Is failing the 80% rule illegal?

No — the four-fifths rule is an evidentiary screen, not a verdict. A failure is "generally regarded as evidence of adverse impact" under the Uniform Guidelines, which shifts the practical burden toward demonstrating that the selection criteria are job-related and consistent with business necessity. Whether a given selection is lawful is a legal judgment that belongs to qualified employment counsel, never to a formula.

Which groups should be tested?

Every demographic dimension the organization has coded — typically sex, race/ethnicity, and age (40+ under the ADEA, where practitioners often analyze impact even though the Uniform Guidelines formally address Title VII categories). The honest practice is to run the screen across all coded fields rather than choosing which comparisons to look at after seeing the data.

What about small groups and small samples?

The 80% rule is unreliable at small numbers — one person's inclusion or exclusion can swing a ratio across the line in either direction. That's why statistical-significance tests (Fisher's exact test for small samples, the two-standard-deviation test for larger ones) are run alongside it: a ratio failure that isn't statistically significant, or a significant disparity the ratio missed, both change the conversation counsel needs to have.

Defensible restructuring

The 80% rule, explained: adverse impact for restructuring teams

OrgTool Learn · 8 min read · Updated July 2026

FOR INFORMATIONAL AND EDUCATIONAL PURPOSES ONLY. NOT LEGAL ADVICE. Statutes and case law change — verify current requirements with qualified employment counsel before acting.

The four-fifths rule is the closest thing U.S. employment analytics has to a standard unit test. It comes from the Uniform Guidelines on Employee Selection Procedures (1978), adopted jointly by the EEOC, the Civil Service Commission, the Department of Labor, and the Department of Justice, and codified at 29 C.F.R. § 1607.4(D). The operative sentence: a selection rate for any race, sex, or ethnic group that is less than four-fifths (80%) of the rate for the group with the highest rate "will generally be regarded by the Federal enforcement agencies as evidence of adverse impact."

The computation

For each group, compute the selection rate: the share of that group's members who received the favorable outcome.
Identify the group with the highest rate.
Divide every other group's rate by that highest rate — the impact ratio.
Any ratio below 0.80 flags presumptive adverse impact for that group.

One subtlety trips up restructuring teams: the rule speaks of selection for a favorable outcome (hiring, promotion). In a reduction in force, the favorable outcome is being retained — so practitioners apply the test to retention rates, or equivalently examine whether any group's termination rate is disproportionately high. Getting this direction right matters; applied naively to "selection for layoff," the arithmetic inverts.

A worked example

Suppose a 400-person division plans a reduction touching 60 roles. Among 240 men, 30 are selected for reduction — a 87.5% retention rate. Among 160 women, 30 are selected — an 81.25% retention rate. The impact ratio is 81.25 / 87.5 = 0.93: above 0.80, no presumptive flag. Now shift the same 60 reductions so that 20 fall on men and 40 on women: retention becomes 91.7% vs 75.0%, and the ratio drops to 0.82 — uncomfortably near the line. A further small shift crosses it. The instructive point: the same business plan, allocated differently across the org, moves from unremarkable to presumptively discriminatory — which is why the screen belongs inside the planning loop, evaluated as the selection is drawn, not run once at the end.

Why the ratio isn't enough

The 80% rule is beloved for its simplicity and criticized for the same reason: it ignores sample size. With small groups, a single individual can flip the result; with very large groups, a ratio can pass while a statistically significant disparity exists. Standard practice therefore pairs it with significance testing — Fisher's exact test (exact, appropriate at RIF-typical sample sizes) and, in the tradition following Hazelwood School District v. United States (1977), a two-standard-deviation test on larger samples. The Uniform Guidelines themselves acknowledge that smaller differences may matter when significant, and larger ones may not when based on trivial numbers. Reading the ratio and the p-value together is the professional standard.

What a failure means — and doesn't

A failed screen does not mean the plan is unlawful, and a passed screen does not immunize it. A failure means the selection, as drawn, carries a statistical pattern that federal enforcement agencies treat as evidence of adverse impact — so the organization needs either a revised selection or a documented, job-related business justification, evaluated by counsel. The screen's real value is timing: discovered during planning, a failure is a design input; discovered in litigation, it's an exhibit. Deterministic, reproducible computation matters for the same reason — months later, the analysis of record must reproduce exactly, or the record itself becomes the dispute.

Key takeaways

The rule: any group's rate below 80% of the highest group's rate = presumptive adverse impact (29 C.F.R. § 1607.4(D)).
In a RIF, test retention rates — the favorable outcome — across every coded demographic field.
Pair the ratio with Fisher's exact test; respect its small-sample fragility.
A failure is a screen, not a verdict; the judgment belongs to counsel.
Run it live during selection design, and keep the computation reproducible.