PREGERSON, Circuit Judge, delivered the opinion of the Court as to Parts I, II, III, IV, V, VI, and VII, in which THOMAS, Circuit Judge, joined. Judge THOMAS delivered the opinion of the Court as to Parts VIII and IX. Judge PREGERSON joined as to Part VIII and dissents specially from Part IX.

RONALD M. GOULD, Circuit Judge joined as to Parts I, II, III, IVA, IVB, V, VI, VII, VIII and IX and filed a dissenting opinion as to Part IVC.

This appeal arises out of a jury verdict in favor of two women who sued their *1178 employer for sex discrimination. The women, Connie Hemmings (“Hemmings”) and Patty Lamphiear (“Lamphiear”), charged their employer, Tidyman’s Inc. (“Tidyman’s”), with discriminating against them on the basis of their sex in violation of federal and state anti-discrimination laws by failing to pay them wages and compensation equal to their male counterparts, failing to promote them, and retaliating against them after they complained of the discrimination.

Tidyman’s appeals the jury verdict and damages award in favor of the plaintiffs on four grounds. First, Tidyman’s contends that the district court erred at trial by admitting into evidence improper statistical expert testimony. Second, Tidyman’s argues that the district court abused its discretion by denying Tidyman’s’ motion for a new trial on the grounds that the evidence was insufficient, that misconduct by the plaintiffs’ counsel permeated the trial and prejudiced the jury, and that the size of the jury verdict was excessive. Third, Tidyman’s challenges the district court’s ruling that the plaintiffs could seek “double damages” under a Washington state law that provides for the doubling of any wages willfully and intentionally withheld from employees. See RCW §§ 49.52.050, 49 .52.070. Tidyman’s argues that the district court erred by not applying the Title VII cap on compensatory damages to the plaintiffs’ damage awards for future losses and for violations of Washington state law. Finally, Tidy-man’s argues that the Washington state law is intended to cover only accrued wages that are not paid, rather than wages not paid because the employer paid a lower wage as a result of discrimination.

Hemmings and Lamphiear cross-appeal on the issue of punitive damages. After the jury awarded the plaintiffs punitive damages, the district court granted Tidy-man’s’ renewed motion for judgment as a matter of law on the basis that the evidence did not support punitive damages. The plaintiffs contend that the district court erred as a matter of law by applying the wrong standard to determine whether the evidence supported the punitive damages awards. The plaintiffs ask that we reinstate the punitive damages awards in full, arguing that we should not apply the Title VII damages cap to these awards because it is unconstitutional. Finally, the plaintiffs argue that the district court erred by excluding costs for depositions and the preparation of certain affidavits from the attorney fees award.

We have jurisdiction under 28 U.S.C. § 1291 . We reverse the district court’s determination that the plaintiffs were not entitled to punitive damages, and we reverse the award of double damages under the Washington state statute. In all other respects, we affirm the district court. We conclude that Title VII’s cap on punitive damages is constitutional, direct the district court to reinstate the jury’s punitive damages award, and apply the Title VII cap to the punitive damages.

I.

Factual Background

A. Connie Hemmings.

In 1973, Connie Hemmings began working in the Billings, Montana office of Tidy-man’s, a chain of grocery stores in the Pacific Northwest. She started as an accounts payable clerk in the bookkeeping department, and was promoted to officer manager. In 1986, the Billings office closed, and Tidyman’s transferred its corporate headquarters to Spokane, Washington. Hemmings moved with her family to Spokane to work in the new office because of the opportunities it offered her for career advancement.

*1179 Hemmings was promoted to controller in 1987, replacing Mike Davis, who became the Chief Financial Officer (“CFO”) and Hemmings’ direct supervisor. Davis consistently gave Hemmings outstanding job performance evaluations. Hemmings decided to go back to college during this period to enhance her career potential. She obtained her bachelor degree in business management and graduated with honors.

Hemmings oversaw a number of employees as part of her job. Hemmings was concerned about the lack of women in management positions at Tidyman’s and what she perceived as roadblocks to their promotions. Hemmings initiated conversations about this topic with various executives and managers of Tidy-man’s, including Jack Heuston, the President. 1 Heuston laughed at the suggestion that more women should be in management and told Hemmings that women in management would “need to lift 50 pounds of potatoes.”

In May of 1996, Hemming’s supervisor, Davis, was promoted to Chief Operating Officer and the CFO position opened. Trial witnesses testified that Hemmings was well-qualified for the CFO position. She had experience overseeing financial statement reporting, supervising staff, working with internal audits, and working with banks and third-party administrators.

Hemmings was interviewed for the position along with another woman and one man. This was the first time Tidyman’s used an interview process to hire for an upper management position; previously, job openings were not posted and individuals were merely informed that they received the promotion. An all-male hiring committee interviewed Hemmings. The hiring committee concluded that Hem-mings demonstrated poor presentation skills during her interview, and hired the male candidate, Lee Clark. Davis told Hemmings that she was not hired because the board “did not want to work with an emotional woman.”

On July 1, 1996, Hemmings and the other plaintiff, Patty Lamphiear, served Tidyman’s with a demand letter outlining their claims of discrimination. Hemmings testified that the new CEO, John Maxwell, intimidated and harassed her in response to the demand letter, and attempted to discuss the possible lawsuit against Tidy-man’s in the absence of her attorney. When Hemmings suggested that she set up a meeting with her attorney present, Maxwell became angry and told Hem-mings that he would “sooner pay $5 million to fight the lawsuit than to pay [her] a penny.”

Hemmings and other Tidyman’s’ employees testified that Hemmings was denied admission to meetings and excluded from the chain of command pursuant to Maxwell’s instructions after the discrimination letter. She no longer had the power to hire and fire staff. Despite the company’s growth, Hemmings was not permitted to hire adequate staff for the expansion. Hemmings’ salary was frozen from 1996 until 1999. She received a raise in March of 1999, days before the trial began. A treating psychiatrist testified that Hemmings developed severe depression as a result of her work environment.

B. Patty Lamphiear.

Patty Lamphiear started working for Ti-dyman’s in 1984 as a part-time data entry clerk. She was promoted as the administrative assistant to Ken Ormsby, the manager of the Customer Prepaid Inventory (“CPI”) department. When Ormsby was *1180 moved to the position of bakery supervisor, Lamphiear assumed the CPI manager duties, but did not receive a salary increase for her new responsibilities. Although as CPI manager, Ormsby was paid $45,000 per year, Lamphiear’s salary remained $9.75 per hour (less than $25,000 per year). In addition to her CPI managerial duties, Lamphiear later assumed responsibility for purchasing all direct store delivery (“DSD”) products. Lamphiear ran the DSD department alone and without supervision, but was not given a title or supervisory powers, and her salary remained at less than $25,000 per year.

Tidyman’s created a new position, grocery supervisor, in 1993. Lamphiear was not given an opportunity to apply for the unposted position, despite her familiarity and experience with the requisite computer and merchandising skills. George Hau-serman was hired for the position, where he remained for one year, at an annual salary of $82,000. When Hauserman left, Lamphiear indicated her interest in the position to Jerry Streeter, the Chief Operating Officer. Streeter’s response was to laugh and tell Lamphiear “there is no way that you could get the position, because the men in the company would run right over you.” Another employee, Gregg Bab-bit, was promoted to the position and became Lamphiear’s supervisor, even though Lamphiear possessed relevant computer skills and experience that Babbit lacked.

In 1995, Lamphiear reported to Tidy-man’s’ management that Babbit had made inappropriate sexual remarks during a meeting. 2 Ron Bashaw, a general manager who resigned from the company in part to protest its unfair treatment of women, testified regarding Babbit’s attitude towards women employees as follows: “I don’t believe [Greg Babbit] thought that [women] were capable of working in management, especially with the pressure that comes with it.” Bashaw related that Bab-bit discussed women behind their backs to management, “[Babbit] put [women] down as far as sometimes the knowledge that they had.”

An internal Tidyman’s’ wage study in August of 1995 indicated that Lamphiear had been underpaid by the company for the past four years and that she should have been paid between $38,000 and $45,000 per year. On August 31, 1995, Lamphiear’s pay was raised from $28,600 to $35,000 — $3000 below the minimum recommended salary.

Lamphiear testified that after she and Hemmings served Tidyman’s with the demand letter in July 1996, she also encountered threats and intimidation from male managers at Tidyman’s. She was threatened by CEO Maxwell and insulted by CFO Davis. Lamphiear experienced the most problems with her new supervisor, Babbit, who verbally abused Lamphiear 3 and made veiled threats about eliminating her job.

Lamphiear’s emotional health deteriorated rapidly. She began having nightmares and trouble sleeping. In 1997, Lamphiear attempted to set up a meeting with her supervisors related to a work project. After confirming that they would attend, none of the men came to the meeting. Lamphiear felt undermined and increasingly insecure at work. Her problems with Babbit continued to escalate. Although Lamphiear was working long *1181 hours, approximately eighty hours a week, Babbit continued to assign her new projects. When Lamphiear tried to refuse the projects, Babbit reported her to senior management as uncooperative. Feeling persecuted at work, Lamphiear experienced severe depression and her therapist placed her on a suicide watch.

At the recommendation of her doctor, Lamphiear left her position on sick leave, due to job-related stress. While on sick leave, Lamphiear was informed by Tidy-man’s that her position had been eliminated. Lamphiear was offered only lower-paid positions, working again under the supervision of Babbit. Lamphiear declined to accept these positions. Although Tidyman’s eliminated Lamphiear’s former position, a new employee was hired for a very similar position, entitled “DSD Buyer,” which Babbit did not supervise.

II.

