EDITH H. JONES, Circuit Judge.

Dr. Marise Gottlieb brought a Title VII action against Tulane University of Louisiana, et. at, alleging that, because of her sex, she was terminated from her position at the School of Public Health and Tropical Medicine and was “demoted” to the rank of Research Associate Professor in the School of Medicine. After a bench trial, the district court held that Gottlieb failed to establish intentional discrimination. In a pleading filed after trial but before the decision on the merits, Gottlieb also contended that after filing her Title VII action against the university, the School of Medicine took retaliatory action against her. The district court refused to reopen the case to consider the retaliation claim because of an asserted lack of jurisdiction. This court is asked to review the district court’s findings against Gottlieb, and the failure to reopen the case. Tulane cross-appeals the denial of its request for attorneys’ fees and costs. We affirm the district court’s handling of the disparate treatment case, but we reverse and remand for further proceedings on Dr. Gottlieb’s retaliation claim.

FACTUAL BACKGROUND

In 1974, plaintiff Dr. Marise Gottlieb, an M.D. epidemiologist, was a tenured member of the Rutgers University Medical School. In the fall of that year, after her husband, Dr. Arthur Gottlieb, accepted a department chairmanship at Tulane University School of Medicine, Dr. Gottlieb began employment discussions with Tulane. In February, 1975, Dr. Gottlieb accepted a full-time dual appointment as an associate professor in the School of Medicine and the School of Public Health,1 with time allowances granted for work at the Louisiana Department of Health and the Louisiana Tumor Registry.

Gottlieb’s appointment, as confirmed by her October 10, 1975 appointment letter, was a non-tenured “special” appointment2 with employment beyond one year dependent on “supportive funds, institutional and program needs and satisfactory performance.” All of Dr. Gottlieb’s subsequent appointments were one year “special” appointments dependent on these same conditions. In 1977, however, in response to a Tulane University Senate directive, the Medical Center began reviewing all special appointments to determine the propriety of converting them to regular appointments. Dr. Gottlieb was never converted to a “regular” appointee.

The following chronology details Dr. Gottlieb’s claims of sex discrimination:

i) The School of Public Health. In October 1978, a departmental faculty committee 3 consisting of three men and six wom*281en, reviewed Dr. Gottlieb’s performance and expressed a number of concerns which they felt jeopardized her reappointment as a part-time associate professor. The committee pointed to an unexpectedly low number of publications since being at Tulane, compared with her past performance; they also felt that Dr. Gottlieb did not “allow colleagues the latitude that their professional status dictates that they deserve”;, and they felt that she had been obstructive and failed to cooperate in furnishing her colleagues with access to data sources. Finally, they expressed “concern about other aspects of her professional behavior toward colleagues within and without the university” and feared “that some of her actions could damage the professional reputation of the department and even that of the school vis-a-vis the community.”

Most of the faculty’s concern resulted from a dispute over a National Cancer Institute (“NCI”) grant for the study of cancer in Southern Louisiana. Dr. Gottlieb attempted to obtain the NCI grant and submitted a proposal for the research project through the School of Medicine. A proposal was also submitted by two of Gottlieb’s colleagues at the School of Public Health in conjunction with the LSU School of Medicine. Gottlieb had not informed the faculty or the dean of the School of Public Health of her proposal. After learning of the competing proposal, and her colleagues’ involvement, Gottlieb requested that Tulane forbid the competing faculty members from participating in the LSU proposal. This request was denied and Dr. Gottlieb subsequently submitted a formal protest. The NCI rejected the protest and commented on Dr. Gottlieb’s critical attitude toward her peers.

Dr. Gottlieb’s conduct in relation to the NCI grant also strained university relations with LSU. Pending her protest, Dr. Gottlieb contacted various hospitals designated to be involved with the NCI research and informed them of her bid protest. The district court concluded that the letter included disparaging remarks about the competing bid and “came very close to requesting non-cooperation with her successful competitor.” Responding to this letter, the dean of the LSU School of Medicine sharply criticized Dr. Gottlieb’s actions and threatened to withdraw university support from the Louisiana Tumor Registry if Dr. Gottlieb was reappointed to that organization. Although Dr. Gottlieb sent a conciliatory letter to the hospitals, she was not reappointed to the Tumor Registry.

In June 1979, the faculty of the School of Public Health unanimously recommended that Gottlieb be given a one-year “terminal” appointment for the academic year 1979-1980. The faculty committee explained that her actions while at Tulane were “non-collegial, overly aggressive and detrimental” to the university. The committee concluded that it “would have a difficult time defending a colleague whose definitions of cooperation and limits of autonomy can depart so radically from those of the majority as yours have in some instances. Given our limited financial resources, we are reluctant to invest in a faculty member to whom we cannot guarantee this kind of support.” Gottlieb accepted this terminal appointment under protest, however, and appealed the decision to the departmental grievance committee. There her termination was again unanimously approved by the grievance committee.

