Short answer: historical data is the ordinary basis for an LL144 bias audit. Test data may be used in circumstances recognized by the DCWP rules when sufficient historical data is unavailable, but the public Summary of Results must explain the data source and why the chosen basis was used.
6 RCNY §5-300 defines historical data as data collected during employer or employment-agency use of an AEDT, and test data as data used for a bias audit that is not historical data. The rules permit test data in specified circumstances and require the public summary to explain the data basis.
How do historical data and test data differ?
| Data basis | Meaning | Audit concern |
|---|---|---|
| Historical data | Collected during actual AEDT use to assess candidates/employees. | Confirm provenance, population, period and applicability. |
| Test data | Bias-audit data that is not historical data. | Document why it is used and whether it adequately represents the tool/output being audited. |
What must be disclosed about the audit data basis?
The public Summary of Results must state the source and explanation of the data used. Internally, the audit record should preserve enough provenance to reproduce and review the calculation set.
Frequently asked questions
Not automatically. The key questions are whether its use fits the rule and whether the data is suitable for the actual AEDT/output being audited.
The rules contemplate historical data from multiple employers or agencies in specified circumstances. The data basis and relevance still need to be documented.
No. The public rule requires specified summary information and a source/explanation of data; it does not require publishing candidate-level raw records.
What should be documented before choosing test data?
Use this page for general information only. For a specific workflow, review the LL144 scope framework or request an independent bias-audit scope review.