IMAGE v. Bailar, 518 F. Supp. 800 (N.D. Cal. 1981)
April 21, 1981
Benjamin F. BAILAR, Joseph F. Morris, Guido Alasia and Austin Simon, Defendants.
United States District Court, N. D. California.
*801 *802 Christine Motley, Harvey Sohnen, Legal Aid Soc. of Alameda County, Oakland, Cal., for plaintiffs I.M.A.G.E., et al.
Deborah Seymour, Asst. U.S. Atty., San Francisco, Cal., for defendants Benjamin F. Bailar, et al.
MEMORANDUM AND ORDER
PECKHAM, Chief Judge.
Plaintiffs in this case are suing on behalf of "Hispanic persons," (see Stipulation and Order Redefining the Class, filed September 12, 1980, at 2) charging discrimination against Hispanics in employment at entry level positions by the United States Postal Service (hereafter "Postal Service") in the Golden Gate District, in violation of Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e et seq. The plaintiffs are focusing on three examinations given by the Postal Service to those desiring certain entry level positions: Test 400, Test 440, and Test 450. This preliminary injunction motion is only concerned with the impact of Test 450, which is given to those applying for a position as mailhandler. Plaintiffs are asking for preliminary relief to alleviate the present impact of Test 450, which they claim has a discriminatory effect upon Hispanic applicants.
In order to issue a preliminary injunction, the court must find that the plaintiffs have shown either (1) a probability of success on the merits and that the plaintiffs would possibly suffer irreparable injury absent an injunction, or (2) that there is a sufficiently serious question on the merits to provide a fair ground for litigation, and that the balance of hardships tips decidedly in favor of the plaintiffs Wright v. Rushen, 642 F.2d 1129 at 1132 (9th Cir., 1981). Aguirre v. Chula Vista Sanitary Service, 542 F.2d 779, 781 (9th Cir. 1976). Because the court finds that plaintiffs have met their burden under the second of the two standards, an injunction will issue. This memorandum will constitute the findings of fact and conclusions of law which support the issuance of the injunction.
Mailhandler is one of the four major entry level hiring positions in the Postal Service. A mailhandler unloads and moves mail in bulk, and does some general sorting. More than 90 percent of the mailhandlers in the Golden Gate District are employed in San Francisco, Oakland, and the Bulk Mail Center in Richmond, California.
Test 450 is given in each separate post office as the need arises. It consists of three parts: Part A address checking; Part B following oral directions; and Part C word meaning. To be eligible for the position of mailhandler, an applicant must score at least 16 points (out of 32) on Part C *803 and achieve a combined score of at least 60 on Parts A and B. If the applicant meets these requirements, his or her score is translated onto a scale of 70 to 100. This point total is what then is referred to as the applicant's "basic" score. Veterans preference points are added to this "basic" score to reach the applicant's "final" score. A hiring register is then prepared, listing candidates in the order of their final point scores. Candidates are called, in the order they appear on the register, for a physical examination and an interview. While the defendants formally adhere to a "rule of three" hiring procedure whereby the next three persons on the hiring register are considered for any available opening the plaintiffs indicated that as a practical matter jobs are offered to candidates in the order in which their names appear on the hiring register. See Plaintiff's Memorandum of Points and Authorities in Support of Motion for Preliminary Injunction Re Test 450, filed December 19, 1980 ("Plaintiff's Memorandum"), at 3. Defense counsel did not dispute this contention, see Defendants' Opposition to Plaintiffs' Motion for Preliminary Injunction of Test 450, filed January 27, 1981 ("Defendants' Opposition"), at 6; and given the semi-skilled nature of the job, it would seem logical that hiring would proceed on that basis.
Under Title VII, a violation may be proved by showing either "disparate treatment" in hiring, or "disparate impact" of a facially neutral hiring procedure. See International Brotherhood of Teamsters v. United States, 431 U.S. 324, 335 n.15, 97 S. Ct. 1843, 1854-1855 n.15, 52 L. Ed. 2d 396 (1977). The claim made in support of this motion for a preliminary injunction is that Test 450 has a disparate impact upon Hispanics. To make a prima facie showing under such a claim, the plaintiffs must show that the test "select[s] applicants for hire in a significantly discriminatory pattern." Dothard v. Rawlinson, 433 U.S. 321, 329, 97 S. Ct. 2720, 2727, 53 L. Ed. 2d 786 (1977). The defendants may then rebut that prima facie showing by showing that the test is an accurate barometer for job performance that it is "job related." This process is commonly referred to as "validation." Id.; Albermarle Paper Co. v. Moody, 422 U.S. 405, 425-36, 95 S. Ct. 2362, 2375-2381, 45 L. Ed. 2d 280 (1975). Even if the defendants make such a showing, the plaintiffs may then prove that there are other methods of selection which would not have such a discriminatory impact and yet would still provide a screening device which would "serve the employer's legitimate interest in `efficient and trustworthy workmanship.'" Albermarle Paper Co. v. Moody, supra, 422 U.S. at 425, 95 S. Ct. at 2375, quoting McDonnell Douglas Corp. v. Green, 411 U.S. 792, 801, 93 S. Ct. 1817, 1824, 36 L. Ed. 2d 668 (1973); Dothard v. Rawlinson, supra, 433 U.S. at 329, 97 S. Ct. at 2727.
