Working Paper No. 04-02

Stated Preference Analysis of Public Goods: Are we asking the right question?
Nicholas E. Flores and Aaron Strong
January 2004


In this paper we develop a simple question response model to dichotomous choice/referendum sytle stated preference questions. In our model, participants gather project information from the survey description, form a prior distribution of project costs, receive additional information through the survey's stated cost, and then respond to the dichotomous choice/referendum question with the belief that a yes increases the probability the project will be provided while a no decreases the probability the project will be provided. Based on our analysis we conclude that participants will generally not respond truthfully in the sense that they may say no to a stated cost that is less than their willingness to pay for a project or yes to a stated cost that is more than their willingness to pay. Given our findings, the resulting mean estimates will not in general be upwardly or downwardly biased. Rather the design of the stated costs used in the survey will influence the mean estimates from various studies in an unpredictable fashion. Our model predicts responses that are consistent with "yea-saying," answering yes to any stated cost amount, and "neah-saying," answering no to any stated cost amount. Further, we offer a potential solution to this problem involving the use of a cost share financing mechanism. Unfortunately this solution is not always applicable.