This case involves a complaint filed under Title VII of the Civil Rights Act of 1964 (as amended), 42 U.S.C. § 2000e-16. After a non-jury trial the district court entered judgment in favor of the defendant finding that the plaintiff, a black woman, had not been denied promotions because of her race. We affirm.
I. STATEMENT OF THE CASE
The federal government's Office of Personnel Management (OPM) maintains its Staffing Service Center (SSC or Center) in Macon, Georgia. The Center, opened in 1975 as part of an experimental project, now stores and provides computerized scoring, reviewing, and rating of all personnel applications submitted for all United States civil service positions throughout the world. Its success has resulted in an increase in the size of its staff from 13 in 1975 to approximately 55 in 1983. Of the original 13 employees, two were black, with one holding a supervisor’s grade of GS-7. Of the 55 employees in 1983,10 were black. However, only one black, a GS-8, was among the 20 employees filling positions graded GS-6 or above.
Louise T. Smith began working for the SSC in May 1976 as a GS-3 Output Clerk. In July 1978 Smith competed and was selected for a GS-4 Staffing Clerk position in the Quality Control section of the Center. Smith, a black, was chosen by Mary Bar-show, a white who supervised the Quality Control section, over three other eligible candidates, all of whom were white. One year later, Barshow promoted Smith to GS-5, the highest grade authorized for Smith’s job. In 1982 Smith applied for a higher grade job as a computer operator in another section at the SCC. Barshow, as Smith’s •supervisor, completed an evaluation of Smith’s work to accompany the application. In the evaluation, Barshow called Smith “an excellent employee” and rated Smith “exceptional” in the categories “ability to communicate orally” and “ability to work independently.” Smith was not selected for the job, however, and remained in her GS-5 position through the events giving rise to this action.
Between 1978 and August 1983, Barshow filled several other vacancies in the Quality Control section. In September 1978 Bar-show selected a black man from a list of two eligible candidates to fill a GS-11 position. After he declined to accept the job, Barshow offered the position to the other eligible individual, a black woman. She too declined. Barshow then received permission to reduce the grade to GS-9 in order to attract more qualified applicants. However, even at the lower, more accessible level, only one candidate applied.1 That person, a white, eventually was selected for the job. In addition, Barshow offered GS-3 and GS-4 positions to four whites and one black. In only two of these instances, however, did the names of both blacks and whites appear simultaneously on a list of eligible candidates.2
*1533In August 1983 the Quality Control section announced that two “career conditional” vacancies existed and that the positions were open only to employees of the Center. Both positions provided an entry-level grade of GS-5 but offered non-competitive promotion opportunities through GS-11. Approximately 20 people, some of whom held grades above GS-5, applied for each job. Smith was among the applicants for both vacancies. In conjunction with Smith’s applications, Barshow completed another evaluation form, rating Smith as “thoroughly competent” or above in the category of “communications skills” and remarking that Smith “ha[d] the potential to perform extremely well in this position.”
All of the applications for the two vacancies were submitted to an administrative officer of the OPM not employed by the Center. That officer screened the applications to ensure that each candidate met the qualifications for the applicable position or positions. This preliminary process resulted in one or two names being removed from consideration for each of the jobs. The remainder of the names were sent to a three-member panel convened to determine which candidates were “best qualified” for the two positions. The panel members used a rating plan to develop a numerical score for each remaining applicant. The scores then were returned to the administrative officer who averaged them for each candidate. Those applicants whose average scores met or exceeded a pre-deter-mined value were deemed “best qualified” and had their names placed on a “certificate of eligibles.” For the first vacancy, that of Personnel Staffing Specialist, 12 names appeared on the certificate listed in alphabetical order within defined groupings.3 Of the 12 “best qualified” candidates for this position, only Smith was black. For the second vacancy, that of Program Analyst, 15 names appeared on the certificate, again listed in alphabetical order within certain groups. Three of these candidates, including Smith, were black.
The two certificates, along with the applications and performance evaluations for those whose names appeared on the certificates, then were submitted to Barshow. Because Barshow supervised the Quality Control section (where the vacancies occurred), she was responsible for choosing the person from each list who would fill each position. Barshow interviewed each candidate listed on the certificates. For the Personnel Staffing Specialist position, Barshow selected Glenda Beck, a GS-6 Secretary to the chief of the SSC. Barshow chose Virginia Ward, a co-worker of Smith, for the job of Program Analyst. Both Beck and Ward are white.
