International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
Psycho-Physical Mapping of Alternatives in Random Regret Minimization Models: Varying Effect of Attribute Intensity
Sunghoon Jang, Soora Rasouli, Harry Timmermans

Last modified: 28 March 2017


Recently, regret-based choice models have been in the spotlight of transportation research. These models define regret as a function of objective (physical) attribute differences between choice alternatives. Contending that perceived rather than objective attributes generate regret, we recently explicitly incorporated the underlying psycho-physical mapping processes, based on the (generalized) Weber law, into random regret minimization models. The Weber law states that individuals perceive differences between stimuli as a constant ratio of the base stimulus, while the generalized Weber law adds a power coefficient to the base stimulus in the denominator of the ratio. Further research found that the Weber law is violated when the intensity (attribute value) of the stimuli is either small or large. Expressed in terms of the generalized Weber law, this empirical regularity can be mathematically depicted by a range-varying power coefficient, which is close to zero when attribute intensity is small, becomes equal to one in the middle intensity range, and then decreases towards zero again with higher intensity. This study elaborates this line of research and starts from the hypothesis that individuals’ psycho-physical mapping of attribute differences depends on the intensity (size) of the attributes of the chosen alternative. Whereas previous research assumed that the curvature of the mapping function is range-invariant, following the (generalized) Weber law, this study assumes that individuals’ perception of attributes differences, which make up the argument of the regret function, systematically varies with attribute intensity. To assess the empirical performance of the proposed models, two data sets are used. Estimation results and out-of-sample validation tests evidence that the model, which incorporates attribute intensity varying psycho-physical mapping functions in random regret minimization models, has a better goodness of fit than the original random regret minimizing models and their variants that include the (generalized) Weber law.


Conference registration is required in order to view papers.