Data and Code for: Causal Effects of AI-Assisted Teaching Tools on Academic Performance and Student Engagement in Primary and Secondary Education

Title:Data and Code for: "Causal Effects of AI-Assisted Teaching Tools on Academic Performance and Student Engagement in Primary and Secondary Education: A Single-Center Multi-Campus Retrospective Cohort Study Based on Target Trial Emulation" Abstract / Overview:This repository contains the de-identified curated dataset and fully reproducible analytical code for the above-named study, which estimates the average and conditional average treatment effects of AI-assisted teaching tools on student academic achievement, engagement, and educational equity using a target trial emulation (TTE) framework. The study includes 8,400 students across 210 classrooms (grades 6–9) during the 2023–2024 academic year. Causal inference methods include stabilized inverse probability of treatment weighting (sIPTW), double machine learning (DML), causal forests, event-study difference-in-differences, instrumental variables, negative controls, decision curve analysis (DCA), and cost-effectiveness analysis (CEA). This dataset contains de-identified individual-level data from a retrospective cohort study of 8,400 students across 210 classrooms (grades 6–9) during the 2023–2024 academic year. Variables include anonymized student/classroom/school IDs, AI exposure intensity quartiles, baseline covariates (demographics, prior achievement, attendance, SES proxies, teacher characteristics, school resources), stabilized inverse probability of treatment weights, and primary outcomes (standardized achievement z-scores and student engagement composite), plus secondary equity and attendance indicators. The data are provided in CSV format with a full codebook (data_dictionary.csv) in the metadata folder. This dataset supports the replication of the statistical analyses reported in the associated manuscript. For the analytical code, please contact the corresponding author.