06-5458. Office of Federal Contract Compliance Programs; Interpreting Nondiscrimination Requirements of Executive Order 11246 With Respect to Systemic Compensation Discrimination; Notice

AGENCY:

Office of Federal Contract Compliance Programs, Employment Standards Administration, Department of Labor.

ACTION:

Notice of final interpretive standards for systemic compensation discrimination under Executive Order 11246.

SUMMARY:

The Office of Federal Contract Compliance Programs is publishing final interpretive standards for systemic compensation discrimination under Executive Order 11246, as amended. This document sets forth the final interpretive standards and discusses comments that OFCCP received in response to proposed interpretive standards published in the Federal Register on November 16, 2004.

EFFECTIVE DATE:

June 16, 2006.

FOR FURTHER INFORMATION CONTACT:

Director, Division of Policy, Planning, and Program Development, Office of Federal Contract Compliance Programs, 200 Constitution Avenue, NW., Room N3422, Washington, DC 20210. Telephone: [phone] (voice) or [phone] (TTY).

SUPPLEMENTARY INFORMATION:

In this preamble, OFCCP summarizes the proposed interpretive standards, discusses the comments received in response to its publication of the proposed standards, and provides a substantive discussion of the final interpretive standards. The substantive discussion of the final interpretive standards substantially restates the preamble of the proposed standards, except that modifications or clarifications were added in response to the comments.

I. Summary of the Proposed Interpretive Standards

On November 16, 2004, OFCCP published a Notice in the Federal Register [hereinafter “Notice”] in which the agency proposed standards interpreting Executive Order 11246 with respect to systemic compensation discrimination. 69 FR 67246 (Nov. 16, 2004). Systemic compensation discrimination was defined in the Federal Register Notice as discrimination under a pattern or practice, disparate treatment theory of discrimination. 69 FR 67246 n. 2. The Notice explained that OFCCP historically has relied on interpretations of Title VII as a basis for interpreting the nondiscrimination requirements of Executive Order 11246, but that OFCCP had not issued any definitive interpretation of Executive Order 11246 with respect to systemic compensation discrimination. 69 FR 67246-47. The Notice also explained that, in the late-1990s, OFCCP informally used a controversial “pay grade theory” of analyzing compensation practices for systemic discrimination. 69 FR 67247-48. Under the pay grade theory, OFCCP compared the compensation of employees who were in the same pay grade or range, based on the assertion that by creating the pay grade, the employer either “has recognized that certain jobs are essentially similar in terms of skill, effort and responsibility” or “has already identified certain jobs as having similar value to the organization.” 69 FR 67247-48. The Notice provided a detailed discussion of OFCCP's reasons for rejecting the grade theory, primarily because the assumptions underlying the grade theory are inconsistent with administrative and judicial interpretations of Title VII and because use of the pay grade theory proved to be a highly ineffective enforcement tool. 69 FR 67248-49.

The proposed interpretive standards had three principal components. The first component of the proposed interpretive standards was adoption of the “similarly situated” standard for comparisons of employees’ compensation. 69 FR 67249-67252. Under the proposed standards, employees are similarly situated if they perform similar work and occupy positions involving similar responsibility levels, skills, and qualifications. Id. OFCCP interpreted Executive Order 11246 ^[1] with respect to systemic compensation discrimination as involving disparate treatment of individuals who are similarly situated under this standard. 69 FR 67251. In adopting the similarly situated standard, OFCCP relied on judicial and administrative interpretations of Title VII. 69 FR 67248-67249. OFCCP stressed that those interpretations were inconsistent with OFCCP's prior “pay grade” method. 69 FR 67248.

The second component of the proposed interpretive standards was adoption of a statistical technique for assessing the combined effects of the multiple, legitimate factors that influence employers' compensation decisions. 69 FR 67250. This statistical technique is called multiple regression analysis. Id. Under the multiple regression analysis, OFCCP would compare the compensation of similarly situated employees, while controlling for legitimate factors that influenced the employers' pay decisions, such as education, experience, performance, productivity, etc. Id. OFCCP explained that it would investigate whether any such factors were actually “tainted” by discrimination, and, if so, OFCCP would not include such factors in the multiple regression analysis. Id. OFCCP also explained that in a particular case it might use a “pooled” regression, in which different groups of similarly-situated employees were combined in a regression while controlling for their membership in their particular similarly-situated group. 69 FR 67250-67251. When using a pooled regression, OFCCP explained, it would test for whether “interaction terms” were required. 69 FR 67251.

The third component of the proposed interpretive standards was its emphasis on the importance of anecdotal evidence of discrimination for a determination of whether systemic compensation discrimination exists. 69 FR 67251. OFCCP noted that it would rarely issue a Notice of Violations alleging systemic compensation discrimination without anecdotal evidence of discrimination to support the statistical evidence of discrimination. Id.

II. Discussion of the Comments Received

OFCCP received 28 comments on the Notice of proposed standards interpreting Executive Order 11246 with respect to systemic compensation discrimination. In response to the comments, OFCCP made several modifications to the proposed interpretive standards, discussed below. In addition, many of the commenters asked for clarification of OFCCP's intent with respect to various aspects of the interpretive standards, which OFCCP provides as appropriate below.

For the following discussion, OFCCP has grouped the comments around the following major subjects: (A) Systemic Compensation Discrimination; (B) The Pay Grade Theory; (C) Similarly Situated Employees; (D) Multiple Regression Analysis; (E) Factors Included in the Regression Analysis; (F) Anecdotal Evidence; and (G) Confidentiality of Compensation and Personnel Information.

A. Systemic Compensation Discrimination

Several commenters, such as the U.S. Chamber of Commerce and HR Analytical Services, Inc., argued that OFCCP should not focus its efforts on investigating systemic employment discrimination, but should instead spend more agency resources on monitoring compliance with OFCCP's affirmative action regulations. OFCCP does not agree with these commenters. OFCCP believes that elimination of systemic workplace discrimination is an important component of its historical mission. Indeed, affirmative action programs are designed to be tools to prevent workplace discrimination. See 41 CFR 60-2.10(a)(3) (“OFCCP has found that when an affirmative action program is approached from this perspective, as a powerful management tool, there is a positive correlation between the presence of affirmative action and the absence of discrimination.”). Further, the commenters' suggestion disregards OFCCP's historical enforcement of Executive Order 11246 by requiring payment of back pay and other make whole relief to victims of discrimination. See 41 CFR 60-1.26(a)(2) (“OFCCP may seek back pay and other make whole relief for victims of discrimination identified during a complaint investigation or compliance evaluation.”). OFCCP's focus on finding and remedying systemic workplace discrimination has provided tangible incentives for contractors to implement affirmative action programs to prevent workplace discrimination.

B. The Pay Grade Theory

Almost all of the commenters addressed the subject of OFCCP's prior “pay grade” method as discussed in the preamble of the proposed standards. Many commenters agreed with OFCCP that the pay grade theory was inconsistent with Title VII standards.^[2]

A few commenters, such as Jude Sotherlund, argued that OFCCP should rely on employer-created classifications such as pay grades because these classifications were designed by compensation professionals for the particular employer. OFCCP does not agree with these comments. Unlike compensation professionals, who design compensation systems to meet a variety of business interests, OFCCP's purpose when investigating an employer's compensation practices is to determine whether the employer has engaged in systemic compensation discrimination prohibited by Executive Order 11246. As noted below, EEOC and courts interpreting Title VII have cautioned against reliance on employer classifications in favor of evidence of actual work activities, responsibility level, and skills and qualifications involved in the job.

A few other commenters, including the Employment Task Force of the Leadership Conference on Civil Rights (ETF), argued against OFCCP's conclusion that the pay grade theory should be rejected because it is inconsistent with Title VII. ETF, for example, generally offered two sets of arguments against OFCCP's rejection of the grade theory.

In the first set of arguments, ETF argued that pay grade information can be an effective indicator of potential pay discrimination. ETF noted that “the pay grade approach serves as a unique investigatory tool” and “provided a suitable starting point for investigators to determine which jobs to compare and analyze.” ETF questioned, “[i]f the pay grade approach is to be abandoned, it is unclear from these proposed standards how OFCCP intends to utilize its limited resources to identify the appropriate cases for further investigation and enforcement.” Several other commenters also expressed concerns about the burden to employers and to the agency if OFCCP conducts the investigation and analysis required by the proposed standards in each compliance review.^[3] OFCCP agrees with ETF that pay grade information has some value as an indicator of potential discrimination. OFCCP also agrees with ETF and the other referenced commenters that the agency does not desire to conduct a full-scale compensation investigation in every compliance review. Thus, the interpretive standards are not intended to restrict OFCCP's use of pay grade information or any other information as an indicator of potential discrimination. Rather, the interpretive standards only foreclose the use of the pay grade theory as the basis upon which OFCCP will allege and establish systemic compensation discrimination in violation of Executive Order 11246 and OFCCP regulations. Indeed, OFCCP has historically used a tiered-review approach in its evaluation of contractors that relies on both pay grade information and individual employee information to determine whether to conduct a comprehensive investigation into the contractor's pay practices. Under the tiered-review approach, OFCCP uses pay grade (or other aggregated compensation) information submitted in response to Item 11 of OFCCP's Scheduling Letter.^[4] Once it receives the Item 11 data, OFCCP conducts a simple comparison of group average compensation by pay grade or other aggregation unit by which the employer has provided the data. If this comparison indicates a significant disparity, OFCCP will ask the contractor for employee-specific compensation and personnel information.^[5] OFCCP intends to continue this tiered-review approach ^[6] and, in fact, recently implemented additional components to further focus compensation investigations on workplaces where there are significant indicators of potential discrimination. In particular, OFCCP now conducts a “cluster regression” using the employee-specific information requested following the desk audit.^[7] If the cluster regression indicates significant disparities, OFCCP conducts a comprehensive evaluation of the pertinent compensation practices, at which point these final interpretive standards govern OFCCP's investigation activity and determinations. OFCCP will afford the contractor an opportunity to provide any additional information and/or analyses that the contractor believes to be pertinent to OFCCP's decision about whether to conduct further investigation of the contractor's compensation practices. OFCCP will consider such information as well as the results of the cluster regression in making a determination of whether further investigation is warranted. Of course, OFCCP will also consider any evidence of discrimination in determining whether to proceed.

Accordingly, OFCCP intends to continue using analysis of pay grade information, supplemented by the cluster regression, as indicators of potential compensation discrimination. However, the pay grade analysis, the cluster regression analysis, and other generalized approaches are only indicators of potential compensation discrimination. These techniques fall far short of the type of fact-intensive investigation and tailored analysis required to make and sustain an allegation of systemic compensation discrimination under Executive Order 11246 and OFCCP regulations. These final interpretive standards fit into the latter part of the OFCCP compliance review process: They serve as the substantive standards interpreting Executive Order 11246 and OFCCP regulations with respect to systemic compensation discrimination. In practical terms, this means that OFCCP must allege and prove facts which meet the interpretive standards in order to establish systemic compensation discrimination in violation of Executive Order 11246 and OFCCP's regulations.

ETF also objected to the provisions of the proposed interpretive standards which mandated prerequisites to issuing a Notice of Violation (NOV). ETF argued that OFCCP should not subject itself to a standard during the “investigatory stage” that is the same standard that OFCCP would be subject to when it pursued enforcement litigation.^[8] OFCCP agrees that its investigations need not adhere to the precise requirements of enforcement litigation in order to issue an NOV. For example, OFCCP need not base its decision to issue an NOV on information that has been obtained in a format which would be admissible in court, e.g., OFCCP can rely on notes of an employee interview during an investigation which may not be admissible in litigation. However, OFCCP disagrees that the substantive standards for whether an employment practice constitutes a violation of Executive Order 11246 can depend on whether the matter is in the “investigation stage” or in litigation. If the pay grade theory assumptions (discussed in the preamble of the proposed interpretive standards and below) do not adhere to legal standards, OFCCP has no authority to rely on such assumptions to allege a violation even during the investigation stage. Because the pay grade assumptions are contrary to legal standards, to base a violation on the pay grade theory during the investigation stage is tantamount to changing the substantive requirements of Executive Order 11246.

ETF offered additional arguments against OFCCP's rejection of the pay grade theory. These arguments were premised on a correct understanding that the interpretive standards ruled out the pay grade theory as a basis for alleging and establishing systemic compensation discrimination under Executive Order 11246 and OFCCP regulations. First, ETF argued that OFCCP should continue to use the pay grade theory, suggesting that it is consistent with interpretations of Title VII. Second, ETF argued that the Title VII cases OFCCP cited do not require rejection of the pay grade theory because the plaintiffs failed in the cited cases when they were unable “to provide additional evidence where employers have put forward a legitimate nondiscriminatory reason.” In this regard, ETF noted that, “[w]hile pay grade information may not have been enough to win these particular cases, such information was clearly instrumental in establishing possible discrimination in the first place.” Finally, ETF argued that the rejection of the pay grade theory could harm or curtail future enforcement efforts or developments in the law.

