and rejected in 1968, 1969 and 1970. Based upon their review of the background files of the individual plaintiffs, Dr. Bernard R. Siskin, plaintiffs’ statistical expert, and Prof. Robert Reinstein, one of plaintiffs’ attorneys, compiled a list of 25 factors, representing every category of derogatory information which appeared in the files and any others thought likely to appear. A form incorporating this list and including a category “other” was developed (P-2); it will be referred to as the coding sheet. Dr. Siskin chose Mr. and Mrs. Ralph Kates, a law student and a graduate student in communications, respectively, to supervise a team of students who were to examine the police files, locate all items of derogatory information, and report their findings on the coding sheet by making a notation opposite the appropriate category. The coders were selected by Prof. Reinstein from the Temple Law School student body, and by Mr. and Mrs. Kates from among people they knew personally and people who responded to a notice placed on a University of Pennsylvania bulletin board. Some of the students were volunteers; others were paid. Mr. and Mrs. Kates instructed their workers in the meaning of the categories on the coding sheet; the sources of information within the background files; the code which was employed to transcribe information from the files to the sheet. Each coder’s entire work product was checked by Mr. or Mrs. Kates until the coder had turned in one full day of error-free work. After that, a random sample consisting of approximately ten percent of each coder’s work was inspected; if any inaccuracies were spotted in any category, each of that coder’s files was checked for errors in that category. In order to insure uniformity, Mr. and Mrs. Kates communicated daily with Dr. Siskin to classify individual bits of information which did not fall clearly within one of the coding sheet’s categories. Defendants challenge the accuracy of the statistics which were developed from the information contained on the coding sheets. Several objections are made. a. The coders were biased. Defendants contend that the coders could not be objective, because they were aware of the purpose of this action, and knew that a judgment for plaintiffs must be based upon a finding of discrimination. Particularly is this asserted to be so in the case of the volunteer coders. I find defendants’ assertions are without merit. The coders were informed of the need for accuracy (N.T. 126, 273); it should have been apparent to them that any errors could have been easily spotted by the defendants (who were given copies of the completed coding sheets (N.T. 823-24)), and that any pattern of errors would detract from the credibility of the study and from the plaintiffs’ likelihood of success. Additionally, 70 to 80 percent of the coding sheets and their respective files were checked by Mr. or Mrs. Kates (N.T. 976). Mr. Kates testified and left no doubt as to his concern for the accuracy of the project. There is no reason to suppose that his wife, whose credentials are excellent (N.T. 941) was any less conscientious. Finally, defendants, who have access to the completed coding sheets, have not 'come forward with any examples of errors. Any significant bias, if reflected in the coders’ work, would be apparent on a comparison of files with coding sheets. b. Some coders tuere more lenient than others. Defendants contend that even if the coders were not biased, they did not all employ the same standards in determining whether an item in the background file should be recorded on the coding sheet. In support of this assertion, Dr. Taylor testified, for example, that the coder designated as number 1 found an average of 2.68 negative factors in accepted blacks, but coder number 2 found 4.58 (N.T. 488-91). However, these statistics are meaningless without evidence that the files examined by the two coders to be compared were comparable. The Police Department provided the coders with files in boxes of fifty to one hundred each, each box containing files of either accepted or rejected applicants, not both, and usually containing files from one year only (N. T. 959). Thus, it is virtually certain that there is no valid basis for comparison between coders. Moreover, Dr. Taylor’s figures (D-8, p. 1) show that the average number of negative factors for accepted blacks increased between 1968 and 1970 by 46%; the discrepancy between the coders’ reports may simply reflect this trend. Again, defendants presented no specific evidence that coders treated factors differently. c. The categories chosen for the coding sheet were ambiguous and lent themselves to subjective determinations. Defendants contend that the coders might have had to exercise judgment in determining the proper category for a particular negative item, or in determining the appropriateness of inclusion of an item on the coding sheet. However, the testimony of Mr. Kates (N.T. 949-58) demonstrated that the categories were well-defined and that negative information which would constitute one of the designated factors would be found, if at all, at one particular location in the file. d. The categories are too broad. Defendants argue that some of the factors chosen