This appeal follows a lengthy employment discrimination trial pitting a long-time professor at Kent State University ("KSU") against the University. After a trial and jury verdict in the plaintiff's favor on sex and age discrimination claims, the district court granted KSU's motion for judgment as a matter of law on these claims. The plaintiff appeals this decision, as well as other district court rulings in KSU's favor. We AFFIRM in part, and REVERSE in part.

After receiving her doctorate from KSU, Plaintiff-Appellant Dorothy Kovacevich began teaching at KSU's College of Education in 1973 as a non-tenure track, one-year Instructor. In 1975, at the age of 46, she was hired into a tenure-track position, working as an assistant professor in the College of Education's Special Education department. KSU offered her a $14,000 academic year salary, and she accepted. Her duties included teaching and coordinating the field experience of Special Education student teachers. In 1978, Kovacevich was granted tenure. Her fortunes at KSU over the ensuing fifteen years--and in particular her history of promotion and salary increases over those years--are the subject of this litigation.

Kovacevich first contends that KSU discriminated against her by promoting her at a snail's pace due to her gender. Before considering her situation in particular, we will review KSU's general procedures for granting promotions.

KSU is organized into six academic colleges, which are divided into separate departments. There are three ranks of tenure-track professor positions at KSU: assistant, associate and full professor. A series of collective bargaining agreements governs the process and criteria used to grant promotions. Promotion applications proceed through a layer of reviews by KSU faculty and administrators. The first reviewing committee is the Faculty Advisory Committee (FAC), comprising faculty members of an applicant's own department who are elected for several-year terms. The FAC issues a recommendation to the chairperson of the department, who makes her own recommendation after independently reviewing the application. The College Advisory Committee (CAC), comprising faculty members from the entire college, performs the next review. The CAC makes a recommendation to the dean of the college in which the department belongs, and the dean makes another independent review and recommendation. Next, the University Advisory Board (UAB), made up of professors from across the university, reviews the application and makes a recommendation to the KSU provost. The provost then makes a recommendation to the president,1 who makes a recommendation to the KSU Board of Trustees based on an assessment of the complete application, including all the recommendations. At any level, a favorable recommendation moves the application to the next level. Any unfavorable recommendation ends the process, barring a successful "appeal" at the next level. In sum, the levels of review are as follows:

To be promoted, a faculty member must demonstrate outstanding work in one ofthree areas: 1) teaching and advising; 2) research and scholarship; or 3) service and creative effort. In addition, a candidate must meet minimum requirements in the other two areas. As part of the application procedure, an applicant must submit substantial paperwork and other materials that evince her professional accomplishments.

Kovacevich first applied for promotion to an associate professor position in 1978. Although the FAC and department chairperson recommended her promotion, the CAC denied her application. She did not appeal this decision. In 1980, Kovacevich once again applied for a promotion; she obtained the recommendation of the FAC, her department chairperson, the CAC and the dean of the College of Education, yet the UAB and the provost denied her application, citing a lack of sufficient scholarship. Kovacevich appealed that denial to then-President Brage Golding, who agreed with the denial. She did not appeal the decision further.2

In 1983, Kovacevich once again applied for promotion to associate professor. She acknowledged at trial that her documentation was "a mess." J.A. at 534-36. After receiving the recommendation of the FAC and department chairperson, the CAC recommended against promotion, as did the dean of the College. The appeals board, the provost, and the president all denied her appeals. Once again, they explained that her research and scholarship were weak.

In 1987, after serving as an assistant professor for fifteen years, Kovacevich applied once again to be an associate professor. Although she received recommendations from the FAC, the department Chairperson, the CAC and the dean of the College of Education, the UAB recommended against promotion. Nevertheless, the provost rejected that recommendation, and recommended granting the promotion, effective for the 1988-89 academic year. The KSU President agreed, concluding that although her scholarship remained weak, it had improved somewhat, and that Kovacevich had a long history of service to KSU.

Kovacevich also argues that KSU discriminated against her, based on gender and age, when granting salary increases during her years there. There are four ways in which KSU faculty salaries are increased: 1) promotion; 2) merit awards; 3) fixed percentage across-the-board salary increases; and 4) equity salary adjustments. Kovacevich here challenges KSU's granting of her merit awards.

Merit awards are only available in those years when sufficient funds are available. They are analogous to a bonus in that they reward faculty members for achievement in teaching or scholarship. Unlike a bonus, however, a merit award becomes a permanent addition to a professor's base salary for future years. In years that merit awards are available, interested faculty members must submit an application documenting their achievements from that year. In the College of Education, the FAC reviews that application, with each FAC member rating the applicant's teaching and research/scholarship. Faculty members may earn merit in one or both categories. Based on the rank-orderings of the compiled FAC evaluations, the department chairperson (who does not know the names of the applicants) ranks the various applications and determines an amount to be awarded to each applicant. The chairperson then submits the list to the dean of the college, who makes the final determination of award amounts.

Since 1977, merit awards were available eleven times. Kovacevich applied forawards on at least eight occasions. KSU granted her merit increases seven times, providing her the minimum award each time. On two occasions, the dean of the College of Education reduced the award amount which Kovacevich's department chairperson had recommended she receive. On another occasion, the dean disallowed a recommended award. The following chart shows Kovacevich's merit award history:

(Merit awards were not available after 1991-92)

J.A. at 74-75, 974.

In 1989, Kovacevich formally requested a salary adjustment. The college dean appointed a committee to consider the request, and the committee ultimately concluded that Kovacevich's record of performance did not warrant the requested adjustment. Kovacevich submitted a second salary adjustment request in 1991, seeking an annual increase of $12,500. On this occasion, a CAC committee found compelling reasons to consider an adjustment when merit funds were available. Because merit funds were not available, Kovacevich and the American Association of University Professors ("AAUP") filed a grievance to gain the adjustment from another funding source. Concluding that her salary was appropriate based on her work record, the KSU provost denied the grievance in March 1993. Kovacevich never received her requested adjustment, and filed her EEO charge of discrimination on January 12, 1994.

In both of her claims, Kovacevich argues that other similarly situated faculty members were treated more favorably.

a) Dr. Zuckerman

In support of her claims of gender-based wage discrimination, Kovacevich presented evidence that Dr. Robert Zuckerman was a similarly situated comparator who was treated more favorably. KSU hired Zuckerman in 1976 as an assistant professor in the Special Education department. He began teaching at an academic year salary of $13,500. Like Kovacevich, Zuckerman applied for promotion as an assistant professor in 1980, and that application was denied. In 1983, Zuckerman reapplied for promotion. Without the unanimous recommendation of either the FAC or CAC, and despite a negative recommendation from the UAB, the President promoted Zuckerman on his appeal.

Zuckerman applied for merit pay in the years that Kovacevich did not--1983 and 1985--and received large merit awards in those years. In 1988, when she received only $500 in merit, Zuckerman received $1,500.

b) Dr. Barbour

Regarding her claim for age discrimination, Kovacevich argues that KSU treated Dr. Nancy Barbour, a younger professor in the College of Education, more favorably because of her age. Around the time that Kovacevich made her adjustment request, Barbour also requested a salaryadjustment, alleging that gender discrimination accounted for her lower salary relative to a male professor doing equal work. As in Kovacevich's case, the dean stated that she would grant an adjustment once merit funds were available. In response, rather than filing a grievance through the AAUP (as Kovacevich did), Barbour filed a complaint with KSU's Office of Affirmative Action. After negotiations, KSU settled Barbour's claim of sex discrimination in March 1993 and granted her a lump sum payment of $15,000 as well as substantial increases in her salary over two years.

On March 17, 1995, Kovacevich filed a complaint against KSU alleging sex and age discrimination in violation of the Equal Pay Act, 29 U.S.C. § 206(d) ("EPA"); Title VII of the Civil Rights Act of 1964,42 U.S.C. § 2000e et seq.; and the Age Discrimination in Employment Act, 29 U.S.C. § 623 ("ADEA"). The parties conducted discovery until August 1996. In September 1996, Kovacevich amended her complaint to add a claim of disparate impact under Title VII. At that time, both parties filed motions for summary judgment. On June 27, 1997, the district court denied both motions. At the final status conference prior to trial, the court granted KSU's motion to bifurcate the Title VII disparate impact claim and granted KSU's motion in limine to: 1) preclude the testimony of Kovacevich's statistical expert; 2) exclude from evidence her statistical study regarding promotion and wage trends at KSU; and 3) exclude a Department of Labor order that KSU show cause why it had not committed wage discrimination.

The trial on the Title VII disparate treatment claim, EPA and ADEA claims began on July 15, 1997, lasting twelve days over a three-week period. At the end of Kovacevich's case on July 29, 1997, KSU moved for judgment as a matter of law under Fed. R. Civ. P. 50(a), but the court deferred ruling on the motion. At the close of its case on July 31, KSU renewed its motion for judgment as a matter of law, and Kovacevich also moved for judgment as a matter of law. On August 1, the district court granted KSU's motion for judgment as a matter of law on Kovacevich's Title VII disparate treatment claims, concluding that the claims were barred by the statute of limitations and that Kovacevich had failed to make out a prima facie case. The court denied Kovacevich's motion, as well as her request to amend her complaint to add a retaliation claim, and again reserved judgment on KSU's motion regarding the other claims.

On August 4, 1997, the jury returned a verdict in favor of Kovacevich on both the ADEA and EPA claims. While granting her $17,477 in damages for the ADEA claim, the jury awarded no damages under the EPA because it found that KSU's actions were not willful. After Kovacevich challenged the EPA verdict as inconsistent, the jury re-deliberated under court instruction and awarded damages of $11,823. The court then excused the jury.

The parties filed several post-verdict motions. KSU moved again for a judgment as a matter of law on the EPA and ADEA claims, and filed a new motion for summary judgment on the Title VII disparate impact claim. Kovacevich in turn moved for reconsideration of the directed verdict on her Title VII disparate treatment claim, filed a motion to strike KSU's new summary judgment motion, and requested a stay on the disparate impact proceedings until she could depose KSU's statistical expert. On May 8, 1998, the district court granted KSU's motion for judgment as a matter of law on the EPA and ADEA claims pursuant to Fed. R. Civ. P. 50(b), denied Kovacevich's motion for reconsideration, and granted KSU's second motion for summary judgment on the disparate impact claim. Kovacevich filed this appeal on June 4, 1998.

