**1 Plaintiffs William L. Johnson, Julian W. Moore, and Keith M. Holley, police officers employed by Defendant Metropolitan Police Department (“MNPD”) of Nashville, Tennessee, appeal the district court's grant of summary judgment on Plaintiffs' reverse discrimination claims. Plaintiffs were passed over for promotion as a result of a departmental policy that they allege favored minority and female candidates. They sued Defendants MNPD; Metropolitan Government of Nashville and Davidson County (“Metro”); MNPD's former chief, Ronal W. Serpas; and other MNPD personnel, alleging violations of 42 U.S.C. § 1983; Title VII of the 1964 Civil Rights Act, 42 U.S.C. § 2000e; and the Tennessee Human Rights Act (“THRA”), Tenn.Code Ann. §§ 4-21-401–4-21-408. The district court dismissed several of Plaintiffs' claims and, after the parties conducted discovery, granted Defendants' motion for summary judgment. Plaintiffs now appeal that judgment, several evidentiary rulings, and the district court's decision to tax costs against them. For the reasons that follow, we AFFIRM the district court's judgment in all respects.
I. The MNPD Promotion Policy
Prior to 2006, it was MNPD's policy to promote officers strictly through the use of standardized tests. An officer applying for a promotion completed a written civil service examination and took part in a performance assessment designed to evaluate the officer's skills and leadership ability. The score received by the officer on each examination was combined into a composite score. The candidates were ranked according to their composite scores, and the departmental chief was required to promote the officers according to ranking, even if the chief believed that a lower-ranked officer was more qualified than a higher-ranked officer.
This policy was the target of criticism in some corners. For one, an outside consultant performed an audit of the MNPD in 2002 and concluded that this method of promoting officers failed to take important criteria into consideration, such as each candidate's managerial skill and past performance. There was also concern about whether the policy inhibited minority candidates from earning promotions, an issue that the local media reported in news articles around 2003. In addition, there was a broader concern internally about the diversity of the MNPD's ranks. For example, Metro and MNPD officials complained to the attorney who represented the police officers union about a lack of diversity in the police force's upper ranks during a meeting regarding negotiations on changes to the promotion policy. These officials *526 theorized that the testing scheme exacerbated the problem.
Well before changes were implemented to MNPD's testing system, MNPD officials made other attempts to increase the department's racial diversity. In December 2003, Deputy Chief Anderson wrote a memorandum to acting Chief of Police Deborah Faulkner urging her to promote fifteen officers to the rank of sergeant before the current slate of officers eligible for promotions expired on January 9, 2004 (the “Anderson memo”). According to the memorandum, the current slate of fifteen officers included five minority candidates, but the upcoming list did not include a single minority officer. As the memorandum further explained, because a list of candidates was effective for several years, and because minority candidates reached the candidate list in lower numbers than white candidates under the test-based promotions policy, it would be impossible for minority candidates to be promoted if they were not selected from the current slate. Faulkner announced the promotion of several officers shortly after the date of the memorandum, and the group included several minority candidates.
**2 Against this background, MNPD's promotion policy underwent significant change in 2004. Defendant Ronal W. Serpas, who was hired as the MNPD's permanent chief in 2004, shared the prevalent concerns about the test-based promotion policy's soundness. He and Metro's Human Resources Department (“Metro HR”) recommended changing the policy, which the Metropolitan Civil Service Commission (“the Commission”) did in 2006. Metro hired an outside contractor to design and implement a new policy to replace the test-based promotion policy.
The new policy continues to feature standardized tests, but now also gives the police chief a measure of discretion in deciding whom to promote. Under the new policy, officers must still complete the written civil service examination and skills and leadership assessment. The test results are again combined into a composite score, with the assessment counting for 80% of a lieutenant candidate's score, and 70% of a sergeant candidate's score. The candidates are then ranked according to their scores.
However, under the new policy, when the chief prepares to promote an officer, he is given an eligibility roster that lists the seven highest-scoring candidates eligible for promotion to the rank of lieutenant or lists the nine highest-scoring candidates eligible for promotion to sergeant. The roster does not rank the candidates according to their composite scores. Rather, the roster simply lists the officers' names in alphabetical order and does not reveal their scores. Each officer on the list is considered equally qualified for promotion. The chief is not restricted in choosing among candidates from the list. If the chief has multiple vacancies to fill, he can fill them entirely from the roster given to him. If, however, the chief finds that none of the candidates are suitable for promotion, he may also choose to promote none of them, and he may instead request a new roster with the next tier of candidates. If the chief obtains another roster, the Commission removes the names of the candidates not selected and then adds the next highest-scoring candidates.
II. The Disputed Promotions
This litigation concerns several lieutenant and sergeant vacancies that became available during 2006 and 2007. Plaintiffs Johnson and Moore applied for promotions to lieutenant, and Plaintiff Holley applied to become a sergeant. Each officer was passed over and now contends that considerations of race and gender improperly affected Serpas' promotion decisions.
*527 Serpas made his selections in three separate rounds between October 2006 and May 2007. Moore's composite test score ranked him seventh overall, and therefore his name appeared on the first roster submitted to Serpas. Moore was not selected, in spite of the fact that a commander allegedly invited Moore to a meeting of higher-ranked officers under the expectation that Moore would be promoted. Johnson was ranked ninth among the candidates and was therefore not on the first roster given to Serpas. Johnson's name appeared on the second roster that was submitted to Serpas, and Serpas did not select Johnson for promotion. According to Johnson and Moore, three female candidates and two black male candidates ranked lower than them were selected instead. Moore and Johnson contend that only race and gender can account for the fact that these candidates were selected in front of them. Defendants note, however, that three white male candidates who ranked below both Moore and Johnson also were selected for promotion.
**3 For his part, Holley was ranked as a “police officer II” and applied to become a sergeant around the same period. Serpas made selections for the sergeant positions in three separate rounds between February and September 2007. Like Johnson, Holley was not on the first roster submitted to Serpas, because Holley's composite test score ranked him sixteenth among the sergeant candidates. In his affidavit testimony, Holley asserted his name eventually moved up into the roster for promotion, but that twenty-nine officers were eventually promoted to sergeant instead of him. Holley does not allege how many female and black male officers were promoted in front of him, but he contends that the highest-ranked black male officer in the initial rankings was ranked tenth and that the next-highest-ranked black male officer was ranked lower than twentieth. In their defense, Defendants assert that Serpas promoted eleven white males who obtained composite scores lower than Holley and declined to promote one white female who scored below Holley.