OFCCP does not find ETF's comments to be persuasive reasons for retaining the pay grade theory as a basis for alleging and establishing systemic compensation discrimination under Executive Order 11246 and OFCCP regulations. As to ETF's argument that OFCCP should continue to rely on the pay grade theory to establish systemic compensation discrimination, OFCCP believes that the pay grade theory was inconsistent with Title VII standards and that there are compelling reasons for ensuring that the nondiscrimination provisions of Executive Order 11246 are interpreted consistently with Title VII. First, this has been OFCCP's historical practice, as well as the practice of the Department of Labor in rendering final agency decisions in cases arising under Executive Order 11246. See note 29, below; see also OFCCP Federal Contract Compliance Manual, at Section 3K00(c) (“It is OFCCP policy, in conducting analyses of potential discrimination under the Executive Order, to follow Title VII principles.”).^[9] Second, OFCCP expects that the federal courts will look to Title VII interpretations when interpreting the nondiscrimination requirements of Executive Order 11246. This is a significant consideration in light of the fact that Department of Labor determinations under Executive Order 11246 are subject to review in federal court under the Administrative Procedure Act. Thus, federal courts are likely to defer to these final interpretive standards because they accord with the weight of authority under Title VII, in addition to deference under traditional deference doctrines. See Barnhart v. Walton, 535 U.S. 212, 217 (2002) (“Courts grant an agency's interpretation of its own regulations considerable legal leeway”); Auer v. Robbins, 519 U.S. 452, 461 (1997) (agency's interpretation of its own regulation is “controlling unless ‘plainly erroneous or inconsistent with the regulation,’ ” quoting Bowles v. Seminole Rock Co., 385 U.S. 410, 413-14 (1945)); Udall v. Tallman, 380 U.S. 1, 16-17 (1965) (agency interpretations of Executive Orders they are charged with enforcing are afforded deference under Bowles v. Seminole Rock Co., 385 U.S. 410, 413-14 (1945)); Reynolds v. Rumsfeld, 564 F.2d 663, 668 (4th Cir. 1977) (OFCCP interpretation of Executive Order 11246 entitled to Seminole Rock deference).

Third, this policy ensures uniformity and consistency with the principal congressional enactment on equal employment opportunity, and with EEOC enforcement standards. OFCCP relied expressly and extensively on the EEOC Compliance Manual chapter on compensation discrimination in developing the interpretive standards. In addition, the EEOC provided written comments for the public record in which EEOC stated, “we are pleased that your approach to addressing compensation discrimination is consistent with EEOC's own view.”

OFCCP also does not agree with ETF's characterization of the authority cited in the preamble of the proposed interpretive standards. First, ETF's comments conflict with the EEOC compensation guidelines, which expressly adopt the “similarly situated” standard. EEOC Compliance Manual on “Compensation Discrimination,” EEOC Directive No. 915.003 (Dec. 5, 2000)[hereinafter, “CMCD”], at 10-5 to 10-8 (“The investigator should determine the similarity of jobs by ascertaining whether the jobs generally involve similar tasks, require similar skill, effort, and responsibility, working conditions, and are similarly complex or difficult.”).

Second, OFCCP does not agree that the plaintiffs in “virtually all” of the cases cited in the preamble of the proposed interpretive standards were able to establish a prima facie case by comparing themselves to individuals who did not perform similar work and whose positions were not similar in the responsibility level, skills, and qualifications involved. It has long been established that plaintiffs must demonstrate that similarly situated employees were treated differently as part of their own prima facie case. See Texas Dep't of Community Affairs v. Burdine, 450 U.S. 248, 258 (1981) (“McDonnell Douglas teaches that it is the plaintiff's task to demonstrate that similarly situated employees were not treated equally.”); see also Quarless v. Bronx Lebanon Hosp. Ctr., 228 F. Supp.2d 377, 383 (S.D.N.Y. 2002) (“In order to establish a prima facie case of discriminatory disparate pay under Title VII, a plaintiff must show * * * that he was paid less than similarly situated non-members of his protected class; * * *”) aff'd, 75 Fed. Appx. 846, 848 (2d Cir. 2003); Lewis v. Smith, 255 F. Supp.2d 1054, 1060-61 (D. Ariz. 2003) (“Plaintiff can establish a prima facie case under Title VII because he can show that * * * he was given greater or similar responsibilities but paid less than [a coworker] who occupied a similar, if not substantially equal, position.”). Indeed, in many of the cited cases, the plaintiffs were unable to establish a prima facie case precisely because they attempted to compare themselves to individuals whose work, responsibility level, and skills and qualifications were not similar to their own. See, e.g., Block v. Kwal-Howells, Inc., No. 03-1101, 2004 WL 296976, at *2-*4 (10th Cir. Feb. 17, 2004) (“The district court concluded Ms. Block failed to establish a prima facie case of discrimination because she failed to prove she occupied a substantially similar position to Mr. Dennis. Aplt. Br., Att. A. at 26. Upon a thorough review of the evidence, we agree. Ms. Block and Mr. Dennis were not similarly situated.”); Williams v. Galveston Ind. Sch. Dist., No. 03-40436, 78 Fed. Appx. 946, 949-50, 2003 WL 22426852 (5th Cir. Oct. 23, 2003) (“Appellants attempt to found their prima facie case on a comparison between their positions and the positions held by Mr. McLarty and Ms. Garcia. However, each employee's responsibilities are plainly dissimilar from the responsibilities of the other three grade 8 employees * * *. The fact that GISD lists all four employees at grade 8 is not significant. Pay grades represent a range of possible salaries, and Appellants concede that salaries can differ within a pay grade.”) ^[10] ; Verwey v. Illinois Coll. of Optometry, 43 Fed. Appx. 996, 2002 WL 1836507, at *4 (7th Cir. Aug. 9, 2002) (“Verwey also argues that the district court erred in granting summary judgment to the College on her wage discrimination claim. She asserts that she raised an inference of discrimination by showing that the three maintenance men in her department received raises after voting against unionizing, but that she, the lone female employee, did not. Verwey's claim fails for several reasons. First, she did not establish that the maintenance men were similarly situated to her. Although they worked in the same department, they had different job titles and responsibilities and therefore did not hold equivalent positions; Verwey was an administrative assistant, not a maintenance worker.”); Rodriguez v. SmithKline Beecham, 224 F.3d 1, 8 (1st Cir. 2000) (“As we set forth above, the uncontested facts before the district court indicate that appellant's job functions and responsibilities were not substantially similar or comparable to those of Document Manager Llivina or Records Management Leader Feo, nor to those of Edwin López. Absent such a showing, plaintiff's Title VII claim fails as a matter of law for lack of a prima facie case.”); Sprague v. Thorn Americas, Inc., 129 F.3d 1355, 1362 (10th Cir. 1997) (“It is apparent from the record that Sprague failed to present genuine issues of material fact which would support her equal pay claim under Title VII. As the district court observed, Sprague contrasts her functions and pay in the jewelry department to those of the assistant product manager of electronics and the assistant product manager of furniture/appliances, both of whom are males. ‘However, the Electronics, Furniture/Appliances, and Jewelry Departments do not contribute equally to [Thorn's] revenues.’ See district court's Memorandum and Order at 5. While the electronics department comprises approximately 50% of revenues and the furniture/appliance department accounts for approximately 45% of revenues, the jewelry department only produces approximately 4% of revenues. Id. * * * Given the evidence presented to the district court, we find that Sprague failed to present a prima facie case of intentional gender discrimination.”); EEOC v. Sears, Roebuck & Co., 839 F.2d 302, 347 (7th Cir. 1988) (“As it turns out, the EEOC's failure to introduce any evidence of actual job content or job performance is fatal to its sex discrimination in wages claim in light of Sears' evidence regarding differences in job content. The EEOC appears to suggest that Sears had the burden of showing the inequality of job content. This line of argument is similar to that which we recognized in Epstein, 739 F.2d at 278: ‘Plaintiff would, it seems, have us infer equal work from the defendants’ failure to prove otherwise.’ We responded that this argument ignores the elementary fact that the burden for proving the prima facie case is on the plaintiff.”); Eastland v. Tennessee Valley Auth., 704 F.2d 613, 624-25 (11th Cir. 1983) (“In the present case Eastland's analyses account for many objective qualifications, but the failure to control for job category casts doubt on whether the regressions are comparing appropriate groups. Given the weakness of the theoretical foundation and the failure to control for job category, the district court did not err in determining that Eastland's regressions were insufficient to establish a prima facie case.”); Lawton v. Sunoco, Inc., No. 01-2784, 2002 WL 1585582, at *7 (E.D. Pa. Jul 17, 2002) (“In order to establish a prima facie case of wage discrimination under Title VII * * * the plaintiffs ‘must demonstrate that they were performing work substantially equal to that of white employees who were compensated at higher rates than they were,’ ” quoting Aman v. Cort Furniture Rental Corp., 85 F.3d 1074 (3d Cir. 1996), but also citing Watson v. Eastman Kodak Co., 235 F.3d 851 (3d Cir. 2000), for “similarly situated” standard).^[11]

ETF's arguments also do not address the fundamental point for which OFCCP cited these cases. OFCCP relied on these cases to identify the factors that courts use to determine whether employees are similarly situated in compensation discrimination claims under Title VII. Under the pay grade theory, OFCCP took the position that employees included in the same pay grade were necessarily similarly situated, without regard to their actual job duties, responsibility levels, and skills and qualifications, and OFCCP persisted in that position, even threatening enforcement action, regardless of the evidence the employer submitted about differences in job duties, responsibility levels and skills and qualifications. Indeed, the defining feature of the pay grade theory was its assumption that employees were similarly situated based solely on the fact that they were included in the same pay grade (or that they were in the same pay grade and their pay could progress to the top of the pay grade without changing jobs). OFCCP has rejected the pay grade theory because it conflicts with courts' interpretations of Title VII.

As noted earlier, ETF expressed concern regarding the stage of the case in which the similarly situated issue arises. However, ETF did not expressly endorse the pay grade assumptions that individuals are similarly situated because they are in the same pay grade. Thus, there are not substantial differences between the final interpretive standards and ETF's position. As noted below, in a particular case the pay grade could coincidentally group employees who in fact performed similar work, and occupied positions involving similar responsibility levels, skills, and qualifications. However, what would make such employees similarly situated is the fact that that they perform similar work and occupy positions involving similar responsibility levels, skills and qualifications, not the fact that they are in the same pay grade. Moreover, ETF apparently accepts that an employer could always justify pay differentials between employees who occupy the same pay grade through evidence that the employees are not similar with respect to the work they perform, their responsibility levels, or the skills and qualifications involved in their positions.^[12]

OFCCP disagrees with ETF's last argument, that the agency should not promulgate the final interpretive standards because they could harm or curtail future enforcement efforts and development of the law. In fact, OFCCP's experience demonstrates that just the opposite is true. OFCCP believes that it is important for the agency to promulgate a definitive interpretation of Executive Order 11246 and OFCCP regulations with respect to systemic compensation discrimination. Most significantly, these final interpretive standards will promote compliance with Executive Order 11246 by helping agency personnel and covered contractors and subcontractors understand the meaning of Executive Order 11246 and OFCCP regulations with respect to systemic compensation discrimination. OFCCP personnel will be guided by written standards which will promote uniformity in OFCCP's enforcement of Executive Order 11246. Together with the Voluntary Self-Evaluation Guidelines, these interpretive standards will help contractors with developing programs for monitoring their own compensation practices. OFCCP also believes these interpretive standards will ensure that OFCCP's enforcement efforts are effective, by providing standards that are consistent with administrative and judicial interpretations of Title VII. In fact, OFCCP has been successful in pursuing systemic compensation discrimination cases under standards quite similar to the standards articulated in these final interpretive standards. In the last three years, OFCCP pursued enforcement litigation in two cases using multiple regression analyses that did not rely on the grade theory. These were the first two compensation cases OFCCP has filed in twenty-five years, and both cases resulted in significant settlements, including a near record $5.5 million settlement. By contrast, OFCCP did not pursue even one case through enforcement litigation during the period in which the agency relied on the grade theory. OFCCP does not believe that it will be effective in establishing and remedying systemic compensation discrimination unless contractors perceive that OFCCP's methods will support a credible threat of successful enforcement litigation.

In sum, OFCCP agrees with ETF that grade information can be useful as an indicator of potential compensation discrimination, and OFCCP intends to continue to use grade information to target agency resources on workplaces where further investigation is warranted. However, OFCCP disagrees with ETF that the grade theory is consistent with Title VII standards or that the grade theory is an efficient and effective method for OFCCP to accomplish its important mission.

C. Similarly Situated Employees

Many commenters approved of OFCCP's proposed interpretive standards for defining similarly-situated employees.^[13] However, several commenters, such as Ellen Shong & Associates, Gaucher Associates, and Society for Human Resource Management (SHRM), argued that OFCCP should adopt the Equal Pay Act standard of “substantial equality” instead of the “similarly situated” standard. OFCCP does not agree with these commenters. As noted, OFCCP has historically relied on interpretations of Title VII to interpret the nondiscrimination requirements of Executive Order 11246. Many courts and the EEOC have interpreted Title VII to allow comparisons of individuals who are “similarly situated” as defined in these final interpretive standards.^[14]

Several commenters, such as TOC Management Services, questioned whether the proposed paragraph 7 of the Standards for OFCCP Evaluation of Contractors' Compensation Practices conflicted with OFCCP's adoption of the similarly situated standard. Proposed paragraph 7 stated that “OFCCP will also assert a compensation discrimination violation if the contractor establishes compensation rates for jobs (not for particular employees) that are occupied predominantly by women or minorities that are significantly lower than rates established for jobs occupied predominantly by men or non-minorities, where the evidence establishes that the contractor made the job wage-rate decisions based on the sex, race or ethnicity of the incumbent employees that predominate in each job.” In response to the comments, OFCCP added a footnote to paragraph 7 of the “Standards for OFCCP Evaluation of Contractors’ Compensation Practices” in the final interpretive guidelines to make clear that the intent of paragraph 7 was not to permit a systemic compensation discrimination theory based on comparison of employees who were not similarly situated. Rather, the intent is simply to permit the type of unique compensation discrimination claim approved of in County of Washington v. Gunther, 452 U.S. 161, 166 (1981) (“[R]espondents seek to prove, by direct evidence, that their wages were depressed because of intentional sex discrimination, consisting of setting the wage scale for female guards, but not for male guards, at a level lower than its own survey of outside markets and the worth of the jobs warranted.”). Unlike the systemic compensation discrimination standards set forth in the final interpretive standards, which involve comparisons of the compensation of similarly-situated employees using multiple regression to control for the joint contributions of the various legitimate factors that influence compensation, the Gunther-type claim “does not attempt by statistical technique or other method to quantify the effect of sex discrimination on the wage rates.” 452 U.S. at 181 & n. 20 (citing Franklin M. Fisher, Multiple Regression in Legal Proceedings, 80 Colum.L.Rev. 702, 721-725 (1980)).^[15]

