The role of the reference alternative in the specification of asymmetric discrete choice models

• We compare asymmetric choice models expressed as deviations and absolute values.
• Two different design setting from two stated choice experiments are compared.
• The way the attribute levels are shown to respondents affects their rational.
• We find differences on the WTP and WTA obtained from the two specifications.
• We provide a guide for the selection of the most suitable asymmetric specification.

Within the discrete choice modelling literature, there has been growing interest in including reference alternatives within stated choice survey tasks. This paper analyses asymmetric discrete choice models by comparing specifications expressed as deviations from the reference point and specifications expressed in absolute values. The results suggest that the most suitable specification reflects the way the attribute levels are presented to respondents, either in terms of deviations or absolute values. Results further suggest significant differences on the willingness to pay and willingness to accept estimates obtained from the two different specifications proposed.