Plaintiffs nonetheless contend that Serpas and other MNPD supervisors abused the discretion granted them under the new promotion policy in order to favor promotions of minority and female candidates. In support of their contention, they cite, in addition to Serpas' selections noted above, several statements that MNPD officials made to local newspapers. For example, MNPD spokesperson Dan Aaron explained the department's rationale for the changes to the promotion policy to The Tennessean
in October 2006. The article stated:
Promoting the best candidates possible was the priority, and the chief was pleased there were diverse candidates to pick from near the top of the list, Aaron said.
“The goal of this police department is to mirror the population of the city to the greatest extent possible,” Aaron said. “When presented with the opportunity to promote bright, qualified minority candidates, you always give those persons consideration.”
wrote another article about the promotions in April 2007, this time emphasizing non-minority officers' complaints about the policy. Aaron was again quoted, this time saying, “[i]f you have two candidates who are essentially equal and believe that both would make very good supervisors, and if your choice is to make the department more diverse, you would probably elect to include diversity in your choice.” In an article from another local news agency, Aaron was quoted as saying that MNPD changed its promotion policy in order to increase diversity among MNPD leadership.
*528 III. The Anonymous Supervisor Surveys
According to Plaintiffs, their claims of reverse discrimination gain further support by Serpas' decision to use anonymous surveys completed by the candidates' supervisors as part of his selection process. At some point in 2006 or 2007, Serpas directed Eric Cardinal, an MNPD technology analyst, to develop a computer-based survey for department supervisors to fill out in order to report their assessments of the various lieutenant and sergeant candidates. As we explain below, Cardinal developed the program and disbursed it digitally to the supervisors, who completed the surveys prior to each round of promotions. Serpas used the surveys as part of his decision about whom to promote. While the instructions for the surveys are in the record, the surveys themselves are not available to us, because Serpas directed Cardinal to dispose of them once Serpas was finished using them. Plaintiffs contend that in allowing the surveys to be deleted, Defendants violated federal law and breached their discovery obligations. Upon learning of the destruction of the surveys, Plaintiffs moved for a default judgment or, alternatively, a negative inference against Defendants. The district court denied both requests.
**4 Cardinal described the survey in his deposition testimony. Cardinal sent an email to the appropriate supervisors that contained an internet link, which, when clicked, opened a page on the MNPD's intranet. The supervisor was directed to fill in the log-in information and was then given a set of instructions (which are contained in the record). The supervisor then arrived at the survey form, titled “Promotion Readiness and Potential Ratings for Sergeant Candidates,” with the lieutenant survey accompanied by a similar title. The survey form explained that the survey's purpose was “to seek the opinions of supervisory and management personnel who have worked directly with, or otherwise have personal awareness of the skills, abilities, and knowledge possessed by, candidates eligible for promotion to the rank of Sergeant [and Lieutenant] as to their overall readiness and skill.” The instructions directed the supervisor to “consider the candidate's potential to lead, manage resources, and successfully perform the duties that he or she will be required to perform.”
The supervisor was then asked to rate each candidate in several categories on a scale of 1 to 3, with 1 standing for “lowest potential” and 3 standing for “highest potential.” In the interest of creating a spread in the candidates' scores, the supervisor was only allowed to award a certain number of 3, 2, and 1 rankings. The supervisor was instructed not to complete a survey for a candidate with whom he was not sufficiently acquainted. When the supervisor completed the survey, the computer program added his ratings to those of other supervisors regarding a particular candidate. When all the supervisors had completed their surveys, the computer program generated an average score for that candidate, referred to by the parties as a “rollup” score. Supervisors at and above the rank of lieutenant completed surveys for the lieutenant candidates, and supervisors at and above the rank of sergeant completed surveys for the sergeant candidates.
After the supervisors completed each round of surveys, Serpas examined each candidate's rollup score. The rollup score showed each candidate's name, the number of supervisors who rated him, and the candidate's raw and average scores. After Serpas examined the rollups, Cardinal deleted the data and notified Serpas that he had done so. Cardinal could have designed the program to save the data, but *529 Serpas did not direct him to do so. Serpas did, however, save the rollup scores for each candidate, and Defendant turned that evidence over to Plaintiffs during discovery.
In his deposition testimony, Serpas testified that he requested the computerized survey because he had not personally met many of the promotion candidates in his three years of leading the MNPD. He testified that he made his promotion decisions after considering the candidates' employment, sick leave, and disciplinary records; their assignment positions; their educational and training experiences; and any citizen complaints filed against them. According to Serpas, the surveys added to the pool of information he used in his selections, and their simplified format allowed him to avoid the time required to personally interview each supervisor about each candidate. Serpas did not keep any notes documenting his promotion decisions. As Plaintiffs point out, the personnel files Serpas consulted disclosed the race and gender of each candidate. Nevertheless, Serpas testified that race and gender played no part in his decision and that he chose other candidates over Plaintiffs because he decided those candidates “were more likely to be successful” than Plaintiffs.
**5 As a general matter, Serpas did not promote any candidate with an average rollup score below 2.0, and each Plaintiff in the instant case scored below that threshold on each of the three supervisor surveys conducted on their performance. Moore received average scores of 1.58 from twelve supervisor surveys in the first round, 1.59 from seventeen supervisor surveys in the second round, and 1.61 from eighteen supervisors in the final round. Johnson received average scores of 1.56 from sixteen supervisors in the first round, 1.58 from twelve supervisor surveys in the second round, and 1.67 from twelve supervisor surveys in the third round. Holley was scored on only two rounds of surveys, and in both rounds thirty-three supervisors scored him for scores of 1.42 and 1.70. These scores placed each Plaintiff near the bottom of their candidate pools.
Nevertheless, Plaintiffs offer several reasons for which they believe that Serpas' use of the surveys was improper. First, Plaintiffs allege that they were not made aware that the surveys would be used as a basis of Serpas' decision, and they argue that it was unfair for Serpas to use the procedure when it had not been disclosed to the candidates. Plaintiffs also consider it improper that a different number of supervisors rated them during each round of surveys. In response, Defendants contend that this inconsistency was actually a virtue of the surveys, because, in their view, it demonstrates that the supervisors followed the survey directions not to rate the candidates with whom they were unacquainted. Finally, Plaintiffs contend that Serpas used the surveys as a subterfuge to favor minority and female candidates, and they argue that he allowed Cardinal to destroy the surveys in order to hide the department's reverse discrimination.
