What drives the performance of Chinese urban and rural secondary schools: A machine learning approach using PISA 2018

Despite Chinese students' excellence in international academic assessment, a clear achievement gap exists between urban and rural areas. This study examines factors driving school-level academic performance disparity between urban and rural areas. Micro-level data from the 2018 Programme for International Student Assessment survey and a machine learning method are employed. Intermediate learning outcomes and student characteristics (e.g., class size, grade repetition rate, parents' socioeconomic status and emotional support, students' learning time, and credibility assessment skills) are essential in the difference in academic performance between the best and worst schools in urban areas. However, in rural areas, school characteristics such as complexity of admission policies, quality of learning environment, and number of computers are important. These findings highlight the need for tailored policies in both urban and rural areas.