Procedural History

The plaintiffs brought suit against Tidy-man’s in federal court in February, 1997 for violations of the Civil Rights Act of 1964, Title VII, 42 U.S.C. § 2000 (e) et seq., and the Washington Law Against Discrimination, Revised Code of Washington (“RCW”) § 49.60 et seq. The plaintiffs alleged that Tidyman’s discriminated against them under both disparate impact and disparate treatment theories. 4 Before trial, both sides brought numerous motions in limine, including a motion by Tidyman’s to exclude the testimony of the plaintiffs’ statistical expert, Dr. Nyak Polissar (“Dr.Polissar”). The district court denied Tidyman’s’ motion and Dr. Polissar testified at trial.

At the close of the evidence, Tidyman’s filed a motion for judgment as a matter of law contending, inter alia, that the Washington statute authorizing double damages for the willful and intentional deprivation of wages was not intended to cover an employer’s failure to pay wages owed because of discrimination. The district court denied the defendant’s motion.

The jury returned large verdicts for both Hemmings and Lamphiear. With respect to Hemmings, the jury found by special verdict that Tidyman’s discriminated against Hemmings under a disparate treatment theory by retaliating against her on the basis of her gender. The jury also found that Tidyman’s discriminated against her under a disparate impact theory by failing to promote her and by failing to provide adequate pay and benefits on the basis of her gender. 5 The *1182 jury awarded Hemmings $120,000 in lost wages and benefits, $1,580,000 in future lost earnings and benefits, and $230,000 in non-economic damages, for a total of $1,930,000.

In Lamphiear’s case, 6 the jury found that Tidyman’s violated the law against disparate treatment discrimination by failing to promote her, retaliating against her, and paying her different compensation on the basis of her gender. The jury further found that Tidyman’s discriminated against Lamphiear under a disparate impact theory by failing to promote her and by failing to provide greater pay and benefits. The jury additionally found that Ti-dyman’s violated the Washington state law against the willful and intentional deprivation of past wages. The jury awarded Lamphiear $596,500 in past lost wages and benefits, $1,024,000 in future lost earnings and benefits, and $650,000 in non-economic damages, for a total award of $2,270,500.

The district court, over Tidyman’s’ objections, sent the issue of punitive damages to the jury. The jury awarded each plaintiff $1 million in punitive damages.

After the jury returned the punitive damages awards, Tidyman’s submitted a renewed motion for judgment as a matter of law on the grounds that the double damages are not available under Washington law in a discrimination case and that the plaintiffs failed to present a sufficient evidentiary basis for the punitive damages awards. At the same time, Tidyman’s also filed a motion to alter or amend judgment in lieu of a new trial and a motion for a new trial. The district court denied the renewed motion concerning the double damages claim, the motion for a new trial, and the motion to alter or amend the judgment. However, the district court granted Tidyman’s’ renewed motion with regard to the punitive damages awards. The district court allocated the jury awards for non-economic damages to the plaintiffs’ state law claims, rather than the Title VII claims. Having done this, the district court concluded that the Title VII *1183 damages cap on non-economic damages did not apply to non-economic damages awarded to plaintiffs for state law violations. Finally, the court granted the plaintiffs’ motion for attorney fees although it denied the plaintiffs’ request for costs for depositions and preparation of the fees motion.

III.

Admission of the Statistical Expert Testimony

Tidyman’s argues that the district court erred by admitting the testimony of the plaintiffs’ statistical expert, Dr. Polissar, because the prejudicial effect of the testimony outweighed its probative value. Tidyman’s also contends that the statistical evidence was insufficient to support the jury finding of disparate impact discrimination in either of the plaintiffs’ cases. We review a district court’s admissibility ruling on expert testimony for abuse of discretion. United States v. Cordoba, 194 F.3d 1053, 1056 (9th Cir.1999). We will not reverse the admissibility ruling unless it is “manifestly erroneous.” Id.

A. Dr. Polissar’s Testimony.

The centerpiece of the plaintiffs’ disparate impact case was the expert testimony of Dr. Polissar. Dr. Polissar testified that statistical analysis of Tidyman’s’ management revealed gender disparities in promotions and wages. In its pretrial motion to exclude Dr. Polissar’s testimony, Tidy-man’s argued that Dr. Polissar’s analysis included improper comparisons, which lacked probative value and would improperly prejudice the jury. The district court denied the motion, and ruled that any flaws in the analysis could be addressed through cross-examination or impeachment. The district court also stated if it determined that portions of Dr. Polissar’s analysis were improper, it would give a limiting instruction to the jury. 7

Tidyman’s provided Dr. Polissar with information about the names, gender, starting salary and position, and ending salary and position of the employees in Tidyman’s’ management. Dr. Polissar testified about the statistical analysis he performed using this data. 8 Dr. Polissar concluded that women in management at Tidyman’s earn an average of $12,000 less than men in management. The mean starting salary for women in management at Tidyman’s was $26,400, while the mean starting salary for men was $38,400.

Dr. Polissar also testified that he analyzed the progression of wages of women and men over time as a method of controlling for factors other than gender — such as experience — that might explain the initial wage differential. Dr. Polissar used regression analysis 9 to control for differences in experience, and reached the same conclusion — that gender predicted a statistically significant wage differential. Dr. Polissar also performed a “step analysis,” which compared the number of women to the number of men within each rank of Tidyman’s’ management hierarchy. Dr. Polissar explained that some of the step *1184 levels had no women employees, which made comparison analysis impossible for those levels. Where comparisons were possible, Dr. Polissar testified that the step analysis revealed wage differentials between men and women for most of the ranks. For example, at the “step four” level, which includes assistant store manager, office manager, system analyst, and controller, the average salary for female workers was $29,400, while the average salary for male workers was $40,800.

Finally, Dr. Polissar analyzed the distribution of men and women in different job categories at Tidyman’s and concluded that the distribution reflected a pattern of segregation of men and women that was unlikely to be due to chance. 10 He also testified that the percentage of women in hourly positions was higher than the percentage of women in salary positions.

On cross-examination, Tidyman’s attempted to show that Dr. Polissar’s analysis was fundamentally flawed because it assumed that each individual was equally qualified. Tidyman’s elicited testimony from Dr. Polissar that his analysis failed to take into account individual qualifications, preferences, motivations, and individual fields.

B. The Standard for Admission of Expert Testimony.

Tidyman’s argues that Dr. Polissar’s testimony lacked probative value and prejudiced the jury. These arguments relate most directly to Federal Rules of Evidence 403 and 702. Rule 403 permits the exclusion of relevant evidence “if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury.” Fed. R.Evid. 403. In applying Rule 403, “[d]is-trict courts enjoy ‘wide latitude’.” Fireman’s Fund Ins. Co. v. Alaskan Pride P’ship, 106 F.3d 1465, 1468 (9th Cir.1997).

Rule 702 governs the admissibility of expert testimony. Fed.R.Evid. 702. Under Rule 702, expert testimony is admissible if the testimony “will assist the trier of fact to understand the evidence or to determine a fact in issue.” Id. Whether testimony is helpful within the meaning of Rule 702 is in essence a relevancy inquiry. See Raskin v. Wyatt Co., 125 F.3d 55, 67 (2d Cir.1997) (“Expert testimony which does not relate to any issue in the case is not relevant, and ergo, non-helpful.”) (quoting Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579, 591 , 113 S.Ct. 2786 , 125 L.Ed.2d 469 (1993)). The trial court acts as “gatekeeper” and determines whether expert scientific testimony is sufficiently relevant and reliable to be admissible. Daubert, 509 U.S. at 589 , 113 S.Ct. 2786 .

C. Tidyman’s’ Challenges.

Tidyman’s attacks the probative value of the statistical evidence on two grounds: (1) that Dr. Polissar used an inappropriate comparison pool; and (2) that Dr. Polis-sar’s analysis did not account for variables such as individual skills and preferences. Both arguments lack merit.

1. The Appropriate Comparison Pool.

Courts have long recognized that statistical evidence may be used to establish a prima facie case of disparate impact discrimination. See, e.g., Hazelwood Sch. Dist. v. United States, 433 U.S. 299, 307-08 , 97 S.Ct. 2736 , 53 L.Ed.2d 768 (1977) (“Where gross statistical disparities can be shown, they alone may in a proper case constitute prima facie proof of a pattern or practice of discrimination.”). The statisti *1185 cal evidence, however, must be drawn from appropriate comparison pools. In the context of statistical analysis, comparison “pools” are the groups of individuals from which data is collected and compared. 11 For example, information gathered from the census would be data from a general population pool.

The plaintiffs’ expert in this case analyzed a data set comprised entirely of Ti-dyman’s’ management employees. Tidy-man’s contends that consideration of this pool was error because: (a) the data set did not included the “qualified” individuals for the at-issue jobs, and (b) the data set improperly included “store” management, instead of only “corporate” management. We reject both of these arguments.

a. Whether the Management Data Set Included the Qualified Individuals.

In support of its argument that the management data set failed to include the qualified individuals, Tidyman’s relies upon the Supreme Court’s opinion in Wards Cove Packing Co. v. Atonio, 490 U.S. 642, 650 , 109 S.Ct. 2115 , 104 L.Ed.2d 733 (1989), the seminal case addressing appropriate comparison pools. In Wards Cove, the plaintiffs alleged, inter alia, that the defendant discriminated against black potential employees by either failing to hire them, or hiring them in the lower-wage cannery positions rather than the higher-wage, non-cannery positions. The plaintiffs attempted to introduce evidence comparing the percentages of black and white employees in the cannery positions with the non-cannery positions. Id. at 650 , 109 S.Ct. 2115 . The Supreme Court found that this comparison analysis could not support a prima facie case of discrimination, because there was no evidence that the employees in the cannery jobs (pool 1) were qualified for the non-cannery jobs (pool 2). Id. at 651 , 109 S.Ct. 2115 . The Supreme Court found that the analysis between the two pools was therefore not a meaningful measure of discrimination, and that the plaintiffs should have instead introduced evidence comparing the percentage of qualified black persons in the labor market (the qualified labor pool) with the percentage of qualified black persons hired in non-cannery positions (the pool actually hired). Id. at 650 , 109 S.Ct. 2115 .