IV. Plaintiffs' Meetings with their Supervisors and Human Resources
After Serpas made his selections, Moore, Johnson, and Holley sought explanations for their non-promotion from department officials. Moore and Johnson met separately with MNPD Deputy Chiefs Steve Anderson, Honey Pike, and Joseph Bishop. In his meeting, Johnson asked the deputy chiefs what information the department used to decide whom to promote. The deputy chiefs told him generally that they used many types of information. Johnson asked the panel whether his work history was used in the decision, and *530 he asserts that Deputy Chief Anderson answered in a manner suggesting that he did not know whether that information was considered. Johnson told the panel that he believed he was the victim of reverse discrimination and asked the deputy chiefs to investigate his complaint, but he alleges that the deputy chiefs never did so.
Johnson also spoke with other MNPD officials and purportedly received information that supported his view that female and minority candidates received favorable treatment. He spoke with a commander who allegedly told him that “diversity issues needed to be dealt with” in the promotions process. Johnson also spoke with an official within Metro HR. When Johnson asked the official whether the candidates' race and gender affected the promotions decision, the official allegedly answered, “We all know what happened, if you know what I mean.”
**6 Moore also discussed his complaint with the deputy chiefs and recorded his conversation. Moore repeatedly raised the issue of race and its effect on the promotions decision, in response to which Deputy Chief Pike conceded that MNPD did not reflect the community's racial and gender diversity. However, Deputy Chief Pike also insisted that the promotion policy was aimed at identifying the most qualified candidates, regardless of minority status. According to Moore, Deputy Chief Pike urged Moore to examine his own performance, evaluate whether he deserved a promotion, and discuss his performance with his direct supervisors. Moore told the deputy chiefs that he had done so and concluded that he possessed the qualities required to serve as a lieutenant. Like Johnson, Moore expressed his belief that he was the victim of reverse discrimination, but the deputy chiefs allegedly did not investigate the complaint.
Finally, Holley discussed his failure to earn a promotion with the three deputy chiefs as well. According to Holley, none of the deputy chiefs offered him useful information about why he was not promoted. Holley explained that his direct supervisor urged him to reapply for a promotion. He also said that he was more involved in community affairs than other officers and was often approached by other officers for answers about departmental policy and the law. Nevertheless, according to Holley, the deputy chiefs simply gave him a piece of paper listing thirty-three names with his name thirty-second among the list.
V. Procedural History
Dissatisfied with the department's promotion decisions, Plaintiffs filed two separate suits in federal district court. Johnson and Moore sued in September 2007, alleging violations of the Equal Protection Clause of the United States Constitution, 42 U.S.C. § 1983, and the THRA. After Johnson and Moore received right-to-sue letters from the Equal Employment Opportunity Commission, they added claims of disparate impact and disparate treatment under Title VII. For his part, Holley filed a similar suit in January 2008, though his complaint contained no Title VII claims, and he did not sue the MNPD. The district court eventually consolidated the cases.
Defendants Pike, Bishop, Anderson, and MNPD filed successful motions to dismiss and all claims against them were dismissed on May 13, 2008. See
Johnson v. Metro. *531 Gov't of Nashville & Davidson Cnty., Tenn., Nos. 3:07–0979, 3:08–0031, 2008 WL 2066475 (M.D.Tenn. May 13, 2008) (“Johnson I
”). Plaintiffs do not appeal those rulings.
Shortly thereafter, Metro filed a separate motion to dismiss Johnson and Moore's disparate impact claim.
(RE 68, Mot. to Dismiss.) Plaintiffs filed a response and also sought leave to amend their complaint. The district court denied Plaintiffs' motion and dismissed the disparate impact claim. Johnson v. Metro. Gov't of Nashville & Davidson Cnty., Nos. 3:07–0979, 3:08–0031, 2008 WL 3163531 (M.D.Tenn. Aug. 4, 2008) (“Johnson II
**7 Following those rulings, Plaintiffs' remaining claims asserted that Metro and Serpas violated the Equal Protection Clause and Defendants' rights as protected by Title VII and the THRA. Defendants moved for summary judgment in June 2010. During the same month, Plaintiffs moved for a default judgment or, alternatively, for an adverse finding against Defendants, based upon Defendants' failure to completely preserve the records from the supervisor surveys.
On August 24, 2010, the district court denied Plaintiffs' motions for a default judgment and an adverse factual finding, granted summary judgment in favor of Defendants, and entered judgment on all claims in favor of Metro and Serpas. Johnson v. Metro. Gov't of Nashville, Nos. 3:07–0979, 3:08–0031, 2010 WL 3342211 (M.D.Tenn. Aug. 24, 2010) (“Johnson III
In November 2010, the clerk of the court taxed costs against Plaintiffs. The clerk awarded to Metro the costs of defending both Metro and Serpas. Plaintiffs challenged the clerk's taxation of costs, but the district court upheld the award. Johnson, et al. v. Metro. Gov't of Nashville, Nos. 3:07–0979, 3:08–0031, 2011 WL 166320 (M.D.Tenn. Jan. 18, 2011) (“Johnson IV
Plaintiffs timely appealed the district court's rulings. Original jurisdiction exists pursuant to 28 U.S.C. §§ 1331 and 1367. This Court exercises jurisdiction pursuant to 28 U.S.C. § 1291.
I. Spoilation of Evidence
Before addressing the merits, we turn first to Plaintiffs' claim that the district court erred in refusing to impose sanctions against Defendants for spoilation of evidence. Plaintiffs contend that they are entitled an inference of discriminatory animus or to a directed verdict because Defendants failed to preserve the individual results of the supervisor surveys and thereby purposefully deprived Plaintiffs of valuable information supporting their claims of reverse discrimination.
This Court reviews a district court's decision regarding spoilation of evidence for abuse of discretion. Beaven v. U.S. Dep't of Justice, 622 F.3d 540, 553 (6th Cir.2010) (citing Adkins v. Wolever, 554 F.3d 650, 652 (6th Cir.2009) (en banc)). Federal law governs the determination of whether spoilation sanctions are appropriate. Adkins, 554 F.3d at 652. A proper spoliation sanction serves both fairness and punitive functions. Id. To accomplish these goals, a district court has “broad discretion” to order sanctions it deems appropriate, including dismissing the case, granting summary judgment, or imposing an adverse inference based on the lost or destroyed evidence. Id. Recently, this Court articulated a three-part standard for determining whether sanctions are appropriate:
[A] party seeking an adverse inference instruction based on the destruction of evidence must establish (1) that the party *532 having control over the evidence had an obligation to preserve it at the time it was destroyed; (2) that the records were destroyed “with a culpable state of mind;” and (3) that the destroyed evidence was “relevant” to the party's claim or defense such that a reasonable trier of fact could find that it would support that claim or defense.
