Opinion for the Court filed by Circuit Judge TATEL.
Opinion concurring in part and concurring in the judgment filed by Circuit Judge SENTELLE.
In this wage discrimination action, the district court found that neither plaintiff established a prima facie case and granted summary judgment for the employer. Although we agree with the district court with respect to one plaintiff, we find the other plaintiff satisfied the minimal evidentiary burden needed to raise a genuine issue of material fact with respect to his prima facie case. We therefore affirm in part, reverse in part, and remand.
I
Appellants Melvia Boling and Edward Coward are African Americans employed by appellee ADT Security Systems, Inc. A supplier of electronic security systems and services, ADT maintains a salary grade structure consisting of grades E-3 through E-20. Each grade has five salary “steps.” Salary ranges within grades overlap: A top-step E-8, for example, earns approximately $11,000 more than a bottom-step E-9. ADT assigns *273job titles, codes, and grades to every employee. Employees with the same job title may have different codes, different grades, and even different duties; together, all of these factors determine salary. Job titles generally reflect duties performed and serve as important, although not dispositive, factors in setting salary.
Employed by ADT since the late sixties, Melvia Boling became a “Data Supervisor” in 1988 with a grade of E-8 and a salary of approximately $85,000. While working on a temporary project in 1995, Boling was given the title “Project Manager.” She received no related increase in either grade or salary. Al ADT Project Managers are white and all but two earn more than Boling. By contrast, Boling earns more than all Data Supervisors.
An E-8 for most of the last seventeen years, Edward Coward was briefly promoted in July 1994 to the position of Telecommunications Network and Facilities Manager (“TNFM”), with a grade of E-9. Approximately nine months later, ADT reclassified him from E-9 back to E-8, later changing his title to “Technical Support.” Coward earns less than most TNFMs, all of whom are white; he earns more than most Technical Support staff.
Alleging that ADT pays them less than their white counterparts in violation of 42 U.S.C. § 1981 (1994), Boling and Coward brought suit in the United States District Court for the District of Columbia. Claiming that neither plaintiff had established a prima facie ease of wage discrimination, ADT moved for summary judgment. The district court agreed, finding that Boling was a Data Supervisor, not a Project Manager; that Coward was Technical Support, not a TNFM; and that so viewed, neither had identified any similarly situated, better-paid white employees.
Boling and Coward now appeal the district court’s grant of summary judgment for ADT. Our review is de novo. Tao v. Freeh, 27 F.3d 635, 638 (D.C.Cir.1994). Mthough we draw all justifiable factual inferences in favor of Boling and Coward, they bear the burden of pointing to “affirmative evidence” establishing a genuine factual dispute. Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 255-57, 106 S.Ct. 2505, 2513-15, 91 L.Ed.2d 202 (1986). “If material facts are at issue, or, though undisputed, are susceptible to divergent inferences, summary judgment is not available.” Tao, 27 F.3d at 638 (citing Alyeska Pipeline Serv. Co. v. U.S. EPA, 856 F.2d 309, 314 (D.C.Cir.1988)).
II
Because Boling and Coward allege intentional wage discrimination, we apply the familiar McDonnell Douglas burden-shifting test. See McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802 & n. 13, 93 S.Ct. 1817, 1824 & n. 13, 36 L.Ed.2d 668 (1973). Only the first stage of that test—the prima facie case—is at issue here. To establish a prima facie case, Boling and Coward must show by a preponderance of the evidence, Texas Dep’t of Community Affairs v. Burdine, 450 U.S. 248, 252-53, 101 S.Ct. 1089, 1093-94, 67 L.Ed.2d 207 (1981), membership in a protected class (uncontested by ADT), and “ ‘that [they] were performing work substantially equal to that of [white employees] who were compensated at [] higher rate[s] than [they were]’ ” (vigorously contested by ADT). Aman v. Cort Furniture Rental Corp., 85 F.3d 1074,1087 (3d Cir.1996) (quoting Hohe v. Midland Corp., 613 F.Supp. 210, 214 (E.D.Mo.1985), ajfd, 786 F.2d 1172 (8th Cir.1986)) (alterations in original); see also Miranda v. B & B Cash Grocery Store, Inc., 975 F.2d 1518, 1529 (11th Cir.1992) (prima facie ease consists of showing membership in protected class and that plaintiffs job is similar to higher paying jobs occupied by non-class members). Mthough “minimal,” St. Mary’s Honor Ctr. v. Hicks, 509 U.S. 502, 506, 113 S.Ct. 2742, 2747, 125 L.Ed.2d 407 (1993), and “not onerous,” Burdine, 450 U.S. at 253,101 S.Ct. at 1094, the prima facie case “serves an important function,” id. at 253-54, 101 S.Ct. at 1094:
[I]t eliminates the most common nondiscriminatory reasons for the plaintiffs rejection____ [T]he prima facie case raises an inference of discrimination only because we presume these acts, if otherwise unexplained, are more likely than not based on the consideration of impermissible factors. Establishment of the *274prima facie case in effect creates a presumption that the employer unlawfully discriminated against the employee.
