2013-04675. Interpreting Nondiscrimination Requirements of Executive Order 11246 With Respect to Systemic Compensation Discrimination and Voluntary Guidelines for Self-Evaluation of Compensation Practices for Compliance With Nondiscrimination ...

AGENCY:

Office of Federal Contract Compliance Programs, Labor.

ACTION:

Notice of final rescission.

SUMMARY:

The Office of Federal Contract Compliance Programs (OFCCP) is publishing a final notice rescinding two guidance documents: The Interpreting Nondiscrimination Requirements of Executive Order 11246 with respect to Systemic Compensation Discrimination and Voluntary Guidelines for Self-Evaluation of Compensation Practices for Compliance with Executive Order 11246 with respect to Systemic Compensation Discrimination. Rescinding these prior guidance documents will improve OFCCP's ability to enforce the Executive Order's ban on pay discrimination. It will eliminate a rarely used, ineffective and burdensome compliance procedure. This rescission allows OFCCP to better direct its resources for the benefit of victims of discrimination, the government, contractors, and taxpayers.

DATES:

Effective February 28, 2013.

FOR FURTHER INFORMATION CONTACT:

Debra A. Carr, 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:

I. Introduction

A. Background

The Department of Labor's OFCCP enforces Executive Order 11246, as amended, which requires Federal Government contractors and subcontractors to provide equal employment opportunity through affirmative action and nondiscrimination based on race, color, national origin, religion, and sex. Compensation discrimination is one form of discrimination prohibited by the Executive Order. In particular, federal contractors ^[1] may not discriminate in “rates of pay or other forms of compensation.” 41 CFR 60-1.4(a)(1). OFCCP enforces this requirement through review and investigation of contractor pay practices, data and other relevant information for potential systemic and individual evidence of discrimination. In addition, contractors must review and monitor their compensation systems to “determine whether there are gender-, race-, or ethnicity-based disparities.” ^[2] Contractors must maintain records, including but not limited to “rates of pay or other terms of compensation.” ^[3]

OFCCP enforces the Executive Order's nondiscrimination provisions, including the ban on compensation discrimination, consistent with Title VII. Title VII forbids discrimination in employment, which includes paying employees differently on the basis of race, sex or other protected class membership. Congress intended for courts to read this ban broadly. Franks v. Bowman Transportation Co., 424 U.S. 747, 763 (1976) (“Congress intended to prohibit all practices in whatever form which create inequality in employment opportunity due to discrimination * * *”) (citations omitted). There have long been three distinct theories of discrimination under Title VII: Individual disparate treatment, “pattern or practice” (systemic disparate treatment), and disparate impact. While courts have developed some specific mechanisms for presenting evidence and satisfying the burden of proof under each theory, they consistently hold that there is no single way to prove discrimination.^[4] Plaintiffs may rely on any evidence of discrimination, whether direct, circumstantial, statistical, anecdotal, or any combination of such evidence.

This flexibility is critical because discrimination may be difficult to identify. Pay discrimination can be easy to spot, like a clear pattern of paying women less than men in the same job, where they are just as qualified. But it can also be complex, like a practice of discriminating against African-American sales workers in handing out territory assignments—so that no matter how well they perform, they can never have the same earnings opportunities as their white counterparts. Title VII addresses all forms of compensation differences, including those that come from channeling a favored group into the better paying entry level jobs with better long-term opportunities, or where glass ceilings or other unfair promotion practices wrongly block advancement of talented workers on the basis of illegal criteria like race or gender. And even where base wages or salaries are fair, discrimination in access to overtime, or higher paying shifts, or bonuses, can add up to unequal take home pay in violation of federal civil rights law.

Further, because there is so much variation in pay practices across industries, employers and types of jobs, investigating compensation discrimination requires considering evidence and data in context, which is the approach that federal courts have embraced when interpreting Title VII. It is not possible to specify in advance a single test, model or framework that accurately and fairly identifies discriminatory pay differences in every case. Attempting to impose a uniform test for pay discrimination without accounting for case-specific facts creates opportunities for error. It means that some contractors who pay fairly will be wrongly identified as discriminating in pay, and that some workers who were underpaid due to discrimination will be left without a remedy. Investigating and addressing compensation discrimination requires a rigorous fact-based assessment of a broad array of pay practices.

Nevertheless, OFCCP has since 2006 narrowed its focus, following two guidance documents: Interpreting Nondiscrimination Requirements of Executive Order 11246 with respect to Systemic Compensation Discrimination (Standards) and the Voluntary Guidelines for Self-Evaluation of Compensation Practices for Compliance with Executive Order 11246 with respect to Systemic Compensation Discrimination (Voluntary Guidelines).^[5] The Standards establish analytical procedures to be followed generally by OFCCP when issuing a Notice of Violation (NOV) alleging systemic compensation discrimination. The Voluntary Guidelines provide a methodology for contractors' self-evaluation of their pay practices, required by 41 CFR 60-2.17(b)(3); following that methodology can provide a “safe harbor” during compliance reviews.

The Standards and Voluntary Guidelines addressed only a single type of pay practice using limited evidence and a highly specified analytic framework. The Standards did not actually apply to or explain investigation procedures, and left many critical details undefined or subject to potential exceptions. The companion document, the Voluntary Guidelines, attempted to tell contractors exactly how to fulfill their regulatory self-monitoring obligations. Yet the Voluntary Guidelines were similarly inadequate, and contractors rarely utilized them to demonstrate compliance.

In 2010, President Obama created the National Equal Pay Task Force, bringing together the Department of Labor (DOL), the Equal Employment Opportunity Commission (EEOC), the Department of Justice, and the Office of Personnel Management to collectively address pay discrimination under their enforcement mandates. The Director of OFCCP, a member of the Task Force, committed OFCCP to review and revise its enforcement guidance and practices to more effectively address compensation discrimination under Executive Order 11246. OFCCP reevaluated all aspects of its existing approach to addressing compensation discrimination by federal contractors, including (1) Guidance documents, (2) procedures for conducting compliance evaluations, (3) training and best practices for investigating and addressing compensation discrimination, and (4) approaches of other federal agencies including the EEOC and the Department of Justice.

As part of this larger review and revision process, OFCCP assessed the role of the Standards and Voluntary Guidelines, concluding they were inconsistent with the Task Force's goal of improving enforcement. On January 3, 2011, OFCCP published a Notice of Proposed Rescission (NPR), proposing to rescind the Standards and the Voluntary Guidelines in their entirety, and soliciting public comment. 76 FR 62. Because neither 2006 guidance document has proved workable or effective in practice, OFCCP is rescinding both guidance documents effective immediately.

In their place, OFCCP is today committing to provide greater clarity for contractors and improve equal employment protection for workers. First, OFCCP will be applying Title VII principles as the basis for determining whether a contractor has violated the Executive Order's ban on pay discrimination, just as the agency does in assessing contractor compliance with respect to all other employment practices. Second, as explained in Section III below, OFCCP is disclosing its interpretation of specific legal and technical issues to assist contractors in evaluating their own practices and promoting greater voluntary compliance. Third, OFCCP will be providing much greater transparency on questions of investigation practices and procedures—issues the Standards did not address—both in this document as well as via ongoing compliance assistance. Collectively, this information should provide ample notice to contractors of their legal obligations as well as assist them in achieving voluntary compliance.

B. Summary of the Reasons for Rescission

The 2006 Standards set forth a single, specified analytical procedure to be used for determining a violation in all systemic compensation discrimination cases, except in unusual circumstances. Under the 2006 Standards, OFCCP was to apply the same analytic framework regardless of the industry, types of jobs, issues presented, characteristics of workers, or available data. In particular, OFCCP may only establish a systemic compensation violation of the Executive Order by testing narrowly defined groupings of employees based on standards typically used in individual disparate treatment cases. 71 FR at 35127-28, 35140. Under the Standards, OFCCP must use multiple regression analysis to test for pay disparities and must have anecdotal evidence to establish a systemic compensation violation, “except in unusual cases.” 71 FR at 35141. As explained above, employment discrimination comes in many forms, which is why Title VII permits a flexible case-specific approach to proof.

The Standards restrict OFCCP's ability to enforce the Executive Order's non-discrimination mandate. The Standards address a single kind of compensation disparity—pay differences among discrete pools of workers limited by job category—to the potential exclusion of other equal opportunity concerns. Pay differences arising from discrimination in job assignments, unequal access to promotional opportunities, channeling and glass ceiling issues can be obscured by the strict grouping requirements of the Standards. The Standards do not favor aggregation and place additional burdens on OFCCP where a pooled regression is used, despite the longstanding legal rule that the proper level of aggregation requires a case by case determination. The regression analysis required under the Standards is not always appropriate or feasible—other approaches may be preferable for certain cases involving very high level or specialized positions or smaller workforces, and cases involving missing or flawed data, among others. The Standards create a special rule for anecdotal evidence in compensation cases that has never been applied in other OFCCP contexts, and which is particularly burdensome for workers who frequently lack meaningful access to information about pay. Fair and effective enforcement requires tailoring the compensation investigation and analytical procedures to the facts of the case based on Title VII principles.

Similarly, the Voluntary Guidelines establish a single one-size-fits-all statistical model that contractors can elect to use in conducting the self-analysis of their pay practices required by 41 CFR 60-2.17(b)(3). As an incentive to encourage contractors to use the analytical procedures contained in the Voluntary Guidelines, OFCCP would deem a contractor, whose self-evaluation meets the procedures outlined in the Voluntary Guidelines, to be in compliance with section 60-2.17(b)(3). OFCCP would then coordinate review of the contractor's compensation practices during a compliance evaluation in the manner specified in the Voluntary Guidelines. 71 FR at 35122. In other words, contractors may provide their own analysis of pay data, based on their own determinations of how to apply the Voluntary Guidelines, and as long as it “reasonably meets” the procedures detailed in the Voluntary Guidelines, OFCCP is bound to accept it. Even if another, equally “reasonable” analytic approach would reveal systemic compensation discrimination against a protected class of workers, OFCCP would seemingly have to consider the contractor in compliance.

