Statistical Efficiency of Double‐Bounded Dichotomous Choice Contingent Valuation

Abstract The statistical efficiency of conventional dichotomous choice contingent valuation surveys can be improved by asking each respondent a second dichotomous choice question which depends on the response to the first question—if the first response is “yes,” the second bid is some amount greater than the first bid; while, if the first response is “no,” the second bid is some amount smaller. This “double‐bounded” approach is shown to be asymptotically more efficient than the conventional, “singlebounded” approach. Using data from a survey of Californians regarding their willingness to pay for wetlands in the San Joaquin Valley, we show that, in a finite sample, the gain in efficiency can be very substantial.