The great difficulty in this case is deciding on the method by which Hispanics are to be identified. Plaintiffs' statistical case depends on the use of data based on self-identification. The defendants wish to rely on data which is compiled by using a list of Spanish surnames. Both methods present some difficulties. The court need not resolve the issue of which of these methods is the proper or preferred one at this juncture of the case. We need only be convinced that the plaintiffs have made a sufficient statistical showing to meet their burden of demonstrating a sufficiently serious question on the merits.
*804 The main difficulty with relying on self-identification data is that the Postal Service was prohibited from collecting such data prior to October of 1980. As a result, the statistical evidence introduced by the plaintiffs showing adverse impact is based only on sittings of Test 450 from October through December 1980. Nonetheless, the results from that period are very significant. Well over 1000 persons sat for the exam, a sample large enough to produce reliable statistics. While the defendants made one unsupported suggestion that an influx of Berkeley students might have skewed the data, most of which apparently comes from the Richmond Bulk Mail Center, see Defendants' Opposition at 12, this court finds nothing in the record to suggest that this data would not be a reliable preliminary indicator of any adverse impact, especially in light of the showing which plaintiffs must make at this stage.
That data does provide evidence of a significant statistical disparity in the success rates of Hispanics and white non-Hispanics. Hispanic applicants passed the test at a rate of 59.5 percent, while white non-Hispanics passed at a rate of 87.1 percent. The disparity grows significantly greater when one looks at the respective rates at which Hispanics and white non-Hispanics achieved "basic" and "final" scores of 80 or more. 50.3 percent of white non-Hispanics, but only 18.3 percent of Hispanics, achieved "basic" scores of 80 or more. 57.4 percent of whites, but only 20.6 percent of self-identifying non-white Hispanics, achieved "final" scores of 80 points or more.
The exact standard for evaluating the adverse impact of a facially neutral testing procedure has never been firmly fixed in the law. Plaintiffs suggest that the court apply the test set out in the Uniform Guidelines on Employee Selection Procedures, 29 C.F.R. Part 1607 (1978), which were developed by the Equal Employment Opportunity Commission. ("E.E.O.C."). The defendants do not seem to dispute the applicability of these guidelines, and the Supreme Court on more than one occasion has indicated that these guidelines are entitled to "great deference" because they are "[t]he administrative interpretation of the Act by the enforcing agency." Griggs v. Duke Power Co., 401 U.S. 424, 433-34, 91 S. Ct. 849, 854-55, 28 L. Ed. 2d 158 (1971); Albermarle Paper Co. v. Moody, supra, 422 U.S. at 431, 95 S. Ct. at 2378. Those guidelines set out a "four-fifths" test for evaluating adverse impact:
A selection rate for any race, sex or ethnic group which is less than four-fifths ( 4/5 ) (or eighty percent) of the rate for the group with the highest rate will generally be regarded by Federal enforcement agencies as evidence of adverse impact ....
29 C.F.R. 1607.4(D). Alternatively, the plaintiffs point to Hazelwood School District v. United States, 433 U.S. 299, 308-09 n.14, 97 S. Ct. 2736, 2742 n.14, 53 L. Ed. 2d 768 (1977) for another standard for measuring adverse impact. The Supreme Court, quoting from its earlier decision in Castaneda v. Partida, 430 U.S. 482, 496-97 n.17, 97 S. Ct. 1272, 1281 n.17, 51 L. Ed. 2d 498 (1977), suggests that a statistical disparity of more than two or three standard deviations between the expected and actual number of black teachers on the school staff there at issue would provide substantial evidence that the hiring procedure was not racially neutral. Plaintiffs here assert that this court may apply the same standard to determine if the statistical disparities in pass rates on Test 450 is significant enough to make a prima facie case under Title VII.
The court does not need to decide here which of these two standards should be used, or whether some other method of *805 evaluating adverse impact should be applied. Under either one of the standards suggested, these statistics indicate that Test 450 has an adverse impact on Hispanics. Affidavit of Richard Drogin, filed December 19, 1980, at ¶ 7 (figures reflect a discrepancy of more than five standard deviations). The disparities are so great that under any reasonable concept of adverse impact, these statistics would support a finding that there was such an impact.
