In the late 1980s, Dr. Marjorie McMillan, head of the radiology department for the Massachusetts Society for the Prevention of Cruelty to Animals ("MSPCA"), learned that she was being paid less than the male heads of the organization's other departments. Defendants MSPCA and Dr. Gus Thornton now appeal the district court's denial of their motion to set aside verdicts in Dr. McMillan's favor on her various pay discrimination claims. They also appeal as excessive the jury's award of compensatory and punitive damages, and the district court's award of attorney's fees. Dr. McMillan cross-appeals the court's grant of judgment as a matter of law on her tortious interference with contractual relations claim against defendant Dr. Paul Gambardella, and its grant of summary judgment on her contract claims against the MSPCA and on her retaliation claim against all three defendants. We affirm the district court's ruling on the pay discrimination, tortious interference, contract, and retaliation claims, and, in part, the jury's compensatory damages verdict; reverse the award of punitive damages and compensatory damages representing lost benefits; and vacate the award of attorney's fees with directions that it be recalculated after remand.

We begin with an overview of general facts and prior proceedings, and leave more detailed recitations to the appropriate contexts.

Defendant MSPCA is a charitable, non-profit organization that combats cruelty to animals through educational programs and veterinary services. It operates Angell Memorial Animal Hospital ("Angell"), whose staff, during the relevant time period, totalled more than 200 employees, including veterinarians, interns, residents in post-graduate veterinary training, and technical and support staff. Defendant Dr. Gus Thornton began working at Angell in 1957, and was its chief of staff from 1966 until 1989, at which time he became president of the MSPCA. Defendant Dr. Paul Gambardella worked as a staff surgeon at Angell from 1975 until 1984, and as the interim director of surgery from 1984 until 1989, when Dr. Thornton appointed him chief of staff.

Plaintiff Dr. Marjorie McMillan was first employed by Angell in 1969 and thereafter was employed in various capacities until she left in 1977 to work in private practice. She returned in 1981 as the director of the radiology department, employed part time. She left Angell again in 1984 to spend one year completing coursework necessary for board certification and returned to Angell in 1985, again as director of radiology on a part-time schedule. From 1987 until 1991, she worked full time as head of radiology. In addition, from 1981 to 1991, she worked approximately seven hours each week at Windhover Bird Clinic ("Windhover"), a part-time private avian practice that she had established in Walpole, Massachusetts.

Until 1988, Angell had seven veterinary departments: clinics, cardiology, intensive care, clinical pathology, pathology, surgery, and radiology. All of the departments were headed by veterinarians, who, in addition to fulfilling their clinical duties, also served as administrative directors of their departments. During this time, Dr. Thornton was responsible for negotiating veterinarians' initial salaries and for setting discretionary annual increases from a fixed amount of funds. Although the department directors were responsible for such tasks as purchasing equipment, training interns and residents, scheduling, and making budget and compensation recommendations to Dr. Thornton, all of the staff reported to Dr. Thornton rather than to the individual directors.

In 1985 Dr. Thornton initiated a plan to restructure Angell's management, giving to the department directors greater responsibility, including the authority to make hiring, firing, compensation, and discipline decisions. As part of the reorganization, Dr. Thornton in 1988 consolidated Angell's seven departments into four departments: radiology, medicine, surgery, and pathology.

Dr. McMillan did not know the salaries of other veterinarians employed by Angell until 1987, when she learned that the salary of a newly-hired radiologist was $38,000. Dr. McMillan, whose salary at that time was $41,000, consulted Dr. Thornton and requested a raise so that her compensation would be comparable to that of the other department heads. Dr. Thornton eventually offered Dr. McMillan a raise to $47,000, which she did not accept because she had been offered a $50,000 salary for a non-administrative veterinary position at Tufts University Veterinary School. Dr. Thornton then agreed to adjust Dr. McMillan's salary to $51,000, effective in January 1988.

In 1989, Dr. McMillan discovered the disparity between her salary and that of the other department heads at Angell when a newspaper published a letter about the MSPCA that listed the various salaries. At the time, her salary was $58,000; her male counterparts in surgery, pathology, and medicine, by contrast, were earning $73,000, $80,244, and $73,199, respectively. On the basis of the salary disparity, Dr. McMillan filed a complaint with the Massachusetts Commission Against Discrimination.

Also in 1989, Dr. Gambardella, who became chief of staff upon Dr. Thornton's accession to the MSPCA presidency, set out to reevaluate and to improve the level of department heads' compensation. He began by creating job descriptions for each of the department heads in which the list of duties for the head of radiology was substantially the same as those for the other department head positions. He also consulted a study of the salaries at another major animal hospital and an informal market survey, on the basis of which he tentatively suggested to Kathleen Collins, Angell's human resources director, that the heads of radiology and surgery receive $88,000, the head of medicine, $90,000, and the head of pathology, $95,000. Collins then undertook an analysis of the department head jobs in order to "rationalize" the MSPCA's salary structure, giving a range of points in a number of categories of responsibility. On the basis of the point totals, she determined an appropriate salary for each of the department heads. On completion of the analysis, in 1990, Dr. McMillan received a raise from $58,295 to $72,000, which was substantially larger than that received by any of the other department heads.1

Also in 1990, Dr. McMillan entered into negotiations with Angell over the status of Windhover. Dr. McMillan suggested that Angell acquire her practice or, alternatively, that it rent her space so that she could carry on her bird practice there or that it pay her a separate, additional amount for her treatment of Angell's avian patients. When the negotiations came to an impasse, Dr. McMillan refused to continue to treat birds at Angell, a duty to which she had been devoting approximately six hours per week.

In 1991, Dr. Gambardella began to receive complaints from veterinarians--other department heads and members of the surgery staff, in particular--that Dr. McMillan was uncooperative and had created an atmosphere of animosity and inflexibility in the radiology department. On November 21, 1991, Dr. Gambardella summoned Dr. McMillan as she was preparing to perform a procedure on an anesthetized dog, fired her, gave her fifteen minutes to gather her belongings, and had her escorted out of the hospital. She was told not to return and was thereafter excluded from the premises. Although Angell had adopted in March 1990 a discipline policy directed at the correction of inappropriate behavior and retention of employees whenever possible, Dr. McMillan was given no advance warning of her termination.