for study are so broad that they encompass both trivial matters likely to be ignored in the Police Department evaluation, and serious, potentially disqualifying, matters. There is considerable merit in this contention, particularly with respect to such categories as “arrests,” “adult convictions,” and “job problems.” Plaintiffs have recognized the overbreadth of some of these factors, and are narrowing their scope in order to increase the accuracy of studies based upon them. (N.T. 985-90.) In the meantime, however, it is appropriate to make use of the studies which presently exist. This is a preliminary stage in the proceedings, and plaintiffs need only demonstrate a substantial likelihood of success on the merits. Obviously, the limitations of data based on inadequate factors must be borne in mind, and some caution in evaluating plaintiffs’ statistical evidence is indicated. On the other hand, defendants have not demonstrated that the derogatory information collected against black applicants in any of plaintiffs’ categories tends to be more damaging than that which exists against white applicants. In the absence of any such evidence, I am not prepared to assume that negative factors attributable to blacks in the “serious” end of the range encompassed by these categories occur so much more frequently than factors attributable to whites as to render plaintiffs’ statistics useless. e. Serious inconsistencies exist between coding sheets representing files which were examined twice. Defendants presented evidence of 85 or 89 files which were ostensibly examined twice or three times. Comparisoris of the coding sheets produced for each file demonstrated inconsistencies which would cast significant doubt upon the accuracy of plaintiffs’ entire study, if in fact the coding sheets did represent files twice examined. However, examination of the sample “duplicates” introduced by defendants and of plaintiffs’ explanation (presented by Mr. Kates, N.T. 962-73) suggests that many of the comparisons are invalid, because the Police Department erroneously copied coding sheets which were incomplete (because the files presented to the coders were incomplete) and compared them with the completed sheets, or made errors in stapling the pages of the coding sheets, or erroneously applied identical numbers to different files. Not all of the discrepancies were explained, and therefore defendants have shown that the coders were not perfectly accurate or consistent. However, it would be surprising if some errors did not appear in an operation of this nature and magnitude. The misleading evidence presented by defendants is insufficient to warrant a finding that the degree of error is greater than that which I would have been prepared to assume. f. The coders designated as “rejected” some applicants who were not rejected on the basis of the background examination. Plaintiffs’ coders were instructed to classify as “other” notations in the file that an applicant was rejected for medical reasons after passing the initial physical examination, or declined an appointment to the Police Department, or was ineligible for failure to meet the residence requirement. Obviously it is erroneous to designate such applicants as rejected on the basis of the background investigation, and their inclusion in this category detracts from the accuracy of plaintiffs’ data, to the extent that no final order could be entered which relied on it. Plaintiffs have taken steps to cure this defect, however (N.T. 999), and the issue at this stage of the case is whether the statistics now available are sufficiently reliable to warrant a finding that plaintiffs are likely to prevail on the merits. The only evidence in the record of the relative proportions of blacks and whites erroneously designated as “rejected” with an “other” factor is defendants’ study of twelve blacks and 47 whites whom Dr. Siskin’s study classified as rejected with one factor. Of the twelve blacks, one (8.33%) was actually rejected as a result of the background investigation; four of the 47 whites (8.51%) were so rejected. In the absence of any showing that the plaintiffs’ error affects black and white applicants disproportionately, I conclude that the statistics compiled by Dr. Siskin may validly be utilized at this point. Finally, with respect to all of defendants’ challenges to the accuracy of plaintiffs’ data, it is significant that apart from these attacks and some minimal efforts to demonstrate that blacks are accepted with more factors than whites (discussed at page 1096, infra), defendants have made no real efforts to demonstrate the absence of discrimination at this stage of the hiring process. Of course, plaintiffs have the burden of proving discrimination, but the Court may consider the nature of the defense not only for the purpose of drawing inferences as to the defendants’ ability to produce rebutting evidence, but also as support for reliance upon plaintiffs’ imperfect statistics, which comprise the only substantial fact-gathering effort in evidence. B. Discriminatory Effect 5. The rejection rate for blacks is far higher than the rejection rate for whites. This disparity has increased since the Police Department assumed the responsibility for