As noted earlier, disparity is particularly great in the statistics for those with "basic" scores of 80 points or more, and those with "final" scores of 80 or more. Plaintiffs assert that a large majority of those hired have scores of 80 points or more. They assert that 70 percent of those who are called in for interviews have basic scores of 80 or more, and that 90 percent of those called in have final scores of 80 or more. They contend this is true because the postal facilities usually give Test 450 again and establish a new hiring register before the old register is fully exhausted.
Defendants dispute the extent of this practice. They have introduced studies which indicate that only 64.6 percent of those who are presently working as mailhandlers achieved basic scores of 80 points or more, and that more than 50 percent of the career mailhandlers in 11 southern states had basic scores on Test 450 of less than 80 points. See Affidavit of Robert E. Anneser, filed January 27, 1981 (attached as Exhibit # 2 to Defendants' Opposition) ("Anneser Affidavit"), at ¶ 6; Affidavit of George H. Butler, filed January 27, 1981 (attached as Exhibit # 1 to Defendants' Opposition) ("Butler Affidavit"), at ¶ 8. Although defendants have not disputed that the hiring registers are sometimes not exhausted before a new sitting of the test is given, counsel asserted at oral argument that it is impossible to ascertain at this time under what circumstances and with what frequency the individual post offices give a sitting of Test 450 prior to exhausting the hiring register.
Given the nature of the preliminary relief which the court is today ordering, it is not necessary for us to determine the regularity at which hiring is cut off above the passing score of 70 points. To the extent that hiring is cut off above the 70 point level, and this greater adverse impact occurs, the preliminary relief will eliminate it. To the extent that hiring would in any case continue until the list was exhausted, the preliminary relief ordered will have no impact on the defendants' procedures and no harm to them will occur.
Defendants did not materially dispute the plaintiffs' statistical analysis of the racial and ethnic self-identification date. Defendants' Opposition at 12; Affidavit of Richard C. Singleton, filed January 27, 1981 (Exhibit # 3 to Defendants' Opposition) ("Singleton Affidavit") at Table 3. Instead, they argue from a different set of data, based on the use of lists of Spanish surnames compiled by the United States Bureau of the Census. The statistics generated by this method indicate that 8.5 percent of those who took Test 450 from 1974-79 were Spanish surnamed, 8.4 percent of those who passed the test during that period were Spanish surnamed, and 8.9 percent of those who achieved a basic score of 80 points or more were Spanish surnamed. The statistics introduced by the defendants also indicate that in the last couple of years, the percentage of persons who passed Test 450 and who achieved a basic score of 80 points or more who were Spanish surnamed exceeded by a large proportion the percentage of applicants who were Spanish surnamed. Defendants' Opposition at 23; Singleton Affidavit at 1-2, Tables 1 and 2.
Defendants also rely on what can be called a "bottom line" defense. They assert that, whatever the impact of Test 450, the hiring process has resulted in the non-discriminatory hiring of Hispanics, as reflected in the statistics on actual hiring. They point out that the percentage of people *806 hired with Spanish surnames averaged 8.7 percent during the period 1974-79, while the applicant rate averaged 8.5 percent Spanish surnamed persons. The figures for the last couple of years are also relatively favorable using this approach: in 1978, 9.1 percent of the applicants were Spanish surnamed, while 8.0 percent of those hired for career positions had Spanish surnames; in 1979, the figures were 12.1 percent and 11.5 percent, respectively. Finally, the defendants argue that 7.3 percent of the present career mailhandlers at the three main facilities are Spanish surnamed, which compares favorably with the applicant flow over the past five years. Defendants' Opposition at 24-26; Singleton Affidavit at 3, Tables 1, 5; Affidavit of Robert N. Davidson, filed January 27, 1981 (Exhibit # 4 to Defendants' Opposition) ("Davidson Affidavit"), at 2.
Defendants defend the use of the Spanish surnames list as a method of identifying Hispanics essentially on three grounds: (1) the widespread reliance on "Spanish-surnamed" as a method of identifying Hispanics in Title VII cases; (2) the use of the Spanish surname list by the Bureau of the Census; and (3) the indication that self identification statistics show that approximately two-thirds of those identifying themselves as Hispanic had Spanish surnames, and the number of those who were Hispanic who did not have Spanish surnames was almost equal to the number of those with Spanish surnames who did not identify themselves as Hispanic. Defendants' Opposition at 22-23; Singleton Affidavit at Table 4.