In addition, the Voluntary Guidelines provide little practical benefit. As the comments discussed below demonstrate, few contractors rely on this model for purposes of a compliance evaluation. This assessment is consistent with OFCCP's own experience since 2006, that “contractors have rarely utilized the analytical procedures outlined in the Voluntary Guidelines when analyzing their compensation practices under section 60-2.17(b)(3).” 76 FR at 63. Further, the Voluntary Guidelines, like the Standards, take an overly narrow approach to analyzing potential systemic pay discrimination—which may lead contractors to shortchange the ongoing monitoring required by OFCCP regulations and best practices. The Voluntary Guidelines do not appear to have improved the level or quality of voluntary compliance.

Non-discrimination in pay is a critical issue for workers and their families, and a cornerstone of OFCCP's equal employment protections. As detailed in the NPR, identifying and remedying compensation discrimination has long been an important goal of OFCCP compliance efforts. 76 FR at 62. The Executive Order and the implementing regulations specifically require contractors to ensure pay equity. They place federal contractors under affirmative duties to maintain data, conduct internal reviews and monitor pay practices for potential discrimination, and comply with the Executive Order's ban on discrimination in the paying of wages, salaries, and other forms of compensation. Sec. 202 of E.O. 11246, as amended, 41 CFR 60-1.12; 60-1.4; 60-2.17(b)-(d). Nevertheless, Bureau of Labor Statistics data and numerous research studies indicate that disparities in compensation on the basis of sex and race continue to exist,^[6] even after accounting for factors such as the type of job, worker qualifications such as experience and education, and other potential explanations.^[7] Further, because many employees do not know how their pay compares to others, OFCCP compliance reviews and contractor voluntary compliance efforts are critical tools for uncovering systemic pay disparities invisible to individual workers. In light of these concerns, OFCCP should no longer limit its analysis or consideration of evidence that points to potential compensation discrimination in violation of the Executive Order. Nor should OFCCP encourage contractors to limit their self-evaluation practices to a single form of inquiry.

The Standards and Voluntary Guidelines may also lead OFCCP to enforcement approaches that are inconsistent with how other federal agencies address pay discrimination. OFCCP is presently working to harmonize its approach with that of other federal enforcement agencies, including the Department of Justice and the EEOC. Neither restricts its analytic and evidentiary framework to a single approach. Along with OFCCP, these agencies have committed to vigorous enforcement of federal non-discrimination mandates. Through this rescission, OFCCP seeks to provide workers the full protection of Title VII anti-discrimination provisions and ensure consistent enforcement in its review of contractor compensation practices.

Finally, by setting special analytical procedures restricting what constitutes proof of discrimination for a particular employment practice, the Standards and Voluntary Guidelines depart from OFCCP's approach to evaluating contractor compliance in other areas. There are no comparable Standards or Voluntary Guidelines for systemic discrimination in hiring, promotion, termination or other employment practices. In those other areas, Title VII principles have proved more than adequate to put contractors on notice of their obligation, to promote voluntary compliance measures, and to define the parameters of a violation. OFCCP has traditionally focused on identifying discrimination through the development of a variety of investigative and analytical tools. Compensation should be no exception.

After considering the comments received, OFCCP concludes that rescinding these prior guidance documents will improve OFCCP's ability to enforce the Executive Order ban on pay discrimination. It will eliminate a rarely used, ineffective and burdensome compliance procedure. This rescission allows OFCCP to direct its resources more efficiently—for the benefit of victims of discrimination, the government, contractors, and taxpayers. These Standards and Voluntary Guidelines have not been useful tools in combating compensation discrimination. OFCCP can better achieve the objectives of the Executive Order—including non-discrimination in pay for the federal contractor workforce—through other methods of investigation and analysis.

Nevertheless, OFCCP takes seriously its obligation to support contractors seeking to comply voluntarily, and wishes to promote transparency and fairness regarding OFCCP practices. The agency will be providing as much clarity as possible regarding its application and interpretation of important legal, factual and technical issues in assessing systemic compensation discrimination, both in this document (see Section III, below) and going forward. OFCCP traditionally has established procedures for investigating compensation discrimination, as well as other forms of discrimination, through instructions for its compliance officers contained in the OFCCP Federal Contract Compliance Manual (FCCM), directives, and other staff guidance materials, and will continue to do so. OFCCP will provide ongoing technical assistance through tools such as written frequently asked questions (FAQs), conference calls, webinars and online chats and will seek opportunities to take questions and get feedback from all stakeholders. OFCCP is not currently contemplating additional formal rulemaking associated with this document.

II. Discussion of the Comments

OFCCP received 22 comments on the NPR from the following: employer associations; employee and other women's and workers' rights associations; named employers, including consultants and law firms focused on employment and personnel practices; a comment from a group of 40 statisticians, economists, sociologists, and psychologists (Social Science Researchers); and one individual comment. OFCCP has considered all of the comments received. Of the 22 comments, ten support the proposed rescission of both the Standards and Voluntary Guidelines, five oppose the proposed rescission of the Standards, and three oppose the proposed rescission of the Voluntary Guidelines—with one comment that recommends partial rescission of the Voluntary Guidelines. The remaining comments do not clearly state a position, but instead comment on particular issues.

With regard to the Standards, comments addressed the following issues: (1) The framework under the Standards for determining the proper comparison groups for analysis; (2) the mandate of the Standards to use multiple regression analysis; (3) the mandate of the Standards that OFCCP have anecdotal evidence; (4) the OFCCP proposal to rely on multiple investigative and analytical methods to address compensation discrimination issues; and (5) cost to contractors should the Standards be rescinded. See Section II.A.

With regard to the Voluntary Guidelines, comments addressed the following issues: (1) Whether the Voluntary Guidelines are effective; (2) substantive limitations of the Voluntary Guidelines; and (3) cost to contractors should the Voluntary Guidelines be rescinded. See Section II.B.

A. Comments Regarding the Standards

The Standards prescribe procedures that limit OFCCP's ability to determine when a contractor has violated the Executive Order. They restrict permissible evidence and require one form of proof of potential systemic compensation discrimination, except in unusual circumstances. These restrictions govern how to group employees for analysis, the use of multiple regression analysis to decide whether wage differences are discriminatory, and the requirement for anecdotal evidence of compensation discrimination except in unusual cases. These procedures are to be followed regardless of the facts of a particular case.

1. How To Compare Workers for Purposes of Compensation Analysis

Under the Standards, OFCCP can generally only establish a systemic compensation violation where there are statistically significant pay disparities comparing highly specified groups of workers. In particular, OFCCP is to begin by establishing groups of “similarly situated” workers on the basis of the positions they hold, and then to test for pay differences only among those workers within each separate group. See 71 FR 35140. The Standards make it more difficult for OFCCP to test for larger patterns across groups of jobs. These restrictions include the formal limits on the use of an aggregate (or pooled) statistical analysis, as well as the job-based comparison requirements, which can make it more difficult to investigate the effect that discrimination in job assignment, level or position has on pay.

Nearly half of the commenters addressed how the Standards require OFCCP to compare workers for purposes of analysis. Two commenters specifically identified concerns with the definition of “similarly situated employees” or the requirement to group workers a specific way—calling it “overly stringent,” “problematic and easily misinterpreted,” and inconsistent with “professional best practices.” Two commenters explicitly supported the appropriateness of comparing similarly situated employees as described and defined by the Standards on legal grounds.

The commenters supporting rescission raised several specific problems with this aspect of the Standards. A women's rights group pointed out a technical problem with performing separate analysis on each group of similarly situated workers. Especially if these groups are small, the analysis may be “underpowered”—and therefore unable to accurately detect discrimination when it exists. Another women's rights group expressed concern that the basis for grouping under the Standards could incorporate discrimination. The Standards define similarly situated employees based on position qualifications, even though qualifications can be illegal barriers where they operate to exclude a protected class from the job. The Social Science Researchers noted that a determination of how to group employees for analysis must be made based on the particular facts and circumstances, Title VII principles, and professional best practices. Multiple commenters pointed out that compensation discrimination takes many forms, and that the OFCCP's analysis should be flexible enough to address all pay issues that may exist in a contractor's workforce.

Commenters supporting the Standards on this point largely relied on the view of applicable law underpinning the Standards themselves. For example, one management law firm referred to the Title VII cases cited in the original Federal Register notice establishing the Standards and noted that the NPR provided no legal authority to the contrary. An employers' association stated that the Standards follow the EEOC's Compliance Manual and that therefore they represent “adherence to Title VII principles.” As explained below, the Standards do not comport with a more comprehensive understanding of applicable legal principles relevant to potential pay discrimination. Nor do they accurately reflect the contents of the EEOC Compliance Manual, which rejects the view that there is a single way to prove discrimination.

OFCCP has concluded that this aspect of the Standards is overly narrow and creates both technical and substantive barriers to effective enforcement. The legal analysis OFCCP used to support its adoption of the Standards in 2006 did not explicitly discuss the most common approaches for proving systemic discrimination. See 71 FR 35126-28. The Standards applied a model typically applied in individual disparate treatment cases, limiting the types of evidence and models of proof in systemic cases. Further, the Standards require a specific technical approach that substantially increases the risk that OFCCP would fail to detect improper pay disparities, and limits investigations to a single form of pay discrimination. Thus, as explained in the paragraphs that follow, the Standards hamper OFCCP's ability to ensure contractor compliance with the Executive Order.