As a preliminary matter, we address whether aspects of Kovacevich's suitagainst KSU are barred by the Eleventh Amendment. First, we recognize that KSU did not raise an Eleventh Amendment defense at the district court level and that the district court did not address this question. We also find that this Court's conclusion in Wilson-Jones v. Caviness, 99 F.3d 203, 206 (6th Cir. 1996)--that "state immunity is jurisdictional in the same sense as the complete diversity requirement," and that "neither the litigants' consent, nor oversight, nor convenience can justify a court's exercise of illegal power"--is no longer good law. In Wisconsin Dep't of Corrections v. Schacht, 524 U.S. 381, 389 (1998), counter to our holding in Caviness, the Supreme Court stated that the Eleventh Amendment is not jurisdictional, and is not akin to the complete diversity requirement. "The State can waive the defense . . . . Nor need a court raise the defect on its own. Unless the State raises the matter, a court can ignore it." Id.; see also Higgins v. Mississippi, 2000 WL 869416 (7th Cir. June 30, 2000), at *1 (stating that Schacht was contrary to the holding in Caviness).

Nonetheless, unlike other affirmative defenses, "'the Eleventh Amendment defense sufficiently partakes of the nature of a jurisdictional bar' that it may be raised at any point of the proceedings," including on appeal. Florida Dep't of State v. Treasure Salvors, Inc., 458 U.S. 670, 683 n.18 (1982)(plurality opinion)(quoting Edelman v. Jordan, 415 U.S. 651, 678 (1974)); see also Schacht, 524 U.S. at 393 (stating that precedent allows the Eleventh Amendment to be first asserted on appeal)(Kennedy, J., concurring); Ford Motor Co. v. Department of Treasury of Ind., 323 U.S. 459, 467 (1945)(considering an Eleventh Amendment defense that was raised for the first time before the Supreme Court). Because, in its supplemental briefing before this Court, KSU has argued that the Eleventh Amendment bars Kovacevich's ADEA and EPA claims, we will consider the defense.

First, Kimel v. Florida Bd. of Regents, 120 S. Ct. 631 (2000) plainly holds that states cannot be sued for money damages under the ADEA because Congress's abrogation of states' immunity was not valid under City of Boerne v. Flores, 521 U.S. 507 (1997). See Kimel, 120 S. Ct. at 644-50; Coger v. Board of Regents, 209 F.3d 485 (6th Cir. 2000) (stating that Kimel concluded that the ADEA's abrogation of Eleventh Amendment immunity "exceeded Congress' authority under Section 5 of the Fourteenth Amendment"), vacating 154 F.3d 296 (6th Cir. 1998). Under Ohio law, KSU is an "arm of the state" entitled to Eleventh Amendment immunity. See Hall v. Medical College of Ohio at Toledo, 742 F.2d 299, 302-07 (6th Cir. 1984). KSU is thus immune from suit under the ADEA, and we affirm the district court's judgment for KSU on that claim.

Conversely, we find that Kimel does not alter this Circuit's conclusion that the EPA constituted a proper abrogation of Eleventh Amendment immunity. See Timmer v. Michigan Dep't of Commerce, 104 F.3d 833 (6th Cir. 1997).

The Judicial power of the United States shall not be construed to extend to any suit in law or equity, commenced or prosecuted against one of the United States by Citizens of another State, or by Citizens or Subjects of any Foreign State.

U.S. Const. amend. XI. Courts have long read this language to mean "that an unconsenting State is immune from suits brought in federal courts by her own citizens as well as by citizens of another State." Edelman v. Jordan, 415 U.S. 651, 663 (1974). Nevertheless, we have recognized that there is no Eleventh Amendment bar in three instances: (1) when the state has consented to suit; (2) when the Ex Parte Young exception applies; and (3) when Congress has properly abrogated states' immunity. See Nelson v. Miller, 170 F.3d 641, 646 (6th Cir. 1999). Whether the Eleventh Amendment bars suits against states under the EPA involves an analysis of this third exception.

After Seminole Tribe of Florida v. Florida, 517 U.S. 44 (1996), Congress can abrogate Eleventh Amendment immunity only under its Section 5 power to enforce the Fourteenth Amendment. Seminole Tribe erected a two-pronged test to determine the validity of Section 5 abrogation. First, we must determine whether Congress "unequivocally expresse[d] its intent to abrogate the immunity." Id. at 55. Second, we must determine whether Congress acted pursuant to a valid exercise of Section 5 power. See id. As the Court explained in Katzenbach v. Morgan, 384 U.S. 641, 651 (1966), Section 5 "is a positive grant of legislative power authorizing Congress to exercise its discretion in determining whether and what legislation is needed to secure the guarantees of the Fourteenth Amendment." Katzenbach went on to define appropriate Section 5 legislation as legislation that: 1) may be regarded as an enactment to enforce the equal protection clause; 2) is plainly adapted to that end; and 3) is not prohibited by but is consistent with the letter and spirit of the Constitution. See id.

The Supreme Court further clarified the extent of Congress's Section 5 power in City of Boerne v. Flores, 521 U.S. 507 (1997), striking down the Religious Freedom Restoration Act ("RFRA") because it exceeded that power. The Court concluded that Congress has the authority to remedy and prevent constitutional violations "even if in the process it prohibits conduct which is not itself unconstitutional and intrudes into 'legislative spheres of autonomy previously reserved to the States.'" Id. at 518 (citation omitted). Nonetheless, Congress can not enforce a constitutional right by changing the substance of that right. "It has been given the power 'to enforce,' not the power to determine what constitutes a constitutional violation." Id. at 519 (citation omitted). The Court acknowledged that this dichotomy between rights-enforcement and rights-creation presents a difficult analysis. "[T]he line between measures that remedy or prevent unconstitutional actions and measures that make a substantive change in the governing law is not easy to discern, and Congress must have wide latitude in determining where it lies . . . ." Id. at 519-20. To assess on what side of that line a given law falls, the Court required that there be "a congruence and proportionality between the injury to be prevented or remedied and the means adopted to that end." Id. at 520 (emphasis added). The Supreme Court in Florida Prepaid Postsecondary Education Expense Board v. College Savings Bank further stated that "for Congress to invoke § 5, it must identify conduct transgressing the Fourteenth Amendment's substantive provisions, and must tailor its legislative scheme to remedying or preventing such conduct." 119 S. Ct. 2199, 2207 (1999).

Under its congruence and proportionality test, the City of Boerne Court struck down RFRA for transgressing Congress's Section 5 power. It did so for three primary reasons. First, the Court found the legislative record barren of examples of modern laws exhibiting religious bigotry,giving it little basis by which to weigh the "evil presented" alongside the remedial measures enacted. 521 U.S. at 530-31; see Florida Prepaid, 119 S. Ct. at 2207 (explaining that in City of Boerne, the Court held that "RFRA failed to meet this test because there was little support in the record for the concerns that supposedly animated the law"). Second, the Court found RFRA "'so out of proportion to a supposed remedial or preventive object' that it could not be understood 'as responsive to, or designed to prevent, unconstitutional behavior.'" Id. (quoting City of Boerne, 521 U.S. at 532). Third, the Court focused on the fact that RFRA re-instituted a compelling interest standard for assessing the validity of burdens on the free exercise of religion--a standard that had been rejected in Employer's Division v. Smith, 494 U.S. 872 (1990)--to underscore that Congress was intending a "substantive alteration of [the Smith] holding." City of Boerne, 521 U.S. at 534. For all of these reasons, the Court held that RFRA "contradicts vital principles necessary to maintain separation of powers and the federal balance." Id. at 536.

Applying the congruence and proportionality test from City of Boerne, the Supreme Court in Kimel held that the ADEA was also an inappropriate exercise of § 5 power. In Kimel, the Court essentially took two steps. First, it found that the "substantive requirements the ADEA imposes on state and local governments are disproportionate to any unconstitutional conduct that conceivably could be targeted by the Act." 120 S. Ct. at 645. The Court noted that the elderly are not a suspect class, and that laws discriminating on the basis of age are presumptively valid under rational basis scrutiny. Thus, the Court found that the ADEA prohibits substantially more state employment decisions and practices than would be illegal under the Fourteenth Amendment. See id. at 6474. Second, the Court found that the legislative record did not support the conclusion that the ADEA constituted reasonable prophylactic legislation rooting out those instances of age discrimination that were sufficiently arbitrary to violate equal protection. The Court parsed the legislative record, finding it void of any concrete showing that states and local governments had been unconstitutionally discriminating against the elderly. Based on what it deemed insubstantial findings, the Court concluded:

Congress had virtually no reason to believe that state and local governments were unconstitutionally discriminating against their employees on the basis of age. . . . Congress' failure to uncover any significant pattern of unconstitutional discrimination here confirms that Congress had no reason to believe that broad prophylactic legislation was necessary in this field.

Id. at 649-50. In light of the "indiscriminate scope of the Act's substantive requirements, and the lack of evidence of widespread and unconstitutional age discrimination by the States," the Court held that the ADEA was an invalid exercise of Congress's §5 power. Id. at 650.

Three years ago, this Court held in Timmer that the EPA was a proper exercise of Congress's Section 5 power, and that Congress's abrogation of states' immunity was therefore proper. See 104 F.3d at 837-42. The Timmer Court concluded that Congress both exhibited an intent to abrogate immunity and possessed the requisite power to abrogate. See id. This decision is controlling authority "unlessan inconsistent decision of the United States Supreme Court requires modification of the decision or this Court sitting en banc overrules the prior decision." Salmi v. Secretary of Health and Human Services, 774 F.2d 685, 689 (6th Cir. 1985). We must therefore determine if Kimel requires modification of the Timmer decision.

We find that Timmer continues to be controlling authority with respect to the first prong under Florida Seminole. The Timmer Court held that Congress clearly expressed its intent to abrogate state sovereign immunity under the EPA. See 104 F.3d at 837-38. Neither Kimel nor any other Supreme Court decision since Timmer has cast doubt on this conclusion, so we are bound by that conclusion.

On the other hand, we find that Kimel does require us to reconsider Timmer's conclusion that the EPA constituted a proper abrogation of states' immunity. After its decision in Kimel, the Supreme Court granted a writ of certiorari and vacated a Seventh Circuit decision concluding, as Timmer did, that the EPA comprised a proper abrogation of states' immunity. See Illinois State Univ. v. Varner, 120 S. Ct. 928 (Jan. 18, 2000), vacating Varner v. Illinois State Univ., 150 F.3d 706 (7th Cir. 1998). The Court remanded the case for further consideration in light of Kimel, see id., so we will likewise reconsider Timmer in light of Kimel. In doing so, we conclude that Kimel only bolsters the conclusion in Timmer that Congress properly abrogated Eleventh Amendment immunity in expanding the EPA to the states.

The Kimel Court suggested in dicta that whether § 5 legislation addresses discrimination against a protected class such as gender or race shapes the "congruence and proportionality" analysis of City of Boerne in important ways. In finding that the substantive requirements of the ADEA were disproportionate to the conceivable unconstitutional conduct targeted by the Act, the Kimel Court contrasted age classifications to those based upon gender.

