This case is an appeal from an entry of summary judgment against the plaintiff, Ellice Luh, in a Title VII race and sex discrimination lawsuit against her employer, J.M. Huber Corporation.
*144The plaintiff alleges that she was terminated from her position due not only to her race, but also because she was a woman who had turned down an unwanted sexual advance from a fellow management employee. Huber maintains that it terminated her as part of a general restructuring of its business following marked downturn in profitability. The district court granted summary judgment for Huber, finding that Mrs. Luh had failed to proffer support for a finding that the reasons advanced by Huber were pretextual. We affirm.
I.
The plaintiff, Ellice Luh, is a Chinese-American woman. She earned an engineering Ph.D. from the University of California at Berkeley in 1988. Following graduation, she spent nine years working for the W.R. Grace Co., developing products for paper coating, electromagnetic glass shielding, and dental silica applications, among others. She was hired away from this position by Huber in June 1997 in order to be the Technology Manager for the company’s dental group located in Havre de Grace, Maryland. Huber is a major supplier of engineered materials and research and development services.
While with Huber, Mrs. Luh’s work was on the development of dental silicas, which are the abrasives used in toothpaste.
Because Mrs. Luh joined Huber shortly' before it undertook a large restructuring of its operations, she had four different supervisors during her time with the company. Her first supervisor, Bob Klem, had also initially interviewed her for the position. Following Klem was Jorma Sakko. Sakko was eventually promoted, and Mrs. Luh was then supervised by Mike Withiam, who then reported to Sakko. In April 2000, Tom Izod was hired as the Director of Technology for the company, and took over the supervision of Mrs. Luh. Our recitation of some of the foregoing sequences of the supervisors may be slightly out of order.
In early 2000, as the result of a large loss in profits, Huber undertook a number of measures to reduce labor costs. These measures included the decision by Izod to combine the Dental Technology Group with the Food, Feed, Health and Care Group into a new group called the Consumer Technology Group. The management of these two groups would be combined, and redundant positions would be eliminated. The plaintiffs was one of the eliminated positions.
Philip Block, who held a position with the Food, Feed, Health and Care Group analogous to Mrs. Luh’s with the Dental Group, received his Ph.D. from the University of North Carolina in 1993. Prior to working at Huber, he worked for Johnson & Johnson developing surgical gloves, sterilization compounds, and skin care products. He also held a position at Unilever developing detergents and soaps. In his only performance review at Huber pri- or to the restructuring, Block had apparently been criticized by some subordinate employees for his management style, including his decision to cancel a morning coffee break.
Prior to selecting between Mrs. Luh and Block, Izod met with each of them, Mrs. Luh on April 19, 2000, and Block the following day. During her meeting, Mrs. Luh informed Izod she was having problems with the other managers at Huber, and stated that, as a result, she could not see herself with the company in five years. This led Izod to conclude that she “was not interested in expanding her expertise outside of the Dental Technology area.” In contrast, Block spoke about his work in skin care and detergents and told Izod that he was thinking about new uses for existing products manufactured by the *145company. This lead Izod to conclude that Block was a “go-getter.”
Izod ultimately selected Block to be the technology manager of the Consumer Technology Group. Izod then informed Mrs. Luh that her position was being eliminated, and her employment would be terminated effective June 30, 2000. She was offered a severance package and outplacement services.
Block’s own time at Huber was limited, however. He was removed from his new position and left Huber entirely by March 2001. Mrs. Luh says that Block later indicated to her that he had been asked to leave due to inadequate performance of his duties, which, however, is not consistent with his performance review in the record.
Mrs. Luh also was not the only Asian eliminated during Izod’s restructuring. Three of the six total employees who were terminated were Asian, although before the terminations, the group had 52 employees, 39 of whom were white and five were Asian.1
Following her termination, Mrs. Luh filed a complaint with the EEOC on February 20, 2001, alleging race and gender discrimination as well as quid pro quo harassment. The EEOC issued her a notice of right to sue, after which she filed this lawsuit on April 10, 2003.
The district court entered summary judgment in favor of Huber on all of Mrs. Luh’s claims on January 25, 2005. Specifically, it determined that she had not provided evidence to survive summary judgment by establishing a pretext under the McDonnell Douglas test. Further, the district court determined that Mrs. Luh could not sustain her quid pro quo claim,2 as she did not provide any support for the claim that her rejection of a sexual advance had been a motivating factor in the decision to terminate her, for much the same reason that she could not demonstrate pretext. Martin was not her supervisor and did not make the decision to terminate her. This appeal of the discrimination claim followed. Mrs. Luh has not pursued an appeal of the district court’s decision on her quid pro quo claim. Br., pp. 1-2 (“STATEMENT OF ISSUES PRESENTED FOR REVIEW.”)
