This project uses Bayesian Network to model impact pathways for interventions. Two interventions, on fruit trees in Kenya and on agricultural policy in Uganda, are evaluated as to their impact on human nutrition. Models are produced during participatory workshops and refined through iterations of modification and feedback. The project will produce a refined methodology, to be described in detail in a methods guide. This is a key activity for CoA5.2, because it explores the use of a new modeling approach, which appears to be a good fit for ex-ante impact assessment, but has not yet been used in development research.