On the Legal Front

Rich Tonowski

Note: The opinions expressed in this article are those of the author and not necessarily those of any government agency. The article should not be construed as legal advice.

EEOC’S Pay Data Collection Plan:
On Again, Off Again, and Maybe a Reboot History (Abbreviated)

The story of the U.S. Equal Employment Opportunity Commission’s (EEOC) collection of private-sector pay data continues. For early background from some of the players, see Silverman (2015); more recent events are covered by Yang and Shiu (2019). A report on data collection feasibility by the National Research Council (2012) of the National Academy of Sciences (NAS) was commissioned by EEOC. That report recommended a pilot study, subsequently conducted by Sage Computing (2015). The instrument for collection was the annual EEO-1 workforce demographics report, now designated as Component 1, submitted annually by large private organizations and federal contractors. Pay data were designated as Component 2 of the report. Data collection was approved by the U.S. Office of Management and Budget (OMB), the agency that oversees, among other things, burden on respondents for federal information gathering. OMB reversed itself under the Trump Administration in 2017. The National Women’s Law Center (2019) sued OMB to get the data collected. Employers with at least 100 employees required to file EEO-1 demographic reports (Component 1) had to submit pay data (Component 2) for 2017 and 2018 by September 30.¹ The court required data collection through April 2021. The U.S. Department of Justice (DOJ) filed an appeal to overturn the order. The EEOC commissioners held a meeting on November 20, 2019 for presentations on the future of the EEO-1. As expected, there was consensus² that Component 1 demographics should be maintained and sharp disagreement on pay data.

Sticking Points

Employer burden.

EEOC now agrees with employer-side arguments that the reporting burden was underestimated by the agency. The employer representatives (American Bankers Association et al., 2019) give a view of the problems and the changes in estimates over time from EEOC. Not every employer has the data-processing resources to get its data into reporting shape, specifically to link up pay data with employee demographic data. With the various demographic categories and 10 pay categories, that’s 3,660 crossed cells to fill. Larger companies file separate reports for locations with 50 employees plus an overall report.

Mitchell (2019) indicated anecdotal evidence of employers having trouble with the current Component 2 reporting mechanism.

Frye (2019) found insufficient detail in how EEOC now calculated Component 2 burden. A reasonable expectation would be that the burden should diminish as employers developed their reporting services or contracted with data analysis providers. This was also mentioned in a letter to the agency from 29 U.S. Senators (Murray et al., 2019; the senatorial letter) urging pay data collection, which noted that employers had already been compelled to provide 2017 and 2018 data; the burden is what is needed to maintain the reporting system not build it from scratch. The point was also raised by Stevenson (2019). In contrast, Clements (2019) argued that employee movement (promotions, transfers, reorganizations) meant that future reports would not be produced by simply pushing a button.

The senatorial letter also noted that pay data were collected from state and local government agencies with the EEO-4 report for years. EEOC indicated that the data were not useful (Mitchell, 2019). This leads to the next sticking point, purpose of the data collection.

 Purpose.

Why EEOC should collect pay data seems to involve two main rationales. First, the data would help in EEOC and the U.S. Office of Federal Contract Compliance Programs (OFCCP) enforcement efforts. The current position of both agencies is that it is not worth the resources to process and use the data. Stevenson (2019) argued that an enforcement consideration should be that the only way to combat unintentional disparities is with intentional evaluation of selection and pay data, such as provided by EEO-1.

Second, data collection would shed light on the gender (and race) pay gap. But this may be merely tangential to the missions of the enforcement agencies. How much of the pay gaps is due to unlawful practices varies with the study. In general, there is a small but persistent gap that cannot be explained by supposedly legitimate factors. A letter signed by 56 members of the House of Representatives (Frankel et al., 2019, p. 1) put the gender pay gap as women earning $0.82 for every $1.00 earned by men, on average. It also indicated that estimates put the unexplained portion “as much as 38%.” That’s less than $0.07  on average for the high-end estimate of potential discriminatory effect. Of course, this matters to anyone shortchanged because of discrimination, but it points to a limitation on expecting antidiscrimination enforcement efforts to solve all pay-gap problems.

