The purpose of this research is to resolve the controversy about what exactly is being measured by subjects? responses in stated preference surveys used to measure environmental and health values. A second objective is to identify how stated preference methods might be modified to enhance the validity of their results.
Critics have argued that stated preference responses lack the properties required of preferences in economic theory, and therefore do not constitute a valid measure of economic value. A major component of this claim involves what has come to be known as the embedding effect (Kahneman and Knetsch, 1992). Phenomena related to the embedding effect play a prominent role in recent critical assessments of stated preference by Payne et al. (2000); Payne, Bettman and Schkade (1999); Kahneman, Ritov and Schkade (1999), and Baron (1997), as well as Diamond et al. (1993).
The investigators plan to conduct a new set of stated preference experiments designed to address the theoretical and conceptual issues raised by stated preference critics in a more systematic and comprehensive manner. The research project will re-examine embedding in a manner that rectifies some flaws we believe exist in these empirical studies, by replicating and correcting, to the extent feasible, the surveys on which the studies are based. The research will also identify the circumstances under which embedding occurs in stated preference surveys, and to investigate whether there are ways of structuring a stated preference survey such that embedding can be controlled or eliminated.
The research will extend and improve on the existing literature in four ways: (1) it will pay more careful attention to considerations of utility theory in formulating the embedding experiments; (2) it will examine whether embedding is affected by the choice of elicitation format, especially closed-ended versus open-ended valuation in stated preference; (3) it will investigate whether embedding effects occur to the same degree in alternative methods of preference elicitation, particularly conjoint analysis; and (4) it will investigate whether embedding effects occur for use values and private market commodities and whether and how these are different from embedding effects associated with non-use values for public goods.
These experiments should provide better insight into the nature of embedding phenomena, which could become a basis for designing preference elicitation procedures that either are less affected by embedding or are effected in a determinate manner such that it would be possible to make corrections when analyzing survey responses in order to adjust for whatever embedding occurs. The end product would be more reliable procedures for measuring individual preferences in non-market valuation.