资源环境科技发展态势分析平台

A key assumption of stated preference methods is that individuals have well-formed preferences that are robust over time. Both the discovered and constructed preference perspectives imply this is not necessarily the case. There can be a large situational component to expressed preferences that add to the uncertainty of sampling error. Most non-market valuation studies only collect data from one point in time so the degree of temporal variability cannot be tested. Test-retest studies that provide data from two points in time generally find significant differences in preference structure and willingness-to-pay (WTP). In this study we test stability of WTP for beach erosion management using a fully ranked discrete choice experiment survey with not one but two retests over a six month period. We find that stability does not improve with the additional repetition as the preference discovery hypothesis implies it might. WTP confidence intervals overlap but the models are significantly different at each point in time, even after allowing for variation in choice error. Either the survey did not facilitate sufficient preference discovery or preferences were reconstructed. However, respondents with high scores of self-reported certainty in their choices in the first survey had significantly more stable WTP estimates. (C) 2017 Elsevier B.V. All rights reserved.