II.
We review the grant or denial of summary judgment de novo. Higgins v. E.I. DuPont de Nemours & Co., 863 F.2d 1162, 1167 (4th Cir.1988). Summary judgment is appropriate when the materials before the court “show that there is no genuine issue as to any material fact and that the moving party is entitled to a judgment as a matter of law.” Fed.R.Civ.P. 56(c).
In her appeal, Mrs. Luh takes issue with a number of evidentiary rulings and interpretations made by the district court in the process of granting summary judgment. In examining a motion for summary judgment, a district court must “view the evidence in the light most favorable to ... the nonmovant, and draw all reasonable inferences in her favor without weighing the evidence or assessing the witness’ credibility.” Dennis v. Columbia Colleton Med. Ctr., Inc., 290 F.3d 639, 644-45 (4th Cir.2002). However, the court must still abide by “the affirmative obligation of the trial judge to prevent factually unsupported claims and defenses from proceeding to trial.” Drewitt v. Pratt, 999 F.2d 774, 778-79 (4th Cir.1993) (internal quotations omitted).
*146In discrimination eases such as this one, we apply the McDonnell Douglas burden-shifting scheme. Under this framework, the plaintiff is responsible for setting forth a prima facie case of discrimination. After that, the burden shifts to Huber to present a legitimate, non-discriminatory justification for its actions. This justification “must be legally sufficient to justify a judgment for” Huber. Tex. Dept. Comm. Affairs v. Burdine, 450 U.S. 248, 255, 101 S.Ct. 1089, 67 L.Ed.2d 207 (1981). Once Huber has done so, the plaintiff bears the burden of demonstrating that the proffered reasons were merely pretextual and the true reason for Huber’s actions was discriminatory animus.
A.
Most issues that Mrs. Luh raises in her appeal take issue with the district court’s use of Izod’s testimony to find that Huber had shown a legitimate, non-discriminatory justification for terminating her.
She argues that reliance on Izod’s testimony at the summary judgment stage is improper. She bases this position on a quotation from Reeves v. Sanderson Plumbing Prods., Inc., 530 U.S. 133, 151, 120 S.Ct. 2097, 147 L.Ed.2d 105 (2000), which states that “the court should give credence to the evidence favoring the nonmovant as well as that ‘evidence supporting the moving party that is uncontradicted and unimpeached, at least to the extent that that evidence comes from disinterested witnesses.’ ”
We agree with the district court that Mrs. Luh’s interpretation of Reeves is inappropriate in the present case. Reeves states the noncontroversial " position that witness testimony that the jury is not required to believe cannot be used to sustain a summary judgment decision, since the jury is not required to believe their testimony. Reeves, 530 U.S. at 151, 120 S.Ct. 2097. However, the Court has long held that there are exceptions to this rule. For example, in Chesapeake & Ohio R. Co. v. Martin, 283 U.S. 209, 51 S.Ct. 453, 75 L.Ed. 983 (1931), the Court held that this rule “does not mean that the jury is at liberty, under the guise of passing upon the credibility of a witness, to disregard his testimony, when from no reasonable point of view is it open to doubt.” 283 U.S. at 216, 51 S.Ct. 453. Instead, the testimony must display some indicia that it is not credible-whether by betraying a lack of candor, or by being unsettled by cross-examination, or by being contradicted by “proof or circumstance,” especially when it could readily have been shown to be inaccurate had it been. Chesapeake & Ohio R. Co., 283 U.S. at 216, 51 S.Ct. 453.
Recognizing this, Mrs. Luh has claimed that Izod’s testimony was internally contradictory and that, moreover, it contradicted evidence that she introduced. Neither of these is sufficient to save her claim, however.
We have held that an employer offering different justifications for the dismissal of an employee, one good and one bad, will justify a conclusion that the good justification is pretextual. Alvarado v. Bd. of Trs. of Montgomery Comm. Coll., 928 F.2d 118, 123 (4th Cir.1991). However, Mrs. Luh fails to establish such contradiction in Izod’s testimony before the court.
