This paper introduces a new method to measure school-level inequality based on Item Response Theory (IRT) models. Categorical data collected by large-scale assessments poses diverse methodological challenges hinder measuring inequality due to data truncation and asymmetric intervals between categories. I use family possessions data from PISA 2015 to exemplify the process of computing the measurement and develop a set of country-level mixed-effects linear regression models comparing the predictive performance of the novel inequality measure with school-level Gini coefficients. I find school-level inequality is negatively associated with learning outcomes across many non-European countries.

中文翻译：

---

基于学校水平的不平等测量的分类数据：一种应用于PISA的新颖方法

本文介绍了一种基于项目反应理论（IRT）模型的衡量学校水平不平等的新方法。大规模评估收集的分类数据由于数据截断和类别之间的不对称间隔而带来了各种方法上的挑战，阻碍了衡量不平等。我使用了来自PISA 2015的家庭财产数据来举例说明该度量的计算过程，并开发了一组国家/地区混合效果线性回归模型，将新型不平等度量的预测性能与学校水平的基尼系数进行了比较。我发现在许多非欧洲国家，学校水平的不平等与学习成果负相关。