Persona Bias Audit

# Persona Bias Audit Reference

Purpose: Detect and remediate bias in persona sets. Cover representation matrix, intersectionality coverage, stereotyping detection, the Inclusive Persona Checklist, name/photo bias, ableism, ageism, and locale bias. Personas with embedded bias propagate downstream into Echo walkthroughs, Spark feature priorities, and Voice copy.

## Scope Boundary

- **cast `bias-audit`**: Persona representation + ethical audit (this document).
- **cast `archetype` (elsewhere)**: Brand / JTBD archetype tagging.
- **Cloak (elsewhere)**: Privacy / consent / PII handling.
- **Oath (elsewhere)**: Regulated context (e.g., insurance non-discrimination).
- **Plea (elsewhere)**: Synthetic user voice (separate concern, but shares ethics surface).

## Why Bias Audits

Personas are reasoning shortcuts — they shape *every* downstream decision. Biased personas produce:

- Excluded users (features that don't fit)
- Stereotyped UX ("women want pink", "elders need huge fonts")
- Discriminatory pricing or eligibility
- Hostile design (assuming all users speak English, have stable WiFi, hold a credit card)
- Reinforcement of marketing tropes

## The Five Bias Lenses

| Lens | Question | Tool |
|------|----------|------|
| **Representation** | "Who's missing?" | Representation matrix |
| **Intersectionality** | "Who exists at the cross of multiple identities?" | Crenshaw-style intersection grid |
| **Stereotype** | "Are attributes stuck together by tradition not data?" | Attribute-correlation audit |
| **Power** | "Who designed this — and whose lens is encoded?" | Designer reflexivity statement |
| **Harm** | "What harm does this persona enable / prevent?" | Inclusive Persona Checklist |

## Representation Matrix