Bid design for non-parametric contingent valuation with a single bounded dichotomous choice format

This paper examines how the number of different bids used in a dichotomous choice contingent valuation exercise influences the non-parametric estimation of the mean willingness to pay. This task has been undertaken by several simulation experiments that combine different non-parametric procedures, number of bids, bid distribution, sample sizes, and willingness to pay distributions over the simulated population. The criteria used to analyze the effect of the number of bids in each experiment are (1) the mean square error of the differences between the actual and estimated mean maximum willingness to pay, and (2) an equality test between actual and estimated mean willingness to pay. The main conclusion is that it is advisable to use a larger number of distinct bids than currently practiced for the non-parametric procedures.