Modeling household discrete choice behavior incorporating heterogeneous group decision-making mechanisms

To represent heterogeneous group decision-making mechanisms in choice behavior, this paper attempts to develop a new household discrete choice model by integrating different types of household choice models based on latent class modeling approach under the principle of random utility maximization, where household utility function at each model is defined to theoretically reflect its members' preferences and intra-household interaction. A latent class corresponds to a particular group decision-making mechanism. The proposed model can deal with not only the choice situations where multiple household members involved in joint decision are known a priori, but also the situations where the involved members are unknown. As a case study, three types of household utility functions are dealt with: multi-linear, maximum and minimum types, in the context of couples' car ownership behaviors. Due to the limited sample size, paired combinations of the three utilities are examined and compared. Using the data collected in two Japanese cities in 2004, the effectiveness of the proposed household decision model is empirically confirmed. It is found that changes of utility combinations result in very complicated variations in both signs and values of the model parameters and choice probabilities by multi-linear utility show small variations across different combinations.