On October 10, 1985, Smith filed a complaint against the OPM under Title VII alleging racial discrimination in employment. At trial, she testified that Barshow treated her differently than her white coworkers. In particular, Smith testified that her requests for personal leave and to use the bathroom were subjected to greater scrutiny by Barshow, and that Barshow discussed Smith’s difficulties publicly while agreeing to counsel the white co-workers in private. Smith also testified that a white co-worker had been dismissed from a temporary duty assignment for making the remark “they ought to teach the niggers how to read instead of how to vote,” that Barshow heard of the remark having been made, and that Barshow stated that she saw “nothing wrong” with the remark. Barshow denied the allegations.
Barshow testified that she selected Beck for the Personnel Staffing Specialist position because of Beck’s good communication skills, particularly as they related to han*1534dling contacts with high-ranking officials. The position required such skills, because the Personnel Staffing Specialist’s duties included remaining in constant contact with high-ranking agency representatives and congressional staff members and dealing tactfully with them. According to Bar-show, Beck had been “in contact with high-ranking officials and [had] handled those contacts extremely well.” Beck’s evaluation, completed by the chief of the SSC, contained comments such as “[w]e continue to receive compliments from our contacts in the Central Office, area offices, and regional offices with respect to the competency, courteousness, and efficiency that Ms. Beck exhibits in her dealings with them” and “[Beck] commands the respect of everyone she deals with and is relied on more and more by units in Washington for information and periodic reports.” Bar-show further testified that Smith’s communication skills did not match Beck’s and that Smith became “too excited” when faced with certain situations.
Barshow also stated that she chose Ward for the Program Analyst position because of Ward’s prior managerial experience, her responsibility, and her ability to “get[] along extremely well with people” and maintain her composure when under pressure. Ward previously had managed a family-owned jewelry business, filling in for several years for her husband after he suffered a severe heart attack. Ward also had 18 years of federal employment experience. Her evaluation form, completed by Barshow, remarked that “[s]he ha[d] excellent potential for the position.”
The district court entered an order supported by its findings of fact and conclusions of law. The court held that the OPM had “clearly articulated legitimate and nondiscriminatory reasons for selecting individuals other than [Smith]” for the two vacancies and that Smith had failed to meet her burden of proving that the OPM’s reasons were but pretexts for discrimination. According to the court, Smith thus had failed to prove intentional discrimination by the OPM.
II. DISCUSSION
On appeal, Smith asserts that the district court erred in several respects. First, Smith contends that the court failed to analyze the facts under the disparate impact theory of discrimination. Second, she asserts that the trial court ignored certain direct evidence of discrimination and thereby utilized an incorrect legal standard in determining the sufficiency of the OPM’s rebuttal argument. Finally, Smith argues that the court erred in not finding OPM’s reasons to be pretextual in light of the evidence presented.
The OPM, to the contrary, asserts that the district court properly used only the disparate treatment analysis because Smith failed to argue the disparate impact theory at trial. The OPM also claims that Smith presented no direct evidence of discrimination and that Smith' failed to meet her burden of proving pretext by a preponderance of the evidence. We agree with the OPM that the disparate impact theory was not argued at trial. We also agree that the district court applied the correct legal standard in evaluating the OPM’s reasons for Smith’s non-promotion and that the court’s factual findings are not clearly erroneous.
A. Disparate Impact
In general, “[a] party cannot raise a new theory on appeal that was not presented to the court below.” Capps v. Humble Oil & Refining Co., 536 F.2d 80, 82 (5th Cir.1976).4 Here, however, Smith begins by contending that she presented the disparate impact theory to the district court. Our review of the record leads us to the opposite conclusion.
