College of Agriculture & Natural Resources
Agricultural & Resource Economics

Seminar: Testing for what? Audit study evidence on the reliability and effectiveness of large-scale test-based accountability in India - Abhijeet Singh - Stockholm School of Economics

Event Series: 
When: 
Wednesday, Mar. 13, 3:30pm
Add to Calendar
Where: 
3121 Symons Hall
Event Type: 
Presenter: 
Abhijeet Singh - Stockholm School of Economics
Description/Notes: 

Abstract:

Public education systems in developing countries are increasingly instituting large-scale regular assessments of student learning – but the reliability of such assessments, and the effectiveness of their current use, remain open questions. In this paper, I present unique audit study evidence in the context of an annual census of student achievement, which covers ~7 million students in Grades 1-8 in government schools in Madhya Pradesh since 2011. I document three main results. First, in a representative audit study covering ~500 schools, I find substantial upwards manipulation in officially-reported levels of student achievement — the reported percentage of students answering a question correctly in official assessments is higher for all common questions, by as much as 100%, in comparison to audit data on the same students. Second, despite this high degree of manipulation in the reported levels of achievement, I see considerable agreement in the ranking of schools and students across official and independently-collected student assessments. Third, I evaluate the effectiveness of the stated current use of the assessments, which is to trigger remedial action, at the school and the student levels, in response to scores falling below pre-specified thresholds which classify continuous scores into letter grades. Using a sharp RD design, I show that there is no evidence that falling into the officially-targeted letter grades has any effect on future achievement of students and, at the school level, or on pedagogy, teacher absence or monitoring of schools. I conclude that, in similar settings with weak governance prone to the manipulation of tests, such assessments are unlikely to be useful for monitoring learning levels or for instituting large-scale test-based accountability systems but may yet be useful for targeting educational interventions 

Maintained by the IET Department of the College of Agriculture and Natural Resources. © 2019. Web Accessibility