**8 Beaven, 622 F.3d at 553 (quoting Residential Funding Corp. v. DeGeorge Fin. Corp., 306 F.3d 99, 107 (2d Cir.2002) (citation omitted)).
In applying this three-part standard, we explained that the obligation element is met where a defendant knows evidence might be relevant to future potential litigation. Id. Where, however, there is no notice of potential litigation, there is less cause to believe the evidence was destroyed intentionally or with the intent to cover up incriminating information. See id.
Nevertheless, we also noted that the “culpable state of mind” element may be satisfied by showing only that “the evidence was destroyed ‘knowingly, even if without intent to breach a duty to preserve it, or negligently.’ ” Id. (internal citation, brackets, quotations omitted).
1In the instant case, Plaintiffs argue that Metro was obligated to preserve the individual survey results scored by each supervisor, in addition to the averaged rollup scores. Plaintiffs cite several regulations promulgated by the Equal Employment Opportunity Commission (“EEOC”) and pursuant to Title VII that obligate employers to preserve employment “records.” See
42 U.S.C. §§ 2000(e) 8, 2000(e) 12; 29 C.F.R. § 1602.31. Defendants counter that employers are not required to keep every single piece of paper created during the employment process, see
Rummery v. Ill. Bell Tel. Co., 250 F.3d 553, 558 (7th Cir.2001), and that the rollup scores were sufficient to satisfy their statutory preservation obligations.
We agree with the district court that Defendants ought to have preserved the individual survey scores. The surveys were part of, and indeed played an integral role in, a significant change to an already controversial promotion system. Whether through formal litigation or otherwise, it was reasonably foreseeable that Defendants would face some sort of challenge to the new promotions system. Defendants should have anticipated that officers who were passed over for promotion might question that decision and that Defendants would need to defend their selections. By deleting the individual surveys, Defendants also deprived the officers of valuable information regarding their individual performance, their likelihood for future promotion, and information that might have been used in litigation. Regardless of whether the rollups were the most efficient way for Serpas to review the survey results, the individual scores had value warranting their preservation beyond his decision-making process.
Furthermore, Defendants were statutorily obligated to preserve the surveys under EEOC and Title VII regulations. The individual surveys are more properly viewed as records in and of themselves, rather than the “rough drafts” or “processes” used to create a final employment record. See
id. at 558. Finally, it was technologically and logistically feasible to retain the survey data, and Defendants have provided no convincing explanation for why they failed to do so.
**9 Having determined that Defendants were obligated to preserve the surveys, the next question is whether Defendants destroyed the evidence with requisite culpable state of mind. Plaintiffs contend that this element is satisfied because the records were destroyed “knowingly” or “negligently” even if the Chief acted “without [the specific] intent to breach [his] *533 duty” to preserve. See
Beaven, 622 F.3d at 553. The district court rejected this argument, reasoning that Plaintiffs could not show that Serpas “acted in bad faith.” Johnson III, 2010 WL 3342211, at * 19.
The district court erred by injecting a bad faith component into its spoilation analysis. In Adkins, we recognized that there may be a “continuum of fault ranging from innocence through the degrees of negligence to intentionality.” 554 F.3d at 652–53. To the extent bad faith is relevant in a spoilation decision, its most appropriately taken into consideration when adjusting the sanction imposed. In the instant case, the record shows that Serpas deliberately chose not to preserve the results and deliberately ordered the destruction of the individual surveys. Although there is no evidence to show that he acted out of bad faith, his conduct was nevertheless intentional and therefore meets the “culpable state of mind” element.
Finally, Plaintiffs must prove that the destroyed surveys are “relevant” to their claims of reverse discrimination. This, however, is where Plaintiffs' request for sanctions must fail. The information Plaintiffs wishes was preserved is of minimal relevance to proving their case for reverse discrimination. First, the surveys themselves were a rather blunt instrument for measuring the supervisors' opinions. The individual results would have consisted only of a string of each supervisor's scoring, rated on a simple 1–to–3 scale, based on instructions which asked the supervisors to take into account a host of qualities demonstrating promotional readiness. The instructions did not tell the supervisors to consider race, gender, or diversity, and to the extent that such motives improperly influenced the scores, they likely would not be immediately apparent from the surveys' simplistic numerical system.
In order to link discriminatory intent to the surveys, Plaintiffs would have to examine each supervisor's rankings for patterns of race or gender discrimination. However, the record does not indicate that the program was enabled in such a way as to accomplish this. Although the supervisors used a log-in “name” and password to access the surveys, the record is unclear as to whether the supervisor's identity was saved alongside with the survey results. Moreover, even if that information was available, multiple inferences would be required to connect the individual surveys relevancy to Plaintiffs' claims. Plaintiffs would need to demonstrate not only an individual supervisor's pattern of discrimination, but also that the pattern of discrimination adversely affected the aggrieved candidate's averaged scores in comparison to his minority counterparts, and by consequence, Serpas' evaluation of the rollup scores. These serious hurdles attenuate the surveys from the ultimate promotion decision at multiple levels. As the proximate causation of the surveys weakens, so does their relevance. Moreover, as we mentioned above, Plaintiffs scored significantly below their promoted counterparts on the supervisor surveys. Accordingly, the likelihood that one or several supervisors' improper discrimination materially altered Plaintiffs' rollup scores is even less likely. The only other way Plaintiffs can conceivably prove the surveys' relevance is by claiming that the rollups were fabricated. Plaintiffs do not seriously press this argument.
*534 **10 Consequently, the district court did not abuse its discretion in denying sanctions.
B. Statutory Framework
Title VII makes it unlawful for an employer “(1) to fail or refuse to hire or to discharge any individual, or otherwise to discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual's race, color, religion, sex, or national origin; or (2) to limit, segregate, or classify his employees or applicants for employment in any way which would deprive or tend to deprive any individual of employment opportunities” on account of one of the same five grounds listed above. 42 U.S.C. § 2000e–2. There are two principal and distinct means of proving employer discrimination under Title VII: disparate treatment and disparate impact. Griggs v. Duke Power Co., 401 U.S. 424, 431, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971). We address Plaintiffs' claims of disparate treatment first.
C. Disparate Treatment Claims
1. Direct Evidence of Reverse Discrimination
Plaintiffs first contend that they offered direct evidence of discrimination proving that they were treated disparately from their minority counterparts. Specifically, Plaintiffs cite to (1) the 2003 Anderson memo; (2) Aaron's public remarks to The Tennessean;
and (3) the comments made by Commander Nash, Deputy Chief Pike, and Metro HR representative Sinor in their meetings with Plaintiffs.