On May 21, 1992, Dr. McMillan sued the MSPCA, Dr. Thornton, and Dr. Gambardella, alleging pay discrimination in violation of Title VII, 42 U.S.C. § 2000e-1; the Equal Pay Act ("EPA"), 29 U.S.C. § 206; and the Massachusetts anti-discrimination statute, Mass. Gen. Laws ch. 151B, and Equal Rights Act, Mass. Gen. Laws ch. 93, § 102. The complaint also alleged retaliation in violation of Title VII, the EPA, and Mass. Gen. Laws ch. 151B and ch. 272, §§ 92A and 98; breach of contract and negligent performance of contractual duties against the MSPCA; and tortious interference with contractual relations against Dr. Thornton and Dr. Gambardella. On March 17, 1995, the district court awarded the MSPCA summary judgment on Dr. McMillan's claims for pay discrimination under Title VII and Mass. Gen. Laws ch. 93, § 102, retaliation under state and federal law, breach of contract and negligent performance of contractual duties, and tortious interference with contractual relations against Dr. Thornton. The court also held that damages against the MSPCA were limited to $20,000, pursuant to the Massachusetts charitable immunity law, Mass. Gen. Laws ch. 231, § 85K. The court later held that chapter 231 did not apply and granted Dr. McMillan's motion for reconsideration of the limitation on damages.

In December 1995, plaintiff tried to a jury the remaining counts: pay discrimination in violation of the EPA against the MSPCA and in violation of Mass. Gen. Laws ch. 151B against the MSPCA and Dr. Thornton, and tortious interference with contractual relations against Dr. Gambardella. The jury found in her favor on all counts. On Dr. McMillan's pay discrimination claim under Mass. Gen. Laws ch. 151B, the jury awarded $183,784.50 as back pay accrued from 1985 through 1991, to which it added $178,915 for interest accrued from 1985. On her tortious interference claim against Dr. Gambardella, it awarded $197,550 as back pay, adding $84,846.50 as interest and $99,375.23 as front pay. Finally, the jury assessed punitive damages against Dr. Thornton and the MSPCA in the amount of $306,912.50. On November 12, 1996, the court awarded Dr. McMillan $447,928.81 in attorney's fees and $18,889 in costs. On the same day, the trial court set aside the jury verdict on the tortious interference claim against Dr. Gambardella but declined to set aside or grant a new trial on the pay discrimination verdict or award of back pay and punitive damages. This appeal and cross-appeal followed.

The MSPCA and Dr. Thornton contend that the trial court erred by denying their Fed.R.Civ.P. 50(b) motion for judgment as a matter of law, thereby ruling that there was sufficient evidence for a reasonable jury to find that they intentionally discriminated against Dr. McMillan, on the basis of her sex, in setting her salary between 1985 and 1991. They first argue that the evidence was insufficient to show that Dr. McMillan's skills, efforts, and responsibilities as a radiologist and department head were substantially similar to those of the male department heads.2 The remainder of their argument addresses the issue of pretext. Focusing on evidentiary questions, defendants contend that the trial court erred by admitting Dr. Gambardella's preliminary suggestions about appropriate department head salaries, and sporadic comments derogatory to women made by Dr. Thornton, as evidence that the reasons provided by Dr. Thornton for the pay disparity were pretextual. They further assert that there was insufficient evidence that Dr. Thornton's stated reasons for his determinations of base salaries and annual salary increases were pretextual. Specifically, they contend that a statistical analysis used to show pay differentials between males and females at the MSPCA could not support the verdict, and that Dr. Thornton's "stray remarks" were insufficient to expose as pretextual his purported method of establishing and raising salaries.3

Defendants challenge the pay discrimination verdicts under both state and federal discrimination law. To establish a prima facie case of sex discrimination based on salary disparity under the federal Equal Pay Act, a plaintiff must show that her employer was subject to the Act, and that she was paid less than her male counterparts who were performing work requiring substantially equal skill, effort, and responsibility under similar working conditions. Defendants must then prove by a preponderance of the evidence that the pay disparity can be explained by a legitimate factor such as seniority or performance. See Corning Glass Works v. Brennan, 417 U.S. 188, 195-96, 94 S. Ct. 2223, 2228-29, 41 L. Ed. 2d 1 (1974); Timmer v. Michigan Dep't of Commerce, 104 F.3d 833, 843 (6th Cir.1997). Thus, in cases brought under the Equal Pay Act, the plaintiff need not show that the defendant was motivated by a discriminatory animus, as required in cases governed by the McDonnell-Burdine burden shifting framework.

Under Massachusetts law, after a plaintiff presents her prima facie case, the requirements of which are the same as under federal law, see, e.g., Petsch-Schmid v. Boston Edison Co., 914 F. Supp. 697, 706-07 (D.Mass.1996), there is a rebuttable presumption of intentional discrimination, see Blare v. Husky Injection Molding Sys. Boston, Inc., 419 Mass. 437, 441, 646 N.E.2d 111 (1995). The defendant then has the burden of producing evidence of a legitimate, non-discriminatory reason for the pay disparity. See id. at 441-42, 646 N.E.2d 111. After the defendant has satisfied this burden, the presumption of intentional discrimination disappears, see id. at 442, 646 N.E.2d 111, and plaintiff assumes the burden of proving, by a preponderance of the evidence, that the defendant's stated reasons for the pay disparity were not the real reasons, see id. at 444-45, 646 N.E.2d 111.4

We review de novo a trial court's decision to deny or to grant a defendant's motion for judgment as a matter of law. See Morrison v. Carleton Woolen Mills, Inc., 108 F.3d 429, 436 (1st Cir.1997). The standard of review for motions for judgment as a matter of law requires us to view the evidence "in the light most favorable to the nonmoving party, drawing all reasonable inferences in its favor." Id. We will not set aside a jury verdict as a matter of law unless there was only one conclusion the jury could have reached. See Conway v. Electro Switch Corp., 825 F.2d 593, 598 (1st Cir.1987).

We agree with the district court that a reasonable jury could have found that Dr. McMillan had satisfied her burden of establishing a prima facie case of pay discrimination. Contrary to defendants' assertion that plaintiff failed to offer evidence comparing the skill, effort, and responsibilities of her position as radiology department head with that of the other department heads, plaintiff proffered testimony that the radiology department is comparable with the other departments in all three respects. In particular, testimony from Dr. Neil Harpster, Angell's cardiology department head from 1985 to 1988, and Dr. Allen Sisson, a staff veterinarian, provided evidence both that the procedures performed by Dr. McMillan in her job were among the most technically difficult procedures performed by any veterinarian at Angell and that the hours Dr. McMillan spent at her job were comparable to those of the other department heads. In addition, the evidence showed that Dr. McMillan supervised more staff than did Dr. James Carpenter, the head of the pathology department, and that, in Kathleen Collins's analysis of department head jobs in the late 1980s, radiology was ranked equally with pathology and medicine in seven out of eight categories of responsibility.5 Moreover, although the job descriptions that Dr. Gambardella formulated in 1989, without more, are not dispositive, the fact that the job requirements for all of the department heads were substantially the same is supportive of other evidence that the department head positions required substantially equal skill, effort, and responsibility. Finally, a market survey published sometime in the first half of the 1980s and admitted into evidence demonstrated that the average salaries of radiologists were comparable to those of other specialists, indicating that radiology is not an intrinsically less demanding area of veterinary medicine. The evidence was thus sufficient to support a finding that plaintiff had presented a prima facie case.

