Decision analysis and Bayesian Network (BN) tools allow the incorporation of disparate data sources and uncertain inputs, to create a representation of the current understanding of cause and effect relationships within the target system. To test these tools, we held a week-long workshop was organized in Nairobi, Kenya in 2017 to gather 20 experts and analysts to collaboratively model the potential livelihood impacts of fruit trees on smallholder farms. A broad range of experts were present, including government and non-government organizations, agricultural technicians and practitioners, as well as academics. They were trained and guided through structured group work procedures, in order to collaboratively build a probabilistic causal model for nutritional outcomes of fruit trees on farms. Through the use of decision analysis methods, and combining features of several participatory procedures into a customized structured conversational process, expert knowledge was used to generate relationships and probable states of variables of importance and parameterize them using a BN. Through the workshop, the team was able to identify the critical determinants of the effectiveness of fruit trees for nutrition, establish the factors determining the success of the steps along the impact pathway, and define the context of a BN for calculation and further analysis. The model indicates that fruit tree planting benefits the nutritional status of households, decreasing chances of hunger and micronutrient deficiency. The resulting model will be used to inform policy and development action in Kenya.
4th Annual meeting of the FLARE network (Forests & Livelihoods: Assessment, Research, and Engagement) FLARE Website.