Finch v. Hercules Inc.

OPINION MURRAY M. SCHWARTZ, Senior District Judge. Plaintiff David G. Finch has filed suit against defendant Hercules Incorporated, alleging he was discriminated against because of his age in violation of the Age Discrimination in Employment Act [“ADEA”], 29 U.S.C. § 621-34. 1 Finch contends Hercules discriminated against him when it terminated his position following a reduction in force at corporate headquarters. In response to Finch’s claim of disparate treatment under the ADEA, Hercules moved for summary judgment. Docket Item [“D.I.”] 117. Prior to oral argument on Hercules’ motion for summary judgment, Finch moved the Court allow him to amend his complaint to allege a claim of disparate impact under the ADEA. D.I. 147. The Court reserved decision on Hercules’ motion for summary judgment, but granted Finch leave to amend the complaint and extended the discovery deadline, solely as to the issue of disparate impact. Hercules then moved for partial summary judgment on Finch’s claim of disparate impact. D.I. 166. On August 18, 1994, the Court entertained oral argument on Hercules’ motion. For the reasons which follow, the Court will deny Hercules’ motion for summary judgment on Finch’s disparate treatment claim and grant Hercules’ motion for summary judgment on Finch’s claim of disparate impact. I. FACTUAL BACKGROUND Hercules hired Finch in 1962 as a Systems Analyst. In 1974, he became Financial Director of Hercules’ Organics Department and was later promoted to manager of the financial analysts for all Hercules’ business groups. D.I. 121 at B45-50. From 1980 until his termination, Finch held the position of General Auditor at Hercules. D.I. 119 at A102. As General Auditor, Finch reported to the Audit Committee 2 of the Board of Directors and Arden B. Engebretsen, Hercules’ Chief Financial Officer. D.I. 121 at B50-51. In the mid 1980’s, responding to its declining economic performance, Hercules began a program of restructuring, including several employee reductions in force [“RIFs”]. D.I. 119 at A15. Hercules lowered its total employment from approximately 25,450 employees in January 1985 to approximately 17,300 employees in January 1991. D.I. 121 at B388-89. In 1988, David Hollingsworth, Hercules’ Chief Executive Officer [“CEO”] at that time, appointed Vice President of Operations Support James D. Beach, Jr. to address a perceived staff imbalance between Hercules’ headquarters and its other locations. D.I. 119 at A227, A270-71. Beach hired the accounting firm of Coopers & Lybrand [“C & L”], Hercules’ outside auditors, to study staff functions and make recommendations for improvements. Beach then hired Thomas S. Litras to help implement C & L’s recommendations, the Indirect Productivity Improvement [“IPI”] study. Id. at All, A203-04, A480-84. Among other things, the IPI study indicated Hercules needed a more proactive, involved audit department. 3 Id. at A485-86. *1111 Titan Missile Problems In late 1989, Hercules was forced to take a $300 million write-off because of a cost overrun at its Titan IV missile plant in Bacchus, Utah. Id. at A55-56, A322. The costs for the Titan were spread over two contracts: a quality development contract which Hercules had, and a production contract it hoped to secure. D.I. 121 at B42-43. According to Arthur C. Nielsen, Jr., the former Chairman of Hercules’ Audit Committee, this accounting treatment was proper, so long as Hercules would realize a profit from the two contracts together. Id. at B139-43. Likewise, both Engebretsen and Francis J. Van Kirk of C & L stated internal audit could not have discovered the Titan problem earlier and was not responsible for the write-off. Id. at B37, B165, B446-61. Defendant, however, contends this write-off led to the first annual loss in Hercules’ history and created a “great uproar” among members of the Audit Committee. D.I. 118 at 9; D.I. 119 at A56-57, A408-09. The Titan cost overrun led to heightened scrutiny of financial areas of the company, with some members of Hercules’ Board of Directors concluding Hercules needed higher quality senior management in internal audit. D.I. 119 at A57-59, A64-65, A408-09. The 1991 Reduction In Force In September 1990, Hercules determined it needed to make additional reductions in its workforce to lower its indirect costs. D.I. 121 at B310. As part of a planned 1991 RIF, Litras conducted a demographic study of Hercules’ corporate headquarters. This study compared the number of employees before the RIF who were under forty years of age and over forty years of age, and the number under fifty and over fifty, with the projected numbers after the proposed RIF. Id. at B276-77. In implementing the RIF, Hercules created a policy designed to retain better employees and avoid discrimination. Hercules also created a Policy Compliance Committee [“PCC”], which was intended to assure both compliance with the law and the retention of qualified personnel. D.I. 119 at A23, A61, A165-67, A193-94, A475. The PCC had authority to approve all RIF decisions, but could not order corporate restructuring or elimination of positions. D.I. 119 at A22, A471; D.I. 121 at B305. Litras also helped Hercules identify positions which could be eliminated and met with the managers of each Hercules’ unit to agree upon a number of persons to be eliminated through the RIF. D.I. 119 at A230. He then published a list for each department of the agreed number, the number he recommended, and a “stretch number” that Litras thought was the outside limit for a feasible RIF. Id. at A235-A235-1. Litras contended the performance appraisal process then in use at Hercules was not helpful in determining which individuals should be terminated through the proposed RIF. D.I. 121 at B313. As a result, he proposed a process of forced ranking and paired comparison by which employees would be grouped into functional categories and then ranked against each other. 4 Starting from the bottom of the list, Hercules would terminate lower ranked employees in the RIF. Id. at B9, B369-71. In ranking employees, Litras proposed considering “performance, education, versatility, flexibility, and continuous service. Age to be the last factor considered, and then only if a ‘tie-breaker’ is required.” D.I. 119 at A488 (emphasis in *1112 original). If there was a tie, Hercules was to retain the older employee. Id. at A217-18. Job performance was to be measured through the performance appraisal system and dialogue with the supervisor. D.I. 121 at B6. Gossage Becomes Hercules’ Chief Executive Officer (CEO) In late 1990, CEO David Hollingsworth announced his plans to retire. In its search for a replacement, Hercules’ Board of Directors interviewed Thomas L. Gossage, the president of a Hercules’ subsidiary, Aqualon. D.I. 119 at A263, A398. During the interview process, Gossage informed the Board that the financial side of the company, including the Audit Department, was weak and that as CEO, he would downsize and cut expenses. Id. at A64, A268-69, A292, A296. After being selected as Hercules’ new CEO, Gossage instructed Beach to proceed with the planned RIF based on Litras’ stretch number. Id. at A14-15, A270. Finch’s position, General Auditor, was not identified in Litras’ January 4,1991 report as a position to be eliminated in the RIF. 5 D.I. 121 at B312-15. Gossage asserts that other than approving the adoption of the stretch number, he had minimal involvement in the RIF. D.I. 119 at A273; see D.I. 159 Ex. 1 at 60 (other than approving its costs and terms, Gossage was not involved in the RIF process). In December 1990, Hercules placed the estimated cost of the RIF, $17,000,000, on its corporate books. Id. at A272, A298. Plaintiff has proffered statements Gossage made during an interview with a local newspaper as evidence of age bias at Hercules. The article quotes Gossage as saying: The young people in the company want us to bring Hercules back to where it ought to be again.... Older people will see friends impacted and will feel bad about it. But we’ll get this behind us. Id. at A539; D.I. 121 at B327. Gossage subsequently stated that the article “captures the spirit of what I said, at least, if it’s not a direct quote.” D.I. 121 at B71. Likewise, at a Hercules’ question and answer session, an employee asked Gossage to explain his statement to the local newspaper. He responded that: What I said was that younger employees want to know when the company will get up and turn itself around, and I acknowledged that to those with long service to the company, it was painful to watch what was going on. I’d say it again. Id. at B288. Defendant contends plaintiff has taken these statements out of context because they refer to Hercules’ voluntary early retirement program. The reference to “older people,” according to defendant, merely indicates more senior employees would qualify for the early retirement program. D.I. 118 at 47. Promotions and Succession Planning To further support his claim that Gossage was motivated by age bias, Finch has proffered deposition testimony of William E. Hosker and C. Doyle Miller. In January 1992, Hosker, the head of Hercules’ Resins Group, asked to meet with Gossage to find out why he had not been promoted to Group Vice President. D.I. 158 Exhibit [“Ex.”] 1 at 19. According to Hosker, Gossage informed him that he had promised the Board of Directors that Hercules would have a cadre of trained people in their early 50s as CEO candidates by the time Gossage retired. 6 Hosker also stated that Gossage informed him that although Hosker was perhaps a better candidate than some of the persons selected for promotion to Group Vice President, Hosker was disadvantaged because of Gossage’s commitment to the Board and the fact Hosker was already 54 years old. Id. at 20-21. Gossage allegedly explained that “someone in their late 50s typically looks, expends their energies in preparing their retirement.... [while] younger people have perhaps more energy and a longer period of *1113 time in which they can perform their duties.” Id. at 27. Finch has also proffered testimony from Hosker’s deposition in which Hosker relates statements that Gossage made to C. Doyle Miller. Hosker testified that during a May 1992 meeting, Miller relayed Gossage’s comments that the Chemical Specialties Group had not aggressively or creatively addressed the problem of “tired warriors.” Id. at 39-40. According to Hosker, Miller stated that tired warriors referred to Hercules’ employees who are in their mid-fifties, contributing to the corporation and who want to work until age 65, “but in so doing are blocking the movement of younger, high potential people into development positions in the corporation.” Id. at 40. In his deposition, Miller testified he did not recall Gossage ever using the phrase “tired warriors,” but acknowledged Gossage was concerned that there were people in Hercules’ management who lacked leadership capability, foresight and “enthusiasm to aggressively lead the business where they needed to go.” D.I. 158 Ex. 3 at 48-49. Miller denied that Gossage ever suggested Hercules needed younger employees. Id. at 50. Hosker also testified that other than his own experience, he was not aware of any Hercules’ employment decisions being made based on age. D.I. 159 Ex. 1 at 67-68. Rather, Hosker asserted that based on his role in the 1991 RIF in the Resins Group, Hercules based its RIF decisions on performance, not age. Id. at 60-61. Finch’s Termination Shortly after assuming responsibilities as CEO, Gossage asked for the resignation of then Vice Chairman of the Board and Chief Financial Officer [“CFO”] Arden Engebret-sen. D.I. 119 at A265-66, A398, A416-17. As part of the search for a new CFO, Gos-sage met with Vice President and Controller George MacKenzie, Treasurer Alex Searl, and Thomas Ciconte, Treasurer of Aqualon. On January 14,1991, Gossage sent a memorandum to MacKenzie and Searl directing that during the search for a new CFO, the Audit Department and the Trading Company would report to MacKenzie, and Hercules’ economist would report to Searl. Id. at A501. Plaintiff asserts that at the time Gos-sage had the Audit Department report to MacKenzie, Gossage was unaware of any deficiencies in Finch’s performance as General Auditor. D.I. 121 at B80-81. Defendant, however, maintains Gossage was aware that the Audit Department and Finch had reputations for poor performance. Id.; D.I. 119 at A416-17. The same day Gossage ordered the new reporting system, MacKenzie met with Gossage and recommended the General Auditor position be eliminated on an interim basis. D.I. 119 at A295, A332, A338, A344. After meeting with Gossage, MacKenzie arranged to meet with Audit Committee Chairperson Joan E. Spero to seek her approval. Id. at A321. MacKenzie informed Spero that Finch’s position as General Auditor would be eliminated and that he would assume those duties until a new CFO was selected. Id. at A343-44, A405-07. After Gossage and Spero approved the plan, MacKenzie included Finch in his forced ranking of Financial Managers. Id. at A334, A367-69; D.I. 121 at B128-29. Plaintiff notes that MacKenzie had supervised Finch for only two days at the time he made his ranking decision and did not consider Finch’s personnel file or seek the opinion of Engebretsen. D.I. 121 at B104-05. Defendant, however, contends MacKenzie had long worked with Finch and the Audit Department and regularly attended Audit Committee meetings. D.I. 135 at C2-3. Of seven financial managers MacKenzie ranked, Finch was sixth. D.I. 119 at A502. 