evaluation of the background reports. Statistics for 1968, 1969, and 1970, the only complete years available, demonstrate that a far higher percentage of white applicants than black applicants are accepted following the background investigation. (P-4, p. A-2.) % white % black accepted accepted 1968 67.7 41.9 1969 59.3 21.1 1970 70.8 28.3 Moreover, the disparity between black and white passing rates has increased significantly since June 1969. Prior to that month the City Personnel Department decided, on the basis of background information obtained by the Police Department, which applicants were to be accepted. Since then, the Police Department has made these determinations. The parties differ as to their characterization of this shift in responsibility, but resolution of that dispute is unnecessary. % white % black accepted accepted 1968 and Jan-June 1969 65.4 36.4 July-Dee 1969 and 1970 67.0 23.6 (N.T. 103.) Defendants’ rebuttal is meaningless. According to defendants’ expert, Dr. Taylor, between 1968 and 1969, the white rejection rate increased by 26%, and the black rate by 35%; this difference is asserted to be “hardly sufficient to support the Plaintiffs’ claim of vicious bias.” Apart from the fact that the difference is in fact a substantial one, Dr. Taylor has set up the straw man of a trend towards greater discrimination against blacks between 1968 and 1969 and rebutted it, but plaintiffs never purported to compare rejection rates for those years. Their comparison was between 1968 to mid-1969 and mid-1969 to 1970. Dr. Taylor has attempted to rebut the latter figures (“false and misleading!”) by challenging the use of pre-1968 statistics in plaintiffs’ calculations. However, as Dr. Siskin makes clear in his report (P-4, p. 3), he did not use pre-1968 figures in these calculations. C. Discriminatory Application of Factors Chosen 6. Plaintiffs have presented prima facie proof that similarly-situated black and white applicants are not treated alike. The approaches taken by the experts on statistics offered by plaintiffs and defendants to the question of discrimination by the Police Department’s background evaluators differ considerably. I have concluded that Dr. Siskin’s analysis on behalf of plaintiffs presents a far more meaningful picture of the selection process than that offered by defendants’ expert. Dr. Siskin’s analysis is in two steps. For each of the 25 factors (plus “other”) on the coding sheets, he calculated the likelihood that an applicant with that factor would be rejected. (P-4, p. A-4.) In all 26 cases, the probability of rejection for a black with a given factor exceeded the comparable white probability. (The table is reproduced in Appendix A.) This result alone is not surprising, and does not demonstrate the presence of discrimination, because other figures show that black applicants have, on the average, more negative factors than white applicants. (In the case of 22 of the 26 factors, the proportion of blacks with that factor exceeded that of whites.) Thus, the fact that 66.2% of white applicants with convictions, but 90.5% of black applicants with convictions, were rejected, might be explained by the likelihood that more of the blacks than the whites had other negative factors and were rejected on the strength of all of the derogatory information in the file. Recognizing the inadequacy of these figures, Dr. Siskin sought to eliminate the influence of the additional factors by holding the number of factors constant. He found that for every number of factors up to eight, and for nine or more, the probability of a black’s rejection exceeded that of a white’s. The disparity was often substantial, as in the case of applicants with three factors. The white rejection rate was 26.-8% ; the black rejection rate was 53.7%. (See Appendix B.) If it is assumed that all of the factors in this survey are considered and tend to reduce an applicant’s chance of acceptance (an assumption which will be dealt with later), these two sets of statistics can be consistent with an absence of discrimination only if black applicants tend to have far more serious factors than white applicants; that is, for example, if the three factors attributable to a black applicant with three factors are likely on the average to be so much more severe than the factors attributable to a white applicant with three factors as to constitute an explanation for the fact that the black rejection rate is twice as high as the white rejection rate for these applicants. Plaintiffs have presented evidence that the factors which tend to occur most frequently among blacks also tend to occur most frequently among whites. (N.T. 256-57, 280-81.) Statistical calculations not in the record may ultimately be presented to show the likelihood, given an applicant has a given number of factors, of any factor’s occurrence on his list. In the absence of any such evidence, I must rely on plaintiffs’ expert’s unrebutted testimony that blacks and whites are likely to have similar factors. Accordingly, I find that the evidence thus far considered does establish a very substantial likelihood that similarly-situated applicants (i. e., those with the same negative information in their files) are not treated