Constructed preference stability: a test–retest

Abstract

Preference construction has been argued to undermine the use of stated preference results like willingness to pay (WTP) values in policy evaluations. Preferences constructed at the time of the valuation task are expected to be unstable and hence unreliable. Choice repetition and experience have been shown to stabilize preferences and increase choice consistency. The objective of this article is twofold. First, to examine the effect of self-reported construction bias on stated preferences during a repeated choice contingent valuation (CV) study. Second, to test the effect of choice repetition on the stability of the articulated preferences in a test–retest experiment using a single and double bound CV referendum elicitation format. As expected, unfamiliarity with the environmental good in question results in a significant impact of the survey instrument on preference construction. This in turn influences the respondent decision to participate in the contingent market, but after controlling for self-selection, cognitive and motivational characteristics, not the size of the WTP value. Evidence is found that choice experience filters out possible construction bias. In the test–retest experiment, choice repetition furthermore generates more stable and consistent preferences compared to the results of the NOAA endorsed ‘one-shot’ single bound approach in CV.

Keywords:

• preference construction
• preference stability
• contingent valuation
• test–retest
• biodiversity
• marine reserves
• North Sea

Acknowledgements

The research presented here was funded by the National Institute for Coastal and Marine Management (RIKZ) in The Hague, the Netherlands. The useful comments and suggestions from Stavros Georgiou, Anthony Ryan and Joop de Boer on a draft version of the paper are gratefully acknowledged. The usual disclaimer applies.

Notes

1. In those cases where respondents face a novel and relatively complex choice task, Payne et al. (1999) developed a ‘building code’ for ‘good’ preference construction. Problematic preference construction may be due, among others, to myopic decision frames, inappropriate problem representation, lack of comprehension of the information provided and avoidance of tradeoffs.

2. Most respondents preferred marine protected areas where some economic activities are allowed under strict conditions (58%), followed by zoning of areas where economic activities are banned altogether (38%). Three percent preferred the current situation. Only 1.5 percent of the sample population was unable to make up their mind and did not know which valuation scenario to choose.

3. The standard errors for the double bounded WTP values are estimated based on the Krinsky and Robb (1986) procedure.

4. This result is confirmed based on the Median test (chi square = 8.808) albeit at a lower confidence level (p < 0.066).

5. The most important protest reason was that respondents felt ‘the North Sea should be protected by law, not by asking people to pay for it’ (11%), followed by ‘the polluter should pay’ (1%), ‘protection of the North Sea is a government task’ (0.3%) and ‘lack of sufficient information’ (0.2%).

6. Test results available from the author.

7. The standard errors for the double bounded WTP values are estimated again based on the Krinsky and Robb (1986) procedure.

8. The outcome of the t-test is t = 0.571 (p = 0.569) for the control group and t = 0.181 (p = 0.857) in the case of the experimental group.