Published: 14 December 2019| Version 1 | DOI: 10.17632/sttvx33637.1
Frank Tchuwa, Kate Wellard, John Morton,


Study sort to investigate processes which affect flow of information in social networks created around soil health innovations. It also looked at behavioral change of farmers accessing soil health information through social networks.Data was collected using egocentric social network analysis method. It involved a survey of 192 ego farmers participating in lead farmer, farmer, field school, farmer research team and farmer research network participatory models in Malawi. Another 282 alter farmers were interviewed (those interacting with egos). Network data (where farmers shared information) was analysed using Netdraw software to generate degree measure of centrality and then visualise network structures. The EI index was calculated to determine ego and alter similarity (presence of homophily). Homophily behaviour is the tendency of individuals to associate with others they perceive as similar to them.EI index ranges from -1 (means perfect homophily) to +1 (means perfect heterophily). A perfect homophily is when the ego has ties to only those alters who belong to his or her group. A perfect heterophily is when ego has ties to only those alters who belong to a different group. When ego’s ties are equally distributed, the EI is zero. The E-NET software was used to compute EI indexes. The generated indexes were then imported into Stata version 13. Non-parametric statistics were used to summarise the indexes. They included median, Kruskal Wallis and Mann-Whitney (U) tests. Dunn’s test was used to compare the medians. Data on sources of information, namely horizontal networks (e.g., discussions with fellow farmers within the village), vertical networks (e.g., extension workers and researchers), and Information and Communication Technology ( (e.g., radio and phones), as well as the data on farmer intention to apply soil health innovations was summarised in Stata. The constraints to application of innovations were subjected to content analysis using ATLAS.ti version 7.5.7 software and visualized using the NETDRAW software. Results reveal dense social network structures with different spaces (nodes) through which soil health information was diffused in the communities. However, diffusion of information mainly occurred during farmer interaction in the learning plots. The networks where characterised by homophily behaviour, whereby information flow was restricted among individuals of similar sex and location. Majority of farmers accessing soil health information through networks did not intend to apply the acquired knowledge on a large area. This data is useful to researchers working on social network analysis, especially those analyzing diffusion of innovations in agricultural extension.



University of Greenwich Natural Resources Institute, Lilongwe University of Agriculture and Natural Resources


Social Sciences