alike. Defendants submit several interrelated statistical rebuttals, none of which is convincing. First, they contend that because accepted blacks in all three years covered by the study had a higher number of negative factors than accepted whites (e. g., in 1970, blacks had 6.46 factors on the average, and whites had 4.12), the Police Department must be more lenient in accepting blacks than whites. But plaintiffs properly point out that black applicants have more factors than whites, and therefore almost any discriminatory or non-diseriminatory policy would result in accepted blacks with more factors than accepted whites. Second, Dr. Taylor criticizes Dr. Siskin’s table (Appendix B), which demonstrates that whites and blacks with equal numbers of factors are treated differently, on the ground that Dr. Siskin failed to take into consideration the fact that a larger proportion of black applicants were rejected. Dr. Taylor’s solution is a table, the significance of which is not readily apparent and is not satisfactorily explained, and the statistics in which are completely consistent with the existence of discrimination. Third, Dr. Taylor considered with respect to several factors the percentages of rejected whites with the factor and the percentage of rejected blacks with the factor, and the percentages of accepted whites and blacks. Where the first two percentages were roughly equivalent, and the last two were, or where the difference between the two sets of percentages greatly exceeded the difference between the black and white percentages, he concluded that there is no disproportionate rejection rate. Thus, 19.3% of all accepted whites and 19.3% of all accepted blacks had falsified their applications. Similarly, 79.-7% of all rejected blacks and 77.1% of all rejected whites were charged with this factor. The size of the difference between rejected and accepted (approximately 60%) greatly exceeds the size of the difference between black and white accepted (0%) or black and white rejected (2.6%), and this purportedly demonstrates the absence of discrimination. That it demonstrates nothing of the sort can readily be shown. Finally, defendants contend that Dr. Siskin’s data is incapable of shedding light on the rejection process because it cannot assist in ascertaining the “cause” of rejection in any particular case. That is, many of the factors chosen by plaintiffs, while present in an applicant’s file, might have been disregarded by the person in charge of evaluating the file. In support of this assertion, defendants refer to Civil Service Regulation 8.02, which sets forth the grounds upon which an applicant may be disqualified ; defendants contend that an applicant without one of the stated disqualifying factors will be accepted. Plaintiffs attempted to rebut this argument with the testimony of two of the plaintiffs. In both cases, though, the applications contained what might be construed as false information, although trivially false if at all and almost undoubtedly not deliberately false. Since deliberate falsification is a ground for disqualification, I cannot find that plaintiffs have demonstrated conclusively that the Civil Service Regulation is not complied with. Evidence that noncompliance is prevalent, if that is a fact, should be readily available to plaintiffs, and may of course be presented at a later stage in the proceedings. But even without such evidence, defendants’ contention does not undermine substantially the validity of plaintiffs’ analysis. Under the Civil Service Regulation, an applicant “may” be disqualified if he has one of the listed factors. (N.T. 187.) There is room for the exercise of discretion. Since all negative factors in an applicant’s file are considered at least to the extent of placing them on a rejection report if the evaluator recommends rejection (Deposition of Chief Inspector Richard F. Bridgeford, P-10, p. 76), it is plain that the categories set forth in the Civil Service Regulation do not constitute the outer limits of the evaluation. It is appropriate to consider all factors which are deemed to be of sufficient importance to warrant inclusion in the file, on the assumption that if they find their way into the file, they will play a role in the evaluation process. There is virtually no evidence of record relating to the decision-making process; none of the six persons who evaluate the background files testified as to what factors they do consider. At minimum, evidence that no attention is paid to derogatory information not set forth in the Civil Service Regulation would be necessary before exclusion of such factors from the analysis would be proper. Moreover, there is some evidence, admissible against some but not all of the defendants, that evaluators have considered factors other than those set forth in the regulation. In addition to logical conclusions that may be drawn from plaintiffs’ fairly complicated statistical data, it may be appropriate to infer unequal treatment from the evidence of rejection rates for the periods before and after the Police Department assumed responsibility for evaluation of background reports. (See finding 5, supra.) The marked decline in black acceptance after June 1969 makes a finding of