39 Fair Empl.prac.cas. 1092,39 Empl. Prac. Dec. P 35,825charles R. Barber, Plaintiff-appellee, v. International Brotherhood of Boilermakers, Iron Shipbuilders, Blacksmiths, Forgers and Helpers,district Lodge No. 57, Defendant-appellant, 778 F.2d 750 (11th Cir. 1985)

George C. Longshore, Birmingham, Ala., for District Lodge.

Robert L. Dameron, Kansas City, Kan., for Dist. Lodge 57.

Carol Ann Rasmussen, Birmingham, Ala., for plaintiff-appellee.

Appeal from the United States District Court for the Northern District of Alabama.

Before HENDERSON and CLARK, Circuit Judges, and HOFFMAN^* , District Judge.

WALTER E. HOFFMAN, District Judge:

In this Title VII case, the district court found that the defendant union had intentionally discriminated against the black plaintiff in its referral practices. The court awarded the plaintiff $8,500 in damages, plus prejudgment interest, and enjoined the defendant from further discrimination. We vacate the judgment of the district court, and remand for further proceedings and findings within the suitable evidentiary framework.

FACTS

The defendant is a construction union that negotiates with employers regarding the working rules and conditions of employment in the boilermaking trade. The fruits of the negotiation are embodied in a collective bargaining agreement entitled the Southeastern States Articles of Agreement ("the Articles"). By the terms of this agreement, local chapters of the union maintain an "out-of-work list" from which boilermakers with varying qualifications are referred to a job when an employer makes a request. As one might expect, boilermakers are paid at different wage rates when referred to a job.

The rate at which boilermakers are paid for referrals depends on their status in the union. Full-fledged boilermakers are referred at the rate established in the Articles, which typically is called the 100% rate. Those who are not fully qualified boilermakers are paid at lesser rates. This group generally includes apprentice boilermakers or members of the union's trainee program. The trainee program, as it is set out in the Articles, provides an 8,000-hour program through which boilermakers who lack formal qualifications for full-fledged status can attain that status. For the first 2,000 hours of the program, the trainee receives 70% of the full-fledged boilermaker rate. The trainee then receives 80% of the rate during the remaining 6,000 hours of the program.

The wage limitations of the trainee program are not ironclad. Even a trainee can receive the 100% rate in four situations: (1) having done ASME code welding¹  during a period when the wage rate for a trainee doing this work had been elevated to the 100% level; (2) having been mistakenly paid by an employer at the 100% rate; (3) having received credit for prior experience; or (4) having been exempted from the wage limits by a grandfather provision. In addition, the 8,000-hour threshold to qualification for the 100% rate can be reduced to a 6,000-hour threshold during periods of manpower shortage.

The plaintiff, Charles Barber, first applied in July 1977, to Local 455 of the defendant union, located in Muscle Shoals, Alabama. He had some prior experience, primarily as a laborer, but he did not inform the union of any experience. Consequently, he was placed in the trainee program, becoming the only black in that local chapter. In 1980, Barber joined the union's National Transient Division²  for which he received a book indicating that he was a full-fledged boilermaker under the terms of the Division's separate collective bargaining agreement. On representing this fact to a Local 455 official, Barber was advised that it did not alter his trainee status.

After Barber began to receive referrals, the 8,000-hour threshold was reduced temporarily to the 6,000-hour threshold, but the parties agreed that he never reached this reduced threshold during the period in which it was effective. Sometime in this general period ASME code welders qualified for 100% wage treatment, although it is unclear whether Barber worked as a code welder during this period.³  In 1979, furthermore, a group of trainees qualified for the 100% wage rate through the operation of a grandfather provision, while others such as Barber continued to receive wages at the lower rate. At the time of trial, Barber consistently had been referred to employers at the 80% rate and had not completed his 8,000 hours in the trainee program.

Barber has sued the defendant union twice. He brought an Equal Employment Opportunity Commission ("EEOC") claim in 1978 that led to the commencement of a suit in 1979. That suit ended on June 30, 1980, with a court-approved settlement by which Barber released "in full compromise settlement and satisfaction of any and all claims and causes of action raised or for which could have been raised." Barber filed the present suit in December 1980, within three months of having filed and had dismissed a second EEOC claim. He alleges disparate treatment in violation of Title VII of the Civil Rights Act of 1964, 42 U.S.C. §§ 2000e to 2000e-17 (1982).

