Summary
Explain why auditing one protected attribute at a time misses harm concentrated at intersections (e.g. older women, young minority men) — and tie it to the Profilers intersectional-gap output.\n\n## Why\nThe Profiler now reports intersectional gaps but theres no explainer for the concept. Also the natural home for the "intersectional bias notebook" roadmap item.\n\n## Tasks\n- [ ] explainers/intersectional-bias.md\n- [ ] Crosstab/subgroup-rate detection code; link to the Open Dataset Profiler\n- [ ] Numbered limitations (small-subgroup instability), further reading; add to site + README + CONTRIBUTING
Summary
Explain why auditing one protected attribute at a time misses harm concentrated at intersections (e.g. older women, young minority men) — and tie it to the Profilers intersectional-gap output.\n\n## Why\nThe Profiler now reports intersectional gaps but theres no explainer for the concept. Also the natural home for the "intersectional bias notebook" roadmap item.\n\n## Tasks\n- [ ]
explainers/intersectional-bias.md\n- [ ] Crosstab/subgroup-rate detection code; link to the Open Dataset Profiler\n- [ ] Numbered limitations (small-subgroup instability), further reading; add to site + README + CONTRIBUTING