During Smith’s opening argument, the court and Smith’s trial counsel discussed Smith’s proposed use of statistical evidence. In responding to a statement made by the court, Smith’s counsel remarked that the case involved both “adverse impact and disparate treatment” claims. How*1535ever, as the trial progressed, it focused on the treatment of Smith individually. Most telling in this regard are the proposed findings of fact and conclusions of law submitted by Smith after the trial was completed. Her proposed conclusions began: “As established in McDonnell Douglas Corp. v. Green, 411 U.S. 792 [93 S.Ct. 1817, 36 L.Ed.2d 668] (1973), there is a three-step burden of proof sequence to be followed in this individual Title VII claim based upon disparate treatment ” (emphasis added). She continued by discussing the McDonnell Douglas proof sequence, writing that “[t]he critical issue ... is whether the defendant’s reasons for [her] non-selection ... are pretextual.” Nowhere, however, did Smith refer to any disparate impact claim. Likewise, Smith’s pre-trial proposed findings and the proposed pre-trial order contained no mention of a disparate impact theory. Finally, Smith’s closing argument addressed solely the McDonnell Douglas test for disparate treatment.5 Based on this record, we conclude that Smith did not present a disparate impact claim to the trial court.6 The court’s failure to analyze the case under that theory, therefore, was not error.
Smith argues in the alternative that this court may consider her disparate impact theory for the first time on appeal. The Supreme Court has stated that “there are circumstances in which a federal appellate court is justified in resolving an issue not passed on below, as where ... injustice might otherwise result.” Singleton v. Wulff, 428 U.S. 106, 121, 96 S.Ct. 2868, 2877, 49 L.Ed.2d 826 (1976). On occasion this court has considered on appeal new theories raising purely legal questions when to ignore them would work a “manifest injustice.” See, e.g., Roofing & Sheet Metal Services, Inc. v. La Quinta Motor Inns, Inc., 689 F.2d 982, 990 (11th Cir.1982); Higginbotham v. Ford Motor Co., 540 F.2d 762, 786 n. 10 (5th Cir.1976). However, the court will not make an excep*1536tion “if additional facts would have been developed in the trial court had the new theory been presented there; in that case judicial economy is served and prejudice is avoided by binding the parties to the facts presented and the theories argued below.” Higginbotham, 540 F.2d at 768 n. 10.
In Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 252 n. 5, 101 S.Ct. 1089, 1093 n. 5, 67 L.Ed.2d 207 (1981), the Supreme Court noted that the factual issues involved in disparate treatment and disparate impact eases are not identical. A defendant in a disparate treatment case, in order to prevail, must articulate a legitimate, non-discriminatory reason for its action if an inference of discriminatory intent would be authorized. In a disparate impact case, however, rather than merely articulating a legitimate, nondiscriminatory reason for its action, the defendant must produce evidence, and indeed bears the burden of proving, that its action was a result of “business necessity.” See Nash v. Consolidated City of Jacksonville, 763 F.2d 1393, 1397 (11th Cir.1985). In this case the OPM asserted that the selecting official did not choose Smith because she believed other applicants to be better qualified, thereby articulating a legitimate, non-discriminatory rationale for Smith’s denial of promotions. Whether this result was dictated by “business necessity,” or whether additional facts could have been developed to help prove business necessity, remain unknown. Hence, this court will not review Smith’s disparate impact claim initially here on appeal.7
B. Direct Evidence
A plaintiff alleging disparate treatment may establish a prima facie case of discrimination in several ways. One such way is to produce credible direct evidence of discriminatory intent. If a plaintiff produces direct evidence, the burden shifts to the defendant to prove by a preponderance of the evidence that the same employment decision would have been reached even absent the discriminatory intent. See, e.g., Lee v. Russell County Board of Education, 684 F.2d 769, 774 (11th Cir.1982). In the alternative, a plaintiff may present a prima facie case of discrimination through circumstantial evidence. Under the framework announced in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 0973),
A plaintiff may establish a prima facie case of promotion discrimination by proving that he or she is a member of a protected minority, was qualified for and applied for the promotion, was rejected despite these qualifications, and that other employees with equal or lesser qualifications who were not members of the protected minority were promoted.
Perryman v. Johnson Products Co., 698 F.2d 1138, 1142 n. 7 (11th Cir.1983).8 If a plaintiff presents a prima facie case through circumstantial evidence, “the bur*1537den shifts to the employer ‘to articulate some legitimate, non-discriminatory reason’ for the alleged discriminatory action.” Id. at 1142 (quoting McDonnell Douglas, 411 U.S. at 802, 93 S.Ct. at 1824). If the defendant satisfies this “exceedingly light” rebuttal burden by producing evidence of a legitimate rationale for its decision, the plaintiff must then prove that the proffered explanation is pretextual. See id.; see also Griffin v. Carlin, 755 F.2d 1516, 1526 (11th Cir.1985).