Several of the commenters who agreed that similarity in job duties, responsibility level, and skills/qualifications is a necessary condition for employees to be similarly situated,^[16] also argued that similarity in these factors is not a sufficient condition for employees to be similarly situated in all cases. These commenters argued that there may be other factors in particular cases that may make individuals dissimilar who would otherwise meet the proposed standard for similarly situated. For example, these commenters noted that otherwise similarly-situated employees may be paid differently for a variety of reasons: They work in different departments or other functional divisions of the organization with different budgets or different levels of importance to the business; they fall under different pay plans, such as team-based pay plans or incentive pay plans; they are paid on a different basis, such as hourly, salary or through sales commissions; some are covered by wage scales set through collective bargaining, while others are not; they have different employment statuses, such as full-time or part-time; etc. OFCCP agrees with these commenters that such factors may be important to whether employees are similarly situated in a particular case. See, e.g., CMCD, at 10-6 (“[T]he fact that employees work in different departments or other organizational units may be relevant, but is not controlling.”); see also Cooper v. Southern Co., 390 F.3d 695, 717 (11th Cir. 2004) (noting that plaintiffs' expert “did not tailor her analysis to the specific positions, job locations, or departmental or organizational structures in question; however, the wide-ranging and highly diversified nature of the defendants' operations requires that employee comparisons take these distinctions into account in order to ensure that the black and white employees being compared are similarly situated”); Goodwin v. General Motors Corp., 275 F.3d 1005, 1012 n.8 (10th Cir. 2002) (holding employees similarly situated for compensation discrimination claim under Title VII because “[a]ll four representatives had the same supervisor, performed identical job duties and were subject to the same company standards and policies”); Webb v. Merck & Co., Inc., 206 F.R.D. 399, 408 (E.D. Pa. 2002) (“We agree with defendant that [the plaintiffs” expert's] analysis of hourly (union) workers is unreliable and irrelevant because it fails to control for the mandated wage rate set by collective bargaining agreements for an employee's position * * *”). OFCCP has added provisions (Paragraph 2 of the “Standards for Systemic Compensation Discrimination Under Executive Order 11246” and Paragraph 3 of the “Standards for OFCCP Evaluation of Contractors’ Compensation Practices”) to the final standards to make clear that the agency will consider the applicability of such additional factors in each case and make a determination based on the facts of the particular case.

Several commenters, including ETF and National Industry Liaison Group (NILG), noted that the proposed interpretive standards were ambiguous about whether similarity of qualifications involves similarity in qualifications required for the position or similarity of qualifications possessed by the individual employees who hold the position. ETF noted that the EEOC Compliance Manual chapter on compensation discrimination relies on the qualifications for the position, not the qualifications of the particular employees. OFCCP agrees with ETF that it is the qualifications involved in the position, not the qualifications of the individuals who occupy the position, that determine whether employees are similarly situated under these final interpretive standards. See CMCD, at 10-7. However, OFCCP generally will consider qualifications of the individuals as an explanatory factor in a regression model because superior qualifications are a legitimate reason for pay differences between similarly-situated employees. Id.; see also Goodwin v. General Motors Corp., 275 F.3d 1005, 1012 n.8 (10th Cir. 2002) (noting in context of disparate treatment compensation discrimination claim under Title VII that plaintiff had superior qualifications to similarly situated male employees: “And Goodwin was one of just two who had master's degrees.”); Klindt v. Honeywell Int'l Inc., 303 F. Supp.2d. 1206, 1223 (D. Kan. 2004) (employer not precluded from considering superior educational qualifications in determining employees' salaries).

Several commenters, such as SHRM and HR Analytical Services, requested that OFCCP provide more guidance on how the agency intends to determine whether employees are similarly situated. OFCCP agrees that further clarification of this issue will be helpful to interested parties. OFCCP intends to gather information on employees' job duties, responsibility levels, and skills and qualifications, and other pertinent factors (as discussed above) through review of job descriptions and interviews of employees, managers, and HR and compensation personnel. Once OFCCP has gathered such information, it will determine which individuals are similarly situated by assessing the information under the standard for similarly situated set forth in these final interpretive standards. Since the final interpretive standards rely on federal court interpretations of Title VII, OFCCP will review applicable caselaw as an aid to making such determinations in particular cases. This review of caselaw typically will involve research for cases that discuss positions that are factually similar to the positions at issue in OFCCP's investigation.^[17] OFCCP will review the reasoning and determinations of the courts in such factually-similar cases for guidance in making a determination on the facts before OFCCP.

Several commenters expressed concern that OFCCP would be forced to group dissimilar employees in order to create groupings of sufficient size for statistical analysis, especially in light of OFCCP's stated desire to cover “most” or “a significant number of” employees.^[18] Several of these commenters also requested that OFCCP explicitly acknowledge that certain employees, such as high-level executives, are unique and are not similarly situated to any other employees. OFCCP agrees with these commenters that it may be expected that certain employees are not similarly situated to any other employee in the organization, workplace, or AAP. Under no circumstances will OFCCP attempt to combine, group, or compare employees who are not similarly situated under these final interpretive standards. If employees are not similarly situated under these final interpretive standards, they will not be included in the statistical analysis, regardless of statistical size requirements or of OFCCP's general objective to include a significant majority of employees in the regression analyses.^[19]

Several commenters, including Equal Employment Advisory Council (EEAC) and ORC Worldwide (ORC), expressed concern with OFCCP's stated intent to review job descriptions and conduct employee interviews to determine whether employees are similarly situated. These commenters noted that job descriptions are often outdated and inaccurate. Several commenters requested that OFCCP also interview managers or supervisors to determine which employees are similarly situated. OFCCP agrees with these commenters that it will be important for agency staff to interview supervisors, managers, and HR and compensation personnel to obtain information needed to determine whether employees are similarly situated, as well as to obtain other pertinent information about the employer's compensation practices.

D. Multiple Regression Analysis

Many commenters agreed that multiple regression analysis is a legally and statistically valid method for evaluating systemic compensation discrimination.^[20] However, several commenters, such as Ellen Shong & Associates, Peopleclick Research Institute (PRI), and David W. Peterson, argued that OFCCP's proposed regression analysis is inaccurate because it does not evaluate pay and personnel decisions directly (or indirectly through a “pay progression study”), but compares employees’ compensation at a particular point in time. OFCCP does not agree with these commenters that multiple regression analysis of current compensation is legally or statistically deficient. Indeed, the Supreme Court has approved of such analysis. See Bazemore v. Friday, 478 U.S. 385, 400 (1986). Without expressing any view as to whether the types of analysis that these commenters suggest may also be legally and statistically acceptable,^[21] OFCCP does not believe that such analysis is preferable to the approach outlined in the final interpretive standards, for two reasons. First, the analysis suggested by the commenters would require OFCCP to gather far more information than required by the regression analysis outlined in these final interpretive standards. For example, under the commenters' approach, OFCCP would have to identify the variety of personnel decisions that influenced employees' compensation over a significant period of time and, as to each decision, evaluate whether the employer treated the employee similarly to other employees who were similarly situated with respect to that particular decision. This would impose significant burdens both on OFCCP and on contractors during OFCCP's investigation to obtain the information needed for the suggested analysis. Second, the commenters' suggested analysis would combine pay, promotion, and perhaps other personnel decisions in the same analysis, making it difficult to define the nature of the alleged discrimination or to determine an appropriate remedy.

Many commenters expressed concern about the complexity of multiple regression analysis and the burden of collecting the data required for such analysis.^[22] Others were concerned that they would need to hire statisticians or other experts.^[23] OFCCP understands that multiple regression analysis is complicated and requires significant compensation and personnel information. However, because OFCCP will use the analysis as a basis for alleging and establishing systemic compensation discrimination, the agency believes that it must conduct an analysis that meets legal and statistical standards. Indeed, the pay grade method undoubtedly was simple, but OFCCP could not prove systemic compensation discrimination by using that method because it did not adhere to legal and statistical standards and it was widely criticized by contractors for those reasons. Thus, there is a natural tension between the accuracy of the analysis and the complexity and burden associated with it. As discussed above, OFCCP has attempted to balance these competing factors by using a tiered-review approach, in which a multiple regression analysis is conducted only after less complex and less intrusive analyses reveal indicators of potential discrimination. Moreover, OFCCP, not the contractor, has the burden of gathering data and conducting the multiple regression analyses. Contractors need not convert their data to electronic format for purposes of a compliance evaluation. If the data is already in electronic format, OFCCP will use it, but if not, OFCCP has the responsibility of taking the raw data and converting it into an electronic format which can be used in the regression analyses. Similarly, contractors are not required to hire experts to conduct the multiple regression analyses, OFCCP will conduct the multiple regression analyses.

Several commenters, such as EEAC and SHRM, requested that OFCCP provide more guidance about how the agency will determine whether to use a pooled regression model.^[24] OFCCP's determination will be based on the general objectives of attempting to cover as many employees as possible—in light of prohibitions on combining or comparing employees who are not similarly situated—and statistical requirements about the size of employee groupings necessary to conduct a meaningful regression analysis. As noted above, OFCCP will not compare employees who are not similarly situated as defined in these final interpretive standards. OFCCP added text to provisions (Paragraph 5 of “Standards for Systemic Compensation Discrimination Under Executive Order 11246” and Paragraph 5 of “Standards for OFCCP Evaluation of Contractors' Compensation Practices”) of the final standards which make clear that pooled regressions must contain category factors that are defined to group only similarly-situated employees as defined in these standards. The pooled regression model affords OFCCP flexibility to conduct an analysis controlling for groupings of similarly-situated employees. However, OFCCP does not intend to use the pooled regression model on a widespread basis as a preferred approach.

Several commenters, including Northeast Region Corporate Industry Liaison Group (NRCILG) and Association of Corporate Counsel (ACC), argued that OFCCP should provide the contractor with the regression model, not just the results of the regression model, in support of any NOV containing an allegation of systemic compensation discrimination. OFCCP agrees that providing such information to contractors will permit the agency to conciliate alleged violations effectively and expeditiously. OFCCP will provide the contractor with enough information about OFCCP's regression model for the contractor to understand the basis for OFCCP's determinations and for the contractor to replicate OFCCP's regression model. OFCCP has revised the interpretive standards (at Paragraph 2 of “Standards for OFCCP Evaluation of Contractors’ Compensation Practices”) to provide that OFCCP will attach such information to NOVs which contain an allegation of systemic compensation discrimination. With such information, contractors have an opportunity to discuss settlement with OFCCP or to attempt to rebut OFCCP's determination.

Several commenters raised technical statistical issues regarding OFCCP's discussion of multiple regression analysis. PRI and David W. Peterson argued that OFCCP should include all interaction terms when using a pooled regression model, not just interaction terms that are statistically significant. These comments raise a statistical controversy regarding factor reduction techniques in regression analysis. While some statisticians disagree on the use of automated stepwise regression techniques to eliminate insignificant factors, most agree that some form of variable reduction is appropriate. As PRI noted, factors which are individually insignificant may in combination have a significant impact on the regression results. However, OFCCP considers there to be greater risks with full-factor modeling procedures. In particular, especially in the analyses of smaller workforces, the statistical precision in the measured disparities decreases as more factors are added to the analysis. As such, if several inconsequential factors are added to the analysis, they will lessen the ability to measure any gender or racial disparities. Furthermore, as the number of factors increases so does the possibility of a statistical problem called “multicollinearity,” which can produce inaccurate results. See Daniel L. Rubenfeld, Reference Guide on Multiple Regression, in Federal Judicial Center, Reference Manual on Scientific Evidence, at 197 (2d ed. 2000) (“When two or more variables are highly, but not perfectly, correlated—that is, when there is multicollinearity—the regression can be estimated, but some concerns remain. The greater the multicollinearity between two variables, the less precise are the estimates of individual regression parameters (even though there is no problem in estimating the joint influence of the two variables and all other regression parameters).”).

Several commenters questioned OFCCP's adoption of a two standard deviation threshold for assessing statistical significance. Some commenters, including ACC, noted that the caselaw is more nuanced and does not support a bright-line rule. OFCCP recognizes that the courts have not announced an exact threshold for statistical significance. However, OFCCP has determined that it is helpful to adopt a bright-line rule of two standard deviations as an enforcement standard based on the need for uniformity and predictability in this area.

Several commenters, including NILG, noted that statistical significance is dependent on sample size and questioned whether OFCCP would take that fact into consideration. OFCCP notes that standard tests for statistical significance already take sample size into account. Since smaller samples have a higher degree of variation, they require a larger observed disparity to achieve statistical significance. OFCCP recognizes when sample sizes become very large, small and potentially non-meaningful disparities may be found to be statistically significant at the two or higher standard deviation threshold. See Daniel L. Rubenfeld, Reference Guide on Multiple Regression, in Federal Judicial Center, Reference Manual on Scientific Evidence, at 181 (2d ed. 2000) (“Other things being equal, the statistical significance of a regression coefficient increases as the sample size increases. Thus, a $1 per hour wage differential between men and women that was determined to be insignificantly different from zero with a sample of 20 men and women could be highly significant if the sample were increased to 200. Often, results that are practically significant are also statistically significant. However, it is possible with a large data set to find statistically significant coefficients that are practically insignificant. Similarly, it is also possible (especially when the sample size is small) to obtain results that are practically significant but statistically insignificant.”); see also David H. Kaye & David A. Freedman, Reference Guide on Statistics, in Federal Judicial Center, Reference Manual on Scientific Evidence, at 127 (2d ed. 2000) (“Significance depends not only on the magnitude of the effect but on the sample size. Thus significant differences are evidence of something besides random error is at work, but they are not evidence that this ‘something’ is legally or practically important. Statisticians distinguish between ‘statistical’ and ‘practical’ significance to make that point. When practical significance is lacking—when the size of a disparity or correlation is negligible—there is no reason to worry about statistical significance.”).

