Efficient and robust willingness-to-pay designs for choice experiments: some evidence from simulations

We apply a design efficiency criterion to construct conjoint choice experiments specifically focused on the accuracy of marginal estimates. In a simulation study and a numerical example, the resulting optimal designs are compared to alternative designs suggested in the literature. It turns out that optimal designs not only improve the estimation accuracy of the marginal, as expected on the basis of the nature of the efficiency criterion, but they also considerably reduce the occurrence of extreme estimates, which also exhibit smaller deviations from the real values. The proposed criterion is there for evaluable for non-market valuation studies as it reduces the sample size required for a given degree of accuracy and it produces estimates with fewer outliers.nrpages: 1-32status: publishe