However, plaintiffs point out a number of difficulties with the use of Spanish surname lists which place the probative value of defendants' statistics in great doubt. Plaintiffs correctly point out that defendants' statistical comparison of success rates on the test compares Spanish surnamed to all others, rather than to the most successful group, which appears to be white non-Hispanics. Of course, it is impossible to make such a comparison for the period prior to October 1980, since there is no way to identify whites in the pool of applicants. Plaintiffs also point out the high discrepancy between the Hispanic group and the Spanish surname group. Their statistical analysis indicates that, based on the self-identification data collected since October 1980, only 58.5 percent of the Hispanics had Spanish surnames, and only 48.3 percent of the Spanish surnamed individuals identified themselves as Hispanic. Affidavit of Dr. Richard Drogin, filed February 2, 1981 ("Drogin Affidavit"), at ¶ 2. The examination of the self-identification data also confirms in part plaintiffs' contention that the large number of Portugese and Filipino residents of the San Francisco Bay Area, groups which are not included in the definition of Hispanics being relied on in this litigation, is responsible for the perhaps unusually large discrepancy between the group of self-identified Hispanics and persons of Spanish surnames. The analysis indicates that almost 51 percent (192 out of 377) Spanish surnamed persons identified *807 themselves as Asians or Pacific Islanders. Only 14.6 percent of those (28) considered themselves to be Hispanic. Drogin Affidavit at ¶ 3.
Plaintiffs also point out, without dispute from the defendants, that the defendants overstate the reliance of the Bureau of the Census on Spanish surname lists. The evidence in the record to date indicates that the Bureau of the Census relies on self-identification as the primary method of identifying Hispanics. See Deposition of Jeffrey S. Passel, dated August 18, 1980, at 7-9 (Exhibit # 9 to Plaintiffs' Memorandum). The Spanish surname list is apparently used as a method of identification in the southwestern states. Plaintiffs' Reply Memorandum in Support of Motion for Preliminary Injunction, filed February 2, 1981 ("Plaintiffs' Reply Memorandum"), at 10. Given the large population of Mexican-Americans in that part of the country, such an approach would presumably result in a closer approximation to the actual number of Hispanics than would its use in the San Francisco Bay Area, where the relative number of non-Hispanic Spanish surnamed Americans is likely to be much higher.Id. at 9.
Moreover, while many Title VII discrimination suits have relied on Spanish surnames as an identifier for evaluating adverse impact and for affecting relief, neither side in this action was able to cite any cases in which the courts have considered this issue when there was a basis for believing that the use of Spanish surnames was not an accurate barometer. This neglect is, of course, understandable, since in many areas of the country the demographics are not such that the discrepancy is likely to be very large. But this court must take note of the rather unusual circumstances in our area which cast doubt on the usefulness of evaluations based on Spanish surname lists.
The court need not nor does it intend to settle this issue now. We only note these problems in connection with our evaluation of whether the plaintiffs have presented sufficient evidence to meet the standard for a preliminary injunction. The problems with the use of Spanish surname lists undermine the statistical rebuttal made by the defendants and so increase, on balance, the relative weight to be given to the plaintiffs' statistical showing based on the self-identification data.
Besides the problems with the use of Spanish surname lists, defendants' arguments which are based on a "bottom line" defense do not carry much weight at this point in the litigation. First, it is uncertain whether such a defense is available under any circumstances under the applicable law. The E.E.O.C. guidelines indicate that, as a matter of prosecutorial discretion, the executive branch is not going to go after employers who are able to demonstrate that the total selection process is non-discriminatory, even if parts of the process seem to have adverse impact. 29 C.F.R. § 1607.4(C). The summary and overview which accompanied the Guidelines, however, make it clear that this prosecutorial decision was not meant to suggest whether an employer could be held liable for such adverse impact under Title VII. The answer to the legal question ultimately depends on whether Title VII is interpreted to be primarily concerned with protecting individuals or in protecting racial and ethnic groups. The summary also indicates that the guidelines were written with the thought in mind that individuals would still have recourse to the courts for denials of jobs which result from the adverse impact of one portion of a selection process. 43 Fed.Reg. 38290-38291 (1978); see also League of United Latin American Citizens v. City of Santa Ana, 410 F. Supp. 873, 894-95 (C.D.Cal.1976).
*808 Second, the defendants, despite the repetition of this defense on a number of occasions, have not to date suggested any other element of the selection process which would account for this alleged counter-balancing of the adverse impact of Test 450. The only evidence in the record indicates that the test results determine the hiring registers from which virtually everybody is offered a job in the order in which their names appear. Without some indication of what other process the defendants engage in that would counter the impact of the test, the mere fact that some statistics indicate that for some unexplained reason the percentage of Spanish surnamed individuals hired equals the percentage who apply would not provide the basis for ignoring the adverse impact of the test, even under the E.E.O.C. guidelines, for the defendants have not demonstrated anything in the "total selection process" to account for the claimed counter-balance. The court has been able to come up with only two possible explanations on its own for these statistics: (1) data based on the use of Spanish surname lists are unreliable, or (2) Hispanics, however identified, accept jobs when they are offered in greater percentages than members of other ethnic or racial groups. If the latter is the case, then the selection process is still working in a discriminatory manner: jobs are still being offered to Hispanics in lower percentages than would occur absent a discriminatory testing procedure. They are simply being accepted at a greater rate. Absent a discriminatory testing procedure, there would be even more Hispanics who would be working as mailhandlers.
