Hiring fairness, answered

What does a bias audit's scope depend on, and how long does it take?

Who is this for?

The short answer

Scope depends mainly on how many tools are audited, what data exists for each, and how cleanly that data records sex and race/ethnicity. Timeline depends most on how quickly usable historical data arrives; the calculations are quick once the data is right. An independence check comes first, before any work.

What sets the scope

The number of tools. Each automated employment decision tool is audited on its own data. BlissPoint's Independent Bias Audit is scoped per tool, with additional tools audited in the same engagement on data from the same period.

The data. The rules (6 RCNY 5-302) require historical data from the tool's use, which may come from one or more employers using it. An employer may rely on an audit that used other employers' data only if it provided its own historical data to the auditor or has never used the tool. Test data is allowed only when there isn't enough historical data, and the summary must explain why and how the test data was made.

The categories. The audit calculates selection or scoring rates and impact ratios for sex, race/ethnicity and the intersectional categories (6 RCNY 5-301). It reports how many people fall in an unknown category, and an independent auditor may exclude a category under 2% of the data from the impact ratio calculations, with a written justification.

What sets the timeline

The steps are: an independence check, a written agreement, the data (a CSV template is provided), the recalculation and review, and the signed report with a summary of results ready to publish. The longest and least predictable step is usually gathering a complete export of the data with the demographic fields filled in.

Plan backward from the date you need to use the tool: the audit must be no more than a year old when the tool is used, and notices go out at least 10 business days before use.

Questions

Can the audit use our vendor's data?
The rules let an audit use historical data from one or more employers that use the tool, but you can rely on one built on other employers' data only if you provided your own data to the auditor or have never used the tool.
What if we don't have enough data?
The rules allow test data when there isn't enough historical data for a statistically significant audit. The summary then has to explain why historical data wasn't used and how the test data was generated and obtained.
Who publishes the results?
The employer publishes the summary of results and the audit date on the employment section of its website. The auditor prepares the summary.

Sources

This page is general information, not legal advice. Laws change and depend on facts this page can't know; talk to an employment lawyer about your situation.

Get started online

Prefer to talk first? Book a free 15-minute call

From an I-O psychology consultant. Covers NYC Local Law 144, the EU AI Act and the four-fifths rule.