ORDER
Barbara Myklebust worked as a research scientist with the United States Department of Veterans Affairs (VA). At the same time, she also worked as an Assistant Professor of Neurology at the Medical College of Wisconsin (MCW). After Myklebust lost her research position at the VA, MCW fired her. She then filed this Title VII action against MCW alleging retaliation due to her prior formal complaint of sexual harassment against another research scientist at the VA who also served as a professor at MCW. The district court granted MCW’s motion for summary judgment. We affirm.
I.
Myklebust worked full-time at MCW from 1985 until she was fired in May 1996. She was an Assistant Professor in MCW’s Department of Neurology. On August 30, 1995, she received a promotion to Associate Professor in the “research track.” Along with this promotion, she was given a three-year contract to run from July 1, 1995, until June 30, 1998. She reported to Dr. Cass Terry, MCW’s Neurology Department Chairman. Terry reported to Dr. Michael Dunn, the Dean of MCW and also its Executive Vice President.
During all relevant times, Myklebust also had a parallel appointment with the VA, working out of a VA Research Lab nearby to MCW. She spent the vast majority of her working time at the VA and she estimates that only ten percent of her time was spent at MCW. Her annual salary was *654$65,000, with only $10,000 of that amount coming from MCW.
Myklebust’s documented problems started due to an alleged sexual harassment incident committed by Dr. Neil Mandel, a Professor of Medicine at MCW and Associate Chief of Staff for Research at the VA. Mandel split his time with MCW and the VA under an arrangement similar to Myklebust’s. In 1993, Myklebust formally charged Mandel with sexual harassment. Her written charge was treated as an internal complaint at both MCW and the VA.
Myklebust asserts that during the 1993 internal proceedings, her supervisor Terry actively participated as part of MCW’s investigation and as a witness in the VA investigation. The MCW investigation revealed that Terry “encouraged Dr. Myklebust to not file a charge as being a career-limiting move.” According to MyMebust, after the 1993 internal investigation concluded, Mandel would keep Terry updated on MyMebust’s EEOC proceedings against Mandel. On April 4, 1995, Mandel sent Terry a three-page letter with attachments concerning MyMebust. This letter, which MyMebust called disparaging, questioned her motives in maMng the harassment charge and accused her of possible ethics violations. According to MyMebust, Mandel wrote the letter because in an earlier phone call, Terry requested that Mandel forward documentation to Terry. MyMebust also complains that during a December 1995 faculty meeting, Terry criticized an unnamed faculty member over a salary issue, and many faculty members understood that he was referring to MyMebust. She then complained to Sheehan, MCW’s Director of Faculty Affairs, that Terry’s comment “was in retaliation for EEO complaints .... ”
MyMebust finally alleges that Mandel met with Dean Dunn in 1995 and together they “reviewed various issues regarding Dr. Barbara MyMebust and the EEOC complaints she has filed.” Myklebust claims that in that meeting Dunn had serious concerns about whether she had misrepresented her academic rank and faculty title. Shortly thereafter, Mandel wrote a letter to Dunn amplifying MyMebust’s alleged academic misrepresentations. Dunn stated that because he did not follow up on Mandel’s letter, the academic misrepresentation allegations were obviously not important to him.
At MCW, MyMebust was primarily paid through grants she obtained for her VA research. In early 1996, the VA notified MyMebust that her grant application was not approved. MyMebust had submitted this grant application in order to obtain funding for her VA lab for a three-year period. Shortly thereafter, in February 1996, the VA advised MyMebust of a pending Reduction in Force (RIF). Due to the RIF, the VA eliminated MyMebust’s research position. On March 4,1996, the VA offered MyMebust a wow-research position of Health Science Specialist and she accepted the position, with the stipulation that she desired to return to research if additional opportunities became available. Her VA salary of $55,000 was to continue in the Health Science Specialist position.
After receiving notice of the RIF eliminating MyMebust’s research position at the VA, Dean Dunn of MCW terminated her MCW position. In a letter dated May 6,1996, Dunn stated that due to the loss of her VA research responsibilities, the basis for her faculty appointment with MCW was removed. Dunn concluded that Myklebust’s MCW employment would end May 12, 1996, the date she was to assume her new VA Health Science Specialist position. MyMebust acknowledges that MCW did not maintain an ownership interest or otherwise control VA operations..
*655Myklebust filed suit against MCW in 1998 alleging sexual discrimination and retaliation under Title VII, as well as breach of employment contract. The district court granted MCW’s motion for summary judgment on the discrimination and retaliation counts. The parties stipulated to the dismissal of the breach of contract claim, and Myklebust appeals only the court’s decision on retaliation.
II.
We review the district court’s decision to grant summary judgment de novo. See, e.g., Bell v. Environmental Protection Agency, 232 F.3d 546 (7th Cir.2000). Title VII prohibits an employer from discriminating against an employee because that employee has opposed any practice deemed unlawful under the Act. 42 U.S.C. § 2000e-3(a). The plaintiff may establish a prima facie case of retaliation and overcome defendant’s motion for summary judgment using either the direct method or the indirect method. See Rhodes v. Illinois Dept. of Transportation, 359 F.3d 498, 508 (7th Cir.2004).
