Full opinion text
OPINION OF THE COURT HUTCHINSON, Circuit Judge. Wolf, Block, Schorr and Solis-Cohen (Wolf) appeals from a judgment of the United States District Court for the Eastern District of Pennsylvania granting relief in favor of Nancy O’Mara Ezold (Ezold) on her claim that Wolf intentionally discriminated against her on the basis of her sex in violation of Title VII of the Civil Rights Act of 1964 (Title VII), 42 U.S.C.A. §§ 2000e to 2000e-17 (West 1981 & Supp. 1992), when it decided not to admit her to the firm’s partnership effective February 1, 1989. At trial Wolf contended that it denied Ezold admission to the partnership because her skills in the category of legal analysis did not meet the firm’s standards. The district court disagreed and found that this articulated reason was a pretext contrived to mask sex discrimination. Wolf argues on appeal that the district court improperly analyzed the evidence before it and that the evidence, properly analyzed, does not support the district court’s ultimate finding of pretext. This case raises important issues that cut across the spectrum of discrimination law. It is also the first in which allegations of discrimination arising from a law firm partnership admission decision require appellate review after trial. Accordingly, we have given it our closest attention and, after an exhaustive examination of the record and analysis of the applicable law, have concluded that the district court made two related errors whose combined effect require us to reverse the judgment in favor of Ezold. The district court first impermis-sibly substituted its own subjective judgment for that of Wolf in determining that Ezold met the firm’s partnership standards. Then, with its view improperly influenced by its own judgment of what Wolf should have done, it failed to see that the evidence could not support a finding that Wolfs decision to deny Ezold admission to the partnership was based upon a sexually discriminatory motive rather than the firm’s assessment of her legal qualifications. Accordingly, we hold not only that the district court analyzed the evidence improperly and that its resulting finding of pretext is clearly erroneous, but also that the evidence, properly analyzed, is insufficient to support that finding and therefore its ultimate conclusion of discrimination cannot stand. We will therefore reverse and remand for entry of judgment in favor of Wolf. This disposition makes it unnecessary to address the issues raised in Wolf’s appeal concerning the remedy the district court awarded to Ezold or those in Ezold’s cross-appeal concerning her claim of constructive discharge. I. Ezold sued Wolf under Title YII alleging that Wolf intentionally discriminated against her because of her sex when it decided not to admit her to the firm’s partnership. She further alleged that she was constructively discharged by reason of the adverse partnership decision. The court bifurcated the issues of liability and damages. After a lengthy bench trial the district court rendered its Findings of Fact and Conclusions of Law on November 29, 1990. See Ezold v. Wolf, Block, Schorr and Solis-Cohen, 751 F.Supp. 1175 (E.D.Pa.1990) (Ezold I). It entered judgment in favor of Ezold on her claim for intentional discrimination and against her on her claim for constructive discharge. The district court held that the nondiscriminatory reason articulated by Wolf for its rejection of Ezold’s candidacy — that her legal analytical ability failed to meet the firm’s partnership standard — was a pretext. It stated: Ms. Ezold has established that the defendant’s purported reasons for its conduct are pretextual. The defendant promoted to partnership men having evaluations substantially the same or inferior to the plaintiff’s, and indeed promoted male associates who the defendant claimed had precisely the lack of analytical or writing ability upon which Wolf, Block purportedly based its decision concerning the plaintiff.... Such differential treatment establishes that the defendant’s reasons were a pretext for discrimination. Id. at 1191-92 (Conclusion of Law (COL) 11). The district court also held that four instances of conduct by Wolf supported its finding of pretext: (1) Ezold was evaluated negatively for being too involved with women’s issues in the firm; (2) a male associate’s sexual harassment of female employees at the firm was seen as “insignificant” and not mentioned to the Associates Committee prior to the partnership decision; (3) Ezold was evaluated negatively for being very demanding, while male associates were evaluated negatively for lacking assertiveness; and (4) Ezold “was the target of several comments demonstrating [Wolf’s] differential treatment of her because she is a woman.” Id. at 1192 (COL 12). In holding that Ezold had failed to establish that she was constructively discharged, the district court stated: A reasonable person in Ms. Ezold’s position would not have deemed her working conditions to be so intolerable as to feel compelled to resign. Id. (COL 16). This holding became relevant to the issue of damages. By way of relief, Ezold sought backpay as well as instatement in the firm as a partner, and if such instatement was impractical, front pay. Wolf argued to the district court that its holding that Ezold was not constructively discharged limited her relief to back pay covering the period from her unlawful denial of admission to the partnership, effective February 1, 1989, until the date of her voluntary resignation from the firm on June 7, 1989. On March 15, '1991, the district court decided that its holding against Ezold on her constructive discharge claim did not preclude her from obtaining relief for the period following her voluntary resignation. See Ezold v. Wolf, Block, Schorr and Solis-Cohen, 758 F.Supp. 303 (E.D.Pa.1991) (Ezold II). The parties then briefed the issue of whether Ezold properly mitigated her damages as required by section 706(g)(1) of Title VII, 42 U.S.C.A. § 2000e-5(g)(l). On July 23, 1991, the district court issued its final memorandum and order. It ruled that Ezold had properly mitigated her damages and that her rejection of Wolfs offer to admit her as a partner as of February 1, 1990 if she accepted responsibility for its domestic relations practice did not toll Wolfs liability for back pay. The court then awarded Ezold back pay in the amount of $131,784.00 for the period from her resignation on June 7, 1989 to January 31, 1991. The parties agreed that if the court’s November 27, 1990 and March 15, 1991 orders were affirmed on appeal, Ezold would be instated as a partner. The court incorporated this agreement into its orders. The district court also awarded Ezold attorney's fees and costs. Wolf timely appealed from the district court’s orders. Ezold filed a protective cross-appeal from the district court’s denial of her constructive discharge claim. II. Ezold was hired by Wolf as an associate on a partnership track in July 1983. She had graduated in the top third of her class from the Villanova University School of Law in 1980 and then worked at two small law firms in Philadelphia. Before entering law school, Ezold had accumulated thirteen years of administrative and legislative experience, first as an assistant to Senator Edmund Muskie, then as contract administrator for the Model Cities Program in Philadelphia, and finally as Administrator of the Office of a Special Prosecutor of the