DOC Learning Objective 2.3: Identify best practices for data collection about people

Motivation Learning objective 2.3: “Identify best practices for data collection about people for their use case and context.”

Data scientists using Fairlearn may need to collect new data to assess fairness-related harms, but may not know best practices for collecting human subjects data, especially for demographic data.

Details to include

• Information about best practices for ethical collection of human subjects data (e.g., from Institutional Review Board guidance)
• Guidance on how to ask participants about demographic categories (see existing issues #792 and #795)
• Privacy implications of demographic data collection (e.g., Andrus et al., 2021)
• Could link to the discussion of construct validity in demographic data collection (see the Construct Validity section of the User Guide)