Id. at 254, 101 S.Ct. at 1094 (citations and quotation marks omitted). If a reasonable factfinder could infer a genuine issue of material fact with respect to any element of the prima facie case—such as whether white employees performing substantially similar work earn more than the plaintiff—summary judgment at that stage is inappropriate.
To establish their prima facie cases, Boling and Coward relied on two types of evidence: a multiple regression analysis comparing salaries of minority and non-minority employees throughout the company; and comparisons of their actual individual salaries to those of immediate co-workers. We consider this evidence in Parts III and IV.
Ill
Controlling only for race and seniority, the multiple regression analysis shows that throughout ADT’s Mid-Atlantic Division, the company pays African American employees approximately twelve percent less than white employees. Because the regression analysis failed to account for education or prior work experience, the district court excluded it from consideration.
The Supreme Court established the rule governing admissibility of regression analyses in discrimination cases in Bazemore v. Friday, 478 U.S. 385, 106 S.Ct. 3000, 92 L.Ed.2d 315 (1986):
While the omission of variables from a regression analysis may render the analysis less probative than it otherwise might be, it can hardly be said, absent some other infirmity, that an analysis which accounts for the major factors must be considered unacceptable as evidence of discrimination. Normally, failure to include variables will affect the analysis’ probativeness, not its admissibility.
Id. at 400, 106 S.Ct. at 3009 (Brennan, J., concurring in part, joined by all Justices) (citations and quotation marks omitted). Qualifying this rule, the Court added that “[tjhere may, of course, be some regressions so incomplete as to be inadmissible as irrelevant.” Id. at 400 n. 10, 106 S.Ct. at 3009 n. 10. We too have explained that although “a defendant cannot undermine a regression analysis simply by pointing to variables not taken into account that might conceivably have pulled the analysis’s [sic] sting ...[,] Bazemore [does not] require acceptance of regressions from which clearly major variables have been omitted—such as education and prior work experience.” Roger v. Reno, 98 F.3d 631, 637 (D.C.Cir.1996) (citations omitted).
“Major factors” that a regression analysis must include depend on the facts and theory of the particular case. In Bazemore, for example, the Court found that omission of a variable accounting for differences in salaries by county, though relevant to plaintiffs’ claim of state-wide wage discrimination, did not require exclusion of the regression analysis. Bazemore, 478 U.S. at 401-03, 106 S.Ct. at 3009-10. Depending on the theory of the case, some variables may be entirely unsuitable. Where plaintiffs allege discriminatory promotion practices, for example, this court considers inclusion of grade variables “inappropriate” because an employee’s grade may itself reflect discrimination. Valentino v. U.S. Postal Serv., 674 F.2d 56, 72 n. 30 (D.C.Cir.1982) (citation omitted).
Applying these standards- to this case, we find that the regression analysis was “so incomplete as to be inadmissible as irrelevant,” Bazemore, 478 U.S. at 400 n. 10, 106 S.Ct. at 3009 n. 10, though for a reason not considered by the district court. The regression analysis failed to account for job title or any other variable representing type of work performed. This omission is fatal because Boling and Coward claim that they should be compared to employees in other job categories who perform similar work but who earn more than they. According to their own theory of the case, therefore, job title, or some other measure of type of work, serves as a “major factor” within the meaning of Bazemore. But because the regression analysis compares all employees in all job categories without accounting for type of work, it tells us nothing about whether employees who perform similar work are paid different*275ly. Boling and Coward therefore cannot use the regression analysis to show that similarly situated white employees earn more. Since the regression analysis is flawed as a matter of law, we need not decide whether the reason given by the district court—the omission of education and prior experience variables— would independently have invalidated the analysis.
IV
This brings us to Boling’s and Coward’s efforts to establish prima facie cases by showing that they personally earn less than similarly situated whites. The district court held that neither did so; we agree with respect to Boling, but not Coward.
