Don Freeman had been working as a Custodial Worker for the Madison Metropolitan School District ("MMSD") for 13 years when, in 1992, he injured his knee while trying to roll a wrestling mat onto a cart. He did not return to work until 1995, at which time he had lost his seniority and was considered a Custodial Worker Trainee. The course of events between his injury and his return form the basis of his Title VII claim that he was discriminated against because of his race. Freeman, who is African-American, argues that MMSD refused to allow him to return to work on a modified basis to accommodate his physical limitations even though MMSD's policy was to provide such work modifications and that policy was followed for white employees.

A trial was held on the Title VII claim, but at the close of Freeman's case the district court granted MMSD's motion for a directed verdict. Freeman appeals that ruling as well as the court's pretrial rulings excluding some evidence.

The evidence at trial revealed numerous and repeated letters, conversations, and documents relating to Freeman's physical condition and his capacity to work. A summary of the more critical evidence is necessary to understand the contentions in this case. To contextualize that evidence, we first examine the physical requirements of the Custodial Worker I position that he was performing prior to the injury. That position involved heavy work, which included constant standing and walking, occasional ladder work, and frequent climbing, squatting and crouching. It further required lifting or carrying 21-50 pounds to waist level occasionally and 51-100 pounds rarely. Prior to his injury, Freeman was assigned to the position of Custodial Worker I Laundry Room Attendant, which he testified was much less strenuous than a regular Custodial Worker I position.

After the injury in April 1992, Freeman's ability to fulfill the physical requirements of the job was severely restricted. On April 28, 1992, his treating physician, Dr. Harrington, completed a physical capabilities form which stated that Freeman was limited to performing light medium work, including lifting 30 pounds maximum, and up to 20 pounds frequently. He further indicated that Freeman could stand no more than « hour per day, walk for one hour per day, and could not climb, squat, kneel, bend, stoop or crouch. Dr. Harrington followed that report with a July 1992 form indicating that Freeman could return to full-time sedentary work with minimal standing and no lifting, bending, stooping or climbing. In response to Freeman's claim for workers compensation benefits, MMSD required an evaluation by Dr. Leonard of the University of Wisconsin Hospital and Clinics, Spine- Sports Medicine Center. Dr. Leonard opined that Freeman could return to work within the limits of his pain and with knee braces. A subsequent functional capacities evaluation report by UW Hospital & Clinics in February 1993, however, found that Freeman could perform only light medium work. That evaluation noted that Freeman's physical limitations in standing, walking and lifting did not meet the requirements of the position. Freeman was paid $47,500 in settlement of his worker compensation claim based upon the permanent partial disability in his knees.

In approximately May or June 1994, Freeman sought to return to his Custodial Worker I position. In a letter of May 1994, Dr. Harrington stated that Freeman was capable of performing light work, and that his limitation to sedentary work no longer applied. Dr. Harrington further recommended that Freeman undergo a work capacity evaluation to address questions regarding Freeman's capabilities in specific job situations. Later that year, in September 1994, Dr. Harrington released Freeman to return to work with some limitations on lifting, including a maximum of 35 pounds, and some limits on squatting and climbing. The restrictions placed him in the medium range of work. Meanwhile, Freeman was engaging in ongoing efforts to return to work with MMSD--to no avail. Dr. Harrington submitted a letter in December 1994 stating that Freeman might be able to perform the job duties of game room monitor (a light work position), craft room custodian (medium work), or Custodial Worker II which Dr. Harrington indicated was similar to Custodial Worker I but involved more supervision and only moderate amounts of squatting and climbing. Because the game room monitor position was only part-time, however, Freeman was uninterested in it unless it could lead to full-time work.

Eventually, in February 1995, the functional evaluation recommended by Dr. Harrington was performed with a goal of determining Freeman's safe functional level for the Custodial Worker II position. That evaluation determined that he was functioning at the heavy level but that he could not perform some of the job requirements. For instance, he was safe for up to a 32-pound repetitive lift, but the position specified 50 pounds. His maximum stand-up lift was 65 pounds, as opposed to the position maximum of 80. Finally, he was safe for occasional chest lifts of 45 pounds which was incompatible with the 50 pound requirement of the position. In response, Dr. Harrington sent a letter the next month indicating that physical deconditioning resulting from the delay in the return to work accounted for some of the functional limits, and that Freeman could return to his job after rehabilitation.

