750 F.2d 867
36 Fair Empl.Prac.Cas. 1289,
36 Empl. Prac. Dec. P 34,953,
17 Fed. R. Evid. Serv. 393
Mary J. MILES, Plaintiff-Appellant,
v.
M.N.C. CORPORATION, Defendant-Appellee.
No. 83-7309.
United States Court of Appeals,
Eleventh Circuit.
Jan. 15, 1985.
Peter H. Martin, Legal Services Corp. of Alabama, Opelika, Ala., for plaintiff-appellant.
Larry E. Forrester, Julie H. Jackson, Daniel S. Bowling, III, Atlanta, Ga., for defendant-appellee.
Appeal from the United States District Court for the Middle District of Alabama.
Before GODBOLD, Chief Judge, JOHNSON and CLARK, Circuit Judges.
CLARK, Circuit Judge:
Mary Miles appeals from an adjudication that defendant-appellee M.N.C. Corporation (M.N.C.) did not discriminate against her on the basis of race in violation of Title VII of the Civil Rights Act. M.N.C. is a company engaged in the manufacture of cardboard boxes in Opelika, Alabama. Mary Miles claims that she was the victim of racial discrimination practiced by M.N.C. when the company did not rehire her after a temporary layoff. Miles started work for Rock-Tenn Company, M.N.C.'s predecessor in interest, on November 7, 1979 and was laid off on December 17, 1979. At the beginning of 1980, M.N.C. replaced Rock-Tenn as the owner of the business. Nick Constan, the major stockholder and president of M.N.C., served as plant manager for Rock-Tenn prior to M.N.C.'s purchase of the company. M.N.C. retained the same work force and managers and produced the same product under the same conditions as Rock-Tenn.1
At the time of Miles' layoff, two other full-time general production workers were also laid off. One of these workers, Lavelle Parmer, was white and the other worker, Sarah Wright, was black. Layoffs were according to seniority and these three women were the junior employees at the plant. Both Parmer and Wright had been employed a few days before Miles. One white male part-time worker, David Kinports, was also laid off. Miles claims that Dave Nichols, the plant supervisor, told her on two occasions that she would be recalled after the first of the year. Around January 23, 1980, Parmer, the white full-time worker, was reemployed and two white male part-time workers were employed. Part-time workers had the same duties as full-time general production workers. In June 1980, Parmer quit her job and was replaced by Karen Wilson, a white worker who had been employed by Rock-Tenn over two years, moved away and then returned to the Opelika area.
After Parmer told Miles about her reemployment, Miles met with Nick Constan in January 1980. According to Miles, Constan told her at that time she would be recalled when work picked up before new employees were hired. Upon hearing about the employment of Wilson in June, Mary Miles spoke with Olin Henderson, plant manager, who told her she would be recalled when business picked up. Miles filed a charge of racial discrimination with the E.E.O.C. on November 1, 1980, received her right-to-sue letter in October 1981 and filed suit in federal court in December 1981.
After hearing testimony and receiving briefs, the district court found against Miles on her claim of racial discrimination. The district court applied a disparate treatment analysis to the facts and found that Miles had made out a prima facie case of discrimination.2 Once the prima facie case is proved, the burden shifts to the employer to articulate a legitimate, nondiscriminatory reason for its acts with regard to the plaintiff. The burden on the defendant is one of production rather than persuasion.3 If the employer carries this burden of production, the presumption raised by the prima facie case is rebutted and the employee must persuade the court that the reasons for not hiring plaintiff offered by the employer were pretextual. Texas Dept. of Community Affairs v. Burdine, 450 U.S. 248, 256, 101 S.Ct. 1089, 1095, 67 L.Ed.2d 207, 217 (1981).
Miles contends that her direct evidence of discrimination, in the form of a racial slur made by Olin Henderson, changed the traditional burden of proof. Miles argues that a defendant cannot refute direct evidence of discrimination by mere articulation of other reasons for its actions. Instead, the defendant must rebut by proving by a preponderance of the evidence that it would have reached the same decision absent the presence of the illegal motive. The district court found that the racial slur evidence was struck from the record. The court held that both Parmer and Wilson were senior to the plaintiff, more experienced and more proficient. The court also noted that at the time the part-time employees were hired there was no need for full-time workers and that the plaintiff never applied for part-time work. The district court found that the statistical evidence lacked probity because it did not show the relative qualifications of black and white workers.
