Partial attribute attendance in environmental choice experiments: A comparative case study between Guangzhou (China)...

Attribute non-attendance (ANA) in discrete choice experiment (DCE) exercises has attracted increasing, yet limited, scholarly attention. This paper attempts to investigate ANA in a comparative case study, with a focus on its patterns and their association with socioeconomic, behavioral and perceptual factors, as well as its impacts on willingness-to-pay (WTP) estimates. We deploy a four-level polytomous scale (always, often, seldom, and never considered) for respondents to state their various degrees of attribute attendance (SANA) in an identical DCE questionnaire about urban river restoration initiatives in two global cities with contrast socioeconomic contexts, yet similar request for restoring polluted and modified urban rivers, Guangzhou (south China) and Brussels (Belgium). The survey results reveal the existence of large proportions of partial attendance in two sampled cities. We use an extended mixed logit model, which incorporates separate parameters delineating each attribute's different attendance groups, to estimate respondents' average WTP values. We find that accounting for SANA could improve the goodness-of-fit of the model and affect the magnitude of mean WTP estimates. Respondents' attribute attendance level pertaining to various attributes is mainly associated with their perceived importance of urban rivers' ecosystem services, but may not be necessarily correlated with the strength of their preference for corresponding attributes as indicated by the mean WTP estimates. Whether this discontinuity between respondents' stated ANA levels and WTP estimates within Guangzhou sample questions the ability of DCEs to generate unbiased welfare estimation and policy guidance in developing countries calls for further studies.