Automated Employment Decision Tool (AEDT) — NYC Local Law 144
Identifying information required under NYC Admin. Code § 20-870 and DCWP Final Rules §§ 5-300 – 5-303
Scope must cover at least one year of NYC employment decisions preceding audit date — DCWP Final Rules § 5-301(a)
Required: impact ratios for all protected categories with ≥ 5 members — DCWP Final Rules § 5-301(a)(3)–(4)
| Category | Applicants (n) | Selected (n) | Selection Rate | Most-Favored Group | Impact Ratio | 4/5ths Rule | p-value (χ²) | Result |
|---|---|---|---|---|---|---|---|---|
| Male | 634 | 81 | 12.78% | — | 1.00 | — | — | Baseline |
| Female | 599 | 74 | 12.35% | Male | 0.97 | PASS (≥ 0.80) | 0.78 | Pass |
| Non-binary / Other | 14 | 1 | 7.14% | Suppressed — n < 30. Documented separately. | Suppressed | |||
| Female/Male impact ratio 0.97 — no adverse impact detected. Chi-square (Yates' correction), α = 0.05. Non-binary category below minimum reporting threshold; employer must document and monitor. DCWP § 5-301(a)(5) | ||||||||
| Category | Applicants (n) | Selected (n) | Selection Rate | Most-Favored Group | Impact Ratio | 4/5ths Rule | p-value | Result |
|---|---|---|---|---|---|---|---|---|
| White (non-Hispanic) | 389 | 52 | 13.37% | — | 1.00 | — | — | Baseline |
| Black or African American | 241 | 31 | 12.86% | White | 0.96 | PASS (≥ 0.80) | 0.84 | Pass |
| Hispanic or Latino | 218 | 28 | 12.84% | White | 0.96 | PASS (≥ 0.80) | 0.81 | Pass |
| Asian | 194 | 27 | 13.92% | White | 1.04 | PASS (≥ 0.80) | 0.92 | Pass |
| Native Hawaiian / Pac. Islander | 42 | 5 | 11.90% | White | 0.89 | PASS (≥ 0.80) | 0.61 | Pass |
| Am. Indian / Alaska Native | 31 | 3 | 9.68% | White | 0.72 | BELOW 0.80 | 0.09 | Monitor |
| Two or More Races | 18 | 2 | 11.11% | Suppressed — n < 30. Fisher's Exact applied. p = 0.41. Documented. | Suppressed | |||
| Unknown / Declined | 14 | — | Excluded from analysis — missing data. Count reported for transparency. | Excluded | ||||
| American Indian/Alaska Native shows impact ratio of 0.72. Below 4/5ths threshold but not statistically significant at α = 0.05 (p = 0.09, n = 31). Employer must investigate and document under DCWP guidance. Monitored quarterly. Fisher's Exact applied to n < 30 cells. Chi-square with Yates' correction applied to n ≥ 30. Bonferroni correction applied. | ||||||||
| Sex × Race Combination | n | Selected | Selection Rate | Impact Ratio | 4/5ths | Result |
|---|---|---|---|---|---|---|
| Male × White | 207 | 28 | 13.53% | 1.00 | — | Baseline |
| Female × White | 182 | 24 | 13.19% | 0.97 | PASS | Pass |
| Male × Black | 122 | 16 | 13.11% | 0.97 | PASS | Pass |
| Female × Black | 119 | 15 | 12.61% | 0.93 | PASS | Pass |
| Male × Hispanic | 108 | 14 | 12.96% | 0.96 | PASS | Pass |
| Female × Hispanic | 110 | 14 | 12.73% | 0.94 | PASS | Pass |
| Male × Asian | 102 | 15 | 14.71% | 1.09 | PASS | Pass |
| Female × Asian | 92 | 12 | 13.04% | 0.96 | PASS | Pass |
| All other combinations | Suppressed — fewer than 30 records per combination. Fisher's Exact results on file. No significant adverse impact detected in any suppressed combination. | |||||
Full methodology required for auditor certification and DCWP examination — DCWP Rule § 5-301(a)(1)
All requirements under NYC Admin. Code §§ 20-869 – 20-871 and DCWP Final Rules §§ 5-300 – 5-303
Independence definition: no financial interest in employer and not employed by employer — NYC Admin. Code § 20-869(b)
SHA-256 integrity chain and MARPP provenance record for evidence immutability
Limitations of this attestation as required by professional standards and DCWP guidance
This attestation is issued pursuant to NYC Local Law 144 of 2021 (NYC Admin. Code §§ 20-869 – 20-871) and the NYC Department of Consumer and Worker Protection Final Rules (§§ 5-300 – 5-303, effective July 5, 2023). It documents the results of an independent bias audit of the identified AEDT. It does not constitute a guarantee of compliance with any other law, including but not limited to Title VII of the Civil Rights Act of 1964, the New York State Human Rights Law, or the NYC Human Rights Law.
HAIEC's bias analysis engine is deterministic. No AI model was used to determine compliance status or generate impact ratios. All calculations are rules-based and reproducible. Same input data produces identical results.