Direct evidence “is that evidence which, if believed, requires the conclusion that unlawful discrimination was at least a motivating factor in the employer's actions.” Amini v. Oberlin Coll., 440 F.3d 350, 359 (6th Cir.2006); Grizzell v. City of Columbus Div. of Police, 461 F.3d 711, 719 (6th Cir.2006) (quoting Kocak v. Cmty. Health Partners of Ohio, Inc., 400 F.3d 466, 470 (6th Cir.2005)). Direct evidence must prove not only discriminatory animus, but also that the employer actually acted on that animus. See
Amini, 440 F.3d at 359. For the following reasons, none of Plaintiffs' proposed evidence meets this standard.
2The Anderson memo does not constitute direct evidence because it would require multiple inferences to reach a finding of discrimination. First, the memo is three years removed from the events of this case, and from it we would need to infer that Anderson's opinions in 2003 motivated his actions in 2005–2006. See, e.g.,
Phelps v. Yale Sec., Inc., 986 F.2d 1020, 1025–26 (6th Cir.1993) (rejecting statements made about the plaintiff nearly a year prior to his layoff). Additionally, Deputy Anderson was not the decisionmaker in the promotions process. Rather, Serpas was responsible for the disputed promotions, and Deputy Anderson only participated in the process by completing *535 supervisor surveys. See
Smith v. Leggett Wire Co., 220 F.3d 752, 759 (6th Cir.2000) (“Statements by nondecisionmakers cannot suffice to satisfy the plaintiff's burden of demonstrating animus.”) (quoting Price Waterhouse v. Hopkins, 490 U.S. 228, 277, 109 S.Ct. 1775, 104 L.Ed.2d 268 (1989) (O'Connor, J., concurring) (internal quotations omitted)).
**11 3Next, Plaintiffs point to several statements allegedly made by MNPD spokesperson Don Aaron to local newspapers in 2006 and 2007. Aaron was quoted as stating, “If you have two candidates that are essentially equal and believe that both would make very good supervisors, and if your choice is to make the department more diverse, you would probably elect to include diversity in your choice.” Aaron was also quoted as saying, “The goal of this police department is to mirror the population of the city to the greatest extent possible.... When presented with the opportunity to promote bright, qualified minority candidates, you always give those persons consideration.”
The district court disregarded these statements as inadmissible hearsay. Plaintiff contends that the district court's evidentiary ruling was incorrect because Aaron's comments should have been considered a statement by a party-opponent. See
Fed. R. Evid. 801(d)(2)(A) (a “party's own statement in either an individual or a representative capacity” is not hearsay if “offered [against] the party”).
We need not consider the district court's evidentiary ruling, because even if admissible, Aaron's statements are not direct evidence of discrimination. Aaron's first statement-answering a hypothetical question about “equally qualified candidates”—requires an inference because the statement did not refer to this employment decision.
Nor is Aaron's second comment direct evidence of discrimination, because it would require us to ignore the equally available inference that the department was simply aware of a lack of minorities within its upper ranks. As we have held, however, statements reflecting a desire to improve diversity do not equate to direct evidence of unlawful discrimination. See
Plumb v. Potter, 212 Fed.Appx. 472, 477–78 (6th Cir.2007) (“[A] jury could find that [the employer] believed it was good to have more women working at the [company], yet still conclude that [the decisionmaker] did not let that personal belief interfere with her decision whether or not to promote a woman over [the plaintiff].”)
4Finally, the statements allegedly made by Commander Nash, Deputy Chief Pike, and HR representative Sinor do not constitute evidence of direct discrimination because there is no evidence to show that these individuals were decisionmakers in the promotion process. See
Smith, 220 F.3d at 759–60. Plaintiffs' attempt to raise a cat's-paw argument about these individuals is also ineffective, because it would require us to draw inferences that are inappropriate under direct discrimination analysis. See
Ercegovich v. Goodyear Tire & Rubber Co., 154 F.3d 344, 355 (6th Cir.1998). Accordingly, Plaintiffs have failed to produce direct evidence of reverse discrimination.
a. Background Circumstances
Establishing the first prong of the prima facie
case in a reverse discrimination claim requires Plaintiffs to show that Defendants are that “unusual employer who discriminates against the majority.” Id. at 603–04. This requirement is not onerous, and can be met through a variety of means, such as statistical evidence; employment policies demonstrating a history of unlawful racial considerations; evidence that the person responsible for the employment decision was a minority; or general evidence of ongoing racial tension in the workplace. See
Treadwell v. Am. Airlines, 447 Fed.Appx. 676, 678 (6th Cir.2011) (citing cases). Indeed, “the mere fact that a racial minority took an adverse action” against the employee is often sufficient to satisfy the background circumstances requirement. Leavey v. City of Detroit, 467 Fed.Appx. 420, 425 (6th Cir.2012) (citing Arendale, 519 F.3d at 603).
5Plaintiffs allege that a background of reverse discrimination may be inferred from: (1) the 2003 Anderson memo; (2) the statements made by Commander Nash, Chief Pike, and HR representative Sinor in the meetings Plaintiffs requested after being passed over for promotion; (3) the fact that Serpas was “chief for a political entity” and motivated to “appease the population of Nashville”; and (4) two unrelated discrimination cases brought against Metro. The district court found this evidence insufficient to prove background circumstances, because there was no evidence to conclude that Defendants' actually unlawfully considered race or gender as factors in the making specific employment decisions. Johnson III, 2010 WL 3342211, at *16.
In so reasoning, the district court went too far. The background circumstances element only requires a plaintiff to “support the suspicion
that the defendant is that unusual employer who discriminates against the majority;” a plaintiff need not prove that the employer's actions actually
were illegally motivated. Boger v. Wayne Cnty., 950 F.2d 316, 324–25 (6th Cir.1991) (emphasis added). Requiring Plaintiffs to show actual discrimination at this stage conflates the background circumstances element with the remainder of McDonnell Douglas' burden-shifting approach. See
Treadwell, 447 Fed.Appx. at 679 (noting that requiring a plaintiff to produce both “the foreground and background evidence to reach a jury” incorrectly collapses the background circumstances prong into McDonnell Douglas' other elements).
