ESTIMATION USING CONTINGENT VALUATION DATA FROM A "DICHOTOMOUS CHOICE WITH FOLLOW-UP" QUESTIONNAIRE by

Dichotomous choice (referendum) contingent valuation questions have gained popularity over the last several years due to their purported advantages for avoiding many of the biases known to be inherent in other value elicitation formats. However, this type of valuation question is inefficient in that a much larger number of observations is required to identify the distribution of values with any degree of accuracy. An alternative questioning strategy introduces a second value threshold which elicits a second discrete response. Previous analyses of these double-bounded referendum surveys have maintained the hypothesis that the identical unobserved resource value motivates both responses. We relax this assumption and find that while the implied resource values deduced from the first and second responses may be drawn from the same distribution, and the implied values are highly correlated, the values elicited by each question are definitely not identical. Furthermore, assuming that they are identical severely distorts the estimated valuation distribution.