The general principle of Wards Cove— that the appropriate comparison pool for statistical analysis is the group from which individuals will be chosen for the job action — is appropriately applied in the plaintiffs’ analysis, which uses a data set of Tidyman’s’ management. In Wards Cove, the plaintiffs challenged the hiring practices of the defendant. The appropriate comparison pool in Wards Cove was thus the pool of potential applicants seeking to be hired- — i.e., qualified individuals from the general population.

In the present case, the plaintiffs challenge the compensation and promotion practices of the defendant. “[I]n cases involving claims of promotion and wage discrimination, the employer’s own workforce (or a portion thereof) may be the best source for data on the qualified labor market.” BARBARA LINDEMAN & PAUL GROSSMAN, EMPLOYMENT DISCRIMINATION LAW 1714 (3d Ed.1996). 12 Tidyman’s fills higher man *1186 agement by internal promotions. Therefore, the potential applicant pool — and thus the appropriate comparison pool — for promotions to upper and middle management jobs at Tidyman’s is comprised of the current employees in lower management positions. The analysis of Tidyman’s’ management data base included the qualified individuals under consideration for promotions. See Shidaker v. Tisch, 833 F.2d 627, 631 (7th Cir.1987) (“Where a company is shown to promote from within, the relevant labor pool of qualified applicants for upper level positions may be the group of employees in the company from which promotees will be drawn.”).

Use of a data set comprised of the employer’s entire management in a case challenging the failure to promote lower management to higher management and pay equal wages is consistent with the principle of Wards Cove, that the comparison pool for analysis should be the group from which individuals will be chosen for the job action, in this case promotion and payment of higher wages. In the instant case, Dr. Polissar testified to the existence of disparities between the percentages of female and male employees of Tidyman’s in terms of their movement from middle or lower management to upper management positions. We find that Dr. Polissar appropriately used the employer management data set for his statistical analysis.

b. Store Management versus Corporate Management.

Tidyman’s next attacks the plaintiffs’ statistical evidence on the ground that Dr. Polissar improperly included the management employees of the local stores within the management data pool- — rather than just the corporate management. Tidy-man’s contends that because the plaintiffs were not qualified for store management positions, the statistical analysis should not have included the store management employees. 13

Rather than argue that corporate positions and store management positions require entirely separate qualifications, Tidy-man’s asserts only that promotion to store management requires “special qualifications” that the two plaintiffs lacked. Tidy-man’s does not dispute that store management may be a career path to corporate management for some employees. 14 In other words, while there was contradictory testimony about whether Lamphiear and Hemmings were individually qualified for the store management positions, it was undisputed that store management positions could result in promotion to corporate management positions.

*1187 Tidyman’s’ argument misunderstands the purpose of including the store management employees. The data set had to include the store management employees to fully capture the potential pool of internal applicants for the upper level corporate management positions. 15 Inclusion of the store management positions was necessary to analyze and compare promotions between men and women. 16

In sum, the district court did not abuse its discretion by admitting the expert testimony based on the analysis of the data set of Tidyman’s’ management employees. 17

*1188 2. Individual Qualifications and Preferences.

Tidyman’s next argues that Dr. Polissar’s statistical analysis should have been excluded because it did not “eliminate all of the possible legitimate nondiseriminatory factors,” including the employee’s qualifications, level of education, and preferences. This argument also fails. We begin our analysis by noting that the law does not require the near-impossible standard of eliminating all possible nondiscriminatory factors. See Bazemore v. Friday, 478 U.S. 385, 400 , 106 S.Ct. 3000 , 92 L.Ed.2d 315 (1986) (“Importantly, it is clear that a regression analysis that includes less than ‘all measurable variables’ may serve to prove a plaintiffs case.”) (Brennan, J).

In Bazemore , the Supreme Court addressed the precise question presented by Tidyman’s’ appeal: if a study fails to account for all variables, how should a court treat the study? Justice Brennan, writing for the court, explained that “[njormally, failure to include variables will affect the analysis’ probativeness, not its admissibility.” 478 U.S. at 400 , 106 S.Ct. 3000 . In other words, in most cases, objections to the inadequacies of a study are more appropriately considered an objection going to the weight of the evidence rather than its admissibility. See, e.g., Wilmington v. J.I. Case Co., 793 F.2d 909, 920 (8th Cir.1986) (“Virtually all the inadequacies in the expert’s testimony urged here by [the defendant] were brought out forcefully at trial.... [T]hese matters go to the weight of the expert’s testimony rather than to its admissibility.”). 18 Vigorous cross-examination of a study’s inadequacies allows the jury to appropriately weigh the alleged defects and reduces the possibility of prejudice. Fireman’s Fund, 106 F.3d at 1468 ; United States v. L.E. Cooke Co., 991 F.2d 336, 342 (6th Cir.1993). In this case, Tidy-man’s cross-examined Dr. Polissar extensively about the fact that the study did not account for all possible variables, such as educational differences.

In some cases, however, the analysis may be “so incomplete as to be inadmissa-ble as irrelevant.” Bazemore, 478 U.S. at 400 n. 10, 106 S.Ct. 3000 . Tidyman’s contends that failure to include the employees’ individual qualifications, preferences, and education in the analysis rendered the analysis inadmissible under this standard. We disagree.

First, Tidyman’s did not prove at trial that any of these factors were important to the subjective and undefined promotion process or compensation awards. We have recognized that a defendant may not rest an attack on an “unsubstantiated assertion of error.” Gen. Tel. Co., 885 F.2d at 580. Rather, the defendant must “produce credible evidence that curing the alleged flaws would also cure the statistical disparity.” Id. at 583. Accord Sobel v. Yeshiva Univ., 839 F.2d 18 , 34 (2d Cir.1988) (“[A] defendant challenging the validity of a multiple regression analysis [must] make a showing that the factors it contends ought to have been included would weaken the showing of a salary disparity made by the analysis.”).

In this case, the plaintiffs’ expert “used the best available data, which [came] from the [defendant] itself.” Adams v. Ameritech Serv. Inc., 231 F.3d 414, 424 (7th Cir.2000). If the defendant believed information about the employees’ educational background, for example, would have explained the differences in promotions and *1189 compensation between male and female upper level employees, Tidyman’s should have provided information about educational level to the plaintiffs, or at a minimum, introduced testimony that education was a central factor in promotions.

We cannot say that the exclusion of preferences, individual qualifications, and education rendered the data set so incomplete “as to be irrelevant.” Bazemore, 478 U.S. at 400 , 106 S.Ct. 3000 . Dr. Polissar’s regression analysis accounted for experience and seniority within the company. By tracking the employees over time, Dr. Polissar was able to control outside factors that may have contributed to women originally being placed in lower management positions. See Gen. Tel. Co., 885 F.2d at 582 (concluding that the plaintiffs statistical analysis was probative of discrimination despite its failure to include employment interests); Adams, 231 F.3d at 427-28 (concluding that the defendant’s objections that the analysis did not control for outside factors went to the weight of the evidence, not its admissibility); Bruno v. W.B. Saunders Co., 882 F.2d 760, 766 (3d Cir.1989) (affirming the district court’s admission of statistical studies, despite the failure of the studies to take “account of the minimum qualifications of the jobs into which promotion or transfer occurred”). 19

In sum, the testimony of Dr. Polissar had probative value. The analysis could have helped the jury determine contested facts and evaluate whether the promotion and compensation practices of Tidyman’s had a disparate impact or reflected disparate treatment against women at the management level. Any inadequacies in the methodology were presented to the jury by the cross-examination of Dr. Polissar. We conclude that the district court did not abuse its discretion by failing to exclude the plaintiffs’ expert testimony.

IV.

Motion for a New Trial

Tidyman’s also appeals the district court’s denial of its motion for a new trial under Federal Rule of Civil Procedure 59 on the grounds that the great weight of the evidence is against the jury’s findings, the verdict was excessive, and that counsel misconduct deprived the defendant of a fair trial. A district court may grant a motion for a new trial based on the insufficiency of the evidence only if the verdict “is against the ‘great weight’ of the evidence or ‘it is quite clear that the jury has reached a seriously erroneous result.’ ” Ace v. Aetna Life Ins. Co., 139 F.3d 1241, 1248 (9th Cir.1998).

We review a district court’s denial of a motion for a new trial for an abuse of discretion. De Saracho v. Custom Food Mach., Inc., 206 F.3d 874 , 880 (9th Cir.2000). We will reverse the denial of a motion for new trial based on the insufficiency of the evidence only if the district court made a legal error in applying the *1190 standard for a new trial or if the record contains no evidence in support of the verdict. See Landes Constr. Co. v. Royal Bank of Canada, 833 F.2d 1365, 1372 (9th Cir.1987).

A. The Weight of the Evidence.

The district court found that the jury verdict was not “against the clear weight of the evidence,” and denied Tidyman’s’ motion for a new trial. Tidyman’s contends that the jury’s findings of: (1) disparate impact in both plaintiffs’ cases; (2) retaliation in both plaintiffs’ cases; (3) discriminatory intent against Lamphiear; and (4) the equal pay violation in Lamphiear’s case were against the great weight of the evidence. We conclude, however, that each of the jury’s findings is supported by sufficient evidence in the record and affirm the district court’s denial of the motion.

To establish a prima facie case of disparate impact under Title VII, the plaintiffs must: (1) show a significant disparate impact on a protected class or group; (2) identify the specific employment practices or selection criteria at issue; and (3) show a causal relationship between the challenged practices or criteria and the disparate impact. Atonio v. Wards Cove Packing Co., Inc., 810 F.2d 1477, 1482 (9th Cir.1987) (en banc). The plaintiffs sought to demonstrate that the subjective processes for awarding promotions, benefits, and salaries had a disparate impact on the two women. Dr. Polis-sar’s statistical conclusions about the large disparities in salaries and the lack of women in upper management, as well as the anecdotal testimony by multiple witnesses about the subjective promotion processes and award of salary and benefits, is sufficient to support the jury’s conclusion that the plaintiffs established a case of disparate impact discrimination.