The Standards inadequately rely on an inquiry relevant to individual disparate treatment cases to evaluate systemic discrimination. Individual disparate treatment cases typically proceed under the McDonnell Douglas burden shifting framework. Reeves v. Sanderson Plumbing Products, Inc., 530 U.S. 133 (2000); McDonnell Douglas Corp. v. Green, 411 U.S. 792 (1973). Many individual disparate treatment cases rely heavily on “comparators”—specifically identified workers who are similarly situated to the plaintiff but outside the protected class. An otherwise unexplained difference in how the employer treated the plaintiff and her comparator permits an inference that the real reason was intentional discrimination.^[8]

Virtually all of the cases OFCCP used to support the Standards focus on how to prove individual instances of pay discrimination. See 71 FR 35127-28. These cases involve disputes over whether a particular comparator is appropriate or not because of differences in their positions. But there are other ways to prove discrimination in individual cases.^[9] More importantly, systemic cases are not based on person to person comparisons but on patterns within a workforce that can transcend specific workers, jobs, locations, or functions.

Proof in systemic disparate treatment cases can go beyond the single scenario of the Standards. Rather than asking whether an employer intentionally discriminated against a specific person, systemic cases ask whether there is a pattern or practice of unequal treatment of a protected class. Plaintiffs must show that discrimination in the workplace manifests as the company's “standard operating procedure.” International Brotherhood of Teamsters v. United States, 431 U.S. 324, 336 (1977); see also Franks v. Bowman Transportation Co., 424 U.S. 747, 772 (1976). Plaintiffs may support their case by presenting evidence of a pattern of discrimination against a protected class of workers, regardless of whether they all have exactly the same jobs, responsibilities, supervisors or work locations.^[10] This approach does not pre-specify how to test for a pattern of pay or other disparities. Whether systemic discrimination exists at all, exists within one particular position, location or function, or spans multiple jobs, facilities or segments of the workforce, is a factual inquiry that turns on case-specific evidence and data.

In a pattern or practice case of compensation discrimination (or any other type of discrimination) relying on statistical evidence, courts permit a wide range of approaches—evaluating each model based on the facts of the case. Proof frequently turns on the results of a statistical analysis of the compensation paid to a protected class, with controls used to ensure comparison of similarly situated employees and accounting for potentially non-discriminatory explanations for statistical disparities. However, there are no hard and fast rules regarding how to group workers, what controls to use, or how to analyze the pay practices at issue. For example, in Segar v. Smith, the court found discrimination based on a regression analysis comparing all African-American special agents to white special agents, even though they held jobs that spanned multiple positions and pay grades. In that case, the evidence demonstrated an overall pattern of racial discrimination in compensation after controlling for qualifications and other factors impacting pay. 738 F.2d 1249 (D.C. Cir. 1984). OFCCP relied on a similar approach in the Harris Bank case involving systemic pay discrimination against workers holding a variety of different positions.^[11]

Courts consistently hold that there is no single correct model or set of factors that must be included in a regression analysis. Bazemore v. Friday, 478 U.S. 385, 400 (1986); McClain v. Lufkin Ind., 519 F.3d 264, 280 (5th Cir. 2008). Proper groupings for purposes of regression analysis are based on a combination of statistical theory and relevant facts, and should be a case-specific determination.^[12] To the extent the Standards mandated specific and very narrowly defined groupings in every case, they are not consistent with Title VII principles and not appropriate as parameters to restrict OFCCP enforcement.

By setting limits on how OFCCP tests for pay differences, and by grounding those limits in job similarity, the Standards make it much harder to detect certain forms of pay discrimination. Where an employer discriminates by channeling workers of a particular race or sex into lower paying jobs, by a glass ceiling preventing advancement, or other promotion or job assignment practices, it may be highly inappropriate to use job similarity as the basis for analysis.^[13] There are also problems with some of the specific factors for determining job similarity, for example where the contractor has a policy linking additional pay to certain qualifications, and reliance on those qualifications may cause adverse impact.^[14] In those situations, the Standards require OFCCP to accept potentially discriminatory decisions or criteria as a neutral justification for pay differences, contrary to longstanding Title VII principles. Defining employees as similar based on their positions is often appropriate, and OFCCP will continue to use that approach.^[15] But it is not the right framework for every single case.^[16] By requiring only a single approach to determining who is “similarly situated,” the Standards made it much more difficult to address the full range of possible unfair pay practices.

Removing the arbitrary restrictions of the Standards will also align OFCCP practice with the EEOC. The EEOC's Compliance Manual rejects the idea that there is one way to prove compensation discrimination, and distinguishes between individual and systemic approaches. The EEOC's Compliance Manual is careful to point out that its approach to disparate treatment analysis is “not intended as an exclusive method” (subsection 10-III.A). And, with respect to using statistics, the Compliance Manual states that “[t]he decision about whether and how to use statistics to aid in investigation should be made on a case by case basis” (subsection 10-III.A.3).^[17]

In addition to the substantive questions about how to group employees, the Standards attempt to dictate the level of aggregation—traditionally a case-specific inquiry. For example, in Velez v. Novartis, the plaintiffs' expert report tested for gender-based pay differences by analyzing all sales employees in all jobs together, including in some versions of the analysis a comparison for job level to differentiate between entry-level and more senior employees. 244 F.R.D. 243, 261-62 (S.D.N.Y. 2004). The defense expert analyzed each job level using a separate regression, and the court ruled that it was up to the fact finder to decide which approach was more persuasive.

Under the Standards, OFCCP generally is to perform a separate regression analysis for each of the defined groups of employees holding similar jobs. While the Standards leave open the option of an aggregate analysis, that approach is not preferred and subject to specific technical limitations. See 71 FR 35131; 35140-41. Courts have consistently held that the decision to aggregate data is a case-by-case inquiry,^[18] and that overly fragmenting a regression analysis makes it harder to detect discrimination when it exists.^[19] Further, certain specific technical requirements for using a pooled regression model under the Standards create additional unnecessary across-the-board hurdles that instead should be case-by-case determinations.^[20]

Proof of discrimination under the Executive Order and Title VII requires evidence sufficient to support a conclusion that discrimination motivated the decision or that an identified employment practice has an adverse impact on a protected class. That evidence can take many forms. What the appropriate comparison groups are depends on the pay practices at issue, the available data, types of workers, and other case-specific factors. It may be important to test for unjustified differences within a set of workers who are similar on the basis of job, but it may be important to consider other approaches. OFCCP will take a more proactive and rigorous approach to analyzing pay differences that does not place unnecessary barriers in the way of effective enforcement or hinder its ability to protect workers from discrimination.

2. Multiple Regression Analysis

Most commenters discussed the use of regression analysis as directed by the Standards. Eight of the seventeen agree with OFCCP that the agency should not formally restrict its analytic method to multiple regression analysis.

The commenters supporting rescission generally stated that multiple regression is often the appropriate tool, but they also agreed that OFCCP should retain the flexibility to consider all possible evidence of discrimination. The Social Science Researchers concluded that “OFCCP * * * should utilize this mode of analysis [regression analysis] in investigating possible compensation where it is feasible and appropriate to do so.” A women's rights group noted that while multiple regression is a “powerful, versatile method of estimation * * * it is not the ideal means for examining every analytical problem, particularly when working with small samples.” Commenters explained that using regression analysis may not be appropriate especially where data or sample size limitations could bias the results or where the underlying technical assumptions necessary to support regression analysis cannot be met. A women's rights group and a civil rights group both noted that OFCCP does not require a regression analysis during the investigatory phase of other types of discrimination cases.

Commenters opposing rescission largely agreed that Title VII does not require regression analysis in all cases; ^[21] however, they challenged the view that the Standards unduly limit OFCCP's choice of methods and expressed concern regarding potential alternatives. For example, an organization of businesses agreed that “a multiple regression analysis may not be the appropriate statistical model to analyze all compensation issues, particularly for small sample sizes.” However, this commenter expressed concern that OFCCP would rescind the Standards without identifying the methods to be used in place of regression analysis. An employers' association makes a similar assessment, stating that “multiple regression analyses may not be the preferred statistical methodology in all cases * * *” but fearing “OFCCP may elect to use less sophisticated statistical analyses in its future compliance evaluations * * *.”

Using a single analytic method to identify compensation discrimination is inconsistent with Title VII's mandate and evidentiary principles. Watson v. Ft. Worth Bank and Trust, 487 U.S. 977, 995 (1988) (Supreme Court's “formulations” for proof of discrimination “have never been framed in terms of any rigid mathematical formula”). Although regression analysis is a common method of proof in systemic cases, courts have considered statistical techniques other than multiple regression as potential evidence of discrimination.^[22] Similarly, published research on discrimination frequently relies on multiple regression, but social scientists use a variety of quantitative and qualitative methods to document differences in hiring, pay or other outcomes on the basis of race and gender.^[23] There is no single method of proving discrimination, and it is critical to consider all relevant evidence in order to draw an appropriate conclusion. Most systemic discrimination cases rely on statistical evidence, but as explained in the prior discussion, there is frequent debate over the choice of models, methods and variables; and courts have permitted a variety of analytic approaches.

Social science principles require choosing a method and a model based on the research question and available data; ^[24] Title VII principles similarly require statistical evidence to be responsive to the issues presented, the underlying facts and the relevant, available data. Multiple regression analysis is frequently the appropriate method; however other statistical or nonstatistical analyses may be better suited, depending on the facts of the case and the available data.

OFCCP has found that the use of multiple regression analysis may be appropriate in some cases and not others. Even in the narrowed context of examining systemic compensation discrimination, its application has limitations. In smaller workplaces, in reviews involving high level or very specialized positions, or in cases where important data are unavailable or unreliable, it may be difficult to identify patterns of discrimination by a single analytic method or type of evidence. OFCCP has not abandoned the use of multiple regression analysis and will continue to use this type of analysis to examine compensation issues where it is feasible and appropriate to do so. Section II.A.4 and Section III discuss more specifically how OFCCP intends to approach the choice of analysis going forward.