Pennsylvania Department of Justice. Ezold was hired at Wolf by Seymour Kurland, then chairman of the litigation department. The district court found that Kurland told Ezold during an interview that it would not be easy for her at Wolf because "she was a woman, had not attended an Ivy League law school, and had not been on law review.” Ezold I, 751 F.Supp. at 1177 (Finding of Fact (FOF) 18). Subsequent to this meeting, but prior to accepting Wolf's offer of employment, Ezold had lunch with Roberta Liebenberg and Barry Schwartz, both members of the litigation department. She did not ask them anything about the firm’s treatment of women. Ezold was assigned to the firm’s litigation department. From 1983-87, Kurland was responsible for the assignment of work to associates in the department. He often delegated this responsibility to partner Steven Arbittier. As Ezold acknowledged, many partners bypassed the formal assignment procedure and directly assigned matters to associates. The district court found that Arbittier assigned Ezold to actions that were “small” by Wolf standards. Id. at 1178 (FOF 24). Ezold’s performance was reviewed regularly throughout her tenure pursuant to Wolf’s evaluation process, which operates as follows: The Associates Committee, consisting of ten partners representing each of the firm’s departments, first reviews the performance of all the firm’s associates and makes recommendations to the firm’s five-member Executive Committee as to which associates should be admitted to the partnership. The Executive Committee then reviews the partnership recommendations of the Associates Committee and makes its own recommendations to the full partnership. The firm’s voting partners consider only those persons whom the Executive Committee recommends for admission to the partnership. Senior associates within two years of partnership consideration are evaluated annually; non-senior associates are evaluated semi-annually. The firm’s partners are asked to submit written evaluations on standardized forms. The partner is asked the degree of contact he has had with the associate during the evaluation period. Partners were instructed that the evaluations were to be completed regardless of the extent of the evaluating partner’s contact or familiarity with the associate’s work. Ten criteria of legal performance are listed on the forms in the following order: legal analysis, legal writing and drafting, research skills, formal speech, informal speech,' judgment, creativity, negotiating and advocacy, promptness and efficiency. Ten personal characteristics are also listed: reliability, taking and managing responsibility, flexibility, growth potential, attitude, client relationship, client servicing and development, ability under pressure, ability to work independently, and dedication. As stated by Ian Strogatz, Chairman of the Associates Committee: “The normal standards for partnership include as factors for consideration all of the ones ... that are contained [on] our evaluation forms.” Joint Appendix (App.) at 1170. Despite format changes, legal analysis was always listed as the first criterion to be evaluated. This criterion was defined on the evaluation forms used in 1987 and 1988 as the “ability to analyze legal issues; grasp problems; collect, organize and understand complex factual issues.” Id. at 3728. Partners provide grades as well as written comments on these criteria. The evaluation forms describe the grades as follows: —DISTINGUISHED: Outstanding, exceptional; consistently demonstrates extraordinary adeptness and quality; star. —GOOD: Displays particular merit on a consistent basis; effective work product and performance; able; talented. —ACCEPTABLE: Satisfactory; adequate; displays neither particular merit nor any serious defects or omissions; dependable. —MARGINAL: Inconsistent work product and performance; sometimes below the level of what you expect from Associates who are acceptable at this level. —UNACCEPTABLE: Fails to meet minimum standard of quality expected by you of an associate at this level; frequently below level of what you expect. Id. at 3464 (emphasis in original). The form asks the evaluating partner to describe any particular strengths or weaknesses of an associate. Partners are also asked to indicate their views on the admission of each senior associate to the partnership. The evaluation lists five possible responses: “with enthusiasm,” “with favor,” “with mixed emotions,” “with negative feelings” or “no opinion.” Partners are also asked to respond “yes” or “no” to the following question: “I would feel comfortable turning over to this Associate to handle on his/her own a significant matter for one of my clients.” Id. at 3467. Given the number of reviewing partners, the evaluations often contain a wide range of divergent views. These evaluations are then compiled and summarized by the firm’s administrative staff and organized in books for review by the Associates Committee. Ezold I, 751 F.Supp. at 1181 (FOF 52). Each member of the Associates Committee is asked to make an initial assessment of the evaluations pertaining to one of the associates or candidates for partnership. That committee member prepares a form entitled “Committee Member’s Associate Evaluation Summary” summarizing his or her personal view of each associate’s evaluations. This form is colloquially referred to as the “bottom line” memo. As found by the district court, the bottom line memo “is intended to be [the Associates Committee member’s] own personal view of what he has gleaned from the evaluations submitted at the time by the partners who submitted evaluation forms, plus anything in addition that [the Associates Committee member] has gleaned from any interviews that he has conducted with respect to those evaluations.” Id. at 1181 (FOF 53) (emphasis in original). The bottom line memo also contains a “grid” reflecting the Associates Committee member’s summary of the evaluated associate’s grades in legal and personal skills. The bottom line memo also assesses a senior associate’s prospects for regular partnership (Category VI) under the following ratings: “more likely than not,” “unclear,” “less likely than not” or “unlikely.” In 1987 and 1988, similar rankings were used to determine the associate’s potential for special partnership (Category VII). The Category VII partnership then in existence conferred a non-equity “partnership” status upon associates who fell below the normal standard for admission as equity partners but whose work nevertheless was making a valuable contribution to the firm. See id. at 1177 (FOF 15). Each member of the Associates Committee receives copies of the bottom line memo for all associates before meeting formally to discuss evaluations. The bottom line memo serves as a starting point for the Associates Committee’s discussion of each candidate. The Committee members, using both the bottom line memo and the administrative summaries of the grades and comments, engage in a process of weighing and comparing each associate’s legal skills and personal characteristics. The Committee also conducts interviews of those partners who failed to submit written evaluations of an associate during an evaluation period, submitted an evaluation that requires clarification or asked for an opportunity to supplement the written evaluation in an interview. Strogatz testified that the Committee has no formal voting procedure. Id. at 1181 (FOF 57). It