Scale sensitivity and question order in the contingent valuation method

Abstract

This study examines the effect on respondents’ willingness to pay to reduce mortality risk by the order of the questions in a stated preference study. Using answers from an experiment conducted on a Swedish sample where respondents’ cognitive ability was measured and where they participated in a contingent valuation survey, it was found that scale sensitivity is strongest when respondents are asked about a smaller risk reduction first (‘bottom-up’ approach). This contradicts some previous evidence in the literature. It was also found that the respondents’ cognitive ability is more important for showing scale sensitivity when respondents are asked about a larger risk reduction first (‘top-down’ approach), also reinforcing the result that a ‘bottom-up’ approach is more consistent with answers in line with theoretical predictions for a larger proportion of respondents.

Keywords:

• cognitive ability
• contingent valuation
• mortality risk
• order effect
• scale sensitivity

Acknowledgements

Financial support from VTI, the Centre for Transport Studies, Stockholm, and the Swedish Civil Contingencies Agency is gratefully acknowledged. The authors would also like to express their gratitude to two anonymous reviewers for their helpful comments on an earlier draft. The usual disclaimers apply.

Notes

1. Scale and scope are used interchangeably in the literature to define the size of the good. Here we have chosen to use scale.

2. Scale sensitivity is also tested by Bateman et al. (2004) and the results indicate scale sensitivity both in a within-sample (internal) test and in the between-sample (external) test.

3. Since a preliminary analysis showed that the area of study had no significant influence on the WTP answers, this analysis based on area of study is omitted.

4. All prices in the paper are 2005 prices. US$1 = SEK 7.48 (www.riksbanken.se, 01/27/06.)

5. For a description of the test of cognitive ability and the questions used, see Andersson and Svensson (2008). The design of the CV survey is available upon request from the authors.

6. The annual bus fare at the time of the survey in Karlstad was SEK 3.690.

7. The baseline risk of this particular age group was also used by Persson et al. (2001) and Carlsson, Johansson-Stenman, and Martinsson (2004). We decided to use it since we wanted a baseline risk other than the respondents’ own age group to increase the chance of respondents perceiving the risks presented to them in the survey as exogenous.

8. The survey asks about the city council selecting from one of three bus companies that are distinguished by the levels of safety, when contracting out the service to a private firm. This corresponds to the actual procedure in the city of Karlstad as well as most other Swedish cities.

9. As we use standard diagnostic tests and regression techniques, we have not included a section where these tests and techniques are described. For a description of the techniques, see any textbook on econometrics.

10. Results are available upon request from the authors.

11. Ln(1.5) = 0.405