3. Anecdotal Evidence

More than half of the commenters addressed the requirement that OFCCP obtain anecdotal evidence to support the issuance of a Notice of Violation (NOV). A majority of these commenters agreed that OFCCP should not specifically require anecdotal evidence to support the issuance of an NOV. The remainder opposed changing the current treatment of anecdotal evidence under the Standards.

Commenters in favor of eliminating this requirement relied on legal and practical considerations. They noted that courts have permitted discrimination cases to go forward without anecdotal evidence. They also stated that anecdotal evidence is much harder to obtain in cases of compensation discrimination because victims are either unaware of the compensation other employees receive or they are expressly prohibited from gaining such information. One women's rights group cited an Institute for Women's Policy Research survey of private and public sector employees in which 50% of respondents and 61% of private sector employees reported that discussing pay was prohibited or discouraged in the workplace.

Commenters in favor of keeping the requirement based their position on either a legal argument or on the view that such a rule places no real burden on the agency. These commenters state that OFCCP is not under a formal restriction, citing to language in the Standards that indicates “[t]here 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.” 71 FR at 35134. They go on to state that it is common in Title VII cases to provide anecdotal evidence to bring “the cold numbers convincingly to life,” as the Supreme Court described in the Teamsters case. 431 U.S. at 339.

OFCCP concludes that the mandate regarding anecdotal evidence operates as a real barrier to enforcement and should be rescinded. Identifying individuals harmed by pay discrimination is particularly difficult. Ledbetter v. Goodyear Tire & Rubber Co., 550 U.S. 618, 645 (2007) (Ginsburg, J., dissenting). Many workers do not know they are underpaid. If OFCCP finds evidence of pay discrimination by federal contractors through its review of data, the agency should not let that discrimination stand simply because the contractor had successfully hidden it from its employees. Federal contractors have special obligations to avoid discrimination, monitor their pay practices and submit to reviews to make certain they are in compliance—regardless of whether any individual applicant or employee actually has knowledge of discrimination.

Further, Title VII does not dictate the use of anecdotal evidence in all systemic cases. As the Supreme Court has explained, statistics may at times be “the only avenue of proof” available “to uncover clandestine and covert discrimination.” Teamsters, 431 U.S. at 339 n.20 (internal citation omitted). In some cases, statistics alone can establish discrimination.^[25]

Although the Standards do allow OFCCP to proceed without anecdotal evidence in certain circumstances, OFCCP finds this exception to the requirement to be too narrow. No anecdotal evidence should be required for any type of case, much less for a compensation case where it may be extremely difficult or impossible to obtain. Regardless, OFCCP will continue to actively seek anecdotal evidence during its investigations. The agency will evaluate all available evidence—statistical and anecdotal—before making a determination regarding contractor compliance.

4. Multiple Investigative and Analytical Methods

The NPR states that OFCCP will continue to adhere to the principles of Title VII in investigating compensation discrimination and will reinstitute flexibility in its use of investigative approaches and tools. Generally, the commenters, whether supporting or opposing rescission of the Standards, acknowledged that multiple investigative and analytical methods for addressing potential compensation discrimination may be used by OFCCP.

A number of commenters expressed support for OFCCP's position in this regard. Specifically, a women's rights group stated that, “[i]t is critical for OFCCP to have a full complement of investigative tools and strategies at its disposal to be used at the various stages of the investigation and litigation process.” A civil rights organization stated, “OFCCP must be permitted to exercise discretion to investigate compensation cases in the same manner that it exercises discretion in other types of cases.” The Social Science Researchers noted that OFCCP should be able to choose an analytic method based on factors such as sample size, data availability, or other circumstances.

Some commenters, opposing rescission of the Standards, raised two concerns with the statement that OFCCP will reinstitute flexibility in its use of investigative and analytical tools as it relates to compensation discrimination. These commenters expressed concern that this would result in inconsistent enforcement and a lack of guidance for contractors. A comment signed by various human resources organizations and a law firm stated that “a contractor has a right to know the standards by which it is being judged.” Further it urges that “a rescission of the Standards without new standards in place would be damaging to both the spirit and enforcement of equal employment opportunity.” Additionally, this commenter challenged OFCCP's statement that it adheres to Title VII principles and asserts that “OFCCP's interpretation of Title VII principles in the proposed rescission is not consistent with the legal standards established in case law * * *.” An employers' association noted agreement with OFCCP's statement that compensation investigations and analytical procedures should be tailored to the facts of the case based upon Title VII principles. However, this commenter also expressed concern that flexibility in OFCCP's use of investigative approaches and tools would result in “inconsistency and confusion.” A comment submitted by a law firm offered that, if “* * * OFCCP believes other methodologies may be appropriate for identifying systemic compensation discrimination under other circumstances, the Standards should be modified appropriately, but not discarded all together.”

These comments involving potential inconsistency and undue flexibility raised one specific past OFCCP practice that involves the so-called “pay grade theory.” This method made a comparison of average pay differences using a particular employer's pay grade, salary band or similar system to draw conclusions about pay discrimination. The method made assumptions that workers in the same pay grade were by definition similarly situated. 71 FR 35136-37. The Notice adopting the Standards explicitly grounded the need for the Standards on the view that this approach was legally untenable. 71 FR 35125-26. Because concerns about the pay grade model animated the original Standards, multiple commenters expressed alarm that this rescission means a return to the prior model.

That is not OFCCP's intent in rescinding the Standards. On the contrary, both approaches suffer from the same flaw. The Standards simply replaced one across-the-board framework with another. Neither permits careful case-specific consideration of the pay practices and workers at issue and the available data and evidence. OFCCP does not view employer pay grades as per se evidence of similarity; rather they are one possible relevant factor among many others. However, OFCCP also has determined that it was a vast overcorrection to address the potential pitfalls of the “pay grade” theory by requiring multiple regression analysis in all cases, or looking to only the narrowest possible comparisons of workers.

OFCCP does not believe that increased flexibility necessarily leads to greater inconsistency, and is committed to ensuring that it does not. Flexibility is needed to allow OFCCP to adapt its approach to the uniqueness of a given case within the framework of Title VII case law. Flexibility also ensures that OFCCP's methodology reflects new legal developments, new analytic practices, and new workplace practices, as well as the relevant nuances of the contractor's workforce and practices. The use of more than one approach to investigate and analyze compensation issues is necessary because of the complexities of these types of investigations. The particular tool, or combination of tools, depends upon the facts of a specific case, and includes consulting with labor economists and other experts, as appropriate.

Further, OFCCP is committed to ensuring consistency in conducting its compliance activities. OFCCP adheres to Title VII principles in developing and applying its compliance policies and procedures. The OFCCP FCCM, directives, and staff training provide necessary guidance to prepare compliance officers to address compensation issues. These tools, used in conformance with the applicable regulations, provide the structure within which compliance officers operate. OFCCP has begun updating materials and implementing a comprehensive training program to ensure that its staff investigate pay discrimination effectively, rigorously, and fairly, consistent with prevailing law and the policy goals animating the Executive Order. In addition, OFCCP will be conducting regular quality audits of its compensation investigations.

Because of the requests from the contractor community for more transparency on OFCCP's procedures for reviewing compensation practices, the agency commits to take specific steps to support future compliance assistance in this area. First, Section III below sets out some specific details regarding how OFCCP intends to apply Title VII principles in the context of its investigations. Second, OFCCP will continue to provide compliance assistance to the contractor community through written materials such as case examples, frequently asked questions, and similar materials. Finally, OFCCP will provide online and in-person opportunities for interactive discussion, such as webinars, online chats, and compliance assistance workshops.

5. Cost to Contractors

A few commenters raised concerns regarding the cost to contractors if the Standards are rescinded, stating that without the Standards in place, contractors will incur unwarranted costs in their attempts to be in compliance. A management law firm noted, “[c]ompensation analysis is not only nuanced and complex, but it also is costly. If contractors are required to navigate the nuance and complexity and absorb these costs, they are at least entitled to transparency in the standards they should use, as well as those OFCCP will use, when doing so.” The commenter recommended that “retaining and modifying the Voluntary [sic] Standards and Guidelines to reflect improvements would be one way to do this.” Simply modifying the existing guidance is not a viable option. OFCCP has not traditionally developed special procedural rules for a single employment practice, instead using directives and other internal guidance to the field. That approach allows the agency sufficient flexibility to respond to changes in case law or the workplace.

For contractors already taking their compliance obligations seriously, the rescission should have little impact on cost. Existing regulations mandate that contractors engage in regular and proactive review of their compensation practices and pay data. Regardless of whether the Standards addressed the full range of potentially discriminatory pay practices, contractors have an independent legal obligation not to discriminate and have affirmatively committed to practice equal employment opportunity as a condition of the privilege of federal contracting. OFCCP is aligning its enforcement procedures with the scope of illegal pay discrimination under Title VII, and ensuring that workers and their families do not bear the cost of unfair discrimination. Further, OFCCP has committed to providing the requested transparency to alleviate potential concerns regarding unnecessary costs as a result of the rescission.

B. Comments Regarding Voluntary Guidelines

1. Contractor Use of the Voluntary Guidelines

In OFCCP's experience, contractors rarely use the Voluntary Guidelines to demonstrate their compliance with the Executive Order. Multiple commenters agreed with OFCCP's assessment, noting that fact warranted rescission of the Voluntary Guidelines. According to an organization of businesses, the Voluntary Guidelines have “limited utility and significant burden” and should therefore be rescinded. A consulting group, while identifying potential benefits of the Standards and Voluntary Guidelines, noted that based on their experience conducting “pro-active compensation reviews for federal contractors,” pay discrimination continues to be a problem. They observed that “the majority of the contractor community did not (unfortunately) go along with the spirit and letter” of the 2006 guidance.