ultimately reaches its own consensus as to each senior associate’s partnership potential and as to each associate’s performance. It also formulates a performance review that will be given to each associate and senior associate by a member of the Committee. The firm’s partners evaluated Ezold twice a year as an associate and once a year as a senior associate from October 1983 until the Associates Committee determined that it would not recommend her for partnership in September 1988. The district court found that “[i]n the period up to and including 1988, Ms. Ezold received strongly positive evaluations from almost all of the partners for whom she had done any substantial work.” Id. at 1182 (FOF 60). In making this finding the district court relied on the evaluations of Wolf partners Seymour Kurland, Robert Boote, Steve Goodman, Barry Schwartz, Alan Davis and Raymond Bradley. Ezold’s overall score in legal skills in the 1988 bottom line memorandum before the Associates Committee was a “G” for good. It was noted that “overall” that year she received “stronger grades in intellectual skills than last time.” Id. at 1183 (FOF 71). Evaluations in Ezold’s file not mentioned by the district court show that concerns over Ezold’s legal analytical ability arose early during her tenure at the firm. In an evaluation covering the period from November 1984 through April 1985, Arbittier wrote: I have discussed legal issues with Nancy in connection with [two cases]. I found her analysis to be rather superficial and unfocused. I am beginning to doubt that she has sufficient legal analytical ability to make it with the firm.... She makes a good impression with people, has common sense, and can handle routine matters well. However these traits will take you just so far in our firm. I think that due to the nature of our practice Nancy’s future here is limited. App. at 3392. That same year Schwartz wrote: I have worked a great deal with Nancy since my last evaluation.... Both cases are complex, multifaceted matters that have presented novel issues to us. While her enthusiasm never wanes and she keeps plugging away — I’m often left with a product that demonstrates uncertainty in the analysis of a problem. After extensive discussions with me, the analysis becomes a little more focused, although sometimes I get the sense that Nancy feels adrift and is just marching as best she can to my analytical tune.... In my view her energy, enthusiasm and fearlessness make her a valuable asset to us. While she may not be as bright as some of our best associates, her talents will continue to serve us well. Id. at 3392. Also in 1985, partner Donald Joseph rated Ezold’s legal analytical ability as marginal and wrote “[i]ts [sic] too early to tell but I have been disappointed on her grasp of the problem, let alone performance.” Id. During her next evaluation period from April through November 1985, Ezold received similar negative evaluations. Arbit-tier, Robert Fiebach and Joseph rated her legal analytical abilities as marginal. Ar-bittier wrote: She took a long time getting [a summary judgment brief] done and I found it to be stilted and unimaginative. One of the main issues — dealing with the issue of notice — she missed completely and did not grasp our position_ Also, in considering whether to file a defensive motion ... she failed to cite me to a clause in the agreement that was highly relevant leaving me with the impression that the motion could not succeed. I think Nancy tries hard and can handle relatively straight-forward matters with a degree of maturity and judgment, but when she gets into more complicated areas she lacks real analytical skill and just does what she is told in a mechanical way. She is not up to our minimal Wolf, Block standards. Id. at 3376. Boote made the following report on his performance review with Ezold after this evaluation period: Nancy appeared to accept the judgment, albeit a little grudgingly, that her analytical, research and writing ability was not up to our standards and that she should focus on the types of matters that she can handle effectively.... We made it very clear to Nancy that if she pursues general civil litigation work she is not on track toward partnership and that her only realistic chance for partnership in our opinion is to develop a good reputation for herself in one of the specialized areas of practice. Id. at 3375. In the evaluation period covering November 1985 to April 1986, Boote wrote the following to the Associates Committee: Nancy continues to get mixed reviews. Her pluses are that she is mature, courageous, pretty good on her feet and has the capacity to inspire confidence in clients. Her minuses are that there is doubt about her analytic and writing abil-ity_ In considering Nancy’s prospects for the long range, I think we should bear in mind that we have made mistakes in the past in letting people go to other firms who really could have filled a valuable niche here. Whether Nancy is such a person, of course, remains to be seen. Id. at 3365. A summary of Ezold’s performance review from October 1986 prepared by Schwartz stated: Nancy was advised that several of the lawyers feel she has made very positive progress as a lawyer, Sy [Kurland] being one of them. However, he told her that other lawyers had strong negative sentiments about her capabilities and they feel she has a number of shortcomings in the way of complicated analysis of legal problems and in being able to handle the big complicated corporate litigation, and therefore, does not meet the standard for partnership at Wolf, Block.... Both Sy and I urged Nancy to seriously consider looking for employment elsewhere as she may not be able to turn the tide. Id. at 3364. Although several partners saw improvement in Ezold’s work, negative comments about her analytical ability continued up until, and through, her 1988 senior associate evaluation, the year she was considered for partnership. A summary of her evalúa-tions for 1987 and 1988, focusing on the grades and comments she received in the category of legal analysis, follows: la?' Evaluations Partner Grade Name (Legal Analysis) Comments Promislo M “I had minimal contact with Nancy, but I thought she did not generate ideas ... or pull the facts together well and exercise the best lawyerly judgement. She seemed somewhat over her head, but I don’t think she should have been.” Recommended partnership with “negative feelings.” Id. at 3854-57. Kurland Alderman A “[T]here seems to be serious question as to whether she has the legal ability to take on large matters and handle them on her own. We have been over this many times and there is nothing I can add to what I’ve already said about Nancy. What I envisioned about her when I hired her as a “good, stand-up effective courtroom lawyer” remains to be true and I think she has proven her case. Apparently she has not proved to the satisfaction of the firm the other qualities considered necessary to rise to the top of the firm.” Recommended partnership “with mixed emotions.” Id. at 3878-81. Slight contact. Recommended partnership with “negative feelings.” Id. at 3886-89. Boote A “Nancy has avoided demonstrating ability in th[e] area [of legal analysis] because I believe she lacks it. On the other hand, in Nancy’s case, other qualities redeem her.... I would not want her in charge of a large legally complex case, the traditional measure of a Wolf Block partner.” Recommended