Workplace Disparity & Incident Pattern Auditor
Statistically evaluate adverse employment decisions (disciplinary write-ups, denied promotions, terminations) using the EEOC Four-Fifths 80% Rule and 2-Standard-Deviation Disparity Tests (Hazelwood / Castaneda standard), while cross-referencing departmental incident patterns.
Audit Parameters
Statistical Findings & Disparity Index
Cohort Breakdown & Rate Comparison
| Cohort | Pool (N) | Action (n) | Action Rate | Impact Ratio | Std Dev (Z) | Legal Flag |
|---|
Formal Compliance Executive Summary
Log Incident / Retaliation Report
Chronological Incident Pattern & Cluster Analysis 5 Logged
Legal Framework & Evidentiary Standards
EEOC 80% (Four-Fifths) Rule
Under 29 C.F.R. § 1607.4(D), a selection or sanction rate for any protected group that is less than four-fifths (80%) of the rate for the highest group will generally be regarded by Federal enforcement agencies as evidence of adverse impact.
Two-Standard-Deviation Rule
Established in Hazelwood School District v. United States (433 U.S. 299) and Castaneda v. Partida (430 U.S. 482). Disparity between expected and observed actions exceeding two or three standard deviations (Z ≥ 1.96 or 2.0) creates a prima facie statistical inference of discrimination under Title VII.
Hostile Work Environment & Inaction
Under California FEHA (Cal. Gov. Code § 12940(k)), employers must take all reasonable steps to prevent discrimination and harassment. Repeated incidents combined with failure to remediate or investigate substantiate systemic liability claims.
How this tool performs the calculations
For adverse actions (such as write-ups, suspensions, and terminations), the benchmark is defined by the cohort with the lowest sanction rate. Any cohort with a higher sanction rate has an Adverse Impact Ratio computed as Benchmark Rate / Subject Rate. An Impact Ratio below 0.80 indicates adverse disparate impact. The Standard Deviation test computes Z = (Observed - Expected) / sqrt(N * p * (1 - p)) under the hypergeometric/binomial distribution null hypothesis.