Smith claims that she produced direct evidence of Barshow’s intent to discriminate. Most significant of the purportedly direct evidence is Smith’s testimony that Barshow ratified a white subordinate’s statement that “they ought to teach the niggers how to read instead of how to vote.”9 Because direct evidence existed, Smith contends, the district court should have placed the heavier rebuttal burden of Lee on the employer.
The difficulty with Smith’s argument is that, at trial, her testimony concerning the racial slur was not used to establish a prima facie case of discrimination. Rather, Smith elected to rely on statistical and other circumstantial evidence to create an inference of discriminatory intent under McDonnell Douglas. The case proceeded under McDonnell Douglas arguments, with the OPM articulating a legitimate rationale for the decisions. Only afterward, in attempting to prove pretext, did Smith allude to the racial statement. Again, most telling are Smith’s post-trial proposed conclusions of law; in them, the only mention of the slur occurs during a discussion of pretext. Nowhere in the proposed conclusions or in closing argument did Smith ever address using direct evidence to establish a prima facie case under Lee.10 Because Smith elected to establish her prima facie case through circumstantial evidence, the district court properly used the McDonnell Douglas standard in evaluating the OPM’s rebuttal.11
C. Pretext
As noted earlier, McDonnell Douglas establishes a three-step burden of proof se*1538quence to be followed in most disparate treatment cases. Under the third prong of the McDonnell Douglas test, once the plaintiff has established a prima facie ease of discrimination and the defendant has articulated a legitimate, non-discriminatory rationale for its action, the burden shifts to the plaintiff to prove by a preponderance of the evidence that the defendant’s reason is but a pretext for discrimination. See McDonnell Douglas, 411 U.S. at 804, 93 S.Ct. at 1825. The ultimate issue remains whether the employer has exercised discriminatory intent. Because intentional discrimination is a question of fact, we may reverse the district court’s decision only if its finding is clearly erroneous. See, e.g., Carmichael v. Birmingham Saw Works, 738 F.2d 1126, 1129-30 (11th Cir.1984).
Smith contends that no basis exists for the conclusion that Beck and Ward were better qualified than Smith for the positions in question. In particular, Smith argues that Barshow’s testimony that Beck possessed superior communication skills is belied by Barshow’s 1982 evaluation of Smith, which rated Smith “exceptional” in the category of “communications skills.” According to Smith, based on the facts presented, the district court’s finding that Smith did not prove pretext “simply does not speak the truth and right of the ease.”
In Canino v. United States EEOC, 707 F.2d 468 (11th Cir.1983), this court was faced with a situation similar to the one presented here. Canino, a GS-14 employee of the EEOC, applied for three vacant GS-15 positions in various parts of the country. His applications were forwarded to ranking panels and assigned numerical scores. In each case, Canino’s name was among those placed on the Promotion Eligibility Listing (i.e., the “certificate of eligibles”). The scores of the candidates did not appear on the listings; instead, the selecting officials were free to choose any of the applicants on each list. Canino was not selected for any of the positions. See id. at 470.
At trial, the selecting officials explained their actions:
Charles Clark, the recommending official for the Kansas City vacancy, explained that he was familiar with the successful applicant’s work and thought that he was best qualified to deal with the problems in the Kansas City office. Donald Hollo-well, the recommending officer for the Jackson position, based his selection on the successful applicant’s 20 years of investigatory and administrative experience, law degree, and work performance as the Jackson Deputy Director. The selecting official for the Detroit directorship chose an applicant who had the most managerial experience and extensive E.E.O.C. experience at the state level and in private industry.
Id. The district court found these reasons to be legitimate. The Eleventh Circuit, utilizing the “clearly erroneous” standard, affirmed. See id. at 471.