The senatorial letter (Murray, et al., 2019)  has a heading (p. 1) that states that discrimination drives the gender wage gap. No other possible causes are mentioned on that page. A qualification appears on p. 2, that the gap “is driven in significant part” by discrimination.

Still, an argument for pay data collection is that, even if crude, the data would tell us more than we know now. One of the disparity factors that may be outside of individual employers’ control (assuming no discriminatory steering into certain jobs) is occupational segregation. Getting a handle on this to differentiate it from discriminatory practices would be useful, and Component 2 can help (Stender, 2019). Tomaskovic-Devey (2019) wrote that employers should welcome data that indicate that not all firms are equally problematic. Component 2 would provide baselines so that firms could judge if they had disparities.

Eastman (2019) indicated that the pay data collection project had problems from the start. EEOC consulted with the National Research Council (2012) of the National Academy of Sciences (NAS). The first thing that NAS indicated was lack of a comprehensive plan for using the data. Eastman sees this as still absent. The second matter was a pilot study. A study was conducted and simulated data were analyzed, but this was absent a comprehensive plan, actual employer data, actual data collection from employers, or analysis of such data. Other recommendations from NAS concerning the nature of the data, analytic capability, and data protection were not followed.³

There seems to be consensus that Component 2 data alone are not detailed enough to be evidence in litigation. The employer concern is that it may be detailed enough to target an employer for a lawsuit. If you are a pay equity advocate, you might think that’s a good idea. If you’re on the employer side, you have another idea: swatting down baseless nuisance suits. Employers can always explain what they’re really doing with pay. But apart from those with something to hide, employers do not want to spend time dealing with bad claims and bad publicity tied to bad data, or to disclose more details on their pay systems to competitors and potential litigants in defending against these claims.

The analysis of EEO-1 pay categories does not look like the detailed analysis that would be done in a pay audit. Mitchell (2019) discussed six differences that could lead to conflicting results.

 Data security versus transparency.

Pay equity advocates have noted that without greater access to how people are paid, it is difficult for employees or enforcement agencies to know if anyone is being shortchanged. By executive order in the Obama administration that is still in place, federal contractors cannot prohibit their employees from discussing pay. But this includes only a subset of employers and does not mandate disclosures from management. Transparency is mentioned in the senatorial letter (Murray et al., 2019) as a purpose for Component 2. EEOC has maintained that employer-level EEO-1 data would not be disclosed. But there has been concern that granular data provided to researchers and other agencies would be released to others. Specifically, OFCCP routinely gets EEO-1 data and that data might be subject to public disclosure under the Freedom of Information Act. But in November, OFCCP (2019) made its position clear: "OFCCP will not request, accept [italics added] or use Component 2 data, as it does not expect to find significant utility in the data given limited resources and its aggregated nature, but it will continue to receive EEO‑1 Component 1 data."

This nondisclosure has been contrasted with United Kingdom’s approach: aggregated company-wide gender pay gaps and public disclosure. It has been termed by some as name and shame. Smith and Mulvaney (2019) reported that litigation had not spiked with disclosure. A similar program in Denmark reportedly led to shrinking the gender pay gap by 7% (which happens to be analogous to the $0.07 mentioned above). Stevenson (2019) proposed that an alternative to the current Component 2 would be less detail to reduce the burden on employers, but public reporting.

 Motivating employers to analyze further.

Yang and Shiu (2019) argued that data collection is needed to institutionalize pay audits beyond current requirements on federal contractors enforced by OFCCP. Getting employers to do more than satisfy Component 2 reporting was discussed in the last “Legal Front.” Whether pay data collection has motivating force currently, especially when the data collection procedure is not aligned with the conventional approach to pay audits, is debatable. Critics of Component 2 note that false positives indicating discrimination might compel employers to undertake more detailed studies to fend off spurious claims; presumably that is not the best motivation to be proactive regarding pay disparities.