Dr. Gottlieb contends that her behavior was neither unreasonable nor unwarranted and that even if she was aggressive and “non-collegial,” this type of behavior was tolerated in the males of the School of Public Health, hence, the justification for her dismissal was merely pretextual. She introduced testimony of colleagues who found her to be very cooperative and unabrasive. Her evidence also suggested that the School of Public Health was male dominated and that aggressiveness in females was not tolerated to the extent that it was in males. Dr. Gottlieb presented statistical evidence that tended to show a disparate salary structure between males and females, suggesting a pattern of discrimination.

*282ii) The School of Medicine. Dr. Gottlieb also claims that she was discriminated against by the Tulane School of Medicine. In the fall of 1979, the faculty senate approved an amendment to faculty guidelines which established a new tier of appointments for the School of Medicine. The amendment created a non-tenured “research appointment” for those whose activities center primarily around research and who performed minimal teaching duties. After completing her 1979 terminal appointment at the School of Public Health and her 1979 part-time appointment at the School of Medicine, Dr. Gottlieb was notified that the School of Medicine had converted her from an Associate Professor to a Research Associate Professor. By its terms, the new appointment depended solely on her ability to secure research grants and other funding. Dr. Gottlieb claims that this change in status was effectively a “demotion” because such a position never leads to tenure, it is considered second-rate, and the title is a disadvantage in obtaining research funds because of its lack of prestige. Moreover, Dr. Gottlieb asserts that her demotion was based on her sex and is consistent with the School of Medicine’s treatment of all women. She points out that the School of Medicine has many women on its faculty, but only one is a full professor and only one other woman holds the rank of associate professor.

INTENTIONAL DISCRIMINATION

i) Prima Facie Case.

Gottlieb attaches particular significance to her argument that the district court misapplied the three-step analysis of McDonnell-Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973), and Texas Department of Community Affairs v. Burdine, 460 U.S. 248, 101 S.Ct. 1089, 67 L.Ed.2d 207 (1981), which, she argues, “impermissibly colored” the district court’s ultimate findings. Her reliance on the district court’s preliminary analysis is misplaced. Following United States Postal Service Bd. of Governors v. Aikens, 460 U.S. 711, 103 S.Ct. 1478, 75 L.Ed.2d 403 (1983), we have often repeated that “by the time a fully-tried case reaches us on appeal, the parties’ showing at the preliminary levels of the framework is largely irrelevant. We need address only the propriety of the ultimate finding of discrimination vel non.” EEOC v. Exxon Shipping Co., 745 F.2d 967, 972 (5th Cir. 1984) (citations omitted). This rule was reiterated most recently in Merrill v. Southern Methodist University, 806 F.2d 600 (5th Cir.1986). Gottlieb’s attack on the district court’s findings at the preliminary stage is thus irrelevant, as we must review the ultimate conclusion that Tulane did not intentionally discriminate against Dr. Gottlieb.4

ii) Intentional Discrimination.

The plaintiff bears the burden of proving intentional discrimination, and the district court's finding on this issue is reviewed under the clearly erroneous standard. Fed. R. Civ.P. 52(a); Pullman-Standard, Div. of Pullman, Inc. v. Swint, 456 U.S. 273, 102 S.Ct. 1781, 72 L.Ed.2d 66 (1982). Unless this court is left “with the definite and firm conviction that a mistake has been committed,” we will not reverse. United States v. United States Gypsum Co., 333 U.S. 364, 396, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948).

After fully reviewing the record, we hold that the district court’s finding that Tulane did not intentionally discriminate against Dr. Gottlieb is not clearly erroneous.5 First, Tulane had legitimate non-discriminatory reasons for terminating Dr. Gottlieb’s part-time special appointment to

*283the School of Public Health. The faculty review committee pointed to Dr. Gottlieb’s limited record of publication since her arrival at Tulane as a partial rationale for her terminal appointment. Additionally, the committee cited her “non-collegial”6 behavior as another justification. Instances of Gottlieb’s abrasive and uncooperative behavior are well documented. It is understandable that a university faculty facing serious budgetary constraints, like the School of Public Health, would be unwilling to recommend for a tenured appointment a member it found uncooperative and whose actions had publicly undermined relations with LSU and institutions in the medical community. Second, the School of Medicine had non-discriminatory reasons for changing Gottlieb’s status from a part-time associate professor to a full-time research associate professor. The School of Medicine stated that, in line with its general policy, her minimum teaching duties and her lack of clinical residency training in internal medicine required changing her status.7