Age classifications, unlike governmental conduct based on race or gender, cannot be characterized as 'so seldom relevant to the achievement of any legitimate state interest that laws grounded in such considerations are deemed to reflect prejudice and antipathy.' Cleburne v. Cleburne Living Ctr., Inc., 473 U.S. 432, 440 (1985). Older persons, again, unlike those who suffer discrimination on the basis of race or gender, have not been subjected to a 'history of purposeful unequal treatment.'

120 S.Ct. at 645 (citation omitted) (emphasis added). Further, the Court emphasized, when a state discriminates on the basis of gender, courts require "a tighter fit" between the discriminatory means and the legitimate ends they serve. See id. at 646 (citing Mississippi Univ. for Women v. Hogan, 458 U.S. 718, 724 (1982)).

Taking these words from Kimel seriously, we conclude that the remedial scheme introduced by the EPA is indeed congruent and proportional to Congress's § 5 enforcement power. First, unlike the ADEA and RFRA, we find that the EPA's substantive requirements are not disproportionate to the unconstitutional conduct targeted by the Act. In fact, both the Supreme Court and this Court have consistently held that the EPA's target is intentional, gender-based wage discrimination. They have articulated a number of reasons for this conclusion.

First, the language of the statute targets intentional discrimination, stating that no employer "shall discriminate . . . between employees on the basis of sex by paying wages to employees . . . at a rate less than the rate at which he pays wages to employees of the opposite sex." 29 U.S.C. § 206(d)(1). As the Supreme Court stated in looking at the statute and its history, the central purpose of the Act wasto remedy the "endemic problem" that wage structures have been based "on an ancient but outmoded belief that a man, because of his role in society, should be paid more than a woman even though his duties are the same." Corning Glass Works v. Brennan, 417 U.S. 188, 195 (1974).5 Looking at both this affirmative prohibition in the statute as well as the structure of an EPA defense, see infra, this Circuit equates EPA liability for discrimination "on the basis of sex" with liability under Title VII for intentional sex discrimination. See Buntin v. Breathitt County Bd. of Educ., 134 F.3d 796, 801 (6th Cir. 1998); Korte v. Diemer, 909 F.2d 954, 959 (6th Cir. 1990).

Moreover, the EPA's remedial scheme is proportional to its anti-discriminatory aims. A state that intentionally discriminates on the basis of gender violates the Fourteenth Amendment unless the gender-based classification serves important governmental objectives and the discriminatory means employed "are substantially related to the achievement of those objectives." Kimel, 120 S. Ct. at 646 (quoting Hogan, 458 U.S. at 724). This showing requires an "exceedingly persuasive justification" for gender discrimination. United States v. Virginia, 518 U.S. 515, 533 (1996). In contrast to the ADEA, which Kimel held "prohibit[ed] very little conduct likely to be held unconstitutional," 120 S. Ct. at 648, the standard for liability under the EPA closely approximates the equal protection analysis for state-sponsored gender discrimination. Congress designed the EPA to target wage differentials due to unexplainable gender-based discrimination, providing a handful of defenses that account for legitimate causes of wage differentials. See 29 U.S.C. § 206(d)(1) (providing employer with defenses that differential payment is due to a seniority system, a merit system, a system which measures earnings by quantity or quality of production, or a differential "based on any other factor other than sex"). These defenses are designed "to confine the application of the Act to wage differentials attributable to sex discrimination." County of Washington v. Gunther, 452 U.S. 161, 170 (1981). While we do not believe that the liability standards under the Equal Protection Clause and the EPA are identical, they are sufficiently similar such that most cases of state-sponsored wage discrimination that have no explanation "other than sex" also constitute equal protection violations under the Constitution. Although the EPA may bring within its sweep some constitutional conduct, this slight overreaching falls well within Congress's power to enact "reasonably prophylactic legislation" to address intentional, gender-based discrimination. Kimel, 120 S. Ct. at 648.

Unlike the ADEA, then, the EPA does not prohibit substantially more state employment decisions and practices than would likely be held unconstitutional under the applicable equal protection standard. Cf. Kimel, 120 S. Ct. at 648. To the contrary, the EPA enforces and remedies already existent constitutional rights. See Hundertmark v. State of Fla. Dep't of Transport., 205 F.3d 1272, 1276-77 (11th Cir. 2000). Congress's abrogation of state immunity pursuant to the EPA is therefore proper,6 and KSU's allegedviolation of the EPA is not shielded by the Eleventh Amendment.

We next address Kovacevich's argument that the district court erred by granting KSU judgment as a matter of law on several claims because she had not made out a prima facie case. First, the court issued a judgment as a matter of law on Kovacevich's disparate treatment claim at the conclusion of KSU's case in part because she had failed to make out a prima facie case. Second, it rendered a post-verdict judgment as a matter of law on Kovacevich's ADEA claim because she failed to make out a prima facie case.7 Kovacevich argues that the district court's approach in each instance contravened circuit caselaw. We agree.

As the district court recognized, this Court's decision in EEOC v. Avery Dennison Corporation, 104 F.3d 858, 862 (6th Cir. 1997), clearly establishes that, after a trial on the merits, a reviewing court should not focus on the elements of the prima facie case but should assess the ultimate question of discrimination. Relying heavily on United States Postal Service Board v. Aikens, 460 U.S. 711, 717 (1983), the Avery Dennison Court stated:

[A] prima facie case is defined as 'sufficient evidence in the type of case to get plaintiff past a motion for directed verdict in a jury case or motion to dismiss in a nonjury case; it is the evidence necessary to require defendant to proceed with his case.' Black's Law Dictionary 1190 (6th ed. 1990). The finding that plaintiff has proven a prima facie case forces the defendant to proceed with its case. It necessarily follows then, that the defendant is not entitled to judgment as a matter of law or summary judgment if a plaintiff has proven its prima facie case. Following a trial on the merits, the district court, therefore, cannot return to a consideration of whether plaintiff has proven its prima facie case. This is a preliminary matter which cannot be revisited at a later time.

104 F.3d at 861. This clear statement of law would appear to resolve the issue here.

However, the district court reasoned that Avery Dennison appears to contradict a prior circuit decision, and also makes little logical sense. The district court stated that it

recognize[d] the existence of Sixth Circuit authority suggesting that a trial court cannot revisit the prima facie case after the onset of trial. [citing Avery Dennison] However, in another Sixth Circuit case the Sixth Circuit Court of Appeals explained that a "plaintiff in adisparate treatment case must prove to the factfinder, by a preponderance of the evidence, that each component of [the] prima facie case has been satisfied." Gafford v. General Elec. Co., 997 F.2d 150, 168 (6th Cir. 1993). These decisions seem wholly inconsistent because if a jury needs to make a determination as to whether or not a plaintiff has proven a prima facie case by a preponderance of the evidence, surely a trial court should be able to make a legal determination that no reasonable jury could conclude that a plaintiff has proven her prima facie case by a preponderance of the evidence.

J.A. at 918. Presumably, by pointing out this conflict, the court is calling on us to heed its reading of Gafford, and not Avery Dennison, since we must defer to a prior case when two panel decisions conflict. See generally Salmi, 774 F.2d at 689. Under this logic, the court found Kovacevich to have failed to make out a prima facie case on both her Title VII and ADEA claims.9

Notwithstanding the district court's doubts, we find that the rule from Avery Dennison rests on firm legal ground, and we do not hesitate to apply its well-established rule in this case. First, the district court incorrectly characterizes Avery Dennison as an isolated decision within the circuit. In fact, it emerges from a series of Circuit cases with analogous analyses of the prima facie case, holding that once there has been a trial on the merits, an appellate court can not revisit whether the plaintiff established her prima facie case. See, e.g., Suggs v. Servicemaster Educ. Food Mgmt., 72 F.3d 1228, 1232 (6th Cir. 1996) ("Since this case proceeded to trial on the merits, on review, this Court does not revisit the McDonnell Douglas analysis of whether the plaintiff established a prima facie case of discrimination. Rather, we proceed directly to the ultimate question of whether the plaintiff carried her burden of proof of discriminatory discharge."); Wilson v. Firestone Tire & Rubber Co., 932 F.2d 510, 513-14 (6th Cir. 1991) (holding that the prima facie question is not reviewed "since the matter proceeded to trial"); Fields v. Bolger, 723 F.2d 1216, 1219 (6th Cir. 1984) (reviewing the "ultimate issue" rather than the prima facie showing). Further, since Avery Dennison, this Circuit has relied on the opinion and the central proposition for which it stands. See Brocklehurst v. PPG Indus., Inc., 123 F.3d 890, 897 n.5 (6th Cir. 1997) ("Because this case has been fully tried on the merits, we may not question whether Brocklehurst established a prima facie case.") (citing Avery Dennison, 104 F.3d at 860)).

The logic anchoring the Suggs-Wilson-Fields line of cases is the same as that in Avery Dennison. Moreover, all of these cases adhere to the Aikens Court's discussion of the role played by the prima facie case. Aikens involved a black postal worker who brought a Title VII suit against the U.S. Postal Service. After a bench trial, the district court entered judgment in favor of the Postal Service, finding that Aikens had not met his prima facie case. After an appellate court reversal,the Supreme Court addressed the district court's decision--and the parties' arguments--which focused on whether Aikens made out a prima facie case.

Because this case was fully tried on the merits, it is surprising to find the parties and the Court of Appeals still addressing the question whether Aikens made out a prima facie case. We think that by framing the issue in these terms, they have unnecessarily evaded the ultimate question of discrimination vel non.

Aikens, 460 U.S. at 713-14. The Court elaborated that the prima facie stage merely poses a "sensible, orderly way to evaluate the evidence" and bring the case to the central factual inquiry--"whether the defendant intentionally discriminated against the plaintiff." Id. at 715 (quoting Texas Dep't of Community Affairs v. Burdine, 450 U.S. 248, 253 (1981)). Thus,

[w]here the defendant has done everything that would be required of him if the plaintiff had properly made out a prima facie case, whether the plaintiff really did so is no longer relevant. The district court has before it all the evidence it needs to decide whether 'the defendant intentionally discriminated against the plaintiff.'

Id. (quoting Burdine, 450 U.S. at 253) (emphasis added). Because the record included defendant's "production" and a dispute over the evidence of intentional discrimination, "the District Court in this case should have proceeded to this specific question [of intentional discrimination] directly." Id. The appellate court should have done the same. See id. at 713-14. By framing the issue in terms of the prima facie case, the courts and parties "unnecessarily evaded the ultimate question of discrimination vel non." Id. at 714.