7 The two highest ranked managers were 51 and 55 years old. Id. On January 26, 1991, MacKenzie met with the PCC, which approved his recommendations. Id. at A24, A472-73. Plaintiff asserts, however, that this did not constitute meaningful review because the PCC had no authority to review organization changes such as eliminating the position of General Audi *1114 tor; its authority was limited to terminations based upon forced ranking decisions. Id. at A22, A471; D.I. 121 at B187, B305. While the decision does not appear in the minutes of the Audit Committee’s January 29, 1991 meeting, defendant contends the Audit Committee also approved MacKenzie’s recommendations. D.I. 119 at A70. Shortly thereafter, on February 2, Mae-Kenzie and James J. Woods, the Director of Operational Reporting at Hercules, informed Finch his position was being eliminated. Id. at A453-56. According to plaintiff, neither MacKenzie nor Woods suggested he was being terminated for poor performance. D.I. 121 at B54-55, B183-84. Further, plaintiff has offered evidence which suggests that at the time he was terminated, the elimination of the General Auditor position was considered permanent. Id. at B13-15; B382. Defendant disputes this evidence, asserting that MacKenzie’s organizational chart indicated that the arrangement was “interim” and that Finch was aware of this fact. D.I. 135 at C47A. Defendant also asserts Finch knew his termination was performance related because he was aware that he had been ranked sixth of seven. Shortly after Hercules decided to terminate Finch, Gossage selected R. Keith Elliott to replace Engebretsen as CFO in March 1991. In May 1991, Elliott talked to Mac-Kenzie about becoming Treasurer and decided that Hercules should hire a Certified Public Accountant [“CPA”] for the now to be reinstituted General Auditor position. D.I. 119 at A33-36, A168-69, A324, A373. Hercules retained an executive search firm to find a new General Auditor. Id. at A43-44, A168-69. While Finch allegedly was considered, Elliott determined Finch should not be interviewed because he was not a CPA and because of his poor performance. Id. at A43-44, A173-79, A180-82, A187. Hercules hired Curtis Tomlin, a 38 year old CPA with an MBA from Harvard University, to be its new General Auditor. Id. at A515-17. Finch’s Job Performance Defendant has placed extensive evidence in the record which tends to question the quality of Finch’s work performance. First, the 1989 IPI study commented that Hercules needed a more proactive and involved audit department. Id. at A485-86. 8 Likewise, C & L’s auditors criticized the timeliness of internal audit reports. Id. at A438, A441. MacKenzie criticized the work of Finch and the Audit Department and noted that one afternoon he found Finch sleeping in his office. Id. at A326-27, A346-52. Further, James Woods, a member of the Controller’s Department who often worked with Finch, stated Finch’s job performance was satisfactory, but felt Finch was too passive. Id. at A465-66. Fred Laquinta and Tom McCarthy, both of Human Resources, also questioned Finch’s ability and that of the Audit Department. Id. at A195-97, A374. Spero, Chairperson of the Audit Committee, stated that she and the other members of the Audit Committee felt Finch was not the type of innovative, proactive person needed in the position of General Auditor. Id. at A418, A424, A433-35; see id. at A65. Some members of the Audit Committee also maintained that Finch and the Audit Department were responsible for the write off of $300 million at the Titan missile plant. Finally, at about the time of the RIF process, Hercules was reviewing personnel at headquarters to determine the amount of incentive bonuses for senior staff employees. Id. at A152, A289-2. The Audit Department was ranked among the three lowest departments at Hercules. Id. at A157, A200. Audit was informed it needed to improve timeliness of audits, risk assessments, and make audits less confrontational. 9 Id. at A188-91, A362-63, A510. *1115 Plaintiff, however, has placed evidence in the record indicating that he was a “very good” employee with long-range potential to be a member of senior management. D.I. 121 at B405-29. Finch received above average salary increases, which also suggests above average performance. Id. at Bl, B20-22. Likewise, Engebretsen, who was Finch’s immediate supervisor for most of his time as General Auditor, stated: Mr. Finch is an effective leader, communicates well, is quiet but decisive, shows strong decision-making capacity and good judgment. He is also a hard worker, and understands the needs of his employes, [sic] Id. at B431. Engebretsen reiterated this view in his statement to the EEOC, noting that he “did not recall hearing any criticism or correction of [Finch’s] work by the Audit Committee, and I was always present at these meetings and discussed Mr. Finch’s performance with them.” Id. at B432-33. Engebretsen also stated he thought Finch was a proactive auditor and gave him credit for expanding Hercules’ global auditing, developing standards for computer auditing and developing risk analysis. 10 Id. at B219-20. Further, Engebretsen stated he was aware of friction between Finch and MacKenzie, attributing it to Finch’s action of pointing out deficiencies in MacKenzie’s performance as Controller. Id. at B32-33. Finally, Nielsen, the former chairperson of the Audit Committee, remarked that he had “high regard for [Finch] as a professional in this field.” Id. at B135; see id. at B230. Like Engebretsen, Nielsen could not recall anyone criticizing Finch’s abilities as General Auditor. Id. at B135-37. Lies, Damned Lies, and Statistics 11 On March 10, 1994, Finch amended his complaint to allege a claim of disparate impact based on Hercules’ RIF process. D.I. 162. The RIF process consisted of several steps. First, Hercules set reduction targets and developed position or functional groupings for implementing forced rankings. Next, Hercules forced ranked persons within each group. RIF recommendations were made at the department level, with lower ranked employees subject to termination. Finally, the PCC either accepted, rejected or modified the department recommendations. D.I. 168 at A7-13, A36-38. Bernard R. Siskin, Ph.D., Hercules’ expert, completed a statistical analysis of the effects of the RIF. By affidavit dated April 29, 1994, Siskin stated that the “bottom-line” results of the RIF indicated that employees ages 40 and over “had a lower likelihood of involuntary termination than did employees less than 40 years of age.” Id. at Al-2 (emphasis in original). Siskin, therefore, concluded the data were inconsistent with Finch’s claim that the 1991 RIF had an adverse impact on age-protected employees. Id. at A2. Finch’s statistical expert, Thomas N. Day-mont, Ph.D., however, concluded there was “strong statistical evidence that the forced ranking process was not age neutral and, instead, that age was a factor in the ranking process.” D.I. 171 at B2 ¶ 5. Daymont compared the likelihood of being ranked near the bottom of a functional group for employees ages 50 and over with the likelihood an employee under age 50 would be ranked near *1116 the bottom. 12 He observed that of the 183 employees 50 and over, Hercules ranked 31.7 percent of them in the bottom 25 percent. Of the 289 employees under age 50, Hercules ranked only 22.5 percent of them in the bottom 25 percent of the forced rankings. Employees ages 50 and over were, therefore, “1.41 times or 41 percent more likely to be ranked in the lowest quarter of their ranking group.” Id. at B6. Applying the Fisher’s Exact Test to account for random variations, Daymont calculated a difference of 2.16 standard deviations. 13 Id. at B7. Daymont also criticized the report Siskin prepared for Hercules. First, Daymont contended Siskin inappropriately restricted his analysis by only comparing employees 40 and over with those under age 40. Such an approach, according to Daymont, fails to consider the possibility that Hercules’ RIF process might have affected employees 50 and over and 55 and over more adversely than younger employees within the protected group. To support his critique, Daymont used a hypothetical company with 200 employees: all of the 50 employees ages 50 and over were terminated; 25 of 100 employees ages 40 to 49 years were terminated; and 25 of 50 employees under age 40 were terminated. Id. at B9-10. Under the facts of this hypothetical, an employee age 50 or older would be three times as likely to be terminated as an employee under 50. In this same hypothetical, however, a comparison of employees 40 and over with employees under 40 “yields the conclusion that the RIF process had no adverse impact on older employees, despite the fact that every single one of the 50 employees 50 and over was RIF’d.” Id. at B10-11. Secondly, Daymont criticized Siskin’s report for analyzing both exempt and nonexempt employees in the same pool. Exempt employees are “management and professional employees who tend to have college degrees and who perform a variety of technical and management functions throughout the Hercules organization. The non-exempt employees tend to be non-degreed employees who perform clerical jobs.” Id. at B14-15 ¶ 5. Further, Hercules evaluated the performance of employees in the two groups differently. Id. at B13 ¶3. Whereas Hercules evaluated exempt employees based on such factors as management perspective, leadership, business ethics and creativity, it evaluated nonexempt employees based on attendance, quality and quantity of work. Id. at B14 ¶4. Finch stated in a sworn affidavit that in the RIF process, Hercules ranked exempt employees against exempt employees and nonexempt employees against nonexempt employees. Id. at B15 ¶ 8. Finally, the PCC, the body which made the final *1117 decision to terminate Finch, “monitored only the reduction-in-force of the salaried ‘exempt employees.’” D.I. 53 ¶2. According to Daymont, even Siskin’s analysis recognized that the likelihood of being terminated in the RIF varied significantly with exempt status. D.I. 171 at Bll. By affidavit dated June 13, 1994, Siskin responded to Daymont’s analysis and critique. In addition to criticizing Daymont for ignoring “the legal definition of the age-protected group,” Siskin concluded Daymont’s report erred by comparing apples to oranges. In analyzing the effects of the RIF, Daymont excluded employees who opted for voluntary separation from Hercules. In his study of the impact of forced rankings, however, Day-mont included the employees who opted for voluntary separation. Id. at C2 ¶3. Secondly, Siskin noted that, based on his understanding from Hercules’ counsel, the difference in criteria applied to evaluate exempt versus nonexempt employees referred to historical performance evaluations, not the criteria used by the department heads in making forced ranking decisions. Siskin concluded any difference in criteria used for historical evaluations was irrelevant. Id. at C3 ¶ 5. Further, assuming there was a difference in the criteria applied in the forced ranking of exempt and non-exempt employees, Siskin concluded the reasoning which would require exclusion of nonexempt employees would require the Court to consider differences among exempt employees. Thus, according to Siskin, excluding nonexempt employees would require the exclusion of exempt employees who were not evaluated by the same decisionmakers. Id. Siskin also stated that there were several rating groups which included both exempt and nonexempt employees, further refuting Day-mont’s assertion that nonexempt employees should be excluded from the analysis. Id. at C4 ¶ 6, C6. Finally, Siskin observed that after examining all employees subject to the same general process and taking statistical recognition of factors such as exempt status and department or functional group, there was no statistically meaningful difference by age of employees ranked in the bottom 25 percent. Even after including persons who opted for voluntary separation and selecting an age break of 50, Siskin computed a difference of only 0.11 standard deviations. Id. at C4 ¶ 9. II. APPLICABLE LAW A. SUMMARY JUDGMENT Federal Rule of Civil Procedure 56 governs motions for summary judgment. Rule 56 requires courts to enter summary judgment “if the pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the moving party is entitled to judgment as a matter of law.” Fed.R.Civ.P. 56(c). An issue is genuine “if the evidence is such that a reasonable jury could return a verdict for the nonmoving party.” Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 248, 106 S.Ct. 2505, 2510, 91 L.Ed.2d 202 (1986). “Material facts” are those which, under applicable law, might affect the outcome of the case. Id. A party moving for summary judgment bears the burden of demonstrating the absence of material issues of fact, regardless of which party has the burden of persuasion. Chipollini v. Spencer Gifts, Inc., 814 F.2d 893, 896 (3d Cir.) (in banc), cert. dismissed, 483 U.S. 1052, 108 S.Ct. 26, 97 L.Ed.2d 815 (1987). If the nonmoving party bears the burden of persuasion at trial, “the party moving for summary judgment may meet its burden by showing that the evidentiary materials of record, if reduced to admissible evidence, would be insufficient to carry the nonmovant’s burden of proof at trial.” Id. To oppose successfully the motion for summary judgment, the nonmoving party must go beyond the pleadings by introducing affidavits and other evidence which will create a genuine issue of material fact. Celotex Corp. v. Catrett, 477 U.S. 317, 324, 106 S.Ct. 2548, 2553, 91 L.Ed.2d 265 (1986). When the moving party bears the burden of persuasion, however, “the standard is more stringent.” National State Bank v. Federal Reserve Bank, 979 F.2d 1579, 1582 (3d Cir.1992). Under this more stringent standard, if the moving party does not establish the absence *1118 of a genuine issue of material fact, the court must deny the motion for summary judgment, even without opposing evidentiary matter. Id. In deciding a motion for summary judgment, “[ijnferences to be drawn from the underlying facts contained in the evidential sources submitted to the trial court must be viewed in the light most favorable to the party opposing the motion.” Goodman v. Mead Johnson & Co., 534 F.2d 566, 573 (3d Cir.1976), cert. denied, 429 U.S. 1038, 97 S.Ct. 732, 50 L.Ed.2d 748 (1977). In short, the court presumes that the nonmoving party’s version of any disputed issue of material fact is correct. Eastman Kodak Co. v. Image Technical Servs., Inc., — U.S. -, -, 112 S.Ct. 2072, 2077, 119 L.Ed.2d 265 (1992). In drawing these inferences, however, the court does not weigh evidence or make credibility determinations. Big Apple BMW, Inc. v. BMW of North America, Inc., 974 F.2d 1358, 1363 (3d Cir.1992), cert. denied, — U.S. -, 113 S.Ct. 1262, 122 L.Ed.2d 659 (1993). B. THE ADEA The ADEA prohibits an employer from discharging or otherwise discriminating against an employee “because of such individual’s age.” 29 U.S.C. § 623(a)(1). The Act’s protection, however, is “limited to individuals who are at least 40 years of age but less than 70 years of age.” Id. § 631(a). Further, it is not unlawful for an employer to terminate an employee based on legitimate factors other than age. Id. § 623(f)(1); see Hazen Paper Co. v. Biggins, — U.S. -, -, 113 S.Ct. 1701, 1707, 123 L.Ed.2d 338 (1993) (“because age and years of service are analytically distinct, an employer can take account of one while ignoring the other, and thus it is incorrect to say that a decision based on years of service is necessarily ‘age-based.’ ”). 1. Disparate Treatment In construing claims of disparate treatment under the ADEA, courts, where appropriate, apply the methods of proof used under Title VII of the Civil Rights Act of 1964. See, e.g., Armbruster v. Unisys Corp., 32 F.3d 768, 778 n. 10 (3d Cir.1994); Lockhart v. Westinghouse Credit Corp., 879 F.2d 43, 48 n. 6 (3d Cir.1989). Accordingly, courts analyze disparate treatment claims under one of two different modes of analysis: mixed motives or pretext. 14 a. Mixed Motives Cases Under the first mode of analysis, the so-called “mixed motives” case, the evidence as a whole suggests that both permissible and impermissible considerations played a role in the employer’s decision and plaintiff offers “direct evidence” that the discriminatory motive was a substantial or motivating factor in the challenged action. Price Waterhouse v. Hopkins, 490 U.S. 228, 244, 276, 109 S.Ct. 1775, 1787, 1804, 104 L.Ed.2d 268 (1989). To qualify as a mixed motives case, the plaintiff must, at a minimum, introduce circumstantial evidence “ ‘of conduct or statements by persons involved in the decision-making process that may be viewed as directly reflecting the alleged discriminatory attitude.’ ” Griffiths v. CIGNA Corp., 988 F.2d 457, 470 (3d Cir.), cert. denied, — U.S. -, 114 S.Ct. 186, 126 L.Ed.2d 145 (1993) (quoting Ostrowski v. Atlantic Mut. Ins. Cos., 968 F.2d 171, 182 (2d Cir.1992)). If plaintiff satisfies this burden, then the defendant may avoid liability only by establishing an affirmative defense, i.e., “that it would have made the same decision even if the forbidden consideration had played no role in the employment determination.” Id. 988 F.2d at 469. At an appropriate “point in the proceedings, of course, the District Court must decide whether a particular case involves *1119 mixed motives.” Price Waterhouse, 490 U.S. at 247 n. 12, 109 S.Ct. at 1789 n. 12. If a plaintiff who intends to advance a mixed motives case fails to provide direct evidence of discrimination, the court should employ the second mode of disparate treatment analysis, “pretext” analysis. Griffiths, 988 F.2d at 470 n. 13 (citing Price Waterhouse, 490 U.S. at 247 n. 12, 109 S.Ct. at 1789 n. 12). b. Pretext Cases In a pretext case, the plaintiff establishes discriminatory intent through the framework established in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973) and Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 101 S.Ct. 1089, 67 L.Ed.2d 207 (1981). This framework proceeds as follows: First, the plaintiff has the burden of proving by a preponderance of the evidence a prima facie case of discrimination. Second, if the plaintiff succeeds in proving the prima facie case, the burden [of production] shifts to the defendant to articulate some legitimate nondiscriminatory reason for the employee’s rejection. Third, should the defendant carry this burden, the plaintiff must then have the opportunity to prove by a preponderance of the evidence that the legitimate reasons offered by the defendant were not its true reasons, but were a pretext for discrimination. Burdine, 450 U.S. at 252-53, 101 S.Ct. at 1093 (citations omitted). To establish a prima facie case, the first step in the pretext mode of analysis, a discharged employee alleging age discrimination must show: “(1) that he belongs to the protected class, i.e., is older than forty; (2) was qualified by training and experience for the job from which he was discharged; and (3) was replaced by a person sufficiently younger to permit an inference of age discrimination.” Turner v. Schering-Plough Corp., 901 F.2d 335, 342 (3d Cir.1990) (citing Sorba v. Pennsylvania Drilling Corp., 821 F.2d 200, 202 (3d Cir.1987), cert. denied, 484 U.S. 1019, 108 S.Ct. 730, 98 L.Ed.2d 679 (1988)). If a discharged employee’s job is eliminated and he or she is not replaced, whether through a reduction in force or otherwise, “the employee need only show that he was laid off from a job for which he was qualified while other workers not in the protected class were retained.” 15 Id. (citing Heady 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)); see Just v. James River, II, Inc., 784 F.Supp. 1145, 1150 (D.Del.1992) (rejecting defendant’s argument that a plaintiff terminated in a reduction in force must show he or she was qualified for another job with the employer). If the plaintiff succeeds in proving a prima facie ease, the burden of production shifts to the defendant to articulate a legitimate nondiscriminatory reason for the employee’s termination. That is, the “defendant must clearly set forth, through the introduction of admissible evidence, reasons for its actions which, if believed by the trier of fact, would support a finding that unlawful discrimination was not the cause of the employment action.” St. Mary’s, — U.S. at -, 113 S.Ct. at 2747 (internal quotations and emphasis omitted). The Supreme Court has emphasized, however, that while the burden of production shifts, the burden of persuasion at all times remains with the plaintiff. Id. In fact, “the determination that a defendant has met its burden of production (and has thus rebutted any legal presumption of intentional discrimination) can involve no credibility assessment. For the burden-of-production de *1120 termination necessarily precedes the credibility assessment stage.” Id. at -, 113 S.Ct. at 2748 (emphasis in original). If the defendant articulates a legitimate nondiseriminatory reason for the challenged employment action, plaintiff has the burden of persuading the trier of fact that the proffered reason is a pretext for discrimination. Plaintiff must show “both that the reason [is] false, and that discrimination [is] the real reason.” Id. at -, 113 S.Ct. at 2752 (emphasis in original). The Supreme Court has also cautioned, however, that “[t]he factfinder’s disbelief of the reasons put forward by the defendant (particularly if disbelief is accompanied by a suspicion of mendacity) may, together with the elements of the prima facie case, suffice to show intentional discrimination.” Id. at -, 113 S.Ct. at 2749. Because the burden of proof remains with the plaintiff in pretext analysis, St. Mary’s does not preclude summary judgment in all cases in which the plaintiff has made out his or her prima facie case. Armbruster, 32 F.3d at 782. Finally, while the extent of plaintiffs ultimate burden in a pretext case is currently in dispute within the Third Circuit, 16 the Court concludes plaintiff must prove by a preponderance of the evidence that a protected characteristic played a role in the employer’s decisionmaking process and that it was a determinative factor in the outcome of that process. Compare D.1.181 (Hercules urging a plaintiff must prove age was the “sole cause” of the employment decision) with D.I. 180 (Finch contending he need only prove age was “a determinative factor” to prevail in a pretext case). Much of the current debate about the standard in a pretext case focuses on Griffiths v. CIGNA Corp., 988 F.2d 457, in which a panel of the Third Circuit Court of Appeals observed that “the plurality in Price Waterhouse [v. Hopkins, 490 U.S. 228, 109 S.Ct. 1775, 104 L.Ed.2d 268 (1989),] specifically noted that the analysis in pretext cases remains unchanged.” Id. 988 F.2d at 471. Shortly after making this observation, the Griffiths court concluded that “it is clear that in pretext eases the claim is that the discriminatory motive was the sole cause of the employment action and therefore it is inappropriate to state that the plaintiff only need show that the discrimination played ‘a motivating’ or ‘a substantial’ role.” Id. at 472 (emphasis in original). The Griffiths court’s conclusions that Price Waterhouse did not affect pretext cases and that a mixed motives standard is inappropriate in pretext cases are both correct. Hercules and others, however, assert Griffiths also requires a plaintiff in a pretext case to prove the discriminatory motive was the sole cause of the employment action. While Griffiths may be read as imposing such a requirement, there are strong reasons it should not be so read. First, the court’s holding was that it is improper to charge the jury that the plaintiff need only show that discrimination was “a motivating” factor in a pretext case. Secondly, under the Internal Operating Procedures of the Third Circuit, only the court sitting in banc may overrule the published opinion of a previous panel. See Third Circuit I.O.P. 9.1; Siegel v. Alpha Wire Corp., 894 F.2d 50, 53 n. 2 (3d Cir.), cert. denied, 496 U.S. 906, 110 S.Ct. 2588, 110 L.Ed.2d 269 (1990) (“a panel of this court may not overrule a decision of another panel.”). As a general proposition, therefore, the Griffiths court lacked the power to overturn a long line of previous panel decisions which provided that the plaintiff must prove the discriminatory motive was a determinative factor, not the sole cause. See, e.g., Gray v. York Newspapers, Inc., 957 F.2d 1070, 1079 (3d Cir.1992); Billet v. CIGNA Corp., 940 F.2d 812, 816 (3d Cir.1991). This is especially true with regard to Chipollini v. Spencer Gifts, Inc., 