Mary Miles now appeals and claims that the district court made the following clearly erroneous findings of fact: 1) that seniority played a role in determining who was entitled to recall; 2) that Wilson's ability to operate a machine was the reason why she was hired; 3) that the plaintiff did not apply for part-time employment; 4) that the statistical evidence was related to racial composition of the area; 5) that Olin Henderson's racial slur had been stricken from the record; 6) that the plaintiff offered no testimony as to the quality of her work; 7) that Parmer was an excellent block setter. The district court made several clearly erroneous findings of fact and erred with respect to the racial slur. The court then wrongfully concluded that M.N.C. would have reached the same decision regarding Miles even absent the illegal motivation. We reverse.
A court of appeals may not set aside district court findings of fact unless they are clearly erroneous. Fed.R.Civ.P. 52(a); Pullman-Standard v. Swint, 456 U.S. 273, 102 S.Ct. 1781, 72 L.Ed.2d 66 (1982). This standard imposes a heavy burden on the appellant in a case in which the evidence was largely testimonial and the district court had the advantage of observing the witnesses and making credibility determinations. Lincoln v. Board of Regents of Univ. System, 697 F.2d 928, 939 (11th Cir.1983). A finding of fact is reversible under Rule 52(a) only when, after reviewing the entire record, the appellate court is convinced that an error has been made. Wright v. Western Electric Co., 664 F.2d 959 (5th Cir.1981). The review this court conducts inquires into whether a finding has substantial evidence to support it. Lincoln, supra at 939.
Because of the employee's easy burden of establishing a prima facie case and the employer's normal ability to articulate some legitimate nondiscriminatory reasons for its actions, most disparate treatment cases turn on the plaintiff's ability to demonstrate that the nondiscriminatory reason offered by the employer was a pretext for discrimination. B. Schlei & P. Grossman, Employment Discrimination Law 1316-1317 (1983). The same is true of the present case. Three types of evidence can be used by a plaintiff to prove pretext: 1) comparative evidence; 2) statistical evidence; and 3) direct evidence of discrimination, in the form of discriminatory statements and admissions. B. Schlei & P. Grossman, supra at 1314. Miles proffered all three types of pretext evidence in this case. The district court made erroneous findings of fact with regard to each proffer: 1) in its determination that M.N.C. had legitimate reasons for hiring the two white workers in preference to Miles, 2) in its analysis of the statistical evidence offered at trial, and 3) in its decision as to the admissibility of the racial slur. The erroneous findings of fact in these three areas are crucial, and thereby affect the substantial rights of the plaintiff, because each determination made by the district court cut against the plaintiff's showing of pretext.4I. Comparative Evidence
Miles attempted to prove that white workers in a comparable employment position to her own were treated more favorably than she was. The district court found that M.N.C. proved by a preponderance of the evidence that the workers in question could not be compared to Mary Miles. Since the district court offered distinct reasons why Parmer, Wilson and the part-time workers could not be compared to Miles, each rationale should be considered separately.
The district court found that Parmer, the first rehired full-time worker, was senior to Miles and more experienced and proficient. The president of M.N.C. indicated that Mary Miles was not considered a call-back employee because she had not served the ninety-day probationary period. Since Parmer had worked only a few days longer than Miles, it is clear that she, too, lacked vested seniority rights. Therefore, seniority could not have acted as a legitimate reason for recalling Parmer but not Miles. The district court's finding that seniority provided a basis for distinguishing between the two women is important. The only other reason for so distinguishing them, the work performance of each, was based upon subjective evaluation.