Both its holding and language demonstrate that Aikens is not a narrow case, as the district court assumes when it argues that Avery Dennison "misinterpreted" Aikens. First, Aikens was not limited to bench trials following a judge's rejection of a directed verdict motion, as the district court has suggested elsewhere. See Petrone, 993 F. Supp. at 1130 (stating that Avery Dennison misapplied Aikens because in Aikens the Supreme Court "concluded that the district court could not revisit whether or not the plaintiffs had proven a prima facie case in a bench trial after previously concluding, at the close of the plaintiffs' case in chief, that the prima facie case had been made"); see also Avery Dennison, 104 F.3d at 865-66 (Ryan, J., dissenting) (distinguishing Aikens in the same manner). Although Aikens involved a bench trial, as well as a decision that followed the district court's refusal to dismiss the plaintiff's case after the defendant moved for a directed verdict, see Aikens, 460 U.S. at 714 n.4, the Court made no effort to narrow its discussion to that procedural circumstance. Indeed, when discussing the role of the prima facie case, the Aikens Court spoke broadly, stating unconditionally that district judges should not revisit a prima facie case once a defendant has presented evidence of its nondiscriminatory reason and the case has been tried on the merits. See 460 U.S. at 714. The Court neither explicitly nor implicitly limited this statement to bench findings of fact, as opposed to those of a jury, as the district court contends. Furthermore, the Court placed little emphasis on the fact that the Postal Service had filed a directed verdict motion in the case, mentioning that fact only in passing. See id. at 714 n.4 ("It appears that at one point in the trial the District Court decided that Aikens had made out a prima facie case.") (emphasis added). Perceiving the breadth of Aikens, an impressive number of appellate courts have applied its holding in the same way that Avery Dennison did--regardless of whether the proceeding below was a bench or jury trial, and regardless of whether thetrial court had previously addressed a motion for judgment on the law.10

Nor was Aikens limited to appellate court review, as the district court suggested when it stated that Avery Dennison is untenable for district courts. Aikens made fundamental observations about the purpose and structure of the McDonnell Douglas test and the steps both trial and appellate courts must take to adhere to that broader purpose. Thus, even though the opinion is often cited in the context of appellate review, see, e.g., Brocklehurst, 123 F.3d at 897 n.5, Aikens and lower court decisions interpreting Aikens have applied its logic to both district court review of verdicts as well as appellate review. See Aikens, 460 U.S. at 714 (expressing "surpris[e]" that the Court of Appeals focused on the prima facie case); id. at 717 (stating that the district court "erroneously focused on the question of prima facie case rather than directly on the question of discrimination"); see also, e.g., Barnes, 778 F.2d at 1100 (criticizing the district court for focusing on the prima facie case after a bench trial on the merits); Hayman, 23 F.3d at 537 ("Because Appellant's claim was fully tried on the merits below, the district court had no reason to base its judgment on supposed deficiencies in Appellant's prima facie case."). The rule from Aikens thus applies to appellate courts and district courts alike.

Finally, the district court's contention that Avery Dennison contradicts this Court's holding in Gafford misreads both opinions. The court focuses on the portion of Gafford rejecting the plaintiff's argument that the district court erred by submitting to the jury the question of whether a prima facie case had been established. See 997 F.2d at 168. Stating that the finder of fact must find the prima facie case satisfied in a disparate treatment case, see id. at 168 (citing Burdine, 450 U.S. at 252-54), the Gafford Court reasoned that whether a prima facie case has been established is a matter of both fact and law. Thus, "it was not error per se for the district court to have submitted the question [] to the jury." Id. at 169. Contrary to the district court's analysis, Avery Dennison does not contradict this conclusion in Gafford any more than Aikens contradicts Burdine. While, as Gafford states, the elements of the prima facie case can indeed go before a jury, Aikens and Avery Dennison simply state that in reviewing a trial that proceeds beyond the prima facie stage, neither the district court nor appellate court should refocus on the question of whether a plaintiff established her prima facie case. They should instead look to the ultimate question of discrimination. See Aikens, 460 U.S. at 714. The district court is simply ignoring or misreading Aikens when it states that if, under Gafford, a jury can determine whether a prima facie case has been proved, "surely a trial court should be able to make a legal determination that no reasonable jury could conclude that a plaintiff has proven her prima facie case by a preponderance of theevidence." J.A. at 91. Aikens instructs us otherwise.

In sum, because Avery Dennison falls within a well-established line of analogous Sixth Circuit caselaw, because it accurately reflects Aikens, and because it does not contradict Gafford, its holding applies in this case. Moreover, there is no basis for KSU to distinguish Avery Dennison as a Title VII case, "not an EPA or ADEA claim." KSU's Br. at 16. Both Aikens and Avery Dennison speak broadly of the purpose of prima facie cases in discrimination suits, and neither those opinions themselves, nor later circuit caselaw applying Aikens, is limited to the Title VII context. See, e.g., Phelps v. Yale Security, Inc., 986 F.2d 1020, 1023-24 (6th Cir. 1993) (applying Aikens in an ADEA case); Soto v. Adams Elevator Equip. Co., 941 F.2d 543, 548 (7th Cir. 1991) (utilizing the Aikens rule for an EPA claim, and pointing to an ADEA case where it was also used). We therefore reject the district court's assertion that Avery Dennison conflicts with prior caselaw and ought not be applied in this instance. To the extent that the district court granted judgment on the ADEA and Title VII claims because Kovacevich failed to make out her prima facie case, it erred as a matter of law.

Finally, we feel compelled to respond briefly to Judge Gilman's criticism of Avery Dennison. First, overreading some of Avery Dennison's dicta, the concurrence misstates that decision's holding. We do not read the decision to have held that "the district court cannot question the proof constituting the plaintiff's prima facie case once the action is allowed to proceed to trial," Post at 55. Like our decision in this case and like Aikens, Avery Dennison stands for the simple proposition that in reviewing the facts of a discrimination claim after there has been a full trial on the merits, a district court or an appellate court must focus on the ultimate question of discrimination rather than on whether a plaintiff made out her prima facie case. Of course, there is nothing to prevent the court from considering evidence that also bears on that prima facie case as long as it does so in order to address the ultimate question of discrimination. Indeed, that is what we do below. See infra. That is also what the D.C. Circuit and First Circuit did in Coffey, 170 F.3d at 326-27 and Barbour, 181 F.3d at 1347, respectively--decisions that, despite Judge Gilman's view to the contrary, we find to be fully consistent with our approach today.

Judge Gilman's discomfort with Avery Dennison appears to be that it allegedly represents the triumph of "semantics," "terminology" and "nomenclature" over substance. But in this, the concurrence overlooks a central point of Aikens, and, indeed, a fundamental element of appellate review. Similar to Avery Dennison, in Aikens the Supreme Court reversed and remanded a bench-trial decision precisely because the deciding court, as well as the appellate court and the parties, had continuously "frame[d] the issue in [] terms" of Aikens's prima facie case even after the case was tried on the merits. 460 U.S. at 714 (emphasis added). While the concurrence would minimize the "terms" by which courts "frame" their review as mere "semantics," the Aikens decision hinged on the fact that the district court's errant "terms" undermined confidence in the decision it had rendered. As the Court explained, given the inappropriate characterization of the applicable review, "we cannot be certain that" the lower court's findings were not influenced by its "mistaken view of the law." Id. at 719. In other words, to a reviewing court, the written language a lower court uses to frame its legal and factual decisions is indispensable to assessing the rationale behind and correctness of those outcomes. Given this aspect of Aikens, if Judge Gilman wishes to battle what he sees as the triumph of "semantics" over substance, he will have to wage that battle at a higher level.

We of course agree with Judge Gilman that, even after a district court uses a faulty framework, a reviewing court can itself look at a complete record to decide the case using the proper mode of review. That is what we do in this case. But that is neither required nor possible in every case, as the concurrence seems to suggest. See Post at 52 (criticizing Avery Dennison for reversing and remanding rather than "affirming on different grounds after correcting the district court's terminology"). In cases where the record is not sufficient to allow the reviewing court to address the ultimate question of discrimination, remand is appropriate. This is precisely what the Supreme Court did in Aikens, see 460 U.S. at 717 (remanding for the district court to reconsider whether the Postal Service discriminated against Aikens using the proper review), and, consistent with Aikens, what this Court did in Avery Dennison. See 104 F.3d at 863 (explaining that a remand was necessary because the record was incomplete and the court was unable to make a determination on the ultimate issue of discrimination).

Finally, we do not perceive the "confusion in the circuit" detected by Judge Gilman. We certainly do not find confusion demonstrated by the fact that Judge Ryan dissented in Avery Dennison itself, or that a lone district court has repeatedly refused to apply that holding. To the contrary, numerous decisions in this circuit and others have understood, explained and applied the central logic of Aikens and Avery Dennison in a variety of circumstances, as we do here. See n. 9, supra. Perhaps the concurrence's puzzlement stems from the fact that it finds parts of Aikens itself to be "rather cryptic," a sentiment we have not found other courts to have expressed about the concise, six-page opinion that enjoyed an eight-Justice majority (with the ninth concurring in the result). Nonetheless, if there is any latent confusion, we hope that our treatment today further clarifies the meaning of both cases. As for Judge Gilman's concern that this decision and Avery Dennison pose a "trap for unwary district court judges," we are confident that this Circuit's district courts are sufficiently wary to take note of binding circuit and Supreme Court caselaw and apply it accordingly.

Kovacevich next argues that the district court erred in granting KSU a judgment as a matter of law on her EPA claim. We agree.

The EPA provides that no covered employer shall discriminate "between employees on the basis of sex . . . for equal work on jobs the performance of which requires equal skill, effort, and responsibility, and which are performed under similar working conditions." Timmer, 104 F.3d at 843 (quoting 29 U.S.C. § 206(d)(1)). To establish a prima facie case for an EPA claim, a plaintiff must show that an employer paid different wages to an employee of the opposite sex for substantially equal work. See id. "'Equal work' does not require that the jobs be identical, but only that there exist 'substantial equality of skill, effort, responsibility and working conditions.'" Buntin, 134 F.3d at 799 (quoting Odomes v. Nucare, Inc., 653 F.2d 246, 250 (6th Cir. 1981)). To determine if work is substantially equal, a court must make "an overall comparison of the work, not its individual segments." Id. Once the plaintiff establishes a prima facie case, the defendant must prove by a preponderance of the evidence that the wage differential is justified under one of the four affirmative defenses set forth in the Equal Pay Act: (1) a seniority system; (2) a merit system; (3) a system which measures earnings by quantity or quality of production; or (4) any other factor other than sex. See 29 U.S.C. §206(d)(1); Corning Glass Works, 417 U.S. at 196-97; Buntin, 134 F.3d at 799; Timmer, 104 F.3d at 843. The burden for provingthat a factor other than sex is the basis for a wage differential is a heavy one. See Buntin, 134 F.3d at 799

Because there was a trial on the merits, Aikens dictates that the district court's and this Court's duty in reviewing that verdict is "not [] to determine whether [Kovacevich] established each element of her prima facie case, or whether [KSU] presented sufficient evidence of an affirmative defense." Soto, 941 F.2d at 549. Rather, looking at the evidence in a light most favorable to Kovacevich, we must decide whether the jury reasonably could have believed that KSU violated the EPA by paying Kovacevich less than Zuckerman because of her gender. See id.; see also Aikens, 460 U.S. at 715-16 (stating that the district and appellate courts must consider the ultimate question of whether intentional discrimination occurred). There must be no genuine issue as to whether the difference in pay is due to a factor other than sex. See Buntin, 134 F.3d at 800. Of course, evidence that bears upon elements of the prima facie case can also come into play in assessing the ultimate question of discrimination. See Reeves v. Sanderson Plumbing Products, Inc., 120 S. Ct. 2097, 2111 (2000)(taking into account the evidence supporting the plaintiff's prima facie case as part of its consideration of the "ultimate question" of intentional discrimination); Barbour, 181 F.3d at 1347; Green, 25 F.3d at 978.

