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 pa ir of jointly distributed (correlated) discrete random variables. The comparison model-the ''conventional'' approach to estimation using the se types of data-contained a programming error. Correction alters the results from the comparison model, but our original claim-that estimat ion by conventional methods can produce misleading inferences if ours is the appropriate specification-still stands. This is especially true if negative TNTP is precluded when E[WTP] is calculated. (C) 1998 Aca demic Press.