Optimal Bid Selection for Dichotomous Choice Contingent Valuation Surveys

This paper develops a model for optimal survey design for the dichotomous choice contingent valuation method that finds the bid amounts as well as the sample sizes corresponding to each bid. The model uses an iterative procedure to select the survey design that minimizes the mean square error of the welfare measure. A set of Monte Carlo simulations shows that the model produces significantly lower mean squared errors for the welfare estimate than the next best model produces. Survey design under alternative assumptions regarding the statistical distribution of the welfare measure is also addressed. The model produced significantly different survey designs depending on whether the distribution was assumed to be symmetric or asymmetric.