The plaintiffs also had to show three elements to establish a prima facie case of disparate treatment discrimination: (1) membership in a protected class; (2) qualification for the position at issue; and (3) an adverse employment action. See Morgan, at 1016. The jury only found disparate treatment of Lamphiear. The jury heard the testimony of a former store manager that Lamphiear’s gender affected the company’s treatment of her, and that she received significantly lower pay than her male counterparts. Tidyman’s’ own internal study found that Lamphiear was underpaid for her position. Multiple witnesses, including the plaintiffs, testified to gender-based discriminatory comments made by Tidyman’s’ male managers to, and about, the plaintiffs. This combined testimony is sufficient to support the jury finding of intentional discrimination against Lamphiear.

To meet their prima facie burden of establishing retaliation, Hemmings and Lamphiear needed to demonstrate: (1) they engaged in a protected activity; (2) they suffered an adverse employment action; and (3) the existence of a causal link between the activity and adverse action. See Morgan v. Nat'l R.R. Passenger Corp., 232 F.3d 1008, 1017 (9th Cir.2000). Testimony by Hemmings that she was denied promotion opportunities and pay raises, excluded from meetings, and otherwise shunned at work following her complaint letter concerning sex discrimination constitutes substantial evidence to support a jury conclusion that Tidyman’s retaliated against her. Testimony by Lamphiear that she was berated, harassed, and given an additional workload following her complaint letter supports a jury finding of retaliation against Lamphiear.

Tidyman’s contends that the great weight of the evidence is against finding an equal pay violation in Lamphiear’s case because the plaintiffs failed to introduce sufficient evidence that Lamphiear’s work *1191 was similar to Ormsby’s, the supervisor whom Lamphiear replaced. Tidyman’s argues that Lamphiear failed to establish that her position shared a “common core” of tasks with the position when held by Ormsby. Cf. Stanley v. Univ. of Califor nia, 178 F.3d 1069, 1074 (9th Cir.1999) (discussing the “common core” of tasks test under the more stringent Federal Equal Pay Act). Contrary to Tidyman’s’ contention, Lamphiear introduced testimony that she assumed the central responsibilities of the job, and was later given additional responsibilities. The jury could have easily credited this testimony and found that the positions shared a common core. Moreover, under the Washington State Equal Pay Act, a plaintiff need only prove that the employer paid different wages to men and women who performed similar work. Adams v. Univ. of Washington, 106 Wash.2d 312 , 722 P.2d 74, 76-78 (1986) (en banc).

In sum, we find that each of the contested jury findings was based on significant evidence in the record. Because evidence in the record supports the verdict, we conclude that the district court did not abuse its discretion by denying Tidyman’s’ motion for a new trial on the ground that the great weight of the evidence did not support the jury’s findings.

B. The Size of the Damages Award.

Tidyman’s next contends that the size of the damages awards meant that the jury must have been motivated by sympathy or sheer guesswork, and that the district court abused its discretion by not granting a new trial on this basis. We will not reverse a district court’s denial of a motion for a new trial unless the damages are “grossly excessive or monstrous.” Los Angeles Memorial Coliseum Com’n v. NFL, 791 F.2d 1356, 1360 (9th Cir.1986).

The plaintiffs introduced testimony of a certified public accountant, Anson Avery (“Avery”), to support their request for damages. Avery testified that approximately $4.4 million in damages were due to the plaintiffs collectively for past lost wages and future lost wages. The jury returned a verdict of $3.3 million for past lost wages and future lost wages. The jury returned a slightly higher verdict for Lamphiear than the amount Avery recommended as due, and returned a significantly lower verdict for Hemmings. 20

The argument by Tidyman’s that the jury verdict is grossly excessive is principally based on quibbles with Avery’s calculations about the front and back pay awards. Avery testified that he calculated the “lost wages” amounts by comparing the salary and compensation packages for male executives in similar positions to the salary and compensation packages of the plaintiffs. In Hemmings’ case, Avery included in his calculation of the “lost wages” the difference between Hemmings’ salary had Tidyman’s promoted her to the position of CFO and her salary without the promotion. For both women, Avery factored inflation and interest rates in calculating the recommended award. The defendant cross-examined Avery about various assumptions in his calculations— *1192 such as whether paid vacations and annuities should be included — and potential mathematical errors.

On appeal, Tidyman’s contends that the alleged problems with Avery’s calculations render the verdict “monstrous” because there is no evidence in the record to support the damage award. This argument clearly fails. As described above, the record supported conclusions of liability. Avery’s lengthy and detailed testimony provided a basis for the jury to translate the liability into dollar amounts. Moreover, the jury’s award was below the amount calculated by Avery. The fact that the jury may have agreed with Avery and rejected the defendant’s contentions, for example, that compensation such as vacation time and annuities should not be included, does not render the verdict “grossly excessive or monstrous.” The district court did not abuse its discretion by rejecting Tidyman’s’ motion for a new trial on this ground.

C. Counsel Misconduct.

Finally, Tidyman’s contends that the district court erred by denying the motion for a new trial because of alleged misconduct by the plaintiffs’ counsel. Recognizing that the district court is “in a superior position to gauge the prejudicial impact of counsel’s conduct during the trial,” we will not overrule a district court’s ruling about the impact of counsel’s alleged misconduct unless we have “a definite and firm conviction that the court committed a clear error of judgment.” Anheuser-Busch Inc. v. Natural Beverage Distribs., 69 F.3d 337, 346 (9th Cir.1995) (internal quotations and citation omitted).

[1] Generally, misconduct by trial counsel results in a new trial if the “flavor of misconduct sufficiently permeate[s] an entire proceeding to provide conviction that the jury was influenced by passion and prejudice in reaching its verdict.” Kehr v. Smith Barney, 736 F.2d 1283, 1286 (9th Cir.1994).

Tidyman’s complains of several comments by plaintiffs’ counsel during closing arguments. 21 The most significant comment was the following statement by the plaintiffs’ attorney during the closing argument:

(Emphasis supplied).

Tidyman’s failed to object to this statement or any other statement made by counsel during closing argument. Nor did Tidyman’s move for a mistrial on the basis of counsel’s misconduct. The first time Tidyman’s complained of the misconduct was in its motion for a new trial. The district court noted that it would have sustained an objection to the comment during closing argument: “I remember the comment and I thought I sua sponte corrected it, but maybe I just thought about it and would have sustained an objection had *1193 one been made because it was totally improper.” Although the district court acknowledged the impropriety of the remark, it denied the motion for a new trial and ruled that a new trial was not necessary to avoid a miscarriage of justice.

The federal courts erect a “high threshold” to claims of improper closing arguments in civil cases raised for the first time after trial. Kaiser Steel Corp. v. Frank Coluccio Constr. Co., 785 F.2d 656, 658 (9th Cir.1986). The rationale for this high threshold is two-fold. First, raising an objection after the closing argument and before the jury begins deliberations “permit[s] the judge to examine the alleged prejudice and to admonish ... counsel or issue a curative instruction, if warranted.” Id. As noted above, the trial judge is in a superior position to evaluate the likely effect of the alleged misconduct and to fashion an appropriate remedy. The second rationale stems from courts’ concern that allowing a party to wait to raise the error until after the negative verdict encourages that party to sit silent in the face of claimed error.

“We will review for plain or fundamental error, absent a contemporaneous objection ..., where the integrity or fundamental fairness of the proceedings in the trial court is called into serious question.” Bird v. Glacier Electric Coop. Inc., 255 F.3d 1136, 1148 (9th Cir.2001).

Plain error review requires: (1) an error, (2) the error is plain or obvious, (3) the error was prejudicial or effects substantial rights, and (4) review is necessary to prevent a miscarriage of justice. See Smith v. Kmart Corp., 177 F.3d 19, 25 (1st Cir.1999) (describing the plain error standard in a civil counsel misconduct case). “Plain error is a rare species in civil litigation, encompassing only those errors that reach the pinnacle of fault envisioned by the standard set forth above.” Id. at 26 (internal citations and quotations omitted).

We readily conclude that the plaintiffs’ counsel’s statement during closing argument was error. Counsel inappropriately referred to other cases that she herself had litigated. The fact that she personally had litigated the cases had no relevance to the lawsuit. Nor should she have suggested that this litigation “proved the ease.” We also find that the error was “plain.” Plaintiffs remarks were obviously improper and blatant enough for the trial judge to recall them easily.

We next must consider whether counsel’s error was prejudicial and fundamentally unfair. “[T]he burden of making a ‘concrete showing of prejudice’ resulting from improper closing argument falls upon appellant.” Moses v. Union Pac. R.R., 64 F.3d 413, 418 (8th Cir.1995). In evaluating the likelihood of prejudice from the comments, we should consider “the totality of circumstances, including the nature of the comments, their frequency, their possible relevancy to the real issues before the jury, the manner in which the parties and the court treated the comments, the strength of the case, and the verdict itself.” Puerto Rico Aqueduct & Sewer Auth. v. Constructora Lluch, Inc., 169 F.3d 68, 82 (1st Cir.1999); see also Cooper v. Firestone Tire & Rubber Co., 945 F.2d 1103, 1107 (9th Cir.1991) (declining to find reversible error where “the alleged misconduct occurred only in the argument phase of the trial ... the remarks were isolated rather than persistent, ... most of counsel’s comments were not objected to at trial and appellants did not move for a mistrial at the end of the argument”).

We note, however, that this remedy is available only in “extraordinary cases.” Bird, 255 F.3d at 1148 . For instance, in Bird , the court concluded that counsel’s closing arguments offended fundamental *1194 fairness where counsel: (1) argued in inflammatory terms; (2) linked the defendant’s behavior to white racism in exploitation of Indians; (3) appealed to historical racial prejudices of or against the white race; and (4) used incendiary racial and nationalistic terms to encourage the all-tribal member jury to award against the non-Indian defendant. Id. at 1152 .