However, other commenters asserted that contractors have in fact used the Voluntary Guidelines—although not for the intended purpose of OFCCP compliance reviews. These comments stated that contractors use the Voluntary Guidelines for internal self-evaluation purposes without taking advantage of the “compliance coordination incentive option.” In the experience of these commenters, contractors perform their compensation analysis under attorney-client privilege and wish to protect it from disclosure. A comment signed by various human resources organizations and a law firm cited two surveys it conducted (with 113 contractors and 33 compensation “experts” responding), which found that 61.3% of the contractors surveyed used the Voluntary Guidelines. This commenter notes that “OFCCP may be confusing a contractor's use of the [Voluntary] Guidelines with contractor's use of the Compliance Coordination Incentive Option (i.e., voluntarily submitting the results of an equity analysis before any triggers have been identified) which our survey indicates is used by fewer than 6% of contractors.” Some commenters expressed concern that without the Voluntary Guidelines, any incentive to self-evaluate would be diminished. A law firm noted that if rescinded “* * * many contractors will be disincentivized from conducting robust self-analysis that permit them to correct problematic disparities [in compensation].”

While it may be true that some contractors privately use the Voluntary Guidelines to predict how OFCCP will evaluate their compliance under the Standards, contractors rarely use them in their interactions with OFCCP. As previously mentioned, OFCCP intends to engage in active compliance assistance regarding compensation analysis. This assistance, as well as the discussion at Section III, below, will provide contractors with notice about how the agency intends to approach investigations of compensation issues and support voluntary compliance activity.

Importantly, even in the absence of the Voluntary Guidelines or some similar explicit instructions for performing pay audits, contractors remain independently obligated to conduct self-evaluations of their compensation practices as required by 41 CFR 60-2.17(b)(3). They are independently obligated to refrain from pay discrimination in violation of the Executive Order and Title VII, so self-monitoring would be prudent even if not required. In addition to the OFCCP, they are subject to potential enforcement actions by the EEOC or Department of Justice or litigation from private plaintiffs. There is no basis to conclude that the Voluntary Guidelines' purely voluntary, rarely utilized and potentially burdensome procedure is the only available mechanism for self-evaluation.

2. Substantive Limitations of the Voluntary Guidelines

In addition to the failure of contractors to use the Voluntary Guidelines, OFCCP in the NPR discussed substantive problems with how the Voluntary Guidelines evaluated potential pay discrimination. A majority of commenters addressed the substantive approach of the Voluntary Guidelines. Over half of those commenters agreed with OFCCP's assessment that the analytical model detailed in the Voluntary Guidelines has not been an effective enforcement strategy, while the remainder defended the approach under the Voluntary Guidelines.

Some commenters noted similar legal and practical deficiencies between the substantive framework of the Voluntary Guidelines and that of the Standards, such as overly narrow groupings and analytic requirements. Several commenters noted the problems with deferring to a contractor analysis of pay, especially where that was not the approach for investigating other types of employment practices.

Other commenters opposed OFCCP's position. A consulting group states that the Voluntary Guidelines are “technically rigorous and sound in almost every regard.” An employers' association stated that it had no objection to rescission of the “coordination” feature of the Voluntary Guidelines but “the remaining portions of the guidelines and the interpretive standards have served as useful blueprints for both OFCCP and federal contractors interested in monitoring compensation patterns for potential systemic discrimination.” Two commenters stated that the Voluntary Guidelines conform to Title VII principles. A comment signed by various human resources organizations and a management law firm, citing to the two surveys which included 113 contractors and 33 compensation experts as participants in the surveys, stated that “[a]lthough contractors and experts might disagree with some of the individual standards of the current Guidelines * * * 84% of contractors [surveyed] believe the Guidelines increase fairness of an audit by standardizing the process.”

Just like the Standards, the Voluntary Guidelines favor a highly limited analysis that may fail to uncover discrimination in pay. The problems with the proposed analytic groupings are the same for the Standards and the Guidelines—as explained in Section II.A., they are overly narrow, inconsistent with Title VII principles and fail to address the variety of potential types of pay discrimination. These limits are magnified by the fact that the Voluntary Guidelines establish specific numerical thresholds to define statistical coverage, group size, and application of regression analysis. The Voluntary Guidelines define those limits across the board, in advance, and without any other information about the pay practices at issue, the types of workers, the number of explanatory factors, or the quantity or reliability of the available data. That one-size-fits-all approach lacks analytic rigor and legal foundation. It is unlikely to be effective at distinguishing between contractors who are in compliance with the Executive Order and those who are not. And it is therefore unlikely to be a useful or appropriate self-evaluation tool.

The Voluntary Guidelines were always optional, but as an officially recommended OFCCP method, these substantive limitations become particularly problematic. Contractors assumed, even if they did not use the Voluntary Guidelines for compliance coordination, that following their dictates would guard against any charges of discrimination in pay. By discouraging any broader examination of pay disparities, the Voluntary Guidelines created a false promise of compliance serving neither the interests of contractors nor of workers.

There is an additional problem specific to the Voluntary Guidelines—the compliance coordination procedure itself. Although rarely used, it is still in conflict with the OFCCP's goal of fully addressing pay discrimination in the contractor workforce. Because compliance coordination requires deference to any analysis that “reasonably meets” the Voluntary Guidelines, and because the Voluntary Guidelines take an overly narrow view of what constitutes discrimination, OFCCP may be prevented from addressing legitimate violations of the Executive Order. There is no reason to have such a compliance coordination mechanism, and especially not one for a specific employment practice. OFCCP does not formally defer to contractor determinations of applicant or promotion pools, steps of hiring procedures, job groups in Affirmative Action plans, or the many other factual issues relevant to evaluating compliance in other areas. Nor should OFCCP defer to contractor decisions about how to test for pay differences.

In the absence of the Voluntary Guidelines, contractors may continue to choose a self-evaluation method appropriate to assess potential pay disparities among their workforce. OFCCP will not be mandating any specific methodology. However, the principles outlined in Section III, below, should be useful to contractors devising a self-audit program. Under section 60-2.17(b)(3), contractors must be assessing specifically “whether there are gender-, race-, or ethnicity-based disparities” in compensation, and under section 60-2.17(d) any self-audit program must be “periodic” and must include specific internal reporting to management of results. OFCCP will assess compliance with these aspects of the regulations by determining whether the scheduled reporting mechanism meets these standards.

3. Cost to Contractors

Several commenters spoke to the issue of cost to the contractors should the Voluntary Guidelines be retained or rescinded. They expressed concern regarding increased costs to the contractors in terms of the “absence” of any guidance. A federal contractor organization, referring to both the Standards and Voluntary Guidelines, noted that “[i]n the absence of such guidance, many employers, particularly smaller and mid-size employers without the `deep pockets' to hire costly third-party experts, will be discouraged from conducting any type of proactive self-analysis.” Taking a different approach to the issue of costs to the contractor, an organization of businesses, supporting rescission of the Voluntary Guidelines, stated that “* * * the [Voluntary] Guidelines ignore the burden associated with developing sophisticated regression models that would satisfy the standards articulated by OFCCP. The cost and complexity of conducting such analyses is too much for many [of our] members to undertake on an annual basis.”

Speculations about potential future costs is not a basis to retain a rarely used, ineffective and potentially burdensome compliance regime. This is particularly true where the current approach may already be costly for some contractors, and where it clearly fails to advance the agency's core policy objective.

OFCCP is taking steps to mitigate any potential cost or burden associated with rescinding the 2006 guidance. In addition to the discussion in Section III below, OFCCP will be providing written materials, such as FAQs, and compliance assistance sessions going forward—clearly describing its investigative procedures and interpretation of key issues. This should make it easier for contractors to assess their own practices. It will also avoid the possibility that the absence of guidance imposes a cost on contractors.

C. General Comments Regarding the Need for Formal Rulemaking

Numerous commenters discussed the OFCCP proposal to communicate its procedures for investigating and analyzing compensation discrimination through the traditional means of using its compliance manual, directives and other staff guidance. A few commenters supported OFCCP's use of the same methodology for establishing policy and procedures as it uses in addressing other discrimination issues, noting that the use of its compliance manual, directives, and other similar guidance have been effective.

Several commenters raised concerns regarding OFCCP's decision not to use formal rulemaking. This was coupled with comments that by not using formal rulemaking, OFCCP is not being transparent in its actions. An employer association noted that if OFCCP rescinds the Standards and Guidelines, “new guidelines should be established through a formal public rulemaking process that mirrors the EEOC's enforcement of Title VII.” A management law firm asserted that the proposed approach “* * * moves from a transparent, consistent approach to compensation analysis by OFCCP to a more covert, possibly ever-changing approach.”

Another management law firm challenged the view that OFCCP has not traditionally addressed investigation standards through formal rulemaking, citing the Uniform Guidelines on Employee Selection Procedures, the Sex Discrimination Guidelines, and the Internet Applicant Rule. The Executive Order implementing regulations establish legal requirements; they do not prescribe or limit the models of proof that the agency may use to demonstrate noncompliance. OFCCP has traditionally established investigation procedures through subregulatory materials such as compliance manuals, directives, and training and will continue to do so.