814 F.2d 893, an opinion in which the Third Circuit Court of Appeals, in banc, stated that a plaintiff under the ADEA “need not prove that age was the sole or exclusive consideration, but must prove that age made a difference in the decision.” Id. at 897. Nonetheless, in the absence of compelling circumstances, a district court owes blind fealty to the latest precedent of *1121 the circuit court. See Vujosevic v. Rafferty, 844 F.2d 1023, 1030 n. 4 (3d Cir.1988) (“It is, of course, patent that a district court does not have the discretion to disregard controlling precedent simply because it disagrees with the reasoning behind such precedent.”). This is so even if the circuit panel failed to follow its own internal operating procedures or the district judge concludes the panel’s pronouncement is both “bad law” and a departure from settled precedent. See May v. Hobart Corp., 839 F.Supp. 309, 315 (E.D.Pa.1993) (granting a new trial because the Third Circuit, in Griffiths, “disapproved its own apparently well-settled jurisprudence_”). In this ease, however, there is a well-recognized basis for declining to follow the latest Third Circuit precedent. Thus, even if Griffiths did require plaintiff to prove a protected trait was the sole cause of the employer’s action, such a standard would not be consistent with subsequent decisions of the United States Supreme Court. See Frangos v. Doering Equip. Corp., 860 F.2d 70, 72 (3d Cir.1988) (“Although a cogent argument could have previously been waged based on past precedent within this circuit, the Supreme Court has recently rendered a decision making the Appellee’s position untenable.”). According to the Supreme Court, a finding that an employer’s proffered nondiscriminatory explanation is a pretext permits, but does not compel, a ruling for the plaintiff. Thus, the trier of fact is free to reject both the plaintiffs claim of discrimination and the defendant’s proffered nondiscriminatory reason. St. Mary’s Honor Ctr. v. Hicks, — U.S. -, 113 S.Ct. 2742, 125 L.Ed.2d 407. Because the pretext mode of analysis does not require an either/or determination, it is improper to require plaintiff to prove the discriminatory motive was the sole cause of the employer’s decision. Likewise, the Supreme Court, after Grif-fiths, has instructed that “a disparate treatment claim cannot succeed unless the employee’s protected trait actually played a role in that process and had a determinative influence on the outcome.” Hazen Paper, — U.S. -, 113 S.Ct. at 1706. In Hazen Paper, the Supreme Court unanimously held a plaintiff under the ADEA can establish a willful violation and a right to liquidated damages if he or she proves the employer acted with knowledge or in reckless disregard of whether its conduct was prohibited by the statute. The Hazen court specifically noted that the plaintiff need not “prove that age was the predominant rather than a determinative factor in the employment decision.” Id., — U.S. at-, 113 S.Ct. at 1710. If a plaintiff need not show that age was “the predominant factor” to establish a claim for liquidated damages, surely he or she need not prove age was the “sole cause” of the employer’s action to prove a nonwillful violation of the Act. Finally, the Court notes that no other circuit has required a plaintiff in a pretext case to prove a protected trait was the sole cause of the employer’s action, suggesting such a standard is inconsistent with the Supreme Court’s pronouncements in St. Mary’s and Hazen Paper. See, e.g., Kralman v. Illinois Dep’t of Veterans’ Affairs, 23 F.3d 150, 156 (7th Cir.1994) (a determining factor); Cone v. Longmont United Hosp. Ass’n, 14 F.3d 526, 529 (10th Cir.1994) (the determinative factor); Mitchell v. Data General Corp., 12 F.3d 1310, 1317 (4th Cir.1993) (a determinative factor); Atkin v. Lincoln Property Co., 991 F.2d 268, 272 (5th Cir.1993) (a determinative factor); Clark v. Coats & Clark, Inc., 990 F.2d 1217, 1228 (11th Cir.1993) (a determining factor). I conclude Griffiths does not require a plaintiff in a pretext case to prove age was the sole cause of the employer’s action. First, the language in Griffiths was directed at distinguishing the mixed motives mode of analysis from the pretext mode of analysis. Secondly, a sole cause requirement is inconsistent with the Supreme Court’s subsequent decisions in St. Mary’s and Hazen Paper. I conclude, therefore, Finch need only show age was “a determinative factor” in Hercules’ decisionmaking process. 2. Disparate Impact In contrast to a claim of disparate treatment, a claim of disparate impact requires no inquiry into the employer’s intent. Rather, the court examines “the consequences of employment practices, not simply *1122 the motivation.” Griggs v. Duke Power Co., 401 U.S. 424, 432, 91 S.Ct. 849, 854, 28 L.Ed.2d 158 (1971) (emphasis in original). For this reason, the focus in a disparate impact case is usually “on statistical disparities, rather than specific incidents, and on competing explanations for those disparities.” Watson v. Fort Worth Bank & Trust, 487 U.S. 977, 987, 108 S.Ct. 2777, 2785, 101 L.Ed.2d 827 (1988). Despite these differences, however, disparate impact cases employ a burden shifting analysis which is similar to that used in disparate treatment pretext cases and, like disparate treatment claims, can be used to challenge both objective and subjective employment practices. See id. at 991, 108 S.Ct. at 2787. To establish a prima facie case of age discrimination under the disparate impact theory, the plaintiff must “show that the employer’s selection process results in unfavorable treatment of a disproportionate number of members of the protected group to which the plaintiff belongs.” Massarsky v. General Motors Corp., 706 F.2d 111, 120 (3d Cir.), cert. denied, 464 U.S. 937, 104 S.Ct. 348, 78 L.Ed.2d 314 (1983). In making this showing, the plaintiff must: (1) identify the specific employment practice that he or she is challenging; (2) show disparate impact; and (3) prove causation by introducing “statistical evidence of a kind and degree sufficient to show that the practice in question has caused the [termination of employment or] exclusion of applicants for jobs or promotions because of their membership in a protected group.” Watson, 487 U.S. at 994, 108 S.Ct. at 2789. If the plaintiff succeeds in establishing a prima facie case of disparate impact, courts proceed to analyze “whether a challenged practice serves, in a significant way, the legitimate employment goals of the employer.” Wards Cove Packing Co. v. Atonio, 490 U.S. 642, 659, 109 S.Ct. 2115, 2126, 104 L.Ed.2d 733 (1989). As in a disparate treatment pretext case, the “employer carries the burden of producing evidence of a business justification for his employment practice. The burden of persuasion, however, remains with the disparate-impact plaintiff.” 17 Id. While the proffered legitimate business justification need not be “essential” or “indispensable,” it must be more than merely rational. The Third Circuit Court of Appeals has explained that an employer satisfies its burden of production only when it proffers some proof that the device serves identified legitimate and substantive business goals. That is, the defendant’s burden [is] to identify the particular employment goal and to present evidence of how the [challenged practice] “serves in a significant way” the identified goal. Merely being abstractly rational, as opposed to arbitrary, would not suffice. The defendant, therefore, has some burden of presenting objective evidence ... factually showing a nexus between the selection device and a particular employment goal. Without evidence of such a relationship it cannot be said that the defendant has presented any evidence that the “challenged practice serves, in a significant way, the legitimate goals of the employer.” Newark Branch, NAACP v. Town of Harrison, 940 F.2d 792, 804 (3d Cir.1991) (quoting Mack A. Player, Is Griggs Dead? Reflecting (Fearfully) on Wards Cove Packing Co. v. Atonio, 17 Fla.St.U.L.Rev. 1, 32 (1989)). Defendant must, therefore, produce evidence both of its legitimate employment goal and evidence of how the challenged practice significantly serves that goal. Finally, if defendant satisfies its burden of production, “the plaintiff must ‘show that other tests or selection devices, without a similarly undesirable [discriminatory] effect, would also serve the employer’s legitimate interest in efficient and trustworthy workmanship.’ ” Watson, 487 U.S. at 998, 108 S.Ct. at 2790 (quoting Albemerle Paper Co. v. Moody, 422 U.S. 405, 425, 95 S.Ct. 2362, 2375, 45 L.Ed.2d 280 (1975)). The plaintiff can “discredit the legitimate busi *1123 ness justification asserted ... [or] suggest a viable alternative to the challenged practice which has the effect of reducing the disparate impact and the employer refuses to adopt the alternative.” Newark Branch, NAACP, 940 F.2d at 798 (emphasis in original) (citing Wards Cove, 490 U.S. at 660-61, 109 S.Ct. at 2126-27). Accord Abbott v. Federal Forge, Inc., 912 F.2d 867, 876 (6th Cir.1990) (“Once the employer identifies a legitimate, non-diseriminatory business reason for the employment practice in question, the burden shifts back to the plaintiff to show that the reason is pretextual or to show the existence of an alternative employment practice that reduces the disparate impact but serves the employer’s legitimate interests.”). In determining whether plaintiffs proposed alternative would be equally effective in serving the employer’s legitimate business concerns, courts should compare the cost and other burdens of the proposed alternative with those of the challenged practice. See Watson, 487 U.S. at 998, 108 S.Ct. at 2790-91. III. FINCH’S DISPARATE TREATMENT CLAIM Finch asserts he has proffered direct evidence of age discrimination which entitles him to the special burden-shifting treatment of the mixed motives mode of analysis. Finch also asserts he has introduced evidence which suggests Hercules’ proffered legitimate nondiscriminatory reason is a pretext for discrimination. The Court will address Finch’s assertions seriatim. If there are genuine issues of material fact as to either mode of analysis, Hercules’ motion for summary judgment on Finch’s claim of disparate treatment must be denied. Further, even if there are no genuine issues of material fact as to one of the two modes of analysis, plaintiff still may present evidence at trial which makes that mode of analysis appropriate. See Armbruster, 32 F.3d at 781 n. 17 (“While the evidence here presented to us at the summary judgment stage does not trigger the Price Waterhouse framework, the evidence presented during trial may.”). A. MIXED MOTIVES To withstand a motion for summary judgment under the mixed motives mode of analysis, a plaintiff under the ADEA must, through the introduction of direct evidence, show that there is a genuine issue of material fact as to whether age was a substantial or motivating factor in the employer’s adverse employment action. Price Waterhouse, 490 U.S. at 244, 276, 109 S.Ct. at 1787, 1804. At a minimum, direct evidence consists of circumstantial evidence “ ‘of conduct or statements by persons involved in the decision-making process that may be viewed as directly reflecting the alleged discriminatory attitude.’ ” Griffiths, 988 F.2d at 470 (quoting Ostrowski, 968 F.2d at 182.). Finch points to several statements made by Gossage, Hercules’ CEO, which he asserts directly reflect age bias by a decision-maker. First, Finch notes Gossage’s statement to the press that “[t]he young people in the company want us to bring Hercules back to where it ought to be again.... Older people will see friends impacted and will feel bad about it.” Gossage repeated the substance of this statement at a Hercules’ question and answer session when he observed “that younger employees want to know when the company will get up and turn itself around, and I acknowledged that to those with long service to the company, it was painful to watch what was going on.” Hercules contends these statements constitute mere stray remarks which are not indicative of age bias. Hercules explains that Gossage’s comments were about Hercules’ recently announced voluntary early retirement program, not about the involuntary RIF. Because only “older,” i.e. more senior employees, could qualify for early retirement, the statement does not constitute evidence of age bias. D.1.118 at 47. Hercules also cites decisions in other circuits in which courts have held a decisionmaker’s stray remarks are insufficient to establish age discrimination. Id. at 48 (citing Waggoner v. City of Garland, 987 F.2d 1160, 1166 (5th Cir.1993) (reference to the plaintiff as “an old fart” and statements that a younger person could do faster work did not preclude grant of employer’s motion for summary judgment)). *1124 The Court disagrees that Gossage’s statements must be deemed mere stray remarks. A reasonable trier of fact could infer either that the statements referred solely to the voluntary early retirement program, as Hercules argues, or that they are indicative of age bias by Gossage, Hercules’ CEO. The Court may not weigh competing inferences on a motion for summary judgment. See Siegel, 894 F.2d at 54-55 (competing inferences from statement that “old dogs won’t hunt” preclude summary judgment). Hercules also asserts that even if the remarks are indicative of age bias, there is no evidence in the record that Gossage was a decisionmaker in Hercules’ decision to terminate Finch. The RIF process had begun before Gossage became CEO and he merely approved use of the “stretch number” for the RIF. D.I. 134 at 3. According to Hercules, MacKenzie was the principal decisionmaker in Finch’s termination. D.I. 163 at 63-64. The Third Circuit Court of Appeals, however, has observed that [w]hen a major company executive speaks, “everybody listens” in the corporate hierarchy, and when an executive’s comments prove to be disadvantageous to a company’s subsequent litigation posture, it can not compartmentalize this executive as if he had nothing more to do with company policy than the janitor or watchman. Lockhart, 879 F.2d at 54. Because of Gos-sage’s central role as Hercules’ CEO, the Court holds that a reasonable jury could infer MacKenzie and other Hercules’ personnel “listened” to Gossage’s statements about young people and older people at Hercules. As additional direct evidence in support of his mixed motives claim, Finch points to the deposition testimony of William E. Hosker. 18 Hosker testified that when he questioned Gossage about why he had not been promoted in January 1992, Gossage acknowledged that Hosker had been disadvantaged because of his age and that employees in their late fifties tended to spend their energy preparing for retirement. Hosker also testified about May 1992 statements Gossage had allegedly made to C. Doyle Miller, relayed by Miller to Hosker, regarding the failure of the Chemical Specialties Group to address the problem of “tired warriors” who blocked the movement of younger, high potential people into development positions at Hercules. Hercules argues the Court must not consider Hosker’s testimony because it is irrelevant to Finch’s claim that he was terminated because of his age and because it is hearsay. First, Hercules objects that Hosker’s testimony is irrelevant because it refers to isolated remarks which are unrelated to, and remote in time from, Finch’s termination. D.I. 134 at 22-23. Specifically, Hercules observes that Hosker’s testimony alleges Gossage considered proximity to retirement age in making promotion decisions for senior executive positions. This issue, according to Hercules, is irrelevant to Finch’s claim. Further, Hercules notes that Hosker alleges Gossage made these statements twelve to eighteen months after Hercules terminated Finch, making the statements too remote to be probative of age discrimination in Finch’s termination. In support of its position that Hosker’s testimony is irrelevant, Hercules has cited several cases from other circuits holding that vague and temporally remote statements are inadmissible. See, e.g., Phelps v. Yale Security, Inc., 986 F.2d 1020, 1025-26 (6th Cir.), cert. denied, — U.S. -, 114 S.Ct. 175, 126 L.Ed.2d 135 (1993) (affirming the district court’s grant of judgment notwithstanding the verdict, the court observed that statements that plaintiff “was too old” and that her fifty-fifth birthday was a cause for concern, made one year before her termination, were too remote and too ambiguous to establish age discrimination); Turner v. North American Rubber, Inc., 979 F.2d 55, 59 (5th Cir.1992) (ruling that a reference to “three young tigers” was irrelevant because it did not refer to plaintiffs age, was not related to his discharge and was made more than a year before his termination); Haskell v. Kaman Corp., 743 F.2d 113, 120 (2d Cir. *1125 1984) (holding the district court erred by admitting employer’s references to “old ladies with balls” and “young turks,” where the statements did not refer to the plaintiff and were made three to eighteen years before his termination). The Court begins by noting that evidence is relevant if it has “any tendency to make the existence of any fact that is of consequence to the determination of the action more probable or less probable than it would' be without the evidence.” Fed.R.Evid. 401. The appropriate inquiry, therefore, is whether the evidence makes the existence of age-bias in Hercules’ decision to terminate Finch more or less likely. In Lockhart, an age discrimination case, the Third Circuit Court of Appeals ruled that the plaintiff, who was terminated in May 1983, could introduce a high-ranking manager’s statement, made in January 1984, that “Westinghouse Credit was a seniority driven company with old management and that’s going to change, ‘I’m going to change that.’ ” 879 F.2d at 54. The Lockhart court reasoned “it would be reasonable for the jury to conclude that his statement of January 1984 was not a recent managerial viewpoint, but that it was merely a cumulative statement of managerial policies that had been sanctioned or favored by top executives at WCC for a considerable time.” Id. While the Court concludes it would be reasonable for the jury to infer Gossage’s views at the time of his January 1992 conversation with Hosker were the same at the time Hercules terminated Finch, Finch has not demonstrated how age bias in promotion decisions makes the existence of age bias in Hercules’ decision to terminate his employment more likely. 19 The statement in Lockhart about a seniority driven company with old management suggests an employee’s termination may have been age-based. In contrast, Gossage’s statements to Hosker that Hosker was not promoted because of his age do not make the existence of age bias in Hercules’ decision to terminate Finch more likely. Age bias in promotion decisions simply does not make age bias in termination decisions more likely. Further, to the extent this portion of Hosker’s testimony is minimally relevant, the Court concludes its probative value is substantially outweighed by the danger of unfair prejudice. See Fed.R.Evid 403. The Court will not consider this portion of Hosker’s testimony. Hosker’s testimony about Gossage’s conversation with Miller, however, may be relevant to Finch’s disparate treatment claim. Hosker testified that Miller relayed Gos-sage’s concern about the failure to address the problem of “tired warriors,” people in their mid-fifties who were “blocking the movement of younger, high potential people.” A jury could fairly conclude that these statements implied Hercules should terminate employees who, like Finch, were in their mid-fifties. Such an inference would also be consistent with Finch’s claim, based on statistical evidence, that age bias infected Hercules’ entire RIF process. It is not clear, however, when Gossage allegedly made these comments to Miller. Further, the Court notes that Miller has testified he does not recall using the term “tired warriors” and Gossage has yet to be questioned about the term. Likewise, while Hosker testified that Miller defined the term tired warriors, there is no indication of what meaning, if any, Gossage ascribed to the phrase. Finally, there is no indication in the record whether the problem of “tired warriors” was limited to the Specialty Chemical Group or applied to Hercules generally. In sum, these issues raise genuine concerns that the probative value of Hosker’s testimony may be substantially outweighed by the danger of unfair prejudice. Hercules also argues the Court must not consider Hosker’s testimony because it is inadmissible hearsay. 20 At this stage of the proceedings, it appears Hosker’s testimony *1126 may fall within the exclusion to the prohibition against hearsay contained in Federal Rule of Evidence 801(d)(2)(D). 21 Hosker, the declarant, testified under oath. Gossage’s statements to Miller, Vice President of Chemical Specialties, were made during the course of Gossage’s employment by Hercules and concerned promotion and succession planning within the Chemical Specialties Group, a matter within the scope of Gos-sage’s employment. Likewise, it appears Miller’s statements to Hosker were made within the scope and during the course of Miller’s employment by Hercules. Each layer of hearsay fits within the exclusion from the prohibition against hearsay contained in Federal Rule of Evidence 801(d)(2)(D). See Lippay v. Christos, 996 F.2d 1490, 1498 (3d Cir.1993) (noting that extra-judicial statements by corporate employees are admissible in civil suits against their superiors “if the factors which normally make up an agency relationship are present as between the employee and the superior.”); 4 Jack B. Wein-stein & Margaret A. Berger, Weinstein’s Evidence ¶ 801(d)(2)(D)[01] (“Once agency, and the making of the statement while the relationship continues, are established, the statement is exempt from the hearsay rule so long as it relates to a matter within the scope of the agency.”) (footnotes omitted). Subject to providing an adequate showing that these statements were made within the scope of Miller’s agency at Hercules and the Court’s concerns about the probative value of Hosker’s testimony, the Court deems Hosker’s statements about Miller’s conversation with Gossage to be nonhearsay. 22 The Court concludes plaintiff has introduced sufficient direct evidence of age bias to create a genuine issue of fact as to whether his age was a motivating factor in Hercules’ decision to terminate him. 23 Gos-sage’s statements in the press and at the Hercules’ question and answer session suggest Gossage, Hercules’ CEO, may have been motivated by age bias. 24 Because persons in *1127 the corporate hierarchy “listen” to a CEO’s comments, the Court will deny defendant’s motion for summary judgment on Finch’s disparate treatment claim. B. PRETEXT The pretext mode of analysis employs a three step, shifting burdens framework. First, the plaintiff must establish a prima facie case of discrimination. If plaintiff establishes a prima facie case, the defendant must produce a legitimate nondiscriminatory reason for the employment action. If the defendant satisfies its burden of production, the plaintiff must show that the employer’s proffered reason is a pretext for discrimination. Burdine, 450 U.S. at 252-53, 101 S.Ct. at 1093-94. To survive summary judgment, the plaintiff must establish a prima facie case and point to admissible evidence which tends to show pretext. Armbruster, 32 F.3d at 781-82. Hercules concedes Finch has established a prima facie case of disparate treatment, but asserts it has presented a legitimate nondiscriminatory reason for his termination, namely that Hercules’ decision-makers believed Finch was a poor performer. 25 D.I. 163 at 4H3. Based on the evidence in the record, the Court concludes Hercules has satisfied its burden of producing a legitimate nondiscriminatory reason for plaintiff’s termination. See D.I. 119 at A32627, A346-52 (MacKenzie criticizing Finch’s work and practice of sleeping at his desk); id. at A418, A424, A433-35 (Spero stating Finch was not an innovative or proactive General Auditor). Finch, therefore, must point to admissible evidence which tends to show Hercules’ proffered reason is pretextual. To support his position that Hercules’ proffered reason is a pretext for discrimi *1128 nation, Finch has introduced evidence suggesting that he was an effective leader, strong decisionmaker, and hard worker who received above average salary increases. D.I. 121 at Bl, B20-22, B431, B135. Secondly, Finch has introduced statistical evidence which suggests Hercules was more likely to rank older employees near the bottom of its forced rankings. D.I. 171 at B6-7. Finally, Finch has introduced evidence of statements by Hercules’ CEO, Gossage, from which a reasonable factfinder could infer defendant’s decision to terminate Finch was based not on performance, but on age. D.I. 121 at B288, B327. The Court concludes Finch has introduced evidence sufficient to create a genuine issue of fact as to whether defendant’s proffered reason is a pretext. Accordingly, the Court will deny Hercules’ motion for summary judgment on Finch’s claim of disparate treatment. IV. FINCH’S DISPARATE IMPACT CLAIM In urging the Court to grant its motion for summary judgment on Finch’s disparate impact claim, Hercules contends: (A) Finch has not established a prima facie case of disparate impact because he has relied upon evidence which improperly divides the relevant statistical pool; (B) it has offered a legitimate nondiscriminatory reason for the employment action; and (C) Finch has not shown that there was an alternative to the challenged practice. 