Again, the court does not mean to imply that it is ready to make a definitive ruling on any of these issues; that would be premature at this point. But consideration of these difficulties with the defendants' rebuttal of the plaintiffs' evidence must figure strongly in the court's consideration of whether the standard for a preliminary injunction has been met.
The court, in evaluating whether there has been a showing by the plaintiffs of a substantial question on the merits, also must take into account any evidence concerning the likelihood that defendants will be able to rebut a prima facie statistical showing of adverse impact by proving the job relatedness of the test. If there is evidence that the defendants would be able to show validity, that would undermine the importance of plaintiffs' ability to show adverse impact. On the other hand, if there is evidence that the likelihood of proving validity is in doubt for any reason, that would further support a holding that a substantial question has been raised by the plaintiffs. In this case, the court finds that there is some evidence that the defendants are likely to have a difficult time proving the validity of the 450 test.
First, defendants concede that there is no evidence available which explains how Test 450 was developed or even any information available describing its origins. This inability to describe the conditions under which the test was developed can only increase the burden on the defendants to provide some evidence now demonstrating the test's current validity. As the Court of Appeals for the Second Circuit noted,
since insufficient spadework usually results in a poor garden, evidence of unsatisfactory *809 preparation imposed upon the defendants a heavier burden of demonstrating that they had created a satisfactory job-related examination.
Kirkland v. New York State Department of Correctional Services, 520 F.2d 420, 426 (2d Cir. 1975). While there is no evidence showing "insufficient spadework" in the development of this test, the inability of the defendants to explain its origins or its developments can only increase the court's concern about the likelihood of showing validity.
The defendants concede that, at best, there is presently no proof that the test is valid, Defendants' Opposition at 30, and the court has been unable to glean any evidence from the record at this point which can really be said to support validation. Anneser Affidavit at ¶ 8 (CVS study "was not intended to and does not show either validity or invalidity.").
There is, moreover, at least a suggestion in two partial validity studies that proof of validity will be difficult. A 1971 study concluded, in part, that
it is difficult to assume that either examination [the Clerk/Carrier or the Mailhandler exams] has any predictive value; measures what it purports to measure; that an examinee would score the same on repeated administrations of the examination; or that different scores on the examination indicate true differences in individuals.
Tiedemann, An Evaluation of the Pilot Recruiting and Selection Program for Distribution Clerks, Letter Carriers and Mail Handlers (August 13, 1971) at 3 (Exhibit A to Plaintiffs' Reply Memorandum). The report also concluded that the test did not show sufficient validity to recommend its use over a pilot procedure also being tested. Id. at 1-2. The report concluded that none of the "objective" comparisons of success on this test of job performance that were conducted as part of this study showed even marginally acceptable positive correlation. Id. at 5. And it concluded, from its evaluation of reports from supervisors, that
[t]here is also a general belief among personnel people that the higher score on the C.S.C. exam does not necessarily mean the applicant has greater aptitude or potential for postal work.
Id. at 6.
Another study of Test 450 was done, entitled Concurrent Validation Study ("CVS"). That report seemed to conclude that there was "no evidence of validity with regard to Part C of Test 450." CVS at 20 (Exhibit # 1 to Plaintiffs' Memorandum). The study also indicated that the slight positive correlation between success on the combined score of Parts A and B and job performance was "too low to be of any practical use." Id. at 18; see also Id. at 24.
Both sides hotly dispute the significance of these partial validity studies, particularly the CVS study. See Defendants' Opposition, at 18-21; Anneser Affidavit, supra, at ¶¶ 7-12; Affidavit of William C. Burns, filed February 2, 1981, (attached to Plaintiffs' Reply Memorandum) ("Burns Affidavit"), at ¶¶ 6-7. The court concludes, after reviewing the relevant evidence, that these studies provide at least some additional reason to believe that validation will be difficult.
This concern is buttressed by the conclusions of William C. Burns, whose expertise and experience in the evaluation of these sort of tests has been relied on by this court in the past. See Plaintiffs' Reply Memorandum at 13. Based on his review of the two partial studies, the nature of the mailhandler position and his own expertise, he has concluded,
there is no shred of evidence that Test 450 is valid. I also conclude that the probability that this test is or can be shown to be job-related is vanishingly small.
*810 Most private employers eliminated these kinds of tests shortly after the Griggs decision was handed down in 1971. They either tried to validate them and failed or they were told that it was not a good use of resources to make the attempt and simply stopped their use. They have experienced no real difficulty in establishing nondiscriminatory alternatives in their place.
Burns Affidavit, supra at ¶ 9. While the defendants' expert has expressed an expectation that they will be able to validate this test, see Anneser Affidavit, supra at ¶ 13, the conclusion reached by an expert such as Mr. Burns must carry weight in this court's preliminary evaluation of whether plaintiffs have presented a substantial question such as would support the issuance of a preliminary injunction.