discrimination almost inevitable (except in the unlikely event that it can be shown that pre-June 1969 practices favored black applicants, and that the shifts merely ended that bias). Although there is evidence that the Police Department may have adopted a broader definition of disqualifying factor than that which the Personnel Department had used, it may also be the case that the Police Department evaluators do not act in as evenhanded a manner as did their predecessors. Whichever of these or any other possible hypotheses is the case, the defendants have permitted to stand unexplained statistics which cast considerable doubt upon their contention that the background examination is nondiscriminatory. One consideration not developed in depth by the parties is the effect of positive information in the file. It is not clear to what extent derogatory notations could be offset by evidence that an applicant is otherwise especially well-qualified for employment. If the parties believe that exploration of this area is indicated, testimony from the background evaluators might be particularly helpful. Another area touched upon is the fairness of the fact-gathering procedure; for example, there is conflicting evidence as to whether an applicant is given an opportunity to explain or deny derogatory information which is collected against him. On the basis of the foregoing, I conclude that plaintiffs have made a prima facie showing that, as between applicants of similar background, blacks are treated less favorably than whites. While it cannot be said that they have conclusively proved that discrimination exists, they have made the necessary showing of a substantial likelihood of ultimate success on the merits to support their claim for preliminary relief. D. Discriminatory Choice of Factors. 1. Even if the evaluation tvere performed in a nondiscriminatory fashion, some of the factors relied upon to disqualify applicants would disqualify a disproportionate number of black applicants. A determination of whether the factors chosen for evaluation tend to disqualify a disproportionate number of black applicants can be made by reference to the table designated by Dr. Sis-kin as “Incidence of Factors by Race” (P-4, p. A-5; duplicated in Appendix C), which lists the probabilities that a black or white applicant would have a particular factor. Nondiscriminatory use of these factors to disqualify applicants would disqualify more blacks than whites in the case of 21 factors (plus “other”); the contrary is true in three cases; both groups would be equally affected in one case. The record is inadequate to allow me to select the factors the use of which creates a discriminatory effect of sufficient magnitude to warrant requiring the defendants to justify their application. There is no evidence, for example, of whether the disparities between black and white frequencies are statistically significant or merely the result of chance. In the case of those disparities which are insignificant because of the infrequency with which the factor has occurred in the population studied, it may be necessary to study a larger population to determine whether any significant discrimination is likely to result from application of the factor in the future; obviously it would be preferable to limit the scope of the inquiry to applicants for the Police Department, but this may not always be feasible. It is clear at this point that use of some of the factors employed in the background investigations will be shown to affect blacks to a significantly disproportionate degree. Although I will not speculate as to what the statistics will ultimately demonstrate, it seems appropriate to set forth as an example of such a factor “Illicit or Immoral Conduct,” which is attributed to 29.4% of black applicants, a very substantial proportion, and 9.7% of whites, resulting in a large disparity. 8. The background investigation has not been demonstrated, to be job-related. It is unquestioned that defendants have concluded no study “validating” the background investigation criteria, i. e., demonstrating that they are job-related. 9. Validation of the background investigation criteria would be feasible. Dr. Barrett testified that it would be feasible to conduct a study to determine whether the background investigation is job-related. It would first be necessary to define successful performance on the job. Next, one could look at the performance records of persons presently on the force who were admitted despite given negative factors, and ascertain whether a correlation exists between those factors and performance. Since none of the factors in plaintiffs’ study as presently defined are disqualifying in all cases, there have been accepted between 1968 and 1970 some applicants with each factor. If the number of policemen with a given negative factor is too small to allow the completion of a statistically valid study, it might theoretically be necessary consciously to decide to accept applicants with that factor in order to correlate the presence with job performance. To avoid the obvious risks that this would entail, Dr. Barrett suggests that common sense and experience, and perhaps study by a panel of experts, would make it possible to