We turn next to the issue of pretext.6 At trial, defendants presented several reasons to explain why Dr. McMillan was paid less than the other department heads. Dr. Thornton testified that, as chief of staff, he set a department head's compensation by conducting an informal survey of what similarly situated veterinarians were being paid at other institutions and by negotiating compensation with the prospective department head. He stated that, in general, radiologists commanded lower compensation than did veterinarians in other specialties. He further testified that, to determine yearly percentage salary increases, he evaluated each department head's productivity and responsibilities. Dr. Thornton stated that, in his view, the head of medicine, Dr. Michael Bernstein, had significantly more responsibilities, including the oversight of more staff, than did Dr. McMillan as head of radiology. Dr. Thornton likewise claimed that the head of pathology, Dr. Carpenter, had more responsibilities than did Dr. McMillan because he supervised more staff, had a larger budget, and was nationally known, and because pathology, unlike radiology, incorporated several disciplines such as clinical and anatomic pathology. The head of surgery, Dr. Gambardella at the time, merited greater compensation in Dr. Thornton's view because he was accountable for the budget and staff hiring and discipline for the department, supervised more staff, and had significant responsibility for patient care and client contact.

Defendants make two primary arguments regarding the pretext prong of plaintiff's pay discrimination claim. First, they argue that the jury's findings that Dr. Thornton's explanations were pretextual are tainted because the district court improperly admitted into evidence (1) Dr. Gambardella's preliminary suggestions in 1989 regarding salary increases as proof of Dr. Thornton's state of mind throughout the time period that he set Dr. McMillan's salary and yearly increases, and (2) evidence of sporadic distasteful remarks that Dr. Thornton made to women colleagues in 1979 and 1989. Second, defendants assert, in essence, that, even taking into account all of the evidence actually admitted, plaintiff failed to prove, as required under Massachusetts law, that Dr. Thornton's stated procedures for setting salaries were pretextual.

We review a district court's evidentiary rulings for abuse of discretion. See United States v. Young, 105 F.3d 1, 8 (1st Cir.1997). A judge should not exclude relevant evidence, unless its probative value is substantially outweighed by the danger of unfair prejudice. See Fed.R.Evid. 403. "Only rarely--and in extraordinarily compelling circumstances--will we, from the vista of a cold appellate record, reverse a district court's on-the-spot judgment concerning the relative weighing of probative value and unfair effect." Freeman v. Package Mach. Co., 865 F.2d 1331, 1340 (1st Cir.1988).

The court's admission of testimony about Dr. Gambardella's 1989 suggestions regarding salary increases was not an abuse of its discretion. One of Dr. Thornton's stated reasons for setting salaries at Angell as he did during the period from 1985 to 1989 was that radiologists commanded lower salaries on the market than did veterinarians in other specialties. The issue was whether this reason was pretextual. The informal 1989 survey on which Dr. Gambardella based his suggestions showed relative parity among salaries of radiologists and other specialists. Accordingly, Dr. Gambardella's suggestions were probative of whether radiology is intrinsically a less-valued veterinary specialty and, therefore, of whether Dr. Thornton's stated reason was, in fact, the real reason. They are probative also of Dr. Thornton's credibility because they were based on what veterinarians in various specialties were worth in the market, the very factor on which Dr. Thornton purported to rely in setting salaries at Angell. Further, although defendants contend that, regardless of the status of radiology at other institutions, the radiology department at Angell was not equal to Angell's other departments with regard to the responsibilities required of its head, we think that the district court did not abuse its discretion in determining that Dr. Gambardella's testimony regarding salary parity at other institutions was not unfairly prejudicial to defendants in their attempt to make this point.

Whether the court's admission into evidence of the so-called stray remarks was proper presents a more difficult question. There is, of course, no question that Dr. Thornton's motivation in paying Dr. McMillan less than the other department heads was a material issue in the case. We note initially that stray remarks may properly constitute evidence of discriminatory intent for the jury to consider in combination with other evidence. See Bevan v. Honeywell, Inc., 118 F.3d 603, 610 (8th Cir.1997); O'Connor v. DePaul Univ., 123 F.3d 665, 671-72 (7th Cir.1997) (" '[S]tray remarks'--made by the decisionmaker but not related to the disputed employment action--may be relevant to the question of pretext...."). Yet even if the remarks are relevant for the pretext inquiry, their probativeness is circumscribed if they were made in a situation temporally remote from the date of the employment decision, see Armbruster v. Unisys Corp., 32 F.3d 768, 779 (3d Cir.1994), or if they were not related to the employment decision in question or were made by nondecisionmakers, see, e.g., Ezold v. Wolf, Block, Schorr & Solis-Cohen, 983 F.2d 509, 545 (3d Cir.1992) ("Stray remarks by non-decisionmakers or by decisionmakers unrelated to the decision process are rarely given great weight....").

The allegations in this case are that, during his time at Angell, Dr. Thornton made various unprofessional remarks to women with whom he worked. In 1979 or 1980, he talked to a lab technician about warming up a vial of medicine in her brassiere. Approximately ten years later, he made a remark to the MSPCA's lobbyist about her cleavage and then later asked whether, to stop bad publicity about the MSPCA by a particular reporter, she had slept with the reporter. Arguably, at least, these remarks may have been relevant and probative as to Dr. Thornton's motivation in setting the salaries of male and female department heads, in the sense that they evince a derogatory attitude toward women.

However, not only did the remarks not involve the employment decision at issue, they did not involve employment at all. Further, at least one of the remarks was made several years before any dispute arose over the pay disparity between Dr. McMillan and other department heads. Such factors heighten the remoteness of the remarks, arguably to the point at which they are no more probative than they are prejudicial. Nonetheless, the question we must address in determining admissibility is not whether we agree with the district court but, rather, whether the district court abused its discretion. See Espeaignnette v. Gene Tierney Co., Inc., 43 F.3d 1, 8 (1st Cir.1994). We think that it did not. Although the comments were unrelated to Dr. McMillan's situation, and although the remarks may, in the end, have fallen on the other side of the imprecise nexus/no-nexus boundary, the court's determination to admit the remarks for the jury to consider with the other evidence was not so one-sided such that it did not act within its discretion.7

Defendants contend in addition that the district court erred by giving the jury an improper instruction on the relevance of Dr. Thornton's remarks.8 They assert that the court both analyzed and added to the evidence by equating Dr. Thornton's remarks with his attitude about the value of women when no evidentiary foundation supplied the nexus. Further, they argue that the instruction misled the jury to believe that stray remarks alone were sufficient evidence of pretext.