The misconduct in this case is substantially different from the misconduct in Bird . Here, the misconduct was an isolated, short comment during a closing statement that covered 66 pages when transcribed. Tidyman’s urges us to weigh other instances where the plaintiffs referred to prior lawsuits against Tidyman’s as evidence that the plaintiffs deliberately violated the district court’s ruling excluding such references. Tidyman’s first points to its testimony by Hemmings and contends that, in conjunction with her testimony, the “district court sustained defense counsel’s objections to testimony that Tidyman’s had been sued before.” A review of the record does not support this contention.

Hemmings testified about conversations that she had with the director of loss prevention, “the discussions that I had with him were that we [Tidyman’s] had been sued before and we lost and we need to get women in management positions.” The defense objected to this statement on hearsay grounds, and the trial court sustained the objection. Hemmings then testified as follows, without objection by defense counsel:

Hemmings went on to describe the failure of the defendant to establish programs that would have supported the movement of women into management programs, despite corporate recognition of the need for such programs. Tidyman’s did not object to any portion of this testimony.

The jury also heard from the plaintiffs’ expert, Dr. Polissar, that Tidyman’s had been sued before by plaintiffs’ counsel:

The defense counsel did not move to strike the portion of Dr. Polissar’s testimony referring to the prior lawsuit against Tidyman’s. The jury thus learned, without objection by the defense counsel, that the plaintiffs’ counsel had sued Tidyman’s before, that Tidyman’s had lost gender discrimination cases in the past, and that the plaintiffs’ expert, Dr. Polissar, worked with plaintiffs’ counsel in another case against Tidyman’s.

*1195 The actions by the court and the parties support an inference that the misconduct was not prejudicial or fundamentally unfair. Although the district court noted that the reference was error, he did not sua sponte issue a correction, instead preferring to wait to issue a limiting instruction until counsel objected. The fact that counsel did not object before the jury was instructed strongly suggests that counsel made a strategic decision to gamble on the verdict and suspected that the comments would not sway the jury.

The strength of the plaintiffs’ case is another factor that weighs against a finding of fundamental unfairness. The plaintiffs’ expert testimony went largely unrefuted. The plaintiffs introduced compelling testimony about a discriminatory culture and the disparate impact of the discrimination on women’s pay and promotion options. In the absence of counsel’s improper statements, we cannot say that we think a different verdict was likely. We conclude that the district court did not abuse its discretion by denying Tidyman’s’ motion for a new trial because of counsel misconduct during the closing statement.

V.

Tidyman’s’ Motion to Cap the Loss Award to $300,000

Title VII, as amended by the Civil Rights Act of 1991, limits compensatory and punitive damages based on the size of the defendant corporation. For a plaintiff suing a company with more than 500 employees, damages are capped at $300,000. 42 U.S.C. § 1981a(b)(3) (1998). The damages limit does not apply to back pay awards, or to relief authorized by 42 U.S.C. § 2000e-5(g). See 42 U.S.C. § 1981a(b)(2). Tidyman’s argues that the district court erred by not applying the § 1981a damages cap to: (1) the plaintiffs’ front pay awards; and (2) to the Washington state law discrimination claims. Both arguments lack merit.

Denial of a motion to amend the judgment is reviewed for abuse of discretion. Kingvision Pay-Per-View Ltd. v. Lake Alice Bar, 168 F.3d 347, 350 (9th Cir.1999). “While the district court generally has discretion regarding how to allocate the damage award, to the extent that the allocation rests on an interpretation of the statute, we review it de novo.” Passantino v. Johnson & Johnson Consumer, 212 F.3d 493 , 509 (9th Cir.2000).

A. Front Pay.

Tidyman’s’ argument that the statutory cap under § 1981a(b)(3) should apply to front pay awards is foreclosed by the Supreme Court’s recent decision in Pollard v. E.I. du Pont de Nemours & Co., 532 U.S. 843 , 121 S.Ct. 1946 , 150 L.Ed.2d 62 (2001). In Pollard , the Supreme Court considered whether the term “compensatory damages” in § 1981a included front pay awards. Following every circuit to address the question except the Sixth, the Supreme Court ruled that front pay is excluded from “compensatory damages” and,thus is not subject to the statutory cap. Id. at 1949.

As the Supreme Court explained, the statutory cap was adopted in 1991 as part of the Civil Rights Act. Id.; Civil Rights Act of 1991, 105 Stat. 1071 , § 2. The Civil Rights Act provisions expanded the remedies available to plaintiffs for intentional discrimination by providing that the plaintiff could recover “compensatory and punitive damages ... in addition to ” relief authorized under § 706(g) of the Civil Rights Act of 1964. 42 U.S.C. § 1981a(a)(1). The new remedies — -not authorized by § 706(g) — in turn are subject to the statutory cap. Pollard, 121 S.Ct. at 1950 . The Supreme Court concluded that front pay awards were available under *1196 § 706(g), and therefore excluded under the term “compensatory damages,” within the meaning of the 1991 Civil Rights Act. Id at 1952. Following Pollard , we reject Ti-dyman’s’ claim that the Title VII statutory cap applied to the plaintiffs’ front pay awards.

B. Application of the Statutory Cap to Washington State Law Claims.

Tidyman’s’ second argument is that the district court abused its discretion by failing to apply the Title VII damages cap to the state damages awards. Tidy-man’s contends that even if the awards were controlled entirely by state law, the federal damages cap applies because Washington courts look to federal law when interpreting their anti-discrimination statutes.

Tidyman’s does not challenge the district court’s allocation of the non-economic damages to the state law claims. The special verdict form in this case did not differentiate plaintiffs’ state and federal law claims. Any portion of the non-economic damages allocated to the federal Title VII claim would be subject to the $300,000 cap. In this case, Hemmings was awarded $230,000 in non-economic damages, an amount less than the federal cap. Tidyman’s’ argument thus potentially applies only to the $650,000 non-economic damages the jury awarded to Lamphiear, which the district court allocated to Lam-phiear’s state law claims.

Tidyman’s argues that we should apply the federal damages cap to the state law discrimination claim because Washington state courts, “in the absence of state authority,” will consider federal law “persuasive” when construing sections of the Washington anti-discrimination law that “parallel” the federal law. Xieng v. Peoples Nat’l Bank, 120 Wash.2d 512 , 844 P.2d 389, 399 (1993) (en banc); Goodman v. Boeing, 75 Wash.App. 60, 877 P.2d 703, 713 (1994). In Passantino, however, we rejected the defendant’s argument that the Title VII cap should apply to the Washington Law Against Discrimination. 212 F.3d at 509-10. We noted that Title VII explicitly prohibits limiting state law remedies and that Title VII was not intended to force plaintiffs to choose among remedial statutes. Id. at 510. We do not find any evidence in the form of case law, statutory language, or legislative history indicating an intent by the Washington legislature to cap damage awards at $300,000. Accordingly, we conclude, consistent with our ruling in Passantino, that the Title VII damages cap does not apply to the Washington state law discrimination claims.

VI.

Punitive Damages

The plaintiffs cross-appeal the district court’s finding that punitive damages are not available as a matter of law. The plaintiffs contend that the district court applied the wrong standard in granting Tidyman’s’ renewed motion that, as a matter of law, the evidence was insufficient to support an award of punitive damages. Specifically, the plaintiffs allege that the district court failed to consider adequately the directives of the Supreme Court’s decision in Kolstad v. Am. Dental Ass’n, 527 U.S. 526 , 119 S.Ct. 2118 , 144 L.Ed.2d 494 (1999). The plaintiffs argue that under Kolstad , the district court should have denied the defendant’s renewed motion for judgment as a matter of law.

“Judgment as a matter of law is appropriate when the evidence, construed in the light most favorable to the non-moving party, permits only one reasonable conclusion, which is contrary to the jury’s verdict.” Gilbrook v. City of Westminster, 177 F.3d 839, 864 (9th Cir.), cert. denied, 528 U.S. 1061 , 120 S.Ct. 614 , 145 L.Ed.2d 509 (1999) (quoting Omega Envtl., Inc. v. *1197 Gilbarco, Inc., 127 F.3d 1157 , 1161 (9th Cir.1997)). We review de novo legal conclusions about the availability of punitive damages. See EEOC v. Wal-Mart Stores, Inc., 156 F.3d 989 (9th Cir.1998).

The district court instructed the jury that the plaintiffs were entitled to punitive damages if they demonstrated that Tidyman’s’ conduct was willful and egregious, or displayed a reckless indifference towards the plaintiffs’ federal civil rights. The jury awarded each plaintiff $1 million in punitive damages. After the verdict, however, Tidyman’s made a renewed motion that the punitive damages were not supported by the evidence as a matter of law. The district court granted the defendant’s motion. The district court found that the plaintiffs “did not present a legally sufficient evidentiary basis for a reasonable jury to find by a preponderance of the evidence that Defendant’s conduct toward either Plaintiff was willful and egregious or displayed a reckless indifference to the Plaintiffs federal rights sufficient to justify an award of punitive damages.” (Emphasis supplied).

In the interim between the jury’s deliberations of the punitive damages award and the district court ruling on the motion for a judgment as a matter of law, the Supreme Court decided Kolstad . In Kolstad , the Supreme Court resolved a circuit split over the appropriate standard for determining the availability of punitive damages under Title VII by establishing a three-part inquiry to address when the evidence supports a punitive damages verdict. 527 U.S. at 530 , 119 S.Ct. 2118 .

