The Trump Administration Takes Aim at Racial-Balancing Rules for Employment Tests

W hile most of Washington was focused on the possibility of a government shutdown and on other political skirmishes this month, the Trump administration took a little-noticed step that could lead to seismic changes in how the Civil Rights Act is interpreted in employment law.

On March 13, Edward R. Martin Jr., the U.S. attorney for the District of Columbia, announced that his office had filed a motion to vacate the 1981 Luevano consent decree. This decree arose from a 1979 class action lawsuit filed against the Office of Personnel and Management (OPM). At the center of the dispute was the use of the Professional and Administrative Career Examination (PACE), a standardized test designed to predict job performance.

PACE was developed to meet the scientific and legal standards of the time. Primarily, the lawsuit focused on PACE’s tendency to create adverse impact across racial and ethnic groups. While 42.1 percent of white PACE examinees obtained a score at or above the required minimum, only 5 percent of black examinees and 12.9 percent of Hispanic examinees did. This adverse impact was the sole reason that PACE became the topic of litigation, which resulted in the later consent decree.

Among other requirements, the consent decree mandated that OPM create a new test that could predict job performance and on which black, Hispanic, and white applicants could perform approximately equally. For a decade, OPM tried to create a cognitive ability test that met both criteria. Six different tests were created, and none of them could predict job performance without disparate impact. Eventually, most federal agencies stopped using tests in hiring or promotion.

If the courts vacate the Luevano consent decree, it could signal a major shift in how the Civil Rights Act applies to employment decisions. Currently, disparate impact is the watchword for the Civil Rights Act: If an employee-selection procedure for hiring or promotion produces disproportionate outcomes for minority groups, then employers are vulnerable to a lawsuit. In the Luevano case, it did not matter that PACE was a scientifically valid test that made better predictions of employee performance than any alternative available. Because members of racial groups performed disparately on it, the test made the government legally vulnerable.

Prioritizing disparate impact over the scientific integrity of employment tests dates back to the 1971 Supreme Court decision Griggs v. Duke Power Company. In that unanimous decision, the Court ruled, in essence, that when a standardized test showed disparate impact, the employer shouldered the burden of proof of demonstrating the need for the test and the absence of a less discriminatory option. Griggs and the subsequent Albermarle Paper Company v. Moody case (1975) severely curtailed the use of standardized tests in employment, and the Griggs standard was codified into the Civil Rights Act in 1991. Using tests of general intelligence in employment decisions fell out of favor because test content had to be job-related, and only the largest employers could afford the expense of creating a scientifically sound standardized test for each job — and even they did so only for their most common job openings. For small- or medium-sized businesses, developing standardized tests was simply financially infeasible, and using off-the-shelf tests made them a potential target for lawsuits. (No wonder the Luevano class action suit targeted the federal government; it was one of the nation’s largest employers and the most prominent one still using a test to fill large numbers of positions.)

Vacating the Luevano consent decree would not nullify Griggs. The consent decree binds only the federal government. However, the filing from Martin’s office contains the seeds to end the Griggs approach to employment tests. The filing takes the logic of Students for Fair Admissions v. Harvard (which banned affirmative action in college admissions) and applies it to employment testing. Moreover, it seeks to apply to government employment the new ban on racial balancing in higher education.

Future cases could build on the government’s legal theory in the Luevano filing. Taken to its logical conclusion, the theory would effectively remove the special status of disparate impact within employment law, just as it has in higher education. If the Supreme Court accepted this application of Students for Fair Admissions to the employment sector, then tools and procedures that produce disparate impact, such as standardized tests, would be much more difficult to challenge in court. Merely identifying that a test — or any other employee-selection tool or policy — results in adverse impact would not be sufficient to shift the burden of proof onto the employer. Plaintiffs would need to prove that the employer had discriminatory intent, which would be much harder than showing simply that different percentages of group members were hired. (Full disclosure: I work for a testing company, RIOT IQ, that could stand to benefit if employers adopt tests in response to a weakened disparate impact doctrine.)

It is difficult to overstate the impact of Griggs. The decision fueled the college-degree arms race because employers lost their ability to quickly and objectively identify the smartest, most competent applicants. Instead, they were forced to rely on indirect measures of intelligence, such as degrees. This is ironic because Warren Burger’s opinion in the Griggs case stated the limited usefulness of such credentials:

History is filled with examples of men and women who rendered highly effective performance without the conventional badges of accomplishment in terms of certificates, diplomas, or degrees. Diplomas and tests are useful servants, but Congress has mandated the common sense proposition that they are not to become masters of reality.

By making employment tests more difficult to administer, Burger and his colleagues unintentionally ushered in an era when “certificates, diplomas, or degrees” would be more important than ever. Griggs also inflamed false accusations of bias against standardized tests because the presence of disparate impact alone disqualified most tests from being used in employment scenarios. By making it practically illegal to use a test for hiring, the Court rendered tests controversial and morally suspect.

Nonetheless, Griggs was correctly decided based on the circumstances of the case. The Duke Power Company was blatantly discriminating against its black employees, and it kept those explicit segregation and discrimination policies until the very last day before it was legally forced to change them. Its new policies maintained de facto segregation and siloed black employees in the lowest-paying jobs by using intelligence tests and other requirements. That needed to stop, and nothing short of a Supreme Court order would do it.

Given the facts available at the time, the Griggs decision was also scientifically correct. In the early 1970s, the scholarly literature on the ability of intelligence tests to predict job performance was chaotic. Conventional wisdom among psychologists was that tests’ predictive power varied from job to job in idiosyncratic ways. Fewer than ten years later, this view would be overturned, and standardized test scores would be shown to be an excellent predictor of job performance across a wide variety of positions. At roughly the same time, testing experts would discover that the surface appearance of questions on cognitive tests was irrelevant, a finding that negated the need (demanded by Griggs) for test content to closely reflect job duties. As recently as this year, independent research showed that a test that meets both criteria of the Luevano consent decree — scientific validity and no adverse impact — is mathematically impossible.

Neither Griggs nor Luevano is based on modern scientific knowledge or the 21st-century employment landscape. More important (from a legal perspective), both rely on a judicial philosophy that has been rejected by the Supreme Court, not just in Students for Fair Admissions but also in other cases dating back to the 1990s. As stated in the filing to dissolve the Luevano consent decree, the Supreme Court is now extremely skeptical of any procedure that attempts to engineer proportionate outcomes for racial groups in education and employment. Such policies often improve outcomes for some groups by discriminating against others.

These developments are still in their infancy. It will take months, perhaps more than a year, before the courts rule on whether the Luevano consent decree can be discontinued. The potential application of the government’s legal theory to private sector employment will take another court case and plaintiffs willing to fund years of litigation with uncertain prospects of success. Still, conservative activists have painted a target on Griggs, and they are willing to play the long game. Overturning Roe v. Wade took 49 years, and Students for Fair Admissions abrogated the affirmative action doctrines that Regents of the University of California v. Bakke had produced 46 years earlier. Is Luevano the beginning of the end for the reign of disparate impact over employment law?