Defendants' objections to the jury instructions were not clearly articulated at trial and, therefore, not properly preserved.9 In such a situation, we review only for error resulting in a "clear miscarriage of justice." Transnational Corp. v. Rodio & Ursillo, Ltd., 920 F.2d 1066, 1069 (1st Cir.1990).

In reviewing jury instructions such as these, we focus principally on "whether they tended to confuse or mislead the jury on the controlling issues." Service Merchandise Co., Inc. v. Boyd Corp., 722 F.2d 945, 950 (1st Cir.1983). In this instance, we discern no problems with the judge's stray remarks instructions. First, contrary to defendants' suggestion, federal judges may comment on the evidence. See Leshore v. County of Worcester, 945 F.2d 471, 474 (1st Cir.1991). Second, in this case the judge merely made explicit what the jury might infer if it chose to credit the stray remarks. This may be, as defendants assert, "connect[ing] the dots," but it is not a suggestion that a particular conclusion must be reached. Moreover, the court made clear to the jury that it was equally plausible to regard the remarks as "random and completely out of character," and thus to infer nothing at all about them. Likewise, the instructions in no way misled the jury to believe that the stray remarks might, alone, be sufficient to prove that defendants' stated reasons for paying Dr. McMillan less were mere pretext. That the jury "might decide that a given remark is evidence of a discriminatory mindset" seems to us to be nothing other than a restatement of why such evidence is presented to the jury in the first place; it implies nothing about how much or how little of such evidence is sufficient for the jury to find in the plaintiff's favor. In any event, the instructions did not result in a "clear miscarriage of justice."

As to the sufficiency of the evidence of pretext, defendants attack the analysis of plaintiff's statistical expert, Dr. Arlene Ash. They argue that the analysis cannot support the jury verdict because it did not incorporate variables that Dr. Thornton argued he applied when he set initial salaries and determined annual incremental increases. They also contend that it was based on a legally insufficient sample size.

We turn first to the argument that Dr. Ash's analysis ignored important factors. Defendants assert that Dr. Ash's model should have included such variables as market factors, negotiations, veterinary experience, performance evaluations, productivity, and veterinarians' overall contribution to the hospital. As a preliminary matter we note that, although a failure to include particular variables in a regression analysis may diminish the probativeness of the analysis, "it can hardly be said, absent some other infirmity, that an analysis which accounts for the major factors ... 'must be considered unacceptable as evidence of discrimination.' " See Bazemore v. Friday, 478 U.S. 385, 400, 106 S. Ct. 3000, 3009, 92 L. Ed. 2d 315 (1986). Dr. Ash incorporated into the analysis such legitimate variables as veterinarians' seniority at Angell, department head status, board-approved specialization, area of specialization, and departmental budget size. Although the models she developed may not have included every relevant variable, her testimony shows solid reasoning in her determinations to exclude certain variables that the defendants argued should have been included. We therefore reject this argument.

Defendants further assert that Dr. Ash's analysis was effectively based on an insufficient sample size because Dr. Ash used factors which could apply only to department heads. This objection is opaque and unpersuasive. Dr. Ash's choice of sample, which was the approximately forty-six veterinarians who worked at Angell from 1981 to 1991, was not fatally deficient. The analysis made a distinction between those veterinarians who were department heads and those who were not. Only one variable--budget size--was specific to department head status, and it was not applied to the veterinarians who were not department heads.

Defendants continue, however, arguing that statistical evidence "should rarely be accorded any weight" in disparate treatment cases. We agree that in a disparate treatment case, such as this one, the necessary focus on the treatment of a particular individual appears incongruous with an analysis that necessarily involves numerous individuals and that, therefore, is intuitively more probative in the disparate impact context. See LeBlanc v. Great Am. Ins. Co., 6 F.3d 836, 848 (1st Cir.1993); Cumpiano v. Banco Santander P.R., 902 F.2d 148, 156 (1st Cir.1990). Nonetheless, such analyses are admissible even in disparate treatment cases unless they are "so incomplete as to be inadmissible as irrelevant."10 See Bazemore, 478 U.S. at 400 n. 10, 106 S. Ct. at 3009 n. 10. Further, we have held that "[s]tatistical evidence is permissible on the issue of pretext in a disparate treatment case since 'statistics as to ... employment policy and practice may be helpful to a determination of whether [the employer's conduct] conformed to a general pattern of discrimination.' " Freeman, 865 F.2d at 1342 (quoting McDonnell Douglas Corp. v. Green, 411 U.S. 792, 805, 93 S. Ct. 1817, 1825-26, 36 L. Ed. 2d 668 (1973)). We agree with the court below that, in a case such as this, "the jury could have found a plausible connection between Dr. Thornton's treatment of women generally at Angell and his treatment of Dr. McMillan in particular."

Leaving aside whether Dr. Ash's analysis was watertight in all respects, we think it important to note that Dr. Ash's testimony was focused, perceptive, and understandably compelling to the jury. Reviewing the record, it is apparent to us that, whatever "infirmities" existed in her analysis, none was so substantive that no reasonable jury could have relied on it, along with the other evidence, in reaching a verdict. Rather, if Dr. Ash's analysis omitted what defendants argue are important variables, or was deficient in other respects, and thereby did not reflect whether Dr. Thornton's salary decisions were based at least in part on gender, it was up to defendants to exploit and discredit the analysis during cross examination. Furthermore, although the conflicting expert testimony they provided placed Dr. Ash's analysis into question, the jury was free to disregard that testimony and place credence in that of Dr. Ash. Indeed, as the district court noted, although the defendants' expert, Dr. Herbert Weisberg, disputed Dr. Ash's finding of a pattern of discrimination, he admitted that her model had produced statistically significant results.11