Considering all the evidence presented in connection with this motion, the court finds that the plaintiffs have met their burden to establish a sufficiently serious question on the merits to provide a fair ground for litigation. The statistical evidence produced by the plaintiffs to date, although limited, when compared to that presented by the defendants, which is burdened by many difficulties, provides the necessary foundation, supported by the evidence which strongly suggests that the defendants will have great difficulty rebutting any prima facie statistical case by showing that Test 450 is job related.
The court also finds that the balance of hardships tips decidedly and sharply in favor of the plaintiffs. There is, first, the simple loss of income, a loss for which it will be impossible, as a practical matter, to award full compensation at the end of trial. As the plaintiffs have noted, it is always difficult in a class action of this nature to locate the individuals at the end of litigation, and it is often difficult for them to establish accurately the amount of individual damages. And, unlike with private employers, it will be impossible for the court to award anything but straight backpay to individuals found to have been discriminated against. Saunders v. Claytor, 629 F.2d 596, 598 (9th Cir. 1980) (absent express statutory authorization, interest, sick leave or vacation adjustments, available against the private employer, cannot be awarded against the United States). Moreover, the loss of work experience, training, the continued denial of the ability to earn a living and support a family, and the simple loss of human dignity which results from deprivation of employment because of discrimination are all factors tipping the balance of hardships heavily towards the plaintiffs. See, e. g., Manhart v. City of Los Angeles, Department of Water and Power, 387 F. Supp. 980, 984 (C.D.Cal.1975) (these factors sufficient to support a finding of irreparable injury).
On the other hand, it is hard to identify any real harm that will befall the defendants by the issuance of the type of preliminary relief which the court is today ordering. The defendants are simply ordered to exhaust their hiring registers before giving the 450 test again. This means that they will simply be required to hire those who they have already identified as passing the test. There is certainly no evidence that those achieving a minimum passing score cannot handle the job fully and capably. Certain veterans and current employees are currently given priority in hiring for mailhandler positions as long as they pass Test 450. There is no evidence that these employees do not perform adequately. While the defendants' expert has expressed an expectation that the defendants will be able to show that relative rank on the test reflects relative productivity, see Anneser Affidavit, supra, at ¶ 13, there is presently no evidence to support such a claim. Moreover, the evidence which has raised doubts about the defendants' ability to prove the job-relatedness of this test certainly raises even greater skepticism about the likelihood that defendants will ever be able to show a significant relationship between relative ranking on the test above a passing grade and capability to perform the job. This is particularly true when one considers that *811 the defendants' counsel indicated that there is no systematic cut-off point in hiring above the passing grade now, so that it is likely that people are presently hired, with some degree of regularity, from all grade levels above passing. It is also hard to see how defendants will be harmed by this form of preliminary relief after their efforts to assure the court that hiring is not systematically cut off at 80 points, and that many present employees scored below 80 points on the test. Defendants' Opposition at 10-11. If that is indeed the case, the injunction will impose little change, let alone any hardship.
This form of relief will not disrupt in any way the use of Test 450, nor will it place on the defendants any significant administrative or substantive burdens. It will also allow more self-identification data to be collected between now and the time of trial in this case, which should be extremely useful in ultimately deciding the case on the merits. At the same time, while it will not totally eliminate any discriminatory impact of Test 450, which may exist even when all those who pass are given an opportunity to come in for interviews and physical examinations, it will, even according to the data presented by the plaintiffs, help to eliminate in the interim prior to trial the greater discriminatory impact which would seem to occur if hiring is cut off at a level above a passing grade. This is one of the alternative forms of interim relief proposed by plaintiffs, and counsel indicated at the hearing on the motion that such a form of relief would satisfy their request. Under all the circumstances of this case, given the showing that has been made to date, the difficulties which remain in the evidence adduced on both sides, and the relative burdens and difficulties which would be produced by the various alternative forms of preliminary relief which have been proposed, it is the conclusion of this court that this form of injunction is most appropriate.
Defendants are not to misinterpret this preliminary relief as an endorsement of the continued use of Test 450. By choosing this route for preliminary relief, this court is not, in any manner or means, providing the defendants with any protection from any liability which may arise from the continued use of this test. The defendants are free at any time, if they so choose, to cease relying on Test 450 for selecting mailhandlers. They may, for that matter, make any other adjustment in their hiring procedures which they choose, either in order to prevent any further potential liability for new applicants subject to discrimination or for any other reason. The court's action today decreases the defendants' potential liability in so far as it will result in some Hispanics being offered employment who might not otherwise be given a job. That is the full extent and effect on the defendants' potential liability which arises from the issuance of the preliminary injunction. The defendants remain fully and absolutely responsible to any individual who is later found to be discriminated against by the use of Test 450, either prior to or after the issuance of this injunction.