Looking at the totality of evidence in the record, we find that Kovacevich introduced sufficient evidence for a reasonable juror to find that her lower salary was a result of gender discrimination. Despite KSU's strong factual defense, we disagree with the district court that KSU was entitled to prevail as a matter of law.

First, looking at the evidence in a light most favorable to Kovacevich, a reasonable juror could find that Kovacevich and Zimmerman were treated differently for comparable work while at KSU. The chronology of their time at KSU is quite similar: having been hired within one year of one another with similar prior experience; undertaking the same workload requirements while there; working within the same department; and having been evaluated under the same criteria by the same FAC in the same years. Despite some differences in their experiences pointed out by KSU, looking at these circumstances in a light most favorable to Kovacevich, we conclude that there is support for her argument that the two professors were roughly comparable professionally. There is also no question that they received different salaries--by the time of Kovacevich's retirement in 1995, her pay was $5,999 less than Zuckerman's.

Although KSU has argued that Kovacevich's lower salary was due to the school's merit system and across-the-board percentage increases, Kovacevich introduced evidence that these justifications did not preclude discrimination on the basis of gender. Most importantly, she adduced evidence showing that rather than a neutral system of merit based on anonymous peer evaluations, the merit award system was driven largely by an opaque, decision-making process at the administrative level, did not necessarily reflect peers' assessment of applicants' performances, and rewarded men disproportionately to women. For instance, the report prepared by a special faculty committee appointed to respond to Kovacevich's charges of discrimination noted that "the nature of current merit procedures suggests that deans and chairs quasi-independently control the size of merit awards:"

[The faculty ranking of peers] is done independently of--and essentially without knowledge of--the dollar amount (if any) to be attached to a particular ranking. Department chairs then use these rankings (as well as other criteria) to recommend actual dollar merit awards. Chairs' recommendations are then forwarded to the Dean, who may support,decrease, or increase the Chairs' recommendations. Moreover, not inconsequential increases over and above departmental recommendations may be awarded to some faculty members by the Dean, since a substantial portion of the college's merit pool . . . is reserved in the Dean's discretionary fund specifically for this purpose. Since the merit system in essence establishes a hiatus between peer evaluations and actual dollars awarded, faculty influence on the size of merit awards is minimized. This condition sometimes makes it extremely difficult to demonstrate the connection between peers' professional judgments of meritorious performance and the size of merit awards.

J.A. at 964-65 (emphasis added). An internal memorandum for KSU administrators also described the college dean's control over a portion of the merit award funds. The record also showed that under this system, the college dean on three occasions either reduced or disallowed entirely a merit award that Kovacevich's department chair had recommended on her behalf. The faculty committee reported that information on the reasons for these reductions was "unavailable," just as "records of any kind relating to merits decisions are virtually nonexistent." J.A. at 960. Finally, Kovacevich discussed at length the results of her own research on the disparity in merit awarded to her department's male and female professors. She stated that from 1988 to 1992, forty percent of males in her department received above average merit awards while only twenty-three percent of its female faculty did so. More broadly, Kovacevich testified that males were disproportionately represented among the top salary-earners in her department, even though women made up forty-seven percent of the faculty. Given this combination of evidence, we believe that Kovacevich created a genuine issue as to whether the difference in her pay and Zuckerman's was due to sex, or due to a factor other than sex. See Buntin, 134 F.3d at 800. For that reason, we find the district court's order of judgment as a matter of law for KSU was in error.

Prior to the jury verdict, the district court granted judgment as a matter of law for KSU on the disparate treatment aspect of Kovacevich's Title VII claim, which encompassed a number of separate allegations of discrimination. After the trial, the court denied Kovacevich's Rule 59 motion to reconsider that decision, and Kovacevich now appeals the denial of that motion.

We conclude that Kovacevich is entitled to judgment in her favor on the wage discrimination element of her Title VII claim. Title VII deems it an unlawful employment practice "to discriminate against any individual with respect to his compensation . . . because of such individual's . . . sex." 42 U.S.C. § 2000e-2(a). Generally, a Title VII claim of wage discrimination parallels that of an EPA violation. See generally Buntin, 134 F.3d at 801; Korte, 909 F.2d at 959. Nevertheless, due to the more demanding threshold of showing a comparator in the EPA context, "claims for sex-based wage discrimination can be brought under Title VII even though no member of the opposite sex holds an equal but higher paying job." Henry v. Lennox Indus., Inc., 768 F.2d 746, 752 (6th Cir. 1985) (quoting Gunther, 452 U.S. at 168). At the same time, the Bennett Amendment to Title VII incorporates the EPA's affirmative defenses into Title VII's prohibition of gender-based wage discrimination. See Gunther, 452 U.S. at 167-68 (interpreting 42 U.S.C. § 2000e-2(h)).

The district court erred as a matter of law when it concluded after KSU's proof that Kovacevich had failed to make out her prima facie case under Title VII. After the defendant has presented its case, the district court can only review whether there is a genuine issue with respect to theultimate question of discrimination. See supra. Moreover, the district court erred when it denied Kovacevich's motion for reconsideration after the verdict by pointing to KSU's defenses under the EPA. For reasons stated above, despite KSU's defenses, a genuine issue remained as to whether Kovacevich's lower salary relative to Zimmerman resulted from gender discrimination. In this Circuit, a finding of liability under the Equal Pay Act requires a similar finding of liability under Title VII where both claims present the same conduct and evidence. See Buntin, 134 F.3d at 801; Korte, 909 F.2d at 959. Because we hold that the EPA verdict was legally sound, Kovacevich is also entitled to prevail on her Title VII disparate treatment claim regarding wage discrimination. We remand for a determination of damages for this claim.

The district court also denied reinstatement of Kovacevich's other Title VII disparate treatment claims--for denial of promotion, disparate workloads, and denial of salary adjustment--because it found them to be barred by the statute of limitations. In doing so, it rejected Kovacevich's "continuing violation" argument. We agree with the district court's conclusion.

The "continuing violation" doctrine provides that when "there is an ongoing, continuous series of discriminatory acts, they may be challenged in their entirety as long as one of those discriminatory acts falls within the limitations period." Haithcock v. Frank, 958 F.2d 671, 677 (6th Cir. 1992). In other words, when a continuing violation is shown, "a plaintiff is entitled to have a court consider all relevant actions allegedly taken pursuant to the employer's discriminatory policy or practice, including those that would otherwise be time barred." Alexander v. Local 496, Laborers' Int'l Union of N. Am., 177 F.3d 394, 408 (6th Cir. 1999); see also Held v. Gulf Oil Co., 684 F.2d 427, 430 (6th Cir. 1982). This Court has held that continuing violations may be shown in one of two ways. The first is "where there is some evidence of present discriminatory activity giving rise to a claim of a continuing violation," and "at least one of the forbidden discriminatory acts [has] occurred within the relevant limitations period." Haithcock, 958 F.2d at 678 (citation omitted) (emphasis in original). The second type of continuing violation exists when a plaintiff has demonstrated a longstanding and over-arching policy of discrimination. See id. This requires a showing by a preponderance of the evidence "that some form of intentional discrimination against the class of which plaintiff was a member was the company's 'standing operating procedure.'" EEOC v. Penton Indus. Publ'g Co., 851 F.2d 835, 838 (6th Cir. 1988) (citation omitted).

Regarding the first category of continuing violation, Kovacevich has not alleged an appropriate event that falls within the 300-day period for filing an EEOC charge. She claims that the provost's denial of her grievance was such an event, but this argument is unpersuasive in light of circuit precedent that the denial of a request for relief from discrimination does not itself constitute a discriminatory act that tolls the statute of limitations. See Janikowski v. Bendix Corp., 823 F.2d 945, 948 (6th Cir. 1987). Kovacevich can thus not claim a category-one "continuing violation."

In the alternative, Kovacevich must show by a preponderance of the evidence that KSU has an over-arching policy--a "standing operating procedure"--of discrimination in promotion or other aspects of employment. Penton, 851 F.2d at 838. This means she must show "something more than the existence of discriminatory treatment in [her] case." Haithcock, 958 F.2d at 679. We conclude that she has not met this high threshold. Regarding the denial of her own promotions, she did not establish any evidence of discrimination against her, acknowledging herself that she had not expected to receive a promotionin 1978 and 1983 and that her documentation in support of the latter application was a "mess." When she next applied for promotion in 1987, KSU granted the promotion. She also failed to show that the reason consistently given for her failure to be promoted--her lack of scholarship--was somehow pretextual. Although she contends that KSU regularly assigned women faculty disparate teaching workloads, that argument is not supported by the record.11

Outside of her claim for wage discrimination, the evidence in the record did not meet the standards required to show a "continuing violation." Kovacevich's non-wage claims under Title VII are thus barred by the statute of limitations.

After the trial, the district court granted summary judgment on the bifurcated disparate impact phase of the Title VII claim. The court reached two conclusions in making that decision. First, it found that the evidence supported KSU's affirmative defense that the merit system was a "legitimate business necessity." Second, it held that once again, the statute of limitations barred Kovacevich's disparate impact claim regarding promotions. We disagree with its first conclusion, although we agree with the second.