In arguing that plaintiff did not show evidence of pretext, Defendants' position seems to be that, aside from evidence of Dr. Thornton's stray remarks and Dr. Ash's statistical analysis, there remains no basis on which the jury might reasonably have found pretext under Massachusetts law and pay discrimination under the EPA. But in addition to the statistical analysis, plaintiff offered other evidence, which, in combination with the statistical evidence, was sufficient to support a finding of pretext. Although defendants posited that the radiology department was smaller than the other departments, plaintiff presented evidence that, in fact, cardiology was the smallest department, with only two employees under the supervision of the department head. The record reveals that the issue of seniority provided an additional basis for a jury finding of pretext. Dr. Thornton testified that both the number of years a veterinarian had been in practice and his or her seniority at Angell were factors he weighed in setting initial salaries for a particular position. And, indeed, it appears from the evidence that he did consider seniority in setting the salaries of Dr. Gambardella when he became head of surgery and of Dr. Michael Aronsohn when he succeeded Dr. Gambardella in that position. Dr. Thornton's testimony indicates, however, that he ignored seniority and years in practice in the case of Dr. McMillan, treating her as a new hire when she returned to Angell in 1985.12 The testimony shows that, even though Dr. McMillan in 1985 had spent two and a half years as radiology department director (albeit part time) and five years at Angell in various capacities, Dr. Thornton did not consider that experience in setting her salary. Rather, he testified that he based her salary on a survey that recommended $29,000 to $50,000 for the least experienced radiologists, and that, in fact, he paid her $35,000 simply because he could. But $35,000 was essentially the same salary Dr. McMillan had been earning in 1983, before she spent a year away from Angell.13 As for annual percentage increases of salaries, although Dr. Thornton testified that he based his decisions on subjective factors such as productivity, contribution to the hospital, and interaction with other employees, he also testified that he did not systematically quantify contributions and interaction and that he did not evaluate the radiology department on the basis of productivity. Finally, testimony from Dr. Bernstein and Dr. Aronsohn placed into question Dr. Thornton's assertion that some department heads were so-called "strong department heads," and had more authority on that basis. Thus, a reasonable jury could have found that Dr. Thornton's stated reason for his salary decisions was not true or, in other words, pretextual.

Accordingly, because we think that there was sufficient properly admitted evidence for a jury to find that Dr. Thornton's stated reasons for the pay disparity were pretextual, we choose not to disturb the district court's denial of judgment as a matter of law with regard to the pay discrimination claims based on Mass. Gen. Laws ch. 151B and on the Equal Pay Act.14

Defendants present a variety of arguments regarding calculation of damages, most of which need not detain us long. At the outset, we note that we do not disturb a jury's award of damages unless it "exceed[s] any rational appraisal or estimate" of what the damages should be. Kolb v. Goldring, Inc., 694 F.2d 869, 871 (1st Cir.1982) (internal citation omitted).

Defendants argue that Dr. McMillan's back pay damages award should be set aside or reduced because the award was excessive and unsupported by the evidence. They contend that the jury's award of back pay based on an average of the salaries of the male department heads between 1985 and 1991 has no legal basis. We disagree. As Dr. McMillan rightly points out, the cases defendants rely on to support their argument, Scarfo v. Cabletron Sys., Inc., 54 F.3d 931, 954 (1st Cir.1995) and E.E.O.C. v. Liggett & Myers, Inc., 690 F.2d 1072, 1075 (4th Cir.1982), are inapplicable in this situation, and, if anything, serve to support rather than to discredit the averaging method. In Liggett & Myers, the court rejected a calculation of back pay based on the average salary in favor of a calculation based on the salary of the highest paid employee comparator. See 690 F.2d at 1074. Although Scarfo holds that back pay should be calculated by comparing the plaintiff's salary to that of "arguably equivalent" employees, it in no way indicates that averaging the salaries of two or more such employees is inappropriate. See 54 F.3d at 954. Here, averaging the salaries of the male department heads was a rational approach. In fact, in this context, a more appropriate method has not been proposed. We thus find that the method the jury apparently used to determine the amount of the back pay award was not inappropriate.

Defendants next assert that Dr. McMillan is not entitled to recovery, as part of her back pay award, for "lost benefits"--health insurance, life insurance, and retirement contributions--because she did not prove that she had lost them. Plaintiff testified that she had been told that such benefits constituted twenty-one percent of an employee's salary and that, in her capacity as department head, she budgeted twenty-one percent of her supervisees' salaries as benefits. On this basis, the jury awarded Dr. McMillan an additional twenty-one percent of her back pay award to represent the value of the lost benefits.

Lost benefits are recoverable only if the plaintiff has offered evidence of out-of-pocket expenses for the same benefits. See Kossman v. Calumet County, 800 F.2d 697, 703-04 (7th Cir.1986) (holding that, to recover damages representing benefits, a plaintiff must show that she actually incurred insurance or medical care expenses); Taylor v. Central Pa. Drug & Alcohol Servs. Corp., 890 F. Supp. 360, 372 (M.D.Pa.1995); Berndt v. Kaiser Aluminum & Chem. Sales, Inc., 604 F. Supp. 962, 965 (E.D.Pa.1985). In this case, even if the budgeted value of benefits corresponding to plaintiff's salary had been less than the budgeted value of benefits corresponding to the salaries of the other department heads, plaintiff presented no evidence that she incurred insurance expenses. In addition, she presented no evidence that any employer-contributed retirement benefits were tied to the amount of her salary. Further, that benefits may have amounted to twenty-one percent of her supervisees' salaries does not mean that benefits constituted an equal percentage of higher salaries. Indeed, it would be logical to expect that employer insurance contributions at all salary levels were substantially the same and that, therefore, benefits were a considerably lower percentage of higher salaries. Because there was no competent evidence from which a reasonable jury could conclude that Dr. McMillan suffered any loss in benefits as a result of her lower salary, Dr. McMillan's back pay award should be accordingly reduced by the amount of the lost benefits award.

Defendants' third argument is that we should import the three-year cap on back pay liability under the EPA, see 29 U.S.C. § 255, to Dr. McMillan's state back pay award under chapter 151B. This argument is without merit. No court decisions impose such a cap under Massachusetts law, and defendants present no rationale as to why state law should be the same as federal law. Further, Mass. Gen. Laws ch. 151B, as interpreted by the Supreme Judicial Court of Massachusetts ("SJC"), mandates "make-whole relief," which encompasses damages which are "the natural and probable consequences of the illegal conduct." Conway v. Electro Switch Corp., 402 Mass. 385, 388, 523 N.E.2d 255 (1988) (internal citation omitted). Indeed, the SJC has declined to follow federal precedent in order to interpret chapter 151B more liberally than the federal courts have interpreted parallel federal law provisions. See Lynn Teachers Union, Local 1037 v. Massachusetts Comm'n Against Discrimination, 406 Mass. 515, 521 n. 7, 549 N.E.2d 97 (1990). We thus decline to adopt defendants' suggestion.

Fourth, defendants contend that Dr. McMillan's punitive damages award against the MSPCA on her pay discrimination claim should be set aside, or, alternatively, that a new trial should be granted on the issue. They challenge the punitive damages award on the basis that it was unwarranted, excessive, and obtained as a result of irrelevant testimony about Dr. Gambardella's termination of Dr. McMillan. We agree with defendants that the award was unwarranted and excessive.