The case was tried on November 28, 1983; the court rendered its decision the next day. Accepting the plaintiff's position, the court found that the defendant did violate Title VII by referring Barber at a wage rate lower than that at which some whites were referred. The court acknowledged that the consent decree in the first suit barred a finding of discrimination based on acts occurring before June 30, 1980. The finding of discrimination therefore rested on the difference between Barber's 80% referrals and the 100% rates that the court found to have been paid after June 30, 1980, to certain white trainees. The court awarded Barber $8,500 in damages, plus prejudgment interest, and enjoined the defendant from further discrimination. After the entry of final judgment, the defendant moved for a new trial on December 6, 1983; the motion was overruled on June 20, 1984. The defendant then appealed the final judgment.

STANDARD OF REVIEW

A succession of United States Supreme Court opinions has refined the scope of an employment discrimination suit. Although a series of evidentiary burdens now forms the skeleton of any Title VII case, the backbone of a disparate treatment suit is a showing that the defendant intentionally discriminated against the plaintiff.⁴  Accordingly, the Supreme Court has warned that, when such a case has been fully tried, the district court should turn its attention directly to the factual issue of discriminatory intent. See United States Postal Service v. Aikens, 460 U.S. 711, 103 S. Ct. 1478, 75 L. Ed. 2d 403 (1983). Because the issue is a factual one, this Court can reverse a district court's finding of intentional discrimination only if the finding is clearly erroneous, is based on clearly erroneous subsidiary findings of fact, or is based on an erroneous view of the law. Pullman-Standard v. Swint, 456 U.S. 273, 102 S. Ct. 1781, 72 L. Ed. 2d 66 (1982); Carmichael v. Birmingham Saw Works, 738 F.2d 1126, 1129 (11th Cir. 1984). Remand generally is necessary when a finding is clearly erroneous unless the record permits only one resolution of the case. Pullman-Standard, 456 U.S. at 291-92, 102 S. Ct. at 1791-92.

Our task, therefore, is to determine whether the district court's finding of intentional discrimination is clearly erroneous. We must note at the outset that our job has been made more difficult by the form of the district court's findings. The judge dictated findings of fact and conclusions of law for the record and, while we do not condemn oral findings outright,⁵  we do feel that in this instance the process contributed to the lack of clarity in the trial court's decision. The difficulty with the district court's decision is not a lack of specificity in the findings. The factual findings are reasonably specific. The difficulty is with the court's failure to arrange the findings clearly within the evidentiary framework constructed by the line of cases following McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S. Ct. 1817, 36 L. Ed. 2d 668 (1973).

In the framework developed by McDonnell Douglas and its progeny, the district court must analyze the proof at trial in three steps. See Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 101 S. Ct. 1089, 67 L. Ed. 2d 207 (1981). First, the court must consider whether the plaintiff has established a prima facie case of discrimination. Second, assuming a valid prima facie case, the court then considers whether the defendant has rebutted the plaintiff's case. And third, the court considers whether the plaintiff has carried its ultimate burden of proof by showing that the defendant's rebuttal is, in fact, a pretext for true discriminatory intent.⁶ 

Although the district court neither cited McDonnell Douglas and its progeny nor used the terminology derived from these cases, both the nature and the sequence of the findings indicate that the district court made its decision within the proper evidentiary framework. The findings progress from consideration of the difference between treatment of the plaintiff and treatment of white workers (prima facie case), to consideration both of what it termed "company records" and of testimony for the union (rebuttal of the prima facie case), to evidence contradicting the rebuttal evidence (pretext evidence). Consequently, we believe that the mandate of Rule 52(a)⁷  has been satisfied and that the district court stated its factual findings and legal conclusions sufficiently to make our review meaningful. See Ramirez v. Hofheinz, 619 F.2d 442, 445 (5th Cir. 1980) (" [W]e are not editors, and, so long as the purposes behind the rule are effectuated, failure to meet the technical requirements of Rule 52 does not warrant reversal or remand.").⁸ 

CORRECTNESS OF FINDINGS

The three-step analysis established by McDonnell Douglas has already been discussed. Now we must determine whether the district court's findings, on each step of the analysis, are clearly erroneous. The district court impliedly found that Barber had met his prima facie case. Because it made a general finding of discrimination,⁹  the court necessarily found that Barber had established his prima facie case. See Wright & Miller, Federal Practice & Procedure Sec. 2579, at 712-13 (1971) ("In some cases if he [the trial court] fails to make a finding on a particular fact it has been assumed that ... he impliedly made a finding consistent with his general finding."); Clinkenbeard v. Central Southwest Oil Corp., 526 F.2d 649 (5th Cir. 1976) (implied finding of agency relationship in general finding of agent's fiduciary duty).

