Local circumstances and adaptations affect water-related interventions and livelihood outcomes. This local variation is crucial for developing resilient livelihood strategies, but also creates significant analytical challenges to assessing the likely impacts of water-related interventions.

This report presents an approach using a probabilistic, ‘fuzzy’ model of the links between water and livelihoods that takes these fundamental uncertainties into account. The model is grounded in the Sustainable Livelihoods framework, and is implemented as a Bayesian network. The approach is applied to data from a previous study in Northeast Thailand, and the research was supplemented by field visits and key informant interviews at farms, communities, and universities in Northeast Thailand.

This report presents a conceptual framework for analysing water-related interventions on poverty, an elicitation approach, and an example application. Also, it presents three innovations that resulted from the project: a novel way to represent institutions within the Sustainable Livelihood framework, a Bayesian approach to representing indicators as indirect evidence of a quantity of interest, and an elicitation technique for the conditional probability tables within a Bayesian network model.

Download publication here (pdf, 1,40 mb)