Under the direct method, the plaintiff must present direct or circumstantial evidence of (1) a statutorily protected activity; (2) an adverse action taken by the employer; and (3) a causal connection between the two. Id. Under the indirect method, the plaintiff must show that (1) she engaged in a statutorily protected activity; (2) she performed her job according to her employer’s legitimate expectations; (3) despite her satisfactory job performance, she suffered an adverse action from the employer; and (4) she was treated less favorably than similarly situated employees who did not engage in statutorily protected activity. Id; Hilt-Dyson v. City of Chicago, 282 F.3d 456, 465 (7th Cir.2002); Haywood v. Lucent Techs., 323 F.3d 524, 531 (7th Cir.2003).
If the plaintiff establishes these elements under the indirect method, the burden shifts to the defendant to come forward with a legitimate, non-invidious reason for its adverse action. Hilt-Dyson, 282 F.3d at 465. Although the burden of production shifts to the defendant under this method, “the burden of persuasion rests at all times on the plaintiff.” Haywood, 323 F.3d at 531 (citing Klein v. Trustees of Indiana Univ., 766 F.2d 275, 280 (7th Cir.1985)). Once the defendant presents a legitimate, non-invidious reason for the adverse action, the burden remains with the plaintiff to show that the defendant’s reason is pretextual. Hilt-Dyson, 282 F.3d at 465.
Here, as an initial matter, there is disagreement over who was the actual decisionmaker. Myklebust claims that Terry was involved in her firing, along with Dean Dunn, while MCW argues that Dunn alone made the decision. This appears to be a legitimate factual dispute, but it is not a material dispute to preclude summary judgment. Assuming that both Terry and Dunn were decisionmakers, there is no evidence of statements by either amounting to direct evidence of retaliation. Instead, Myklebust complains that Terry commented to her in 1993 that filing a discrimination charge could be a “career limiting move.” She also claims that Terry made a remark at a 1994 faculty meeting regarding a salary issue with an unnamed faculty member, supposedly Myklebust. Finally, she submits the angry, retaliatory letters and phone calls of Mandel, her alleged harasser, received by Terry and Dunn.
As for circumstantial evidence under the direct method, none of the above evidence submitted by Myklebust comes close to the “convincing mosaic” of proof necessary to proceed under this method — nor does she even argue causation, a required element *656of a direct case. See, e.g., Oest v. Illinois Dept. of Corrections, 240 F.3d 605, 611 (7th Cir.2001) (“[I]f the remarks are not contemporaneous with the discharge or causally related to the discharge decision making process they are insufficient to create a triable issue of material fact regarding discrimination.”). It is important to emphasize that Myklebust’s protected activity (her sexual harassment complaint) occurred in 1993, but her firing was not until 1996. This court has previously found a delay of as little as three months between the protected activity and the alleged retaliation to be “too attenuated to raise an inference of discrimination.” Id. at 616 n. 8. Here, not only is there a three-year delay from the time Myklebust complained, but she also received a promotion in 1995. These facts are similar to Albrechtsen v. Board of Regents of the University of Wisconsin System, 309 F.3d 433 (7th Cir.2002), in which this court overturned a jury verdict in the plaintiffs favor on a retaliation claim, partly because the plaintiff had received a promotion after the protected activity and before the adverse action occurred. For these reasons, even if Myklebust had cited legal authorities addressing her burden to show causation under the direct method, the three-year gap between her protected activity and her firing (along with her intervening promotion) nevertheless foreclose her from presenting a jury issue.
Turning to the indirect method, her prima facie case fails on two separate and independent elements. An element of the prima facie case is that the employee must show that she was adequately performing her job at the time of termination. Here, as stated above, the VA eliminated Myklebust’s research position due to the RIF. MCW cannot be held accountable for the actions of the VA, as Myklebust concedes. In addition, Myklebust concedes that after the RIF, she accepted a full-time job at the VA with no research responsibilities in a program totally unrelated to MCW’s endeavors. She agrees the basis for her MCW appointment thus ceased to exist. See App. 100-01, UK 117, 123. Accordingly, although Myklebust can blame the loss of her job at MCW on the VA RIF, she cannot dispute that due to the loss of her VA lab facilities, she was no longer adequately performing the research MCW required.1 Myklebust counters by arguing, in essence, that the MCW employee handbook did not mention or contemplate a firing due to a VA RIF in its “termination section.” This backdoor argument attempting to establish retaliation through the alleged breach of an employment contract is irrelevant because Myklebust has dropped her breach of contract claim.
The analysis need not proceed any further because Myklebust fails one element of the prima facie case. However, in the interests of a thorough analysis, Myklebust also fails to meet the final element of a prima facie case, i.e., that a similarly situated faculty member who did not engage in protected activity received preferential treatment. She correctly points out that on previous occasions when a faculty member with an employment contract had temporarily lost grant funding, MGW had accommodated the employee. However, to establish this element of the prima facie case, Myklebust must point to another employee “directly comparable in all material respects.... ” Patterson v. Avery Dennison Corp., 281 F.3d 676, 680 (7th Cir.2002). She is unable to point to another instance where a different employee lost funding, lost their labs, and accepted a full-time, *657non-research position unrelated to MCW’s programs.
III.
The district court correctly granted MCW’s motion for summary judgment. Myklebust is unable to show retaliation under the direct method because three years elapsed between her protected activity and her firing — and she received a promotion in the interim. She cannot establish a prima facie case under the indirect method because she was not adequately performing her research job at MCW due to the loss of research facilities at the VA. For these and the foregoing reasons, we AFFIRM the decision of the district court.
. MCW apparently does not own hospital research facilities of its own, but instead contracts out with other hospital facilities for research space.