January 2022

IZA DP No. 15041: Improving Survey Quality Using Paradata: Lessons from the India Working Survey

Deepti Goel, Rosa Abraham, Rahul Lahoti

We describe the design and implemention of a paradata based method to reduce interviewer induced measurement error in a household survey in India. Our method identifies enumerators exhibiting deviant field practices, and provides them feedback to correct potentially faulty behavior. A novel feature is the emphasis on dynamic benchmarking within a group of enumerators facing similar field conditions. This helps to correctly pin down steady state levels of multiple data generating processes that exist within our survey. We also present evidence that our method succeeded in changing actual enumerator behavior in the field. Furthermore, we provide a complete prototype of how to operationalize paradata use in a resource constrained environment. At each step, we highlight the trade-offs involved, share insights from our own shortcomings, and provide recommendations to help make more informed choices. We hope our work will encourage the use of paradata to improve survey quality, especially in low- and middle-income countries where their use is still rare.