The impact of school quality and social interactions on student performance: Insights from the middle school educational reform in Beijing's Eastern City District

This dissertation consists in three chapters on the impact of school quality and of social interactions at school on educational outcomes. It uses random assignment of students to different middle schools introduced by the educational reform in Beijing, China.; Exploiting this natural experiment, the first chapter examines how school and teacher characteristics affect students' test scores. This is the first study to examine school effects using random assignment from the educational reform in China. Results show that the education reform has reached its goal of reducing inequalities in student performance across middle schools through random assignment. One additional teacher of high quality increases students' overall scores by 0.14 to 0.21 standard deviation of the overall test score. The teacher's effect is stronger for students of weaker performance. However, teacher's length of tenure decreases the overall score, expectedly due to job burnout. These results are robust to the possible attritions in the random assignment and peer influence.; The second chapter examines social interactions among students, and determinants of student's popularity. Students with special talents and stronger academic and family backgrounds, or similar to the majority in the school are more popular. This pattern is more evident when comparing among boys, and among students who were randomized out of their first choice schools. Segregation according to academic and family backgrounds exists. However, the reform seems to promote interactions among students of diverse backgrounds by randomizing students out of their first choice schools.; The third chapter investigates the impact of popularity and classroom heterogeneity on student academic performance. The results show that classroom diversities are more detrimental to the performance of students with weaker academic and socioeconomic backgrounds, but less so to students who were randomized out of their first choice schools. Using travel time and travel modes from home to school as instruments for popularity, the results show that popularity has a significant positive effect on student performance.