Disparate Impact Realism
Amy L. Wax
University of Pennsylvania Law School
William & Mary Law Review, Vol. 53, p. 621, 2011
U of Penn, Inst for Law & Econ Research Paper No. 11-12
U of Penn Law School, Public Law Research Paper No. 11-11
In Ricci v. DeStefano, 129 S. Ct. 2658 (2009), the Supreme Court recently reaffirmed the doctrine, first articulated by the Court in Griggs v. Duke Power Company, 401 U.S. 424 (1971), that employers can be held liable under Title VII of the 1964 Civil Rights Act for neutral personnel practices with a disparate impact on minority workers. The Griggs Court further held that employers can escape liability by showing that their staffing practices are job related or consistent with business necessity.
In the interim since Griggs, social scientists have generated evidence undermining two key assumptions behind that decision and its progeny. First, the Court in Griggs noted the absence of evidence that the selection criteria in that case (a high school diploma and an aptitude test) were related to subsequent performance of the service jobs at issue, and expressed doubt about the existence of such a link. But research in industrial and organization psychology (IOP) has repeatedly documented that tests and criteria such as those at issue in Griggs (which are heavily “g”-loaded and thus dependent on cognitive ability) remain the best predictors of performance for jobs at all levels of complexity. Second, Griggs and its progeny rest on the implicit assumption, reflected in the so-called 4/5 rule, that fair and valid hiring criteria will result in a workforce that roughly reflects the representation of each group in the background population. Work in psychometrics and labor economics shows that this assumption is unjustified. Because blacks lag significantly behind whites on measures of cognitive ability, most valid job selection criteria will have a substantial adverse impact on this group. The combination of well-documented racial differences in cognitive ability and the consistent link between ability and job performance generates a pattern that experts term “the validity-diversity tradeoff”: job selection devices that best predict future job performance generate the smallest number of minority hires in a broad range of positions. Indeed, the evidence indicates that most valid screening devices will have a significant adverse impact on blacks and will also violate the 4/5 rule under the law of disparate impact.
Because legitimately meritocratic (that is, job-related) job selection practices will routinely trigger prima facie violations of the disparate impact rule, employers who adopt such practices run the risk of being required to justify them – a costly and difficult task that encourages undesirable, self-protective behaviors and may result in unwarranted liability. To alleviate this burden, the article proposes to adopt a new regime of “disparate impact realism” that abandons the 4/5 rule in favor of sliding scale ratios pegged to measured disparities in group performance and the selectivity of particular positions. Alternatively, the disparate impact rule should be repealed altogether. The data indicate that pronounced differences in the background distribution of skill and human capital, not arbitrary hurdles imposed by employers, explain racial imbalances in the workforce. Moreover, blacks lag behind whites in actual on-the-job performance, which indicates that employers are not unfairly excluding minorities but rather bending over backwards to include them. Disparate impact litigation, which does nothing to correct existing disparities and distracts from the task of addressing them, represents a cumbersome, misplaced effort that could better be directed at the root causes of workforce racial imbalance.
Number of Pages in PDF File: 93
Keywords: Employment Discrimination, Title VII of the Civil Rights Act, Anti-Discrimination Law, Disparate Impact, Screening Criteria, Aptitude, Business Necessity, Selectivity in Hiring
JEL Classification: J71, K31, K39, M51Accepted Paper Series
Date posted: March 28, 2011 ; Last revised: January 19, 2012
© 2014 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollo3 in 0.437 seconds