Title VII proscribes "not only overt discrimination but also practices that are fair in form, but discriminatory in operation." Griggs v. Duke Power Co., 401 U.S. 424, 431 (1971). Thus, when bringing a disparate impact claim, plaintiffs need not show that a defendant intended to discriminate, but must instead prove that a particular employment practice, although neutral on its face, has produced a significant adverse effect on a protected group to which the plaintiff belongs. See Alexander, 177 F.3d at 405; Scales v. J.C. Bradford & Co., 925 F.2d 901, 907 (6th Cir. 1991); Abbott v. Federal Forge, Inc., 912 F.2d 867, 872 (6th Cir. 1990). The plaintiff must first establish a prima facie case by identifying and challenging a specific employment practice, and then show an "adverse effect" by offering statistical evidence "of a kind or degree sufficient to show that the practice in question has caused the" adverse effect in question. Scales, 925 F.2d at 908 (citation omitted). When a prima facie case is established, the defendants must articulate a legitimate business reason for the employment practice. See id. Once a defendant has articulated that legitimate business reason, the burden shifts back to the plaintiff to show either that the employer's reason is a pretext for discrimination, or that there exists an alternative employment practice that would achieve the same business ends with a less discriminatory impact. See id. In a wage discrimination case, the Bennett Amendment allows a defendant to offer one of the affirmative defenses spelled out in the EPA, so long as it is a legitimate business reason. See EEOC v. J.C. Penney Co., 843 F.2d 249, 253-54 (6th Cir. 1988).

We do not find the district court's analysis sufficient to affirm its summary judgment for KSU on the wage discrimination claim. The court dismissed the claim with the following reasoning:

[T]he undisputed evidence at trial revealed that KSU was entitled to a defense under the EPA as to Dr. Kovacevich's salary. Moreover, the evidence clearly showed that the merit system and across the board percentage increases, both of which were implemented pursuant to the Collective Bargaining Agreement, were adopted for legitimate business reasons. Accordingly, Dr. Kovacevichcannot maintain an action for disparate impact as to her wages.

J.A. at 98. While the court is certainly free to look beyond the prima facie statistical assessment and proceed directly to an assessment of KSU's defense, cf. J.C. Penney Co., 843 F.2d at 253-54, in doing so it must also consider whether Kovacevich showed either that the employer's reason was a pretext for discrimination, or that there exists an alternative employment practice that would achieve the same business ends with a less discriminatory impact. See Scales, 925 F.2d at 908. As stated supra, viewed in a light most favorable to her, Kovacevich presented sufficient evidence at trial to create a material issue as to whether the merit system was a pretext for discrimination. Moreover, we can not discern from the court's opinion or the joint appendix whether Kovacevich made a direct argument that there is an alternative employment practice which could achieve the same ends with a less discriminatory impact--an alternative showing available to her even had she failed to show pretext. We thus do not believe that the court's mere statement that the merit system was adopted for legitimate business reasons is sufficient to justify summary judgment. In light of this conclusion, we remand for the district court to conduct a more complete analysis of whether trial is required for Kovacevich's disparate impact claim. Because the case has yet to be tried, the court may address whether Kovacevich's statistical evidence was sufficient to make out a prima facie case in the first place. The record before us is not sufficiently clear for us to make such a determination.

We agree with the district court that the statute of limitations bars Kovacevich's disparate impact claim regarding promotion. Given that she was last denied a promotion in 1983, and was granted promotion when she last applied in 1987, the 300-day statute of limitations under Title VII clearly has passed.

At the end of trial, Kovacevich moved to amend her complaint to add a retaliation claim. She intended to show that KSU retaliated for work she undertook on her 1988 sabbatical examining gender-based wage disparities in higher education. The court denied her motion. She appeals this decision, arguing that because KSU had itself discussed the sabbatical in presenting its case, the motion to amend should have been granted. We review a decision to permit or deny a Rule 15(b) motion to amend for abuse of discretion. See Strickler v. Pfister Associated Growers, Inc., 319 F.2d 788, 791 (6th Cir. 1963).

When issues not originally raised by the pleading "are tried by express or implied consent of the parties," Rule 15(b) authorizes a district judge to permit the filing of an amended pleading to the extent necessary "to conform to the evidence" presented at trial. Fed. R. Civ. P. 15(b); see Hasselbrin v. Speelman, 246 F.2d 34, 39 (6th Cir. 1957). "Implied consent" requires considerable litigation of a matter--"[i]t must appear that the parties understood the evidence to be aimed at the unpleaded issue." Yellow Freight Sys., Inc. v. Martin, 954 F.2d 353, 358 (6th Cir. 1992). The rule does not exist simply "to allow parties the change theories mid-stream." Donald v. Wilson, 847 F.2d 1191, 1198 (6th Cir. 1988). The district court did not abuse its discretion by denying Kovacevich's motion to amend. We do not find the evidence introduced at trial to establish a case of Title VII retaliation.

Kovacevich argues that the district court improperly excluded several types of evidence she planned to show at trial: a statistical study, an expert witness, and a show-cause order by the OFCCP. We disagree.

We review a district court's decision to exclude evidence under Fed. R. Evid. 403 for abuse of discretion, viewingthe excluded evidence in a light most favorable to the proponent. See Robinson v. Runyon, 149 F.3d 507, 512 (6th Cir. 1998). An abuse of discretion exists where the reviewing court is firmly convinced that a mistake has been made. See id. Even if a court abused its discretion in excluding evidence, a decision will not be disturbed if substantial injustice did not result. See McGowan v. Cooper Indus., Inc., 863 F.2d 1266, 1271 (6th Cir. 1988).

The district court excluded the statistical study as irrelevant under Fed. R. Evid. 402 and unfairly prejudicial under Fed. R. Evid. 403. Under the federal rules, a judge should not admit evidence that is irrelevant--that is, evidence that does not have "any tendency to make the existence of any fact . . . more probable or less probable than it would be without the evidence." Fed. R. Evid. 402. Federal Rule of Evidence 403 prohibits the admission of evidence if there is a danger that its probative value will be substantially outweighed by unfair prejudice. See Robinson, 149 F.3d at 514-15; United States v. Bonds, 12 F.3d 540, 567 (6th Cir. 1993). "Unfair prejudice does not mean the damage to a defendant's case that results from the legitimate probative force of the evidence; rather it refers to evidence which tends to suggest a decision on an improper basis." Robinson, 149 F.23 at 515 (citation omitted).

Excluding this study was not an abuse of discretion. In disparate treatment challenges to university wage and tenure decisions, courts disapprove of studies and other evidence that describe trends of an entire university as opposed to the particular department in question. See, e.g., Evans v. Cleveland State Univ. Bd. of Trustees, No. 90-3759, 1991 WL 93094 at *5 (6th Cir. 1991) (unpublished opinion) (rejecting the comparison of professors in different university departments); Dugan v. Ball State Univ., 815 F.2d 1132, 1136-37 (7th Cir. 1987) (finding plaintiff's statistical showing of discrimination unpersuasive because it was "for the university at large," rather than for the department in question); Molthan v. Temple Univ., 778 F.2d 955, 963 (3d Cir. 1985) (concluding that "statistical evidence of a general underrepresentation of women in the position of full professor adds little to a disparate impact claim"); Zahorik v. Cornell Univ., 729 F.2d 85, 95 (2d Cir. 1984) (concluding that gross statistics are "meaningless absent a departmental breakdown"). These courts have concluded that when both wage and promotion decisions are largely made by individual departments rather than the broader university administration, as they are in this case, "the most important data would be data regarding the individual department." Dugan, 815 F.2d at 1137.

While Kovacevich was permitted to present her department-based data on wages to the jury, the study that the court did not allow was based on the entire population of the university faculty, including all departments and KSU's seven regional campuses. J.A. at 74212. As in the cases cited supra, however, the distribution of merit awards is determined by each separate college--not the university as a whole. J.A. at 747. Thus, the study's descriptions of wage differentials for the entire university bear little on the operation of the merit system within the College of Education, and how that system may have specifically impacted on Kovacevich's wage over time. Accordingly, the court's decision to exclude that evidence was not an abuse of discretion. Likewise, it wasnot an abuse of discretion to exclude the show-cause order issued by the United States Department of Labor, Office of Federal Contract Compliance, which was based on the university-wide study.

Finally, Kovacevich argues that the district court should have disqualified KSU's counsel from trying this case. She argues that trial counsel Richard Cunningham (special counsel appointed by the attorney general) had been a long-time member of KSU's board of trustees, including during the time when the events in question were unfolding. From 1987 to 1995, he served on the board of directors of the KSU Foundation, from which he resigned to represent the University in this case. Kovacevich now argues that the district court erred by not granting her motion to disqualify, since Cunningham's role as trustee in those years gave an appearance of impropriety and showed a likely conflict of interest.

When reviewing a district court's factual findings as part of a decision not to disqualify, we apply a clearly erroneous standard. See Dana Corp. v. Blue Cross & Blue Shield Mut. of N. Ohio, 900 F.2d 882, 889 (6th Cir. 1990). We agree with the district court that there is no case for disqualification here. In addition to the reasons offered by KSU--primarily, that there was no evidence that Cunningham was involved in any of the alleged discriminatory acts, and that Cunningham did not appear as a witness at the trial--it appears that the only party who could have been conceivably harmed by the potential conflict was KSU itself. See generally Melamed v. ITT Continental Banking Co., 592 F.2d 290, 293 (6th Cir. 1979). Kovacevich makes no attempt to explain how the alleged conflict of interest harmed her.

For these reasons, we REVERSE the district court's judgment with respect to the EPA claim and the Title VII claims for wage discrimination, and REMAND for proceedings not inconsistent with this opinion. We AFFIRM the district court's judgment with respect to the remaining Title VII claims and the ADEA claim, as well as its orders excluding the university-wide data, precluding Kovacevich from adding a retaliation claim, and denying her motion to disqualify KSU's counsel.

RONALD LEE GILMAN, Circuit Judge, concurring in part and concurring in the judgment.

I concur in the judgment of the court and in much of its reasoning, though I believe that the jury could at least as easily have found that Kovacevich's lagging pay was the result of her apparently cavalier attitude towards research and publication, and the admittedly subpar documentation of her work, rather than discrimination on the basis of gender. But Kent State University has the burden of showing that no reasonable jury could have found in Kovacevich's favor based on the evidence presented, and I agree with the court that it has failed to do so. The fact that one or more reviewing judges might have weighed the evidence differently provides no basis to overturn the jury's verdict. See Williams v. Nashville Network, 132 F.3d 1123, 1130-31 (6th Cir. 1997); Scott v. County of Ramsey, 180 F.3d 913, 916-17 (8th Cir. 1999). I write separately, however, because I question the soundness of the decision in EEOC v. Avery Dennison Corp., 104 F.3d 858 (6th Cir. 1997), and the court's defense of that opinion.

