International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
Identifying Individual Choice Mechanism Profiles for Air Travel Behaviour
Felipe Gonzalez-Valdes, Sebastian Raveau

Last modified: 28 March 2017


Understanding and modelling travellers’ decisions, by correctly analysing their preferences and being able to forecast their choice, is essential in the planning and operation of any transport system. The air travel industry is certainly not an exception, particularly in a context of high competition, where airline carriers seek to attract customers by differentiating themselves from their competitors.

To the best of the authors’ knowledge, applications of  discrete choice models to the air travel context reported in literature are based on Random Utility Theory with few exceptions. One of these exceptions is the work of Gonzalez-Valdes and Raveau (2017), whose modelling approach enables people to follow, stochastically, one of two choice mechanism, Elimination by Aspects (Tversky, 1972a, 1972b) and Random Utility Maximization (McFadden, 1973).

Previous studies support the fact that an individual could follow different choice mechanisms in different (or even the same) context (Denstadli et. al., 2012; Wansink and Sobel, 2007). This study seeks to identify and characterize different choice mechanisms that individuals can follow when facing different different choice scenarios at an individual level. In the proposed model, individuals would have a probability of following a given choice mechanism; such probability is different across the population, but could be the same for observations of the same individual. Different sociodemographic characteristics, as well as features from the choice scenarios themselves, are used to characterize the consideration of different choice mechanism. Then Bayesian conditioning is done to identify individual probabilities.

The proposed modelling approach is tested in an air travel context. The data consists of stated preferences for travel itineraries from Singapore to 24 different destinations in Australasia, which were divided in 12 short-haul trips and 12 long-haul trips. All alternative itineraries correspond to real flights for given travel dates and are characterized according to seven attributes: fare, total time, number of connections, connecting time (if any), if the airline is a low-cost carrier, if the airline is part of an alliance, and if the itinerary requires to be at the airport at an “inconvenient” time (earlier than 9:00am and/or later than 9:00pm).

An exclusively compensatory approach, such as Random Utility Maximization (which is traditionally adopted in the literature), might not be suitable for modelling air travel preferences. This study considers a multiple choice mechanisms framework, that combines a compensatory approach (Random Utility Maximization, RUM) and a non-compensatory approach (Elimination by Aspects, EBA). The proposed framework allows to capture different individual decision-making processes within a population.

The estimation results are analyzed in terms of goodness-of-fit, forecasting capabilities and marginal rates of substitution for the RUM model (which are considerably affected when different EBA classes are included). As individual choice profiles are obtained, an analysis of  how the sociodemographic characteristics affect the consideration of each choice mechanism is conducted.

Since the proposed framework could be interpreted as a general version of the RUM model, the likelihood ratio test is used to assess the advantages of the multiple choice mechanisms approach. In order to prevent overfitting issues, the model is compared with information criteria and cross validation techniques. Cross validation enables to test whether prediction accuracy differences are statistically different.

A significant contribution of the proposed model approach is the explicit characterization of non-compensatory behaviours and the identification of how the individual profile relates to the preference/aversion towards certain attributes. This study also highlight different opportunities for further development both for researchers (e.g. opening the possibilities to use multiple choice mechanisms in the flight travel market) and practitioners (e.g. showing how marginal rates of substitutions are highly dependant of the behavioural assumptions of the analyst).


Denstadli, J.M., R. Lines and J. de D. Ortúzar. Information processing in choice-based conjoint experiments: a process-tracing study. European Journal of Marketing, Vol. 46, 2012, pp.422-446.

Gonzalez-Valdes, F., Raveau, S., 2017. Modelling Air Travel Behaviour with Heterogeneous Choice Mechanisms. Paper submitted to the 96th Meeting of the Transportation Research Board.

McFadden, D., 1973. Conditional logit analysis of qualitative choice behavior, in: Frontiers of Econometrics. New York: Academic Press, New York.

Tversky, A., 1972a. Choice by elimination. Journal of Mathematical Psychology 9, 341–367.

Tversky, A., 1972b. Elimination by aspects: A theory of choice. Psychological Review 79, 281–299.

Wansink, B. and J. Sobel. Mindless eating: the 200 daily food decisions we overlook. Environment and Behavior, Vol. 39, 2007, pp. 106–23.

Conference registration is required in order to view papers.