It's Not Over Until It’s Over

The regulatory agenda of the EEOC (2019) envisions further discussion of “meaningful” pay data collection in September 2020. Some employers would like to see any rulemaking follow the Administrative Procedures Act, which could include early lawsuits against a proposal. EEO-1 authorizations have followed the Paperwork Reduction Act. Data collection authority is written into Title VII.

Meanwhile . . .

One blockbuster ($400M) pay equity case brought by OFCCP against Oracle took a new turn. The company has been litigating this through the agency’s hearings system since 2017. Oracle America Inc. v. U.S. Department of Labor et al. (2019) has now gone to federal court to complain that the system is unconstitutional. The complaint alleges that OFCCP had been running its own procedure with no authorization since the 1970s, when it should have referred cases to EEOC or DOJ. “Without authority from any act of Congress—indeed, in contravention of congressional legislation—a group of unelected, unaccountable, and unconfirmed administrative officials have cut from whole cloth this adjudicative agency-enforcement scheme” (Oracle, 2019).

Complaints against the system are not new, with employers alleging that legal interpretations and procedural safeguards do not match those in federal courts. The suit comes at a time when OFCCP has been addressing procedural and transparency grievances from employers. Being more employer-friendly apparently has not diminished enforcement activity; it has been a record-breaking time for monetary recovery from contractors ($81M, October 2016–September 2019).

Implications for I-Os

We have EEOC’s former leadership in conflict with current leadership, each with a cohort of supporters to question the accuracy of whatever the other side has done. The agency seems open to discussion to get it right in the fall of 2020. Not that there will be unanimity on “right.” Equal pay advocates likely would have full, open disclosure of pay practices. Employers, including the good actors, likely would want to forego any hassle with pay reports. The issue of whether there should be any pay data collection and (if so) what form it might take goes beyond only our profession, but I-Os have the expertise for designing performance evaluation, compensation, and reward systems, and statistical smarts for conducting pay audits and related analyses. We, as individuals and as an organization, should at the least be providing comments on proposals when EEOC gears up.

You Don’t Have a Disability Until You Have a Disability

The Americans with Disabilities Act (ADA) protects workers who are not disabled but who are regarded by employers as such. This raises a question: Does ADA cover people who are not disabled nor regarded as disabled now, but who are regarded as likely to become disabled in the future?

EEOC is saying yes, but the courts are saying no. EEOC was on the losing side recently in two cases in separate appellate circuits. In EEOC v. South Tampa Massage Envy (2019), a woman was fired because she went to Africa against her employer’s wishes. The employer feared she would contract Ebola (she didn’t) and bring it back to the workplace. EEOC sued, the case was dismissed, and the appeal went to the Eleventh Circuit. The court ruled that her employer did not regard her as disabled when she was fired. Therefore, ADA does not apply. Because the employee was in an at-will situation,⁴ there was no further recourse.

In the Seventh Circuit, EEOC filed an amicus brief, urging the court to read use of the present tense in the ADA as including the future. The case, Shell v. Burlington Northern Santa Fe (BNSF) Railway (2019), involved obesity. BNSF raised safety concerns possibly arising from obesity but asserted that it did not consider the employee to be disabled currently. The court was persuaded by the employer’s argument.

Commentators have noted that these rulings are in line with similar conclusions in other circuits, but they may involve how fact patterns are viewed in specific cases. It may make a difference whether a risk of becoming suddenly incapacitated involves driving a truck on a freeway or driving a mouse at a computer. It may also matter if the concern for disability is that the person would become disabled because of job conditions that interact with that person’s physical condition. The U.S. Supreme Court held in Chevron v. Echazabal (2002) that the employer had a business necessity justification in not having an employee sensitive to certain toxins working where the toxins would be encountered. EEOC regulation favoring Chevron was upheld.

None of these cases has involved physical ability testing, although parallels can be drawn. For example, an employer uses a test to screen out those who can do repetitive heavy lifting now but who are likely to become injured if they do the job over time. Besides the issue of whether this is testing to uncover a disability, a future court case could involve the probability of injury, its severity, and the length of time in which the injury is expected to occur.

