The Effects of Changing Cost Vectors on Choices and Scale Heterogeneity

Abstract

Choice Experiments (CE) are widely used to estimate the values of changes in non-market goods and services. A cost attribute is typically included in a CE questionnaire to enable the estimation of monetary values for changes in the non-market attributes presented. Notwithstanding the central importance of the cost attribute, relatively little research has been undertaken on the impacts of varying cost attribute levels on value estimates, or on individual heterogeneity. In this paper, I present results from mixed logit and generalised mixed logit models that account for unobserved idiosyncratic preference and scale heterogeneity. Respondents are found to anchor their choices on the relative cost levels presented in the survey with results suggesting that people are more sensitive to relative rather than absolute cost vectors. However, the higher cost levels do not lead to significantly higher value estimates, partly because of observed preference heterogeneity towards the environmental attributes. An important observation is that scale heterogeneity is important: accounting for scale— as well as preference—heterogeneity in the generalised mixed logit model leads to significantly improved model fit. The results indicate significant unobserved error variance across respondents, unrelated to whether a high or low cost vector is used.