In EEOC v. Avery Dennison Corp., 104 F.3d 858 (6th Cir. 1997), Ronald Willis, the charging party, settled a race discrimination suit against Avery Dennison Corporation, which was then Willis's employer. Avery Dennison agreed to a cash settlement of the claim and to provide Willis with a letter of reference. In exchange,Willis agreed to resign his position at Avery Dennison and look for work elsewhere. After Willis interviewed for another job, his prospective employer contacted Avery Dennison, seeking information about Willis's employment. The Avery Dennison employee who responded to the request for information reported that Willis had been frequently absent from work, and had left "because of an arbitration case that awarded him a cash settlement on the condition that he terminate his employment" with Avery Dennison. Id. at 859. Not surprisingly, Willis did not get the job.

Willis then filed a charge with the Equal Employment Opportunity Commission, claiming that the bad employment reference was in retaliation for his participation in protected activities, i.e., protesting allegedly discriminatory company policies and Title VII violations, filing EEOC claims, and prosecuting Title VII lawsuits. The EEOC filed a complaint in the United States District Court for the Northern District of Ohio on Willis's behalf, alleging retaliation in violation of Title VII. Willis was permitted to intervene.

The EEOC and Willis filed motions for summary judgment. (Although the majority opinion in Avery Dennison describes the motions as "cross-motions," Avery Dennison, 104 F.3d at 860, it in fact appears that Avery Dennison had not cross-moved for summary judgment and that the only motions for summary judgments before the district court were the ones filed by the EEOC and Willis.) The district court denied the motions, concluding that, based on the summary judgment record, a reasonable factfinder would not be compelled to conclude that the bad job reference was causally related to Willis's engaging in protected activities.

Because no party had requested a jury trial, the district court then conducted a bench trial, after which it issued findings of fact and conclusions of law as required by Rule 52 of the Federal Rules of Civil Procedure. Among the district court's findings of fact was the key point that the EEOC had failed to establish by a preponderance of the evidence that the adverse employment action was causally related to Willis's engaging in protected activities. Accordingly, the district court concluded that the EEOC had not proved its case by a preponderance of the evidence, and entered judgment in favor of Avery Dennison. The EEOC and Willis appealed.

A divided panel of this court reversed, with Judge Ryan dissenting. The majority concluded that the district court's denial of the EEOC's and Willis's motions for summary judgment meant that the district court "believed that genuine issues existed with respect to the prima facie case," id. at 860, including whether or not Willis's bad job reference was causally related to his engaging in protected activities. According to the majority, "[t]he fact that the [district] court was faced with the prima facie case question at the summary judgment stage and then allowed the case to go to trial, could be construed as a tacit acknowledgment of [the EEOC's] prima facie case." Id. at 861. What the district court should have done, the majority concluded, was "draw[] appropriate reasonable inferences and rule[] on whether or not a prima facie case had been made, reserving for trial only the ultimate issue of discrimination." Id. at 860-61.

Reasoning that a "finding that [the] plaintiff has proven a prima facie case forces the defendant to proceed with its case," id. at 861, the majority went on to hold that "[i]t necessarily follows, then, that the defendant is not entitled to judgment as a matter of law or summary judgment if a plaintiff has proven its prima facie case" because "[t]his is a preliminary matter which cannot be revisited at a later time." Id. at 861. (Putting aside all other issues for the moment, it appears that the majority overlooked the fact that a district court's entry of judgment following a bench trial is not "judgment as a matter of law or summary judgment."). Based on the above analysis, this court reversed thedistrict court's entry of judgment for Avery Dennison, and remanded the case to the district court for "a determination on the ultimate issue of discrimination." Id. at 863.

On remand, the district court re-entered judgment for Avery Dennison, this time finding that the EEOC had failed to prove that Avery Dennison's legitimate, non-discriminatory explanation for its conduct was pretextual. This court affirmed the judgment of the district court in a short unpublished opinion. See EEOC v. Avery Dennison Corp., No 98-3490, 1999 WL 1073675 (6th Cir. Nov. 15, 1999).

A principal problem with Avery Dennison is its suggestion that the district court, by allowing the case to go to trial, "tacit[ly] acknowledged . . . Plaintiffs' prima facie case." Avery Dennison, 104 F.3d at 861. Actually, by denying the plaintiffs' motions for summary judgment, the only thing the district court decided was that, on the basis of the summary judgment record, a reasonable trier of fact would not be compelled to find in their favor. Specifically, the district court concluded that a reasonable trier of fact would not be compelled to find that Willis's bad job reference was causally related to his engaging in protected activities. That is not the same as concluding that a reasonable trier of fact could find that the job reference was causally related to Willis's engaging in protected activities. (Again, Avery Dennison had not moved for summary judgment.)

Another problem with Avery Dennison is its assumption that a district court's denial of summary judgment is entitled to preclusive effect, as if it becomes "the law of the case." It does not. See, e.g., Perez-Ruiz v. Crespo-Guillen, 25 F.3d 40, 42 (1st Cir. 1994) (noting that interlocutory orders do not become "the law of the case" and "remain open to trial court reconsideration"); Dessar v. Bank of Am. Nat'l Trust & Sav. Ass'n, 353 F.2d 468, 470 (9th Cir. 1965) (observing that the denial of a motion for summary judgment "merely postpones decision of any question; it decides none" and that "[t]o give it any other effect would be entirely contrary to the purpose of the summary judgment procedure. The court did nothing more than it purported to do, that is, refuse to grant the motion"); Dictophone Prods. Co. v. Sonotone Corp., 230 F.2d 131, 134-45 (2d Cir. 1956) (L. Hand, J.) ("No one will suggest that the [trial judge] may not change his mind and overrule his own order . . . . 'In the absence of statute, the phrase, "law of the case," as applied to the effect of previous orders . . . merely expresses the practice of courts generally to refuse to reopen what has been decided, not a limit on their power.'") (quoting Messinger v. Anderson, 225 U.S. 436, 444 (1912) (Holmes, J.)).

District courts may in their discretion permit renewed or successive motions for summary judgment, particularly when the moving party has expanded the factual record on which summary judgment is sought. See, e.g., Whitford v. Boglino, 63 F.3d 527, 530 (7th Cir. 1995) ("[T]he denial of summary judgment has no res judicata effect, and the district court may, in its discretion, allow a party to renew a previously denied summary judgment motion or file successive motions, particularly if good reasons exist."). If the denial of a motion for summary judgment were entitled to preclusive effect, district courts presumably would not be allowed to entertain renewed or successive summary judgment motions.

In addition, Avery Dennison fails to acknowledge the difference between a summary judgment record and a trial record. Motions for summary judgment are decided "exclusively on the basis of a 'paper' record . . . ." Shields v. Eli Lilly & Co., 895 F.2d 1463, 1466 (D.C. Cir. 1990). Deposition transcripts and affidavits are used as stand-ins for the live-witness testimony that would be presented at trial. The trial record, on the other hand, consists principallyof the witness testimony for which the deposition transcripts and affidavits were surrogates (assuming that the parties have not agreed to a trial on stipulated facts).

A party might not always introduce at trial all of the evidence it put into the summary judgment record--whether for tactical purposes, lack of witness availability, or other reasons. Or, rather than presenting too little evidence, it might present too much, and wreck its case in the process. If the evidence in the trial record is not the same as the evidence in the summary judgment record, it would make little sense to say that a district court's decision to deny summary judgment forecloses it from granting judgment as a matter of law on the same issue after a jury trial.

But even if the contents of the trial record appear to match up with the contents of the summary judgment record, live witness testimony "plays" differently than affidavits and deposition fragments. See DeCarlo v. Fry, 141 F.3d 56, 61 n.1 (2d Cir. 1998) ("The fact that [the district court] held prior to trial that genuine issues of material fact precluded summary judgment as to [the defendant's] liability does not necessarily mean that a verdict as a matter of law after the trial would be unwarranted. The way the evidence plays out at trial may sufficiently alter the contours of the liability issue such that a reasonable jury could reach only one conclusion."). Sometimes it is simply easier to determine what inferences can reasonably be drawn once one has actually heard the testimony in question, rather than to make the determination from a paper preview of that testimony.

The rationale of Avery Dennison is rooted in the Supreme Court's observation in United States Postal Service Board of Governors v. Aikens, 460 U.S. 711 (1983), that, following final judgment for the defendant in a Title VII case after a bench trial, it was "surprising to find the parties and the Court of Appeals still addressing the question whether [the] plaintiff made out a prima facie case." Id. at 714. Because the case had been tried on the merits, the Supreme Court concluded that "by framing the issue in these terms," the parties and the Court of Appeals had "unnecessarily evaded the ultimate question of discrimination vel non." After all, the Court reasoned, if "the defendant has done everything that would be required of him if the plaintiff had properly made out a prima facie case, whether the plaintiff really did so is no longer relevant." Id. at 715.

But what if there is a dispute about whether the defendant "has done everything that would be required of him if the plaintiff had properly made out a prima facie case?" Even if the district court has denied summary judgment to the defendant, the plaintiff still has to prove his case at trial. See, e.g., St. Mary's Honor Ctr. v. Hicks, 509 U.S. 502, 511 (1993) ("The factfinder's disbelief of the reasons put forward by the defendant (particularly if disbelief is accompanied by a suspicion of mendacity) may, together with the elements of the prima facie case, suffice to show intentional discrimination. Thus, rejection of the defendant's proffered reasons will permit the trier of fact to infer the ultimate fact of intentional discrimination.") (emphasis added).

At trial, the plaintiff must still persuade the trier of fact that a "preponderance of the evidence establishes the facts of a prima facie case." Id. at 509-10. If the trier of fact is not persuaded, the plaintiff loses--even if the defendant has completely failed to come forward with evidence in support of any legitimate nondiscriminatory explanation for its actions. See id. at 509-10 & n.3. Indeed, this is apparently the only way the plaintiff can lose if the defendant has failed to sustain its burden of producing evidence that would permit a finding that illegaldiscrimination was not the real reason for its action. If the plaintiff has persuaded the factfinder of the facts constituting a prima facie case, and the defendant has not sustained its burden of production, judgment for the plaintiff is required. See id. But cf. Fisher v. Vassar College, 114 F.3d 1332, 1389 (2d Cir. 1997) (en banc) (Winter, J., dissenting) (explaining why "[i]n reality . . . there is no such thing as a silent defendant"), cert. denied, 522 U.S. 1075 (1998).

Avery Dennison held that the district court should have been concerned only with "the ultimate issue of discrimination." Avery Dennison, 104 F.3d at 862. But when the plaintiff in a Title VII retaliation case has failed to persuade the judge at a bench trial that the defendant's adverse action was causally related to the plaintiff's engaging in protected activities, there can be only one rational conclusion on "the ultimate issue of discrimination," i.e., that the plaintiff has simply not proven his case. The plaintiff's problem is not that he failed to make a showing sufficient to get to trial; his problem is that he failed to prove his case at trial. Similarly, when a plaintiff in a Title VII disparate treatment case fails to persuade the factfinder, for example, that he is similarly situated to others who were treated more favorably, the defendant is entitled to judgment. And if the plaintiff fails to produce evidence sufficient for any reasonable factfinder to conclude that he was similarly situated to others who were treated more favorably, the defendant is entitled to judgment as a matter of law.