In this case Barshow testified that she believed that Beck best would be able to deal with high-ranking agency officials. Beck’s performance evaluations and prior work history support Barshow's belief. Likewise, Ward’s evaluation and prior experience support the conclusion that she might be best qualified for the managerial aspects of the Program Analyst position. “If an employer selects the person it believes is best qualified, an argument of pretext ordinarily will fail.” Clark v. Huntsville Board of Education, 717 F.2d 525, 527 (11th Cir.1983). The OPM does not dispute that Smith also was well-qualified for the challenged positions.12 However, “Title VII does not require an employer to hire or promote the most qualified applicant.” McCarthney v. Griffin-Spalding County Board of Education, 791 F.2d 1549, 1552 (11th Cir.1986). Nor *1539does it require that a candidate be given preference over other applicants because of race. Rather, it merely requires that employers not disadvantage certain employees based on race, sex, religion, color, or national origin. Where, as here, several candidates are well-qualified for a single position, and the district court accepts the employer’s testimony that it chose the person it thought best qualified for the job, that finding ordinarily will not be overturned on appeal. See, e.g., Canino, 707 F.2d at 470-71. Beck and Ward both possessed important skills relevant to the two positions. Barshow explained to the trial court why she chose the two women and why only one black worked in the Quality Control Section. Barshow testified that she had not treated Smith any differently than her other employees and denied having heard or ratified the racial slur. Her testimony was internally consistent and not contradicted by any evidence of record. Based on the evidence presented, therefore, the district court’s conclusion that Smith would not have been promoted regardless of her race is not clearly erroneous.
The judgment of the district court is AFFIRMED.
. According to Barshow, although the vacancy was advertised throughout the country, few people likely possessed the specialized experience needed for the job.
. Barshow testified that in 1978 she selected a white woman for a GS-4 slot from a list of three whites. That woman departed after two months, forcing Barshow to refill the vacancy as well as fill another that had been authorized *1533for her section. In this instance Barshow chose two whites from a list of four whites and one black. In 1979 Barshow selected a black woman for a GS-3 position from a list of two whites and one black. After the black did not accept the job, Barshow chose one of the remaining two white applicants.
. The names of the eligible candidates were listed in three groups: those who held positions graded above GS-5 (the initial employment grade for the vacancy), those who held GS-5 rank, and those who held grades below GS-5. The numerical scores computed by the panel were not placed on the certificate, in accordance with the Center’s standard procedures.
. In Bonner v. City of Prichard, 661 F.2d 1206 (11th Cir.1981) (en banc), this court adopted as precedent all of the decisions of the former Fifth Circuit decided prior to October 1, 1981.
. Smith’s trial counsel actually uttered the phrase "disparate impact” near the beginning of the closing argument, apparently accidentally. Taken in context, the utterance does not undermine the conclusion that the argument did not present a disparate impact claim:
All right, sir. I believe she's shown that she’s obviously a member of the protected race, that she applied for a job for which she was qualified, I believe that’s undisputed, that the job — she was denied that job and that the job was given to a white individual, in this case two white individuals, the same would apply for both jobs. Statistics, Your Honor, are relevant in a disparate impact, disparate treatment case, which is basically what we’ve got, that being the first prong of attack Ms. Smith has against the defendant, and the statistics are crystal clear and there’s been no evidence introduced by the defendant whatsoever to explain the statistics.
After reviewing the statistical evidence and Smith’s testimony, Smith’s trial counsel continued:
The explanation now that the Government has come forward with this prima facie case having been established by the plaintiff that you've got lousy statistics out there that jump out at you and you've got different treatment based upon the testimony of the plaintiff, humiliating treatment from her white supervisor — to counter that, the Government has got to establish that they have a legitimate reason for the non-selection, a legitimate reason, a non-protectual [sic non-pretextual] reason, and their reasons are, starting with the personnel staffing specialist position, the one that Ms. Beck got, was — as I heard Ms. Barshow’s testimony, is that she considered Ms. Smith, the plaintiff, not to have the necessary communicative skills. Ms. Smith has come forward with evidence which I believe shows that that’s a pretext.
This is a McDonnell Douglas argument. See infra part II. C.
. A new lawyer represented Smith on appeal. Appellate counsel initially appeared to agree with the conclusion that the disparate impact theory was not presented at trial. During oral argument, the following exchange occurred:
PANEL MEMBER: ... The case wasn’t tried on disparate impact theory—
SMITH’S COUNSEL: No it wasn’t. But the evidence was—
PANEL MEMBER: And the law says, as I understand it, that you cannot argue disparate impact when it wasn’t presented to the district court.
Later, however, Smith’s lawyer began to back away from her earlier comment:
PANEL MEMBER: Are you arguing a theory to us that you didn’t argue to the district judge, that’s what I’m getting at. If you are, you’re on tenuous grounds.
SMITH’S COUNSEL: Yes, Sir. I think that the theory was presented. I think that the ball was there. I think that the ball was dropped....
While we doubt that the ball ever was "there," we agree that if it was, it indeed was "dropped.”