The efforts to return Freeman to his position continued in the ensuing months, and in September 1995 a conference was held at the State Workers Compensation Division to explore the cause of the delay in his reinstatement. MMSD and Freeman agreed to devise a plan for returning him to work, and sought Dr. Harrington's opinion regarding the necessary physical conditioning. Dr. Harrington recommended that Freeman undergo a work hardening and rehabilitation program at Meritor Hospital, and obtain a more current work capacity evaluation. MMSD initially opposed this recommendation, arguing that Dr. Harrington's connections with Meritor rendered it unacceptable because of the possibility that he could influence the outcome. In December 1995, however, MMSD agreed to the plan, and Freeman completed that program in February 1996. The reports from Meritor indicated that he was a very motivated participant and that he progressed rapidly. The final report indicated that he met the requirement for heavy work, and could lift up to 100-105 pounds rarely (up to 10% of the day), 50 pounds occasionally (up to 33% of the day), and 25 pounds frequently (66% of the day). It further stated that his capabilities were consistent with the requirements of the position as set forth in the job description. The report acknowledged that its conclusions constituted recommendations that were subject to Dr. Harrington's acceptance. On February 17, 1996, Dr. Harrington sent a letter stating that in light of the Meritor report, a trial return to work was appropriate and that Freeman should be judicious in the amount of squatting and stair climbing he performed, particularly with heavy loads. That language in his letter was fodder for yet another correspondence war between MMSD and Freeman's attorney. MMSD asserted that it required formal restrictions so that it could determine what work was appropriate for Freeman, and that it could not be expected to interpret "judicious." It further sought clarifica tion of the language indicating a "trial" return. After a less ambiguous response was received, Freeman was finally allowed to return to work. Because the lengthy delay resulted in the loss of his seniority under the contract, Freeman was required to begin as a part-time Custodial Worker Trainee.

The parties present vastly different characterizations of the protracted process described above. According to MMSD, it was willing to return Freeman to work as soon as he was physically capable of doing so, but he did not receive an unrestricted release until March 1996. MMSD further asserts that it was distrustful of Dr. Harrington's switch from indicating permanent partial disability to stating that Freeman could return to work. Freeman, on the other hand, acknowledges that he could not return to unrestricted work for some time, but argues that the MMSD policy is to return workers to a modified position until they are back at full capacity. He argues that MMSD's refusal to return him to work until he was unrestricted was a result of race discrimination.

At the close of Freeman's case-in-chief, the district court granted MMSD's motion for a directed verdict. The court held that Freeman met his burden of establishing a prima facie case of race discrimination. The court then declared that once the prima facie case is established, the jury must determine whether race was a motivating factor in MMSD's decision not to rehire and whether MMSD's decision would have been the same regardless of Freeman's race. Transcript at 184. In its written memorandum and order on the motion, the court phrased it slightly differently but with the same import, stating that the prima facie case was met and "[t]he issue before the jury is whether the defendant discriminated against plaintiff because of his race." Ct. Order at 2. The court went on to state that a plaintiff must establish intentional discrimination, and that no facts were presented from which race discrimination could be inferred.

We review de novo the grant of the directed verdict under Federal Rule of Civil Procedure 50(a). Payne v. Milwaukee County, 146 F.3d 430, 432 (7th Cir. 1998). In considering whether it was properly granted, we must consider the evidence in the light most favorable to the non-moving party to determine whether there was no legally sufficient evidentiary basis for a reasonable jury to find for the non-moving party. Id. "Credibility determinations, the weighing of the evidence, and the drawing of legitimate inferences from the facts are jury functions, not those of a judge, whether he is ruling on a motion for summary judgment or for a directed verdict." Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 255 (1986).