The defendants are hereby enjoined from reopening the hiring registers for the position of mailhandler at any facility prior to exhausting the hiring register for that facility. In order to insure the smooth operation of the hiring procedure, the defendants may conduct a new sitting of Test 450 prior to the time that the hiring register is actually completely exhausted. However, the results of those examinations may not be added to the hiring register for use in selecting applicants before all those who are already on the hiring register have been considered. The purpose of this injunction is to assure that all those who are placed on the hiring register by virtue of having achieved a passing score on Test 450 are considered for employment, rather than the reopening of hiring registers before the list is exhausted, which has apparently resulted in some instances and with some regularity in those with scores under 80 points not being considered for employment.
The plaintiffs have also requested that the defendants be ordered to include in their validation studies consideration of alternatives to the tests in question which *812 might have less adverse impact on Hispanics and which would still provide a screening device related to job performance. The court is not prepared to do so at this time. The case law indicates quite clearly that if the defendants are successful in countering a prima facie statistical showing of adverse impact by demonstrating job relatedness, the burden is then on the plaintiffs to show that there are suitable alternatives with less of an adverse impact. Albermarle Paper Co. v. Moody, supra, 422 U.S. at 425, 95 S. Ct. at 2375; Dothard v. Rawlinson, supra, 433 U.S. at 329, 97 S. Ct. at 2727. To order the defendants to include such considerations in their validation studies would in effect shift the burden of proof. Plaintiffs point out correctly that the E.E.O.C. guidelines require that validity studies conducted in accordance with those guidelines include consideration of alternative selection procedures which have less adverse impact. 29 C.F.R. § 1607.3(B). Whatever weight might be given the failure to follow that instruction in the consideration of the value of the validity studies presented in evidence or in consideration of suitable relief after a finding of liability, this court does not find that this statement in the E.E.O.C. guidelines should be enforced by injunction, particularly in the face of the case law distributing the burden of proof differently. Defendants are presently conducting these validity studies as an element of preparation of their case; they have made it quite clear that they are not ready to concede the threshold issue of the adverse impact of Test 450. Given the preliminary nature of the consideration of the evidence today by the court, and given that the court is only prepared to find today that the plaintiffs have shown that there is a substantial question worthy of litigation, it is inappropriate for the court to use its injunctive power to enforce this instruction which is part of these regulatory guidelines.
Plaintiffs have also requested that some form of notice of this pending case be sent to all those who take Test 450 and who identify themselves as Hispanic. Since the defendants will send notification to all applicants informing them whether or not they passed the examination, it would seem to cause no additional burden for a notice to be included in those letters going to Hispanics. This court is well aware, however, that costs of notice of a class action are not to be shifted to the defendant at a preliminary stage of the proceedings. Eisen v. Carlisle and Jacquelin, 417 U.S. 156, 177-79, 94 S. Ct. 2140, 2152-53, 40 L. Ed. 2d 732 (1974). Therefore, the court will require any such additional costs to be borne by the plaintiffs. At the least, this will mean that plaintiffs will be responsible for having the notice printed. They will also be responsible for any additional costs for postage and processing which the defendants can demonstrate to this court will be incurred by including this notice in the mailing to applicants.
Plaintiffs also raised this issue of notice for the first time only a few days prior to the hearing on the motion. The defendants have therefore quite properly requested an opportunity to consider the request more carefully and to file a written response to the request. Therefore, while the court is generally inclined to grant the request within the bounds discussed supra, it will hold off making a final determination until defendants have had an opportunity to file their response. Therefore, the court is hereby ordering that plaintiffs file with the court, no later than May 8, 1981 a proposed form of notice which will be sent to these Hispanic applicants. Defendants shall file, no later than May 8, 1981 any response they wish to make to this request. They shall also file, at the same time, an itemization and explanation of any costs which they claim they would incur from the inclusion of this notice in the mailing. Any such claims must be substantiated by affidavits and other evidence sufficient to convince the court that these are reasonable costs which will be incurred solely because of the addition of this notice in the mailing. Each side shall then have until May 15, 1981 to file any papers in response to the submissions made in this connection by the other party.
Finally, the plaintiffs have asked that they be furnished with periodic reports on hiring and projections of future hiring by *813 the Postal Service in the Golden Gate District. This request was also raised by the plaintiffs for the first time shortly before the hearing on this motion, and it was not clear that some informal arrangement could not be worked out between the parties on this point without the intervention of the court. The court therefore will not enter any order on this issue, but urges the parties to attempt to reach an accommodation among themselves so that plaintiffs will be reassured that they are being kept abreast of any significant fluctuations in actual hiring and projected hiring. If such an accommodation cannot be reached, the plaintiffs may then renew their request for an order from the court, and it will be briefed and heard in accordance with the normal local procedures for motions in civil cases.
