The Role of Consequentiality in the External Validation of Stated Preference Methods

This study examines consequentiality and information effects of stated preference methods by taking advantage of a unique opportunity to compare survey responses with a parallel, financially binding public referendum held in the Town of Middleborough, Massachusetts, concerning the adoption of a conservation and preservation policy to be funded by a property tax surcharge. Our survey setting departs from previous work in this area in that (1) many survey respondents were unaware of the upcoming referendum and (2) the survey “referendum” mirrors that of the public referendum. The survey and analysis are designed to elicit and control for respondent beliefs regarding the policy consequences of respondent choices and to take subsamples before and during a period of public information immediately prior the referendum vote. Using the survey sample of verified voters, we find no statistical differences between survey and referendum votes at either the aggregate or precinct levels for the consequential respondents, but do find evidence of negative bias at the precinct-level for the full sample. Highly significant negative bias is found in the inconsequential voter sample. Negative hypothetical bias is reinforced by comparing inconsequential estimates of willingness to pay (WTP) to those of consequential respondents. Tests of information effects revealed no evidence of bias in either consequential or inconsequential cases