In the first step, the Supreme Court clarified the requisite mental state of employers, the analysis primarily at issue in this case. Under Title VII, the jury may award punitive damages if the moving party demonstrates that “the respondent engaged in a discriminatory practice or discriminatory practices with malice or with reckless indifference to the federally protected rights of an aggrieved individual.” 42 U.S.C. § 1981a(b)(l). Interpreting this section, the Supreme Court concluded that Congress intended to impose a heightened standard of liability for the award of punitive damages, but rejected the argument that the heightened standard requires that an employer’s behavior be “egregious.” Kolstad, 527 U.S. at 534-35 , 119 S.Ct. 2118 . Instead, the Court concluded that Congress intended for punitive damages to apply in intentional discrimination cases where the plaintiff can show that the employer knowingly or recklessly acted in violation of federal law. Id. at 535 , 119 S.Ct. 2118 .

The Court found that the questions of malice and reckless indifference are subjective questions concerning the employer’s motive or intent, rather than an objective inquiry into whether the employer’s behavior is “egregious.” Id. at 535-38 , 119 S.Ct. 2118 . The defendant is appropriately subject to punitive damages if it acts “in the face of a perceived risk that its actions will violate federal law.” Id. at 536 , 119 S.Ct. 2118 . The Supreme Court explained that although egregious conduct could be evidence of an intentional violation of the law, it was not a necessary element. Id. at 535 , 119 S.Ct. 2118 .

The Supreme Court also held that the plaintiff must “impute liability for punitive damages to respondent.” Id. at 539 , 119 S.Ct. 2118 . Under this step, the plaintiff must show that the intentional discrimination by an employee is attributable to the employer by using traditional agency principles, e.g., that a managerial employee acted within the scope of his or her employment. Id. at 540-41 , 119 S.Ct. 2118 .

Finally, the Supreme Court clarified that the defendant employer may raise as an affirmative defense its good faith efforts to comply with Title VII, if such efforts were contrary to the actions of its *1198 managerial agents. Id. at 545-46 , 119 S.Ct. 2118 . As we noted in Passantino, the Supreme Court did not eliminate the rule that a sufficiently senior agent may be treated as the corporation’s proxy. Passantino, 212 F.3d at 516. The affirmative defense is thus unavailable to the employer for the actions of agents sufficiently senior to be considered proxies. Id. at 516-17.

Although the district court was aware of Kolstad, 22 it did not apply the Kolstad framework to the question of the availability of punitive damages in this case. Importantly, the district court ap- plied the old, pre-Kolstad standard for demonstrating the relevant mental state of the defendant by requiring that the plain- tiff establish sufficient evidence that the defendant’s conduct toward either plaintiff “was willful and egregious or displayed a reckless indifference.” (Emphasis sup- plied). As discussed above, the plaintiffs did not need to demonstrate “egregious” conduct. We conclude that the district court erred by applying the incorrect stan- dard for determining the availability of punitive

Both parties urge us to review the rec- ord to determine if the evidence was legal- ly sufficient to support an award of puni- tive damages by a reasonable jury. 23 We conclude that it was and reinstate the pu- nitive damages

Under Koistad, we must begtn by asking whether Tidyman's acted with malice or reckless indifference-whether Tidyman's "acted in the face of a perceived risk that its actions will violate federal law." Kolstad, 527 U.S. at 536 , 119 S.Ct. 2118 . In 218 this case, the district court found that the jury verdicts of intentional discrimination in the form of retaliation against both plaintiffs and failure to promote Lam- phiear were supported by sufficient evidence.

[11] We noted in Passantino that after Kolstad , “in general, intentional discrimi- nation is enough to establish punitive dam- ages liability.” 212 F.3d at 515. We then cited the examples noted by the Supreme Court of situations in which intentional discrimination would not give rise to puni- tive damages liability. Id. For example, if the plaintiff relied upon a novel theory of discrimination, the evidence might not sup- port a conclusion that the defendant acted with the awareness that it might violate the law. Id. Alternatively, in rare cases, the defendant might actually be unaware of Title VII’s anti-discrimination mandates. Id.

In Passantino, we found that the mental state element was satisfied because the jury could have concluded that the defen- dant acted with reckless disregard for plaintiffs rights based on evidence that the defendant’s employees lied to the plaintiff to conceal their discriminatory actions. Id. at 516. Other circuits have found puni- tive damage awards appropriate under the Kolstad framework where the plaintiff demonstrated that the defendants were aware of anti-discrimination principles. See, e.g., Bruso v. United Airlines Inc., 239 F.3d 848, 858 (7th Cir.2001) (holding that a plaintiff may satisfy the mental state “element by demonstrating that the relevant individuals knew of or were *1199 familiar with the anti-discrimination laws and the employer’s policies for implementing those laws”); Romano v. U-Haul Int’l, 233 F.3d 655, 669 (1st Cir.2000) (finding liability because the manager was aware of the company’s anti-discrimination policies); Lowery v. Circuit City Stores, Inc., 206 F.3d 431, 443 (4th Cir.2000) (finding that the employer discriminated in the face of a perceived risk of violating federal law where the manager was aware of the existence of federal anti-discrimination laws); Equal Employment Opportunity Comm’n v. Wal-Mart Stores, Inc., 187 F.3d 1241 , 1246 n. 3 (10th Cir.1999) (finding liability for punitive damages where the defendant “was aware of the prohibitions against workplace discrimination and ... had a company wide policy designed to ensure compliance”).

The record contains substantial evidence from which a reasonable juror could conclude that Tidyman’s was aware of anti-discrimination laws and acted in the face of this awareness. The plaintiffs introduced testimony that Tidyman’s hired additional women in store management positions, in part because of its concern about legal exposure for discriminatory practices. The plaintiffs also introduced testimony that the defendants excluded plaintiffs from decision making processes and harassed the plaintiffs after they sent their discrimination complaint. The jury could have easily concluded from this testimony that Tidyman’s intentionally discriminated against the plaintiffs in face of the risk that it was violating the law.

Based on the record, there is no doubt that discriminatory actions by Tidyman’s’ management officers can be imputed to Tidyman’s. The plaintiffs’ claims of intentional discrimination were based on the conduct of Jerry Streeter, the Chief Operating Officer; John Maxwell, the Chief Executive Officer; Mike Davis, the Chief Financial Officer; and Gregg Babbit, the grocery supervisor, among others. Tidy-man’s does not argue that the actions of management were contrary to the good faith non-discriminatory policies of the company, probably because on appeal Ti-dyman’s continues to assert that the actions themselves were non-discriminatory. 24

Based on our review of the record, we cannot say that the only reasonable conclusion is contrary to the jury’s verdict awarding punitive damages. Testimony and trial evidence support a determination by a reasonable jurist that Tidyman’s acted in the face of a perceived risk that its actions would violate federal law. We therefore reverse the district court’s ruling that, as a matter of law, the evidence was insufficient to support the jury’s awards of punitive damages.

VII.

Attorneys Costs

The district court awarded $62,638.16 in attorneys fees and costs, adopting in whole the plaintiffs’ cost affidavit with the exception of deposition costs and the costs for assembling the affidavit in support of attorneys fees. The plaintiffs contend that the district court abused its discretion by excluding the costs for depositions, $18,060.65, and the costs for preparing the *1200 attorneys fees motion, $2,775.21, from the fee award. Attorneys fee awards are reviewed for abuse of discretion. Fischer v. SJB-P.D., Inc., 214 F.3d 1115, 1118 (9th Cir.2000).

Both deposition costs and costs for preparation of attorneys fees motions are recoverable under Title VII in certain circumstances. See Harris v. Marhoefer, 24 F.3d 16 , 19 (9th Cir.1994) (holding that plaintiffs may recover out-of-pocket expenses and expenses related to discovery and expert witnesses under § 1988). “Normally [the attorneys fees award] will encompass all hours reasonably expended on the litigation,” however, the “district court should ... exclude from this initial fee calculation hours that were not ‘reasonably expended.’ ” Hensley v. Eckerhart, 461 U.S. 424, 433-34 , 103 S.Ct. 1933 , 76 L.Ed.2d 40 (1983).

In this case, the district court could have determined within its discretion that the hours preparing some of the witnesses or taking the depositions were overinflated or unnecessary. See Van Germen v. Guarantee Mutual Life Co., 214 F.3d 1041 , 1045 (9th Cir.2000) (holding that district courts should not include hours that are “excessive, redundant, or otherwise unnecessary”) (quoting Hensley v. Eckerhart, 461 U.S. 424, 434 , 103 S.Ct. 1933 , 76 L.Ed.2d 40 (1983)). We therefore find that the district court did not abuse its discretion by denying costs for depositions or preparation of the motion for attorneys fees.

THOMAS, Circuit Judge, in which Judges PREGERSON and RONALD M. GOULD joined:

VIII.

The Constitutionality of the Title VII Cap on Punitive Damages

The statutory limitation on damages contained in 42 U.S.C. § 1981a(b)(3), which imposes a $300,000 cap on compensatory and punitive damages in Title VII suits against employers with more than 500 employees, does not violate the doctrine of separation of powers or the Seventh Amendment.

The § 1981a damages cap does not represent an impermissible intrusion by the legislature into the province of the judiciary. Whether Congress may reopen a final judgment of an Article III court is not in question; the law is clear that Congress may not do so. Plaut v. Spendthrift Farm, Inc., 514 U.S. 211, 240 , 115 S.Ct. 1447 , 131 L.Ed.2d 328 (1995). However, Congress has significant power to define and circumscribe self-created causes of action. Indeed, almost two decades ago, the Supreme Court articulated a vital distinction between common law causes of action and actionable rights created by Congress. Specifically, the Court noted that

Northern Pipeline Constr. Co. v. Marathon Pipe Line Co., 458 U.S. 50, 83 , 102 S.Ct. 2858 , 73 L.Ed.2d 598 (1982). Title VII epitomizes such a eongressionally created right; as such, its remedies — including type and quantity of damages — may be legislatively determined without violating separation of powers. Cf. Lauf v. E.G. Skinner & Co., 303 U.S. 323, 330 , 58 S.Ct. 578 , 82 L.Ed. 872 (1938) (upholding an act of Congress removing the jurisdiction of the district court to issue injunctive relief in a labor dispute).

