New publication | Improving administrative data at scale: Experimental evidence on digital testing in Indian schools
Education systems in many low- and middle-income countries face a “learning crisis,” where students are enrolled but achieve low levels of knowledge. Reliable data on student performance is crucial for improving outcomes, but existing assessments are often inaccurate due to cheating or manipulation.
A new study by Abhijeet Singh from the Stockholm School of Economics investigated whether digital tablet-based testing could improve data integrity in Indian schools. The study involved 2,400 schools in Andhra Pradesh, where students in grade 4 were tested either via traditional paper-based tests or tablet-based assessments.
The results showed clear differences. In schools using paper-based tests, students scored 16–20 percentage points higher on average than in an independently proctored retest, indicating widespread score inflation. In contrast, tablet-based tests closely matched the independently proctored results, suggesting significantly reduced cheating.
The study also found that procedures for detecting suspicious answer patterns in test data—developed in previous research on cheating in Italy—could be effectively applied in this new context. These methods could provide a scalable solution for ensuring assessment integrity.
“Digital testing shows real promise for improving data quality in challenging settings,” says Abhijeet Singh, Associate Professor of Economics at the Stockholm School of Economics. “With more reliable data, policymakers can better understand learning gaps and design targeted interventions.”
Reducing the risks of distorted data
Cheating in official assessments poses a major challenge in education reforms. Traditional paper tests are vulnerable to manipulation, particularly in environments with weak monitoring. Tablet-based testing makes copying harder by displaying one question at a time and prevents teachers from altering responses. Furthermore, the presence of an external invigilator in digital test settings adds another layer of accountability.
Wider implications for education policy
The findings suggest that introducing well-designed digital assessments could improve data reliability at scale, potentially benefiting efforts to monitor learning progress and inform education reforms. Digital testing may be especially useful in settings where traditional assessment methods are prone to manipulation.
