Ooh la la: Testing the one-and-one-half bound dichotomous choice elicitation method for robustness to anomales

Although attractive in terms of its incentive compatibility, the standard single bound (SB) dichotomous choice technique for eliciting willingness to pay (WTP) responses in contingent valuation surveys has a major drawback in terms of its low statistical efficiency. While alternatives such as the double bound (DB) approach (which supplements an initial SB style question concerning a specified bid amount with a subsequent follow-up question concerning a different bid amount) offer improved statistical efficiency, they do so at the cost of compromised incentive compatibility and have also been shown to be vulnerable to a number of response anomalies. An innovative alternative, the one-and-one-half-bound (OOHB) dichotomous choice approach, has recently been proposed by Cooper, Hanemann and Signorello (2002). The OOHB differs from the DB in a number of important respects; the most important being that while each respondent is again exposed to two bid amounts, these are presented prior to any response as upper and lower limits on the cost of schemes. This preserves the incentive compatibility of responses concerning those two limits while generating most of the efficiency gains afforded by the DB method. However, Cooper, Hanemann and Signorello fail to test the method for robustness against response anomalies. Such a test is provided by the present paper. A number of theoretical consistency hypotheses are formulated by contrasting standard expectations with those derived from non-standard reference dependent utility theory. These are tested through the first application of the OOH method within its intended public goods context in a study concerning WTP for remediating impacts upon water quality associated with climate change. Data is collected through a face-to-face survey of over 1250 UK households. Results reject the theoretical consistency of elicited WTP responses showing that the OOHB is highly vulnerable to a number of anomalies. In particular acceptance rates for a given bid amount varied according to which other amount it was paired with and the order in which responses were elicited. We speculate upon the implications of these findings.