partnership “with favor.” Id. at 3894-97. Flaherty A Slight contact. Recommended partnership with “mixed emotions.” Id. at 3918-21. Joseph “I have been singularly unimpressed with the level of her ability.... She may be fine to keep for certain smaller matters, but I don’t see her skills as being those for our sophisticated practice.” Id. at 3930-33. Recommended partnership with “negative feelings.” Id. at 3933. Schwartz G “She is excellent in court and loves to be in that arena.... She remains a little weak in her initial analysis of complex legal issues.” Id. at 3954-56. Dubrow A “[I]n my one experience we lost a client, but I think Nancy performed satisfactorily." No opinion as to partnership admission. Id. at 3990-93. Roberts G Slight contact. Recommended partnership “with favor.” Id. at 4052-55. Spitzer G “Little contact, most favorable impression.” Recommended partnership “with favor.” Id. at 4060-63. During the 1987 evaluation period, two partners viewed Ezold’s eventual admission to the partnership “with enthusiasm,” sixteen “with favor,” eight “with mixed emotions” and seven with “negative feelings.” Id. at 3346. The Associates Committee formed a consensus that Ezold’s analytical ability fell below the firm’s standards. It predicted her partnership chances as “unclear” and if she was made a partner it would most likely be a Category VII partner, because there was substantial doubt as to her legal ability. Id. at 3349. At trial Ezold acknowledged that during her evaluation meeting for this period she was told that “there were partners who criticized [her] writing ability and questioned [her] ability to handle complex litigation, perhaps criticized or questioned [her] ability in the area of legal analysis.” Id. at 666. 1988 Evaluation Partner Grade Name (Legal Analysis) Comments Rosenblum A “On a very complicated matter primarily involving financial analysis, I am not sure whether or not [Ezold] grasped analysis fully. (I am not sure that others working on project did either_).” Recommended partnership with “mixed emotions.” Id. at 3488-91. Temin A Slight contact. Recommended partnership “with mixed emotions.” Id. at 3508-11. Davis A “She will never be a legal scholar — but we have plenty of support in that area.” Recommended partnership with “enthusiasm.” Id. at 3512-15. Arbittier A “Barely adequate legal skills”; “Her abilities are limited. She makes a good impression but she lacks real legal analytical ability.” Recommended partnership with “mixed emotions.” Id. at 3520-23. Fiebach M “Nancy has certain strengths.... If directed, she will do a good job — except that she has limitations with respect to complex legal issues. However, when left on her own she does not do what has to be done until [the] case is in crisis and she does a poor job in keeping [the] client informed.” Recommended partnership with “negative feelings.” Id. at 3544-47. Goldberger Would feel comfortable turning over a significant matter for one of his clients “if not too complex.” “Nancy reputedly can handle many of our matters on her own. If so and reliable others bear these rumors out, partnership may be in the cards.” Recommended partnership with “mixed emotions.” Id. at 3552-55. Joseph “[H]er abilities to grasp legal issues from the little I observed was insufficient to trust her in major litigation on her own.” Recommended partnership with “negative feelings.” Id. at 3560-63. Poul G Slight contact. Recommended partnership “with favor.” Id. at 3580-83. Simon “Probably ancient history — but I do recall my perception that she does not write well and lacks intellectual sophistication.” Recommended partnership with “negative feelings.” Id. at 3596-99. Fala G “Nancy handled a moderate sized lawsuit for a client of mine. Job was done well and responsibly. Result was good.” Id. at 3656. Roberts G Slight contact; recommended partnership with “mixed emotions.” Id. at 3688-91. Garber “Experience with her years ago was unsatisfactory.” No opinion on partnership recommendation. Id. at 3756-59. Berriman G Slight contact; recommended partnership “with enthusiasm.” Id. at 3776-79. Kaplinsky A “She has done a very nice job on the Home Unity shareholder litigation. ... I am probably not as complimentary as Alan [Davis] might be. I was never convinced that she had a complete grasp of the accounting issues in the case.” Recommended partnership “with favor.” Id. at 3452-55. McConomy G "Only worked on one matter for me. She is doing a super job.” Recommended partnership “with favor.” Id. at 3464-67. In 1988, ninety-one partners submitted evaluations of Ezold. Thirty-two, a little more than one-third, made recommendations with varying degrees of confidence, for Ezold’s admission to partnership. Seven of those partners recommended that Ezold be made a partner “with enthusiasm,” fourteen “with favor,” six with “mixed emotions,” four with “negative feelings,” and one with “mixed emotions/negative feelings.” Id. at 3318. Three of the four partners who voted for partnership with negative feelings were members of Ezold’s department. After reviewing Ezold's evaluations and conducting interviews, the Associates Committee voted 9-1 not to recommend Ezold for Category VI partnership. In a discussion initiated by Davis, the Associates Committee also debated modifying the partnership standard as a matter of general policy or specially in Ezold's case because of her other positive attributes. Davis believed: although [Ezold] was not up to par on her legal analytical ability, ... deficiencies in a particular area, even though it was a traditional area where we required a certain superior level, could be overlooked or relaxed to where there were sufficiently compensating skills in other areas, because I felt as chairman that in staffing a case, I could put together the right skills, and we had enough business where we could fit everybody in usefully and productively. Id. at 1665, 1686. He thought the firm "just ought to come off [its] standards and be a little more creative in melding different abilities that different people might bring.” Id. at 1685. The other Committee members ultimately rejected this suggestion. The Associates Committee told Ezold that she would not be recommended for admission as a “Group VI” regular partner effective February 1, 1989 because “too many partners did not believe she had sufficient legal analytical ability to handle complex legal issues.” Ezold I, 751 F.Supp. at 1189 (FOF 136). It did vote, however, to recommend her for the status of “Group VII” special partner that the firm had heretofore made available to associates who are valuable but fall below the firm’s high standards for full partnership. The continuing existence of that category was, however, then under review by the firm’s Executive Committee. It was in fact later eliminated. Out of a total of eight candidates in Ezold’s class, five male associates and one female associate were recommended for regular partnership. One male associate, Associate X, was not recommended for either regular or special partnership. The Executive Committee decided to review the Associates Committee’s negative recommendation of Ezold and also to conduct an independent review of the negative recommendation of Associate X. William Rosoff, former chairman of the Executive Committee, conducted the inquiry. Rosoff reviewed Ezold’s evaluation documents and interviewed four litigation department partners: Schwartz, Boote, Arbittier and