The Reliability of Willingness-to-Pay Responses in the Contingent Valuation Method: A New Single-Sample Approach

The Contingent Valuation (CV) Method, like other stated preference techniques, seeks to measure economic preferences for public goods. Throughout the development and application of this non-market procedure, the accuracy of the measured preferences has been front-and-centre among practitioners and potential users. The most important issue of debate has been the extent to which the method can reliably measure economic preferences. In this article, a new methodology is described that enables multiple indicators of latent preferences. Multiple-indicator CV (MCV) enables the application of reliability analyses that are well established in psychology and sociology and represent the foundation of evaluating the measurement of latent variables. Furthermore, with the new MCV approach, the reliability of measurement at the individual level can be assessed in a single administration of the MCV survey thereby alleviating any need for longitudinal methodologies or comparison of mean estimates with other valuations of the same ecosystem service or public good should these be available. Once adequate reliability is established, the multiple-indicator framework supports the estimation of mean values via existing econometric techniques. With greater confidence in the reliability of measured contingent values, the interpretation of validity tests is enhanced.