The Role of Standardized Tests in Admissions: Predictions vs. Decisions
Talkat Bridging Prediction and Intervention Problems in Social Systems (Simons Institute, 2026) · 2026-01-12..01-18
summary
Analysis uses a held-out predictive model on 16,000 applications to first-year grades; adding GREs raises predictive fit from R² from .19 to .31 but simulated admissions (top 20% admitted by predicted GPA) increases mean admitted GPA only about +0.05 grade points (3.54 → 3.59). Using parsed transcripts, letters and other application features further shrinks the GRE marginal gain to ~.03. Test scores mainly reweight applicants by undergraduate institution; downstream decisions change little because most admits are the same or near the admission threshold.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:13Standardized tests debate
- 4:41Schools reversing policies
- 6:15Predictions vs decisions
- 7:23Replicating predictive results
- 10:22Predictive power vs decisions
- 17:16Weak baselines in research
- 21:23Predictive boosts don’t help
- 25:24Fairness and test scores
- 28:47Full application data baselines
- 30:41Summary and policy
- 32:32Should tests be used?