The focus of this article is on the ‘educational attainment’ of social groups in India, defined as the proportion of 21–29-year-olds in the different social groups that are graduates. Within this context, the article makes three contributions. The first is in terms of decomposition methodology: It shows how decomposition analysis can be used to breakdown graduate educational attainment (GEA) by population subgroups and region to offer insight into the policy levers available for improving the GEA of certain groups. The second is to a methodology for computing probabilities (termed ‘recycled predictions’) for isolating the group-specific probabilities of GEA which can be totally attributed to group identity. The third is to use these probabilities to break down the differences in proportions between social classes into a ‘social group effect’ and an ‘endowments effect’. The data for this article is from India’s Periodic Labour Force Survey for the period 2017–2018.

中文翻译：

---

印度社会群体研究生教育程度的差异：教育偏好与教育友好型捐赠

本文的重点是印度社会群体的“教育程度”，定义为不同社会群体中21-29岁毕业生的比例。在此背景下，本文做出了三点贡献。第一个是分解方法论：它展示了如何使用分解分析按人口亚组和地区分解研究生教育程度（GEA），以深入了解可用于提高某些群体的 GEA 的政策杠杆。第二个是计算概率的方法（称为“循环预测”），用于隔离 GEA 的特定群体概率，这些概率完全归因于群体身份。第三是利用这些概率将社会阶层之间的比例差异分解为“社会群体效应”和“禀赋效应”。本文的数据来自 2017-2018 年印度定期劳动力调查。