*537 **13 As is required on summary judgment, we must draw disputed inferences in Plaintiffs' favor and conclude that Plaintiffs have met the background circumstances element. Plaintiffs supplied sufficient evidence to raise at least a material issue of fact that there was ongoing racial tension within the MNPD, and particularly, tension surrounding the manner in which promotions were being awarded within the department's ranks. See, e.g.,
id. at 678; Boger, 950 F.2d at 324–25; McCloud v. Testa, 97 F.3d 1536, 1549 (6th Cir.1996); Zambetti v. Cuyahoga Cmty. College, 314 F.3d 249, 256 (6th Cir.2002). Plaintiffs presented depositions by persons at all levels of MNPD's hierarchy who testified that the department was interested in promoting diversity and that it was acutely conscious of its long history of racial and gender imbalance.
And while the district court was correct that such a focus may only reflect a legitimate desire to comply with equal opportunity requirements, it also presents a context sufficient to meet the background circumstances element.
b. Qualifications and Similarly Situated Candidates
6As to the remaining prongs of Plaintiffs' prima facie
case, Plaintiffs suffered an adverse employment decision when they were passed over for promotion, but Defendants dispute whether Plaintiffs were qualified for promotion and similarly situated to the promoted candidates. These prongs touch on essentially the same issues, so we address them together.
Whether Plaintiffs were qualified for promotion and similarly situated to the promoted candidates depends partly upon how we compare the candidates. The district court disregarded the surveys in deciding the qualification issue, but considered them in its similarly situated analysis. We agree with this approach.
Under the new scoring procedure, candidates who received the highest scores were placed on the eligibility roster in alphabetical order without revealing the composite scores. We agree with the district court that this procedure meant that all candidates whose names appeared on the roster were deemed qualified for promotion. Johnson II, 2010 WL 3342211, at *3. Hence, we need not take into account the candidates' survey scores when considering the qualification element, and we conclude that Plaintiffs have met their burden to prove that they were qualified for promotion.
By contrast, a candidate's score is relevant in deciding whether Plaintiffs were similarly situated to the officers that were selected for promotion. Inclusion on the eligibility roster only qualified a candidate for promotion, but did not guarantee it. Because the chief had the discretion to choose among candidates, a candidate could be passed over for promotion entirely or promoted later than a comparatively lower-scored candidate.
Accordingly, the revised system established a two-stage promotions process. Scoring well on the written and assessment tests was not a guaranteed basis for *538 promotion; it only qualified a candidate for promotion. At the second stage, the chief held the sole discretion to choose among candidates. The system did not limit the considerations the chief could take into account, which could conceivably include a wide variety of qualities, including: education, training, length of service, readiness, ability to lead, disciplinary infractions, or a host of other factors. Any of these factors might distinguish Plaintiffs from their promoted peers and render Plaintiffs not similarly situated.
**14 We have explained that in failure-to-promote cases we only require a plaintiff to show that he possesses “similar qualifications” to the employee who received the promotion. Provenzano v. LCI Holdings, Inc., 663 F.3d 806, 814 (6th Cir.2011). During the prima facie
stage, our inquiry is not finely distinguished, given that it “is not realistic from a human standpoint” to expect the plaintiff to prove that he has identical qualifications to the chosen candidate. Id. Rather, we perform a “more searching evaluation of the relative qualifications of the two candidates” when examining pretext. Id. at 816. However, even if we granted leniency to Plaintiffs at the prima facie
stage, they still have made no effort whatsoever to show that they were similarly situated to the chosen candidates on even the most basic level. Plaintiffs utter failure to address the similarly situated element fatally undercuts their prima facie
case, as well as their argument for pretext.
Moreover and as noted above, the survey results are appropriately considered when evaluating the similarly situated element. In the instant case, the survey results render Plaintiffs dissimilar from the candidates who were selected for promotion. Serpas ordered the surveys in an effort to capture the supervisors' opinions as to which candidates were most suited for promotion. These opinions were considered, along with each candidate's personnel record and overall qualifications, when Serpas made his promotions decisions. And while Serpas did not take notes about his decisions, he testified that he did draw at least one clear line in his reasoning; he did not promote any candidate whose rollup score fell below a “2.0”
In fact, no candidate was promoted who fell below that line, and Plaintiffs concede that they scored well below this threshold. As such, they were not similarly situated to those candidates chosen for promotion. See
Gaffney v. Potter, 345 Fed.Appx. 991, 993 (6th Cir.2009) (holding that differences in performance ratings render two employees not similarly situated). Accordingly, Plaintiffs' prima facie
case fails because they have failed to show that they were similarly situated to the promoted candidates.
7Although Plaintiffs have not met their prima facie
burden, we note that similar deficiencies prevent them from demonstrating pretext.
Serpas explained that he did not select Plaintiffs for promotion because of their poor survey results. He testified that the scores demonstrated that those who were familiar with Plaintiffs' qualifications did not believe they would be as “successful” or “ready” for promotion in comparison to the other candidates. Serpas averred that he did not take race or gender into consideration when deciding whom to promote. Thus, Defendants have provided a permissible, non-discriminatory basis for the adverse employment decision, and Plaintiffs bear the burden of producing a triable issue of fact that the basis was a pretext for reverse discrimination.
**15 We have explained that a closer comparison of candidates is appropriate when the parties dispute whether an employee was *539 chosen for promotion based on qualifications versus discriminatory animus. Provenzano, 663 F.3d at 816. In conducting our evaluation, we do not act as a “ ‘super personnel department,’ overseeing and second guessing employers' business decisions,” Bender v. Hecht's Dep't Stores, 455 F.3d 612, 626 (6th Cir.2006), but rather simply compare characteristics such as “[d]ifferences in job title, responsibilities, experience, and work record,” in order to make an informed determination about whether an employment decision was based on pretext, Leadbetter, 385 F.3d at 691.
As noted above, however, Plaintiffs have provided the Court with none of the information required to draw an informed comparison. Rather, the only pertinent information we have before us shows that both minority and majority candidates with better survey scores were promoted ahead of Plaintiffs. Likewise, candidates with lower survey scores of both races and genders were not selected for promotion. These facts render fatal Plaintiffs' claims of pretext.
Finally, we note that the move to the new promotions procedure was apparently based on a variety of factors. Plaintiffs concede that the test-based promotions method was criticized for a number reasons—one of which was the disadvantage it imposed on minority candidates—but also for other reasons that were not based on race or gender. The MNPD, its officers, and the consulting company that worked on the updated procedure all agreed that the old method did not take into account other factors that critically affected the candidate's suitability for promotion—including length of service, educational background and training, and the opinions of the candidates' supervisors. The fact that the survey instructions mentioned these permissible factors, and not those impermissible factors of race or gender, also undermines Plaintiffs' claims for pretext.
