Abstract. We test the robustness of visitor dichotomous choice willingness to pay (WTP)for
IntroductionMany water resource trade-offs involve reducing water for recreation to augment municipal or irrigation supplies. Since public agencies often do not charge market clearing prices for access to rivers for recreation, economists estimate the value of river recreation using either revealed preference methods such as the travel cost method [Ward, 1983[Ward, , 1987 they have traded money the form of travel costs for access to the river for recreation. In our study, sampled visitors took an average of 12 trips per year, thus providing them with the opportunity to learn as they would with repeated market transactions. This suggests the first ROC is satisfied since multiple trips provide the visitor with familiarity with the site they were asked to value. There is also little uncertainty about the good being valued as they were asked to value a trip to the site they are currently visiting. Finally, our survey elicited WTP.Nonetheless, there has been a great deal of recent research in the area of uncertainty of preferences and responses in CVM surveys. Li and Mattsson [1995] develop an approach to preference uncertainty that combines a dichotomous choice CVM question with the respondent's rating of their certainty in their response. Their empirical application required households to value forest access. They found explicitly incorporating the household's degree of uncertainty of response into the likelihood function reduced mean WTP, but it was not a statistically significant decrease. Ready et al. [1995] provide another alternative way to query the respondents about how certain they are regarding their WTP answers. Instead of a continuous measure, the respondents check-off whether their response to the bid amount they are asked to pay is "definitely yes," "probably yes," "maybe yes," "maybe no," "probably no," and "definitely no." In the Ready et al., study this polychotomous choice process resulted in higher WTP than dichotomous choice CVM. Champ et al. [1995] used a 10-point rating scale of respondent certainty to recode any dichtomous choice responses that were not "definitely yes" to a "no." When the recoded data were analyzed with a logit model, the resulting WTP value was closer to the actual cash value than the original dichotomous choice WTP.Our approach differs from the previous approaches in that we check for uncertainty by (1) asking visitors with affirmative responses to their bid if they are certain they would pay this amount. This provides them a second chance to rethink and reconsider their answer: (2) rephrasing the WTP question from 1187