Home » JEDI Corner

Achieving Diversity in Labor Market Needs JEDI, Advocacy Groups

1 December 2022 435 views No Comment
The Justice, Equity, Diversity, and Inclusion (JEDI) Outreach Group Corner is a regular component of Amstat News in which statisticians write about and educate our community about JEDI-related matters. If you have an idea or article for the column, email the JEDI Corner manager at jedicorner@datascijedi.org.

Joseph Gastwirth is a statistics and economics professor at The George Washington University and a fellow of the American Association for the Advancement of Science, American Statistical Association, and Institute of Mathematical Statistics. His book and articles have been used by the US Equal Employment Opportunity Commission and Office of Federal Contract Compliance Programs and cited in several legal opinions concerning discrimination.

The Civil Rights movement, which led to the desegregation of schools and the Civil Rights Acts of 1964 and 1972, helped improve the economic status of Blacks relative to whites from the 1950s to the early 1980s. According to “Black-White Earnings Over the 1970s and 1980s: Gender Differences in Trends” in The Review of Economics and Statistics, however, socioeconomic disparities have persisted due to, in part, legal, governmental, and societal practices that systematically deny resources and opportunities to racial minorities. It is noteworthy that plaintiffs win only 15–20 percent of equal employment cases in contrast with their 40–50 percent rate in other civil cases. This commentary suggests changes in the way courts have interpreted the equal employment laws since the mid-1980s likely contributed to sustained socioeconomic inequality.

Title VII is the part of the Civil Rights Act protecting individuals from employment discrimination on the basis of their race, ethnicity, gender, national origin, or religion. Plaintiffs can base their claim on one of two theories: disparate impact or disparate treatment. The Griggs v. Duke Power (1971) case concerned the appropriateness of a high-school education requirement when previously the company only required 10 years. Using census data from 1960, plaintiffs showed the proportion of Black males who had a high-school education was much less than the corresponding proportion of whites, a consequence of the myriad barriers Black children faced in segregated and under-funded schools. The court concluded the requirement had a disparate impact on Blacks. Then, the defendant needed to prove the requirement was necessary for successful performance of the job, which it could not.

The disparate treatment approach requires the plaintiffs to prove they received less favorable treatment than an individual from the favored group. At the outset, the plaintiff needs to establish a prima facie case by showing they were qualified for the job but were rejected. Plaintiffs can support their claim with statistical evidence showing the hiring or promotion rates of individuals from their legally protected group were significantly less than those from favored groups with similar qualifications. In response, the defendant must articulate a legitimate nondiscriminatory reason for its action.

In Texas Dept. of Community Affairs v. Burdine (1981), the court noted that while the stated reason need not be the true one, it must be reasonably specific, as the plaintiff will need to use the evidence about the employer’s practices obtained during pre-trial discovery to show the offered reason is pretextual (i.e., a “cover” for discrimination). In contrast with disparate impact cases, where the defendant must prove the requirement is job-related, in disparate treatment cases, the employer only needs to produce a justification, as the burden of proof remains with the plaintiff.

The Civil Rights Act (1991) requires plaintiffs to show their protected status was a motivating factor in their adverse treatment. Then, the employer is given the opportunity to show a legitimate reason had greater influence on its decision, and the plaintiff will try to discredit that explanation.

For discrimination cases brought under other statutes, plaintiffs need to demonstrate their legally protected status was a “but for” cause of their negative treatment (i.e., that a similarly qualified member of the favored group would have received the position in question). These different standards of proof are described in the Gross v. FBL Financial Services, Inc. (2009), University of Texas Southwestern Medical Center v. Nassar (2013) and Bostock v. Clayton County, Georgia (2020) decisions.

Women face similar problems in winning compensation cases under the Equal Pay Act (EPA), which prohibits discrimination on the basis of sex. The use of regression analysis in this context is discussed by Mary Gray in her 1993 Statistical Science article, “Can Statistics Tell Us What We Do Not Want to Hear? The Case of Complex Salary Structures,” and Michael Sinclair and Qing Pan’s 2009 Law, Probability, and Risk article, “Using the Peters-Belson Method in Equal Employment Opportunity Personnel Evaluations.” While the employer has the burden of proof in justifying a pay differential, plaintiffs are disadvantaged because even with strong statistical evidence, they need to compare themselves to an actual male employee with similar qualifications, not a hypothetical one.