26 D.I. 174. Finally, Hercules also asserts a claim of disparate impact is not cognizable under the ADEA. Id. at 17-19. Following the mode of analysis of the Third Circuit Court of Appeals, the Court will, in resolving Hercules’ first three contentions, assume the ADEA permits claims of disparate impact. See Massarsky, 706 F.2d at 120 (assuming, without deciding, that the ADEA permits claims of disparate impact). A. THE PRIMA FACIE CASE: STATISTICAL ANALYSIS Hercules contends flaws in Finch’s statistical analysis require the Court to enter summary judgment in its favor. Specifically, Hercules objects that Finch’s expert’s report improperly: (1) restricts its analysis to “exempt” employees; (2) uses subsets within the ADEA’s protected class; and (3) includes persons who opted for voluntary retirement. Hercules also asserts Finch’s statistics are legally insufficient to prove a claim of discrimination. D.I. 174 at 3-12. First, Hercules asserts the relevant labor pool for Finch’s disparate impact claim consists of both exempt and nonexempt employees because both groups were eligible for the RIF and subject to forced ranking. Id. at 4. In his sworn affidavit, however, Finch states that “[f]or the purposes of the RIF, exempt employees were ranked against exempt employees and non-exempt employees were ranked against non-exempt employees pursuant to the RIF guidelines.” D.I. 171 at B15 ¶ 8. Finch also asserts that because Hercules evaluated job performance differently for the two groups, the RIF ranking process varied between exempt and nonexempt employees. Id. Because the determination of whether exempt and nonexempt employees should be considered in the same statistical pool necessarily involves weighing and interpreting the evidence presented, the Court concludes this is a question of fact. See Faulkner v. Super Valu Stores, Inc., 3 F.3d 1419, 1430 (10th Cir.1993). In light of this conclusion and because there is a genuine issue as to whether the RIF ranking process applied in the same manner to exempt and nonexempt employees, the Court may not, on this ground, grant Hercules’ motion for summary judgment. 27 *1129 Secondly, Hercules objects to Finch’s expert’s practice of dividing the protected class, persons between the ages of forty and seventy, into subsets. 28 D.I. 174 at 8-12. Finch’s expert compared: (1) Hercules’ employees who were ages 55 and over to those who were under 55; and (2) employees who were ages 50 and over to those who were under 50 years old. D.I. 171 at B6-7. Hercules, citing Lowe v. Commack Union Free School District, 886 F.2d 1364 (2d Cir.1989), ce rt. denied, 494 U.S. 1026, 110 S.Ct. 1470, 108 L.Ed.2d 608 (1990), contends this practice is an attempt to bypass the ADEA’s plain language, which protects all persons between the ages of forty and seventy. In Lowe, plaintiffs, like Finch, sought to establish a prima facie case of age discrimination by comparing the effects of the defendant’s actions on persons over 50 with those under 50. The Second Circuit Court of Appeals held the ADEA’s protection of persons between the ages of forty and seventy precluded a disparate impact claim based on an adverse effect on a subset of the protected group. Id. 886 F.2d at 1370-74. In support of its holding, the Second Circuit noted that [i]f appellants’ approach were to be followed, an 85 year old plaintiff could seek to prove a discrimination claim by showing that a hiring practice caused a disparate impact on the “sub-group” of those age 85 and above, even though all those hired were in their late seventies. Id. at 1373. Accord Barnes v. GenCorp Inc., 896 F.2d at 1467 n. 12 (subgroup analysis “may not” apply to discriminatory impact eases); Leidig v. Honeywell, Inc., 850 F.Supp. 796, 803 n. 7 (D.Minn.1994) (“comparison of employees over the age of fifty to employees under the age of fifty in support of an ADEA disparate impact claim is also troublesome, since the ADEA confers protected status on anyone forty or older.”). The Court concludes, assuming claims of disparate impact are cognizable under the ADEA, that adopting the holding of Lowe would frustrate the purpose of the ADEA. See Connecticut v. Teal, 457 U.S. 440, 453-55, 102 S.Ct. 2525, 2533-35, 73 L.Ed.2d 130 (1982) (concluding Title VII does not permit employers to discriminate against some employees while favorably treating others within the protected group). As suggested by Finch’s expert, Dr. Daymont, failure to accord protection to subsets of the protected class would allow an employer to adopt facially neutral policies which had a profoundly disparate impact on individuals over age 50 or 55, so long as persons under age 50 or 55 received sufficiently favorable treatment that the adverse impact on individuals over 40 was minimal. 29 Such policies, while facially neutral, could well reflect the specific type of arbitrary age discrimination Congress sought to prohibit by enacting the ADEA. See 29 U.S.C. § 621. If the Court were to adopt the holding of Lowe, such a policy would be free from judicial scrutiny. Further, the Court concludes the Lowe court’s concern about where to draw the line on the use of subgroups is answered by the nature of a disparate impact claim. If a plaintiff attempts to define the subset too narrowly, he or she will *1130 not be able to obtain reliable statistics upon which to prove a prima facie case. See Mas-sarsky, 706 F.2d at 121 (“An adverse effect on a single employee, or even a few employees, is not sufficient to establish disparate impact.”). Assuming, without deciding, that the ADEA permits claims of disparate impact, I am persuaded Finch may seek to prove Hercules’ forced ranking process had a disparate impact on a subgroup of which he is a member. 30 See Graffam v. Scott Paper Co., 848 F.Supp. 1, 4 (D.Me.1994) (rejecting Lowe for similar reasons). Thirdly, Hercules, citing LeBlanc v. Great American Insurance Co., 6 F.3d 836, 848 (1st Cir.1993), cert. denied, — U.S. -, 114 S.Ct. 1398, 128 L.Ed.2d 72 (1994), asserts Finch’s statistical analysis of the effect of the forced ranking process is flawed because it includes employees who opted for voluntary termination. According to Hercules, voluntary departures have no bearing on Finch’s disparate impact claim. The Court disagrees. In LeBlanc, plaintiff offered statistical evidence of the defendant’s employment practice to support a claim of disparate treatment. After noting that statistical evidence was less probative in disparate treatment cases than in disparate impact cases, the court concluded plaintiffs statistics, which included all employees who had either voluntarily or involuntarily left the employer, could not reasonably be used to infer intentional age discrimination in plaintiffs termination. See id. In contrast to LeBlanc’s claim of intentional discrimination, Finch has alleged Hercules’ forced ranking process had a disparate impact on older employees. He has challenged the forced ranking itself, not the resulting terminations. Further, the Court notes that Hercules required its managers to submit their forced rankings by mid-January 1991, while the deadline for the voluntary reduction in force was not until January 28,1991. See D.I. 119 at A140; D.I. 121 at B287. All Hercules’ employees, therefore, were forced ranked, regardless of whether they later opted for voluntary retirement. 31 Simply put, I cannot conclude that the rank of persons who later opted for voluntary retirement is irrelevant to Finch’s disparate impact claim. Cf. Healy, 860 F.2d at 1218 (in reviewing a disparate treatment pretext case, the court noted the employer’s statistics “may have been different if the older employees who accepted early retirement had been accounted for.”). Finally, Hercules asserts Finch’s statistics are “insufficient to prove a claim of discrimination.” D.I. 174 at 11. The Court disagrees. While no particular number of standard deviations are required for a plaintiff to establish a prima facie case of disparate impact, the United States Supreme Court has suggested that a difference of two or three standard deviations generally is statistically significant. Watson, 487 U.S. at 995 n. 3, 108 S.Ct. at 2789 n. 13; Hazelwood School Dist. v. United States, 433 U.S. 299, 308 n. 14, 97 S.Ct. 2736, 2742 n. 14, 53 L.Ed.2d 768 (1977). Hercules’ employees ages 50 and over were 1.41 times more likely to be ranked in the bottom 25 percent of the forced rankings, a difference of 2.16 standard deviations. Employees ages 55 and over were 1.60 times more likely to be ranked in the bottom 25 percent of the forced rankings, a difference of 2.68 standard deviations. Under the facts of this case, the Court holds a difference approaching three standard deviations is sufficiently significant to support Finch’s prima facie case of disparate impact. B. HERCULES’ LEGITIMATE NONDISCRIMINATORY REASON If a plaintiff successfully establishes a prima facie case of disparate impact discrimination, the burden shifts to the defendant to produce a business justification for the employment practice. Wards Cove, 490 U.S. at 659, 109 S.Ct. at 2126. Defendant must produce evidence both of its legitimate employment goal and evidence of how the challenged practice significantly serves that *1131 goal. Newark Branch, NAACP, 940 F.2d at 804. Hercules contends the forced rankings were necessary to discern the least qualified performers for termination in the RIF. D.I. 174 at 13; see D.I. 171 at B91-95; D.I. 119 at A205, A209, A482. The Court agrees Hercules has identified a legitimate employment goal, retaining the better qualified personnel. See Lucas v. Dover Carp., 857 F.2d 1397, 1403 (10th Cir.1988) (noting employer’s efforts to retain more qualified employees). Finch, while conceding there were deficiencies in Hercules’ previous performance appraisal system, maintains that Hercules has not demonstrated that the forced ranking process served its legitimate employment goals. D.I. 170 at 20-21. Hercules is not, however, required to prove how forced ranking helped guarantee better employees would be retained. Rather, Hercules need only produce evidence which, if believed by the trier of fact, would permit the conclusion that forced ranking significantly served such a purpose. This it has done. The very nature of the forced ranking guidelines required evaluators to assess each employee’s present performance, not potential. D.I. 119 at A495. Further, the forced ranking process used paired comparisons between employees within the same group, a process designed to assure each employee was appropriately ranked based on RIF criteria. Id. at A496-99. Accordingly, the Court holds Hercules has satisfied its burden of producing evidence of both a legitimate business goal and how the forced ranking process significantly served that goal. C. A LESS DISCRIMINATORY ALTERNATIVE Finally, if the defendant provides a legitimate business justification for the specific employment practice challenged by a disparate impact plaintiff, the plaintiff must offer evidence which discredits the asserted business justification or suggest a viable, less discriminatory alternative [“LDA”] to the challenged practice. Wards Cove, 490 U.S. at 660-61, 109 S.Ct. at 2126-27; Newark Branch, NAACP, 940 F.2d at 798. If the plaintiff proposes an LDA, thé court should consider the cost and other burdens it would impose on the employer. Watson, 487 U.S. at 998, 108 S.Ct. at 2790-91. At oral argument, Finch’s counsel expressly disavowed any assertion that Hercules’ business justification was a pretext. Thus, Finch must suggest a viable LDA for Hercules’ forced ranking process. The exact nature of plaintiffs burden at this stage of a disparate impact case has not been well developed in the ease law. In United States v. City of Warren, 759 F.Supp. 355 (E.D.Mich.1991), at the summary judgment stage, the district court concluded the United States had established a prima facie case of disparate impact as to Warren’s use of a preapplication residency requirement for employment. Id. at 363. Referring to a deposition of a former city councilmember, the court noted Warren proffered its good faith reliance on then-existing state law to justify its adoption of a residency requirement, as well as urging the residency requirement promoted the tax base, enhanced community pride and improved the ability of uniformed employees to respond to emergencies. Id. Citing the United States’ brief, the court observed that Warren’s goals could have been served by a