This case arises out of an action in the district court challenging the legality, under Title VII of the Civil Rights Act of 1964, of the policy of defendant Washington Hospital Center that requires respiratory therapists to pass a written examination as a condition for retaining their jobs. Plaintiffs moved for a preliminary injunction restraining further use of the examination pending resolution of this challenge. Plaintiff's motion was denied, and they appeal.

In the hearing on their motion for a preliminary injunction, plaintiffs introduced evidence that, between 1961 and 1975, 26 white and nine black members of defendant Hospital's Respiratory Therapy Department took the test; and that all of the white employees passed the test but only four of the black employees were able to do so. Plaintiffs also introduced the testimony of a statistician, who stated that the odds of such a disparity in the pass rates of the two groups occurring by chance were four in one thousand. Finally, plaintiffs attempted to demonstrate that the test was not job-related (by evidence that some employees who had failed the test were highly regarded by their superiors; and by testimony that the test was not employed at other hospitals in the metropolitan area).

Defendants adduced evidence designed to show that the test was job-related. This consisted essentially of testimony describing the relationship of the test questions to practical knowledge of the respiratory apparatus, respiratory ailments, and therapeutic maneuvers; delineation of the responsibilities of respiratory therapists; and description of the clinical settings in which they operate.

In its memorandum and order denying plaintiffs' motion for a preliminary injunction, the district court focused its attention upon the statistical data provided by plaintiffs in their effort to establish a prima facie case of discriminatory impact. See Griggs v. Duke Power Co., 401 U.S. 424, 431, 91 S. Ct. 849, 28 L. Ed. 2d 158 (1971). While the testimony of plaintiffs' statistician was uncontroverted, the court held that it "represents too slender a reed on which to rest the weighty remedy of preliminary relief." 431 F. Supp. 873 at 876 (filed May 18, 1977). The court thus appeared to reject, without adequate explanation, the undisputed testimony of a witness who had been qualified by the court as an expert. While the numbers involved may have appeared small, statistical analysis showed them to reflect a discriminatory impact that could not reasonably be ascribed to chance alone.

The action of the district court would be palpably infirm if it were based solely on its rejection of the validity of plaintiffs' statistical evidence.1

But there is language in the court's opinion that suggests that it relied on considerations other than statistical inadequacy. For this reason we remand the case for reconsideration and clarification, as outlined below.

The proper framework for determining plaintiffs' motion is established by Virginia Petroleum Jobbers Association v. FPC, 104 U.S.App.D.C. 106, 259 F.2d 921 (1958) and Washington Metropolitan Area Transit Comm'n v. Holiday Tours, 182 U.S.App.D.C. 220, 559 F.2d 841 (1977) (decided after the action of the district court that is now before us). The court must consider: (1) the likelihood that plaintiffs will prevail on the merits;2 (2) the prospect of irreparable injury to plaintiffs if relief is withheld; (3) the possibility of harm to other interested parties (here, the Hospital) if preliminary relief is granted; and (4) the public interest. Plaintiffs have the burden of establishing a substantial case on the merits. The merits of the case at bar will, of course, turn on whether plaintiffs can show discriminatory impact, and on whether defendants, in turn, can rebut this showing by demonstrating job-relatedness. Griggs v. Duke Power Co.,supra. The burden of demonstrating job-relatedness is on defendants in a proceeding to decide the merits. However, on a motion for relief pending litigation, the court may take into account the possibility that defendant will be able to establish job-relatedness without a determination that it has shown a positive likelihood of doing so. Even as to issues where the burden at trial is on defendant, when the court is of the view that the motion for relief pending litigation (and the attendant change in the status quo ante) threatens harm to defendant, and the defendant has come forward with some evidence tending to support a defense, the court may take into account whether there is a substantial case for the defense without requiring defendant to establish a mathematical probability of success on the merits.

This is a mirror image of the rule of Holiday Tours (Supra note 2) permitting plaintiff to obtain a preliminary injunction on showing a substantial possibility of success. In Holiday Tours, a plaintiff seeking to maintain the status quo may obtain a preliminary injunction on a showing of a substantial possibility of success assuming the decree serves other elements of public interest. The analogue rule provides that a defendant seeking to maintain the status quo may oppose a preliminary injunction on a showing of a substantial possibility of success assuming that maintenance of the status quo serves other elements of public interest.

Applying this analytical framework to the case and decision before us, we note that the court's disposition of plaintiffs' motion may have been based on a variety of considerations other than the factor of size of the employee groups that have taken the test (which was embraced in the expert's testimony). The statistical evidence was relevant only to the question of whether the observed disparity in pass rates as between white and black employees was likely to have been due to chance; it did not address the related but distinct question of how large a margin of error is associated with the observed disparity. The court was concerned with the problem of the effect of a difference in a mere "handful of test results"3 which we discern as a factor that might have dramatically narrowed the disparity in pass rates. The court noted the likelihood that material supplemental data will be available to the court in the trial on the merits.4

Further, we discern that the court may have believed that there was a substantial possibility that at trial defendants would be able to demonstrate job-relatedness. While the court's Memorandum Opinion does not expressly discuss this issue, evidence relevant to it was presented by both parties; and the court did note that the challenged examination is "related to the delivery of critical medical care to gravely ill patients." The court may have been influenced on the issue of job-relatedness by such considerations as the apparent facial validity of the test and the nature of the role played by respiratory therapists in the hospital setting. While practical ability may be dominant, the issue tendered by defendants is the importance, at least on occasion, of facility of communication and command of the language of background and context.

The critical nature of the work of the therapists and the grave condition of their patients also bears on the public interest issue that is material on a motion for the relief that is to be accorded pending the determination of the litigation. That must be weighed alongside the public interest in early termination of discriminatory conditions if the discrimination is not justified.

Because of our inability to clearly discern the basis upon which the district court denied preliminary relief, we remand for reconsideration. Upon remand, the court may take additional evidence and it also has discretion to combine the hearing on the preliminary injunction with a trial on the merits.5 The case may be sufficiently ready for trial, now or in the near future, to warrant proceeding on trial proof rather than speculation as to its content.

Remanded for further proceedings not inconsistent with this opinion.