*1201 “Title VII of the Civil Rights Act of 1964 is not a statute to which would apply the traditional presumption in favor of all available remedies. That statute did not create a general right to sue for employment discrimination, but instead specified a set of circumscribed remedies.” Landgraf v. USI Film Prods., 511 U.S. 244 , 286 n. 38, 114 S.Ct. 1483 , 128 L.Ed.2d 229 (1994). The § 1981a damages cap perfectly illustrates the extent of Congressional power to delineate contours of such a Con-gressionally-created cause of action. Prior to the enactment of the Civil Rights Act of 1991, Title VII forbade imposition of punitive, or of compensatory, damages; the original Act provided only for equitable remedies.

In 1991, Congress determined that victims of employment discrimination were entitled to additional remedies. Pollard v. E.I. du Pont de Nemours & Co., 532 U.S. 843 , 121 S.Ct. 1946, 1951 , 150 L.Ed.2d 62 (2001). But, as legislative history makes clear, the 1991 Act would not have been passed by Congress but for the inclusion of a punitive damages cap:

S. Rep. 102-286 at *2 (May 21,'1992). As the Sixth Circuit has observed, “[t]he fact that the judicial branch is limited in the amount of damages which it may award does not mean that its ability to decide cases is being impaired by Congress.” Pollard v. E.I. DuPont de Nemours Co., 213 F.3d 933, 946 (6th Cir.2000), rev’d on other grounds by Pollard, 532 U.S. 843 , 121 S.Ct. 1946 , 150 L.Ed.2d 62 (2001).

Plaintiffs further argue that, even if the damages cap does not violate separation of powers, it unconstitutionally infringes on the province of the jury as defined in the Seventh Amendment. Essentially, Plaintiffs claim that, once Congress in 1991 provided for a jury trial in Title VII cases, it could not then circumscribe the ability of that jury to make factual findings, including a finding of the proper amount of punitive damages. Again, we disagree.

The Supreme Court has not specifically addressed this question, although it has noted that courts of appeals have upheld such caps against Seventh Amendment challenges. Gasperini v. Ctr. for Humanities, 518 U.S. 415 , 429 n. 9, 116 S.Ct. 2211 , 135 L.Ed.2d 659 (1996). However, Justice Stevens noted in his concurrence to Gas-perini that

Id. at 2227 (Stevens, J., concurring).

The Supreme Court’s recent ruling in Pollard impliedly affirms Justice Stevens’ conclusion. Pollard did not disturb the Sixth Circuit’s holding that the Title VII damages cap is constitutional: The Court held only that front-pay awards are not an element of compensatory damages, and thus are not subject to the cap. Id. at 1949. In fact, in defining the ambit of the Title VII damages cap, the Court seemingly sanctioned the Sixth Circuit’s conclusion. See id. at 1951 (“However, compensatory and punitive damages awarded under *1202 § 1981a may not exceed the statutory limitations set forth in § 1981a(b)(3) .... ”); see also Feltner v. Columbia Pictures Television, Inc., 523 U.S. 340, 355 , 118 S.Ct. 1279 , 140 L.Ed.2d 438 (1998) (holding that the Seventh Amendment includes a right to a jury determination of statutory damages under § 504(c) of the Copyright Act, but not addressing the fact that the Act limits damages to amounts between $500 and $20,000).

The Seventh Amendment does not provide unlimited protection to jury determinations. For example, when a punitive damage jury verdict “can fairly be categorized as ‘grossly excessive’ in relation to [legitimate state] interests,” the Due Process Clause requires judicial intervention. BMW of North Am. v. Gore, 517 U.S. 559, 568 , 116 S.Ct. 1589 , 134 L.Ed.2d 809 (1996). Here, Congress has merely definitively explained what constitutes excessive damages. Congress may “prescribe” a “rule of decision” in such a context. See Robertson v. Seattle Audubon Soc., 503 U.S. 429, 440 , 112 S.Ct. 1407 , 118 L.Ed.2d 73 (1992) (noting that in the context of statutorily created rights, Congress may amend a statute in such a way as to have the effect of prescribing rules of decision in specific cases). Other Circuits to have considered this issue agree. See Madison v. IBP, Inc., 257 F.3d 780, 804 (8th Cir.2001). As the Eighth Circuit explained it:

We further note the paradoxical implications of Plaintiffs’ claim: If a judge cannot limit damages found by a jury in accordance with a statute, how can a judge impose statutorily mandated double or treble damages without also imposing on the jury’s province as sole factfinder? And yet “[ajwards of double or treble damages authorized by statute date back to the 13th century ... and the doctrine was expressly recognized in cases as early as 1763.” Browning-Ferris Ind. of Vermont, Inc. v. Kelco Disposal Inc., 492 U.S. 257, 274 , 109 S.Ct. 2909 , 106 L.Ed.2d 219 (1989).

In sum, what Congress can create, Congress can define. Title VII’s damages cap does not interfere with the proper exercise of authority by the judicial branch, nor does it offend the Seventh Amendment. Because Title VII’s damages limitation passes constitutional muster, we affirm the district court’s judgment applying it.

THOMAS, Circuit Judge, in which Judge RONALD M. GOULD joined:

IX.

Double Damages Under Washington State Law

The Plaintiffs are not eligible to receive double damages under RCW § 49.52.070, which provides that:

Plaintiffs argued that an award of double damages pursuant to RCW § 49.52.070 was justified pursuant to RCW § 49.52.050(2), which provides in relevant part that:

In short, the Plaintiffs argue that because the jury found that the Defendant willfully and intentionally violated federal and state anti-discrimination statutes, RCW § 49.52.050(2) was violated as a matter of law, entitling Plaintiffs double damages under RCW § 49.52 .070.

In interpreting state law, federal courts are bound by the pronouncements of the state’s highest court. Davis v. Metro Prods., Inc., 885 F.2d 515 , 524 (9th Cir.1989). If the particular issue has not been decided, federal courts must predict how the state’s highest court would resolve it. Dimidowich v. Bell & Howell, 803 F.2d 1473, 1482 (9th Cir.1986), modified at 810 F.2d 1517 (9th Cir.1987). In making that prediction, federal courts look to existing state law without predicting potential changes in that law. Moore v. R.G. Inds., Inc., 789 F.2d 1326, 1327 (9th Cir.1986).

The Washington legislature enacted the statute at issue with a group of related provisions in 1939. Ellerman v. Centerpoint Prepress, Inc., 143 Wash.2d 514 , 22 P.3d 795, 798 (2001) (en banc). “Sometimes referred to as the ‘Anti Kickback’ statutes, they were enacted to prevent abuses by employers in a labor-management setting, e.g., coercing rebates from employees in order to circumvent collective bargaining agreements.” Id.

Washington courts have not extended RCW § 49.52.050 to situations where employers violate anti-discrimination statutes. Rather, violations of § 49.52.050 have been upheld where an employer consciously withholds a quantifiable and undisputed amount of accrued pay. See, e.g., Ellerman, 22 P.3d at 798 (failure to pay wages); Schilling v. Radio Holdings, Inc., 136 Wash.2d 152 , 961 P.2d 371, 377 (1998) (failure to issue regular paychecks)’.

The language of the statute does not support the expansive interpretation urged by the Plaintiffs. In ascertaining legislative intent, “the language at issue must be evaluated in the context of the entire statute.” Ellerman, 22 P.3d at 798 . The key word in the statute is “obligated.” If the Washington legislature intended for the provision to apply to a situation such as Plaintiffs’, it could have stated that any employer who violates any statute is subject to double damages. The insertion of the word “obligated” indicates a pre-exist-ing duty imposed by contract or statute to pay specific compensation. Thus, a willful and intentional withholding of accrued pay legally owed the employee would subject the employer to double damages. Here, the Defendant’s “obligation” to pay Plaintiffs the specific amount at issue had not legally accrued prior to the jury verdict. It did not stem from a “statute, ordinance, or contract;” rather, it resulted from a retrospective jury verdict.

Further, the Washington Supreme Court has not extended the reach of RCW § 49.52.050 to instances, such as the one at bar, in which there is a bona fide dispute as to whether the employer is obligated to *1204 pay the amounts in question. Schilling, 961 P.2d at 375 . As the Washington Supreme Court put it in Chelan County Deputy Sheriffs’ Ass’n v. Chelan County, 109 Wash.2d 282 , 745 P.2d 1 (1987):

Id. at 11.

Thus, the. Plaintiffs were not entitled to double damages pursuant to RCW § 49.52.070.

CONCLUSION

We affirm the district court’s admission of the plaintiffs’ expert testimony and the district court’s denial of Tidyman’s’ motion for a new trial. We also affirm the district court’s ruling that the Title VII cap on compensatory and punitive damages does not apply to front pay awards or the Washington state law claims.

We reverse the district court’s judgment as a matter of law that double damages were available to Lamphiear under the Washington statute, RCW 49.52.070. We also reverse the district court’s judgment as a matter of law that punitive damages were not available in this case. Finally, we uphold the constitutionality of the Title VII cap on compensatory and punitive damages.

Accordingly, we AFFIRM in part, REVERSE in part, and REMAND to the district court to reduce the back pay damages to Lamphiear, reinstate the punitive damage awards to Lamphiear and Ham-mings, and apply the Title VII damages cap to the punitive damages awards.

Each side to bear its own costs on appeal.

. Hemmings also discussed with Jim Armstrong, the director of risk management, the need to hire more women in management to avoid liability in future lawsuits.

. Jack Rippee, a Tidyman’s' store director who was present at the meeting, testified for die defense that his memory of the analogy was that Lamphiear "wouldn't recognize a good deal if it was three naked guys for a dollar."

. Lamphiear's therapist testified that Lam-phiear was upset during visits because of Babbit’s increasing use of inappropriate sexual innuendos, including references to naked men and women and cutting off a penis.