#You, Too

Eradication of sexual (and racial) harassment in the workplace is a compelling matter for our time. Menaker v. Hofstra University (2019) provides a caveat that should be noted by I-Os and other consultants involved with policy making or training in this area. A male university sports coach denied a full scholarship to a female student athlete then on partial scholarship. The student claimed that she had been promised the full scholarship by the coach’s predecessor but no record of this was found. Allegedly, the student retaliated by claiming sexual harassment under Title IX, which applies to educational institutions. The coach met with university officials, provided records of correspondence with the student, and suggested other students to interview. The coach was told by an official that one instance of harassment was known to be unsubstantiated, and he would be kept informed of the progress of the investigation. The witnesses mentioned by the coach were not interviewed. The coach was summoned to a meeting without being told the purpose and was fired for “unprofessional conduct.” No specific charge was cited as warranting termination; it was the “totality” of the accusations.

The coach sued under Title VII, arguing that his sex was the factor underlying his allegedly unjust termination. District court did not agree and dismissed the case for want of an actionable claim. As an at-will employee, the coach had no legal standing without a discrimination complaint. He appealed to the Second Circuit.

The appellate court found that the coach had established at least a minimal inference of sex discrimination. The university previously had been under criticism for not acting on harassment complaints by women, had not followed its own procedures in his case for handling harassment complaints, and may have allowed itself to be manipulated by the student to show that it was now taking sexual harassment against women seriously. The case was ordered back to district court for trial.

The standards for surviving dismissal of a case are not as demanding as the standards for winning it. But the case is a reminder that procedural justice matters. Title VII also applies to the accused.

Blind Acceptance

Many of us have heard of Goldin and Rouse’s (2000) study, how blind auditions for musicians seeking positions in orchestras substantially increased the hiring of women. Sommers (2019), citing a statistical critique provided by Gelman (2019), indicated that the study proved nothing about blind selection. As Gelman noted, contraindication was in plain sight, as stated by the authors themselves: large sample overall but small numbers for key findings, not using conventional levels of statistical significance (but claiming “economic” significance), and a finding in the wrong direction. But what grabbed public attention was the line near the end of the article: “Using the audition data, we find that the screen increases—by 50 percent—the probability that a woman will be advanced from certain preliminary rounds and increases by severalfold the likelihood that a woman will be selected in the final round.” As the recent critiques indicated, the importance of the article had been expanded in its apparently uninformed retelling.

None of this should detract from using methods for reduction of possible bias in selection procedures, including blind evaluations. But claims for a method should not go beyond the data.

Thinking Outside the Banned Box

 EEOC v. Dolgencorp (2019) regarding use of criminal history for employment ended with a settlement valued at $6 million. The retailer had been making conditional offers of employment pending examination of criminal history and then applied a table of exclusions. Exclusions disproportionately fell on some demographic groups. Should the company continue use of criminal history for employment decisions, it is required to develop a revised procedure with a criminologist named in the settlement. Unlawful discrimination with the current procedure was neither admitted nor proven.

EEOC scored no major court victories in criminal history cases but got some favorable settlements.

We may be entering a new phase with employment for ex-offenders. Local ban-the-box laws have become widespread. Re-integrating ex-offenders into the workforce has become a concern across the political spectrum. Where do we go from here?

As Hyman (2019) pointed out, issues are not settled. Banning the box on the employment application may not mean much if many applicants are subsequently found unsuitable; this was the issue with Dollar General’s procedure. By not asking about criminal history, some employers may be inclined toward “statistical discrimination;” certain demographic groups are more likely to have people with criminal convictions, so applicants from those groups are to be avoided. There are legitimate employer concerns that need to be addressed: what constitutes an appropriate evaluation of criminal history, applicants needing additional training, and the threat of a negligent-hiring suit. Some of this is addressed by tax credits and bonding; some states have provisions for establishing due diligence to avoid negligent hiring. For I-O practitioners, the issue is assessment of both qualifications and suitability—preferably, a generally accepted professional practice regarding employer risk in this area.

