Estimating Causal Spillover Effects in Smallholder Farmer Networks in Western Kenya

Abstract

This is ongoing work. As farmers share information on practical agricultural techniques with each other, we want to understand what demographic and social network factors lead to greater information sharing and adoption. The goal of this research is to understand and estimate how efficiently information on disease management behaviors (DMB) spreads in smallholder farmer networks. In this research, we are concerned with DMBs with respect to plant and crop disease. More formally, the goal is to causally estimate the network spillover effects of specifc DMBs in smallholder farmer networks in Western Kenya. We intend to achieve this goal through a cluster randomized trial. Methods employed include re-randomization during the design phase, randomization tests to detect interference and estimating direct and indirect effect size via exposure networks.