Disparate Impact Analysis: Four-Fifths Rule, Tests, and Deadlines

Disparate impact analysis is the method used to test whether a facially neutral employment policy, such as a hiring test, education requirement, or background-check rule, screens out a protected group at a significantly higher rate than others. It relies on outcome data rather than proof of intent, and under federal law an employer can be held liable for a policy that produces unequal results even when no one meant to discriminate.¹ The framework is codified at 42 U.S.C. § 2000e-2(k) and structures the case as a shifting sequence of proof between the person challenging the policy and the employer defending it.²

Why Intent Is Not the Question

Disparate treatment claims allege that an employer deliberately singled someone out because of race, sex, religion, or another protected characteristic. Disparate impact works differently. The focus is entirely on outcomes: if the numbers show a policy is screening out a protected group at a significantly higher rate, the employer has to justify the policy whether or not anyone meant harm.² That is what makes the analysis useful for surfacing structural barriers that no one is consciously maintaining.

Title VII covers private employers with 15 or more employees for at least 20 calendar weeks in the current or prior year, along with federal, state, and local governments regardless of size, and labor unions and employment agencies.³ Employers below that threshold sit outside the federal framework, though some state laws set lower cutoffs.

The Four-Fifths Rule

Federal regulators use a ratio called the four-fifths rule to flag potential adverse impact. Under 29 C.F.R. § 1607.4(D), if the selection rate for any racial, sex, or ethnic group is less than 80 percent of the rate for the group with the highest selection rate, enforcement agencies treat the gap as evidence of adverse impact.⁴

The math is simple. Say 60 percent of Group A passes a pre-employment test and 40 percent of Group B passes the same test. Divide the lower rate by the higher: 40 ÷ 60 = 0.667, or about 67 percent. That falls well below 80, so the test would be flagged. For Group B to clear the bar in this example, its pass rate would need to reach at least 48 percent (60 × 0.80). Run the calculation separately for every protected group in the applicant pool.

Falling below the four-fifths line does not automatically make the policy illegal. The regulation itself acknowledges that small sample sizes can produce unreliable ratios, and that special recruiting efforts can skew the applicant pool in ways that distort the comparison.⁴ Courts often supplement the ratio with standard-deviation tests and other statistical methods when the raw numbers are ambiguous. But the four-fifths rule is the first screen most agencies apply, and crossing it triggers a much closer look.

What Happens After a Policy Is Flagged

Disparate impact litigation follows a three-step sequence under 42 U.S.C. § 2000e-2(k). Each step shifts responsibility between the challenger and the employer.

Step One: Identifying the Practice and the Disparity

The person bringing the claim must show that a specific employment practice causes a disparate impact based on race, color, religion, sex, or national origin. Pinpointing the particular practice is required. If the employer’s decision-making process bundles multiple criteria together in a way that cannot be separated, the whole process can be challenged as a single practice.² This is where the four-fifths rule and other statistical evidence come in.

Step Two: Job Relatedness and Business Necessity

Once a statistical disparity is established, the employer must show that the practice is related to the job and consistent with business necessity. The Supreme Court set this standard in Griggs v. Duke Power Co., striking down a high school diploma requirement and two aptitude tests that bore no meaningful connection to the jobs they screened for. Employment practices must be a reasonable measure of actual job performance, not just a convenient filter.⁵

Practically, the employer needs evidence linking the test or requirement to actual duties. A warehouse requiring applicants to lift 50 pounds has an easier time justifying a strength test than an office imposing the same requirement on data-entry clerks. The link has to be more than theoretical. Employers who cannot produce documentation tying a policy to job tasks lose at this stage.

Step Three: Less Discriminatory Alternatives

Even if the employer proves business necessity, the policy can still be struck down if the challenger identifies an alternative that serves the same purpose with less adverse impact and the employer refuses to adopt it.² Replacing a written aptitude test with a hands-on work sample that better reflects actual duties may produce equivalent predictive value while narrowing the pass-rate gap.

The alternative does not need to be costless or perfect, but it does need to be roughly as effective at identifying qualified candidates. The Uniform Guidelines on Employee Selection Procedures also direct employers to investigate less discriminatory alternatives during any validation study, rather than waiting for a lawsuit to force the issue.⁶

Validating a Selection Procedure

When a selection procedure produces adverse impact, the employer must validate it under one of three strategies recognized by the Uniform Guidelines.⁷

Criterion-related validity uses a statistical study showing that scores on the test correlate with actual job performance measures, such as supervisor ratings or productivity data.
Content validity demonstrates that the test directly samples important tasks or knowledge required by the job. A typing test for a transcription role is the classic example.
Construct validity shows that the test measures a psychological trait (like spatial reasoning) and that the trait is important for success in the role.

Content validity is the most commonly used approach because the test mirrors the job. Criterion-related studies require larger sample sizes and more sophisticated statistical analysis, which puts them out of reach for smaller employers. Construct validity is the hardest to defend and the least frequently relied on. Whatever the method, the validation study should be documented and updated periodically, because job duties evolve and a study that was solid five years ago may no longer reflect current requirements.