Personal Note

 I am not listing an organizational affiliation because I am leaving EEOC. It’s been a good 18 years. But I’m ready to move on, and other people have the situation well in hand. My interest in I-O psychology, employment law, and equal employment opportunity is unabated. Also unabated is my admiration for my EEOC colleagues, especially those with whom I served in Research and Analytic Services (RAS). Attorneys have an ethical obligation to provide their clients with zealous advocacy. In RAS we advocated for the truth that our professional expertise led us to discern and present. Through all the years and leadership changes, this is what the agency expected of us. We did our best.

Notes

1 That date has been extended several times. The court’s decision will keep current data filing open into 2020.

2 But see Dwyer (2019) on possible inconsistencies in Component 1 estimates. He argues for an expanded Component 2 that would make Component 1 redundant.

3 However, EEOC has undertaken general upgrading of its analytic and IT capabilities in recent years.

4 Employment at will is legal doctrine that, absent an employment contract (or a collective bargaining agreement) stating otherwise, either the employer or the employee can end the employment relationship at any time, for any reason that is not unlawful.

References

American Bankers Association, Associated Builders and Contractors, Associated General Contractors of America, HR Policy Association, National Federation of Independent Business, National Retail Foundation, Society for Human Resource Management, and U.S. Chamber of Commerce. (2019, November 12). Letter to Bernadette B. Wilson, Executive Officer, EEOC. Retrieved from https://www.regulations.gov/document?D=EEOC-2019-0003-7275
Chevron U.S.A. Inc. v. Echazabal, 536 U.S. 73 (2002).

Clements, L.A. (2019, November 20). Written testimony. EEOC convenes public hearing on the proposed revision of the Employer Information Report (EEO-1). Retrieved from https://www.eeoc.gov/eeoc/meetings/11-20-19/index.cfm

Dwyer, M. (2019, September 11). Comment on the Equal Employment Opportunity Commission “Agency Information Collection Activities: Existing collection.” Retrieved from https://www.regulations.gov/document?D=EEOC-2019-0003-11508

Eastman, M. J. (2019, November 20). Written testimony. EEOC convenes public hearing on the proposed revision of the Employer Information Report (EEO-1). Retrieved from https://www.eeoc.gov/eeoc/meetings/11-20-19/index.cfm.

Equal Employment Opportunity Commission. (2019, November 20). Amendments to the regulations at 29 CFR Part 1602 to provide for a pay survey. Retrieved from https://www.reginfo.gov/public/do/eAgendaViewRule?pubId=201910&RIN=3046-AB15.

Equal Employment Opportunity Commission v. Dolgencorp (Dollar General), No. 13 C 4307 (N.D. Ill November 18, 2019) (settlement).

Equal Employment Opportunity Commission v. South Tampa Massage Envy, LLC., Nos. 18-11121, 18-12277 (11^th Cir. September 12, 2019).

Frankel, L. et al. (2019, October 2). Letter to Janet Dhillon, Chair, EEOC. Retrieved from https://www.regulations.gov/document?D=EEOC-2019-0003-11508

Frye, J. C. (2019, November 20). Written testimony. EEOC convenes public hearing on the proposed revision of the Employer Information Report (EEO-1). Retrieved from https://www.eeoc.gov/eeoc/meetings/11-20-19/index.cfm

Gelman, A. (2019, May 11). Did blind orchestra auditions really benefit women? Statistical Modeling, Causal Inference, and Social Science. Retrieved from https://statmodeling.stat.columbia.edu/2019/05/11/did-blind-orchestra-auditions-really-benefit-women/

Goldin, C. & Rouse, C. (2000). Orchestrating impartiality: The impact of “blind” auditions on female musicians. American Economic Review, 90(4), 715-741.

Hyman, J. (2019, November 19). A pox on ban the box. Workforce. Retrieved from https://www.workforce.com/2019/11/19/a-pox-on-ban-the-box

Menaker v. Hofstra University, No. 18-3089 (2^nd Cir. August 15, 2019) (vacate and remand).