Several commenters, including HR Analytical Services and Northeast Region Corporate Industry Liaison Group, requested that OFCCP provide, post online, or otherwise make available to contractors, the statistical software that OFCCP will use in evaluating whether contractors engaged in systemic compensation discrimination. OFCCP uses SAS software, which was purchased through the normal procurement process. Other software may be available to perform the evaluation. This listing does not constitute any endorsement of SAS software, but rather is provided pursuant to several commenters' requests.

Several commenters, including NILG and SHRM, requested that OFCCP provide a grace period or a pilot stage before full implementation of the final interpretive standards. As OFCCP has explained, the agency does not require or expect the contractor to gather data, build databases, or perform multiple regression analyses. OFCCP will do all of those activities. In fact, OFCCP has been using aspects of the analyses discussed in these final interpretive standards in a substantial number of compliance reviews over the last several years. Because OFCCP is not requiring contractors to engage in any activity to implement these final interpretive standards, OFCCP disagrees that a grace or pilot period are appropriate.

E. Factors Included in the Regression Analysis

Several commenters, including the U.S. Chamber of Commerce, were concerned that the listing of factors in the proposed guidelines could result in agency investigators presuming that the listed factors must be used in all cases. These commenters asked OFCCP to clarify that the factors to be used in the regression analysis must be determined by the facts of the particular case. By contrast, several commenters, such as HR Analytical Services, requested that OFCCP provide more guidance on the factors that the agency would use in the regression analysis. OFCCP agrees that the factors must be determined based on the facts of the particular case. OFCCP listed several of the typical factors to provide some general idea of the types of factors that may be used, not to identify an exhaustive list that is presumed to apply in every case. Because the factors must be based on the facts of the particular case, OFCCP is unable to provide additional guidance on which factors may be used in a case. OFCCP agrees that there are many other factors that may be important in a particular case, such as significant leaves of absence, employment with a predecessor company, whether the educational degree is related to the employee's position, etc.

Many commenters noted that contractors frequently do not collect data in their HRIS systems on all of the factors that may influence compensation decisions, and that some of the factors used in making compensation decisions cannot be quantified.^[25] As noted above, OFCCP does not expect a contractor to maintain all of the data necessary to conduct a multiple regression analysis in its HRIS system. Nor does OFCCP require that contractors collect such data and build a database to turn over to OFCCP during a compliance review. Instead, OFCCP will gather the pertinent information through interviews and though review of personnel files and other pertinent documents. Once OFCCP gathers the necessary information, OFCCP staff will build a database. OFCCP does not presume that every factor that may influence compensation is necessarily quantifiable. OFCCP may attempt to account for such factors in the regression model through categorical variables or proxies, if possible. OFCCP also may assess whether unquantifiable or inherently qualitative factors explain multiple regression results through non-statistical methods.

ETF argued that OFCCP should include only factors that the employer actually relied on in making pay decisions. OFCCP agrees that the factors that are included in the multiple regression analysis must be factors that actually had an influence on the employer's compensation practices. However, OFCCP does not agree that the factor must have been overtly considered by a particular decisionmaker when making a particular compensation decision. A legitimate factor may influence compensation without having been a factor that the employer's decisionmakers overtly relied on in making a particular compensation decision. For example, a department manager responsible for setting merit pay increases in a particular year may only have limited discretion to determine merit increases because of constraints established by budget decisions made by other decisionmakers and by the employer's compensation guidelines. Thus, the merit increase decisions actually involved a host of other decisions by other decisionmakers at an earlier point in time. As noted above, some commenters criticized the proposed standards because the referenced regression model evaluates current compensation, not each and every individual pay decision that contributed to current compensation (or compensation at a particular point in time). OFCCP rejected those commenters' suggestion of using an analysis that focuses more directly on compensation decisions. Because the regression approach OFCCP adopts in the final standards uses compensation at a particular point in time, the factors that influence compensation may not necessarily be factors that the employer's decisionmakers relied on overtly in making particular pay decisions. However, OFCCP can obtain an indication through the multiple regression analyses whether a particular factor had an influence on specific employees' current compensation (or compensation at the particular point in time).

F. Anecdotal Evidence

Several commenters, including ETF, ACC, NILG, EEAC, and ORC, commented on OFCCP's interpretive standard relating to anecdotal evidence. ETF commented that OFCCP's proposed standard places additional burdens on OFCCP not required by Title VII or Executive Order 11246 because the proposed standards suggest that anecdotal evidence is required to establish a violation of systemic compensation discrimination. OFCCP disagrees with ETF's characterization of the interpretive standard relating to anecdotal evidence. The interpretive standard on anecdotal evidence is not intended to place burdens on OFCCP in establishing a violation beyond what is required by interpretations of Title VII. Rather, the interpretive standard sets forth OFCCP's interpretation that anecdotal evidence is important in establishing systemic compensation discrimination and its position that rarely will a Notice of Violation be issued by OFCCP alleging systemic compensation discrimination absent anecdotal evidence.

OFCCP's strong preference for anecdotal evidence and the important role that such evidence plays in determining whether systemic compensation discrimination exists is supported by case law. For example, in EEOC v. Morgan Stanley & Co., Inc., No. 01 Civ. 8421, 2002 WL 1431685, at *1 (S.D.N.Y. July 1, 2002)[footnote omitted], the court discussed the importance of anecdotal evidence to the EEOC's case:

The Court agrees that the EEOC is entitled “to develop its case, including the circumstances surrounding discrimination against individual women,” see Plaintiff's Opp. at 3, with the safeguards put in place by Judge Ellis. While the EEOC's case “depends on a statistical analysis of promotion and compensation data of an entire class of women, the [EEOC] is also entitled to put on proof of anecdotal evidence of discrimination.” Plaintiff's Opp. at 3; see Rossini, 798 F.2d at 604 (recognizing the importance of anecdotal evidence in employment discrimination cases) (citing Intl'l Brotherhood of Teamsters v. United States, 431 U.S. 324, 339, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977)); see also Coser v. Moore, 739 F.2d 746, 751-752 (2d Cir.1984) (“where a pattern and practice of discrimination is alleged, [statistical evidence alone] must be weighed in light of the failure to locate and identify a meaningful number of concrete examples of discrimination * * *.”).

Similarly, in Obrey v. Johnson, 400 F.3d 691, 698 (9th Cir. 2005), the court noted the important role of anecdotal evidence:

It is commonplace that a plaintiff attempting to establish a pattern or practice of discriminatory employment will present some anecdotal testimony regarding past discriminatory acts. See, e.g., Rossini v. Ogilvy & Mather, Inc., 798 F.2d 590, 604 (2d Cir. 1986) (“In evaluating all of the evidence in a discrimination case, a district court may properly consider the quality of any anecdotal evidence or the absence of such evidence.”); Coates v. Johnson & Johnson, 756 F.2d 524, 532 (7th Cir. 1985) (“The plaintiffs” prima facie case will thus usually consist of statistical evidence demonstrating substantial disparities in the application of employment actions as to minorities and the unprotected group, buttressed by evidence of * * * specific instances of discrimination.”); Valentino v. United States Postal Serv., 674 F.2d 56, 69 (D.C. Cir. 1982) (“[W]hen the statistical evidence does not adequately account for the diverse and specialized qualifications necessary for (the positions in question), strong evidence of individual instances of discrimination becomes vital to the plaintiff's case.”) (internal quotation marks omitted); Garcia v. Rush-Presbyterian-St. Lukes Med. Ctr., 660 F.2d 1217, 1225 (7th Cir. 1981) (“We find very damaging to plaintiff's position the fact that not only was their statistical evidence insufficient, but that they failed completely to come forward with any direct or anecdotal evidence of discriminatory employment practices by defendants. Plaintiffs did not present in evidence even one specific instance of discrimination.”).

OFCCP cited additional cases that support the important role of anecdotal evidence in the preamble of the proposed interpretive standards. See, e.g., Bazemore, 478 U.S. at 473 (noting that statistics were supported by “evidence consisting of individual comparisons between salaries of blacks and whites similarly situated”); Morgan v. United Parcel Service of America, Inc., 380 F.3d 459, 471 (8th Cir. 2004) (“One of the most important flaws in Plaintiffs” case is that they adduced no individual testimony regarding intentional discrimination. As mentioned above, Plaintiffs' purported anecdotal evidence was insufficient for the working-conditions claim, and we see none with regard to pay. Although such evidence is not required, the failure to adduce it ‘reinforces the doubt arising from the questions about validity of the statistical evidence.’ EEOC v. Sears, Roebuck & Co., 839 F.2d 302, 311 (7th Cir.1988) (quoting Griffin v. Board of Regents, 795 F.2d 1281, 1292 (7th Cir.1986))”); Dukes v. Wal-Mart Stores, Inc., 22 F.R.D. 137, 165-66 (N.D. Cal. 2004) (“[P]laintiffs have submitted * * * 114 declarations from class members around the country * * *. [who will] testify to being paid less than similarly situated men, * * *, and being subjected to various individual sexist acts.”); Bakewell v. Stephen F. Austin Univ., 975 F. Supp. 858, 905-06 (E.D. Tex. 1996) (“The paucity of anecdotal evidence of discrimination severely diminishes plaintiffs' contention that a pattern or practice of salary discrimination against female faculty members prevails at SFA.”).^[26] OFCCP's position is also consistent with EEOC's guidance on compensation discrimination. See CMCD, at 10-13 n.30 (“A cause finding of systemic discrimination should rarely be based on statistics alone.”). OFCCP's Federal Contract Compliance Manual for many years has included a section on anecdotal evidence and a description of its use in systemic discrimination cases. See OFCCP's Federal Contract Compliance Manual, at Section 7D05(e) (“While courts have held that statistics alone may be sufficient to prove discrimination where disparities are gross; i.e., at least two standard deviations, supporting evidence strengthens statistical cases and should always be sought. One type of supporting evidence is anecdotal evidence. Anecdotal evidence consists of statements from minorities or women who can show that they met all of the contractor's requirements but still did not receive the benefit at issue, and any first hand accounts of discriminatory acts on the part of the contractor that support the statistical inference. Thus, anecdotal evidence is not limited to independent examples of comparative disparate treatment.”).

OFCCP agrees with ETF that anecdotal evidence need not be, and in most cases likely will not be, in the form of “‘smoking gun’ evidence of discrimination,” or what is known in the caselaw as “direct evidence” of discrimination. See, e.g., Desert Palace Co. v. Costa, 539 U.S. 90, 97 (2003) (noting that Ninth Circuit defined direct evidence as “‘substantial evidence of conduct or statements by the employer directly reflecting discriminatory animus,”’ quoting Costa v. Desert Palace, Inc., 268 F.3d 882, 884 (9th Cir. 2001)). OFCCP's reference to “anecdotal evidence” in these final interpretive standards is to evidence that leads to an inference that the employer subjected a particular employee or particular employees to disparate treatment in compensation. See, e.g., Bazemore, 478 U.S. at 473; Morgan, 380 F.3d at 471; Dukes, 22 F.R.D. at 165-66; CMCD, at 10-13 n.30 (“Where possible, evidence of individual instances of discrimination should be used to bring the ‘cold numbers convincingly to life,’ Teamsters, 431 U.S. at 339, 340 * * *”); Obrey v. Johnson, 400 F.3d 691, 698 (9th Cir. 2005); EEOC v. Morgan Stanley & Co., Inc., No. 01 Civ. 8421, 2002 WL 1431685, at *1 (S.D.N.Y. July 1, 2002). OFCCP agrees with ETF that witness testimony from management officials and employees concerning the employer's pay practices would help establish the appropriate factors for the regression analysis and OFCCP will seek such evidence in evaluating whether there is systemic pay compensation discrimination. See, e.g., Eastland v. Tennessee Valley Auth., 704 F.2d 613, 623 (11th Cir. 1983) (“By evaluating the basis upon which the party selected the variables included in its regression the court may assess the model's validity. ‘Three kinds of evidence may be offered in support of a regression model; direct testimony as to what factors operated in the decision-making process under challenge, what kinds of factors generally operate in decision-making processes of the kind under challenge, and expert testimony concerning what factors can be expected to influence the process under challenge according to principles of economic theory.’ D. Baldus & J. Cole, Statistical Proof of Discrimination Sec. 8.22 at 70 (1980 & 1982 Supp.) (hereinafter Baldus & Cole). The strength of the factual foundation supporting a regression model may be a factor in assessing whether the group status coefficient indicates discrimination or the influence of legitimate qualifications which happen to correlate with group status. Baldus & Cole, supra, Sec. 8.021 at 66 (1982 Supp.).”). However, in addition to this type of evidence, OFCCP will seek the anecdotal evidence described above.

Several commenters, including ACC, NILG, and NRCILG, were concerned that OFCCP's investigation for anecdotal evidence of discrimination would unduly disrupt the employer's operations when agency staff interviewed employees. These commenters argued that OFCCP should afford the contractor an opportunity to rebut OFCCP's regression analysis or settle the case before the agency conducts such employee interviews. OFCCP is sensitive to the commenters concerns that employee interviews may disrupt the employer's operations and OFCCP will accommodate the employer's legitimate business needs in scheduling the interviews. At the same time, however, OFCCP disagrees with the commenters that the agency should allege a violation or offer the contractor an opportunity to rebut a regression analysis or settle with OFCCP prior to the completion of the agency's investigation under the final interpretive standards. In this regard, the proposed standards reflect OFCCP's strong preference for developing anecdotal evidence in establishing systemic compensation discrimination.

