Primary Submission Category: Applicants in Social Sciences
Do Test Scores Help Teachers Give Better Track Advice to Students? A Principal Stratification Analysis
Authors: Fabrizia Mealli, Javier Viviens, Andrea Ichino,
Presenting Author: Fabrizia Mealli*
We study whether access to standardized test scores improves the quality of teachers’ secondary school track recommendations. We frame teachers’ advice as a decision problem and evaluate efficiency and fairness by building on literature on algorithm-assisted human decisions and using a principal stratification framework. We target Average Principal Causal Effects (APCEs), which measure how access to test scores affects student graduation outcomes within latent strata defined by students’ potential graduation under each teacher’s decision. APCEs quantify the gains from improved targeting and the losses arising from misallocation. Approaches in the literature identify APCEs under strong assumptions (e.g., unconfoundedness, exclusion restriction). We extend this literature by identifying the APCEs without these assumptions. We also provide a comprehensive welfare evaluation that allows for differential weighting of short-term and long-term gains and losses. Finally, we study principal fairness, assessing whether access to test scores improves or worsens equity with respect to certain protected attributes. We find that providing test results to teachers induces fairer recommendations for immigrant and low-SES students; it also increases the share of students successfully placed in more demanding tracks, but misplaces some of the weaker students. However, only implausibly high weights on the short-term losses of weaker students would justify prohibiting test-score-based upgrades.