Latent Class Model

We used discrete choice experiments to explore the potential adoption of alternative agricultural systems. We analyse the heterogeneity of farmers’ preferences and heuristics of choices using a latent class model, where class can include different heuristics such as attribute non attendance, and elimination by aspect.

We developed a best–worst scaling (BWS) experiment, in which farmers have to declare the “most” and the least “important” goals they use when making decisions. We first derive a ranking of the goals according to the population average, which showed the importance of rice self-sufficiency and transmission of farm capital. We then use a scale-adjusted latent class analysis. We identified four groups of homogenous preferences among farmers.