The requisites for a prima facie case in a Title VII suit are flexible. Direct proof of discriminatory intent is not necessary. Teamsters v. United States, 431 U.S. 324, 335 n. 15, 97 S. Ct. 1843, 1854 n. 15, 52 L. Ed. 2d 396 (1977) ("Proof of discriminatory motive is critical, although it can in some situations be inferred from the mere fact of differences in treatment."). McDonnell Douglas listed four elements that a plaintiff must establish:

(i) that he belongs to a racial minority; (ii) that he applied and was qualified for a job for which the employer was seeking applicants; (iii) that, despite his qualifications, he was rejected; and (iv) that, after his rejection, the position remained open and the employer continued to seek applicants.

411 U.S. at 802, 93 S. Ct. at 1824. McDonnell Douglas was a hiring case, but variants of the four factors will apply in the context of referrals at unequal wage rates. Id. n. 13, 93 S. Ct. at 1824 n. 13 ("The facts necessarily will vary in Title VII cases, and the specification above of the prima facie proof required from respondent is not applicable in every respect in different factual situations."). In cases of the present type, the plaintiff can state a prima facie case if he shows that he is the member of a minority, that he was properly on the union's out-of-work list, and that he was not referred at the same wage rate as were white trainees. Cf. Lincoln v. Board of Regents of the University System of Georgia, 697 F.2d 928, 937 (11th Cir.), cert. denied, 464 U.S. 826, 104 S. Ct. 97, 78 L. Ed. 2d 102 (1983) (modifying elements of a prima facie case to fit wrongful discharge).

The district court grounded its finding of intentional discrimination, and thus its finding of a prima facie case, on a comparison of Barber's referrals with the referrals of three whites--John Sharit, Damon Temple, and James Ray. Sharit, formerly a trainee, testified that his referrals always were as a welder at the 100% rate. If Sharit's referrals were as a code welder¹⁰  during the temporary wage increase for these welders, the explanation for the 100% referrals would be plain. The period during which this exception was effective, however, is unclear. See supra note 4. Consideration of Sharit's 100% referrals as a trainee nevertheless was inappropriate because they occurred before the 1980 consent decree.

Sharit received his 8,000 hours in February or March of 1980, qualifying him for the 100% rate. Tr. at 65. The consent decree in Barber's first suit against the union was entered on June 30, 1980. If the decree precludes consideration of any prior acts of discrimination, therefore, evidence of Sharit's 100% referrals as a trainee would be barred. We hold that this evidence is indeed barred.

A consent decree can bar relitigation of an issue through the doctrine of issue preclusion, traditionally called collateral estoppel. See Kaspar Wire Works, Inc. v. Leco Engineering & Machine, 575 F.2d 530 (5th Cir. 1978). The familiar elements of issue preclusion are that the fact or point now in issue is (1) identical to an issue in the former action, (2) actually litigated and determined by the parties, and (3) necessarily so determined. See Williams v. Bennett, 689 F.2d 1370, 1381 (11th Cir. 1982). The nature of consent decrees forces a twist in this traditional analysis. The very purpose of such decrees is to avoid litigation, so the requirement of actual litigation necessary to preclusion always will be missing. See Kaspar Wire Works, 575 F.2d at 539; James, Consent Judgments as Collateral Estoppel, 108 U. Pa. L. Rev. 173, 179 (1959). Consequently, the court in Kaspar Wire Works adopted the following principles from the Restatement (Second) of Judgments:

In the case of a judgment entered without contest by confession, consent, or default, none of the issues is actually litigated. Therefore, the rule of [issue preclusion/collateral estoppel] does not apply with respect to any issue in a subsequent action. The judgment may be conclusive, however, with respect to one or more issues, if the parties have entered an agreement manifesting such intention.

Kaspar Wire Works, 575 F.2d at 539 (quoting Restatement (Second) of Judgments, Sec. 68 comment e (Tent. Draft No. 1, 1973)).