The first instance she identifies as a contradiction stems from the April 5, 2001 statement that Huber made to the EEOC, which she claims indicated that the decision to hire Block was a joint one by Izod and Rob Edmonds, another Huber executive. The relevant passage in the letter was that “Tom Izod, the Director of the Consumer and Industrial Global Technology, in discussions with Rob Edmonds, the Vice President of Consumer and Industrial Markets, looked carefully at each employee’s managerial strengths, and determined *147that Block was far better qualified for this management position.” Mrs. Luh contends that this contradicts Huber’s response to an interrogatory where it stated simply that “Tom Izod selected Philip Block ... Mr. Izod made his selection after assessing the managerial skills that he believed were important ...” She adds that these statements, indicating careful consideration of the choice, should be seen as conflicting with Izod’s testimony that “time was of the essence” in making the decision. We do not believe that she reads the record fairly. First, Edmonds was Izod’s superior, and had to approve Izod’s decision. Mrs. Luh presents no evidence, however, that Edmonds did anything other than approve Izod’s choice. Simply because Izod did not have authority to act without first obtaining the approval of his supervisor is not sufficient to find that any statement that he, Izod, “made the decision” is false or otherwise a misrepresentation that undercuts his testimony. Similarly, the mere fact that he had to reach a conclusion quickly does not mean that, in so doing, he did not conduct an adequate assessment of Mrs. Luh’s qualifications as compared to Block’s. Without evidence to support her claims, Mrs. Luh’s attempt to import contradictions into these statements fails.
Similarly, another claimed contradiction is based upon a misreading of the record. She contends that Huber indicated in its interrogatories that there was a reduction in force plan approved by the company’s human resources department, but insists that, in actuality, there was only a list of potential terminations maintained by Izod. Examining the interrogatory response, however, demonstrates that the two are not in conflict. The response states that “the HEM management requested that HEM Human Resource Managers and/or Directors prepare a reduction in force plan. In response, Tom Izod made the determination that [Mrs. Luh’s and Block’s departments be combined], thereby eliminating a technology manager position.” Mrs. Luh has not introduced any evidence that Izod’s list was not, in fact, the reduction in force plan that was requested by Huber’s human resources department, and we decline to imply a contradiction where none exists based on the record.
Finally, Mrs. Luh attempts to demonstrate an inconsistency in Izod’s testimony concerning Block’s skill at managing employees. The first statement she highlights is drawn from a letter to the EEOC, in which Huber’s counsel stated that “Izod and Edmonds felt that Block was the superior manager in terms of planning, effective leadership, and approaching issues with a team orientation.” This, she says, is contrary to Izod’s statement in his deposition that Block was “an effective manager of people.” Further, Mrs. Luh argues that both of these statements contradict Izod’s knowledge that Block’s staff was very upset at the changes he introduced to his group, particularly reducing their coffee break time.
This claimed contradiction is also without merit. First, although “superior” is often used to describe objectively excellent performance, it is also a comparative term, and the statement in the EEOC letter clearly indicates that Izod and Edmonds believed that Block was a superior manager to Mrs. Luh, and that they were not attempting to provide an objective assessment of his capabilities. This is confirmed by reading the full quotation from Izod’s deposition. He stated that Block is “an effective manager of people. He had some management skills in the past ... you can understand both of them [Block and Mrs. Luh] are young managers.”
Secondly, we disagree that these statements are opposite to the staff complaints. As we have noted, “it is the perception of the decision maker that is relevant ... not *148the opinions of [a plaintiffs] co-workers or other third parties.” Tinsley v. First Union Nat’l Bank, 155 F.3d 435, 444 (4th Cir.1998). As a result, we look not to Block’s subordinates’ complaints, rather to Huber’s own assessment of Block’s performance. Here, Huber presented ample evidence that Izod considered Block’s performance satisfactory, and these staff complaints a good thing. Izod considered the Havre de Grace facility “the laziest place in our organization,” and was pleased that Block had instituted firmer policies to bring the facility more in line with the standards in the rest of the company. At no point did Mrs. Luh introduce evidence contrary to this explanation by Huber.
Thus, we find no error in the district court’s determination that Huber advanced a nondiscriminatory justification for its termination of Mrs. Luh.
B.
Since Huber proffered a nondiscriminatory justification for its actions, the burden then shifted back to Mrs. Luh to establish pretext. The district court determined that she had not done so. She appeals this determination, arguing that the district court improperly disregarded evidence that she had introduced that was sufficient to send the case to the jury.
Mrs. Luh’s claim to have established pretext rests on three separate foundations. First, she contends that the evidence shows that Izod’s evaluation of her and Block’s qualifications was unreasonable. Second, she argues that evidence of Block’s failures in the position demonstrate that she was discriminated against. Finally, she contends that the district court improperly ignored statistical evidence that she introduced. We disagree, and find that the district court’s actions in each of these instances was proper.
