种姓认同与教师的偏见期望：来自印度比哈尔邦的证据
Caste identity and teachers’ biased expectations: Evidence from Bihar, India
Journal of Development Economics · 2025 · [{"name": "Ritwik Banerjee", "affiliation": ["Indian Institute of Management Bangalore"]}, {"name": "Satarupa Mitra", "affiliation": ["Manipal Academy of Higher Education", "T A Pai Management Institute"]}, {"name": "Soham Sahoo", "affiliation": ["Indian Institute of Management Bangalore"]}, {"name": "Ashmita Gupta", "affiliation": ["Asian Development Research Institute"]}]
中文摘要
我们利用一项大规模、详细且具有代表性的印度比哈尔邦公立学校调查数据，考察教师是否基于种姓身份对学生持有系统性偏误的预期。学生参加标准化测试以确定其实际学业排名，而教师则独立评价每名学生在班级中的相对位置。学生实际排名与教师感知评价之间的差距，构成了我们衡量教师评价偏误的指标。采用教师固定效应方法，我们发现，与同一班级中由同一教师教授的前进种姓同伴相比，前进种姓教师系统性地低估后进种姓学生的表现。无论采用后进种姓身份的替代定义，还是采用不同的学生表现度量，这些基于种姓的评价偏误差异均保持稳健。
Abstract
We investigate whether teachers hold systematically biased expectations about students based on caste identity, using data from a large, detailed, and representative survey of public schools in Bihar, India. Students take standardized tests that determine their actual academic rank, while teachers independently rate each student’s relative standing within the class. The gap between a student’s actual rank and their teacher’s perceived rating gives us a measure of the teacher’s Evaluation Bias. Using a teacher fixed-effect approach, we find that forward caste teachers systematically underestimate the performance of backward caste students compared to forward caste peers taught in the same class. These caste-based differences in Evaluation Bias remain robust to alternative definitions of backward caste status and different measures of students’ performance.