The central inquiry in determining the preclusive effect of a consent decree is the intention of the parties as manifested in the decree or otherwise. "Whatever type of repose is sought to be invoked as a result of a judicial consent decree, a court should take into account the fact that it was rendered by consent and determine its impact by the issues actually intended to be precluded by the parties." Kaspar Wire Works, 575 F.2d at 539; see also James, supra, at 180 (contractual intent is the key); 1B Moore's Federal Practice, p 0.443, at 768 (1984) (consent decree may have preclusive effect because of parties' intent). The intent can be inferred from the words of the agreement or from the record. See Kaspar Wire Works, 575 F.2d at 540; James, supra, at 193. Applying these principles, the court in Kaspar Wire Works held that issue preclusion did not apply to the particular consent decree in that case. 575 F.2d at 542. Before entry of the consent decree in a suit for a declaratory judgment of patent invalidity, counsel for plaintiff stated on the record that the agreement did not prevent his client from raising patent invalidity as a defense if the client subsequently was sued. Thus, the contractual intent essential to the preclusive effect of a consent decree clearly was missing. Id. at 539-40.

In contrast to the decree in Kaspar Wire Works, the consent decree settling the first suit between Barber and the union clearly manifests their intention to give the decree preclusive effect. After outlining the relief to be afforded Barber, the final paragraph¹¹  of the court-approved settlement agreement states the following:Upon satisfaction of the terms and conditions set forth in this Settlement Agreement, plaintiff releases and discharges defendant in full compromise settlement and satisfaction of any and all claims and causes of action raised or for which could have been raised in Barber v. International Brotherhood of Boilermakers, Iron Shipbuilders, Blacksmiths, Forgers and Helpers (AFL-CIO) Local Lodge # 455, et al, Civil Action No. 79-P-1069-S.

The language is clear. The parties agreed to give the release preclusive effect on "all claims and causes of action raised or for which could have been raised" in Barber's first suit. We respect the parties' clear contractual intent and accordingly hold that consideration of the referrals prior to the consent decree is improper. See Yachts America, Inc. v. United States, 673 F.2d 356, 230 Ct. Cl. 26 (1982), cert. denied, 459 U.S. 839, 103 S. Ct. 86, 74 L. Ed. 2d 81 (1984) (consent decree, based on settlement agreement expressing clear intent to resolve issues once and for all, precluded relitigation of issue in prior suit); Green v. Ancora-Citronelle Corp., 577 F.2d 1380 (9th Cir. 1978) (same); Harrison v. Bloomfield Building Industries, 435 F.2d 1192 (6th Cir. 1970) (same); see also James, supra, at 193 ("Where the agreement upon which a consent judgment is based is fairly to be construed as providing that the parties should be bound collaterally upon a certain point, that agreement will and should generally be given effect."); Comment, The Consent Judgment as an Instrument of Compromise and Settlement, 72 Harv. L. Rev. 1314, 1320-21 (1959) (" [B]ecause of [a consent] judgment's consensual nature, there seems to be no reason why, if the parties expressly so stipulate, collateral-estoppel effect should not be accorded, provided it is limited to those claims which might reasonably have been foreseen at the time of the agreement.").

The evidence revolving around Temple's referrals also does not support the plaintiff's prima facie case of disparate treatment. The evidence showed that Temple, a supervisor's son, was referred no more than four or five times. The union's work record listed him as having four referrals, two before the 1980 consent decree and two after. Although the record listed Temple's primary job classification as boilermaker, the entries for each particular referral indicates that only the first referral was in this classification; the rest were as a trainee at the 70% rate. The court found as follows: " [C]ompany records reflect that Mr. Temple--that his primary job class is that of boilermaker, which would entitle him to the one hundred percent rate. The court, therefore, does not credit [union business manager] Claunch's testimony that Mr. Temple was never sent out at the one hundred percent rate." Tr. at 158.