Several commenters, such as EEAC and ORC, argued that OFCCP should never allege systemic compensation discrimination without anecdotal evidence of discrimination, nor should the agency ever allege systemic compensation discrimination based only on anecdotal evidence. OFCCP disagrees with these commenters. There may be cases in which the statistical analysis is so compelling that an allegation of systemic discrimination is warranted even in the absence of anecdotal evidence of compensation discrimination.^[27] Similarly, the amount, weight, and reliability of anecdotal evidence found in a case may support an inference of systemic discrimination, even in the absence of statistical evidence.^[28] Of course, the anecdotal evidence of systemic compensation discrimination in such a case would have to support an inference that the employer compensated similarly situated employees differently based on gender or race and that the employer's compensation “discrimination was the company's standard operating procedure—the regular rather than the unusual practice.” Bazemore, 478 U.S. at 398 (quoting Teamsters, 431 U.S. at 336).

G. Confidentiality of Compensation and Personnel Information

Many commenters expressed concern about the confidentiality of compensation and personnel information contractors will be required to submit or make available to OFCCP under the proposed interpretive standards. These commenters requested that OFCCP provide express assurances that the agency would not disclose such information to third-parties or other enforcement agencies. In response to these comments, OFCCP has added a provision (Paragraph 8 of the “Standards for OFCCP Evaluation of Contractors’ Compensation Practices”) to the final interpretive standards under which “OFCCP will treat compensation and other personnel information provided by the contractor to OFCCP during a systemic compensation investigation as confidential to the maximum extent the information is exempt from public disclosure under the Freedom of Information Act, 5 U.S.C. 552 * * *.” OFCCP borrowed this text from its regulations at 41 CFR 60-2.18(d).

III. Substantive Discussion Regarding the Final Standards

A. OFCCP Compliance Reviews Focus on Systemic Compensation Discrimination

The Department of Labor's Office of Federal Contract Compliance Programs (OFCCP) enforces Executive Order 11246, which prohibits covered federal contractors and subcontractors from making employment decisions on the basis of race, color, national origin, religion, or sex.^[29]

OFCCP conducts compliance reviews to determine whether covered contractors have been engaging in workplace discrimination prohibited by Executive Order 11246. As part of its compliance review process, OFCCP investigates whether contractors' pay practices are discriminatory.

OFCCP compliance reviews typically produce cases that involve allegations of systemic discrimination, not discrimination against a particular individual employee. OFCCP systemic compensation discrimination cases typically are proven under a disparate treatment, pattern or practice theory of discrimination.^[30] The burdens of persuasion necessary to succeed on a discrimination claim differ depending on whether the case involves allegations of a pattern or practice of discrimination or allegations that a particular individual was subjected to discrimination. In a case involving alleged discrimination against a particular individual, the plaintiff must establish by a preponderance of the evidence that the employer made the challenged employment decision because of the individual's race, color, religion, sex, or national origin. United States Postal Service Bd. of Governors v. Aikens, 460 U.S. 711, 715 (1983). In a pattern or practice case, “plaintiffs must ‘establish by a preponderance of the evidence that racial discrimination was the company's standard operating procedure—the regular rather than the unusual practice.’ Teamsters v. United States, 431 U.S. 324, 336 (1977).” Bazemore v. Friday, 478 U.S. 385, 398 (1986).

In addition to differences in the burdens of persuasion as between cases involving alleged discrimination against a particular individual and an alleged pattern or practice of discrimination, the burdens of production necessary to survive a motion for summary disposition are different between the two types of cases. In both types of cases, a plaintiff bears the initial burden of presenting a prima facie case of discrimination. There is no precise set of requirements for a plaintiff's prima facie case. “The facts necessarily will vary in Title VII cases, and the specification * * * of the prima facie proof required from [a plaintiff] is not necessarily applicable in every respect to differing factual circumstances.” Int'l Bhd. of Teamsters v. United States, 431 U.S. 324, 358 (1977) (quoting McDonnell Douglas, 411 U.S. at 802 n. 13). “The importance of McDonnell Douglas lies, not in its specification of the discrete elements of the proof there required, but in its recognition of the general principle that any Title VII plaintiff must carry the initial burden of offering evidence adequate to create an inference that an employment decision was based on a discriminatory criterion illegal under [Title VII].” Teamsters, 431 U.S. at 358.

In an individual case, the plaintiff typically must rely on evidence pertaining to his or her own circumstances to establish a prima facie case of discrimination. The prima facie case creates a presumption of discrimination that the employer may rebut by articulating a legitimate nondiscriminatory reason for the alleged discriminatory employment decision. McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802 (1973). The employer must produce admissible evidence of a legitimate, nondiscriminatory reason for the challenged employment decision. Texas Dep't of Community Affairs v. Burdine, 450 U.S. 248, 254 (1981). “Th[e] [employer's] burden is one of production, not persuasion; ‘it can involve no credibility assessment.’ ” Reeves v. Sanderson Plumbing Products, Inc., 530 U.S. 133, 142 (2000) (quoting St. Mary's Honor Center v. Hicks, 509 U.S. 502, 509 (1993)). Once the employer articulates a legitimate nondiscriminatory reason for the challenged employment decision, the plaintiff is afforded the opportunity to prove that the employer's articulated reason is a pretext for discrimination. McDonnell Douglas, 411 U.S. at 804; Reeves, 530 U.S. at 142. “Proof that the [employer's] explanation is unworthy of credence is simply one form of circumstantial evidence that is probative of intentional discrimination * * *.” Reeves, 530 U.S. at 147. “Other evidence that may be relevant to any showing of pretext includes * * * [the employer's] general policy and practice with respect to minority employment * * *. On the latter point, statistics as to [the employer's] employment policy and practice may be helpful to a determination of whether [the employer's actions] * * * conformed to a general pattern of discrimination * * *” McDonnell Douglas, 411 U.S. at 804-05.

In a pattern or practice case, the plaintiffs' “initial burden is to demonstrate that unlawful discrimination has been a regular procedure or policy followed by an employer * * *.” Teamsters, 431 U.S. at 360. “The burden then shifts to the employer to defeat the prima facie showing of a pattern or practice by demonstrating that the [plaintiffs'] proof is either inaccurate or insignificant.” Id. “The employer's defense must, of course, be designed to meet the prima facie case of the [plaintiffs] * * *.” which typically focuses on “a pattern of discriminatory decisionmaking.” Id., at 360 n. 46. However, there are no “particular limits on the type of evidence an employer may use.” Id.

Despite these differences in the burdens of persuasion and production, however, once the plaintiff has offered evidence that is sufficient to establish a prima facie case, and the employer has produced evidence that is sufficient to rebut the prima facie case, then the factfinder must decide whether plaintiffs have demonstrated discrimination by a preponderance of the evidence. “[O]ur decision in United States Postal Service Board of Governors v. Aikens, 460 U.S. 711 (1983), although not decided in the context of a pattern-and-practice case, makes clear that if the defendants have not succeeded in having a case dismissed on the ground that plaintiffs have failed to establish a prima facie case, and have responded to the plaintiffs' proof by offering evidence of their own, the factfinder then must decide whether the plaintiffs have demonstrated a pattern or practice of discrimination by a preponderance of the evidence. This is because the only issue to be decided at that point is whether the plaintiffs have actually proved discrimination. Id., at 715.” Bazemore, 478 U.S. at 398.

B. OFCCP Has Not Issued Significant Interpretive Guidance on Systemic Compensation Discrimination Under Executive Order 11246

In 1970, the Department of Labor published “Sex Discrimination Guidelines,” codified at 41 CFR part 60-20, which included a section (60-20.5) on “[d]iscriminatory wages.” 35 FR 8888 (June 9, 1970). The Sex Discrimination Guidelines (SDG) do not provide specific standards for determining systemic compensation discrimination for OFCCP or a contractor.^[31] Rather, the SDG provide that “[t]he employer's wages (sic) schedules must not be related to or based on the sex of the employees,” and contains a short “note” that references the “more obvious cases of discrimination * * * where employees of different sexes are paid different wages on jobs which require substantially equal skill, effort and responsibility and are performed under similar working conditions.” 41 CFR 60-20.5(a) (2004). OFCCP has not promulgated any definitive interpretation of the SDG, nor has a definitive interpretation arisen through longstanding agency practice.^[32]

Instead, OFCCP has provided only a general policy statement about compensation discrimination in the preamble to a May 4, 2000 Notice of Proposed Rulemaking (NPRM). In the May 4, 2000 NPRM, OFCCP formally expressed the Department of Labor's policy regarding compensation analysis:

More recently, an additional objective of the proposed revision has been to advance the Department of Labor's goal of pay equity; that is, ensuring that employees are compensated equally for performing equal work.

65 FR 26089 (May 4, 2000).

This stated policy was reflected in several significant settlements in systemic compensation discrimination cases in which OFCCP relied on sophisticated multiple regression analyses to remedy an alleged violation of Executive Order 11246. OFCCP has not, however, published formal guidance providing any interpretation of Executive Order 11246 with respect to systemic compensation discrimination.

C. OFCCP's Informal Approaches to Systemic Compensation Discrimination in the Late 1990s Involved the Controversial “Pay Grade Theory”

In the late-1990s several OFCCP regions began to use a controversial “grade theory” approach to compensation discrimination analysis.^[33]

The basic unit of analysis under the grade theory is the pay grade or pay range. Under this theory, it is assumed that employees are similarly situated with respect to evaluating compensation decisions regarding such employees if the contractor has placed their jobs in the same pay grade:

By the very act of creating a grade level system, where each employee has approximately the same potential to move from the minimum to the maximum of his/her grade range dependent upon performance, the employer has recognized that certain jobs are essentially similar in terms of skill, effort and responsibility.

“Systemic Compensation Analysis: An Investigatory Approach” (hereinafter “SCA”), at 5. A later paper, “Update on Systemic Compensation Analysis” (hereinafter, “Update”), also described this pay grade assumption:

Where we determine that each employee in a salary grade system has the same opportunity, subject to performance, to move to the maximum rate of the salary grade range without a change in job title, we believe the employer * * * has already identified certain jobs as having similar value to the organization.

Update, at 6.^[34]

After identifying employees in the same pay grade, one version of the grade theory method called for a comparison of the median compensation of males versus females, and minorities versus non-minorities in each pay grade. SCA, at 6; Update, at 7. If there was a “significant” difference (although “significant” was not defined) in median compensation between males/females or minorities/non-minorities within a given pay grade, then the next step was to assess whether this disparity is explained by median or average differences in other factors, such as time in grade, prior experience, education, and performance. SCA, at 7; Update, at 11. However, this method did not use tests of statistical significance in determining whether a pattern of compensation discrimination exists. If a “pattern” of pay disparities (although “pattern” was not defined) emerged not explicable by analysis of median or average differences in time in grade, prior experience, or other factors, OFCCP alleged that the contractor violated the nondiscrimination requirements of Executive Order 11246. Update, at 15.

In another version of the grade theory method used by some OFCCP regions in the late 1990s,^[35] the pay grade was included as a factor in a regression model that typically covered all exempt employees in the workplace within a single, “pooled” regression. The regression typically included factors such as time in grade, experience, and education. This method did rely on tests of statistical significance, although rarely did OFCCP develop anecdotal evidence to support the statistical analysis under this method.

D. The Pay Grade Theory Is Inconsistent With Title VII Standards

OFCCP has discontinued using these pay grade methods because the agency has determined that the methods' principal assumptions related to pay grade or pay range do not comport with Title VII standards as to whether employees are similarly situated. OFCCP recognizes that, with respect to compensation discrimination, similarity in job content, skills and qualifications involved in the job, and responsibility level are crucial determinants of whether employees are similarly situated under Title VII. See, e.g., CMCD, at 10-5 to 10-8; Block v. Kwal-Howells, Inc., No. 03-1101, 2004 WL 296976, at *2-*4 (10th Cir. Feb. 17, 2004); Williams v. Galveston Ind. Sch. Dist., No. 03-40436, 78 Fed. Appx. 946, 949-50, 2003 WL 22426852 (5th Cir. Oct. 23, 2003); Verwey v. Illinois Coll. of Optometry, 43 Fed. Appx. 996, 2002 WL 1836507, at *4 (7th Cir. Aug. 9, 2002); Lang v. Kohl's Food Stores, Inc., 217 F.3d 919, 922-23 (7th Cir. 2002); Rodriguez v. SmithKline Beecham, 224 F.3d 1, 8 (1st Cir. 2000); Coward v. ADT Sec. Sys., Inc., 140 F.3d 271, 274 (D.C. Cir. 1998); Aman v. Cort Furniture Rental Corp., 85 F.3d 1078, 1087 (3d Cir. 1996); Sprague v. Thorn Americas, Inc., 129 F.3d 1355, 1362 (10th Cir. 1997); Tomka v. Seiler Corp., 66 F.3d 1295, 1310-11 (2d Cir. 1995), abrogated on other grounds by Burlington Indus., Inc. v. Ellerth, 524 U.S. 742 (1998); Mulhall v. Advance Sec., Inc., 19 F.3d 586, 598 (11th Cir. 1994); Brinkley-Obu v. Hughes Training, Inc., 36 F.3d 336, 343 (4th Cir. 1994); Miranda v. B&B Cash Grocery Store, Inc., 975 F.2d 1518, 1526-31 (11th Cir. 1992); EEOC v. Sears, Roebuck & Co., 839 F.2d 302, 343-53 (7th Cir. 1988); Marcoux v. State of Maine, 797 F.2d 1100, 1107 (1st Cir. 1986); Eastland v. Tennessee Valley Auth., 704 F.2d 613, 624-25 (11th Cir. 1983); Woodward v. United Parcel Serv., Inc., 306 F. Supp.2d 567, 574-75 (D. S.C. 2004); Lawton v. Sunoco, Inc., No. 01-2784, 2002 WL 1585582, at *7 (E.D. Pa. Jul 17, 2002); Stroup v. J.L. Clark, No. 99C50029, 2001 WL 114404, at *6 (N.D. Ill. Feb. 2, 2001); Donaldson v. Microsoft Corp., 205 F.R.D. 558, 563 (W.D. Wash. 2001); Dobbs-Weinstein v. Vanderbilt Univ., 1 F. Supp.2d 783, 803-04 (M.D. Tenn. 1998); Beard v. Whitley Co. REMC, 656 F. Supp. 1461, 1471-72 (N.D. Ind. 1987); Dalley v. Michigan Blue Cross/Blue Shield, Inc., 612 F. Supp. 1444, 1451-52 (E.D. Mich. 1985); EEOC v. Kendall of Dallas, Inc., No. TY-80-441-CA, 1984 WL 978, at *9-*12 (E.D. Tex. Mar. 8, 1984); Presseisen v. Swarthmore Coll., 442 F. Supp. 593, 615-19 (E.D. Pa. 1977), aff'd 582 F.2d 1275 (3d Cir. 1978) (Table).

