A Comparison of Approaches to Calculating Confidence Intervals for Benefit Measures from Dichotomous Choice Contingent Valuation Surveys

This paper compares the performance of four approaches to calculating confidence intervals around dichotomous choice contingent valuation method (DC CVM) benefit estimates. The performance of the approaches is compared using Monte Carlo simulation techniques for the two most common specifications for the welfare estimate. The results indicate that all four methods tended to perform well on average but the methods differed in the frequency with which they performed well. The results indicate the best choice depends on the sample size, on the distribution of the welfare estimate, and on the choice of functional form for the welfare estimate.