It is well-established that a plaintiff may establish a Title VII violation even absent direct evidence of race discrimination through the burden- shifting method of McDonnell-Douglas Corp. v. Green, 411 U.S. 792 (1973). Kaniff v. Allstate Ins. Co., 121 F.3d 258, 263 (7th Cir. 1997). Because Freeman lacked any direct evidence of race discrimination, the court properly recognized that he was proceeding under the McDonnell-Douglas test. Under that test, a plaintiff must first establish a prima facie case of discrimination based on race. Stewart v. Henderson, 207 F.3d 374, 376 (7th Cir. 2000). If that is accomplished, the plaintiff has established a presumption of discrimination, and the defendant then bears the burden of production to provide a legitimate, non-discriminatory reason for the challenged action. Id. Once the defendant meets that burden, the plaintiff must establish that the reasons proffered by the defendant were pretextual, by presenting direct evidence that his race played a role in the challenged action or indirectly by creating a genuine issue of material fact regarding the sincerity of the proffered reasons for that action. Collier v. Budd Co., 66 F.3d 886, 892 (7th Cir. 1995). Indirect evidence of pretext showing that an employer's proffered reasons are not credible can include evidence that the reasons are without basis in fact, did not actually motivate the challenged action, or were insufficient to motivate the discharge. Id.

The district court, however, erroneously applied that test in ruling on the motion for a directed verdict. After holding that Freeman indeed met his prima facie burden, the court did not pause to consider whether evidence was introduced that MMSD had a legitimate, non- discriminatory reason for the challenged action. Instead, the court considered only whether Freeman had ultimately proven discrimination. That step is critical, not only because MMSD bears the burden of production at that stage, but because the pretext analysis necessarily focuses on the reason provided. See Plair v. E.J. Brach & Sons, Inc., 105 F.3d 343, 349 (7th Cir. 1997) (to prove pretext, plaintiff must squarely rebut the specific reason articulated by the defendant).

Our examination of the trial transcript reveals little regarding MMSD's explanation for its ongoing refusal to return Freeman to work. That is unsurprising given the context of this appeal--a motion for a directed verdict granted before MMSD presented its case. The only testimony relevant to that second prong is the testimony of adverse witness Robert Nadler, who was the benefits manager for MMSD. Nadler testified that MMSD had a policy of allowing workers with temporary disabilities to return to work and accommodations are made to allow them to continue working until they regain their full ability. He further testified that MMSD's position was that Freeman had a permanent rather than temporary disability, and that position was based upon Dr. Harrington's medical opinion. Dr. Harrington, of course, repeatedly authorized Freeman's return to work and documented Freeman's steady physical improvement, but Nadler asserted that MMSD was confused by Dr. Harrington's later opinions that the disability was not permanent given his initial diagnosis. Thus, the only race-neutral explanation provided at this point in the trial is MMSD's contention that it believed he was permanently disabled based on Dr. Harrington's initial opinion that he had a permanent partial disability of his knees, and thus he did not fall within the policy covering temporarily-disabled employees.

Freeman, however, presented abundant evidence indicating that MMSD could not honestly have believed that his disability was permanent, and thus indicating that the race-neutral explanation was pretextual. From the time of Dr. Harrington's initial assessment of permanent partial disability in November 1992, substantial medical evidence demonstrated that his condition was improving. A functional capacities evaluation in February 1993 indicated that he could perform light medium work, although Nadler testified that in awarding workers compensation benefits that report was disbelieved in favor of Dr. Harrington's testimony. Moreover, Dr. Harrington specifically disavowed his earlier prognosis, indicating that Freeman's improvement had exceeded his expectations. Dr. Harrington's letters chart improvement in his limitations from sedentary work in 1992, to light work in May 1994, to medium work in September 1994. Any lingering doubts regarding Dr. Harrington's credibility could have been laid to rest by the functional capacities evaluation in February 1995, which indicated that he was functioning at the level of heavy work. Although that evaluation indicated that he still had some restrictions on lifting, it rebuts Nadler's testimony that MMSD believed Freeman was permanently disabled. In addition, at the conference at the State Workers Compensation Division, MMSD agreed to a plan designed to allow for Freeman's return to work, which also refutes Nadler's statement that MMSD believed him to be permanently disabled and incapable of improving enough to resume his job. Finally, Nadler's contention that it continued to believe Freeman was permanently disabled because it was "confused" by Dr. Harrington's change in position is belied by MMSD's failure to seek a second opinion to allay its confusion. The documented progression in Freeman's physical ability and MMSD's own action--and inaction--thus provides evidence that its asserted race-neutral explanation was not honestly held. That is sufficient to raise a jury issue of pretext, and thus of discrimination. Of course, MMSD may have an entirely different explanation that it has not yet had an opportunity to assert, or it may well be able to establish that its reasons are not pretextual. We express no opinion on the merits, but merely hold that the court erred in granting the motion for a directed verdict.