This circuit has frequently noted the problems associated with this type of worker assessment and noted that subjective evaluations involving white supervisors provide a ready mechanism for racial discrimination. Parson v. Kaiser Aluminum & Chemical Corp., 575 F.2d 1374, 1385 (5th Cir.1978); Rowe v. General Motors Corp., 457 F.2d 348, 359 (5th Cir.1972); Robbins v. White-Wilson Medical Clinic, 660 F.2d 1064, 1067 (5th Cir. Unit B 1981), vacated on other grounds, 456 U.S. 969, 102 S.Ct. 2229, 72 L.Ed.2d 842 (1982); Harris v. Birmingham Bd. of Educ., 712 F.2d 1377 (11th Cir.1983).5 This is because the supervisor is left free to indulge a preference, if he has one, for one race of workers over another. In addition, subjective and vague criteria may be insufficient reasons given by an employer for its failure to rehire because such criteria do not allow a reasonable opportunity for rebuttal.6 The employee is left without any objective criteria to point to in order to show competence. Lee v. Russell County Bd. of Educ., 684 F.2d 769, 775-76 (11th Cir.1982). All of the worker evaluations at M.N.C. were done by Olin Henderson. Evidence in the case established that there were no guidelines for evaluating performance, written worker evaluations or regular checks done on the employee's work habits. This system appears to have resulted in management having more familiarity with the performance of some workers than others. Miles testified that in the period she worked at the company she never met her supervisor. M.N.C. president Constan testified that he relied on Henderson for his evaluation of Miles' performance. Henderson's evaluation of Miles was that she was not a very good worker. At the same time, Henderson and Constan testified from personal knowledge Parmer was a superior worker who was quite efficient at all aspects of the work. What the court is left with as M.N.C.'s legitimate nondiscriminatory reason for failure to rehire Mary Miles is the subjective evaluation, without more, that she was not a good worker and Lavelle Parmer was. Considering the fact that Miles offered evidence which would tend to show that the evaluation process was tainted, see infra, M.N.C. has not met its burden with respect to Lavelle Parmer.
The district court held that at the time the part-time workers were hired there was no need for a full-time worker and that Miles had not applied for part-time employment. There was no evidence introduced into the record to show that Miles did not apply for part-time work or expressed an aversion for it. The only evidence is that from January until June of 1980 Mary Miles contacted M.N.C. several times about the possibility of working.
II. Statistical Evidence
As another part of her case of pretext, appellant offered statistical evidence to show that there was an overall discrimination in M.N.C. hiring in favor of whites. The district court found that the statistical evidence had methodological problems and, therefore, was not probative. In doing so, the court mischaracterized the evidence. The plaintiff's claim did not rest, as found by the district court, on the fact that M.N.C.'s general production positions were filled by white women in a community where the work force was predominantly black. Instead, Miles offered statistical proof which tended to show that the hiring rate for white applicants, referred to M.N.C. from the Alabama state employment service, far exceeded that of black applicants.7 The statistical evidence introduced in the case compared the racial composition of the pool of qualified applicants with those actually hired. This is one of the appropriate statistical methods for demonstrating intentional discrimination. Hester v. Southern Railway Co., 497 F.2d 1374, 1379 (5th Cir.1974).
The district court indicated that the statistical study lacked probative value because there was no data showing whether the white applicants were more experienced, more qualified or had better experience. At the same time, however, the court correctly noted that there were no specific qualification or criteria for employment at M.N.C. Various employees testified to widely differing past employment experiences. General production work was described as unskilled labor for which there were no particular qualifications. Several of the employees testified that the entire hiring process at M.N.C. consisted of filling out an application and having a very short conversation with the plant supervisor. Evidence introduced at trial showed that M.N.C. submitted job descriptions to the Alabama employment service and that office referred the applicants who were the basis for the statistical study. As a result, the statistics were not invalidated because they lacked variables to assess experience and qualifications.
It is true that the statistical study offered did not encompass all of the 1980-1981 hiring at M.N.C. Plaintiff's expert witnesses admitted that the study only compared the hiring rates between white and black workers referred from the state employment service and that out of the forty-nine persons reported hired during the relevant period only nineteen were referred from the Alabama state employment service.8 The expert witness, however, did examine the racial composition of the pool hired from other sources and found that there was almost a three to one preference of whites over blacks. This information led the expert to conclude that the data available to him would underestimate any discrimination that would exist. Nevertheless, a study of the hiring rates did indicate what the expert characterized as significant discrimination in favor of whites over blacks. It is also true that the statistical study only measured hiring and therefore did not offer any information on rehiring.
However, the Supreme Court recognized in McDonnell Douglas v. Green, 411 U.S. 792, 805, 93 S.Ct. 1817, 1825, 36 L.Ed.2d 668, 679 (1973), that statistics as to a defendant's employment policy and practice might be helpful to a determination of whether the refusal to rehire in a particular case conformed to a general pattern of discrimination against blacks. The Court also held that such general determinations, while helpful, may not control individualized hiring decisions, particularly in the presence of otherwise justifiable reasons for refusing to rehire. 411 U.S. at 805 n. 19, 93 S.Ct. at 1826 n. 19, 36 L.Ed.2d at 679 n. 19. Miles never claimed that the statistical evidence proved her case of disparate treatment, but only that it showed a pattern of favoring whites over blacks at M.N.C. The statistical evidence supports such a claim. The district court was clearly erroneous in finding that the statistical evidence was compared to the general area population and that it was not useful because it did not reflect worker qualifications. The statistical evidence in this case provided a background against which to assess Miles' individual claim.
