Using choice experiments to improve the design of weed decision support tools

Abstract

The potential for computer-based decision support tools (DSTs) to better inform farm management decisions is well-recognised. However, despite considerable investment in a wide range of tools, the uptake by advisers and farmers remains low. Greater understanding of the demand and the most valued features of decision support tools has been proposed as an important step in improving the impact of DSTs. Using a choice experiment, we estimated the values that Australian farm advisers attach to specific attributes of decision support tools, in this case relating to weed and herbicide resistance management. The surveys were administered during dedicated workshops with participants who give weed management advice to grain growers. Results from various discrete choice models showed that advisers’ preferences differ between private fee-charging consultants, those attached to retail outlets for cropping inputs, and advisers from the public sector. Reliably accurate results were valued, but advisers placed a consistently high value on models with an initial input time of three hours or less, compared to models that are more time demanding. Results from latent class models revealed a large degree of personal preference heterogeneity across advisers. Although the majority of advisers attributed some value to the capacity for DST output that is specific to individual paddocks, approximately one quarter of respondents preferred generic predictions for the district rather than greater specificity. The use of a novel non-market valuation approach can help to inform development of decision support tools with attributes valued by potential users.