We also find no merit in Dr. Gottlieb’s contention that Tulane’s justifications were merely pretextual. Gottlieb’s argument relies on statistics compiled by her expert, Dr. Gordon Henderson, which showed disparate mean salaries between males and females at the Tulane Medical Center. While such statistics are admissible for the purpose of showing motive or intent, they are not determinative of an employer’s reason for the action taken against the individual. Plemer v. ParsonsGilbane, 713 F.2d 1127, 1135 (5th Cir.1983) (citing Terrell v. Feldstein Co, 468 F.2d 910, 911 (5th Cir.1972)). This indirect evidence tended to show unequal treatment within the Medical Center, but the district court gave the statistics little or no weight because it found them “inaccurate in several areas, insignificant in others, and based upon questionable assumptions.” The district court cited three significant flaws in the statistical analysis. First, Dr. Henderson’s data-set included all full-time faculty members in both the School of Public Health and the School of Medicine during the 10-year period 1974 to 1984. Not only are part-time employees excluded from this sample, but Gottlieb dropped from the sample as well in 1981 when she began to work for Tulane on a part-time basis. Second, the variable “length of service” is limited to a maximum of 10 years, regardless of actual length of employment. The existence of a substantial number of males with high salaries and/or high rank as well as more than 10 years service was thus arbitrarily excised from the overall salary analysis. Finally, the use of the mean, rather than the median, skewed salary comparisons in favor of males. Based on these justified criticisms of Gottlieb’s statistics, the district court did not abuse its discretion in declining to attach controlling significance to them. See, e.g., Merrill v. Southern Methodist University, 806 F.2d at 606.

THE RETALIATION CLAIM

Following trial, and while her case was under submission to the district court, Gottlieb filed a motion for a new trial or to reopen on the question whether Tulane re*284taliated against her for exercising Title VII rights. She also sought a temporary restraining order to enjoin Tulane from enforcing its conflict of interest policy and refusing to pay her for work on a specific university grant. The judge denied both motions and concluded that because Gottlieb failed to file an EEOC charge on her retaliation claim a jurisdictional prerequisite was absent.8 Dr. Gottlieb properly contests this ruling.

In Pettway v. American Cast Iron Pipe Co., 411 F.2d 998, reh’g and reh’g en banc denied, 415 F.2d 1376 (5th Cir.1969) (Pettway II) this court held that the district court erred in refusing to treat a petition for injunctive relief based on a Title VII retaliation claim as ancillary to Pettway I, the original Title VII action which was then on appeal. Gupta v. East Texas State Univ., 654 F.2d 411 (5th Cir. 1981), clarified Pettway II, concluding that “it is unnecessary for a plaintiff to exhaust administrative remedies prior to urging a retaliation claim growing out of an earlier charge.” 654 F.2d at 414. Tulane strenuously seeks to distinguish Gupta based on the relative procedural posture of each case. Gupta brought his retaliation claim to the attention of the district court before trial, while Gottlieb asserted her retaliation claim after trial. This is a “distinction without a difference.” The concerns of policy and judicial economy that were discussed in Gupta are applicable here. Requiring Dr. Gottlieb to resort to the EEOC a second time on a retaliation claim would serve no purpose “except to create additional procedural technicalities when a single filing would comply with the intent of Title VII.” Gupta, 654 F.2d at 414 (citations omitted). Because Dr. Gottlieb’s initial Title VII intentional discrimination claim was properly before the court, it had ancillary jurisdiction over the retaliation claim. Consequently, this portion of the case must be remanded for further proceedings.

COSTS AND ATTORNEY’S FEES

Fighting each step of the way, the parties dispute the award of costs and Tulane contests the failure of the district court to award prevailing defendant’s attorneys’ fees under Christiansburg Garment Co. v. EEOC, 434 U.S. 412, 98 S.Ct. 694, 54 L.Ed.2d 648 (1978). Christians-burg permits a discretionary award of fees if the district court determines that the plaintiff’s claim was groundless, without foundation, vexatious, frivolous, or unreasonable. The district court found against Tulane, and we ascertain no abuse of its discretion.

The district court taxed $6,325.75 costs against Gottlieb. Gottlieb seeks a reduction in the amount of costs in an unspecified amount because: 1) a significant portion of the costs of deposition transcripts (almost $6,000) represents depositions taken in preparation for trial on both federal Title VII and state breach of contract claims, therefore, the costs awarded should be apportioned to the federal claim only; and 2) certain copies of deposition transcripts ($337.50) were made exclusively for the convenience of the defense counsel. The district court adopted the magistrate’s report and recommendation on costs and refused to reduce the award. No abuse of discretion appears in the record on this issue.

