Analysing social data on adoption of conservation practices: Exploring Bayesian networks

Australia's Natural Resource Management (NRM) regions are required to report to the Australian Government on the impact their investments have on natural resource condition. Reporting to date has typically been limited to describing the nature of investment and on-ground activities that have taken place. In some cases activities are linked by crude assumptions to an expected change in condition. With the Australian Government pushing for regional reporting that focuses on outcomes (change in condition) rather than outputs (activity and dollars spent), the regions are seeking ways to improve their current investment planning and resource condition reporting. This requires an improved understanding of biophysical systems to establish causal links between management actions and change in resource condition, and social research to better understand who in the community is likely to respond to which type of environmental programs and why. Techniques exist to analyse large complex social data sets including statistical and social-psychology models. Bayesian Networks (BNs), which are increasingly being used to model environmental systems, have not often been used in the analysis of social data. A BN (Figure 1) is a dynamic way to analyse complex cause and effect relationships. This paper explores the utility of BNs for analysis of social data sets and compares this approach to other more commonly applied techniques. Survey data from the Wimmera Catchment Management Authority was used to develop a BN model of the social drivers of conservation activity adopted by landholders to protect native vegetation. The resulting BN shows relationships between a landholder's likelihood of fencing native vegetation and their values, knowledge, attitudes, income and access to government support. It clearly illustrated the importance of government funding in the uptake of conservation, but also showed that a significant amount of fencing was carried out in the absence of government programs. In the absence of government funding, on-farm income was found to be critical to the uptake this activity, illustrating that whilst landholders may have been willing to adopt recommended practice, the behaviour depended on their 'capacity to change' (in this case, financial capacity). This paper suggests that BNs based on social research could be used by managers to support their decisionmaking and reporting, and has value for researchers as a tool for analysing, interpreting and communicating social data. (Figure Presented)