III. Direct Evidence of Discrimination
Appellant's final offer of evidence to prove pretext consisted of direct evidence of discrimination. Miles introduced a racial slur about the work abilities of blacks, made by Olin Henderson. Confusion has arisen over whether the racial slur was or should have been stricken from the record. Perhaps the best way to clarify the confusion is to quote from the record sequentially the court's rulings on the admissibility of the evidence. Plaintiff called Betty Rogers a white who had gone to work for the employer in 1979 and remained until 1982. She testified that there were no qualifications for the job. She stated that when she applied for the job two or three black women were simultaneously being interviewed for an opening but none of them were hired. The following from pages 82 to 111 are excerpts from the testimony of Betty Rogers and rulings of the court:
Q. Did you ever have occasion to discuss with Olin Henderson during normal business hours his reasons for not hiring black women as employees at the company?
[R. 86].
(Objection and colloquy between the court and counsel).
Q. Mrs. Rogers, do you of your own personal knowledge know who did the hiring of employees at this company in 1982?
* * *
* * *
A. Olin Henderson did the hiring.
[R. 89].
Q. Did you have occasion during the course of your employment during the normal company hours to discuss with Olin Henderson his reasons for not hiring black women as employees at the company?
A. Yes.
Q. What did he say?
[R. 90].
(Objection and colloquy between the court and counsel).
A. I asked Olin Henderson why they didn't have any blacks. He said, "Half of them weren't worth a shit."
[R. 92].
At page 105, the defendant moved to strike the testimony of Betty Rogers with respect to the racial slur on a ground that the evidence had failed to show that Henderson was responsible for hiring in 1980. At page 109, the trial court made the following ruling:
And if Olin Henderson was not in charge of hiring and firing at the time he made the statement, it is not admissible against the company and I will sustain the motion to strike it.
Because of questions propounded to Nick Constan, president of M.N.C., the court reversed his rulings with respect to the admissibility of the racial slur. The following statements appear from the transcript:
THE COURT: Well, it seems to me that you have, by examination--both of you--of this witness proved that Olin Henderson was competent to make the statement he made earlier about the hiring of blacks, and I will let that in.
I will also let in the questions that you asked him, and you may state your questions and I will let him answer it.
[R. 166].
* * *
* * *
THE COURT: All right. I think these ... I think they opened the door for you by proving that Olin Henderson was very muchly in the picture as regards hiring and firing. By that, I think the statement he made about it is admissible.
[R. 167].
Without any further mention of the subject during the trial, the trial court in its final order finding in favor of the defendant ruled that racial slur was not admissible. The following appears from page 285 of the record:
The Plaintiff's claim is largely resting upon the fact that the Defendant's general production positions are largely filled by white women in a community wherein the larger percentage of the work force is black, and the four positions--two full time and two part time--filled by re-employment after the Plaintiff's layoff were filled by white persons. Two white applicants who applied for full-time production work testified that they, along with black applicants, applied for work in full-time production and that they, but not the black applicants, were offered jobs. Neither of these are employed by Defendant. One of these, Ms. Rogers, testified without denial, as Plaintiff's brief points out, that Mr. Henderson was the hiring authority when he said they had no black employees because "they ain't worth a s---". Later, however, Ms. Rogers testified that Mr. Wiggins, not Mr. Henderson, did the hiring and firing at the time the remark was made and that Mr. Henderson had no authority to hire or fire. On motion, the Court struck from the evidence Mr. Henderson's racial slur which cannot be considered against the Defendant because Mr. Henderson had no authority to hire, fire or speak for the Defendant Company. This statement having been excluded, there was no occasion for Mr. Henderson to deny it. There is no evidence of the relative merits of the applicants, but there were no written criteria to guide one entrusted with hiring for the Defendant.