Contrary to these standards, the grade theory assumed that employers' pre-existing job-groupings, such as pay grades or pay ranges, are absolute indicia of similarity in employees' job content, skills and qualifications involved in the job, and responsibility level. While all of the courts in the above string cite have implicitly rejected the grade theory by emphasizing the importance of facts about the work employees actually perform, several of these courts have expressly rejected the proposition that a pay grade offers absolute indicia of similarity in job content, qualifications and skills involved in the job, and responsibility level. See Williams, 78 Fed. Appx. at 949 n. 9; Cort Furniture, 85 F.3d at 1087; Woodward, 306 F. Supp.2d at 574-75. The facts about employees' actual work activities, the skills and qualifications involved in the job, and responsibility levels in a particular case may, of course, happen to coincide with the employer's pay grade or pay range, but the crucial determinant of whether the employees are similarly situated is their actual work activities, not the fact that the employees have been placed in the same pay grade or range.^[36]

Based on these considerations, the Department interprets Executive Order 11246 and the SDG as not permitting the pay grade theory approach to systemic compensation discrimination. Instead, the Department interprets Executive Order 11246 and the SDG as prohibiting systemic compensation discrimination involving dissimilar treatment of individuals who are similarly situated, based on similarity in work performed, skills and qualifications involved in the job, and responsibility levels.

E. The Department Has Decided To Promulgate Interpretive Standards on Systemic Compensation Discrimination To Guide Agency Officials and Covered Contractors and Subcontractors

The Department of Labor has decided to formally promulgate detailed standards interpreting Executive Order 11246 and the SDG with respect to systemic compensation discrimination. The final interpretive standards will provide guidance and methods for OFCCP evaluations of contractors' compensation practices during compliance reviews. This will ensure that agency personnel and covered Federal contractors and subcontractors understand the substantive standards for systemic compensation discrimination under Executive Order 11246. The Department believes that contractors and subcontractors are more likely to comply with Executive Order 11246 if they understand the substantive standards which determine whether there is systemic compensation discrimination prohibited by Executive Order 11246. Further, agency officials will have a stronger basis for pursuing investigations of possible systemic compensation discrimination because of the transparency and uniformity provided by these standards.

These final standards are intended to govern OFCCP's analysis of contractors' compensation practices, and in particular, OFCCP's determination of whether a contractor has engaged in systemic compensation discrimination. In addition, these final standards are intended to constitute a definitive interpretation of the SDG and Executive Order 11246 with respect to systemic compensation discrimination.

F. Discussion of the Final Interpretive Standards

OFCCP adopts final standards interpreting Executive Order 11246 and the SDG with respect to systemic compensation discrimination. The systemic compensation discrimination analysis as set forth in these final standards has two major characteristics: (1) The determination of employees who are “similarly situated” for purposes of comparing contractor pay decisions will focus on the similarity of the work performed, the levels of responsibility, and the skills and qualifications involved in the positions; and (2) the analysis relies on a statistical technique known as multiple regression.

Under OFCCP's final standards, employees are similarly situated with respect to pay decisions where the employees perform similar work, have similar responsibility levels, and occupy positions involving similar qualifications and skills. See discussion and cases cited under Section IIID, supra.^[37]

The determination of whether employees are similarly situated must be based on the actual facts about the work performed, the responsibility level of the employees, and whether the positions involve similar skills and qualifications. The employer's preexisting groupings developed and maintained for other purposes, such as job families or affirmative action program job groups, may provide some indication of similarity in work, responsibility level, and skills and qualifications. However, these preexisting groupings are not dispositive, and OFCCP will not assume that these groupings contain similarly situated employees. For example, it cannot be assumed that employees are similarly situated merely because they share the same pay grade or range, or because their pay can progress to the top of a pay grade or range without changing jobs.^[38] Thus, OFCCP will investigate whether such preexisting groupings do in fact contain employees who perform similar work, and whose positions involve similar skills, qualifications, and responsibility levels, by looking at job descriptions and conducting employee interviews. Based on sufficient empirical data (e.g., job descriptions and employee interviews), OFCCP will determine which employees are in fact similarly situated. There may be other factors that have a bearing on whether employees are similarly situated, in addition to work performed, responsibility level, and skills/qualifications involved in the positions. For example, additional factors may include department or other functional unit of the employer, employment status (e.g., full-time versus part-time), compensation status (e.g., union versus non-union, hourly versus salaried versus commissions), etc. OFCCP will consider the applicability of these additional factors in each case and make a determination based on the facts of the particular case.

In addition to similarity in work performed, skills and qualifications, and responsibility levels, systemic compensation discrimination under Executive Order 11246 requires that the comparison take into account legitimate factors that affect compensation. In order to account for the influence of such legitimate factors on compensation, a statistical analysis known as “multiple regression” must be used. Multiple regression is explained as follows:

Multiple regression analysis is a statistical tool for understanding the relationship between two or more variables. Multiple regression involves a variable to be explained—called the dependent variable—and additional explanatory variables that are thought to produce or be associated with changes in the dependent variable. For example, a multiple regression analysis might estimate the effect of the number of years of work on salary. Salary would be the dependent variable to be explained; years of experience would be the explanatory variable. Multiple regression analysis is sometimes well suited to the analysis of data about competing theories in which there are several possible explanations for the relationship among a number of explanatory variables. Multiple regression typically uses a single dependent variable and several explanatory variables to assess the statistical data pertinent to these theories. In a case alleging sex discrimination in salaries, for example, a multiple regression analysis would examine not only sex, but also other explanatory variables of interest, such as education and experience. The employer-defendant might use multiple regression to argue that salary is a function of the employee's education and experience, and the employee-plaintiff might argue that salary is also a function of the individual's sex.

Daniel L. Rubenfeld, Reference Guide on Multiple Regression, in Federal Judicial Center, Reference Manual on Scientific Evidence, at 181 (2d ed. 2000).

The multiple regression model must include those factors that are important to how the contractor in practice makes pay decisions. “Such factors could include the employees' education, work experience with previous employers, seniority in the job, time in a particular salary grade, performance ratings, and others.” CMCD, at 10-18. OFCCP generally will attempt to build the regression model in such a way that controls for the factors that the investigation reveals are important to the employer's pay decisions, but also allows the agency to assess how the employers' pay decisions affect most employees. One factor that must be controlled for in the regression model is categories or groupings of jobs that are similarly situated based on the analysis of job similarity noted above (i.e., similarity in the content of the work employees perform, and similarity in the skills, qualifications, and responsibility levels of the positions the employees occupy, and additional factors as discussed above). This will ensure that the analysis compares the treatment of employees who are in fact similarly situated.

In addition, OFCCP will investigate the facts of each particular case to ensure that factors included in the regression are legitimate and are not themselves influenced by unlawful discrimination, which is often discussed in case law as a factor “tainted” by discrimination. However, OFCCP will not automatically presume that a factor is tainted without initially investigating the facts of the particular case. OFCCP will determine whether a factor is tainted by evaluating proof of discrimination with respect to that factor, but not based on the fact that the factor has an influence on the outcome of a regression model that includes the factor. See, e.g., Morgan v. United Parcel Service of America, Inc., 380 F.3d 459, 470 (8th Cir. 2004) (“Plaintiffs” only evidence of discrimination in past pay is the apparent correlation between race and center-manager base pay during the class period. But that correlation is what Plaintiffs have evidence of only by omitting past pay. They have no evidence, statistical or otherwise, that past pay disparities were racially discriminatory. This sort of bootstrapping cannot create an inference of discrimination with regard to either class-period base pay or past pay.”); Smith v. Xerox Corp., 196 F.3d 358, 371 n. 11 (2d Cir. 1999) (“Absent evidence tending to show that the CAF scores were tainted they should have been included in a multiple regression analysis in an effort to eliminate a relatively poor performance compared to coworkers as a cause of each plaintiff's termination. Certainly, performance is a factor Xerox was permitted to consider in deciding whom to retain.”); Ottaviani v. State Univ. of New York, 875 F.2d 365, 375 (2d Cir. 1988) (“The question to be resolved, then, in cases involving the use of academic rank factors, is whether rank is tainted by discrimination at the particular institution charged with violating Title VII. Although appellants reiterate on appeal their claim that rank at New Paltz was tainted, it is clear that the district judge accepted and considered evidence from the parties on both sides of this issue, and that she rejected the plaintiffs' contentions on this point. At trial, the plaintiffs failed to adduce any significant statistical evidence of discrimination as to rank. As the district court stated in its opinion, the plaintiffs' studies of rank, rank at hire, and waiting time for promotion ‘were mere compilations of data’ which neither accounted for important factors relevant to assignment of rank and promotion, ‘nor demonstrated that observed differences were statistically significant.’ Ottaviani, 679 F.Supp. at 306. The defendants, on the other hand, offered persuasive objective evidence to demonstrate that there was no discrimination in either placement into initial rank or promotion at New Paltz between 1973 and 1984, and the district court chose to credit the defendants' evidence. Upon review of the record, we cannot state that the court's rulings in this regard were clearly erroneous.”); CMCD, at 10-18 (discussing use of performance rating in multiple regression analysis for assessing systemic compensation discrimination).

The factors that influence pay decisions may not bear the same relationship to compensation for all categories of jobs in the employer's workforce. For example, performance may have a more significant influence on compensation for a high-level executive, than for technicians or service workers. This issue must be addressed through either of two methods. One method is to perform separate regressions for each category of jobs in which the relationship between the factors and compensation is similar (while including category factors in each regression that control for groupings of employees who are similarly situated based on work performed, responsibility level, and skills and qualifications). If separate regressions by categories of jobs would not permit OFCCP to assess the way the contractor's compensation practices impact on a significant number of employees, OFCCP may perform a “pooled” regression, which combines these categories of jobs into a single regression (while including an OFCCP-developed category factor in the “pooled” regression that controls for groupings of employees who are similarly situated based on work performed, responsibility level, and skills and qualifications). However, if a pooled regression is used, the regression must include appropriate “interaction terms” ^[39] in the pooled regression to account for differences in the effects of certain factors by job category. OFCCP will run statistical tests generally accepted in the statistics profession (e.g., the “Chow test”), to determine which interaction terms should be included in the pooled regression analysis.

Systemic compensation discrimination under Executive Order 11246 must be based on disparities that are “statistically significant,” i.e., those that could not be expected to have occurred by chance. “While not intending to suggest that ‘precise calculations of statistical significance are necessary in employing statistical proof,’ the Supreme Court has stated that ‘a fluctuation of more than two or three standard deviations would undercut the hypothesis that decisions were being made randomly with respect to [a protected trait].’ Hazelwood Sch. Dist. v. United States, 433 U.S. 299, 311 n.17 (1977).” CMCD, at 10-14 n.32. To ensure uniformity and predictability, OFCCP will conclude that a compensation disparity is statistically significant under these final standards if it is significant at a level of two or more standard deviations, based on measures of statistical significance that are generally accepted in the statistics profession.

OFCCP will seldom make a finding of systemic discrimination based on statistical analysis alone, but will obtain anecdotal evidence to support the statistical evidence. See, e.g., Teamsters, 431 U.S. at 338-39 (“The Government bolstered its statistical evidence with the testimony of individuals who recounted over 40 specific instances of discrimination * * *. The individuals who testified about their personal experiences with the company brought the cold numbers convincingly to life.”); Bazemore, 478 U.S. at 473 (noting that statistics were supported by “evidence consisting of individual comparisons between salaries of blacks and whites similarly situated”); Morgan, 380 F.3d at 471 (“One of the most important flaws in Plaintiffs' case is that they adduced no individual testimony regarding intentional discrimination. As mentioned above, Plaintiffs' purported anecdotal evidence was insufficient for the working-conditions claim, and we see none with regard to pay. Although such evidence is not required, the failure to adduce it ‘reinforces the doubt arising from the questions about validity of the statistical evidence.’ EEOC v. Sears, Roebuck & Co., 839 F.2d 302, 311 (7th Cir.1988) (quoting Griffin v. Board of Regents, 795 F.2d 1281, 1292 (7th Cir.1986))”); Dukes v. Wal-Mart Stores, Inc., 22 F.R.D. 137, 165-66 (N.D. Cal. 2004) (“[P]laintiffs have submitted * * * 114 declarations from class members around the country * * *. [who will] testify to being paid less than similarly situated men, * * *, and being subjected to various individual sexist acts.”); Bakewell v. Stephen F. Austin Univ., 975 F. Supp. 858, 905-06 (E.D. Tex. 1996) (“The paucity of anecdotal evidence of discrimination severely diminishes plaintiffs' contention that a pattern or practice of salary discrimination against female faculty members prevails at SFA.”); see also CMCD, at 10-13 n.30 (“A cause finding of systemic discrimination should rarely be based on statistics alone.”).

IV. Standards

Standards for Systemic Compensation Discrimination Under Executive Order 11246

1. As used herein, “systemic compensation discrimination” is discrimination under a pattern or practice theory of disparate treatment.