less discriminatory alternative, post-hire move-in requirements. Id. at 364. The court granted the United States’ motion for summary judgment, concluding both that Warren’s business justification was unpersuasive because good faith was irrelevant in disparate impact cases and “that there were alternatives available to Warren, other than such a requirement, which could have served to achieve the goals stated by Warren as part of the decision to use the residency requirement.” Id. at 365. Thus, for the court in the Eastern District of Michigan, confronted with an LDA that was obviously less discriminatory than the challenged practice, argument in a brief was sufficient proof of a less discriminatory alternative. See also Byrne v. City of Naperville, No. 88-C-5375, 1991 U.S.Dist. LEXIS 15873, at *16 (N.D.Ill. Oct. 30, 1991) (denying defendant’s motion for summary judgment because “whether there exists a business justification and whether Byrne can present evidence of reasonable alternatives *1132 are genuine issues of material fact which preclude summary judgment.”). Not surprisingly, other courts have required plaintiff to do more than argue the presence of an LDA; they have required evidence from which the trier of fact could conclude there was a viable, less discriminatory alternative. Thus, in Colby v. J.C. Penney Co., 926 F.2d 645 (7th Cir.1991), the court concluded that because plaintiff “introduced no facts suggesting that Penney’s head of household rule is pretextual or that Penney has less discriminatory means of achieving its broad beneficial purposes, summary judgment was appropriate on the disparate impact claims.... ” Id. at 649 (emphasis supplied). See also Fitzpatrick v. City of Atlanta, 2 F.3d 1112, 1118 (11th Cir.1993) (affirming summary judgment for defendant because plaintiffs “faded to create a genuine issue as to the City’s contention that the ban on shadow beards is necessitated by safety concerns.”). In light of these conflicting views on the nature of plaintiffs burden of demonstrating an LDA, the Court finds it appropriate to return to the standards governing disparate impact litigation and summary judgment. Under Wards Cove, the plaintiff retains the burden of proof throughout the litigation, including the burden of proving a viable LDA. To oppose successfully a defendant’s motion for summary judgment, a disparate impact plaintiff, like Finch, must go beyond the pleadings by introducing affidavits and other evidence which will create a genuine issue of fact. See Celotex, 477 U.S. at 324, 106 S.Ct. at 2553. Thus, the Court concludes a plaintiff in a disparate impact case governed by Wards Cove must, when confronted by a motion for summary judgment in which the defendant has introduced evidence of a business justification, put evidence in the record which supports a viable LDA. Mere argument will not suffice. As noted, Finch’s counsel expressly disavowed any assertion that Hercules’ business justification was a pretext. Rather, counsel argued Hercules’ forced ranking process was too subjective because it permitted decisionmakers to consider performance, versatility, flexibility and continuous service without documenting how employees ranked in each category. Plaintiffs proposed LDA is also forced ranking, but based on interpersonal skills, sense of urgency, quantity, technical competence, analytical skills and knowledge of business skills. See D.I. 171 at B84. In the proposed LDA, the RIF decisionmaker would assign a point value to each of these specific weighted categories, creating documentary support for the rankings derived from the categories. See D.I. 170 at 21; D.I. 171 at B82-84. Finch points to the May 1992 “Personnel Displacement List” for the Information Management — Computer Systems Department as an example of his proposed LDA. See D.I. 171 at B82-84. Plaintiff, however, has provided no evidence to support the proposition that the forced ranking process used in the Displacement List is in fact less discriminatory. While plaintiff at summary judgment is not required to prove his proposed alternative is less discriminatory, he must at least introduce some evidence to support such a conclusion. In short, because plaintiff has placed no evidence in the record of an alternative with a less discriminatory effect, he has failed to create a genuine issue of fact on his claim of disparate impact. Further, the Displacement List, if it could be considered an alternative practice, appears to have a discriminatory effect which is greater than the process of which Finch complains. The Displacement List is a forced ranking of 25 employees, of whom Hercules terminated the bottom five. 32 It contains eighteen employees under age 50 and seven ages 50 and over. Of the six employees ranked in the bottom 25 percent, three were under 50 years of age and three were age 50 and over. The expected number, however, would have been approximately 4.5 employees under age 50 and 1.5 employees ages 50 and over. Thus, from the figures in the Displacement List, it appears Hercules’ employees ages 50 and over were 2.53 times more likely than their younger counterparts *1133 to be ranked in the bottom 25 percent of plaintiffs proposed LDA. The forced ranking process of which Finch complains, however, resulted in employees ages 50 and over being only 1.41 times more likely to be ranked in the bottom 25 percent. Thus, even after carefully examining the limited materials plaintiff has placed in the record, the Court concludes plaintiff has not proffered any evidence of a viable LDA. See Kuhn v. Island Creek Coal Co., No. 91-6325, 1992 WL 207942, at *4-5,1992 U.S.App. LEXIS 20996, at *18 (6th Cir. Aug. 27, 1992), cert. dismissed, — U.S. -, 114 S.Ct. 421, 126 L.Ed.2d 367 (1993) (summary judgment for defendant proper where plaintiff “has not demonstrated an equally effective alternative to the selection criteria_”). Accordingly, defendant’s motion for summary judgment on the disparate impact claim must be granted. V. CONCLUSION For the foregoing reasons, the Court will deny Hercules’ motion for summary judgment on Finch’s claim of disparate treatment and grant Hercules’ motion for summary judgment on Finch’s claim of disparate impact. An appropriate order will issue. Footnotes [1] Finch also asserted a second, state law claim which the Court previously dismissed. Finch v. Hercules, Inc., 809 F.Supp. 309, 313 (D.Del. 1992). [2] The Audit Committee oversees the inside and outside auditors, as well as management of Hercules' assets. D.I. 119 at A55. [3] Plaintiff contends the IPI study is not relevant to Finch's suit against Hercules because it was intended to help the company reduce its indirect costs, not evaluate job performance. Further, plaintiff asserts the IPI study's existence “was not realized until the middle of this litigation ..D.I. 120 at 27. The Court agrees and will not consider the IPI study. To be admissible in either a mixed motives or pretext case, evidence must be relevant to the proofs or rebuttals of defendant's discriminatory motivation. See St. Mary’s Honor Ctr. v. Hicks , — U.S. -, -, 113 S.Ct. 2742, 2752, 125 L.Ed.2d 407 (1993) (level of specificity refers to "the specific proofs and rebuttals of discriminatory motivation the parties have introduced[,]” not to whether defendant's asserted reason is *1111 true or false.); Price Waterhouse v. Hopkins, 490 U.S. 228, 252, 109 S.Ct. 1775, 1791, 104 L.Ed.2d 268 (1989) (an employer in a mixed motives case cannot prevail "by offering a legitimate and sufficient reason for its decision if that reason did not motivate it at the time of the decision.’’). The Third Circuit Court of Appeals recently has instructed that “after-acquired evidence is inadmissible, because irrelevant, at the liability stage of a cause of action under Title VII or ADEA.” Mardell v. Harleysville Life Ins. Co., 31 F.3d 1221, 1238 (3d Cir.1994). Because the persons making forced ranking decisions were not aware of the IPI study until after the decision to terminate Finch had been made, the study could not have affected their motivation in terminating plaintiff. As a consequence, the Court will not consider the IPI study for purposes of summary judgment. [4] In a forced ranking, "evaluators are instructed to identify the 'best' employee and the 'worst' employee with regard to a particular factor, and then the second best and the second worst, and so on down the line." N. Thompson Powers, Reductions in Force Under the Age Discrimination in Employment Act, 2 Lab.Law. 197, 216-17 (1986). [5] In fact, Litras was not aware that Finch had been terminated until after Finch filed this suit. D.I. 119 at A260. [6] The parties acknowledge that Hercules experienced significant difficulties in selecting Gossage as a new CEO, due primarily to a lack of candidates within Hercules' management. See D.I. 163 at 74. [7] According to Litras, supervisors were to rank "new” employees in the middle of the list. D.I. 121 at B94-95. [8] Because Hercules was not aware of the IPI study until after it decided to terminate Finch, the Court will not consider it. See supra note 3. [9] Plaintiff contends these ratings were only intended to determine “the size of a department's and an individual's bonus, not to provide a basis for the termination of an individual.” D.I. 120 at 29 (citing D.I. 121 at B87-88, B93, B134). More importantly, plaintiff notes that this evaluation process was not completed until after the decision had been made to terminate Finch. The evaluation comments were "forwarded to Gos-sage on January 30, 1991, four days after the PCC had approved Finch's termination.” Id. (citing D.I. 121 at B83-86). Because, according to defendant's timetable for the decision process, both the PCC and the Audit Committee had approved the decision to terminate Finch before the *1115 evaluation comments were available, the comments could not have affected Hercules' motivation in terminating Finch. See Mardell, 31 F.3d 1221, 1238 (holding after-acquired evidence is inadmissible in determining liability under the ADEA). Accordingly, the Court will not consider these evaluations as evidence in deciding defendant's motion for summary judgment. [10] Defendant has provided substantial record evidence that Finch and Engebretsen have known each other for over twenty years, are both members of the Church of Jesus Christ of Latter Day Saints [the "Mormons”], and that in his role as a bishop of the Mormon church, Finch counselled Engebretsen’s children. See D.I. 119 at A78, A93-94, A128. Defendant has also put evidence in the record to suggest Engebretsen protected Finch from criticism at Hercules. See id. at A348, A464. Defendant acknowledges, however, that the Court should not consider this evidence in ruling on its motion for summary judgment. D.I. 163 at 65. [11] According to Mark Twain, "[t]here are three kinds of lies: lies, damned lies, and statistics.” Mark Twain’s Autobiography 246 (1924) (remark attributed to Disraeli). [12] Based on Hercules' workforce reduction of approximately 26.5 percent, Daymont selected the bottom 25 percent of the forced rankings to define near the bottom. D.I. 171 at B6; D.I. 119 at A504. [13] A standard deviation is "the square root of the average of the squares of the deviations from the mean.” Webster's Third New Int'l Dictionary 2223 (1971). It is considered the most reliable measure of dispersion. T.G. Connolly & W. Sluckin, An Introduction to Statistics for the Social Sciences 49 (1971). In a normal curve, approximately 95 percent of all cases will fall within plus or minus two standard deviations. Id. at 84. Thus, "where a disparity is larger than 2 standard errors, fewer than 5 percent of the disparities in a random system would be that large, and if the disparity exceeded 3 standard errors we would know that fewer than 1 percent of random selections would result in larger disparities.” David C. Baldus & James W.L. Cole, Statistical Proof of Discrimination § 9.0 (1980). In his report, Daymont also compared the likelihood of an employee age 55 and over being ranked in the bottom 25 percent of the forced rankings with the likelihood an employee under 55 years old would be ranked in the bottom 25 percent. The difference was 2.68 standard deviations. D.I. 171 at B7. Daymont further evaluated the likelihood of older employees being ranked in the bottom 20 percent and 30 percent of the forced rankings. The 20 percent mark resulted in standard deviations of 2.54 for employees 50 and over and 2.75 for employees 55 and over. When comparing the likelihood older employees would be ranked in the bottom 30 percent, Daymont computed standard deviations of 2.82 for employees 50 and over and 3.08 for employees 55 and over. Id. at B8-9. Finally, to account for the