In a lawsuit such as this one, in which state law provides the basis of decision, "the propriety of an award of punitive damages for the conduct in question, and the factors the jury may consider in determining their amount, are questions of state law." Browning-Ferris Ind. v. Kelco Disposal, 492 U.S. 257, 278, 109 S. Ct. 2909, 2921, 106 L. Ed. 2d 219 (1989). In reviewing an award of punitive damages, our role is to determine whether the district court abused its discretion in determining that the jury's award of punitive damages was within the confines of state law and that defendants are not entitled to a new trial or remittitur. See id. at 279, 109 S.Ct. at 2922. Although juries have wide discretion in determining the amount of punitive damages, and the trial court has broad discretion to affirm the jury's award of damages, see Fishman v. Clancy, 763 F.2d 485, 489-90 (1st Cir.1985), appeals courts should reduce or set aside awards that are "grossly excessive" or "shocking to the conscience," id. at 489 (citing LaForest v. Autoridad de Las Fuentes Fluviales de P.R., 536 F.2d 443, 447 (1st Cir.1976)).

The jury in this case assessed punitive damages under chapter 151B against the MSPCA in the amount of $135,662.50 and against Dr. Thornton in the amount of $171,250. Although the statute authorizes punitive damages, see Mass. Gen. Laws ch. 151B, § 9, it does not specify when they should be awarded. However, other punitive damages provisions in Massachusetts law provide that such damages shall apply only on a finding that defendants' conduct was wilful, wanton, or reckless, see, e.g., Mass. Gen. Laws ch. 229, § 2, and the SJC has further refined the criteria, holding that punitive damages are warranted "where a defendant's conduct warrants condemnation and deterrence," Bain v. City of Springfield, 424 Mass. 758, 767, 678 N.E.2d 155 (1997).

Dr. McMillan contends that, because the jury found intentional discrimination, she was eligible for an award of punitive damages if the jury so decided in its discretion. We observe, however, that nowhere does Massachusetts law state that a finding of intentional discrimination necessarily justifies an award of punitive damages. This accords with the logic that, even in situations involving intentional misconduct, compensatory damages may provide sufficient punishment and deterrence.

The district court's jury instructions on punitive damages were that the jury could award punitive damages if it found defendants' conduct to have been "egregious and beyond the pale of tolerable," and that such damages are "intended to punish the defendants as a warning, both to them and other like-minded individuals, that society will not tolerate grievous discriminatory behavior." The district judge further stated that "[a]ny sum you award as punitive damages should be commensurate with your own conscience, the outrageousness of the conduct that you intend to punish, and the character of the MSPCA as both an employer and as a nonprofit enterprise." As these instructions imply, above the inquiry for finding intent, the jury had to conduct a second inquiry to determine whether defendants' intentional conduct was egregious enough for an award of punitive damages. It follows, then, that just as we may review the record to determine whether a reasonable jury could have found that the defendants intentionally discriminated against Dr. McMillan, we may also review the record to determine whether a reasonable jury could have found defendants' conduct sufficiently worthy of condemnation and deterrence to justify exemplary damages.

Although we affirm the jury's finding of discrimination, we cannot say, on the basis of the record before us, that Dr. Thornton's and the MSPCA's intentional misconduct calls for punishment and deterrence beyond that secured by the compensatory award. Determining what conduct rises to the level at which an award of punitive damages is appropriate is a difficult task, but the evidence shows that the actions at issue in this case do not. We therefore hold that the district court abused its discretion in upholding the award of punitive damages, and that the award of punitive damages in this case constitutes a grossly excessive award of damages that shocks the conscience.15

Fifth, defendants argue that the MSPCA's liability is limited to $20,000 by a damages cap for charitable organizations under the Massachusetts charitable immunity law, which provides that liability for any "cause of action based on tort brought against a [charitable] corporation" shall not exceed $20,000 exclusive of interest and costs. See Mass. Gen. Laws ch. 231, § 85K. Although this statute applies to negligent, reckless, and intentional torts in the employment context, see St. Clair v. Trustees of Boston Univ., 25 Mass.App.Ct. 662, 665-66, 521 N.E.2d 1044 (1988), it has been unclear whether a claim under chapter 151B is a "cause of action based on tort" and thus subject to the $20,000 limitation. We conclude that it is not.

Defendants and plaintiff engage in extensive exercises of interpretation to suggest the relevance of a variety of cases to this issue. Ultimately, we agree with Dr. McMillan that Linkage Corp. v. Trustees of Boston Univ., 425 Mass. 1, 27, 679 N.E.2d 191 (1997), provides the most guidance. In that case, the SJC held that the chapter 231, section 85K damages limitation does not apply to damages awarded under Mass. Gen. Laws ch. 93A, the Massachusetts Consumer Protection Statute, on the basis that chapter 93A "creates an independent statutory basis of liability" and "forbid[s] conduct not previously unlawful under the common law of contract and tort or under any prior statute," id. (internal citation omitted). Like chapter 93A, chapter 151B creates rights that did not exist under the common law, see, e.g., Melley v. Gillette Corp., 19 Mass.App.Ct. 511, 512-13, 475 N.E.2d 1227 (1985), aff'd, 397 Mass. 1004, 491 N.E.2d 252 (1986); the causes of action to which it gives rise thus cannot properly be called causes of action in tort. Accordingly, we hold that the damages award to Dr. McMillan pursuant to chapter 151B is not subject to the constraints of chapter 231.

Defendants next contend that the district court misapplied the relevant legal standard when it awarded attorney's fees because it did not award the appropriate hourly rates for the different types of services performed and instead allowed Dr. McMillan's counsel to recover her standard hourly rate ($285 per hour) for performing tasks appropriate to either a less experienced lawyer or a secretary or paralegal. We agree.

We have established that "clerical or secretarial tasks ought not to be billed at lawyers' rates, even if a lawyer performs them." Lipsett v. Blanco, 975 F.2d 934, 940 (1st Cir.1992). Thus, "[t]ime spent on clerical or secretarial tasks by attorneys should be compensated at a rate commensurate with the nature of the tasks." Massachusetts Dep't of Pub. Health v. School Comm. of Tewksbury, 841 F. Supp. 449, 460 (D.Mass.1993); see Deary v. City of Gloucester, 789 F. Supp. 61, 66 (D.Mass.1992), aff'd, 9 F.3d 191 (1st Cir.1993). We think that it was an abuse of discretion to award the same rate for all tasks performed by Dr. McMillan's counsel, without regard to the nature of the tasks. Accordingly, we remand this issue in order that the court may determine those tasks which could have been adequately performed by a less-experienced lawyer or by a secretary or paralegal. The compensation for tasks so designated should be reduced to an appropriate level.