Fie-bach. Rosoff had learned of the policy disagreement among some of the firm’s partners as to whether the partnership standard should be relaxed in light of Ezold’s other attributes. He reported to the Executive Committee that it should not recommend Ezold’s admission unless it was prepared to reduce the firm’s partnership standards. The 5-member Executive Committee voted unanimously not to recommend Ezold’s admission as a regular partner. On November 16, 1988 Executive Committee Chairman Charles Kopp met with Ezold and informed her of the decision. He also told her that two domestic relations partners had informed the Committee several days earlier that they were leaving the firm and that this immediate vacuum presented an opportunity for her. Id. at 1189 (FOF 137). He promised that if she agreed to work in this department, she would be made a regular partner in one year. Other associates passed over for partnership in the past had sometimes agreed to specialize in a certain area where the need arose and had later made partner. Although Kopp had little contact with Ezold, he believed that Ezold could handle the work because of the positive evaluations of her skills with clients and in the courtroom and because the practice area did not require the same complex analysis as the firm’s commercial litigation practice. See id. at 1189 (FOF 137-38). Ezold declined the offer. Kopp told Ezold that the firm nevertheless wanted her to stay and she could stay on as a litigation associate as long as she wanted. Id. at 1189 (FOF 139). Ezold also met with Rosoff concerning the domestic relations offer. The district court found that Rosoff “told her that although he could not assure her of a partnership in the future if she declined the domestic relations partnership offer, she would be considered for partnership in the future.” Id. at 1190 (FOF 142). He also told her that she would receive a substantial pay raise the following July when semiannual raises are given to associates, but she would not receive the pay raise being given to the other members of her class who were recommended for partnership. Ezold remained at the firm, none of her cases were taken away from her, and Davis, then chair of the Litigation Department and one of Ezold’s supporters, continued to assign her new cases. On January 25, 1989, one day after the firm’s partners voted on the admission of new partners, Ezold began looking for other employment. She ultimately signed a one-year contract as president of an environmental consulting firm, a former client of Wolf, and also took an “of counsel” position with a suburban law firm. Ezold resigned from the firm on June 7, 1989. III. The district court had subject matter jurisdiction under 28 U.S.C.A. § 1331 (West Supp.1992) and 42 U.S.C.A. § 2000e-5(f)(3) (West 1981). We have jurisdiction over the final orders of the district court pursuant to 28 U.S.C.A. § 1291 (West Supp.1992). IV. Ezold claims Wolf intentionally discriminated against her because of her sex. Intentional discrimination in employment cases fall within one of two categories: “pretext” cases and “mixed-motives” cases. See Price Waterhouse v. Hopkins, 490 U.S. 228, 247 n. 12, 109 S.Ct. 1775, 1789 n. 12, 104 L.Ed.2d 268 (1989) (plurality). In pretext cases, the familiar McDonnell Douglas/Burdine analysis applies. In a mixed motives case the McDonnell Douglas/Burdine analysis does not apply, and the plaintiff has the burden of showing by evidence tied to a discriminatory animus that an illegitimate factor had a “motivating” or “substantial” role in the employment decision. Id. at 258, 109 S.Ct. at 1794. This theory has been codified in the Civil Rights Act of 1991. See 42 U.S.C.A. § 2000e-2(m) (West Supp.1992). If the plaintiff makes such a showing, the burden shifts to the employer to prove by a preponderance of the evidence “that it would have reached the same [employment] decision ... even in the absence of” the impermissible factor. Hopkins, 490 U.S. at 244-45, 109 S.Ct. at 1787-88. There is some uncertainty in the law about the sort of evidence a plaintiff must show to shift the burden to an employer in a mixed motives case, see Tyler v. Bethlehem Steel Corp., 958 F.2d 1176, 1183 (2d Cir.), cert. denied, — U.S. -, 113 S.Ct. 82, 121 L.Ed.2d 46 (1992), but we need not address that issue here as Ezold proceeded only under the McDonnell Douglas/Burdine framework. See Transcript of Oral Argument, at 46-47 (“Your Honor ... I intended to say that this case followed standard McDonnell Douglas and Burdine.... This is a pretext case.”). This is not a mixed-motive case. The issue in this case is “whether illegal or legal motives, but not both, were the ‘true’ motives behind the [partnership] decision.” Lockhart v. Westinghouse Credit Corp., 879 F.2d 43, 49 (3d Cir.1989); see also Price Waterhouse, 490 U.S. at 260, 109 S.Ct. at 1796 (White, J., concurring). Therefore, before considering Wolf's contentions, we think it wise to revisit the alternating burdens of proof in a Title VII discrimination case under the now familiar process set forth in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802, 93 S.Ct. 1817, 1824, 36 L.Ed.2d 668 (1973) and Texas Dep’t of Community Affairs v. Burdine, 450 U.S. 248, 252-56, 101 S.Ct. 1089, 1093-95, 67 L.Ed.2d 207 (1981). See, e.g., Bennun v. Rutgers State Univ., 941 F.2d 154, 170 (3d Cir.1991), cert. denied, — U.S. -, 112 S.Ct. 956, 117 L.Ed.2d 124 (1992); Roebuck v. Drexel Univ., 852 F.2d 715, 726-27 (3d Cir.1988). Ezold relied on this particular method of circumstantial proof of discrimination at trial. The plaintiff must first establish by a preponderance of the evidence a prima facie case of discrimination. Burdine, 450 U.S. at 252, 101 S.Ct. at 1093; Bellissimo v. Westinghouse Elec. Corp., 764 F.2d 175, 179 (3d Cir.1985), cert. denied, 475 U.S. 1035, 106 S.Ct. 1244, 89 L.Ed.2d 353 (1986), abrogated on other grounds, Price Waterhouse v. Hopkins, 490 U.S. 228, 109 S.Ct. 1775, 104 L.Ed.2d 268 (1989). The plaintiff can establish a prima facie case by showing that she is a member of a protected class; that she was qualified for and rejected for the position; and that non-members of the protected class were treated more favorably. Roebuck, 852 F.2d at 726 (citing McDonnell Douglas, 411 U.S. at 802, 93 S.Ct. at 1824); see Burdine, 450 U.S. at 252-53, 101 S.Ct. at 1093-94. After the plaintiff has established a prima facie ease, the burden shifts to the defendant to produce evidence of a legitimate, nondiscriminatory reason for the employee’s rejection. Burdine, 450 U.S. at 252, 101 S.Ct. at 1093; Bellissimo, 764 F.2d at 179. If the defendant’s evidence creates a genuine issue of fact, the presumption of discrimination drops from the ease. Burdine, 450 U.S. at 254-55, 101 S.Ct. at 1094-95; Bellissimo, 764 F.2d at 179. Then, the plaintiff, since she retains the ultimate burden of persuasion, must prove, by a preponderance of the evidence, that the defendant’s proffered reasons were a pretext for discrimination. Burdine, 450 U.S. at 257, 101 S.Ct. at 1095; Chipollini v. Spencer Gifts, Inc., 814 F.2d 893, 898 (3d Cir.) (in banc), cert. denied, 483 U.S. 1052, 108 S.Ct. 26, 97 L.Ed.2d 815 (1987); Bellissimo, 764 F.2d at 180. The parties do not dispute the district court’s conclusion of law that Ezold demonstrated a prima facie case, in particular that she was “qualified” for admission to the partnership. While “more than a denial of promotion as a result