reject applicants society cannot afford to make policemen. (N.T. 361-64.) On the basis of Dr. Barrett’s unrebutted testimony, I conclude that it would be possible to conduct a study to determine which criteria utilized in the background investigations are job-related. Problems with respect to forced hiring in order to conduct the study can be dealt with as they arise; in many of these cases, it is likely that use of a factor to disqualify will be so obviously appropriate that no statistical showing of job-relatedness would be necessary. See Developments in the Law-Title VII, 84 Harv.L.Rev. 1109, 1151-52 (1971). V. Promotion 10. The written examination for promotion to ranks above patrolman plays a significant role in determinations of eligibility and eliminates a disproportionate number of blacks. There is little evidence in the record concerning the promotion process. It is uncontroverted, however, that a written examination is a very significant factor in determining eligibility for promotion. In the ease of promotion to corporal, detective, sergeant and lieutenant, the grade obtained on the examination comprises 90% of an applicant’s objective rating, which is considered in conjunction with his supervisor’s assessment of his performance; for promotion to higher ranks, the written examination score is assigned a weight of 50%. Plaintiffs’ statistics, which are unrebutted, demonstrate that whites pass at least three of the written promotion examinations at substantially higher rates than blacks: WHITE PASS BLACK PASS TEST RATE RATE Sergeant 18.4% 11.7% Detective 27.2 17.7 Corporal 36.2 23.8 There is no evidence as to the statistical significance of these figures, nor as to the passing rates for the other promotion examinations. 11. The written promotion examinations have not been proven to be job-related. It is admitted that defendants possess no proof that the written examinations are job-related. Although defendants contend that the tests “satisfy technical criteria for content validity,” no evidence of validity has been presented. There is expert evidence (testimony by Dr. Barrett) that to some extent these examinations merely test an applicant’s test-taking ability, and are therefore unlikely to be valid. 12. Underrepresentation of blacks in ranks higher than patrolman may be due in part to discrimination in the selection of patrolmen. The percentage of officers who are black decreases for each rank as rank ascends. The statistics do not indicate whether the increasing underrepresentation is caused by a discriminatory written examination alone, or by that in conjunction with other factors. For example, if the entrance examination is deficient in that it measures mere test-taking ability, an applicant might improve his ability by repeated applications (and examinations) until his proficiency was sufficient to allow him to pass. (One of the plaintiffs took the examination eight times before passing.) If such a person would have passed a valid, job-related examination on his first attempt, the discriminatory nature of the invalid examination will have deprived him of some seniority. Since, seniority is considered in determining eligibility for promotion (seniority constitutes 10% of an applicant’s objective grade), it is apparent that discrimination in hiring may-affect subsequent promotion. 13. There has been no showing that discrimination occurs in stages of the promotion process other than the written examination. Plaintiffs have not introduced evidence of discrimination in the other two steps of the promotion process: supervisor’s rating of performance (for all ranks) and oral examination (for ranks of captain and above). The only evidence bearing on these processes is that all of the administrators of the oral examination are white, which alone does not support a finding of likelihood of proving discrimination. 14. The range of test scores among blacks who pass the written examinations is smaller than the range among whites who pass. It is undisputed that the difference between high and low passing scores is less in the case of black applicants, and that blacks tend to be “clustered” at the middle and lower positions on the eligibility lists. Accordingly, the longer a given eligibility list has been used as the source of promotions, the greater the likelihood that the percentage of blacks promoted will more nearly approach equality with the percentage of black applicants. VI. Future Contemplated Hiring and Promotions The Police Department’s normal attrition rate is approximately 35 to 40 policemen per month. Typically, approximately 100 applicants are selected to commence training at the Police Academy approximately every two-and-one-half months. Additionally, efforts are being made to secure federal funds in order to hire 900 new police officers. (N.T. 10.) Plans for future promotions are less definite, since promotions tend to be announced sporadically. (See N.T. 1081.) VII. Irreparable Harm 15. Plaintiffs and the members of the class of plaintiffs would be likely to suffer irreparable harm in the absence of preliminary relief. Continued use of hiring and promotion practices which discriminate against blacks necessarily causes irreparable injury to those