Tulane University also appeals, requesting that witness fees of $27,753.77 be added to its costs and taxed against Gottlieb. Recently, in International Woodworkers of America, AFL/CIO Local No. 5-376 v. Champion International Corp., 790 F.2d 1174, 1181 (5th Cir.) (en banc) cert. granted, — U.S.-, 107 S.Ct. 568, 93 L.Ed.2d 573 (1986), we held that in Title VII cases “the fees of non-court-appointed expert witnesses are taxable by federal courts in non-diversity cases only in the amount specified by [28 U.S.C. § 1821], ex*285cept that fees in excess of that amount may be taxed when expressly authorized by Congress” or when one of the three narrow equitable exceptions to the American Rule applies.9 No express authorization by Congress for the award of Tulane’s excess fees is claimed and none of the equitable exceptions noted in International Woodworkers is applicable. Therefore, Tulane may not recover its expert witness fees as costs. We note that Tulane never requested nor did the district court award the minimum statutory fee provided for in 28 U.S.C. § 1821(b). Accordingly, the decision to award Tulane total costs of $6,825.75 is affirmed.

For the foregoing reasons, the judgment of the district court is AFFIRMED in part, REVERSED in part, and REMANDED for further proceedings on appellant’s claim of retaliation.

. Together, the School of Medicine and the School of Public Health and Tropical Medicine make up the Tulane University Medical Center.

. Article II, section 4 of the Statement on Academic Freedom, Tenure, and Responsibilities contained in the 1974 Tulane Faculty Handbook provided that there are two types of faculty appointments: “special” and "regular.” Section 5 provided:

A special appointment neither gives tenure nor is to be regarded as a probationary appointment that may lead to tenure. The following are examples of special appointments: ******

(d) an appointment for a definitive time limit, and (e) an appointment concerned with a project for which continued financial and salary support is not assured.

Section 6 provided:

A regular appointment may be either probationary, with the prospect of tenure, or continuous, namely an appointment with immediate tenure.

The record reflects, and the district court concluded, that due to budgetary constraints all faculty appointments from 1973 through 1978 were designated "special" appointments, with the exception of “chair persons” and department heads.

. The preamble to Tulane’s Statement on Academic Freedom, Tenure, and Responsibilities states that “faculty status, including appointments, reappointments, decisions not to reappoint, promotions, the granting of tenure, discipline and dismissal, is primarily a faculty responsibility." 1977 Tulane University Faculty Handbook.

. Gottlieb also asserts that the district court failed to set forth complete findings of fact so as to "forestall any meaningful review by the appellate court.” This is untenable, in light of the meticulous 17-page recitation of findings and conclusions by the district court. There was ample detail upon which to predicate appellate review.

. Gottlieb’s argument that the district court’s findings are clearly erroneous is based on myriad references to facts contained in the record. We address only what we consider to be the most significant challenges. We have, however, considered all of her contentions, none of which lead us to conclude that the district court erred in its determination.

. While there is a danger in allowing subjective evaluations to be used in rebutting a plaintiffs claim of discrimination, we have recognized the necessity for such evaluation techniques in the university context. Wilkins v. University of Houston, 654 F.2d 388, 401 (5th Cir.1981) vacated on other grounds, 459 U.S. 809, 103 S.Ct. 34, 74 L.Ed.2d 47 (1982). See also, Zahorik v. Cornell Univ., 729 F.2d 85, 93 (2d Cir.1984) (holding that "peer judgments as to departmental needs, collegial relationships and individual merit may not be disregarded absent evidence that they are a facade for discrimination”); Lieberman v. Gant, 630 F.2d 60, 67 (2d Cir.1980) (Friendly, J.) folding that no inference of discrimination can be drawn from reasonable, although partially subjective, evaluations of academic qualifications).

. We here assume without deciding that Dr. Gottlieb’s claimed "demotion” can be the basis of a Title VII action. Although Dr. Gottlieb asserts that a loss of status accompanied the "research" appointment, the district court concluded that since Dr. Gottlieb moved from one non-tenured position to another, the new appointment was not a demotion.

. The filing of an administrative complaint is ordinarily a jurisdictional prerequisite of a Title VII action. Ray v. Freeman, 626 F.2d 439, 442 (5th Cir. 1980), cert. denied, 450 U.S. 997, 101 S.Ct. 1701, 68 L.Ed.2d 198 (1981).

. Even if the abuse of discretion standard were to apply to the district court’s decision regarding the award of expert witness fees as costs, See Fed.R.Civ.P. 54(d), we are unable to conclude that the district court abused its discretion in not taxing Gottlieb with expert witness fees in this case.