Rural migrant concentration and performance inequality in Chinese middle schools: A machine learning approach

This study proposes a new methodological approach by utilizing a machine learning-based clustering algorithm to measure academic performance inequality in Chinese middle schools. Unlike traditional methods that use single summary statistics, our approach clusters schools based on the entire empirical cumulative distribution function of student test scores, capturing more complex patterns of inequality. We classify schools into three distinct clusters reflecting varying degrees of inequality. Our findings reveal that schools with higher concentrations of rural migrant students are more likely to fall into more unequal clusters, where students face greater academic challenges. By comparing our method with traditional measures, we demonstrate its ability to detect subtle inequality patterns that traditional measures may overlook. This methodology provides valuable insights for targeted policy interventions to address disparities.