Finally, defendants argue that the trial court erred by computing interest on the back pay award from the date on which the cause of action accrued, rather than from the date plaintiff filed her complaint. This argument has no merit. As the district court stated, "the purpose of interest is to insure that the present value of an award fully compensates a plaintiff for her losses." Moreover, the SJC has held that an award of interest may be used to make an aggrieved party whole in chapter 151B actions. See Conway, 402 Mass. at 390 n. 7, 523 N.E.2d 255. Here, an award of interest from the date Dr. McMillan was paid less than her counterparts comports with these standards.

Plaintiff cross-appeals from the district court's grant of judgment as a matter of law in favor of defendants setting aside the jury's finding that Dr. Gambardella had tortiously interfered with Dr. McMillan's employment relationship with the MSPCA. She argues that she presented ample evidence during the trial to show that Dr. Gambardella's actions leading up to and including her termination were based on "an improper motive, active animosity and spite." We do not agree.

Viewing the evidence in a light most favorable to the plaintiff, see Morrison, 108 F.3d at 436, we conclude that the evidence did not permit a reasonable jury to find for the plaintiff. Under Massachusetts law, to make out the tort of intentional interference with a contractual relationship, a plaintiff must present evidence of "actual malice" or a "spiteful, malignant purpose, unrelated to the legitimate corporate interest." Shea v. Emmanuel College, 425 Mass. 761, 764, 682 N.E.2d 1348 (1997) (quoting Wright v. Shriners Hosp. for Crippled Children, et al., 412 Mass. 469, 476, 589 N.E.2d 1241 (1992)). In addition, "[w]hen an employer or supervisor is acting within the scope of his employment responsibilities, the hiring and firing decisions are privileged unless he acted with malevolence." Walker v. Waltham Hous. Auth., 44 F.3d 1042, 1049 n. 1 (1st Cir.1995).

Dr. McMillan presented evidence that Dr. Gambardella questioned her decisions and actions on various occasions, that he had her replaced as the head of a hospital committee, that he did not reprimand another doctor who made a derisive comment to her during a staff meeting, and that he terminated her without notice, had her summoned while she was involved in a complicated radiological procedure, and then barred her from the hospital grounds. To support her claim of tortious interference, Dr. McMillan points to O'Brien v. New England Tel. & Tel. Co., 422 Mass. 686, 687, 664 N.E.2d 843 (1996), in which the SJC held reasonable a jury's finding that an employer tortiously interfered with an employment relationship when he incessantly harassed employee, called her names in front of coworkers, and actively sought to make her cry. The facts of O'Brien are, however, much more egregious than those in this case. Insensitive though Dr. Gambardella's actions may have been, they did not demonstrate that type of ill will that the employer's actions in O'Brien did. We think that the district court properly concluded that these actions and statements did not demonstrate "anything more than a personality conflict with Dr. McMillan." For the most part, Dr. Gambardella's actions were properly within the scope of his duties as chief of staff at Angell, and, in any event, they failed to evince malevolent intent. No reasonable jury could have concluded otherwise.

Dr. McMillan argues next that the district court erred by granting the MSPCA, Dr. Thornton, and Dr. Gambardella summary judgment as to her claim that her termination was in retaliation for her filing of an administrative complaint of sex discrimination in 1989. She asserts that the district court misapplied the law in ruling that she had failed to establish a causal connection between the filing and her termination because too much time had lapsed between the two events. She also argues that the court improperly viewed the facts from the defendants' perspective.

We review the district court's grant of summary judgment to defendants de novo. See Associated Fisheries of Me., Inc. v. Daley, 127 F.3d 104, 109 (1st Cir.1997). Further, to determine whether the summary judgment record is sufficient to support a finding that the employment action was retaliatory, we review only those facts that were available at the summary judgment stage, construing all facts and inferences therefrom in favor of the plaintiff. See Aybar v. Crispin-Reyes, 118 F.3d 10, 13 (1st Cir.1997).

To succeed on a retaliation claim under both state and federal law, when, as here, there is no direct evidence of retaliatory animus, a plaintiff must both establish a prima facie case and prove that the defendants' legitimate business reasons for terminating the plaintiff were pretextual. See Fennell v. First Step Designs, Ltd., 83 F.3d 526, 535 (1st Cir.1996) (discussing Title VII's analytical framework); Lewis v. Gillette, Co., 22 F.3d 22, 24-25 (1st Cir.1994) (observing that Massachusetts retaliation law parallels federal law). Specifically, to establish a prima facie case of retaliation, a plaintiff must show that (1) she engaged in protected conduct under federal or Massachusetts law; (2) she suffered an adverse employment action; and (3) a causal connection existed between the protected conduct and the adverse action. See Fennell, 83 F.3d at 535. Once the plaintiff has made a prima facie showing, the burden of production shifts to the defendant to articulate a legitimate, non-retaliatory reason for its employment decision. See id. If the defendant does so, the plaintiff must show that the defendant's proffered reason was not, in fact, the real reason for the decision and that the decision was the result of the defendant's retaliatory animus. See id.

We conclude that Dr. McMillan did not establish a prima facie case because she presented no evidence of a causal link between her termination and the filing of her complaint. The strongest argument Dr. McMillan makes in this regard is that Dr. Gambardella urged her not to file an MCAD complaint, stating that her doing so would make their relationship difficult. But, in light of undisputed facts in the summary judgment record, this does not suggest a causal connection between Dr. McMillan's filing of the discrimination claim, on one hand, and the actual deterioration and eventual termination of the employment relationship, on the other. The MSPCA and Dr. Gambardella asserted that Dr. McMillan had had an acrimonious relationship with other department heads and staff veterinarians at Angell, that she refused to treat birds after Angell refused to purchase her bird practice or pay her separately for the time she spent with avian patients, and that Dr. Gambardella received complaints from staff members about Dr. McMillan's attitude. Indeed, Dr. Gambardella stated that, by late 1991, his relationship with Dr. McMillan had deteriorated to the extent that he told Dr. Thornton that he would himself resign if Dr. McMillan were not terminated.

In addition, we agree with the district court that, even if Dr. McMillan had established a prima facie case, she did not present material facts that would permit a reasonable jury to find that Dr. Gambardella's explanation for her firing was pretextual and influenced instead by a retaliatory motive. Although plaintiff presented evidence that she continued to work hard and to maintain a good relationship with much of the veterinary staff, she simply did not proffer evidence to refute defendants' showing that the veterinary staff complained to Dr. Gambardella about her performance and demeanor. In light of this undisputed evidence, even if Dr. Gambardella may have exercised arguably poor business judgment in terminating her, and in doing so in an abrupt, clumsy, and unfeeling manner, it would be unreasonable on the evidence for a factfinder to conclude that his explanation for the termination was, in fact, not the true reason. We therefore decline to order a new trial on this issue.

