2011 Volume 4 Issue 1 Pages 23-31
This study introduces a new method of investigating whether results from a contingent valuation (CV) survey are rationally understandable. The method checks whether differences in response times to survey questions are realized across survey participants in a manner consistent with theoretical expectations. The new method is applied to an experimental CV survey of Japanese climate policy. The survey divides its respondents into two groups. One group is subject to conditions where the results of the survey are expected to have a consequence in government policy formulation. The other group is subjected to conditions where the survey results have virtually no impact on government policy. The survey is computerized and the response time for each page of its questionnaire is automatically measured. The average response time among the first group is found to be longer than that of the second. The response times are regressed on explanatory variables including characteristics of the questionnaire pages and personal attributes. A panel regression technique is employed. The signs of the coefficients of the explanatory variables among the first group perfectly meet theoretical expectations, whilst most signs of the coefficients for the second group fail to meet the expectations. Using this new method of investigating response times together with conventional methods of investigating stated values will improve the procedure for assuring the correctness of CV results.