D. Disparate Impact Claims
We turn next to Plaintiffs' claims of disparate impact brought under Title VII. “Disparate impact causes of action penalize employment practices that are ‘fair in form but discriminatory in operation.’ ” Phillips v. Cohen, 400 F.3d at 388, 397 (6th Cir.2005) (quoting Griggs, 401 U.S. at 431, 91 S.Ct. 849). “Thus, disparate impact analysis is intended to make sure that employers do not use ‘neutral’ decision-making mechanisms that in fact work to eliminate a greater portion of otherwise-qualified protected group members than they do members of other groups.” Id. Accordingly, a disparate impact claim does not require a showing of intent to discriminate. Id. Instead, the plaintiff must allege that a specific employment practice had an adverse effect on a group of employees, despite its facial neutrality. Id.
Disparate impact analysis requires the Plaintiff to identify—with specificity—the particular procedure having an adverse effect. Dunlap v. TVA, 519 F.3d 626, 629 (6th Cir.2008). A three-part burden-shifting test is then used to examine a disparate impact claim: (1) the plaintiff must establish a prima facie
case for discrimination; (2) the employer may then show that the challenged procedure is justified by “business necessity”; and (3) the plaintiff can rebut the employer's justification by showing that other tests or selection protocols would serve the employer's interest without creating the undesirable discriminatory effect. Albemarle Paper Co. v. Moody, 422 U.S. 405, 425, 95 S.Ct. 2362, 45 L.Ed.2d 280 (1975).
1. Procedural Posture
**16 Before turning to the merits of Plaintiffs' disparate impact claim, we must consider *540 its procedural posture and Plaintiffs' argument that the district court improperly denied them leave to amend in order to supplement their disparate impact arguments.
In the instant case, there are three components of the new promotions procedure that Plaintiffs could have alleged caused a disparate impact: (1) the change from the test-based promotions procedure to the eligibility roster procedure; (2) Serpas' unilateral discretion to select candidates for promotion off the roster based on unspecified criteria; and (3) the introduction of the supervisors' subjective opinions by way of the survey results. Of these, and as further explained below, Plaintiffs only alleged the first of the three had a disparate impact.
At the close of discovery, Plaintiffs moved to amend their complaint, in order to allege “all facts as known to the plaintiffs as of today with regard to these charges.” The additional paragraphs detailed information Plaintiffs learned during the discovery process, including additional details about how the survey process was implemented. However, the motion to amend did not seek to add any additional charges of disparate impact beyond Plaintiffs' original challenge to the eligibility roster procedure.
Accordingly, Plaintiffs never sought to add a claim that Serpas' discretion caused a disparate impact. Likewise, Plaintiffs never claimed that the supervisor surveys caused a disparate impact. Plaintiffs however, did allege that these aspects of the promotions procedure caused a disparate impact in their response to Defendants' motion to dismiss. Plaintiffs' arguments, however, went beyond the strict allegations raised in their proposed amended complaint—that the “sliding-band” procedure caused a disparate impact—and went on to say that “the Captain's surveys and the input from Serpas caused a disparate impact on the promotions favoring minorities over white males.”
The district court denied Plaintiffs' motion to amend, noting that the amendment deadline passed two months prior; that Plaintiffs had requested extensions on several other deadlines, but not on the amendment deadline; and that Plaintiffs' filing violated Federal Rule of Civil Procedure 16(b)(4) (requiring good cause to modify a scheduling order) and Local Rule 7.01(a) (requiring the filing of an accompanying memorandum of law where a motion “may require the resolution of an issue of law”). Finally, the court commented that Plaintiffs had already amended their claims multiple times and that the latest reason given—“to allege all facts as known to the plaintiffs as of today with regard to these charges”—was “insufficient to justify a fourth amendment.”
Shortly thereafter, the district court granted Defendants' motion to dismiss the disparate impact claim. The district court noted the pleading errors described above and commented that, while “plaintiffs [ ] somewhat convincingly argue[d]” a disparate impact claim related to the surveys, the argument could not be considered because Plaintiffs failed to raise it in their complaint. See
Johnson II, 2008 WL 3163531, at *6 (citing Kostrzewa v. City of Troy, 247 F.3d 633, 643 (6th Cir.2001)).
2. Denial of Leave to Amend
**17 Rule 15(a)(2) of the Federal Rules of Civil Procedure provides that a court may freely grant leave to amend a pleading when justice so requires, in order to make certain that a case is tried on its merits “rather than [on] the technicalities of the pleadings.” Moore v. City of Paducah, 790 F.2d 557, 559 (6th Cir.1986). “In deciding whether to grant a motion to amend, courts should consider undue delay in filing, lack of notice to the opposing *541 party, and futility of amendment.” Brumbalough v. Camelot Care Ctrs., Inc., 427 F.3d 996, 1001 (6th Cir.2005). Unless leave was denied on the basis of futility, this Court reviews the district court's decision for abuse of discretion. Ziegler v. Aukerman, 512 F.3d 777, 786 (6th Cir.2008).
Rule 15 provides that leave to amend “shall be freely given when justice so requires.” Fed.R.Civ.P. 15(a). In the absence of reasons such as those listed above, leave should generally be granted. Foman v. Davis, 371 U.S. 178, 83 S.Ct. 227, 9 L.Ed.2d 222 (1962). Nevertheless, a district court retains its discretion on such matters, and provided that the court announces a clear reason for its decision, this Court will generally defer to its reasoning. Id.
Rule 15 is augmented by Rule 16, which states that the generally wide latitude to amend may be restricted by the court's other scheduling orders. See
Fed.R.Civ.P. 16(b). Rule 16 requires that a party seeking to amend outside a scheduling order may do so only on a showing of “good cause.” This limitation is “designed to ensure that ‘at some point both the parties and the pleadings will be fixed.’ ” Leary, 349 F.3d at 906. In balancing Rules 15 and 16, this Court considers the movant's “diligence in attempting to meet the case management order's deadlines” and “whether the opposing party will suffer prejudice by virtue of the amendment.” Id.
8Plaintiffs contend that the district court abused its discretion when it refused to grant their motion to amend. They contend that the district court unfairly denied the motion solely on procedural technicalities and that, by doing so, it denied them the opportunity to allege facts that “could not have been discovered [ ] without having conducted written discovery.” Given that the district court perceived some merit in a disparate impact claim based on the surveys, Plaintiffs contend the court's refusal to allow amendment was in error.