On the first issue, we have held that a plaintiff may establish pretext by demonstrating that the decisionmaker conducted the selection process in a “peculiarly informal” manner in order to discriminate against the plaintiff. Dennis, 290 F.3d at 647. Mrs. Luh contends that Izod’s evaluation process fit this category. She focuses on what she sees as Izod failing to investigate her employment record with Huber. Her evidence for this is Izod’s admission at deposition that he' was unaware she was in the company’s Leadership Development Program in 1997 and that he had reviewed only the most recent of her annual performance evaluations. Additionally, she argues that Izod did not consult with any of her prior managers.
Again, we look to our decision in Tinsley that “it is the perception of the decision maker that is relevant ... not the opinions of [a plaintiffs] co-workers or other third parties.” 155 F.3d at 444. Here, it is clear that Izod had established criteria for the promotion that he carried to completion. During his interviews with both Block and Mrs. Luh, Izod stressed that he was emphasizing forward-thinking ability and creativity, rather than merely past performance. Additionally, Izod testified that he received and considered information from both Block and Mrs. Luh’s supervisors, peers and subordinates. Her contention that he disregarded the various letters written in support of her candidacy appears to be based upon nothing more than his decision to select Block for the position rather than her. An assertion such as this, without supporting evidence, is not sufficient to demonstrate pretext and defeat summary judgment.
Apart from improprieties in the selection process, we have held that post-selection performance failures are relevant to a pretext inquiry when the selectee exhibited similar performance failures during the decision process and where the decision-maker was aware of these failures. Grano *149v. Dept. of Development, 637 F.2d 1073, 1081 (6th Cir.1980). Mrs. Luh argues that Block had later difficulties at Huber, which is denied, and that his departure from Huber should ground a finding of pretext.
We decline to consider this issue, as Mrs. Luh did not raise the issue in the district court. Issues raised for the first time on appeal, as this, generally will not be considered. Nat’l Wildlife Fed. v. Hanson, 859 F.2d 313, 318 (4th Cir.1988). In any event, Block’s Performance Summary signed by Izod and Edmonds contradicts the assertion.
Finally, we turn to the district court’s consideration of Mrs. Luh’s statistical evidence. Her position is copied verbatim from her brief on appeal, pp. 58, 59:
Before the restructuring/reduetion in force, the workforce population managed by Izod consisted of 46 employees, of whom 39 were White, two were Black and five were Asian. (JA 854; JA 706-720) Thus, Asians made up less than 11% of Izod’s workforce, while whites accounted for 85% and Blacks 4%.
From this group, Izod selected or helped to select a total of six employees for dismissal in connection with the “restructuring.” Of these, three were Asian, including Plaintiff, and three were White. (JA 720-21) Thus, 50% of the employees selected for “definite” termination in Izod’s plan were Asian, and 50% were white. However, the three whites whom Izod selected for “definite” termination represented only 7.5% of the 39 whites in his workforce. (JA 724) By contrast (as Izod acknowledged), the three Asians whom Izod selected for “definite” termination represented 60% of the five Asians in his workforce. (JA 723) Thus, his Asian employees were eight times more likely than their white co-workers to be selected by Izod for termination in the reduction in force.
In sum, it cannot be denied that the burden of Izod’s termination decisions fell disproportionately upon the Asian employees he supervised, at a rate that evidences unlawful intent.
The district court analyzed this evidence and decided that “Luh has not provided sufficient evidence to suggest that the reasons offered by Huber are false or a pretext for racial or gender discrimination.” There was no expert testimony as to methodology justifying the sought conclusion of discrimination or relevance of the statistics to the plaintiffs claim. In such a case we have held that “[i] f a plaintiff offers a statistical comparison without expert testimony as to methodology or relevance to plaintiffs claim, a judge may be justified in excluding the evidence.” Carter v. Ball, 33 F.3d 450, 457 (4th Cir.1994). Carter relied on Williams v. Cerberonics, Inc., 871 F.2d 452, 455 n. 1 (4th Cir.1989) for the same proposition in which a plaintiff had sought to rely on bare statistics without testimony as to why the statistics were compiled or how they were related to the plaintiffs claim. In the case before us, in addition to the absence of such evidence relating the statistics to the case, the district court noted the small number of people involved, that some of the Asian employees involved were from a different facility over which Izod only had partial responsibility and that Huber articulated an individual reason for Mrs. Luh’s layoff.
We conclude that the district court properly analyzed the evidence presented in this case. We find no error in the reasoning set forth by the district court in its opinion, either as to Huber’s demonstration of a legitimate, nondiscriminatory justification or as to Mrs. Luh’s failure to establish the necessary showing of pretext.
The judgment of the district court is accordingly
AFFIRMED.
. The number of employees involved in the layoff may vary slightly, depending on the place in the record which is read.
. The parties considered the claim of Mrs. Luh, that one Martin had suggested a sexual liaison, to be a claim quid pro quo. Martin denied the claim.