 There are two types of employees who can be working at any one time as mailhandlers for the Postal Service. Besides "career" mailhandlers, who have been hired for permanent, full-time positions, there are "casual" employees. These "casuals" are hired for vacation relief, other temporary needs, and the Christmas season. By contract with the postal employees' union, the Postal Service may only hire any person as a "casual" for only two terms, each of which cannot last for more than 90 days, during each calendar year, and one additional period, not to exceed 21 days, at Christmas time. See Declaration of George H. Butler, filed January 27, 1981 (attached as Exhibit # 1 to Defendants' Opposition to Plaintiffs' Motion for Preliminary Injunction of Test 450) ("Butler Declaration"), at ¶ 14. All the employment figures used in this opinion refer only to career mailhanders.
 Qualified veterans and certain other persons receive special preference by requirement of law. Depending on their status, veterans may have five or ten points added to their score. "Compensable" veterans who pass the examination receive an absolute preference, i. e., they are automatically placed at the top of the hiring register. See Butler Declaration at ¶ 10-¶ 13.
 Plaintiffs have defined "Hispanic" persons to include "persons of Mexican, Puerto Rican, Cuban, Central or South American or other Spanish origin or culture, regardless of race." See Defendants' Opposition at 16. This is the same definition used by the E.E.O.C. in its guidelines, 29 C.F.R. § 1607.4(B), and is the same definition used by the Postal Service in collecting self-identification data since October of 1980. See Butler Declaration at ¶ 18.
 The Postal Service asks two questions, one to identify race and the other to identify ethnicity. The applicant chooses under the race question "White," "Black," "Asian or Pacific Islander," or "Other." The applicant also identifies himself as either of "Hispanic Origin" or "Not of Hispanic Origin." The definition of Hispanic provided is the same definition being used by the plaintiffs in this action to define the class, see supra note 3. This double identification is the reason why Hispanics are compared to a sub-group of "White non-Hispanics." See Butler Declaration at ¶ 18.
 The term "affidavit" is used interchangeably in this opinion for both verified declarations and affidavits.
 At the hearing on this preliminary injunction, defendants raised another slightly different version of their "bottom line" defense. They argued that the pass rates for Spanish surnamed applicants on the tests given since October 1980 are about the same as the pass rates of the self-identified Hispanics. Yet the figures indicate, defendants assert, that despite this "adverse impact" on Test 450, the Hispanics, as identified by use of Spanish surname lists, are being hired at a rate commensurate with the rate of application.
 Defense counsel at one point argued that the success rate of Hispanics should be compared to the success rate of blacks, rather than whites, because at first glance the blacks seem to be the most successful ethnic or racial group, in that they constitute about nine percent of the labor force in this area and yet presently constitute 55 percent of the mailhandlers in the Golden Gate District. Defendants' Opposition at 32-34. However, defense counsel failed to note that the self-identification data collected since last October indicate that approximately 50 percent of the applicants for the position were black. See Affidavit of Richard C. Singleton, filed January 27, 1981 (attached as Exhibit # 3 to Defendants' Opposition), at Table 3. Absent a better showing otherwise, the court agrees with plaintiffs' assertion that the proper comparison under the law is between the success rates of Hispanics and white non-Hispanics.
 Of course, discrepancies caused by the marriage of Hispanics with non-Hispanics would exist in any part of the country.
 In Castro v. Beecher, 334 F. Supp. 930, 934 n.2 (D.Mass.1971), an employment discrimination case, the court takes note of the fact that "Spanish Surnamed" persons did not accurately describe the group of Hispanic persons with which the suit was concerned. Despite this consciousness of the problem, the court went on, without further comment, to rely on statistics concerning "Spanish Surnamed" persons in considering the adverse impact of the testing procedure in question. 334 F. Supp. at 942.
 Plaintiffs have also pointed out that one set of data collected by the defendants indicates that only 5.8 percent of the workforce of mailhandlers in San Francisco between the years 1974 and 1977 were Hispanic, and only 4.0 percent were Hispanic throughout the entire district for those years. Plaintiffs' Memorandum at 18; Affidavit of Richard Drogin, filed December 19, 1980, at ¶¶ 4, 5. This data was collected by the "eyeball method," whereby a Postal Service personnel clerk would code each appointee's papers according to the clerk's guess of the appointee's ethnic/racial classification just by looking at the appointee. See Defendants' Opposition at 14-15, 24. Plaintiffs would, we think, concede that this method of identification is undoubtedly fraught with inaccuracies. It is, of course, impossible to know if it is more or less accurate than the use of Spanish surname lists. In any case, the unreliability of such data and the fact that plaintiffs were relying on it for only secondary evidence in support of their main evidentiary claims leads the court to severely discount the importance of this evidence in evaluating the plaintiffs' arguments in support of their motion.
 Anneser, Concurrent Validation Study of O/N 450 for Selection of Mailhandler 2315-01 (March 1980).
 See note 2.