Because OFCCP adopted the Standards and Voluntary Guidelines by means of the notice and comment process, OFCCP has decided to take subsequent action regarding the specific published guidance in the same manner. However, OFCCP's general practice has been to develop specific investigative procedures for all of its programs through training programs, internal guidance documents, the FCCM, and similar materials. OFCCP has developed and conformed its investigative procedures based on its interpretation of Title VII principles as the law has developed over time. OFCCP will continually refine these procedures to ensure that they are as effective and efficient as possible. In addition, OFCCP plans to provide written materials and compliance assistance as explained above. Going forward, OFCCP will provide as much transparency and public disclosure as possible about its procedures for investigating compensation discrimination. Technical assistance will include tools such as written Frequently Asked Questions, webinars, conference calls, online chats, and presentations, which also provide opportunities for stakeholder dialogue and feedback. The comments received in response to the NPR do not present a compelling argument for OFCCP to unnecessarily restrict its ability to be responsive and timely in this regard.

D. Statutory and Regulatory Reviews

Executive Order 12866 and Executive Order 13563 (Regulatory Planning and Review)—This rule has been designated an “other significant” regulatory action, although not economically significant, under Executive Order 12866. The public was provided a meaningful opportunity to provide input on this document through a 60-day comment period on a Notice of Proposed Rescission issued on January 3, 2011.

Paperwork Reduction Act—The Paperwork Reduction Act (PRA), 44 U.S.C. 35, does not apply to this document because it does not involve any collection of information subject to the approval of the Office of Management and Budget. The information reviewed under the Title VII framework described in this document is collected and reviewed as a result of a desk audit of a contractor's or subcontractor's employment practices. The information collected during the desk audit is covered under OMB Control Number 1250-0003. The compensation analysis described in the Notice occurs after OFCCP compliance officers identify one or more indicators of compensation discrimination during the desk audit that warrant a more in-depth investigation or a compliance evaluation. Pursuant to 5 CFR 1320.4(a)(2), the PRA does not apply to information collections during an “administrative action, investigation, or audit involving an agency against specific individuals or entities.”

Regulatory Flexibility Act—OFCCP determined that, pursuant to the Regulatory Flexibility Act (RFA), 5 U.S.C. 601, et seq., this rescission does not require a regulatory flexibility analysis. Agencies must conduct a regulatory flexibility analysis for any regulatory action that requires a notice of proposed rulemaking. 5 U.S.C. 603(a). The Notice provides subregulatory guidance to contractors and subcontractors regarding OFCCP's application of Title VII principles to compensation discrimination evaluations. Therefore, a regulatory flexibility analysis is not required.

E. Conclusion

After careful consideration of these comments, OFCCP concludes that the Standards and Voluntary Guidelines impede the agency's ability to detect and investigate compensation discrimination, which disserves workers, contractors, and the agency. They require an overly narrow definition of what may constitute systemic compensation discrimination, encourage a less rigorous approach to self-evaluation, and preclude full enforcement of the Executive Order ban on pay discrimination. There should be no unnecessary barriers to enforcing the promise of equal opportunity for workers, and certainly not with respect to ensuring non-discrimination in pay.

OFCCP has concluded that the Standards and Voluntary Guidelines have failed to meet the objectives they were designed to address. They significantly undermine the ability of the agency and contractors to vigorously investigate and identify compensation discrimination consistent with Title VII principles. OFCCP has developed and will continue to develop more effective methods for investigating and addressing compensation discrimination. OFCCP rescinds the Standards and Voluntary Guidelines in their entirety.

Going forward, OFCCP will follow Title VII principles in investigating and analyzing compensation discrimination. The agency proposes to make its treatment of compensation cases consistent with other types of OFCCP discrimination investigations. With the rescission of the Standards and Voluntary Guidelines, OFCCP will focus on the case-by-case assessment of compensation discrimination investigation procedures, and provide clear and consistent guidance to its staff, contractors, and the public regarding its approach.

III. Applying Title VII Principles To Evaluate Whether Contractor Pay Practices Comply With Executive Order 11246

As explained above, OFCCP is rescinding the 2006 guidance documents to ensure its enforcement practices address all forms of pay discrimination that may violate Title VII. In order to assist contractors seeking to comply, and to provide transparency, OFCCP is setting forth its interpretation of certain significant legal and technical issues. This will provide notice of the standards OFCCP intends to rely upon when conducting compliance evaluations, and the standards OFCCP will be instructing its compliance officers to follow.

A. Investigation Procedures

Under Executive Order 11246 and its implementing regulations, contractors may not discriminate in “rates of pay or other forms of compensation;” ^[26] and must review and monitor their compensation systems to “determine whether there are gender-, race-, or ethnicity-based disparities.” ^[27] Contractors must maintain records, including but not limited to “rates of pay or other terms of compensation.” ^[28] During compliance evaluations, OFCCP requests compensation data and analyzes contractors' compensation systems and practices to determine if discrimination exists and, if so, how it should be remedied.

OFCCP's approach to investigating and enforcing non-discrimination in compensation follows Title VII principles. The approach involves factual investigation, and data and legal analyses, which allow OFCCP to identify and remedy all forms of compensation discrimination. OFCCP will tailor the compensation investigation and analytical procedures to the facts of the case as appropriate under Title VII. This case-by-case approach to compensation discrimination includes the use of a range of investigative and analytical tools. Statistical analyses and non-statistical analyses, such as the use of comparators or cohort analysis, will be applied as feasible and appropriate given available data and evidence, and the factual issues being studied. OFCCP will seek anecdotal evidence, but will investigate and remedy instances of compensation discrimination regardless of whether individual workers have reported being underpaid.

This approach is designed to eliminate unnecessary barriers to OFCCP's ability to protect workers from discrimination. It ensures OFCCP fully takes into account any possible explanations or responses from contractors, and that OFCCP conducts an analysis tailored to a contractor's specific compensation systems and practices.

B. Reviewing Contractor Pay Practices

In particular, OFCCP will consider five principles when reviewing contractor pay practices: 1. Determine the most appropriate and effective approach from a range of investigative and analytical tools; 2. Consider all employment practices that may lead to compensation discrimination; 3. Develop appropriate pay analysis groups; 4. Investigate large systemic, smaller unit and individual discrimination; and 5. Review and test factors before including them in analysis. Each of these is explained in more detail below.

1. Determine the Most Effective and Appropriate Approach From a Range of Investigative and Analytical Tools

Investigation of potential compensation discrimination presents complex and nuanced issues. The choice of the best approach for a case depends upon the underlying facts, the available data, and the contractor's compensation system and practices. As such, OFCCP takes a case-by-case approach to analyzing compensation issues. In every case there are three key questions to be addressed: a. Is there a measurable difference in compensation on the basis of sex, race, or ethnicity? ^[29] b. Is the difference in compensation between employees comparable under the contractor's wage or salary system? c. Is there a legitimate (i.e. nondiscriminatory) explanation for the difference? OFCCP will conduct an appropriate factual investigation, data and legal analyses to address each of these questions. An investigation may include analysis of workforce data and contractor compensation policies and practices; interviewing personnel and employees; examining payroll and Human Resource Information Systems (HRIS) data and records; conducting statistical analyses, such as regression analysis; and non-statistical analyses, such as comparative and/or cohort analysis, and consulting with statistical analysts, labor economists and other experts; as well as examining other relevant information.

At the early phase of a scheduled compliance evaluation, OFCCP may use a range of preliminary analysis techniques to determine whether further review is warranted to make a final determination of compliance, and to assist offices in prioritizing investigative resources. As a compliance evaluation moves from the desk audit to an onsite investigation and a final determination regarding compliance, OFCCP will review and refine the approach in light of further information provided by the contractor or developed through investigation. All ultimate determinations of compliance will be based on a rigorous, appropriate and legally sound analysis of the facts and data.

2. Consider All Employment Practices That May Lead to Compensation Discrimination

OFCCP will examine all employment practices that have the potential to lead to compensation disparities that are relevant given the case-specific facts and data. Compensation includes any payments made to, or on behalf of, an employee as remuneration for employment, including but not limited to salary, wages, overtime pay, bonuses, commissions, vacation and holiday pay, allowances, insurance and other benefits, stock options, profit sharing, and contributions to retirement. The compensation a group of employees or an employee receives may be negatively affected by denial of equal access to certain earnings opportunities. OFCCP will examine employee access to opportunities affecting compensation, such as: Higher paying positions or job classifications, work assignments, training, preferred or higher paid shift work, and other such opportunities. OFCCP will also examine policies and practices that unfairly limit a group's opportunity to earn higher pay, such as: “Glass ceiling” issues; and access to overtime hours, pay increases, incentive compensation, and higher commission or desired sales territories. OFCCP will tailor the approach and tools to be used to examine possible unequal access and denial of opportunity issues based on the compensation practices relevant to a particular case. Differences may be observed with regard to base salary; job assignment or placement; opportunities to receive training, promotions, and other opportunities for advancement; earnings opportunities; and differences in access to salary increases or add-ons, such as bonuses.

3. Develop Appropriate Pay Analysis Groups

If the data allow, OFCCP will begin by testing for statistical significance on large groups of employees. The analysis may be based on groups that are larger than individual job titles and job groups. By combining employees into appropriate pay analysis groups, using statistical controls as necessary for title or level, OFCCP will be able to more easily identify potential systemic discrimination needing further investigation and potential remedy. Additionally, if the data allow, OFCCP will analyze pay disparities based on protected class status that cannot be explained by neutral job-related factors, e.g., identifying potential placement or classification issues for further investigation.

A pay analysis group is a group of employees subject to a single statistical framework, model or test. For compensation analysis, a group may be limited to a single job or title, may be performed separately on multiple distinct units or categories of workers, or may be a pooled regression analysis that combines employees from multiple job titles, units, categories and/or job groups that are comparable for purposes of the contractor's pay practices. Where a combination of job titles or jobs at multiple levels is used, it may be appropriate to control for title and level within the group, in order to ensure comparison of similarly situated workers (see below). The size and definition of a group, including questions such as whether to include title or level as a control in the analysis, depends on available data and evidence and the compensation practices at issue. Reasonable differences may exist among workers in a pay analysis group as long as these differences are properly accounted for in the statistical analysis to be conducted. OFCCP will conduct regression analysis on the pay analysis groups to determine whether statistically significant disparities in compensation exist. Statistical testing for practices that impact compensation such as job assignment may require a different model than tests for within job compensation differences.

