First, we estimated one model for each crop with only the attributes
of the experiment. Then, we estimated two models, where a number of
socio-economic characteristics interact with the attributes in order to
capture observed heterogeneity. The models are estimated with
simulated maximum likelihood, using Halton draws with 500 replications.8
Although the experiment was generic, we included alternative
specific constants that are allowed to vary with the choice sets, since we
wanted to test whether any factors other than the attributes themselves
affected the choices. We have tested the hypothesis of different price
coefficients for farmers with different income levels, but in all cases we could not reject the hypothesis of equal coefficients. We therefore
estimate the models restricting the price coefficient to be constant
across subjects.