discriminated against, as well as to the public at large. United States by Mitchell v. Hayes International Corp., 415 F.2d 1038 (5th Cir. 1969). SUMMARY AND CONCLUSIONS It is not the function of this Court or any other court to supervise the operation of the Philadelphia Police Department, or to decide who should and who should not be appointed or promoted to positions in the Police Department, or to establish standards of eligibility for appointment or promotion. Within the limits established by the Constitution and applicable legislation, all such matters are within the province of the executive branch of government. The sole obligation of the court is to insure that, in the formulation and implementation of policies and procedures for the hiring and promotion of policemen, constitutional and statutory limitations are not violated. In the seminal case of Griggs v. Duke Power Co., 401 U.S. 424, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971), the Supreme Court established a simple and straightforward doctrine: any selection requirement or procedure which in fact disqualifies a disproportionately high percentage of blacks is illegal in the absence of affirmative proof that it bears a “manifest relationship to the employment in question” (at p. 432, 91 S.Ct. at p. 854). The absence of discriminatory intent is irrelevant. The Griggs case involved private employment, and was decided at a time when Title VII of the Civil Rights Act of 1964 did not apply to employment by governmental agencies. Nevertheless, it has uniformly been held that the standards of Title VII and the guidelines established by the Equal Employment Opportunities Commission, as interpreted in Griggs, provide “persuasive analogy” for the decision of similar questions in cases involving public employment. E. g., Chance v. Bd. of Examiners, 458 F.2d 1167 (2d Cir., 1972); Castro v. Beecher, 459 F.2d 725 (1st Cir., 1972); Western Addition Community Organization v. Alioto (Waco II), 340 F.Supp. 1351 (N.D.Cal., 1972). Moreover, by virtue of the Equal Employment Opportunity Act of 1972, P.L. 92-261, effective March 24, 1972, 86 Stat. 103. Title VII is now directly applicable to public employment as well as private. I recognize that the complaint in this case does not invoke Title VII directly, that administrative remedies thereunder have not been exhausted, and that it may be questionable whether pre-1972 statistics alone can provide an adequate basis for invoking Title VII directly. However, the present statutory law may properly be considered in determining the appropriate form of interim relief. Not all of the findings set forth above are preliminary in nature. While it is conceivable that, on final hearing, the evidence may justify some modification of the findings relating to discriminatory impact of the examinations, discriminatory application of the background investigation process, and discriminatory selection of disqualifying-factors, it is an established fact that no attempt has ever been made to validate the background screening process as job-related, and that the examinations have not been validated as job-related. It is virtually inconceivable that the defendants can avoid the ultimate necessity of either validating their present testing and screening procedures in relation to performance on the job, or devising new testing and screening procedures shown to be thus valid. Unfortunately, the defense of this case to date has been characterized by overreaetion at the emotional level, and underreaction at the rational, practical level. The raw statistics, and the probable results of plaintiffs’ statistical studies, have been known for more than a year. The defendants have made no independent statistical studies of their own, and have been notably unsuccessful in refuting plaintiffs’ conclusions. The law is clear, and has been since before this suit was filed. Surely it should be apparent to all concerned that further delay in undertaking the necessary validations would not be in the best interests of the Police Department or the public generally. In view of the sensitive nature of the controversy, I believe it particularly important to emphasize the following facts: 1. Requiring that hiring and promotion in the Police Department be done on a basis which does not discriminate against blacks except for reasons related to job performance does not imply a “lowering of standards,” but rather an improvement of standards to make certain that they accurately determine, on’ a non-diseriminatory basis, who is and who is not qualified. The fact that some blacks who have been disqualified under the existing procedures may be found qualified under revised procedures does not mean that the standards will have been lowered, in the absence of proof that the existing standards are accurate. In point of fact, while I have not found it necessary to make express findings on this subject, there is persuasive evidence in the record that the present entrance examination is so structured that the passing grade has been lowered to the point where a candidate can pass if he knows the answers to roughly one-third of the questions, and blindly guesses at the rest of the answers. No one flunks out of the Police Academy. Performance on the entrance examination and at the Police Academy have no bearing upon initial job assignments. The only police witness who testified on the subject stated, when asked whether he thought the Police Academy course was helpful in his work as a policeman, only that he thought it was helpful in enabling him to pass subsequent promotional examinations. 