Freeman sought to introduce additional evidence of discrimination but was prevented from doing so by some pretrial rulings. In light of our holding that the directed verdict was improper, we must address those challenges because they will impact the retrial. First, the district court held that the statute of limitations precluded evidence of discrimination that occurred prior to September 13, 1995, which was the last date within the 300-day period of limitations. See 42 U.S.C. sec. 2000e- 5(e). The court rejected Freeman's contention that the actions prior to that date were part of a continuing violation not barred by the statute of limitations.

It is well-established that a Title VII plaintiff may recover for acts beyond the limitations period if she can demonstrate that such acts were part of a "continuing violation." Jones v. Merchants Nat. Bank & Trust Co. of Indianapolis, 42 F.3d 1054, 1058 (7th Cir. 1994); United Air Lines, Inc. v. Evans, 431 U.S. 553 (1971). The continuing violation doctrine applies only if the plaintiff identifies acts of discrimination that occurred within the limitations period as well, rather than simply the persisting effects of past discrimination. Merchants Nat. Bank, 42 F.3d at 1058. In determining whether the pre-limitations period conduct constitutes a continuing violation rather than discrete acts of discrimination, the court considers factors such as: "(1) whether the acts involve the same subject matter; (2) the frequency at which they occur; and (3) the degree of permanence of the alleged acts of discrimination, 'which should trigger an employee's awareness of and duty to assert his or her rights.'" Filipovic v. K & R Exp. Systems, Inc., 176 F.3d 390, 396 (7th Cir. 1999), quoting Selan v. Kiley, 969 F.2d 560, 565 (7th Cir. 1992). Courts have identified a number of different fact patterns that indicate continuing violations. One such pattern encompasses decisions, usually related to hiring and promotions, where the employer's decision-making process takes place over a period of time, making it difficult to determine the actual date that the allegedly discriminatory act occurred. Merchants Nat. Bank, 42 F.3d at 1058. In such instances, the statute of limitations does not begin to run until the date that the plaintiff knows the allegedly discriminatory decision has been made. Id.

The facts of this case fall within that category of continuing violation cases. There is no specific date that MMSD can identify on which Freeman was informed that he would not be accommodated with a work modification. Instead, the record reveals a constant back-and-forth between Freeman and MMSD, in which Freeman would supply medical information and letters from his doctors and attorneys, and MMSD would identify its concerns with the information it was receiving, and a need for further information or clarification. At most, MMSD expressed doubts about Freeman's ability to return to work, but it never rejected the possibility outright. Moreover, he was never informed that MMSD considered him to be permanently disabled and thus not subject to the policy that provided work accommodations for temporarily disabled employees. Nothing in the record would have put Freeman on notice that the various delays during that time period were attributable to race discrimination. This is a classic case in which the violation unfolded over a long period of time and continued into the limitations period. The record reveals no specific point in time at which Freeman should have been aware that he was being discriminated against, because MMSD shifted its reasons for its decisions as the circumstances changed, thus creating at least the appearance of an employer attempting to work toward the desired accommodation. Thus, only with the benefit of hindsight, after the series of discriminatory acts, could Freeman have realized he was the victim of unlawful discrimination. See Moskowitz v. Trustees of Purdue Univ., 5 F.3d 279, 281-82 (7th Cir. 1993). Therefore, the district court erred in holding the continuing violation doctrine inapplicable to this case. Because at least some of the decisions delaying his return to work were made within the limitations period, Freeman may proceed to challenge the entire series of allegedly discriminatory decisions.

