dissenting:
I agree with the majority that the district court’s judgment in this case cannot stand, for the district court incorrectly placed the burden of persuasion on United States Steel to rebut Jackson’s prima facie case. However, I cannot agree with so much of the majority opinion which holds that the district court properly found that Jackson had established a prima facie case of discrimination in that the company had failed to assist her in finding alternate employment. Accordingly, I would reverse the judgment for Jackson outright, instead of remanding for reconsideration, leaving open the possibility of a new trial.
In the context of Jackson’s claim that United States Steel discriminatorily failed to find an alternative position for her, the elements of a Title VII prima facie case are as follows:
(1) that Jackson belonged to a protected class (women),
(2) that she was qualified for and desired jobs that United States Steel might have had available,
(3) that despite her qualifications, United States Steel failed to look for alternative positions for her, and
*444(4) that United States Steel did look for alternative positions for similarly situated males.
Cf. McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802, 93 S.Ct. 1817, 1824, 36 L.Ed.2d 668 (1973) (failure-to-hire case), quoted in Majority Op., supra, at 440; Scott v. University of Delaware, 601 F.2d 76, 80-81 & n.3 (3d Cir. 1979) (discriminatory discharge case), cert. denied, 444 U.S. 931, 100 S.Ct. 275, 62 L.Ed.2d 189 (1979).
I do not believe Jackson presented sufficient evidence to prove the fourth prong of the prima facie case. To be sure, the district court made findings of fact that this fourth prong was satisfied and that Jackson had established a prima facie ease, App. at 34a-35a, 48a-49a, 51a, but unless those findings are based upon evidence, and I cannot find such evidence in the record, those findings must be set aside as clearly erroneous.
The district court’s opinion demonstrates clearly that the court relied on statistics and comparisons of Jackson’s treatment with various male employees over certain periods of time. Id. at 34a-35a, 48a-51a. Nowhere did the district court purport to rely, as suggested in Majority Op., supra, at 441, on the testimony of Donald R. Patterson, United States Steel’s employment manager. Yet the majority interprets a statement by Patterson as an admission that United States Steel had a policy of placing terminated employees in alternative positions but failed to do so for Jackson. This mischaracterizes Patterson’s testimony. Patterson testified as follows in response to questions asked him by the district court:
Q. When you have the elimination of a position of that kind, don’t you have a policy of attempting to place people by some form of senority [sic]?
A. There is no form of senority [s/c] as far as the exempt and the non-union individuals are concerned. We do attempt to make placement if at all practical within the general area where the individuals are located.
Q. All right. What effort did you make here for placement of this girl who had been with the company for 19 years?
A. We were ready to attempt to place Mrs. Jackson on a job that would come available in the Pittsburgh office if she was interested in it.
Q. No, I am not talking about jobs that would become available. Did you make any effort, whatsoever to compare her availability and experience and so on with any positions that were then filled?
A. No, we did not.
App. at 176a-77a (emphasis supplied). Fairly read, Patterson merely recognized that some attempts at placement are made if practical, and “within the general area where the [particular] individuals are located,” and that such assistance was available to Jackson. Nowhere did Patterson testify that United States Steel had a policy of canvassing all currently available openings whenever an employee was terminated. Nor did the district court apparently understand Patterson to have admitted that such a policy existed, since the district court asked only whether the company had tried to relocate Jackson in a position that was already occupied by someone else. Thus, I do not understand how the majority can hold that this ambiguous testimony satisfies and makes out a prima facie case.
An examination of the statistics and the district court’s discussions of them reveals that they are woefully inadequate to support the inference that there was any company policy to assist males whose jobs were eliminated in finding alternative positions within the company. The district court, without any basis in logic or expert opinion testimony, arbitrarily and artificially selected as the relevant comparison group exempt supervisors, as opposed to all exempt employees; and as the relevant period 1973 to 1976, instead of the entire 1966 to 1978 period covered by the data submitted by Jackson. Moreover, the district court inconsistently cited two nonsupervisory employees’ cases as examples of the company’s policy. Finally, the district court’s choice of “supervisors” as a general class to which Jackson could be said to belong does not comport with its finding respecting Jack*445son’s claim of discrimination during her employment. The latter claim, which arose from her asserted lack of promotion, was denied. The district court found that she was uniquely qualified for her own position but no higher supervisory position or salary. Id. at 39a-46a. The district court’s statistical approach violates a basic canon of statistical analysis — that one should draw conclusions from data and not arrange that data post hoc to justify a particular result.
Because I believe that Jackson failed to present sufficient evidence to establish a prima facie case of discrimination with respect to United States Steel’s failure to find her alternative employment, I would reverse the judgment of the district court outright and direct that judgment be entered for the defendant, United States Steel.1 I therefore respectfully dissent from the remand prescribed by the majority opinion.
. The majority opinion, recognizing the fact that the district court judge who initially heard this case has since died, suggests that there may be an alternative to the holding of a completely new trial. Majority Op., supra, at 442 n. 12. In view of the fact that credibility determinations will be required for the findings of fact made necessary by the majority opinion, 1 seriously question whether there is any alternative to a new trial. However, as the majority opinion points out, this will be a determination to be made by the district court in the first instance.