2. An interim requirement that hirings and promotions be effected on the basis of at least one black for every two whites does not amount to the imposition of a quota system as such, nor does it discriminate against whites. It merely means that, unless and until the defendants can justify excluding a disproportionate percentage of blacks, they are not legally permitted to do so. The defendants are apparently satisfied that the individuals on existing eligibility lists are properly qualified for appointment or promotion as the case may be. The interim restraints heretofore ordered permit the use of these lists on a neutral basis, i. e., in the same racial proportions as existed among the applicants from whom the eligibles were selected. 3. The morale of the Police Department, and the possible temporary adverse effects upon whites on existing eligibility lists, are matters of grave concern. However, I am unwilling to accept the notion that police morale would be adversely affected by requiring compliance with constitutional and statutory mandates, or that police morale is so fragile that questioning the validity of the procedures which have led to the establishment of existing eligibility lists would shatter it. The morale of minority members of the force and would-be members of the force must also be considered. Moreover, a great many members of the force are the beneficiaries of federal largesse, and substantial additional federal funds to underwrite expansion of the Police Force are now being sought. It would not enhance police morale to place these funds in jeopardy. Finally, it is important again to emphasize that the Court has not made any finding of intentional discrimination. Neither is there any suggestion that the City of Philadelphia compares unfavorably with other cities in minority employment. On the contrary, it appears probable that Philadelphia has done better than most cities in this respect. However, that fact does not remove the necessity of validating hiring and promotion procedures and eliminating whatever discrimination does exist. The Police Department of Philadelphia does lag behind other City departments in minority employment. On the present record, the assumption that this is due to the fact that fewer blacks are qualified for police service than for other forms of public employment is precisely that, an untested assumption. The record shows clearly that, since the Police Department was given the final responsibility for hiring policemen, in mid-1969, there has been a dramatic decrease in the numbers and percentages of blacks hired as policemen. This may very well be attributable to the (perhaps natural) unconscious supposition that the persons best qualified for police work are those who most nearly resemble in attitudes, characteristics, and life styles, the policemen making the selections. But in a City whose black population amounts to nearly 40% of the total, any such approach suggests the conclusion that minorities are incapable of contributing substantially to their own self-government. I find any such supposition totally unacceptable. An order will be entered which continues in effect the existing restraints on hirings and imposes similar restraints on promotions until final judgment in this case. Needless to state, this does not prohibit the defendants from undertaking immediate steps to validate the present testing and promotional procedures in relation to job performance, or to devise new tests and procedures which are so validated. APPENDIX A Probabilities of Rejection Given ground Exam (Table developed by Dr. Siskin). Factor on Back-plaintiffs' expert, FACTOR WHITE BLACK Conviction 66.2% 90.5% Arrests 72.0 93.0 Police Contacts 40.5 83.3 Traffic Offenses 32.4 67.5 Juvenile Delinquency 81.5 91.2 Juvenile Arrests 54.2 84.6 Juvenile Police Contacts 23.4 57.1 Court Martial Convictions 86.7 94.7 Summary Offenses In Military 50.0 83.5 Military Arrests 50.0 100.0 Military Discharge 75.3 94.4 No Valid Driver's License 57.8 89.3 Falsification of Application 63.1 87.4 Fired 74.5 94.2 Job Problems 70.3 89.4 Unemployed and/or on Welfare 38.2 75.0 Bad Credit 68.8 89.7 Education Academic Problems 52.3 85.8 Education Discipline Problems 59.9 91.9 Bom out of Wedlock 36.3 66.7 Divorce 67.9 79.4 Illicit or Immoral Conduct 64.7 90.0 Alleged Threats or Violence 81.9 95.5 Improper Conduct of Friends or Relatives 41.8 71.5 Bad Appearance 45.3 77.2 Other 52.9 79.2 APPENDIX B Probabilities of Rejection Given a Given Number of Factors (Table developed by plaintiffs' expert Dr. Sis-kin). NUMBER OF FACTORS WHITE BLACK 1 11.0% 29.5% 2 15.9 24.6 3 26.8 53.7 4 41.6 70.4 5 56.6 87.0 6 74.4 93.0 7 76.9 90.6 8 88.6 92.9 9 or more 92.2 94.1 APPENDIX C Incidence of Factors by Race (Table developed by plaintiffs' expert, Dr. Siskin.) FACTOR WHITE BLACK B%/W% Convictions 6.3 % 9.0% 1.4 Arrests 11.6 18.2 1.