Mitchell, J. W. (2019, November 20). Written testimony. EEOC convenes public hearing on the proposed revision of the Employer Information Report (EEO-1). Retrieved from https://www.eeoc.gov/eeoc/meetings/11-20-19/index.cfm

Murray, P., Cantwell, M., Baldwin, T., Feinstein, D., Blumenthal, R., Sanders, B., Warren, E., Booker, C., Klobuchar, A., Wood Hassan, M., Gillibrand, K., Menendez, R., Duckworth, T., Durbin, R., Hirono, M., Jones, D., Rosen, J., Kaine, T., Shaheen, J., …Merkley, J. (2019, November 12). RE: Comment on notice of information collection-request for new control number for a currently approved collection: Employer Information Report (EE0-1) Component l; Revision of Existing Approval for EE0-1 Component 2 (Docket Number: EEOC-2019-0003). Retrieved from https://www.regulations.gov/document?D=EEOC-2019-0003-11504

National Research Council. (2012). Collecting compensation data from employers. Washington, DC: National Academies Press. Retrieved from https://www.nap.edu/catalog/13496/collecting-compensation-data-from-employers

National Women’s Law Ctr. v. OMB, 358 F. Supp. 3d 66 (D.D.C. 2019), appeal docketed, No. 19-5130 (D.C. Cir. May 8, 2019).

Office of Federal Contract Compliance Programs. (2019, November 25). Intention not to request accept or use employer information report EEO-1 Component 2 data. Retrieved from https://www.federalregister.gov/documents/2019/11/25/2019-25458/intention-not-to-request-accept-or-use-employer-information-report-eeo-1-component-2-data

Oracle America Inc. v. U.S. Department of Labor et al., No. 1:19-cv-03574 (D.D.C. November 27, 2019) (complaint filed).

Sage Computing. (2015). Final report to conduct a pilot study for how compensation earning data could be collected from employers on EEOC’s survey collection systems (EEO-1, EEO-4, and EEO-5 survey reports) and develop burden cost estimates for both EEOC and respondents for each of EEOC surveys (EEO-1, EEO-4, and EEO-5). Retrieved from https://www.eeoc.gov › employers › eeo1survey › pay-pilot-study

Shell v. Burlington Northern Santa Fe Railway Company, No. 19-1030 (7th Cir. October 29, 2019).

Silverman, L. (2015, March 18). EEOC quietly moves forward with plans to collect compensation data. HR Daily Advisor. Retrieved from https://hrdailyadvisor.blr.com/2015/03/18/eeoc-quietly-moves-forward-with-plans-to-collect-compensation-data/

Smith, P., & Mulvaney, E. (2019, September 16). U.S. pay data effort could fall short without public scrutiny. Bloomberg Law News. Retrieved from https://www.bloomberglaw.com/document/X62RVQOC000000?bna_news_filter=daily-labor-report&jcsearch=BNA%25200000016cf892d9c0a96cfeb66a750002#jcite

Sommers, C. H. (2019, October 20). Blind spots in the “Blind Auditions” study. Wall Street Journal. Retrieved from https://www.wsj.com/articles/blind-spots-in-the-blind-audition-study-11571599303

Stender, J. (2019, November 20). Written testimony. EEOC convenes public hearing on the proposed revision of the Employer Information Report (EEO-1). Retrieved from https://www.eeoc.gov/eeoc/meetings/11-20-19/index.cfm

Stevenson, B. (2019, November 20). Written testimony. EEOC convenes public hearing on the proposed revision of the Employer Information Report (EEO-1). Retrieved from https://www.eeoc.gov/eeoc/meetings/11-20-19/index.cfm

Tomaskovic-Devey, D. (2019, November 19) The EEOC should continue to collect pay data. Law360. Retrieved from https://www.law360.com/articles/1221086/the-eeoc-should-continue-to-collect-pay-data

Yang, J.R. & Shiu, P.A. (2019). Brief for Jenny R. Yang and Patricia A. Shiu as Amici Curiae Supporting Appellees. National Women’s Law Ctr. v. OMB, No. 19-5130 (D.C. Cir. October 25, 2019).