OFCCP will develop pay analysis groups by considering the following, at a minimum: The particular industry, the types of jobs and compensation at issue, the contractor's actual compensation practices and available data. Compensation practices may differ by role (e.g., executives, managers, supervisors and individual contributors), by level (with higher-level employees tending to receive additional or alternate forms of compensation), by function (such as sales employees who are paid on commission), by unit (department, division, location, etc.) and/or by job classification (exempt or non-exempt, part time or full time, bargaining unit, etc.). This information may be found through a review of the contractor's policy or training documents, description of its compensation system or practices, compensation data, records and coding, job descriptions, and other facts relevant to determining groups, such as the ability of workers to rotate or transfer among different positions within a business unit, a common hiring or selection process, a common performance review practice or other common identifiable employment practice relevant to compensation.

As the results of the initial analysis and facts warrant, OFCCP will refine the analysis, and may conduct subsequent statistical and/or non-statistical tests of smaller units or individuals.

4. Investigate Large Systemic, Smaller Unit and Individual Discrimination

OFCCP will investigate possible large systemic, smaller group or unit, and individual compensation discrimination. Pay analysis groups are to be developed to examine possible systemic issues. Systemic discrimination may be a pattern or practice of discrimination or an identified employment practice with adverse impact that affects multiple employees or groups of employees or applicants. When OFCCP completes analysis of larger pay analysis groups, or in cases where the data are inappropriate or insufficient for regression analysis, the agency will examine the data to further address possible compensation discrimination involving specific job titles, particular units or locations, or other smaller groupings. These additional analyses will be used to confirm, refine or supplement the larger analysis.

After analyzing the data for potential systemic discrimination in larger and smaller groups, OFCCP may conduct comparative analyses of very small groups or individuals to determine if discrimination has occurred, and if there is evidence sufficient to support an inference that pay differences are due to discrimination. The mere fact that there are pay differences between comparators, without any other evidence of pretext or other indicia of possible discrimination, generally is not sufficient to find a violation of E.O. 11246.

For purposes of evaluating compensation differences, the determination of similarly situated employees is case specific. Relevant factors in determining similarity may include tasks performed, skills, effort, level of responsibility, working conditions, job difficulty, minimum qualifications, and other objective factors. In some cases, employees are similarly situated where they are comparable on some of these factors, even if they are not similar on others. For example, when evaluating a job assignment issue, workers are similarly situated when their qualifications are comparable, but they are assigned to jobs at different levels. Employees are similarly situated when they are comparable on factors relevant to the compensation issues presented. Who is similarly situated for purposes of an individual analysis or review of a single specific employment decision may be determined based on different criteria than when conducting a systemic discrimination analysis.

5. Review and Test Factors Before Including Them in Analysis

OFCCP will evaluate, on a case-by-case basis, information from the contractor regarding the factors the contractor considered in making compensation decisions. A factor is an element that the contractor offers to explain differences in employee compensation under its compensation system and practices. Factors may include internal and external elements potentially affecting compensation. A factor may be a qualification or skill that the worker brings to the position such as education, experience, etc. It may also be a job-related element such as position, level or function; tenure in position; performance ratings, etc.

As in any investigation, OFCCP will review and test the factors offered before accepting them as appropriate for inclusion in the analytical model and/or comparative analysis to be conducted. OFCCP will evaluate whether these factors actually explain compensation, whether they are implemented fairly and consistently applied, whether data regarding that factor is accurate, and whether they should be incorporated into the analysis to be conducted. Where a factor that explains pay differences is based on an identified employment practice, such as a specific qualification, performance review instrument, job assignment policy, or a similar policy or practice, OFCCP will evaluate it for potential disparate impact or disparate treatment before determining whether to include it in the analysis.

C. Application to Pending Compliance Evaluations

The procedures and principles described in this document apply to all OFCCP reviews scheduled on or after February 28, 2013.

The 2006 Compensation Standards and Voluntary Guidelines will govern determinations regarding the issuance of an NOV for systemic compensation discrimination in any OFCCP review scheduled, open or otherwise pending on the effective date of this Notice of Rescission. Contractors may elect to waive application of the 2006 Guidelines, and/or to have pending reviews conducted under these procedures, by notifying OFCCP in writing.

Authority: E.O. 11246, 30 FR 12319, 3 CFR, 1964-65 Comp., p. 339, as amended by E.O. 11375, 32 FR 14303, 3 CFR, 1966-70 Comp., p. 684. E.O. 12086, 43 FR 46501, 3 CFR, 1978 Comp., p. 230.

Dated: February 22, 2013.

Patricia A. Shiu,

Director, Office of Federal Contract Compliance Programs.

Footnotes

1.  The term “federal contractor” or “contractor” used in this notice refers to federal contractors, subcontractors, and federally-assisted construction contractors and subcontractors.

2.  41 CFR 60-2.17(b)(3); see also 41 CFR 60-2.17(d) (required internal auditing and reporting system must include compensation).

3.  41 CFR 60-1.12.

4.  See, e.g., McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802 n.13 (1973) (“The facts necessarily will vary in Title VII cases, and the specification of the prima facie proof required from respondent is not necessarily applicable in every respect to differing factual situations”); Furnco Constr. Corp. v. Waters, 438 U.S. 567, 575 (1978) (title VII approaches to proving discrimination “not intended to be an inflexible rule”).

5.  The Standards and Voluntary Guidelines were published on June 16, 2006. See 71 FR 35124 (June 16, 2006) (Standards) and 71 FR 35114 (June 16, 2006) (Voluntary Guidelines).

6.  According to the latest Bureau of Labor Statistics data, women's weekly median earnings are about 81% of men's. In 2010, women on average earned .81 for every dollar earned by a man. Bureau of Labor Statistics, Women at Work (2011). Through the first three quarters of 2012, that figure increased slightly. See also BLS, Current Population Survey, Labor Force Statistics from Current Population Survey, available at http://www.bls.gov/​cps/​earnings.htm#demographics;​; updated 2012 CPS earnings figures by demographics by quarter available at http://www.bls.gov/​news.release/​wkyeng.t01.htm. And looking at annual earnings reveals even larger gaps—approximately 23 cents less on the dollar for women compared with men. U.S. Bureau of the Census, Income, Poverty and Health Insurance Coverage in the United States, Current Population Reports 2011 (Sept. 2012), available at http://www.census.gov/​prod/​2012pubs/​p60-243.pdf. Analyzing the weekly figures can be more precise in certain ways, like accounting for work hours that vary over the course of the year, and less accurate in others, like certain forms of compensation that don't get paid as weekly wages. No matter which number you start with, the differences in pay for women and men really add up. According to one analysis by the Department of Labor's Chief Economist, a typical 25-year-old woman working full time in 2011 would have already earned $5,000 less than a typical 25-year-old man. If that earnings gap is not corrected, by age 65, she will have lost hundreds of thousands of dollars over her working lifetime. White House Council on Women and Girls, The Key to an Economy Built to Last (April 2012), available at http://www.whitehouse.gov/​sites/​default/​files/​email-files/​womens_​report_​final_​for_​print.pdf. For women of color, the gap is even greater, approximately .70 on the dollar for African-American women and approximately .60 for Latinas compared with white men based on BLS data, and .64 for African American women and .56 for Latinas based on Census data.

7.  A March 2011 White House report entitled “Women in America: Indicators of Social and Economic Well-Being,” found that while earnings for women and men typically increase with higher levels of education, the male-female pay gap persists at all levels of education for full time workers (35 or more hours per week), according to 2009 BLS wage data. Potentially non-discriminatory factors can explain some of the gender wage differences. See, e.g., June Ellenoff O'Neill, The Gender Gap in Wages, Circa 2000, American Economic Review, May 2003, at 309. Even so, after controlling for differences in skills and job characteristics, women still earn less than men. Explaining Trends in the Gender Wage Gap, A Report by the Council of Economic Advisers (June 1998). See also, e.g. Ariane Hegewisch, Claudia Williams, Vanessa Harbin, The Gender Wage Gap by Occupation (2012) (women's median earnings less than men in virtually all occupations); Anthony T. LoSasso, et al, The $16,819 Pay Gap For Newly Trained Physicians: The Unexplained Trend Of Men Earning More Than Women, 30 Health Affairs 193 (2011). Ultimately, the research literature still finds an unexplained gap exists even after accounting for potential explanations, and finds that the narrowing of the pay gap for women has slowed since the 1980s. Joyce P. Jacobsen, “The Economics of Gender 44 (2007); Francine D. Blau & Lawrence M. Kahn, The U.S. gender pay gap in the 1990s: slowing convergence, 60 Industrial and Labor Relations Review, 45 (2006). In addition to the gender pay gap, scholars have found race and ethnicity-based pay gaps that put workers of color at a disadvantage. Joseph G. Altonji and Rebecca M. Blank, Race and Gender in the Labor Market, in, Orley Ashenfelter and David Card, eds., Handbook of Labor Economics 3143 (1999).

8.  For example, in Williams v. Galveston Ind. Sch. Dist., No 03-40436, 78 Fed. Appx. 946 (5th Cir. 2003), the court found that differences in duties and supervisory roles explained the difference in pay between the individual plaintiffs and the comparators.