The other evidentiary challenge raised by Freeman involves the court's decision on the morning of trial precluding testimony that Freeman sought to present of two similarly-situated white MMSD employees who were treated differently. Those MMSD employees sustained injuries and were unable to do the type of work that their job description required, but MMSD nevertheless allowed them to return to work and temporarily assigned them considerably less strenuous work than their original jobs. The district court excluded the testimony regarding those two employees because their injuries and work modifications began in approximately February 1997. The court held that testimony of similarly situated employees should be restricted to conduct which occurred during the time period beingconsidered by the jury, which was September 1995 through April 1996 (given the court's holding that there was nocontinuing violation.) Because the proposed testimony involved conduct that occurred after that time period, the court granted MMSD's verbal motion in limine and excluded the testimony.

The court erred in holding that individuals could not be similarly situated if their testimony involved conduct that occurred outside the time period of the alleged acts of discrimination.1 It is the rare case indeed in which there is a nearly exact temporal overlap between the allegedly discriminatory conduct and the conduct regarding similarly situated individuals. The last date of the allegedly discriminatory conduct is not a bright line beyond which the conduct of the employer is no longer relevant in a discrimination case. Otherwise, clearly relevant evidence would be arbitrarily excluded; for instance, a plaintiff in a race discrimination case would then be precluded from producing evidence that the week after he was fired, a white employee escaped discipline for the exact same conduct. The focus must remain on whether the evidence is relevant to demonstrate that discrimination played a role in the decision, and that determination is not served by a bright- line temporal restriction. Here, the proffered testimony involved conduct by the employer approximately ten months after the last challenged act regarding Freeman. That is not a very long period of time given that the policy at issue here involved employees who became disabled and sought alternative job duties during their recoveries-- presumably not a daily occurrence. MMSD's benefits manager, Nadler, admitted at trial that the policy at issue regarding Freeman was in place when Nadler assumed his position in 1993, and continued to be the policy unchanged from that time until the time of trial. Therefore, both the policy and the person implementing that policy were the same for Freeman and his proposed witnesses, and no change in circumstances is apparent within the ten months between April 1996 and February 1997. On the limited record before us, there is no basis for a finding that the employees are not similarly situated.

After rejecting the testimony because it involved conduct that occurred after the last act of alleged discrimination, the court further opined that the jobs and injuries were different as well, and that those differences would have to be explored. Because it concluded the testimony was time-barred, however, the court did not further explore that issue. We note, however, that in determining whether employees are similarly situated, the inquiry varies depending upon the type of employer conduct at issue. For instance, where a male employee fired for sexual harassment claimed that women who engaged in similar conduct were not terminated, "similarly-situated" employees would not necessarily be those who held the same job that he held, but rather would be those female employees who had been the subject of comparable complaints of sexual harassment. Morrow v. Wal-Mart Stores, Inc., 152 F.3d 559, 561 (7th Cir. 1998). Here, the uncontradicted testimony was that the policy at issue was applied to all employees regardless of job description. Moreover, it matters not that the similarly-situated employees proffered by Freeman had injured their backs whereas Freeman had injured his knees. Nothing in the policy rendered that distinction meaningful. Similarly-situated employees, for the purpose of this discrimination case, should include employees covered by the policy who were injured and unable to perform the functions required by their job description, but who were allowed to return to work and assigned different tasks during their recovery. MMSD's argument that the policy did not apply to Freeman because his injury was permanent rather than temporary is an issue for the jury, and does not alter the class ofpersons who are similarly situated here. On this record, the court erred in concluding that the proposed employeewitnesses were not similarly situated and in excluding their testimony.

For the above reasons, the decision of the district court is reversed, and the case is remanded for further proceedings in accordance with this opinion.