of a dispute over qualifications” must be shown to prove pretext, see Molthan v. Temple Univ., 778 F.2d 955, 962 (3d Cir.1985), such a dispute will satisfy the plaintiffs prima facie hurdle of establishing qualification as long as the plaintiff demonstrates that “[s]he was sufficiently qualified to be among those persons from whom a selection, to some extent discretionary, would be made.” Bennun, 941 F.2d at 171 (quoting Roebuck, 852 F.2d at 726). In Title VII cases involving a dispute over “subjective” qualifications, we have recognized that the qualification issue should often be resolved in the second and third stages of the McDonnell Douglas/Burdine analysis, to avoid putting too onerous a burden on the plaintiff in establishing a prima facie case, but we have refused to adopt a blanket rule. Fowle v. C & C Cola, 868 F.2d 59, 64 (3d Cir.1989). Because the prima facie case is easily made out, it is rarely the focus of the ultimate disagreement. Healy v. New York Life Ins. Co., 860 F.2d 1209, 1214 n. 1 (3d Cir.1988), cert. denied, 490 U.S. 1098, 109 S.Ct. 2449, 104 L.Ed.2d 1004 (1989). We agree with the district court’s conclusion that favorable evaluations from partners with whom Ezold worked, and a score of “G” on her 1988 bottom line memo, demonstrate that she was qualified for partnership consideration. See Ezold I, 751 F.Supp. at 1191 (COL 6). The defendant may rebut the presumption of discrimination arising out of the plaintiff’s prima facie case by producing evidence that there was a “legitimate, nondiscriminatory reason” why the plaintiff was rejected. Burdine, 450 U.S. at 254, 101 S.Ct. at 1094. McDonnell Douglas, 411 U.S. at 802, 93 S.Ct. at 1824. The Supreme Court in Burdine said: [T]he defendant must clearly set forth, through the introduction of admissible evidence, the reasons for the plaintiff’s rejection. The explanation provided must be legally sufficient to justify a judgment for the defendant. If the defendant carries this burden of production, the presumption raised by the prima facie case is rebutted, and the factual inquiry proceeds to a new level of specificity. Placing this burden of production on the defendant thus serves simultaneously to meet the plaintiff’s prima fa-cie case by presenting a legitimate reason for the action and to frame the factual issue with sufficient clarity so that the plaintiff will have a full and fair opportunity to demonstrate pretext.... 450 U.S. at 255-56, 101 S.Ct. at 1094-95. The burden then shifts to the plaintiff to show that the defendant’s articulated reasons are pretextual. Id. at 256, 101 S.Ct. at 1095. This burden merges into the plaintiff’s ultimate burden of persuading the court that she has been the victim of intentional discrimination. Id. The plaintiff must demonstrate “by competent evidence that the presumptively valid reason[] for [the alleged unlawful employment action] [was] in fact a coverup for a ... discriminatory decision.” McDonnell Douglas, 411 U.S. at 805, 93 S.Ct. at 1825; Explicit evidence of discrimination — i.e., the “smoking gun” — is not required. See Bennun, 941 F.2d at 171; Lockhart v. Westinghouse Credit Corp., 879 F.2d 43, 48 (3d Cir.1989). A plaintiff can establish pretext in one of two ways: “either directly by persuading the court that a discriminatory reason more likely motivated the employer or indirectly by showing that the employer’s proffered reason is unworthy of credence.” Burdine, 450 U.S. at 256, 101 S.Ct. at 1095. In proving that the employer’s motive was more likely than not the product of a discriminatory reason instead of the articulated legitimate reason, sufficiently strong evidence of an employer’s past treatment of the plaintiff may suffice. See Patterson v. McLean Credit Union, 491 U.S. 164, 188, 109 S.Ct. 2363, 2378, 105 L.Ed.2d 132 (1989); McDonnell Douglas, 411 U.S. at 804, 93 S.Ct. at 1825. The employer’s “general policy and practice with respect to minority employment” may also be relevant. McDonnell Douglas, 411 U.S. at 804-05, 93 S.Ct. at 1825-26. Alternately, if a plaintiff produces credible evidence that it is more likely than not that “the employer did not act for its proffered reason, then the employer’s decision remains unexplained and the inferences from the evidence produced by the plaintiff may be sufficient to prove the ultimate fact of discriminatory intent.” Chipollini, 814 F.2d at 899. Wolf’s articulated nondiscriminatory reason for denying Ezold’s admission to the partnership was that she did not possess sufficient legal analytical skills to handle the responsibilities of partner in the firm’s complex litigation practice. Ezold attempted to prove that Wolf’s proffered explanation was “unworthy of credence” by showing she was at least equal to, if not more qualified than, similarly situated males promoted to partnership. She also contended that her past treatment at the firm showed Wolf’s decision was based on a discriminatory motive rather than the legitimate reason of deficiency in legal analytical ability that the firm had articulated. V. From this overview of the law, we turn to the specifics of the district court’s analysis, its findings and the parties’ contentions concerning them. The district court compared Ezold to eight successful male partnership candidates, Associates A-H. It found: The test that was put to the plaintiff by the Associates Committee that she have outstanding academic credentials and that before she could be admitted to the most junior of partnerships, she must demonstrate that she had the analytical ability to handle the most complex litigation was not the test required of male associates. Ezold I, 751 F.Supp. at 1183 (FOF 73). The district court then concluded: Ms. Ezold has established that the defendant’s purported reasons for its conduct are pretextual. The defendant promoted to partnership men having evaluations substantially the same or inferior to the plaintiff’s, and indeed promoted male associates who the defendant claimed had precisely the lack of analytical or writing ability upon which Wolf, Block purportedly based its decision concerning the plaintiff. The defendant is not entitled to apply its standards in a more “severe” fashion to female associates.... Such differential treatment establishes that the defendant’s reasons were a pretext for discrimination. Id. at 1191-92 (COL 11) (citations omitted). Wolf says this finding of pretext is wrong. Analyzing its contentions, we perceive two reasons why this is so. First, the district court’s finding that Ezold was required to have outstanding academic credentials before she could be admitted to partnership is without factual support in the record. The only evidence in the record that Wolf considered Ezold’s academic record is limited to the original decision to hire Ezold and to assignments given to Ezold early in her employment with Wolf, issues we consider in Part IX,' infra. Second, in its analysis, the district court did not focus on Wolf’s articulated reason for denying Ezold partnership — lack of analytic ability to handle complex litigation. Instead, the district court first substituted its own general standard for the qualities Wolf believed were essential to law firm partnership. Then, applying its own incorrect standard of comparison, the district court did not realize that a comparison of Ezold’s legal analytic ability with that of the sue-cessful males could not support a finding of pretext. Overall, Ezold's