Third, Dr. McMillan argues that the district court improperly granted the MSPCA summary judgment on her claim that the MSPCA's failure to comply with the progressive discipline policy outlined in its Rules and Regulations Memorandum constituted a breach of contract because the policy was a contract governing her employment. She contends that the court improperly applied in a "wooden" fashion factors set forth in Jackson v. Action for Boston Community Dev., 403 Mass. 8, 14-15, 525 N.E.2d 411 (1988), for determining whether a personnel manual's policy guidelines constitute a binding employment agreement. For support, she relies on O'Brien, 422 Mass. at 691-93, 664 N.E.2d 843, in which the SJC held that the terms of an employees' manual become part of an implied employment contract if the parties so agree, even if the terms of the manual have not been negotiated, and even if, in a continuing employment relationship, an employee only reasonably believed that the employer "was presenting the manual as a statement of the conditions under which employment would continue." Id. at 693, 664 N.E.2d 843.

We agree with the district court's reasoning on this issue and its observation that, although personnel manuals may sometimes constitute an employment agreement, see Jackson, 403 Mass. at 13, 525 N.E.2d 411, the facts Dr. McMillan alleged do not support such an inference. As the district court observed, "Dr. McMillan does not claim to have bargained for any of the terms of the manual, to have signed the manual, or to have executed any other employment agreement with the MSPCA. Moreover, the manual described itself as a disciplinary guideline that 'is not all inclusive.' " McMillan v. MSPCA, 880 F. Supp. 900, 910 (D.Mass.1995). Contrary to Dr. McMillan's assertion, the court's analysis was not an application of inflexible prerequisites to finding an employment contract. Rather, the court's analysis employs the factors appropriately, as mere guidelines in the determination. See O'Brien, 422 Mass. at 692, 664 N.E.2d 843. We do the same. In addition to the factors highlighted by the district court, we note additionally that there is no evidence that Dr. McMillan reasonably expected that the MSPCA would follow the procedures with regard to her conduct or that of other supervisory veterinarians at Angell.

Dr. McMillan's next argument is that she should be granted a new trial on her claim for emotional distress damages. Dr. McMillan withdrew that claim after the district court adopted a magistrate's ruling ordering the disclosure of records of Dr. McMillan's therapist, which defendants had requested during discovery. Dr. McMillan argues that we must review this matter de novo, in light of Jaffee v. Redmond, 518 U.S. 1, 116 S. Ct. 1923, 135 L. Ed. 2d 337 (1996), in which the Supreme Court recognized a federal privilege against the compelled disclosure of counseling records. In the alternative, she argues that the district court erred as a matter of law and abused its discretion in its determination of the privilege issue.

We need not resolve these issues, however, because, as defendants rightly point out, Dr. McMillan's withdrawal of her claim for emotional distress damages constituted a waiver of the claim. This rule holds regardless whether new law has emerged which potentially may bolster a party's argument.

Finally, Dr. McMillan argues that the district court abused its discretion by determining that attorney's fees representing work on Dr. McMillan's contract and emotional distress claims were noncompensable and, accordingly, by reducing the attorney's fee award by twenty percent. She argues that the court was wrong in concluding that the contract claims were severable and not connected to the discrimination claims and thus not compensable. Further, she contends that the claims together represent no more than three percent of the total time documented.

As the district court correctly observed, fee awards are appropriate only for successful claims; unsuccessful claims warrant a fee award only if they are connected to the successful ones. Krewson v. City of Quincy, 74 F.3d 15, 19 (1st Cir.1996). Having determined that Dr. McMillan's unsuccessful emotional distress and contract claims were not connected with her successful claims, the court thought appropriate "some apportionment between successful and unsuccessful claims," and correspondingly reduced the award by twenty percent.

"As a general rule, a fee-awarding court that makes a substantial reduction in either documented time or authenticated rates should offer reasonably explicit findings, for the court, in such circumstances, 'has a burden to spell out the whys and wherefores.' " Brewster v. Dukakis, 3 F.3d 488, 493 (1st Cir.1993) (quoting United States v. Metropolitan Dist. Comm'n, 847 F.2d 12, 18 (1st Cir.1988)). Although the district court did not explicitly state why it regarded the contract claims as unconnected to the discrimination claims, we do not think the court abused its discretion in making that determination and in denying fees for the time spent on that claim,16 or on the emotional distress claim, which is even more clearly unrelated to Dr. McMillan's successful claims.

It is not so easily discernable, however, that twenty percent was a proper reduction. Although the district court has considerable discretion in determining fee awards, and although, as defendants point out, the fee petition is not amenable to easy analysis, our review of the fee petition does not convince us that the time plaintiff's counsel spent on the contract and emotional distress issues could reasonably be regarded as twenty percent of the total time documented.17 In such a situation, the district court should, at a very minimum, make clear why it chose to reduce the fee award by such a substantial percentage. Because the court in this case provided no elaboration for the reduction, we remand this issue to the district court for reconsideration.18

We affirm the district court's denial of judgment as a matter of law on plaintiff's pay discrimination claims and, in part, the jury's award of compensatory damages, and we vacate the jury's award of punitive damages. In addition, we remand for a recalculation of the damages in light of our determination that an amount representing lost benefits should not have been included in the award, and for recalculation of attorney's fees with regard to hourly compensation rates and the fee reduction for time spent on plaintiff's contract and emotional distress claims.

SO ORDERED.

Before TORRUELLA, Chief Judge, GODBOLD* and CYR, Senior Circuit Judges, SELYA, BOUDIN, STAHL, and LYNCH, Circuit Judges.

The panel of judges that rendered the decision in this case having voted to deny the petition for rehearing and the suggestion for rehearing en banc having been carefully considered by the judges of the Court in regular, active service and a majority of said judges not having voted to order that the appeal be heard or reheard by the Court En Banc,

It is ordered that the petition for rehearing and the suggestion for rehearing en banc, be denied.

SELYA, BOUDIN, and LYNCH, Circuit Judges.

On consideration of the petition for rehearing en banc filed by plaintiff and on review of the panel opinion, three of the court's five active judges believe that the viability of Blare, or at least the most common and expansive reading of it, has been cast in doubt by Matthews, contrary to the view expressed in footnote 4 of the panel opinion. In particular, we are far from sure that Massachusetts law now requires a jury to be told that if it finds pretext, it must decide the case in favor of the plaintiff.

However, the defendants have not petitioned for rehearing en banc and there is some question whether the issue was properly preserved in the district court. Accordingly, we go no further than to say that if the issue had been properly preserved and was before the en banc court, we would be inclined to certify this issue to the Massachusetts Supreme Judicial Court because it is a matter of continuing importance.