نوع مقاله : مقاله پژوهشی
1 استاد، دانشکدة مهندسی عمران، دانشگاه علم و صنعت ایران، تهران، ایران
2 دانشجوی کارشناسی ارشد، دانشکدة مهندسی عمران، دانشگاه علم و صنعت ایران، تهران، ایران
3 دانش آموخته دکتری، دانشکده مهندسی عمران، دانشگاه علم و صنعت ایران، تهران، ایران
عنوان مقاله [English]
This research was done with the aim of investigating the necessity of involving the decision making process and latent variables in discrete choice models to provide a more realistic representation of these models. Entering latent variables in discrete choice models is initiated from the fact that the concept of Behavioral theory is concerned with the manner in which a decision is made, while discrete choice models which are based on random utility hypothesis and economic consumer theory, are concerned with the observatory input and output variables. Discrete choice models are often proposed as an optimizing black box since this model attaches the observable inputs directly to the observable output (choice) and does not explain anything about the process of choice formation, while behavioral theory tries to consider the process of forming and existing a decision to have a more realistic representation of the choice process. In creating more realistic behavioral models, it can be said simply that, this method, considers hidden structures such as attitudes, perceptions and all concepts that influence the choice. In this method, psychometric measures are used which are features of fundamental latent variables. The aim of the representation of this structure is developing a comprehensive framework and methodology, to incorporate latent factors, as descriptive variables in discrete choice models. The method used in this research is a staged paradigm which is modeled in the first part of the latent variables using structural equation modeling and the output of this part is used in the choice model construction. The research findings also showed that entering latent variables in vehicle choice model has a statistical significant effect and causes the model with latent variables to have a better estimate compared with the models without these variables.