Beyond Prediction Performance: How Modeling Decisions Shape Fairness Outcomes in Statistical...
Talkat Bridging Prediction and Intervention Problems in Social Systems (Simons Institute, 2026) · 2026-01-12..01-18·
summary
German administrative labor market records (2% sample) predict long‑term unemployment across about 700,000 episodes, with regression and machine‑learning models reaching ROC‑AUC 0.70–0.77 yet disagreeing on flagged individuals (pairwise Jaccard ≈ 0.45). Modeling choices—feature construction, thresholds, architecture, training on caseworker decisions or via a reinforcement model—shift false‑positive rates between Germans and non‑Germans and thus who receives scarce interventions. Proposes secure, annotated environments (Coleridge Initiative style) and using linked historical data to train LLMs that capture temporal context.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:33Modeling decisions impact
- 5:59Long-term unemployment case
- 9:16Multiverse of modeling
- 13:08Human vs automated decisions
- 16:56Secure data environments
- 21:58Future directions
- 23:33Q&A
related talks
39mBeyond Bias: Crafting Fairness in Pythonic Machine Learning Systems
17mExplaining models: an empirical study of how explanations impact fairness judgment
34mInterpretability and Algorithmic Fairness
30mBuilding fairer models for finance
55mEthical Machine Learning
19mAdversarial Fairness in the Labor Market
28mMeasuring Model Fairness
31mMeasuring Model Fairness