We disagree. Plaintiffs have made no real attempt, either before the district court or before us now, to argue that “good cause” supported an amendment sought over two months after the deadline in the court's scheduling orders. Plaintiffs simply perfunctorily allege that they could not have discovered the facts required to support their claims before written discovery was conducted. However, Plaintiffs never state with any specificity when the evidence was received or why they could not have met or sought an extension on the district court's original deadline.
**18 We note that the disparate impact claims Plaintiffs now put forth are largely based on information that Plaintiffs appear to have known from the outset of this suit. For instance, as Plaintiffs' own affidavits indicate, they knew that Serpas had total discretion to select candidates for promotions off the eligibility roster, and they knew, or at least discovered shortly after the disputed promotions were announced, that an anonymous supervisor survey had been conducted. Accordingly, Plaintiffs had access to the basic facts required to allege their new disparate impact claims, even if they lacked certain specifics.
Moreover, even if we were to consider the new evidence Plaintiffs sought to add by way of an amended complaint, the fact remains that Plaintiffs failed to expand the legal bases for their disparate impact claims. A close reading of the documents reveals that the district court was correct. Plaintiffs only properly alleged a disparate impact claim based on the eligibility roster method. Although Plaintiffs' response to the motion to dismiss expanded their disparate impact claims, the district court was limited, as are we, to the facts and legal claims as raised in the pleadings. See
*542 Moore's Federal Practice § 12.34. (“The court may not ... take into account additional facts asserted in a memorandum opposing the motion to dismiss, because such memoranda do not constitute pleadings under Rule 7(a).”). Accordingly, the district court did not abuse its discretion in denying Plaintiffs leave to amend.
3. Merits Analysis
9Thus, we are left only to consider the disparate impact claim that was properly before the court—whether the move to the eligibility roster imposed a disparate impact on non-minority, male candidates. The district court rejected the challenge, noting that the complaint “alleged that MNPD ignored the objective criteria it gathered from the written examination and the assessment because of an intent to discriminate.”
Johnson II, 2008 WL 3163531, at *6 (emphasis added). The district court reasoned that this type of allegation was more akin to a disparate treatment claim than it was to a disparate impact claim. Id. We agree.
Finally, we briefly point out Ricci v. DeStefano, 557 U.S. 557, 129 S.Ct. 2658, 174 L.Ed.2d 490 (2009), a case issued by the Supreme Court about a year after the district court dismissed Plaintiffs' disparate impact claim. In Ricci, the Supreme Court found that the city of New Haven, Connecticut imposed a disparate impact on certain Caucasian firefighters when it refused to certify examination results used for making promotions when the results failed to produce any minority candidates. The fact that the City changed its procedures during the middle of the promotions process was a key factor in the Supreme Court's decision:
[Once a promotion process is established] and employers make clear their selection criteria, they may not then invalidate the test results, thus upsetting an employee's legitimate expectation not to be judged on the basis of race. Doing so, absent a strong basis in evidence of an impermissible disparate impact, amounts to the sort of racial preference that Congress disclaims, and is antithetical to the notice of a workplace where individuals are guaranteed equal opportunity regardless of race.
**19 Id. at 585, 129 S.Ct. 2658.
To be sure, guidance from Ricci might have helped Plaintiffs plead their allegations with requisite specificity, had that case been available to them. However, the present case remains easily distinguishable from Ricci, in that there is no evidence to show that Defendants disregarded any valid scores. And while Plaintiffs might have argued that their legitimate expectations were upset by the delayed introduction of the supervisor surveys, as explained above, they failed to properly allege this argument in their complaint.
Accordingly, we affirm the district court's decision dismissing Plaintiffs' disparate impact claim.
10The district court awarded Metro the costs of defending both Serpas and Metro against this lawsuit. On appeal, Plaintiffs raise two challenges to this decision. First, Plaintiffs argue that, regardless of the fact that Metro actually paid for Serpas' costs and legal fees, Serpas should have filed a separate bill of costs. Second, they argue that the district court did not carefully scrutinize whether the cost of a second set of transcripts for Serpas' attorneys was truly “necessary.”
This Court reviews a district court's award of attorneys costs for abuse of discretion. Singleton v. Smith, 241 F.3d 534, 538 (6th Cir.2001). Awards decisions are “fact-intensive” and a trial court's determination of those facts is entitled to substantial deference. See
Imwalle v. Reliance Med. Prods., 515 F.3d 531, 551 (6th Cir.2008). Applying this standard, neither of Plaintiffs' arguments have merit.
Federal Rule of Civil Procedure 54(d)(1) provides that “costs other than attorney's fees [ ] should be allowed to the prevailing party.” The presumption in favor of awarding costs may only be reversed upon a “heavy showing” that the court abused its discretion. Baji v. N.E. Reg'l Bd. of Dental Examiners, Inc., 3 Fed.Appx. 352, 360 (6th Cir.2001) (citing White & White, Inc. v. Am. Hosp. Supply Corp., 786 F.2d 728, 730 (6th Cir.1986)). In order to award costs to a prevailing party, the court must determine that:
the expenses are allowable cost items and that the amounts are reasonable and necessary. As long as statutory authority exists for a particular item to be taxed as a cost, [the court] shall not overturn a district court's determination that the cost is reasonable and necessary, absent a clear abuse of discretion. Thus, we shall review carefully whether an expense is recoverable, but once we determine that it is, we defer to the district court, which is in the best position to determine whether the cost is recoverable.
**20 Baker v. First Tenn. Bank Nat'l Assoc., 142 F.3d 431 (6th Cir.1998) (table) (citing Northbrook Excess & Surplus Ins. Co. v. Procter & Gamble Co., 924 F.2d 633, 642 (7th Cir.1991)).
Rule 54(d) provides that costs “should be allowed to the prevailing party,” but it does not specify how to file a bill of costs. The Middle District of Tennessee's Local Rule 54.01 is somewhat more specific and requires that “a cost bill, with supporting documentation, shall be filed by the prevailing party.
” (emphasis added). However, this rule does not provide that each prevailing party must file separately, and we have found no cases requiring that. Lastly, we note that Plaintiffs did not cite to Local Rule 54.01 before the district court. Accordingly, the district court did not abuse its discretion.
Plaintiffs next argue that the second set of transcripts was not reasonable or necessary, and they fault the district court for failing to carefully scrutinize whether Serpas' attorneys had an individual need for the transcripts. This claim is meritless. The depositions in question—those of Chief Serpas, Deputy Chiefs Pike and Anderson, and Eric Cardinal—were at the core of Plaintiffs' claims of reverse discrimination against Serpas and critical to his defense against those charges. Accordingly, *544 the deposition transcripts were necessary and the award of costs was appropriate.