. " ‘Disparate treatment’... is the most easily understood type of discrimination: The employer simply treats some people less favorably than others because of their race, color, religion, sex, or national origin.... Claims of 'disparate impact’ ... involve employment practices that are facially neutral in their treatment of different groups but that in fact fall more harshly on one group than another and cannot be justified by business necessity.” Int'l Bhd. of Teamsters v. United States, 431 U.S. 324 , 335 n. 15, 97 S.Ct. 1843 , 52 L.Ed.2d 396 (1977).

. The following questions and answers are excerpted from the special jury verdict form for Connie Hemmings:

. The following questions and answers are excerpted from the special jury verdict form for Patty Lamphiear:

IF YOUR ANSWER TO QUESTION NO. 3 IS "YES”, YOU MUST DOUBLE THE PORTION OF ANY AWARD FOR WAGES WILLFULLY AND INTENTIONALLY DEPRIVED PATTY LAMPHIEAR BY DEFENDANT WHEN CONSIDERING QUESTION NO. 4.

Special Jury Verdict form for Patty Lam-phiear, April 13, 1999.

. Tidyman’s did not renew its objection to Dr. Polissar’s testimony at trial and no limiting instruction was given.

. At trial, Dr. Polissar testified that he had a masters and a doctoral degree in statistics from Princeton University, and that he had been a faculty member in the Department of Bio-statistics at the University of Washington for fifteen years before becoming a full-time consultant. Tidyman's did not object at any point on the ground that Dr. Polissar was unqualified, or that the statistical analysis he presented was outside of his area of expertise.

."A regression analysis is a common statistical tool ... designed to isolate the influence of one particular factor — e.g. sex — on a dependent variable — e.g. salary.” Equal Employment Opportunity Comm’n v. Gen. Tel. Co. of Northwest, Inc., 885 F.2d 575 , 577 n. 3 (9th Cir.1989).

. Polissar testified to the following percentages of women and men employees by department: grocery, 100% men; meat, 83% men; produce, 94% men; floral, 100% women; pharmacy, 73% men; bakery, mixed; expres-so, 93% women; store support, mixed; store management, 86.2% men; bookkeepers, 100% women.

. See generally Kevin Gilmartin, Identifying Similarly Situated Employees In Employment Discrimination Cases, 31 JURIMETRICS J. 429 (1991).

. See also Julie Lee & Caitlin Lui, Measuring Discrimination in the Workplace: Strategies for Lawyers and Policymakers, 6 U. CHI. L. SCH. ROUNDTABLE 195, 198 (1999) ("If promotions or terminations within an organization are being examined, an internal benchmark would likely be used: the affected' pool of employees who seem to have differential rates of promotion or termination might be compared against benchmark pool of employees who are not ‘affected.’ ”); Stender v. Lucky Stores, 803 F.Supp. 259, 295-98 *1186 (N.D.Cal.1992) (upholding the statistical analysis based entirely on employer workforce data where the plaintiffs challenged the placement and promotion practices of the employer).

. Tidyman's' attack involves disputed facts concerning the store management — namely, whether the two plaintiffs were qualified for store management positions and whether store experience was a prerequisite to store management positions. For example, Lam-phiear testified that she was qualified to be a store manager. Ron Bashaw, a former Tidy-man's' store manager, also testified that Lam-phiear was qualified for a store management position. The plaintiffs introduced testimony that no objective criteria were used in awarding store promotions. Tidyman's presented testimony that the individuals actually promoted to store manager and assistant manager positions all had store experience, which Lamphiear and Hemmings lacked.

. There is evidence that Tidyman’s itself treated the corporate and store management employees as a group. For example, Tidy-man's treated corporate staff and store management in the same group for the purpose of its salary policy. Tidyman’s' internal corporate management flow charts include store director and manager positions.

. Even assuming that the two individual plaintiffs were unqualified for store management positions, the inclusion of the store management employees in the management data set was proper given the fluid movement of employees between the store and corporate management. In Aiken v. City of Memphis, 37 F.3d 1155 (6th Cir.1994), the Sixth Circuit considered a similar argument in the context of alleged discrimination in promotion in the municipal fire and police departments. Id. at 1163 . The Court rejected the argument that the inclusion of all patrol officers and fire privates in the pool was inappropriate because not all of the patrol officers and fire privates were qualified for promotion. Id.

. Moore v. Hughes Helicopters, 708 F.2d 475, 482 (9th Cir.1983), a case relied on by Tidy-man’s, is inapposite. In Moore, this court concluded that the plaintiffs failed to establish a prima facie case of discrimination by pointing to the low percentage of women in high job grades alone. Id. at 484. The court found that the statistical analysis lacked probative value because the plaintiff women did not have the skills for the higher job grades, and the higher job grades were heavily filled by applicants from outside of the company. Id.

The instant case is easily distinguishable: here, no one disputes that Hemmings and Lamphiear were qualified for upper level management jobs. Rather, Tidyman's challenges whether the plaintiffs were qualified for each of the management positions, including positions below their positions as controller and senior buyer. Tidyman's’ organizational diagrams indicate that Hem-mings' position as controller was higher in Tidyman's' hierarchy than the store management positions. Lamphiear's position as senior buyer was below the highest level of store management, on an even level with some of the store management positions, and above others.

Moore is distinguishable on other grounds as well. The statistical analysis used by Hem-mings and Lamphiear involved interpretative statistical tools such as regression analysis rather than the straight percentage used in Moore. Additionally, in Moore the plaintiffs attempted to use statistics from an employer data set when the higher positions were filled from outside rather than internal applicants. In this case, Tidyman’s promoted from within, and the use of the employer management data set is thus appropriate.

.An additional line of case law supports rejecting Tidyman's' argument that the store management employees should have been excluded from the data set. Federal courts have rejected defendant's arguments that statistical analysis should be limited to a pool of "qualified applicants" where the job qualification standards are themselves discriminatory. Failure to post vacancies and the use of subjective promotion practices — both practices of Tidyman's prior to the plaintiffs' lawsuit in 1996 — may be evidence of discrimination. Paxton v. Union Nat’l Bank, 688 F.2d 552 , 563-64 (8th Cir.1992). The Eleventh Circuit described the relevance of a lack of formal promotion procedures for statistical analysis:

Forehand v. Florida State Hosp., 89 F.3d 1562, 1574 (11th Cir.1996).

In the instant case, the plaintiffs introduced testimony that one of the informal requirements to a store management position was prior experience in a night crew position. A former store manager of Tidyman’s testified that women were directed away from such positions.

. "Litigation generally is not fussy about evidence; much eyewitness and other nonquan-titative evidence is subject to significant possibility of error, yet no effort is made to exclude it if it doesn’t satisfy [a particular] statistic test.” Kadas v. MCI Systemhouse Corp., 255 F.3d 359, 362 (7th Cir.2001).

. Tidyman's cites Ottaviani v. State Univ. of New York, 875 F.2d 365 (2d Cir.1989) for the proposition that a statistical study must include all possible nondiscriminatory factors. However, Ottaviani involved a specific challenge to the plaintiff's expert's decision to exclude available data about academic rank from the statistical analysis. Id. 374-75. The district court in Ottaviani ruled that academic rank was not in itself a discriminatory measure, and thus appropriate for inclusion in the analysis. Based on this ruling, the district court considered only the plaintiffs analysis that included rank. Id. at 375. Unlike in Ottaviani, the plaintiffs’ expert in this case did not have available data of the criteria which Tidyman's complains are excluded, such as data about the employees’ educational background. Ottaviani, which involved the inclusion of available data about one factor, cannot be read to stand for the broad proposition that a statistical study must include all possible criteria.

. In Lamphiear’s case, Aveiy recommended an award of $1,405,865, and the jury returned an award of $1,620,500. The special verdict form instructed the jury to double the back pay award if it found that Tidyman’s willfully deprived the plaintiff of wages. The jury concluded that Tidyman’s did willfully deprive Lamphiear of wages, and presumably doubled the amount to reach a back pay award of $616,000. Without the doubling, the total jury award for front and back pay damages in Lamphiear's case would have been $1,311,500, an amount smaller than the amount Avery recommended. In Hemmings' case, the jury awarded $1,700,000 in front pay and back pay damages, although Avery recommended $3,004,497.

. Tidyman's also complains that the plaintiffs' attorney referred to the defendant’s case as a "jack rabbit defense,” that a verbal assault on Lamphiear by one of the defendant’s employees was a rape of her mind, that a female board member and a paralegal sitting at the counsel table were mere "tokens.” While in poor taste, these comments do not rise to the level of misconduct. Tidyman's itself made similar remarks in its closing arguments. For example, it referred to the plaintiffs case as "a whole big bag of rabbits the plaintiffs brought in, [because] their case has no merit and they want to distract you.”

. Plaintiffs' counsel addressed the Kolstad opinion at oral argument on the defendant’s motion for judgment as a matter of law.

. We agree with other circuits that the rul- ing in Kolstad was not so novel as to require a remand to provide the defendant an opportu- nity to supplement the record. See Rubinstein v. Adm'r of the Tulane Educ. Fund, 218 F.3d 392, 406 (5th Cir.2000) (“We will consider Tulane's motion for a judgment as a matter of law, with respect to punitive dam- ages, by applying Kolstad , in full to the record before us, without concern that our retroac- tive application is unfair.”); see also Blackmon v. Pinkerton Sec. & Investigative Servs., 182 F.3d 629, 636 (8th Cir.1999).

. Tidyman's does appear to concede that the sexually inappropriate joke by Babbit could be evidence of discrimination towards women. Tidyman’s contends that management swiftly responded by directing Babbit towards counseling after learning of this remark. Certainly, this type of evidence might establish a defense under Kolstad . However, we must review the evidence in the light most favorable to the plaintiffs. The plaintiffs introduced contrary evidence that Tidyman's' management permitted Babbit to forgo the recommended counseling. More importantly, the plaintiffs’ cases of intentional discrimination were based on the actions of multiple senior officers, not just Babbit.