Social networks are an important mechanism for diffusing information when institutions are missing, but there may be distributional consequences from targeting only central nodes in a network. After implementing a social network census, one of three village-level treatments determined which treated nodes in the village received information about composting: random assignment, nodes with the highest degree, or nodes with high betweenness. We then look at how information diffuses through the network. We find information diffusion declines with social distance, suggesting frictions in the diffusion of information. Aggregate knowledge about the technology did not differ across targeting strategies, but targeting nodes using betweenness measures in village-level networks excludes less-connected nodes from new information. Women farmers are less likely to receive information when betweenness centrality is used in targeting, suggesting there are important gender differences, not only in the relationship between social distance and diffusion, but also in the social learning process.
January 31, 2018