A model for the simultaneous inference of attribute nonattendance and taste heterogeneity
Access status:
Open Access
Type
Working PaperAbstract
An extensive literature has recognised that when travel choices are made, only a subset of the attributes of the choice alternatives may be considered or attended to by each decision maker. This paper introduces the random parameters attribute nonattendance (RPANA) model. Attribute ...
See moreAn extensive literature has recognised that when travel choices are made, only a subset of the attributes of the choice alternatives may be considered or attended to by each decision maker. This paper introduces the random parameters attribute nonattendance (RPANA) model. Attribute nonattendance (ANA) is estimated through the model, which does not rely on stated ANA, although the latter can be introduced as a covariate, in recognition that whilst concerns have been raised about the reliability of stated ANA, it may provide the analyst with valuable information. The model employs a latent class structure to capture an elevated mass of taste coefficients at zero; a technique widely employed in the literature. However, preference heterogeneity can additionally be captured as a continuous distribution, with random parameters. The latent class component introduced in this model is highly flexible and parsimonious, and can exploit any independence of ANA across the attributes, whilst also handling correlation in ANA across subsets of attributes, as needed. Results are presented from a stated choice experiment investigating short haul flights. Non-zero probabilities of ANA are estimated for all attributes, and the RPANA model represents an improvement on the random parameters logit model in terms of model fit. Specific issues with model identification are discussed, and some potential pitfalls noted.
See less
See moreAn extensive literature has recognised that when travel choices are made, only a subset of the attributes of the choice alternatives may be considered or attended to by each decision maker. This paper introduces the random parameters attribute nonattendance (RPANA) model. Attribute nonattendance (ANA) is estimated through the model, which does not rely on stated ANA, although the latter can be introduced as a covariate, in recognition that whilst concerns have been raised about the reliability of stated ANA, it may provide the analyst with valuable information. The model employs a latent class structure to capture an elevated mass of taste coefficients at zero; a technique widely employed in the literature. However, preference heterogeneity can additionally be captured as a continuous distribution, with random parameters. The latent class component introduced in this model is highly flexible and parsimonious, and can exploit any independence of ANA across the attributes, whilst also handling correlation in ANA across subsets of attributes, as needed. Results are presented from a stated choice experiment investigating short haul flights. Non-zero probabilities of ANA are estimated for all attributes, and the RPANA model represents an improvement on the random parameters logit model in terms of model fit. Specific issues with model identification are discussed, and some potential pitfalls noted.
See less
Date
2013-03-01Department, Discipline or Centre
ITLSShare