To help choose among job applicants or candidates for promotions, many employers use tests or other selection devices such as educational requirements aimed at predicting who is likely to do well in the relevant position. One risk of doing so is that such tests or devices, although neutral on their face, may screen out a disproportionate number of women, minorities or members of other protected groups. Upon learning that a test would have such an effect, an employer wary of facing a discrimination suit may be tempted not to use the test in deciding who to hire or promote.
On June 29, the U.S. Supreme Court held that such an approach can easily backfire. In seeking to avoid unconscious discrimination against some employees, an employer may intentionally discriminate against others. In Ricci v. DeStefano, the court held that scrapping a test because a statistically significant proportion of a particular race or sex did poorly on the test amounts to intentional discrimination against those who did well on it, unless the employer can show a “strong basis in evidence” for fearing that using the test would lead to liability.
Employers who learn that a test or other selection device has a discriminatory impact on some employees now find themselves between a rock and a hard place. Using the test may lead to one kind of discrimination claim, while discarding the test may lead to another. That makes it more important than ever that employers think through the possible issues before using any test or selection device in making employment decisions.
Two kinds of discrimination
Title VII of the Civil Rights Act of 1964 prohibits two types of employment discrimination based on race, color, national origin, sex and religion. First, the law prohibits intentionally treating one employee better than another because of such protected characteristics. This is called “disparate treatment” discrimination. Second, the law prohibits unintentional discrimination that results when an employer uses a test, selection device or practice that is neutral on its face but has a disproportionate adverse impact on members of a protected group. This is “disparate impact” discrimination. In the Ricci case, the court addressed what happens when an employer may face both kinds of claims from different groups of employees.
The court’s decision
The city of New Haven, Connecticut, used a written test to help decide which firefighters would be eligible for certain promotions. In reviewing the test results, the city discovered that the test had a statistically significant adverse effect on African-Americans. Not only was the passing rate for black firefighters only about half of what it was for whites, but none of the employees with top scores—the only ones eligible for promotion under city rules—was black. Understandably, the city worried that using the test would lead black employees to file, and probably win, a suit alleging that the test had a discriminatory racial impact, so the city decided not to use the test in deciding who could be promoted.
In what likely appeared to the city as a case of “damned if you do, damned if you don’t,” it was then sued by 18 firefighters (17 white and one Hispanic) who had done well on the test, alleging that the city had discriminated against them based on race by refusing to use the test and thus denying them a chance at promotions.
By a 5-4 vote, the Supreme Court ruled in favor of those non-black employees, holding that they had been subject to intentional race discrimination by the city. The court acknowledged that the city’s test had a statistically significant adverse effect on blacks, but said that alone did not justify throwing out the results and thereby discriminating against white and Hispanic employees based on their race, given that the city had taken reasonable steps to make sure that the test was job-related.
To justify discarding a test, the court held, an employer must be able to show a “strong basis in evidence” that it will lose a disparate impact suit brought by employees who would be screened out if the test were used. Statistics alone are not enough. The employer must also present strong evidence that it will not be able to show that its test was “job related and consistent with business necessity,” or fend off a claim that it refused to use other methods that would have had a less discriminatory impact. In effect, the court said that there is only one way that an employer can avoid liability for disparate treatment discrimination if it walks away from a test it has already given, based on concern about possible liability for disparate impact discrimination. In such a case, the employer would have to show that the test was unrelated to the job for which it was used or did not serve an important business need, or that the employer had ignored ways of selecting employees that would have met the employer’s needs with a less discriminatory effect.
What the case means for employers
If an employer gives a test or uses a selection device and then discovers that it has a disproportionate adverse effect on members of a protected group, it is caught on the horns of a dilemma. If the employer uses the test, it may be sued by members of that group for “disparate impact” discrimination. But if it decides not to use the test, it may be sued by those who did well on the test for “disparate treatment” discrimination. After Ricci, an employer facing the latter type of claim can prevail only by showing that its own test was largely irrelevant to the job for which it was being used, that the test was not closely aligned with the employer’s business needs, or that the employer rejected less discriminatory ways of meeting those needs. But having chosen and used a test, presumably based on some evidence that the test would accurately predict success in a job, few employers will be able to meet that burden.
For that reason, it is more important than ever for an employer to make a thorough evaluation of a test or other method of selecting among employees before giving the test or using that method.
An employer must weigh whether the test or method may have a disproportionate adverse effect on certain groups, whether other alternative tests or methods are available that may have a less discriminatory impact, and—most important—how accurate the test or method will be in correctly selecting those employees who are best able to perform the relevant job. By taking those steps in advance, employers will be best prepared to defend any resulting discrimination claim.
In particular, before administering a test or other method of selecting among employees, an employer should:
- Consider whether the test or method is the best way to measure the skills, knowledge and qualifications needed to succeed in the job.
- Do everything reasonably possible to make sure that the test, its passing rate and its cutoff scores will not have a disparate impact based on any protected characteristic. For instance, the employer may involve persons of diverse backgrounds in developing the questions.
- Measure how employees who are currently succeeding in the relevant job do on a proposed test. If they do not do well, it will be hard to show that the test is valid.
- Before using any test recommended by a vendor, ask the vendor these questions:
- Has the test been vetted for disparate impact based on sex, race, national origin and other protected characteristics? If so, when and how? How confident are you that the test will not have significant disparate impact if we use it?
- Do you offer alternative tests that are equally valid but have less disparate impact?
- Has the test been professionally validated in compliance with the Uniform Guidelines on Employee Selection Procedure? If so, how did the jobs for which the test was validated compare with the jobs for which we plan to use it, in terms of skills, qualifications, required knowledge and responsibilities? What studies have been done to measure the correlation between successful performance on this test and successful performance on equivalent jobs?
- Do other employers use the test by itself or in combination with other factors? If the latter, what weight are test results given compared to the other factors?
- Has the test been challenged as discriminatory by any government agency or employees? If so, what happened? How will you help us if that happens? Will your testing experts testify at depositions or at trial? At what cost?
- If you give a test or use a selection device and discover statistically significant disparate impact:
- Do not adjust the test scores or change cutoff scores to deal with the problem, because that is squarely prohibited by Title VII.
- Do not throw out the results, or refuse to use the test or device, unless you are prepared to show that you will lose a Title VII discrimination suit brought by those disfavored by the test. Under Ricci, that is illegal discrimination.
- Recognize that the holding in this case applies not just to employment tests, but to all policies or practices used to select among employees, such as factors used in deciding who to terminate in a reduction in force.
- By taking these steps, employers who use tests or other selection devices can maximize their likelihood of navigating between both types of discrimination claims.
This article originally appeared in Workforce Management, September 2009.