vaiying degrees of risk in each Hercules' department, Daymont repeated his analysis based on the RIF target for each department. This process revealed differences of 2.21 standard deviations for employees 50 and over and 2.64 standard deviations for employees 55 and over. Id. at B9. [14] As a formal matter, there is a third category of disparate treatment cases: pure discrimination cases. Mardell, 31 F.3d at 1224. Such a case occurs when, after the plaintiff has established a prima facie case, the employer fails to satisfy its burden of producing a legitimate nondiscriminatory reason for its action. See Texas Dep’t of Community Affairs v. Burdine, 450 U.S. 248, 254, 101 S.Ct. 1089, 1094, 67 L.Ed.2d 207 (1981) (failure to produce a legitimate nondiscriminatoty reason requires the court to enter judgment for the plaintiff). In practice, however, an employer almost always satisfies its burden of production. [15] While Hercules forced ranked Finch using its RIF guidelines, it later reinstituted the position of General Auditor. Defendant acknowledges that whether this case is characterized as a reduction in force case has no legal effect on Finch's prima facie case. D.I. 163 at 28-39. This acknowledgment is consistent with the Third Circuit Court of Appeals' observation that because the prima facie case is easily made out, the prima facie case is rarely the focus of the ultímate disagreement. Rather, "the exigencies of a reduction in force can best be analyzed at the stage where the employer puts on evidence of a non-discriminatory reason for the [discharge].” Healy, 860 F.2d at 1214 n. 1 (quoting Coburn v. Pan American World Airways, Inc., 711 F.2d 339, 343 (D.C.Cir.), cert. denied, 464 U.S. 994, 104 S.Ct. 488, 78 L.Ed.2d 683 (1983)). [16] The Third Circuit Court of Appeals recently has granted argument, in banc, to resolve this dispute. Miller v. CIGNA Corp., No. 93-1773, 1994 WL 420284, 1994 U.S.App. LEXIS 21371 (3d Cir. Aug. 12, 1994). [17] Sections 104 and 105 of the Civil Rights Act of 1991, codified at 42 U.S.C. §§ 2000e, 2000e~ 2(k), overruled Wards Cove by providing that once a plaintiff establishes a prima facie case, the burden of persuasion, as well as the burden of production, shifts to the defendant. At oral argument, Hercules and Finch agreed that the 1991 Act does not affect Finch's claims of disparate impact and disparate treatment. The Court, therefore, will not consider the 1991 Act. [18] While Gossage's statements in the press and at the Hercules' question and answer session provide sufficient direct evidence to create a genuine issue of fact as to whether age was a motivating factor in Hercules’ decision to terminate Finch, the Court will discuss Hosker's testimony for the purpose of guiding the parties in their trial preparation. [19] Likewise, the Court finds Hosker’s memorandum to file, which purports to summarize his conversation with Gossage, is irrelevant. The Court further notes that Finch has not provided the necessary foundation for admitting the memorandum. [20] Because Hosker’s testimony, except for comments from Miller about “tired warriors,” is either irrelevant under Federal Rule of Evidence 401 or unduly prejudicial under Rule 403, the Court limits its consideration of Hercules' hearsay objection to the portion of Hosker’s testimony in which he relates Miller's comments. [21] Rule 801(d)(2)(D) provides that "[a] statement is not hearsay if [t]he statement is offered against a party and is a statement by the party's agent or servant concerning a matter within the scope of the agency or employment, made during the existence of the relationship[.]” Fed.R.Evid. 801(d)(2)(D). [22] Hercules, citing White Industries v. Cessna Aircraft Co., 611 F.Supp. 1049, 1064 (W.D.Mo. 1985), asserts that Miller's repetition to Hosker of Gossage's statement precludes application of Rule 801(d)(2)(D) because "the agent's statement 'must amount to an intended assertion of the fact of the matter by the speaker, rather than being a mere repetition of another's assertion.' " D.I. 159 at 3-4 (emphasis in original). The Court rejects Hercules' position because neither case cited by the court in White Industries, Zenith Radio Corp. v. Matsushita Electric Industrial Co., 505 F.Supp. 1190 (E.D.Pa.1980) or Mahlandt v. Wild Canid Survival & Research Center, Inc., 588 F.2d 626 (8th Cir.1978), supports such a proposition. In Zenith Radio, the court considered whether incomprehensible diary entries which did not clearly assert the truth of any proposition could be admitted under Rule 801(d)(2). 505 F.Supp. at 1240-42. Likewise, in Mahlandt, the court expressly rejected the propositions that Rule 801(d)(2)(D) required either communication to an outsider or the declarant’s personal knowledge. 588 F.2d at 630. The issue, under the plain language Rule 801(d)(2)(D), is whether the statement was made within the scope and during the existence of the agency relationship. While Hercules' objection might be construed as asserting that Miller's repetition of Gossage’s statement was not within the scope of Miller's agency with Hercules, it appears such an objection lacks merit. The record suggests Miller was relaying Gossage’s views about the succession planning of the Specialty Chemical Companies, a process in which both Miller and Hosker were involved. See D.I. 158 Ex. 1 at 37-40. The Court holds, based on the present record, that this statement is nonhearsay. If Finch is unable to demonstrate the relevance of this testimony, however, the Court will not reach the hearsay issue. Finally, Hercules has argued that Hosker is not a reliable source because of his anger about not being promoted. D.I. 163 at 48. This argument is best made to the jury regarding the weight it should accord Hosker’s testimony. Likewise, Hercules can use the differences in the testimony of Miller and Hosker to suggest the jury should not credit Hosker’s testimony. These arguments do not, however, affect admissibility. [23] This holding does not, however, constitute a ruling by the Court that Finch's case will necessarily be entitled to the special burden-shifting treatment of the mixed motives mode of analysis at trial. [24] Finch also has suggested his statistical evidence constitutes direct evidence entitling him to special mixed motives treatment. D.I. 163 at 103. The Court disagrees. While statistical evidence may be used to support a plaintiff's claim of disparate treatment under either a mixed mo *1127 tives or pretext analysis, such evidence does not "directly reflect” a discriminatory attitude. See, e.g., Healy, 860 F.2d at 1217 ("We have approved the use of statistical evidence to support a plaintiff's claim of disparate treatment.”); Griffiths, 988 F.2d at 470 (to qualify for special mixed motives treatment, the plaintiff’s evidence " ‘must be tied directly to the alleged discriminatory animus.” Thus, 'purely statistical evidence would not warrant [shifting the burden];' ”); Ostrowski, 968 F.2d at 182 (purely statistical evidence would not warrant a mixed motives charge); Barnes v. GenCorp, Inc., 896 F.2d 1457, 1469 n. 17 (6th Cir.), cert. denied, 498 U.S. 878, 111 S.Ct. 211, 112 L.Ed.2d 171 (1990) (“statistical proof alone cannot establish the existence of a mixed motive case”). [25] In its briefs and at oral argument, defendant has asserted its decisionmakers "honestly” and "sincerely” believed Finch was a poor performer. See, e.g., D.I. 118 at 40 ("Finch may claim he was a stellar performer but this is beside the point; the evidence is overwhelming that the decisionmakers within the company disagreed and sincerely held the different view discussed above.”); D.I. 134 at 2-3 ("Plaintiff has not addressed the proper issue in this case, i.e. did Hercules’ decisionmaker sincerely believe Finch should be terminated in the RIF_”); D.I. 163 at 66 (“The question is sincerity of belief.”). Citing McCoy v. WGN Continental Broadcasting Co., 957 F.2d 368 (7th Cir.1992), defendant argues that even if its decisionmakers mistakenly believed Finch was a poor performer, Hercules cannot be liable for age discrimination. In McCoy, the Seventh Circuit affirmed the district court's grant of summary judgment for the employer because plaintiff's "efforts to ward off summary judgment by showing pretext bear more on the issue of mistake on WGN’s part than on the issue of whether WGN honestly believed in the reasons it has offered for its actions.” Id. at 373. According to defendant, Finch’s efforts to demonstrate that he was a good performer do not show its proffered reason is a pretext for discrimination; it only suggests Hercules’ deci-sionmakers may have been mistaken. The Court notes that Hercules' position is consistent with the nature of disparate treatment analysis, which is based on a claim that the employer intentionally discriminated against the plaintiff. On the other hand, to require a plaintiff in a pretext case to disprove the employer’s subjective belief would require the type of "smoking gun" evidence which the Supreme Court has recognized is often absent in employment discrimination cases. I conclude that, while an employer should not face liability for intentional discrimination when its decision is based on a good faith mistake, the same is not necessarily true if the plaintiff introduces evidence from which the factfinder could reasonably conclude that the employer's "honest belief" was unreasonable. If the plaintiff introduces evidence sufficient to create a genuine issue of fact as to the reasonableness of the employer’s mistaken belief, it is for the trier of fact to determine whether the employer's belief was so unreasonable as to be pretextual. See St. Mary's, -U.S. at -, 113 S.Ct. at 2749 ("The fectfin-der's disbelief of the reasons put forward by the defendant (particularly if disbelief is accompanied by a suspicion of mendacity) may, together with the elements of the prima facie case, suffice to show intentional discrimination.”). [26] In its opening brief, Hercules also asserted Finch had failed to identify the specific employment practice he was challenging. D.I. 169 at 14-16. In his answering brief, Finch identified the RIF’s forced ranking process as the facially neutral policy producing a disparate impact. D.I. 170. In its reply brief, Hercules' sole challenge to Finch's prima facie case was with regard to Finch's statistical evidence. D.I. 174. The Court concludes, therefore, that Finch has identified a specific employment practice for purposes of establishing his prima facie case. [27] Hercules further argues that if the Court restricts the statistical pool to similarly situated employees by separating exempt and nonexempt employees, the Court must limit the pool to Finch’s department at Hercules because only *1129 those persons were ranked by a common evaluator. D.I. 174 at 7-8. The Court disagrees. Other than arguing such a limitation would be a "logical extension” of the decision to exclude nonexempt employees from the disparate impact analysis, Hercules has produced no legal authority in support of its position. Further, the record indicates there is an issue of fact as to whether Hercules applied the same criteria to exempt and nonexempt employees in its forced ranking decisions; there is nothing in the record to indicate that individual evaluators applied the RIF criteria in a different manner. [28] The parties agree that this issue presents a legal question which must be decided by the Court. [29] Assume a hypothetical company of 200 employees, of whom: 50 are age 50 and over, 100 are in their 40s, and 50 are under age 40. Pursuant to a reduction in force, the company terminates all employees age 50 and over, 25 employees who are in their 40s, and 25 employees under age 40. If the effects of the RIF are analyzed by comparing the likelihood of an employee under 40 being terminated with the likelihood of an employee age 40 and over being terminated, the RIF appears completely age neutral, as there is a 50 percent chance of being terminated in either age group. If, however, one compares the chance of being terminated for employees age 50 and over (100%) with the chance of being terminated for employee under age 50 (33%), the older employees are three times as likely to be terminated, a difference of more than nine standard deviations. D.I. 171 at B10. [30] Any subgroup, of course, must be within the ADEA’s protected class of individuals ages forty to seventy. [31] In fact the incentive Hercules provided to employees who opted for voluntary retirement was the same package of benefits provided to employees who were involuntarily terminated in the RIF. D.I. 121 at B327. [32] Only one of the 25 employees was age 55 or over, making any computation of the impact at ■ that age break meaningless.