9.  Other circumstantial or direct evidence of discrimination can support an individual claim even in the absence of a formal comparator. See Satz v. ITT Financial Corp., 619 F.2d 738, 745-46 (8th Cir. 1980). As the EEOC Compliance Manual explains, “A claim of compensation discrimination can be brought under [title VII] * * * even if no person outside the protected class holds a `substantially equal,' higher paying job.” http://www.eeoc.gov/​policy/​docs/​compensation.html.

10.  See, e.g., McReynolds v. Sodexho Marriott Svs. Inc., 349 F. Supp.2d 1, 9, 21 (D.D.C. 2004) (companywide statistical model covering jobs in multiple grades and locations sufficient for prima facie case of a pattern or practice); Beckman v. CBS, 192 F.R.D. 608, 618 (D. Minn. 2000) (summary judgment not appropriate where plaintiff alleged a pattern of segregating women in less well-paying jobs and comparisons covered multiple types of jobs); Stender v. Lucky Stores, 803 F.Supp. 259, 336 (N.D. Cal. 1992) (proof of systemic discrimination supported by pattern of lower earnings for women across multiple jobs); Greenspan v. United Auto Club of Mich., 495 F.Supp. 1021, 1029-33 (E.D. Mich. 1980) (same). Applying an analogous principle in ruling on class certification, courts have agreed that proof of systemic discrimination can be supported by evidence of patterns that span jobs or locations. See, e.g., Hnot v. Willis Grp. Holdings, 228 F.R.D. 476, 483-84 (S.D.N.Y. 2005) (disputes over statistical models to be resolved by factfinder at liability); Satchell v. Fed. Express, 2005 WL 2397522, at *7 (under Teamsters, proof of pattern and practice can be based on statistical evidence covering workers in multiple jobs and locations); Warren v. Xerox Corp., 2004 WL 1562884, at *9 (E.D.N.Y. 2004) (race-based disparities in multiple pay grades relevant to merits of discrimination claim). Courts have also approved consent decrees settling pattern or practice or disparate impact claims of pay discrimination covering multiple positions, levels and locations, see e.g., Ingram v. The Coca-Cola Co., 200 F.R.D. 685 (N.D. Ga. 2001); Shores v. Publix Supermarkets, 1997 WL 714787 (M.D. Fla.).

11.  OFCCP v. Harris Bank, 1978-OFC-2, ALJ's Recommended Decision on Remand (Dep't. of Labor Dec. 22, 1986). See also OFCCP v. St. Regis Corp., 78-OFC-1, ALJ's Recommended Decision (Dep't. of Labor Dec. 28, 1984).

12.  For example, in Harris Bank the ALJ noted that analysis of two large groups of jobs that spanned multiple titles and levels—professional and clerical—was appropriate based on the theory of discrimination.

13.  See, e.g., Beckman v. CBS, 192 F.R.D. 608 (D. Minn. 2000); Stender v. Lucky Stores, 803 F.Supp. 259 (N.D. Cal. 1992); OFCCP v. St. Regis Corp., supra.

14.  When the qualifications are not job related or consistent with business necessity, their use would be illegal under title VII, such as offering a higher paying job to workers able to pass a lifting test when the job does not require lifting and women more often fail the test. Cf., Valentino v. U.S Postal Service, 674 F.2d 56, 67-68 (D.C. Cir. 1982) (relying on qualifications to define groups is potentially more appropriate for high level or specialized positions than for general administrative, technical or clerical jobs that share skill levels).

15.  OFCCP v. Astra Zeneca, 2010-OFC-0005, ALJ Consent Decree and Order (Dep't. of Labor June 6, 2011).

16.  Harris Bank, at 25 (“Dr. Killingsworth's decision to group the professional and clerical hires together in one study was clearly correct for this particular issue. The key determination is whether the distinction itself was based upon discriminatory criteria. It is bard [sic] to visualize how the question could have been properly examined without a simultaneous comparison of the employees directly affected. Regressing each subgroup individually would have assumed Harris' initial employment decisions were correct, and included this assumption in the probits.”).

17.  EEOC, Directives Transmittal, Compliance Manual, Section 10: Comp. Discrimination (Dec. 5, 2000), available at http://www.eeoc.gov/​policy/​docs/​compensation.html.

18.  See, e.g., Stagi v. Amtrak, 391 F. App'x. 133, 11 (3d Cir. 2010); McReynolds, 349 F.Supp. 2d at 14.

19.  Courts have recognized the value of aggregate data in a variety of circumstances. Lilly v. Harris Teeter Supermarket, 720 F.2d 326, 336 n.17 (4th Cir. 1983) (aggregate data across years preferred over single year); Eldredge v. Carpenters 46 N. California Cnty Joint Apprenticeship and Training Comm., 833 F.2d 1334, 1339 (9th Cir. 1987) (aggregate data over years provides a “more complete and reliable picture”); Cook v. Boorstin, 763 F.2d 1462, 1468-69 (D.C. Cir. 1985) (evidence of discrimination across multiple jobs is relevant to discrimination in a particular job group); Capaci v. Katz & Besthoff, 711 F.2d 647, 654 (5th Cir. 1983) (improper to fragment data in ways that make statistical tests “less probative”).

20.  For example, the Standards require a control for job similarity in every pooled analysis. This will frequently be appropriate, but as explained above, in cases involving a concern about discrimination in assignment to job or level, it may not be appropriate. The Standards also require testing for interaction terms in every case where OFCCP is considering pooled analysis, specifically mentioning the Chow test as an example. While this is a standard statistical test, applying it in this particular context is highly disputed by experts, is not always technically feasible, and it has not been required by courts. See, e.g., Taylor v. District of Columbia Water & Sewer Auth., 241 F.R.D. 33, 43 (D.D.C. Mar. 13, 2007); Rossini v. Ogilvy & Mather, 615 F.Supp. 1520, 1522-23 (S.D.N.Y. 1985), vacated and remanded on other grounds, 798 F.2d 590 (2d Cir. 1986); Vuyanich v. Republic Nat'l Bank, 505 F. Supp. 224, 299, 314 (N.D. Tex. 1980), vacated on other grounds, 723 F.2d 1195 (5th Cir. 1984).

21.  A management law firm noted “* * * statistical regression analysis may not be required by Title VII (and in fact, no specific methodology is), it is clear from case law that regression analysis is an appropriate method for evaluating pay.” (Emphasis in original text). A comment signed by various human resources organizations and a management law firm, similarly noted that “multiple regression has a long and well-established history in Title VII compensation cases.” Another management law firm went even further, asserting that regression analysis is required under Title VII case law. A consulting group's comment addressed the issue from a different perspective—focusing on the analytic procedures used to prioritize cases for investigation. The consulting group asserted that “a more robust, widespread, and consistent compensation evaluation system should be installed.” This commenter recommended that annual submission of electronic compensation data by federal contractors will contribute to an “improved” system and using a more robust “tipping point test” is needed. The comment concludes that with these changes, “multiple regression should be used as the sole tool for identifying systemic pay disparities.” OFCCP has sought input on whether to ask contractors to submit annual data as the comment suggests and how to analyze that data. 76 FR 33372 (June 8, 2011.)

22.  For example, courts have considered descriptive statistics about the representation of protected groups in certain jobs, or statistical analyses other than multiple regression, in combination with other evidence, in determining proof of systemic discrimination. See, e.g., Beckman, 192 F.R.D. at 611; Greenspan, 495 F.Supp. at 1029-1033. In addition, in St. Regis the ALJ concluded there was a pervasive pattern of wage disparities disfavoring women using statistical techniques other than regression analysis. The wages of numerous small groups of comparable male and female employees were compared. In statistically significantly more groups, the wages of males were higher than the wages of females. The ALJ concluded these differences were attributable to discriminatory job assignments.

23.  When researchers need to delve deeper into potential explanations for differences in outcomes for particular groups, or overcome biases and limitations of linear regression models for particular cases, they have considered alternative techniques. See, e.g., Jaume Garcia, Pedro Hernández and Angel López-Nicolás, How wide is the gap? An investigation of gender wage differences using quantile regression, 26 Empirical Economics 149 (2001)(adjusting estimates of the wage gap to account for increases in the wage scale). Social scientists have also increasingly applied experimental techniques such as paired comparison testing to identify race or sex discrimination in employment and other contexts. See, e.g., Marianne Bertrand and Sendhill Mullainathan, Are Emily and Brendan More Employable Than Lakisha and Jamal? American Economic Review (2004).

24.  See Daniel Rubinfeld, Reference Guide on Multiple Regression, in Reference Manual on Scientific Evidence 179, 186-91 (Federal Judicial Center 2000).

25.  Teamsters, 431 U.S. at 339 (“We have repeatedly approved the use of statistical proof, where it reached proportions comparable to those in this case, to establish a prima facie case of racial discrimination in jury selection cases, see, e.g., Turner v. Fouche, 396 U.S. 346; Hernandez v. Texas, 347 U.S. 475; Norris v. Alabama, 294 U.S. 587. Statistics are equally competent in proving employment discrimination.”) (S.Ct. and L. Ed. citations omitted). Accord, Palmer v. Schultz, 815 F.2d 84, 90-91 (DC Cir. 1987); Rossini, 798 F.2d at 604; OFCCP v. Greenwood Mills Inc., 89-OFC-39, Decision and Order of Remand at 3 (Dep't. of Labor Nov. 20, 1995); OFCCP v. Jacksonville Shipyards, 89-OFC-1, Decision and Order of Remand at 6, (Dep't. of Labor May 9, 1995).

26.  41 CFR 60-1.4

27.  41 CFR 60-2.17(b)(3), (d)

28.  41 CFR 60-1.12

29.  In situations where there is sufficient data and analytic power to use regression analysis, a measurable difference generally means a statistically significant difference, two standard deviations, consistent with title VII. In the situation of disparities in small group and/or individual compensation, a measurable difference and sufficient evidence will be determined in conformance with title VII principles.