2. Employees are similarly situated under these standards if they are similar with respect to the work they perform, their responsibility level, and the skills and qualifications involved in their positions. In determining whether employees are similarly situated under these standards, actual facts regarding employees' work activities, responsibility, and skills and qualifications are determinative. Preexisting groupings, such as pay grades or Affirmative Action Program (AAP) job groups, are not controlling; rather, such groupings may be relevant only to the extent that they do in fact group employees with similar work, skills and qualifications and responsibility levels. To determine whether such preexisting groups are relevant one must evaluate and compare information obtained from job descriptions and from employee interviews. The determination that employees are similarly situated may not be based on the fact that the contractor or subcontractor has grouped employees into a particular grouping, such as a pay grade or pay range, or that employees' pay can progress to the top of the pay grade or range based on performance or without changing jobs. Rather, such preexisting groupings may only be used if employees within the group perform similar work, and occupy positions involving similar skills, qualifications, and responsibility levels, which may be determined only by understanding employees' actual work activities. In addition to work performed, responsibility level, and skills/qualifications involved in the positions, other factors may have a significant bearing on whether employees are similarly situated. Such additional factors may include, for example, department or other functional unit of the employer, employment status (e.g., full-time versus part-time), compensation status (e.g., union versus non-union, hourly versus salaried versus commissions), etc.

3. Systemic compensation discrimination exists where there are statistically significant compensation disparities between similarly situated employees (as defined in Paragraph 2, above), after taking into account legitimate factors which influence compensation. Such legitimate factors may include education, experience, performance, productivity, location, etc. The determination of whether there are statistically significant compensation disparities between similarly situated employees after taking into account such legitimate factors must be based on a multiple regression analysis. However, legitimate factors that influence compensation may be qualitative or otherwise unquantifiable, in which case non-statistical methods must be used to explain the multiple regression analyses.

4. A compensation disparity is statistically significant under these standards if it is significant at a level of two or more standard deviations, based on measures of statistical significance that are generally accepted in the statistics profession.

5. If a pooled regression model is used, this must be accompanied by statistical tests generally accepted in the statistics profession (e.g., the “Chow test”), to determine which interaction terms should be included in the pooled regression model. Any pooled regression model must contain category factors defined in such a way as to group only similarly situated employees (as defined in Paragraph 2, above).

Standards for OFCCP Evaluation of Contractors' Compensation Practices

1. OFCCP will investigate contractors' and subcontractors' compensation practices to determine whether the contractor or subcontractor has engaged in systemic compensation discrimination under these standards. OFCCP will issue a Notice of Violations alleging systemic discrimination with respect to compensation practices based only on these standards.

2. OFCCP will make a finding of systemic compensation discrimination in those cases where there is anecdotal evidence of discrimination (as discussed in Paragraph 6, below, which notes that, except in unusual cases, OFCCP will not issue a Notice of Violation (NOV) alleging systemic compensation discrimination without providing anecdotal evidence to support OFCCP's statistical analysis) and where there exists a statistically significant (as defined in Paragraph 4, below) compensation disparity based on a multiple regression analysis that compares similarly situated employees (as defined in Paragraph 3, below) and controls for factors that OFCCP's investigation reveals influenced employees' compensation. OFCCP may reject inclusion of such a factor upon proof that the factor was actually tainted by the employer's discrimination. OFCCP will attach the regression analyses and results to, and summarize the anecdotal evidence in, the Notice of Violations issued to the contractor or subcontractor.

3. Employees are similarly situated under these standards if they are similar with respect to the work they perform, their responsibility level, and the skills and qualifications involved in their positions. In determining whether employees are similarly situated under these standards, OFCCP will collect and rely on actual facts regarding employees' work activities, responsibility, and skills and qualifications. In addition, OFCCP will investigate whether preexisting groupings, such as pay grades or AAP job groups, do in fact group employees with similar work, skills and qualifications and responsibility levels, by evaluating and comparing information obtained from job descriptions and from employee interviews. OFCCP will not base its determination that employees are similarly situated on the fact that the contractor or subcontractor has grouped employees into a particular grouping, such as a pay grade or pay range, or that employees' pay can progress to the top of the pay grade or range based on performance or without changing jobs. Rather, OFCCP will investigate whether such preexisting groupings do in fact group employees who perform similar work, and who occupy positions involving similar skills, qualifications, and responsibility levels, by looking at job descriptions and conducting employee interviews. In addition to work performed, responsibility level, and skills/qualifications involved in the positions, other factors may have a significant bearing on whether employees are similarly situated. Such additional factors may include, for example, department or other functional unit of the employer, employment status (e.g., full-time versus part-time), compensation status (e.g., union versus non-union, hourly versus salaried versus commissions), etc. OFCCP will consider the applicability of these additional factors in each case and make a determination based on the facts of the particular case.

4. A compensation disparity is statistically significant under these standards if it is significant at a level of two or more standard deviations, based on measures of statistical significance that are generally accepted in the statistics profession.

5. OFCCP will determine whether a pooled regression model is appropriate based on two factors: (a) the objective to include at least 80% of the employees (in the workforce subject to OFCCP's compliance review) in some regression analysis; and (b) whether there are enough incumbent employees in a particular regression to produce statistically meaningful results. If a pooled regression is required, OFCCP will conduct statistical tests generally accepted in the statistics profession (e.g., the “Chow test”), to determine which interaction terms should be included in the pooled regression model. In any pooled regression model, OFCCP will include category factors defined in such a way as to group only similarly situated employees (as defined in Paragraph 3, above).

6. In determining whether a violation has occurred, OFCCP will consider whether there is anecdotal evidence of compensation discrimination, in addition to statistically significant compensation disparities. Except in unusual cases, OFCCP will not issue a Notice of Violation (NOV) alleging systemic compensation discrimination without providing anecdotal evidence to support OFCCP's statistical analysis. In unusual cases, OFCCP may assert a systemic discrimination violation based only on anecdotal evidence, if such evidence presents a pattern or practice of compensation discrimination.

7. OFCCP will also assert a compensation discrimination violation if the contractor establishes compensation rates for jobs (not for particular employees) that are occupied predominantly by women or minorities that are significantly lower than rates established for jobs occupied predominantly by men or non-minorities, where the evidence establishes that the contractor made the job wage-rate decisions based on the sex, race or ethnicity of the incumbent employees that predominate in each job. Such evidence of discriminatory intent may consist of the fact that the contractor adopted a market survey to determine the wage rate for the jobs, but established the wage rate for the predominantly female or minority job lower than what that market survey specified for that job, while establishing for the predominantly male or non-minority job the full market rate specified under the same market survey.^[40]

8. OFCCP will treat compensation and other personnel information provided by the contractor to OFCCP during a systemic compensation investigation as confidential to the maximum extent the information is exempt from public disclosure under the Freedom of Information Act, 5 U.S.C. 552. It is the practice of OFCCP not to release data where the contractor is still in business, and the contractor indicates, and through the Department of Labor review process it is determined, that the data are confidential and sensitive and that the release of data would subject the contractor to commercial harm.

Signed at Washington, DC, this 12th day of June, 2006.

Victoria A. Lipnic,

Assistant Secretary for the Employment Standards,

Charles E. James, Sr.,

Deputy Assistant Secretary for Federal Contract Compliance.

Footnotes

1.  Executive Order 11246 has been amended several times since its original promulgation. For ease of reference, “Executive Order 11246” as used hereinafter refers to Executive Order 11246, as amended.

2.  See, e.g., Association of Corporate Counsel, Equal Employment Advisory Council, Gayle B. Ashton, Gaucher Associates, National Industry Liaison Group, ORC Worldwide, Society for Human Resource Management, Sonalysts, TOC Management Services, U.S. Chamber of Commerce, and World at Work. As discussed below, some of these commenters argued that OFCCP should adopt the Equal Pay Act's “substantial equality” standard.

3.  See, e.g., American Society of Employers, Berkshire Associates, Maly Consulting LLC, National Industry Liaison Group, Sonalysts, and the U.S. Chamber of Commerce.

4.  Item 11 of the Scheduling Letter currently requests “annualized compensation data (wages, salaries, commissions, and bonuses) by either salary range, grade, or level showing total number of employees by race and gender and total compensation by race and gender.”

5.  OFCCP is studying potential alternatives to use of pay grade information so that the agency can better target its investigative resources.

6.  OFCCP may modify the investigation process leading up to the application of these final interpretive standards, so as to maximize agency resources and efficiency.

7.  The “cluster regression” creates comparison groups by relying on job titles and, where a particular job title does not contain at least 30 employees and at least 5 from each comparator group (females/males, minorities/non-minorities), groups job titles based on the average compensation within each job title. In particular, the cluster regression groups job titles with the closest average compensation values until the 30/5 size requirements are reached. The cluster model uses only two or three explanatory factors in the regression, including age as a proxy for experience, and education level. As noted below, the cluster regression does not comport with Title VII standards for grouping similarly-situated employees, nor does the cluster regression include factors that were determined from an investigation of the employer's pay practices. For these reasons, the cluster regression will be used only as an indicator of potential systemic compensation discrimination; it is not a sufficient basis to issue a Notice of Violation.

8.  This is one of the arguments presented in the publication circulated in support of the pay grade theory. See “Update on Systemic Compensation Analysis,” at 1 (“It is not OFCCP's policy or practice to ‘litigate’ the merits of investigation findings at the investigatory stage of a review.”). However, the “Update on Systemic Compensation Analysis” also noted that “OFCCP has always applied Title VII principles to its methods of investigation.”Id.

9.  Section 3R(a) of OFCCP's Federal Contract Compliance Manual (FCCM) provides that “compensation discrimination” encompasses “[d]isparate treatment in pay in relationship to the established range for a job, whether at entry or later; e.g., Blacks with similar backgrounds to Whites on the legitimate factors considered for initial salary are hired at less money, etc. * * *.” To the extend that this reference, or any other reference in the FCCM, implies the pay grade theory or any other theory of compensation discrimination that permits comparison of compensation of individuals who are not similarly situated under these final interpretive standards, or otherwise conficts with these interpretive standards, these interpretive standards supercede the FCCM in that regard.

10.  ETF argues that the fact that Williams was unpublished and, under Fifth Circuit rules, cannot be cited as precedent, “undermines the case's significance.” However, under Rule 47.5.4 of the Local Rules of Appellate Procedure for the United States Court of Appeals for the Fifth Circuit, “[a]n unpublished opinion may, however, be persuasive. An unpublished opinion may be cited, but if cited in any document being submitted to the court, a copy of the unpublished opinion must be attached to each document. The first page of each unpublished opinion bears the following legend: Pursuant to Loc. R. 47.5, the court has determined that this opinion should not be published and is not precedent except under the limited circumstances set forth in Loc. R. 47.5.4.” A district court in the Fifth Circuit has found the reasoning in Williams to be persuasive. See Dean v. Kimberly-Clark Corp., No. 3:02-CV-1682-K, 2005 WL 309509, at *2 (N.D. Tex. Feb. 8, 2005) (“Plaintiff claims that Kimberly-Clark discriminated against him by failing to compensate him at the same rate it compensated its Process Specialists, although he admits he was a Production Officer, not a Process Specialist. “If a plaintiff's job responsibilities are significantly different from the responsibilities of employees [he] cites as a point of comparison, then the plaintiff has not made out a prima facie case.” Williams 78 Fed. Appx. at 949.”). In addition to Williams, the district court in Woodward v. United Parcel Serv., Inc., 306 F. Supp. 2d 567, 574-75 (D. S.C. 2004), expressly rejected the pay grade theory as a basis for establishing a prima facie case of compensation discrimination: “In order to establish a prima facie case of pay discrimination, Woodward must show that he * * * was paid less than similarly situated employees who were outside his protected class * * *. Woodward has not identified any relevant group of similarly situated comparators to support his claim of pay discrimination * * *. In 1998, Woodward transferred to the District Assessor position in the South Carolina District—a job in which he had no comparators because the other six Grade 16 managers in the IE department during 1998 and 1999 (while Woodward was the Assessor) all held positions with significantly different duties * * *. In summary, Woodward has failed to identify any comparators who are similarly situated with respect to pay. Woodward has made no effort to demonstrate that any of the alleged comparators that he has identified held positions whose duties were the same as or substantially similar to his own. Instead, Woodward relies solely on his unsupported assertion that all Grade 16 level employees are similarly situated with respect to pay.”

11.  By contrast, plaintiffs were successful in their claims when they offered evidence that they were similarly situated based on the work they performed, and the responsibility level, skills, and qualifications involved in their positions. See, e.g., Brinkley-Ubo v. Hughes Training Inc., 36 F.3d 336, 343 (4th Cir. 1994) (“The plaintiff may establish a prima facie case by demonstrating * * * that the job she occupied was similar to higher paying jobs occupied by males.”); Miranda v. B&B Cash Grocery Store, Inc., 975 F.2d 1518, 1526-31 (11th Cir. 1992) (“We agree with the trial court that Miranda carried her burden of proof and established that B & B discriminated against her because of her gender. The plaintiff establishes a prima facie case of sex discrimination under Title VII by demonstrating that she is female and that the job she occupied was similar to higher paying jobs occupied by males. The trial court found that Miranda's description of the type of duties she performed as a buyer, as well as testimony from defendant's witnesses established that she shared the same type of tasks as the other buyers.”).

12.  Of course, if OFCCP used pay grade as the initial grouping, subject to the employer's rebuttal that the jobs were dissimilar, employers typically would argue that the pay grade grouped positions that were dissimilar, as they did throughout the period that OFCCP used the pay grade theory. However, in the past, OFCCP generally did not investigate the employer's contention that the jobs were dissimilar because the pay grade theory assumed that employees were similarly situated if they were in the same pay grade, regardless of whether they were similar or dissimilar in the work they performed, their responsibility levels, or the skills and qualifications involved in their positions. However, if OFCCP used grade as the initial grouping subject to the employer's rebuttal that the jobs were dissimilar, OFCCP could not simply accept the employer's contention that jobs were dissimilar, but would have to investigate whether the facts supported the employer's contention. This would require OFCCP to conduct the same type of factual investigation specified in these final interpretive standards.