. Nor do we believe that to ignore this theory would work a "manifest injustice." Smith’s “naked” statistics do show that few blacks at OPM were employed at grades above GS-5. However, some evidence was introduced that might show that factors other than the selection process contributed to this result. For example, Barshow testified that she offered a GS-11 position to two blacks, both of whom declined to accept it. Another black, holding a GS-7 or GS-8 rating at the Center, departed for personal reasons. The OPM also introduced evidence detailing various hiring freezes throughout the Carter and Reagan administrations. Although these freezes apparently had little effect on internal promotions, they did affect the hiring and promoting of those not already employed by the SSC. Whether Smith’s statistical evidence would have sustained a disparate impact claim is unclear; however, we do not believe her evidence so compelling that to “ignore" it for purposes of the disparate impact theory would work a manifest injustice.
. A prima facie case of classwide disparate treatment also may be established by statistics alone if they are sufficiently compelling. See Griffin v. Carlin, 755 F.2d 1516, 1525 (11th Cir.1985); Carmichael v. Birmingham Saw Works, 738 F.2d 1126, 1131 (11th Cir.1984); Eastland v. Tennessee Valley Authority, 704 F.2d 613, 618 (11th Cir.1983), cert. denied sub nom. James v. Tennessee Valley Authority, 465 U.S. 1066, 104 S.Ct. 1415, 79 L.Ed.2d 741 (1984). In an individual case of disparate treatment, such as this one, statistical information “can be relevant and important.” Carmichael, 738 F.2d at 1131. However, statistics alone cannot establish a prima facie case of individual disparate treatment. See id.
. Smith argues that Barshow’s prior selections and the fact that the government interviewed black employees of the Center the day before trial also constitute direct evidence of discriminatory intent. We fail to see how this additional evidence is directly probative of an intent to discriminate, however.
. We also note that, in the same discussion that Smith now contends supports her claim that the disparate impact theory was presented to the trial court, the following exchange occurred:
TRIAL COURT: ... You don't allege any direct racial discrimination. You know, by that I mean some statement—
SMITH’S COUNSEL: This is like most — I believe most discrimination cases, at least in my experience, you don't have a smoking gun.
TRIAL COURT: Well, I know. I just — I know you very seldom see those any more, but I just wanted to know what you were relying on.
SMITH’S COUNSEL: We don’t have racial statements.
TRIAL COURT: Okay. Well, I think I understand exactly where you’re going. Thank you.
. Of course, the method which a plaintiff chooses to establish a prima facie case affects only the internal structure of the legal analysis. The external structure, representing the question of whether the defendant has intentionally discriminated against the plaintiff, remains the same under both Lee and McDonnell Douglas. Thus, Smith’s use of her testimony as evidence of pretext rather than direct evidence of discrimination should not change the overall result.
However, one significant difference in the internal analyses under Lee and McDonnell Douglas manifests itself here on appeal. Barshow denied having made or ratified the racial slur about which Smith testified, thus creating a conflict in the evidence. The district court, although accepting virtually all of Barshow’s testimony as true, made no express finding regarding whether Barshow ratified the racial remark. Had Smith presented her testimony as direct evidence of discrimination, the district court’s failure to “specifically state whether or not it believe[d] plaintiff’s proffered direct evidence of discrimination” would have constituted error. Thompkins v. Morris Brown College, 752 F.2d 558, 564 (11th Cir.1985). No such requirement exists regarding statements used to show pretext. While we would have preferred an express finding here, we cannot fault the district court for not making one. Smith’s lawyer initially indicated that no racial statements had been made, see supra note 10, and never argued that the district court should use analysis normally reserved for cases involving direct evidence of discrimination. Quite appropriately, then, the district court focused its attention on the OPM’s reasons for denying Smith the promotions and the question of whether they were pretextual. Likewise, we now turn to that issue.
. Smith argues that she was better qualified for the positions than Beck and Ward. Barshow’s earlier rating of Smith’s communication skills as "exceptional,” albeit in the context of a promotion to a different position, indeed presents some evidence of Smith's ability to perform the two jobs in question, both of which required good communication skills. However, this evidence does not render Beck and Ward unqualified for the jobs; nor does it render their communication or other skills inferior to Smith’s. Thus, the earlier rating does not prove that Barshow’s partial reliance on communication skills was pretext.