For the same reason that consideration of Sharit's referrals as a trainee is improper, consideration of Temple's first two referrals would be improper.¹²  The third and fourth referrals thus would be the only ones capable of supporting a finding of discrimination through disparate treatment. In making its subsidiary finding that Temple was referred at the 100% rate, the court relied on the work record only for the general indication of boilermaker classification, ignoring the more specific entries indicating trainee status on his referrals after the 1980 consent agreement. Based on the inference that it derived from considering only the general entry, the court disregarded the testimony of the union business manager that Temple had never been referred at the 100% rate. Temple did not testify at trial. His name arose when Barber, responding to the court's examination, stated that Temple was paid at the 100% rate even though he was not a journeyman and did not qualify under the grandfather provision. Barber gave no specific instances to support his claim and did not indicate whether he was pointing to referrals that occurred before or after the 1980 consent agreement. A finding is clearly erroneous "when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed." United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S. Ct. 525, 541, 92 L. Ed. 746 (1948). We have such a conviction about the district court's subsidiary finding that Temple had been referred at the 100% rate after the 1980 consent decree, and therefore consider the finding clearly erroneous.

The union's referrals of Ray, however, support Barber's prima facie case of disparate treatment. The evidence showed that Ray consistently received 100% referrals, continuing after the 1980 consent decree. There is some question whether Ray was admitted as a full-fledged boilermaker, qualifying him for the 100% wage rate. See infra notes 14 & 15. Nonetheless we will assume, for the purposes of analyzing the sufficiency of plaintiff's prima facie case, that Ray was admitted as a trainee. Cf. Burdine, 450 U.S. at 253, 101 S. Ct. at 1093 (plaintiff's "burden of establishing a prima facie case of disparate treatment is not onerous"); Schlei & Grossman, Employment Discrimination Law, ch. 36, at 154 & n. 23 (1983 Supp.) (indicating that the burden has not been rigorously enforced).

The district court relied on Ray's referrals at the 100% rate in finding that Barber established a prima facie case of disparate treatment. Although the referrals of Sharit and Temple do not support the finding, the referrals of Ray may well support such a finding. Assuming that Ray was a trainee just like Barber, the disparity between Ray's 100% referrals and Barber's 80% referrals, without further explanation, supports an inference of disparate treatment. Thus, we cannot say that the district court's finding of a valid prima facie case is clearly erroneous.

Once Barber established his prima facie case, the union had the burden of articulating some legitimate, nondiscriminatory reason for the disparate treatment. See Burdine, 450 U.S. at 253-56, 101 S. Ct. at 1093-95 (considering this a burden of production). Only then would the plaintiff have to establish that the defendant's reasons were, in reality, a pretext for discrimination. See id. at 253, 101 S. Ct. at 1093 (considering this an ultimate burden, one of persuasion).

In making its general finding of discrimination, see supra note 9, the district court necessarily found that the defendant's trainee program was not legitimate or was not properly applied (no rebuttal of prima facie case), or that the program was a pretext for discrimination (ultimate burden of proof carried).¹³  The district court analyzed both "company records" and testimony offered for the defendant and appears to have rejected its rebuttal evidence. The court also relies on, in its ultimate finding, evidence that a union official had called the plaintiff "a smart ass nigger," which would seem to be evidence of pretext. We hold, in any event, that either a finding of inadequate rebuttal or a finding of pretext is clearly erroneous in light of the record.

The 8,000-hour trainee program, sufficiently described by the testimony and the exhibits, is unobjectionable in itself. Experience requirements are uniformly approved. See, e.g., Cooper v. Allen, 493 F.2d 765 (5th Cir. 1974). These requirements in effect help to avoid disparate treatment by providing reasonably objective standards by which the value of workers with different qualifications can be measured. Considering the way in which the program was applied, however, is the more pointed inquiry because even the most ideal of programs can be carried out unfairly.

The union presented evidence of a legitimate, nondiscriminatory reason for having referred Ray at the 100% rate. Ray admitted that he misrepresented his qualifications on first applying to the union. Tr. at 62. He told the union that he was a National Transient Division member when in fact he was not, although he truthfully told them of his seven to eight years of welding and rigging work in shops and with Pittsburgh Des Moines Steel, a large contractor that he had just left. This may have led the union to admit Ray as a full-fledged boilermaker. The evidence shows that he repeatedly was referred at the 100% rate, and his referrals are consistent with the status of a full-fledged boilermaker.¹⁴ 

Even assuming that Ray was a trainee like Barber,¹⁵  he could have qualified for the 100% wage rate because of his prior experience. One of the exceptions to the 80% rate for trainees is for prior experience.¹⁶  Unlike Barber, Ray told the defendant at the outset not only that he had extensive welding experience but also that he was a National Transient Division member, even though this last representation was untrue.¹⁷  The union, therefore, had a legitimate, nondiscriminatory reason for referring Ray at the 100% wage rate even if he was a trainee. See Burdine, 450 U.S. at 254, 101 S. Ct. at 1094 ("The defendant need not persuade the court that it actually was motivated by the proffered reasons.... It is sufficient if the defendant's evidence raises a genuine issue of fact as to whether it discriminated against the plaintiff.") (citation omitted). Because we are left with the definite and firm conviction that a mistake has been committed, we consider clearly erroneous the district court's finding that the union did not rebut the plaintiff's prima facie case. See United States Gypsum Co., 333 U.S. at 395, 68 S. Ct. at 541.