Because Boling claims that she is underpaid given her duties, experience, and seniority, and because she earns more than all Data Supervisors, her ease depends on her assertion that she is similarly situated to Project Managers, all of whom are white and most of whom earn more than she. In support of her claim, Boling points to ADT’s own records which refer to her as a Project Manager. She also argues that during the pendency of this case, ADT changed her job title from Project Manager back to Data Supervisor. Admitting that Boling briefly held the title of Project Manager, ADT argues that the title was temporary, that it should have been changed when the project ended, and that Boling performs none of the duties of other Project Managers. ADT points out that Boling’s own complaint identifies her as a Data Supervisor. It also relies on her deposition testimony acknowledging not only that she was unaware of the title change at the time, but that she knows nothing about Project Manager duties.
Under these circumstances, we agree with the district court that Boling failed to create a genuine issue of material fact with respect to a critical element of her prima facie ease— that she is similarly situated to Project Managers. The “Project Manager” title in ADT’s records is her only evidence suggesting that her duties and skills amount to anything other than those of a Data Supervisor. She has pointed to no other evidence that would support an inference that she is a Project Manager or should be paid like one. Although job titles usually serve as strong evidence of an employee’s actual skills and duties, see, e.g., Burger v. New York Inst, of Tech., 94 F.3d 830, 833 (2d Cir.1996) (under ADEA, similarity of jobs could be inferred from parallel job titles), Boling’s..admission that she knew nothing about the duties of Project Managers or even that she held the title, together with ADT’s undisputed explanation that the title was temporary, eliminates any genuine dispute about the real nature of her work. We thus compare Boling’s salary not to salaries of Project Managers, but to salaries of Data Supervisors. Because no Data Supervisor makes more than she, the district court properly granted ADT summary judgment with respect to Boling.
Unlike Boling, Coward has pointed to evidence that creates a genuine factual issue about whether similarly situated white employees earn more than he. Coward’s service record reflects his “promotion” to the position of TNFM. Unlike Boling, who testified that she knew nothing about the duties of Project Managers, Coward described himself and his duties as managerial and supervisory, consistent with TNFM status. ADT “aeknowledge[d] there is some overlap between [Coward’s] position and that of a TNFM.” Appellee’s Br. at 15. Indeed, at oral argument counsel for ADT conceded the existence of a genuine issue of material fact over whether Coward is a TNFM. Accordingly, Coward’s TNFM status should not have been resolved at summary judgment.
The district court also found that Coward failed to raise a genuine issue of material fact about whether he was either the lowest graded or lowest paid TNFM. With respect to the question of grade, the district court “sup-posted] that a reasonable juror could conclude that white TNFMs were slotted in higher grades than Coward,” but nevertheless rejected this possible evidence of wage discrimination, pointing out that under ADT’s overlapping salary structure Coward theoretically could earn as much as higher-graded TNFMs. But the court had before it ADT’s own chart listing all TNFMs and showing that employees with higher grades *276generally have higher salaries. A reasonable juror thus could infer wage discrimination from Coward’s lower grade. By rejecting this inference in Coward’s favor, the district court usurped the jury’s factfinding role.
The company’s list of TNFMs further supports Coward’s contention, also rejected by the district court, that he earned less than most TNFMs. According to the chart, eight TNFMs (one E-9 and seven E-lOs) earn more than Coward and only three (one E-8 and two E-9s) earn less. All TNFMs listed are white. That some TNFMs make less than Coward, a fact heavily relied on by ADT, makes no difference. Coward need not show that all similarly situated white employees receive better treatment. Identifying some similarly situated employees—even one—establishes a prima facie case. See Parker v. Secretary, U.S. Dep’t of Hous. and Urban Devel, 891 F.2d 316, 321 (D.C.Cir.1989) (“[T]he fact that ... a black female was denied educational opportunities when another white male employee was not, gives rise to an inference of discriminatory treatment.”); Pittman v. Hattiesburg Mun. Separate Sch. Dist., 644 F.2d 1071, 1074 (5th Cir. Unit A May 1981) (“The usual unequal pay case involves two employees, one black and one white, employed at the same time and doing substantially the same work.”). Having identified eight higher-paid white employees, the majority of TNFMs, Coward has satisfied the minimal requirements of a prima facie case.
V
Since the district court ceased its inquiry at the prima facie stage, deciding neither whether ADT proffered a nondiscriminatory reason for Coward’s salary and grade nor whether Coward rebutted such a proffer, see McDonnell Douglas Corp., 411 U.S. at 802, 93 S.Ct. at 1824, we reverse and remand Coward’s case so the district court can complete the summary judgment inquiry. As to Boling, we affirm.
So ordered.