Running the Analysis on Your Own Workforce

A disparate impact analysis is only as reliable as the data feeding it. You need demographic information for every individual in the applicant pool or affected workforce, paired with the outcome being measured, whether that is a hiring decision, promotion, test score, or termination.

Federal law requires employers to invite applicants and employees to voluntarily self-identify their race and ethnicity. The disclosure must be voluntary, and refusing to provide it cannot lead to adverse treatment. If someone declines, the employer must determine the information through visual observation or other available means.⁸ All data is reported using seven standardized categories: Hispanic or Latino, White, Black or African American, Asian, Native Hawaiian or Other Pacific Islander, American Indian or Alaska Native, and Two or More Races.

Organize the data so that each record pairs a demographic identifier with a pass-fail or selected-not-selected result. Applicant tracking systems typically export this in a format you can drop into a spreadsheet. Run the four-fifths calculation for every protected group against the highest-performing group, and do it separately for each selection stage if the process has multiple steps. Incomplete or inconsistently coded data is one of the most common reasons an internal analysis produces misleading results. If your HRIS codes ethnicity differently in the hiring module than in the performance module, the numbers will be wrong before the math starts. An employer that fails to maintain adverse impact data may also face an inference of discrimination from enforcement agencies based on that gap alone.⁴

Deadlines If a Charge Is Filed

An employee or applicant who believes a policy has a disparate impact must file a charge with the EEOC within 180 days of the discriminatory act. The window extends to 300 days if a state or local anti-discrimination law also covers the claim.⁹ The clock runs from the date of the challenged decision, not the date the employee realized the statistical pattern. Missing the deadline can permanently bar the claim.

After a charge is filed, the EEOC notifies the employer within 10 days and may offer mediation. If mediation is declined or fails, investigation takes roughly 10 months on average. If the agency finds reasonable cause, it attempts a voluntary settlement through conciliation. If conciliation fails, the EEOC decides whether to file its own suit or issue a right-to-sue notice.¹⁰ Once you receive a right-to-sue notice, you have exactly 90 days to file in federal court. Courts dismiss cases filed even one day late.¹¹

What Liability Looks Like

Back pay covers lost wages and benefits from the date the discrimination occurred, reaching back up to two years before the charge was filed.¹² When reinstatement is not feasible, courts may award front pay for anticipated future losses. Back pay is not subject to statutory damage caps.¹³

Compensatory damages (for emotional distress, medical costs, and job-search expenses) and punitive damages (for especially reckless or malicious conduct) are capped by employer size:¹³

15 to 100 employees: $50,000 combined limit
101 to 200 employees: $100,000 combined limit
201 to 500 employees: $200,000 combined limit
More than 500 employees: $300,000 combined limit

The caps apply per complainant, not per lawsuit, so a class action involving hundreds of affected applicants can generate liability well into the millions even at the lowest tier. Courts can also order injunctive relief, requiring the employer to stop using the practice and sometimes to overhaul its selection process. Attorney’s fees are recoverable by the prevailing party. Back pay for a class of rejected applicants accumulates quickly, and the total bill in a large-scale case almost always dwarfs the statutory caps.

Where the Analysis Applies Beyond Hiring

Title VII is the most familiar context, but the same logic reaches other areas. The Supreme Court confirmed in 2015 that the Fair Housing Act also supports disparate impact claims, so zoning decisions, lending criteria, and insurance underwriting policies can be challenged if they disproportionately harm a protected group, even without proof of discriminatory intent.¹⁴ The analytical approach is the same: identify a neutral policy, demonstrate its unequal effects with data, and evaluate whether the policy serves a legitimate, non-discriminatory purpose.

For employers, that means disparate impact thinking does not stop at the HR department door. Policies on employee housing assistance, relocation benefits, or background-check criteria incorporating credit history or criminal records can face scrutiny under multiple statutes at once. An organization that runs the analysis only on hiring tests and ignores its other policies is looking at one corner of the risk.

1
Civil Rights Division. Laws We Enforce
2
Office of the Law Revision Counsel. 42 US Code 2000e-2 – Unlawful Employment Practices
3
Office of the Law Revision Counsel. 42 USC 2000e – Definitions
4
eCFR. 29 CFR 1607.4 – Information on Impact
5
Justia U.S. Supreme Court Center. Griggs v Duke Power Co
6
eCFR. 29 CFR 1607.3 – Discrimination Defined
7
U.S. Equal Employment Opportunity Commission. Questions and Answers to Clarify and Provide a Common Interpretation of the Uniform Guidelines
8
U.S. Equal Employment Opportunity Commission. Alternative Suggested Employee Questionnaire
9
U.S. Equal Employment Opportunity Commission. Time Limits for Filing a Complaint
10
U.S. Equal Employment Opportunity Commission. What You Can Expect After You File a Charge
11
U.S. Equal Employment Opportunity Commission. Filing a Lawsuit
12
GovInfo. 42 USC 2000e-5 – Enforcement Provisions
13
Office of the Law Revision Counsel. 42 USC 1981a – Damages in Cases of Intentional Discrimination in Employment
14
Congressional Research Service. Disparate Impact Claims Under the Fair Housing Act