Adverse Impact
Adverse impact is a legal concept describing the disproportionate exclusion of a protected group by a facially neutral selection practice. It is the main framework under which AI screening tools are evaluated for discrimination risk under Title VII and the EEOC Uniform Guidelines.
Editorial review notice
This page has not received legal review and is for informational purposes only. Adverse impact analysis involves complex legal and statistical judgments. Consult qualified employment counsel and an industrial-organisational psychologist or statistician before conducting or relying on adverse impact analyses.
What is adverse impact?
Adverse impact (also called disparate impact) occurs when a selection practice that appears neutral on its face produces a substantially different — and lower — rate of selection for members of a protected class relative to others. Under US law, an employer need not intend to discriminate; the statistical outcome is sufficient to establish a prima facie case.
Protected classes under Title VII include race, color, religion, sex, and national origin. Additional classes are covered by the Age Discrimination in Employment Act (ADEA), the Americans with Disabilities Act (ADA), and the Genetic Information Nondiscrimination Act (GINA).
Legal framework
The adverse impact doctrine originates in Griggs v. Duke Power Co., 401 U.S. 424 (1971). The EEOC's Uniform Guidelines on Employee Selection Procedures (1978) established the four-fifths rule as a primary indicator and provided the framework still used for analysis today.
If adverse impact is established, the employer must demonstrate that the selection procedure is job-related and consistent with business necessity. Even if that burden is met, the plaintiff may show that a less discriminatory alternative exists.
Measuring adverse impact
Two primary methods are used:
- Four-fifths (80%) rule: If a group's selection rate is less than 80% of the highest group's selection rate, adverse impact is indicated. This is a rule of thumb, not a legal bright line.
- Statistical significance testing: Tests such as the chi-square test or Fisher's exact test assess whether observed differences could be due to chance. Both standard deviation analysis and practical significance are relevant.
Analysis should be conducted at each selection step — application screening, resume review, interview, and final selection — not only at the end of the process.
AI tools and adverse impact
AI screening tools can create or amplify adverse impact through:
- Training data that reflects historical underrepresentation of certain groups
- Features that correlate with protected characteristics (e.g., zip code, school name, employment gaps)
- Video or voice analysis that may be less accurate for certain demographic groups
The EEOC has stated that employers may be responsible for adverse impact caused by third-party AI tools under the same analytical framework applied to any other selection procedure. Employers should request adverse impact analysis data from vendors and monitor selection rates after deployment.