Making Networks Work for Policy

This project tests whether theories of learning on social networks can be leveraged to improve agricultural extension outcomes. Over a four-year period, researchers addressed this gap by examining the impact of targeting information based on social networks as a way to communicate messages. In particular, they mapped social networks, then selected lead farmers based on predictions from theory and tracked information dissemination and adoption of an improved maize cultivation practice called “pit planting” in 200 villages in Malawi. Researchers specifically tested variations to the standard method of selecting contact farmer for extension services, such as mapping village social networks to identify contact farmers with the highest potential to share information. Identifying good contact farmers mattered: in villages where extension selected contact farmers via business as usual there was a 50% chance that no other farmers ever adopted pit planting. In contrast, in 85% of villages where contact farmers were selected using network theory, at least some social diffusion takes place, and adoption rates are significantly higher than in benchmark villages. The researchers find that the most successful treatment is one which guarantees as many farmers as possible multiple contact points, which is consistent with learning models where multiple contact points are necessary to motivate adoption. Several other patterns in the data confirm this insight.