Estimation Using Contingent Valuation Data from a “Dichotomous Choice with Follow-Up” Questionnaire: Reply

Our original article proposed that the two yes–no willingness-to-pay (WTP) responses produced by this survey format should be viewed as a pair of jointly distributed (correlated) discrete random variables. The comparison model—the “conventional” approach to estimation using these types of data—contained a programming error. Correction alters the results from the comparison model, but our original claim—that estimation by conventional methods can produce misleading inferences if ours is the appropriate specification—still stands. This is especially true if negative WTP is precluded when E[WTP] is calculated.