In Equal Pay Act and disparate treatment cases, courts often reject comparators in different departments or having different supervisors, even when other job-related characteristics are similar. Furthermore, in addition to accepting justifications of pay differentials based on seniority, merit, or productivity, the law allows employers to use any factor other than sex. Some circuits require this factor to be job-related while others do not.

Even when plaintiffs prevail, the compensation they receive does not bring them to the level of the successful majority employees. Suppose a large firm hires 105 male and 105 female recent graduates. All majored in a job-related field and had similar job-related summer internships. At the end of their six-month probationary period, 100 males and 100 females are retained with similar favorable evaluations. Six months later, the employer promotes 40 males and 10 females. The females file suit relying on the 30 percent difference in promotion rates, which is highly significant (p-value <10-5). Because of the very similar background qualifications, the employer cannot produce a legitimate explanation; the plaintiffs win and the court follows the shortfall method adopted by the relevant government agencies (US Equal Employment Opportunity Commission and Office of Federal Contract Compliance Programs). Women were half the eligible pool and should have received 25 (0.5*50) promotions. Therefore, they deserve the monetary equivalent of 25–10, or 15 positions.

Suppose females were awarded the promotions rather than money. There would be 25 females and 40 males at the higher level and females will form 38.5 percent (25/65) of the eligible pool for the next promotion. In a fair system, where women formed 50 percent of the qualified pool, they should form about 50 percent of those eligible for future promotion.

The conundrum is resolved by reviewing the underlying logic of the court’s calculation. The expected number of promotions was based on the assumption that males and females have the same rates (null hypothesis). At trial, this assumption was contradicted by the data; therefore, subsequent calculations relying on it are erroneous. An appropriate calculation assumes the two promotion rates are equal, so 40 women should have been promoted. Thus, females deserve 40–10=30 additional positions; twice what the current system awards them.

Congress is unlikely to remove these impediments faced by individuals experiencing discrimination by allowing plaintiffs in all types of discrimination cases to demonstrate their protected status was a “substantial or motivating factor” in their adverse treatment, requiring explanations to be job-related, and broadening the range of potential comparators.

Individuals committed to advancing justice, equity, diversity, and inclusion in our schools and workplaces should raise awareness of these important issues, advocate for strengthening the educational system, improve access to health and housing, and hold policymakers accountable for their inaction.

Additional Reading
Federal Judicial Center. 2000. Reference manual on scientific evidence (2d Ed.).
(Especially, the chapters on statistics by Kaye and Freedman, surveys by Diamond, and regression by Rubenfeld)

Goldbeck, A. L. (Ed.) 2021. Leadership in statistics and data science. Springer Nature: Switzerland.
(Contains several relevant articles, including the author’s, on which this commentary is based.)

Bayer, P., and K.K. Charles. 2018. Divergent paths: A new perspective on earnings differences between black and white men since 1940. The Quarterly Journal of Economics. 1459–1501.

Clermont, K.M., and S.J. Schwab. 2009. Employment discrimination plaintiffs in federal court: From bad to worse. Harvard Law and Policy Review 3:103–132.

Gastwirth, J.L., W. Miao, and Q. Pan. 2017. Statistical issues in Kerner v. Denver: A class action disparate impact case. Law, Probability, and Risk 16:35–53.
(A reanalysis of both expert reports in a disparate impact case)

Miao, W., and J.L. Gastwirth. 2016. Statistical issues arising in class action cases: A reanalysis of the statistical evidence in Dukes v. Wal-Mart II. Law, Probability, and Risk 15:155–174.
(An alternative analysis of the data indicating that the plaintiffs had a stronger case)

Selmi, M. 2001. Why are employment discrimination cases so hard to win? Louisiana Law Review 61:555–575.

1 Star2 Stars3 Stars4 Stars5 Stars (No Ratings Yet)
Loading...

Comments are closed.