Appellant now complains that the trial court's conclusion that the racial slur was struck from the record is error because it was properly admissible under Fed.R.Evid. 801(d)(2)(D). The issue, therefore, is whether the slur was admissible and therefore properly admitted into evidence. Any determination of admissibility turns on the nature of the supervisory positions held by Olin Henderson throughout the time periods in question. Olin Henderson was assistant superintendent in charge of employees between 1978 and 1979. During this period, he made the racial slur. According to his own testimony, he remained in that position until March 1980 when he replaced Dave Nichols as Production Superintendent. The Production Superintendent appears traditionally to have been in charge of hiring and firing. Nevertheless, there is evidence in the record, from several sources, including Henderson, which leads to the conclusion that the lines of authority with respect to hiring decisions were not clear-cut at M.N.C. It appears that Henderson as employee supervisor, Nichols as Production Superintendent, and Constan as president of M.N.C. all participated in hiring determinations. Cross-examination testimony of Henderson elicited the fact that he hired one or two employees during 1978-79 when the Production Superintendent at that time was away. Henderson testified that he did not know whether he had hiring and firing authority in November of 1979 when Miles was hired and Nichols was still the Production Superintendent. He also testified that he did not know whether he or Nichols rehired Lavelle Parmer when the job became open in January of 1980. Henderson made the recommendation to Constan that Parmer be rehired. According to his testimony, he also in 1980 told Constan, who testified to the contrary, that Miles was a poor worker and that it would not be wise to take her back. As noted earlier, Constan testified that he and Henderson held regular meetings about personnel and other matters. It appears that Henderson was closely involved in the hiring evaluations and decisions from the time he made the statement through and past Miles' tenure at the plant. The trial court so found when holding that Henderson's statement was admissible, supra. Henderson's response to the question about the absence of black workers at the plant was demonstrated by his influence over employment decisions. The statement was made by an agent about a matter within the scope of his employment and therefore was admissible under 801(d)(2)(D). Although the district court noted that the statement was excluded and therefore Henderson had no occasion to deny it, we note that at the time of Henderson's testimony the statement was properly admitted. Henderson never did deny the statement or the opinion behind it.
The district court's post-trial decision to strike the statement was critical to the plaintiff's case. Normally, the defendant in an employment discrimination case would only have to articulate legitimate, nondiscriminatory reasons for its failure to rehire an employee. The plaintiff who always retains the burden of persuasion would then have to show that the employer's reasons were pretextual. The McDonnell Douglas method of proving an employment discrimination case, however, pertains to situations where direct evidence of discrimination is lacking. Bell v. Birmingham Linen Service, 715 F.2d 1552, 1556 (11th Cir.1983); Lee v. Russell County Bd. of Educ., 684 F.2d 769, 774 (11th Cir.1982). Where a case of discrimination is proved by direct evidence, the defendant bears a heavier burden.9 If the evidence consisted of direct testimony that the defendant acted with a discriminatory motive, and it is accepted by the trier of fact, the ultimate issue of discrimination has been proved. The employer can then rebut only by proving by a preponderance of the evidence that the same decision would have been reached even absent the presence of the discriminatory motive. Bell, supra, 715 F.2d at 1557. The inquiry required is one which makes the employer show that it would be the same decision absent the discriminatory motive. Since the district court did not consider the statement admitted, the judge did not make any findings about whether Henderson put aside his bias in the rehiring recommendations and evaluations he gave to Constan. In a similar situation, the court in Bell v. Birmingham Linen Service found that absent a similar finding in that case it was impossible to conclude that Birmingham Linen Service met its burden of proving under a Mt. Healthy-type analysis, that it would have reached the same decision absent discrimination and remanded to the district court to evaluate the evidence under the Mt. Healthy standard. 715 F.2d at 1557. In the present case, however, it appears doubtful that M.N.C. could meet the heavier burden imposed under Bell. The person in charge of making employee evaluations and suggestions for rehiring and the person making the racial slur were the same individual.
Discrimination by M.N.C. in employment practices was proved by 1) Henderson's remark, 2) the statistical evidence, 3) the testimony of Betty Rogers that no blacks worked in production when she was employed in 1978 and Louise Lyles' testimony that she was the only black woman out of 20 production workers when employed in August 1979, and 4) Plaintiff's Exhibit 36 which showed that no blacks were employed in 1980. The district court found the plaintiff qualified. Consequently, we remand to give the defendant an opportunity to prove by a preponderance of evidence that the same employment decision would have been reached absent the presence of clearly discriminatory employment practices.