In Avery Dennison, the district court's use of nomenclature was perhaps imprecise. Rather than hold that the plaintiff failed to establish a prima facie case, the district court should have held that the EEOC had not persuaded the court that the facts constituting a prima facie case had been proved. As a practical matter, this would have required a finding against the EEOC on "the ultimate issue of discrimination" and judgment in favor of Avery Dennison. The opinion of the court in the present case, in fact, accepts that "evidence that bears upon elements of the prima facie case can also come into play in assessing the ultimate question of discrimination," Op. at 30, even as it castigates the district court for its error in "refocus[ing] on the question of whether a plaintiff established her prima facie case" rather than "look[ing] to the ultimate question of discrimination." Op. at 824.

This distinction appears to me more a matter of semantics than of substance. If, based on the evidence presented at trial, a reasonable jury could not find that the facts constituting a prima facie case were proved by a preponderance of the evidence, then the defendant will be entitled to judgment as a matter of law. Contrary to the court's opinion, Aikens does not "instruct[] us otherwise," Op. at 26, and Aikens certainly "does not stand for the proposition that a district court is flatly prohibited from considering the adequacy of the prima facie case after the case as a whole has proceeded to trial, even when the defendant has never asked the district court to consider the question." Avery Dennison, 104 F.3d at 865 (Ryan, J., dissenting).

To be sure, parts of Aikens are rather cryptic. Nevertheless, I cannot agree with the court's statement in the present case that "an impressive number of appellate courts have applied [Aikens's] holding in the same way that Avery Dennison did." Op. at 24. I am, in fact, not aware of any other appellate court that has applied Aikens in quite the same way that Avery Dennison did.

For example, Barbour v. Browner, 181 F.3d 1342 (D.C. Cir. 1999), cited by the court in the present case, see Op. at 24 n.10, was a Title VII case that was tried to a jury. The jury found for the plaintiff, and the district court denied the defendant's motion for judgment as a matter of law. On appeal, the District of Columbia Circuit reversed, concluding that on the basis ofthe evidence presented at trial, a reasonable jury could not have concluded that the plaintiff was similarly situated to the person whose relatively favorable treatment the plaintiff claimed was circumstantial evidence of discrimination. See Barbour, 181 F.3d at 1345.

Barbour duly noted that because the case had been fully tried, "the question whether [the plaintiff] established a prima facie case is irrelevant." Id. at 1347 (citing Aikens). But then Barbour concluded that "[t]his does not mean, however, that in our analysis of 'the ultimate question of discrimination vel non,' we are obliged to pretend that there is evidence supporting a prima facie case when in fact there is not." Id. (citation omitted). I believe that the District of Columbia Circuit's reasoning was correct. According to Avery Dennison and the opinion of this court today, however, Barbour revisited the prima facie case after trial and thus did exactly (or almost exactly) what Aikens supposedly forbids.

Similarly, in Coffey v. Dobbs International Services, Inc., 170 F.3d 323 (2d Cir. 1999), also cited by this court, see Op. at 24 n.10, the Second Circuit overturned a jury verdict for the plaintiff on a Title VII retaliation claim, concluding that the district judge erred by not granting the employer's post-verdict motion for judgment as a matter of law. See Coffey, 170 F.3d at 327. Coffey cites Aikens for the proposition that "[b]ecause this case has been fully tried on the merits, we need not determine whether [the plaintiff] established a prima facie case." See id. at 326. Then Coffey concludes that because the plaintiff failed to present any evidence suggesting that she was treated differently than other individuals who were not retained when Dobbs, the employer, sold its Albany, New York operations to a new company, and because she had never requested that Dobbs transfer her to one of its other locations, "no reasonable jury could find that Dobbs's failure to transfer [the plaintiff] was retaliatory." Id. at 327.

The Second Circuit in Coffey expressly stated that it was examining "the ultimate question of discrimination vel non," id. at 326, and there can be little doubt that a plaintiff who has introduced no evidence that she was treated differently than others who did not engage in protected activities will be unable to prove to a reasonable jury that she was the victim of retaliation. See, e.g., Morris v. Oldham County Fiscal Court, 201 F.3d 784, 792 (6th Cir. 2000) (noting that a causal connection between the protected activity and the adverse employment action or harassment is an element of a prima facie case for Title VII retaliation). But if one were inclined to take issue with the end result in Coffey, one could just as easily say that what the Second Circuit did was revisit the prima facie case (i.e., whether the plaintiff presented sufficient evidence that the adverse employment action was causally related to her engaging in protected activities) after the case had already been deemed worthy of submission to a jury.

The outcome of cases utilizing the McDonnell Douglas-Burdine burden-shifting formula should not turn on whether the district court concludes after trial that the plaintiff failed to establish a prima facie case (technically the wrong inquiry), or concludes that the plaintiff failed to produce sufficient evidence to permit a reasonable trier of fact to find in his favor on the "ultimate issue of discrimination"--because he produced insufficient evidence to support one of the facts constituting the prima facie case (technically the correct inquiry). Because the difference is essentially one of semantics, it is difficult to understand why Avery Dennison reversed the judgment of the district court, rather than affirming on different grounds after correcting the district court's terminology.

In view of the factual finding of the district court in Avery Dennison that the employer's actions were not causally related to Willis's participation in protected activities, there was no way for any reasonable trier of fact to find in the EEOC'sfavor on "the ultimate issue of discrimination"--unless, of course, the district court's finding of no causal relationship was clearly erroneous. If that finding had been clearly erroneous, then judgment for the EEOC might logically have been required. But Avery Dennison did not conclude that the district court's finding of no causal relationship was clearly erroneous. Instead, it purported to find a legal error that required reversal, and then sent the case back to the district court, where judgment was re-entered for Avery Dennison (supposedly on other grounds) and then affirmed by another panel of this court.

Perhaps the reason that Avery Dennison attempts to limit the district court from revisiting the prima facie case once the matter goes to trial is because it overestimates what the plaintiff is then required to prove in order to win on the merits. The majority in Avery Dennison wrote that once the plaintiff has made out a prima facie case and the defendant has produced evidence of a legitimate nondiscriminatory reason for its actions, the "plaintiff is required to overcome the additional obstacle of the defendant's rebuttal and convincingly demonstrate the existence of discrimination," and that "St. Mary's v. Hicks places the additional burden on Plaintiffs that they must prove not only that the defendant's reason was a pretext, but that the real reason was discrimination." Avery Dennison, 104 F.3d at 861-62.

This seems to me to be a misreading of St. Mary's Honor Center. To be sure, the plaintiff in a Title VII case, in order to succeed, must persuade the trier of fact that the real reason for the defendant's actions was discrimination. But in the ordinary course, the trier of fact may reasonably infer that the real reason for the defendant's actions was illegal discrimination simply by virtue of being persuaded (1) of the facts of the plaintiff's prima facie case, and (2) that the defendant's proffered legitimate nondiscriminatory explanation is pretextual. See, e.g., Reeves v. Sanderson Plumbing Prods., Inc., 120 S. Ct. 2097, 2108-09 (2000) (reasoning that because "discrimination may well be the most likely alternative explanation" once the trier of fact is persuaded that the employer's legitimate nondiscriminatory explanation is false, "a plaintiff's prima facie case, combined with sufficient evidence to find that the employer's asserted justification is false, may permit the trier of fact to conclude that the employer unlawfully discriminated"); Furnco Constr. Corp. v. Waters, 438 U.S. 567, 577 (1978) ("[W]hen all legitimate reasons for rejecting an applicant have been eliminated as possible reasons for the employer's actions, it is more likely than not that the employer . . . based his decision on an impermissible consideration.").

Recently, the Supreme Court has suggested that there may be cases in which, "although the plaintiff has established a prima facie face and set forth sufficient evidence to reject the defendant's explanation, no rational finder could conclude that the action was discriminatory." Reeves, 120 S. Ct. at 2109 (partially disapproving Kline v. TVA, 128 F.3d 337 (6th Cir. 1997)). But one can only assume that such cases will be rather unusual. (St. Mary's Honor Center, however, was evidently one of those cases; the district court, the trier of fact in that case, disbelieved the defendants' proffered explanation and found that there was "a crusade to terminate" the plaintiff, but declined to infer that the crusade was racially motivated. St. Mary's Honor Ctr., 509 U.S. at 508.).

But let us be realistic. The most reasonable inference for jurors to draw, once they disbelieve the defendant's proffered explanation for its actions, will ordinarily be that the real reason the defendant acted as it did was illegal discrimination. See Sanderson, 120 S. Ct. at 2108-09; see also Fisher, 114 F.3d at 1373(Newman, C.J., dissenting) ("Though I agree . . . that 'a finding of pretext, together with evidence comprising a prima facie case, is not always sufficient to sustain an ultimate finding of intentional discrimination,' it will be a rare case where this is not so.") (citation omitted).

In short, I believe that if an appellate court is persuaded that the plaintiff in a Title VII case produced sufficient evidence at trial for a reasonable jury to have found in the plaintiff's favor, and that as a consequence the district court's post-trial grant of judgment as a matter of law for the defendant was erroneous, the appellate court should simply say so. When that happens, the district court's error is that it improperly substituted its judgment for that of the jury, not that the district court somehow contradicted now-binding implied findings that it supposedly made simply by letting the case go to trial.

If Avery Dennison's prohibition against revisiting the prima facie case is taken literally, meaning that the district court cannot question the proof constituting the plaintiff's prima facie case once the action is allowed to proceed to trial, then the decision is erroneous for all of the reasons set forth above. I also believe that this literal interpretation would place Avery Dennison in conflict with this court's prior decision in Gafford v. General Electric Co., 997 F.2d 150 (6th Cir. 1993), for the reasons well-stated by both the dissent in Avery Dennison and by the district court below. If, on the other hand, the thrust of Avery Dennison is only intended to require the district court to focus on the ultimate issue of discrimination once all the proof is in--a focus that includes consideration of whether the plaintiff has produced sufficient evidence to support all of the elements that constitute a prima facie case--then Avery Dennison is simply a trap for unwary district court judges who use the wrong terminology.

Either way, it appears that Avery Dennison has caused unnecessary confusion within the circuit, as evidenced by Judge Ryan's dissent in Avery Dennison itself and by the opinions of the district court below in both this and other cases. I thus join in the judgment of the court, but disagree with its defense of Avery Dennison, which, in my opinion, only perpetuates the confusion.