Measuring the Difference in Mean Willingness to Pay When Dichotomous Choice Contingent Valuation Responses Are Not Independent

Dichotomous choice contingent valuation surveys frequently elicit multiple values in a single questionnaire. If individual responses are correlated across scenarios, the standard approach of estimating willingness to pay (WTP) functions independently for each scenario may result in biased estimates of the significance of the difference in mean WTP values. This paper applies an alternative bivariate probit approach that explicitly accounts for correlation across errors in the estimation of WTP and mean WTP. Using data from three separate dichotomous choice contingent valuation studies, this correlation is demonstrated to have an effect on the significance level of mean WTP difference tests.