the 4/5 rule is the floor · two tests apply in litigation · audit before deploy
Adverse impact at the strategic level
PHR-level adverse impact is the four-fifths-rule mechanics. SPHR-level is system design — AI tool audits, pre-deployment bias review, and proactive prevention so the question never becomes litigation.
The selection-rate ratio chart
below the line = adverse impact indicated
highest-rate group
100%
EEOC 4/5 line
80%
⟵ ratio below this triggers indicated adverse impact
protected-class group
68%
below the line — defend with business necessity, or redesign
In litigation, two tests apply
The four-fifths rule is the EEOC\'s rule-of-thumb screen — fast and rough. Statistical significance testing (z-test, chi-square) is what plaintiffs and the DOL actually litigate on. A 4/5ths rule failure may not reach statistical significance with small samples; statistical significance can flag adverse impact even when the 4/5ths ratio looks fine. Senior HR works with statisticians, not just spreadsheets.
disparate treatment
Intentional discrimination
Different rules applied to different groups. Easier for plaintiffs to prove with direct evidence (statements, documents). Different proof structure, different remedies.
disparate impact
Facially neutral · disproportionate effect
A neutral practice (e.g., a height requirement, a credit check) that screens out a protected class disproportionately. No intent required. Defense is job-relatedness + business necessity + validation study.
the SPHR-level posture
Audit before deployment. Not after litigation.
AI hiring tools, selection processes, comp models — all need pre-deployment bias review. NYC Local Law 144 requires it for AEDTs. EU AI Act classifies employment AI as high-risk. The exam reliably treats “defend in court” as the wrong answer when prevention was an option.
Exam Traps
Strategic adverse impact ≠ tactical compliance
PHR-level focus is on the four-fifths rule mechanics. SPHR-level focus is on system design, AI tool audits, and proactive disparate impact prevention.
Disparate impact vs disparate treatment
Disparate impact: facially neutral practice with disproportionate effect. Disparate treatment: intentional. Different proof structures and remedies.
AI tools can create adverse impact
Algorithms trained on biased historical data perpetuate bias. Senior HR audits AI hiring/comp tools regardless of intent.
Statistical significance vs practical significance
A 4/5ths rule failure may not reach statistical significance. Both tests apply in litigation. Senior HR works with statisticians.
Court of public opinion
Numbers may pass the four-fifths rule but fail the court of public opinion. Senior HR considers reputational impact, not just legal threshold.
Audit before deployment
AI tools, selection processes, comp models — audit for adverse impact BEFORE deployment, not after litigation.
For each protected class, what percent are selected at each stage.
Compare lowest-rate group to highest. If ratio < 80%, adverse impact indicated.
Z-test, chi-square. Some courts require statistical significance, not just 4/5ths.
If indicated, defend with job-relatedness and business necessity. Validation studies required.
Senior HR redesigns the system to prevent adverse impact, not just defend in litigation.
Ready to test your SPHR knowledge?
1,500+ practice questions written by certified professionals.
Start Practicing SPHR