Primary Submission Category: Applicants in Social Sciences
Uncovering Treatment Effect Heterogeneity in New York City’s Gifted and Talented Program using BART
Authors: Katherine Strickland, Wei Li, Jennifer Hill,
Presenting Author: Katherine Strickland*
This study examines the heterogeneous effects of New York City’s Gifted and Talented (G&T) program on student achievement. Using administrative data from the 2010-2023 school years, we use Bayesian Additive Regression Trees (BART) to estimate treatment effects at the individual level, then aggregate these to examine heterogeneity across ethnic and socioeconomic groups. Our findings reveal small but consistent impacts of G&T participation on Grade 6-8 Mathematics and English Language Arts (ELA) performance across all student groups. We explore treatment heterogeneity by student demographics, program type, entry timing, program duration, cohort, and school district, finding effect sizes ranging from 0.09 to 0.28 standard deviations. These results suggest targeted expansion of G&T programs could potentially reduce achievement gaps while supporting high-achieving students from underrepresented groups.