13.  See, e.g., Association of Corporate Counsel, Equal Employment Advisory Council, HR Analytical Services, National Industry Liaison Group, ORC Worldwide, TOC Management Services, U.S. Chamber of Commerce, and World at Work.

14.  See, e.g., Sprague v. Thorn Americas, Inc., 129 F.3d 1355 (10th Cir. 1997); Mulhall v. Advance Sec., Inc., 19 F.3d 586 (11th Cir. 1994); Brinkley-Obu v. Hughes Training, Inc., 36 F.3d 336 (4th Cir. 1994); Miranda v. B & B Cash Grocery Store, Inc., 975 F.2d 1518 (11th Cir. 1992); Crockwell v. Blackmon-Mooring Steamatic, Inc., 627 F. Supp. 800 (W.D. Tenn. 1985).

15.  Because Gunther-type claims are unique, OFCCP has not included a paragraph regarding such claims in the “Standards for Systemic Compensation Discrimination Under Executive Order 11246.”

16.  See, e.g., Equal Employment Advisory Council, Morgan, Lewis & Bockius LLP, Northeast Region Corporate Industry Liaison Group, ORC Worldwide, and Picha & Salisbury, Society for Human Resource Management.

17.  OFCCP has cited cases in this preamble that discuss whether specific positions are similarly situated. There are hundreds of other federal court pay discrimination cases that discuss whether other positions are similarly situated based on facts about the specific positions involved in each of those cases.

18.  See, e.g., Equal Employment Advisory Council, Gaucher Associates, and World at Work.

19.  OFCCP reserves the right, in rare cases, to perform non-statistical analyses on the wages of those employees who are not similarly situated to any other employee, such as high-level executives.

20.  See, e.g., Berkshire Associates, Equal Employment Advisory Council, HR Analytical Service, Society for Human Resource Management, U.S. Chamber of Commerce, and World at Work.

21.  Unfortunately, these commenters did not cite any cases in which the court accepted these types of analysis to prove systemic compensation discrimination. OFCCP currently is studying methods for evaluating promotion practices for systemic discrimination and does not intend this discussion to foreclose exploration of such analysis for that purpose.

22.  See, e.g., American Society of Employers, Gaucher Associates, Glenn Barlett Consulting Services, HR Analytical Services, National Industry Liaison Group, and Picha & Salisbury.

23.  See, e.g., Berkshire Associates Inc., HR Analytical Services, and Northeast Region Corporate Industry Liaison Group.

24.  As noted in the preamble of the proposed interpretive standards and restated below, if separate regressions by categories of jobs would not permit OFCCP to assess the way the contractor's compensation practices impact on a significant number of employees, OFCCP may perform a “pooled” regression, which combines these categories of jobs into a single regression (while including an OFCCP-developed category factor in the “pooled” regression that controls for groupings of employees who are similarly situated based on work performed, responsibility level, and skills and qualifications).

25.  See, e.g., DCI Consulting, Equal Employment Advisory Council, Gaucher Associates, Gayle B. Ashton, Glenn Barlett Consulting Services, Peopleclick Research Institute, and Society for Human Resource Management.

26.  OFCCP's strong preference for anecdotal evidence does not imply that the agency believes that anecdotal evidence is sufficient to refute statistical or other evidence of a pattern or practice of discrimination. OFCCP's use of anecdotal evidence fits into the pattern-or-practice framework established by the Supreme Court in Intl'l Bhd. of Teamsters v. United States, 431 U.S. 324, 360 & n. 46 (1977) (citations omitted):

“The plaintiff in a pattern-or-practice action is the Government, and its initial burden is to demonstrate that unlawful discrimination has been a regular procedure or policy followed by an employer or group of employers. At the initial, “liability” stage of a pattern-or-practice suit the Government is not required to offer evidence that each person for whom it will ultimately seek relief was a victim of the employer's discriminatory policy. Its burden is to establish a prima facie case that such a policy existed. The burden then shifts to the employer to defeat the prima facie showing of a pattern or practice by demonstrating that the Government's proof is either inaccurate or insignificant * * *. The employer's defense must, of course, meet the prima facie case of the Government. We do not mean to suggest that there are any particular limits on the type of evidence an employer may use. The point is that at the liability stage of a pattern-or-practice trial the focus often will not be on individual hiring decisions, but on a pattern of discriminatory decisionmaking. While a pattern might be demonstrated by examining the discrete decisions of which it is composed, the Government's suits have more commonly involved proof of the expected result of a regularly followed discriminatory policy. In such cases the employer's burden is to provide a nondiscriminatory explanation for the apparently discriminatory result.”

27.  As discussed in the cases cited above, one would expect some anecdotal evidence of compensation discrimination if the employer has engaged in systemic compensation discrimination. However, there may be unusual factors, applicable in a particular case, which explain why OFCCP was unable to uncover anecdotal evidence during its investigation despite the statistical evidence of systemic compensation discrimination.

28.  This issue does not arise in a Gunther-type claim, which does not involve statistical evidence. See discussion in text above.

29.  The Administrative Review Board, and, before its creation, the Secretary of Labor, have turned to Title VII standards for determining compliance with the nondiscrimination requirements of Executive Order 11246. See, e.g., OFCCP v. Greenwood Mills, Inc., 89-OFC-039, ARB Final Decision and Order, December 20, 2002, at 5; OFCCP v. Honeywell, 77-OFCCP-3, Secretary of Labor Decision and Order on Mediation, June 2, 1993, at 14 and 16, Secretary of Labor Decision and Remand Order, March 2, 1994. The EEOC has issued guidance on compensation discrimination in the form of a chapter in the EEOC Compliance Manual on “Compensation Discrimination.” EEOC Directive No. 915.003 (Dec. 5, 2000). EEOC is the agency with primary enforcement responsibility for Title VII and its interpretations of that statute constitute a body of experience and informed judgment to which courts and litigants can turn for guidance. See, e.g., Clackamas Gastroenterology Assocs., P.C. v. Wells, 538 U.S. 440, 449 n.9 (2003) (citing with approval and quoting from an EEOC Compliance Manual chapter applicable to Title VII).

30.  The term “systemic compensation discrimination” used hereinafter references compensation discrimination under a disparate treatment, pattern or practice theory of discrimination. These interpretive standards address only systemic compensation discrimination. However, nothing in these final interpretive standards precludes OFCCP from investigating and alleging compensation discrimination under an individual disparate treatment theory or under a disparate impact theory of compensation discrimination in accordance with applicable law.

31.  By contrast to sex-based compensation discrimination, OFCCP has published regulations providing specific guidance with respect to hiring discrimination. Thus, OFCCP is a signatory to the Uniform Guidelines on Employee Selection Procedures (UGESP), which provide formal guidance as to how OFCCP evaluates contractors' selection procedures to determine compliance with Executive Order 11246. See 41 CFR part 60-3. Before being published as a final rule, 43 Fed. Reg. 38290 (August 25, 1978), UGESP was published in the Federal Register as a proposed rule and subject to public comment. See 42 Fed. Reg. 65542 (December 30, 1977).

32.  The final interpretive standards contained in this Notice are intended to provide definitive interpretations of both the SDG and Executive Order 11246 with respect to systemic compensation discrimination, regardless of the specific basis (e.g., sex, race, national origin, etc.) of the discrimination.

33.  Although used in practice by several OFCCP regions for several years, the grade theory was never formally adopted by OFCCP.

34.  OFCCP officials informally distributed the SCA and the Update in the late 1990's. They were not published by OFCCP nor did they bear any indication of formal agency approval, e.g., they were not printed on OFCCP letterhead.

35.  This method was not described in materials made available to the general public. The method was used primarily in OFCCP's Southeast Region.

36.  OFCCP's principal basis for rejecting the grade theory is that it allows for comparison of employees who are not similarly situated under applicable legal standards, as discussed in the text. However, an alternative reason for OFCCP's rejection of the grade theory applies specifically to attempts to justify the use of pay grades to compare dissimilar employees or jobs on the grounds that the employees perform or the jobs entail (dissimilar) work that has equal or similar “value” or “worth” to the employer. See Update, at 6 (justifying use of pay grade on grounds that by creating pay grades the employer has “identif[ied] certain jobs as having similar value to the organization.”). Regardless of whether the worth or value of the dissimilar work or jobs is alleged to have been established by the employer (i.e., by placing the employee or the employee's job into a particular pay grade along with other, dissimilar employees or jobs) or by someone other than the employer, the attempt to compare employees who are performing dissimilar work or who occupy dissimilar jobs based on the “value” or “worth” of the work or jobs, constitutes the comparable worth theory of compensation discrimination, which has been widely discredited by the courts. See American Federation of State, County, and Municipal Employees v. State of Washington, 770 F.2d 1401, 1404 (9th Cir. 1985) (“The comparable worth theory, as developed in the case before us, postulates that sex-based wage discrimination exists if employees in job classifications occupied primarily by women are paid less than employees in job classifications filled primarily by men, if the jobs are of equal value to the employer, though otherwise dissimilar.”); Colby v. J.C. Penney Co., 811 F.2d 1119, 1125-26 (7th Cir. 1987 (describing comparable worth theory as “bas[ing] liability on the fact that the[] employer paid higher wages to workers in job classifications predominantly occupied by men than to workers in job classifications predominantly occupied by women, though it paid the same wages to men and women within each classification”); American Nurses Association v. Illinois, 783 F.2d 716, 720-22 (7th Cir. 1986) (considering plaintiffs “charge that the state pays workers in predominantly male job classifications a higher wage not justified by any difference in the relative worth of the predominantly male and the predominantly female jobs in the state's roster.”); Lemons v. City and County of Denver, 620 F.2d 228, 229 (10th Cir. 1980) (“In summary, the suit is based on the proposition that nurses are underpaid in City positions, and in the community, in comparison with other and different jobs which they assert are of equal worth to the employer.”); Christensen v. Iowa, 563 F.2d 353, 354-56 (8th Cir. 1977) (“Appellants, who are clerical employees at UNI, argue that UNI's practice of paying male plant workers more than female clerical workers of similar seniority, where the jobs are of equal value to UNI, constitutes sex discrimination and violates Title VII”); see also County of Washington v. Gunther, 452 U.S. 161, 165 (1981) (“Respondents’ claim is not based on the controversial concept of “comparable worth” under which plaintiffs might claim increased compensation on the basis of a comparison of the intrinsic worth or difficulty of their job with that of other jobs in the same organization or community.” [footnotes omitted]); Gunther, 452 U.S. at 203 (Rehnquist, J., dissenting) (“The opinion does not endorse the so-called ‘comparable worth’ theory: though the Court does not indicate how a plaintiff might establish a prima facie case under Title VII, the Court does suggest that allegations of unequal pay for unequal, but comparable, work will not state a claim on which relief may be granted. The Court, for example, repeatedly emphasizes that this is not a case where plaintiffs ask the court to compare the value of dissimilar jobs or to quantify the effect of sex discrimination on wage rates.”); Judith Olans Brown et al., Equal Pay for Jobs of Comparable Worth: An Analysis of the Rhetoric, 21 Harv. C.R.-C.L. Rev. 127, 129 (1986) (“ ‘Comparable worth’ means that workers, regardless of their sex, should earn equal pay for work of comparable value to their common employer * * *. The basic premise of comparable worth theory is that women should be able to substantiate a claim for equal wages by showing that their jobs and those of male workers are of equal value to their common employer.”); Hydee R. Feldstein, Comment, Sex-Based Wage Discrimination Claims After County of Washington v. Gunther, 81 Colum. L. Rev. 1333, 1333 (1981) (noting comparable worth “theory holds that employees performing work of equal value, even if the work they do is different, should receive the same wages.”).

37.  Federal courts disagree on whether the Equal Pay Act's standard of “substantial equality” applies to gender-based pay discrimination claims under Title VII, absent direct evidence of discrimination. See, e.g., Conti v. Universal Enter., Inc., 50 Fed. Appx. 690, 2002 WL 31108827, at *7 (6th Cir. Sept. 20, 2002); Clark v. Johnson & Higgins, 181 F.3d 100, 1999 WL 357804, at *3-*4 (6th Cir. May 28, 1999) (Text in Westlaw); Loyd v. Phillips Bros., Inc., 25 F.3d 518, 525 (7th Cir. 1994); EEOC v. Sears, Roebuck & Co., 839 F.2d 302, 243-53 (7th Cir. 1988); Merrill v. S. Methodist Univ., 806 F.2d 600, 606 (5th Cir. 1986); McKee v. Bi-State Dev. Agency, 801 F.2d 1014, 1019 (8th Cir. 1986); Plemer v. Parsons-Gilbane, 713 F.2d 1127, 1133-34 (5th Cir. 1983); see also CMCD, at 10-6 n.18. Because an OFCCP enforcement action may be subject to APA review in a federal court that does not adopt the “similarly situated” standard, the Department will address this issue on a case by case basis.

38.  In this respect, OFCCP will not rely on the grade theory assumptions discussed supra., at Sections IIIC and IIID.

39.  An “interaction term” is a factor used in the regression model whose value is the result of a combination of subfactors, which allows the factor to vary based on the combined effect of the subfactors. For example, a performance by job level interaction term would allow performance to have a different impact on compensation depending on the job level.

40.  See County of Washington v. Gunther, 452 U.S. 161, 166, 180-81 (1981) (“We emphasize at the outset the narrowness of the question before us in this case. Respondents' claim is not based on the controversial concept of “comparable worth,” under which plaintiffs might claim increased compensation on the basis of a comparison of the intrinsic worth or difficulty of their job with that of other jobs in the same organization or community. Rather, respondents seek to prove, by direct evidence, that their wages were depressed because of intentional sex discrimination, consisting of setting the wage scale for female guards, but not for male guards, at a level lower than its own survey of outside markets and the worth of the jobs warranted.”).