The plaintiff must produce evidence of pretext if the defendant rebuts the plaintiff's prima facie case. As has already been discussed, the district court apparently found that the defendant failed to rebut the plaintiff's prima facie case. In its ultimate holding, however, the court stated as a reason for the plaintiff's intentional discrimination "the view as expressed by one of the union officials that the plaintiff was a smart ass nigger for having complained of what the court finds to be an unlawful employment practice." Tr. at 158. This resembles a finding of pretext. If it is, the finding is clearly erroneous. The statement, according to the evidence, was made by a union official who was not part of Local 455 and who was not, in fact, directly charged with referring Barber. In addition, the mere utterance of an ethnic or racial epithet is insufficient to carry the ultimate burden of proof. See Henson v. City of Dundee, 682 F.2d 897, 904 (11th Cir. 1982) (" [T]he 'mere utterance of an ethnic or racial epithet which engenders offensive feelings in an employee' does not affect the terms, conditions, or privileges of employment to a sufficiently significant degree to violate Title VII"; harassment must be pervasive) (quoting Rogers v. EEOC, 454 F.2d 234, 238 (5th Cir. 1971), cert. denied, 406 U.S. 957, 92 S. Ct. 2058, 32 L. Ed. 2d 343 (1972)); see also Cariddi v. Kansas City Chiefs Football Club, 568 F.2d 87 (1977) (per curiam) (isolated ethnic slurs insufficient to show pretext).IV.

PROCEEDINGS ON REMAND

The evidence at trial did not support a finding of intentional discrimination in the union's referral practices. Evidence of Sharit's referrals fails to show disparate treatment, by any interpretation, because of the 1980 consent decree. The evidence concerning referrals of Temple and Ray was, at best, problematic.

The evidence did not establish whether Temple was referred at the 100% rate after the 1980 consent decree. If he was referred at this rate, proof of such referrals would support Barber's prima facie case. The court chose not to rely on union business manager Claunch's testimony that Temple had not been so referred. This it had a right to do. Yet the court, in finding that Temple had been referred at the 100% rate, relied on a work record that is internally inconsistent. This it could not properly do. On remand, further proceedings will be necessary to establish whether Temple was referred at the 100% rate after June 30, 1980. The additional evidence, of course, should then be analyzed by the court according to the suitable evidentiary framework. See supra note 6 and accompanying text.

Likewise, the evidence failed to establish whether Ray was referred, as a trainee, at the 100% rate after the consent decree. We assumed so for purposes of analyzing the sufficiency of Barber's prima facie case. If the court finds on remand that Ray had been considered a full-fledged boilermaker all along, however, then it must conclude that his referrals in this status do not support Barber's prima facie case. Disparate treatment would be lacking.

Even if the court finds Ray to have been a trainee at the time of his 100% referrals after June 30, 1980, the inquiry must continue. The court then should consider whether Ray was credited by the defendant for his prior experience, qualifying him for an exception to the 80% referral rule. See supra notes 14-17 and accompanying text. If he was so credited, a finding to this effect will establish, at the least, that the defendant had a legitimate, nondiscriminatory reason for referring him at the 100% rate while not referring Barber at that rate. The court would then consider whether Barber produced any pretext evidence other than the reference to him as "a smart ass nigger," which we have held insufficient to carry Barber's ultimate burden of proof. Further proceedings may be necessary on the issue of Ray's referrals, particularly relating to whether the union credited him for his prior experience.

CONCLUSION

The district court's ultimate finding of intentional discrimination is clearly erroneous because the evidence at trial failed to establish disparate treatment. Further proceedings and additional findings are necessary. Accordingly, the judgment of the district court is VACATED, and the case is REMANDED for further proceedings consistent with this opinion.