evaluations in that category were not as good as that of even the least capable male associate who was offered a partnership position. VI. Wolf contends that in all aspects of its analysis the district court improperly substituted its own subjective judgment—not only concerning what the firm’s partnership standard should be—but also concerning whether Ezold met this standard. Specifically, it alleges that the district court ignored the negative evaluations concerning Ezold’s legal analytical ability that are in the record; looked beyond the criterion of legal analysis, Wolf’s articulated nondiscriminatory reason, in comparing Ezold to male associates admitted to the partnership; failed to make findings concerning male associates denied admission to the partnership based on their deficient legal analytical ability; and excluded from evidence the evaluation files of female associates admitted to the partnership who received criticisms similar to male associates admitted to the partnership in areas other than legal analysis. Initially, Wolf argues our review of these issues is plenary. Wolf relies on Logue v. International Rehabilitation Associates, Inc., 837 F.2d 150 (3d Cir.1988), for the proposition that we exercise plenary review over the district court’s determinations on these questions. Ezold responds that Wolf is trying to obtain plenary review by couching a challenge to the sufficiency of the evidence as legal error in the selection of the appropriate standards for determining discrimination. In Logue the defendant asserted on appeal that the district court incorrectly applied the legal standard for sex discrimination by failing to address and make findings of fact on all of the legitimate, nondis-eriminatory reasons it offered in support of its termination of the plaintiff’s employment. Id. at 153. We held that by failing to address all of the defendant’s proffered reasons the district court erred as a matter of law, misapplied the legal standard governing sex discrimination and deprived the defendant of the full trial process contemplated by Burdine. Id. at 154. This case is distinguishable from Logue. Here, the district court did consider Wolf’s articulated nondiscriminatory reason and did make findings upon it. Wolf contends the district court’s findings are incomplete and that those it did make do not support its ultimate finding of pretext. Plenary review is appropriate in order to determine the extent to which essential findings are missing. The district court’s refusal to credit or make findings concerning all of Wolf’s proffered evidence, however, does not subject its express findings to plenary review. Those findings cannot be set aside unless they are clearly erroneous. A finding becomes clearly erroneous “when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.” Anderson v. City of Bessemer City, 470 U.S. 564, 573, 105 S.Ct. 1504, 1511, 84 L.Ed.2d 518 (1985) (quoting United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948)). When there are two permissible views of the evidence, the district court’s choice of one view cannot be clearly erroneous. Id. at 574, 105 S.Ct. at 1511. The district court’s resolution of the ultimate issue whether Wolf’s reason for denying Ezold’s admission to the partnership was a pretext is a finding of fact subject to the clearly erroneous standard set forth in Federal Rule of Civil Procedure 52(a). See id. at 573, 105 S.Ct. at 1511; Bellissimo, 764 F.2d at 179. We may reverse the district court on this finding of fact only if the evidence is insufficient to permit a rational factfinder to infer that Wolf’s assertion that Ezold was wanting in legal analytic ability was a mask for unlawful sex discrimination. Wolf’s disagreement with the method of analysis the district court employed leads naturally to its challenge to the sufficiency of the evidence to support the district court’s finding of pretext and its ultimate conclusion of unlawful discrimination. Thus, Wolf contended at oral argument before this Court: “[tjhere is no proof, in this case, of a gender-driven result." Transcript at 59. In considering a challenge to the sufficiency of the evidence, we must determine based on our own “comprehensive review of the entire record” whether Ezold has satisfied her ultimate burden of proving intentional' sex discrimination. Bennun, 941 F.2d at 170; Bellissimo, 764 F.2d at 178-79. In doing so, we view the evidence in the light most favorable to Ezold. See Roebuck, 852 F.2d at 727-28 (citing Dreyer v. Arco Chem. Co., 801 F.2d 651, 654 (3d Cir.1986), cert. denied, 480 U.S. 906, 107 S.Ct. 1348, 94 L.Ed.2d 519 (1987)). We again defer to the district court’s factual findings, including once more its ultimate finding, and we cannot reverse any of them unless they are clearly erroneous. Bellissimo, 764 F.2d at 178-79. VII. Wolfs articulated reason for refusing to offer Ezold a partnership was its belief, based on a subtle and subjective consensus among the partners, that she did not possess sufficient legal analytic ability to handle complex litigation. Wolf never contended that Ezold was not a good courtroom lawyer, dedicated to her practice, and good with clients. Instead, many partners felt, because of the level of her legal analytical ability, that she could not handle partnership responsibilities in the firm’s complex litigation practice. Absent evidence to show that legal analytic ability was not a necessary precondition for partnership at Wolf, the district court’s opinion about Ezold’s comparative strengths in the other categories on the evaluation form is immaterial. A. The record does not show that anyone was taken into the partnership without serious consideration of their strength in the category of legal analytic ability. The evaluations specifically asked each partner whether he or she would feel comfortable turning over to the partnership candidate “to handle on his/her own a significant matter for one of my clients.” See App. at 3423. Several of the partners’ responses to this question on Ezold’s evaluations show clear concern about the depth of her legal analytical capabilities. See, e.g., App. at 3348 (“I would not want her in charge of a large legally complex case, the traditional measure of a Wolf, Block partner.”). This same question, reflecting a requirement that an applicant exhibit analytical abilities sufficient to meet Wolf’s perception of the firm’s standard, was considered throughout the firm’s evaluations of the male associates with whom Ezold was competing. See, e.g., App. at 4257 (“I just am concerned if he could ‘first chair’ a case.”); App. at 4823 (“He [Associate H] can handle the most complex litigation we have.”); App. at 4532 (“Based on [Associate C’s] ability to analyze a legal problem I could feel comfortable in turning over my best client to him for a significant matter.”); App. at 5044 (“[There are] questions about his intellectual strength, his ability to manage complex transactions and his level of attention to detail”); App. at 4696 (“[H]e just doesn’t have the high level of intelligence we need to handle complex legal questions.”). Ezold herself acknowledged at trial that because of the nature of Wolf’s litigation practice, its litigators devote much more time to legal analysis than in-court trial work. Davis, a member of the Associates Committee who favored partnership for Ezold, testified that he recognized her shortcomings in the area of legal analytic