Scaling Up Peer Education with Farmers in India
Kweku A. Opoku-Agyemang, Bhaumik Shah, Tapan S. Parikh · 2017
Researchers and agricultural advisory services often use data-driven approaches to target content to farmers. We demonstrate that allowing farmers to direct targeting procedures on their own is a more efficient way of scaling educational content than targeting based on demographic and agronomic indicators. Within the Khedut Saathi mobile message forwarding system, farmers received audio content on their phones, and could anonymously forward messages to up to five other phone numbers. Using Ordinary Least Squares regressions, we find that relative to targeting based on simple demographic and agronomic indicators, farmers are significantly more effective in identifying and targeting farmers most likely to be interested in messages -- based on how long farmers listen, and how likely they are to listen to complete messages before hanging up. Significant volumes of incoming calls were also made to our service from unrecognized phone numbers, implying a diffusion of the service offline by farmers. These results demonstrate the importance of farmers' own social connections for targeting and scaling educational content.