The judgment of the district court is reversed and remanded.
REVERSED AND REMANDED.
The district court was correct in finding that M.N.C. met the test under E.E.O.C. v. MacMillan Blodell Containers, 503 F.2d 1086, 1094 (6th Cir.1974), of whether a successor corporation is liable for the acts of its predecessor
The court made that determination by applying the Eleventh Circuit modification of the McDonnell Douglas test for discharge cases. According to Lee v. Russell County Bd. of Education, 684 F.2d 769, 773 (11th Cir.1982), the plaintiff must prove that he/she 1) is a member of a protected class; 2) was qualified for the position held; 3) was discharged; and 4) was replaced by a person outside the protected class. Mary Miles proved a prima facie case
As the Court made clear in Texas Dept. of Community Affairs v. Burdine, 450 U.S. 248, 254, 101 S.Ct. 1089, 1094, 67 L.Ed.2d 207, 216 (1981), the defendant does not have to persuade the court that it was actually motivated by the proffered reasons. It is sufficient if the defendant raises a genuine issue of fact as to whether it discriminated
The plaintiff is always trying to show that the employer had a discriminatory motive for the actions it took. In trying to prove pretext, the plaintiff can demonstrate that the proffered reasons were not the true reasons for the employer's decision either by directly persuading the court a discriminatory reason more likely motivated the employer or indirectly by showing that the employer's proffered explanation is unworthy of credence. Burdine, supra, 450 U.S. at 256, 101 S.Ct. at 1095, 67 l.ed.2d at 217
M.N.C. suggests that the principle in Rowe and its progeny applies only to cases where a class of plaintiffs is trying to show that the lack of formal evaluation processes can aid in establishing the existence of a pattern or practice that has a discriminatory impact. This circuit has not limited its concern about the dangers of subjective evaluations to cases involving disparate impact. Robbins v. White-Wilson Medical Clinic, Inc., supra (disparate treatment); Parson v. Kaiser Aluminum & Chemical Corp., supra (discriminatory impact and discriminatory treatment); Harris v. Birmingham Bd. of Educ., supra (appellate court analyzed the claim of one individual--disparate treatment). Furthermore, this court finds no reason to accept the idea that the problems which can stem from subjective evaluations would be any less serious in a disparate treatment case. Discriminatory motive can be expressed just as easily in actions against one individual as it can in an employment practice used against many. This case demonstrates this fact
Plaintiff's witness, Louise Lyles, a co-worker with Miles at the plant, was not allowed to testify as to the quality of Miles' work. Lyles did testify that in early 1980 she asked both Henderson and Constan, separately, when Miles was going to be rehired. They both assured her that Miles would be recalled when there was an opening
The statistical evidence introduced by the plaintiff's expert showed: 1) among all applicants to M.N.C. the hiring rate was 38.7% for whites and 18.6% for blacks; 2) for general production positions the hiring rate for whites was 41.7% compared to 12.1% for blacks; and 3) for female applicants for general production positions (the same situation as the plaintiff) the hiring rate was 40.9% for white females and 12.2% for black females. The expert's conclusion was that the result showed a significant discrimination in favor of whites over blacks
According to the appellant, the statistics were taken from the Alabama employment service referral lists because M.N.C. president Constan testified in deposition that employees at M.N.C. were hired by this method
M.N.C. argues on appeal that size of the statistical pool makes it impossible to draw any reliable inference of discriminatory motivation. While it is true that Dr. Gundlach testified that the larger the statistical pool, the more accurate the conclusions, he also testified that the differences he found in the hiring rate were statistically significant for a sample of that size.
As explained by the court in Lee v. Russell County Bd. of Education:
Where the evidence for a prima facie case consists, as it does here, of direct testimony that defendants acted with a discriminatory motivation, if the trier of fact believes the prima facie evidence the ultimate issue of discrimination is proved; no inference is required. Defendant cannot rebut this type of showing of discrimination simply by articulating or producing evidence of legitimate, nondiscriminatory reasons. Once an unconstitutional motive is proved to have been a significant or substantial factor in an employment decision, defendant can rebut only by proving by a preponderance of the evidence that the same decision would have been reached even absent the presence of that factor.
684 F.2d 769, 774 (footnotes omitted).