ability. Thus, he advocated a relaxation of the partnership standard to accommodate her because he believed that her other skills “outweighed whatever deficiencies she had in the legal ability area.” App. at 1684. The Associates Committee and the Executive Committee ultimately refused to relax the firm’s standards. Such a refusal to relax standards, however, is not evidence of discrimination. Wolf reserves for itself the power to decide, by consensus, whether an associate possesses sufficient analytical ability to handle complex matters independently after becoming a partner. It is Wolf’s prerogative to utilize such a standard. In Billet v. CIGNA Corp., 940 F.2d 812 (3d Cir.1991), an age discrimination case, we stated that “[bjarring discrimination, a company has the right to make business judgments on employee status, particularly when the decision involves subjective factors deemed essential to certain positions.” Id. at 825. We stated again that “[a] plaintiff has the burden of casting doubt on an employer’s articulated reasons for an employment decision. Without some evidence to cast this doubt, this Court will not interfere in an otherwise valid management decision.” Id. at 828 (citing Lucas v. Dover Corp., 857 F.2d 1397, 1403-04 (10th Cir.1988) (a court will not second guess business decisions made by employers, in the absence of some evidence of impermissible motives)); see Loeb v. Textron, Inc., 600 F.2d 1003, 1012 n. 6 (1st Cir.1979) (“While an employer’s judgment or course of action may seem poor or erroneous to outsiders, the relevant question is simply whether the given reason was a pretext for illegal discrimination.”). The partnership evaluation process at Wolf, though formalized, is based on judgment, like most decisions in human institutions. A consensus as to that judgment is the end result of Wolf’s formal process. In that process, the Associates Committee has the role of collecting and weighing hundreds of evaluations by partners with diverse views before reaching its consensus as to a particular associate’s abilities. The consensus the Associates Committee reaches is then passed on to the Executive Committee. After its review and, at least in Ezold’s case, additional independent investigation, the Executive Committee submits its final recommendation to the partners for a vote. The differing evaluations the partners first submit to the Associates Committee are often based on hearsay or reputation. No precise theorem or specific objective criterion is employed. Cf. Bennun, 941 F.2d at 179 (not “unwarranted invasion” of college’s tenure process for district court “to determine that [professor] was held to higher standards in objective terms, i.e. number of publications”) (emphasis added). We have cautioned courts on several occasions to avoid unnecessary intrusion into subjective promotion decisions in the analogous context of academic tenure. While such decisions are not insulated from judicial review for unlawful discrimination, it is clear that courts must be vigilant not to intrude into that determination, and should not substitute their judgment for that of the college with respect to the qualifications of faculty members for promotion and tenure. Determinations about such matters as teaching ability, research scholarship, and professional stature are subjective, and unless they can be shown to have been used as the mechanism to obscure discrimination, they must be left for evaluation by the professionals.... Id. at 181 (Sloviter, C.J., dissenting from denial of petition for rehearing) (quoting Kunda v. Muhlenberg College, 621 F.2d 532, 548 (3d Cir.1980)). These cautions against “unwarranted invasion or intrusion” into matters involving professional judgments about an employee’s qualifications for promotion within a profession inform the remainder of our analysis. B. In Ezold’s case, the district court correctly recognized the legal premise that should have governed its result: Title VII prohibits only “discrimination.” Therefore, “consideration of the practices of the [firm] toward the plaintiff must be evaluated in light of its practices toward the allegedly more favored group, in this case males.” Kunda, 621 F.2d at 538. The district court, however, failed to apply this legal premise to the evidence before it. It disagreed not only with Wolf’s assessment of Ezold’s ability to meet Wolf’s standards, but also with Wolf’s partnership standards themselves. For example, it found: In the magnitude of its complexity, a case may have a senior partner, a younger partner, and an associate(s) assigned to a case. Accordingly, requiring the plaintiff to have the ability to handle on her own any complex litigation within the firm before she was eligible to be a partner was a pretext. Ezold I, 751 F.Supp. at 1188 (FOF 121). The district court disagreed with Wolfs decision not to overlook Ezold’s deficiency in legal analysis because of her other skills and attributes, but the court is not a member of Wolfs Associates Committee or Executive Committee. Its belief that Wolfs high standard of analytical ability was unwise in light of the staffing of senior partners on complex cases does not make Wolfs standard a pretext for discrimination. The evaluations that the district court did rely upon in making its finding of pretext praised Ezold for skills other than legal analysis, such as client relations and ability in court, that Wolf never disputed she possessed. Where an employer produces evidence that the plaintiff was not promoted because of its view that the plaintiff lacked a particular qualification the employer deemed essential to the position sought, a district court should focus on the qualification the employer found lacking in determining whether non-members of the protected class were treated more favorably. Without such a limitation, district courts would be routinely called upon to act as members of an employer’s promotion board or committee. It would subjectively consider and weigh all the factors the employer uses in reaching a decision on promotion and then make its own decision without the intimate knowledge of the history of the employer and its standards that the firm’s decisionmakers use in judging the degree to which a candidate exhibits a particular qualification that the employer has decided is of significance or primary importance in its promotion process. Pretext is not established by virtue of the fact that an employee has received some favorable comments in some categories or has, in the past, received some good evaluations. See, e.g., Billet, 940 F.2d at 826; Turner v. Schering-Plough Corp., 901 F.2d 335, 343-44 (3d Cir.1990); Healy, 860 F.2d at 1215; see also Frieze v. Boatmen’s Bank, 950 F.2d 538, 541 (8th Cir.1991) (“An employer rating an employee as competent discredits the employer’s stated reason for discharging the employee, however, only when the employer’s stated reason is the employee’s general incompetence.”) (emphasis added). It was not for the district court to determine that Ezold’s skills in areas other than legal analysis made her sufficiently qualified for admission to the partnership. The district court’s method of comparing Ezold to “similarly situated” male associates admitted to the partnership points up this initial flaw in its analysis. It engaged in a “pick and choose” selection of various comments concerning the male associates’ personalities, work habits, and other criteria besides legal analy