Log in / Register
Home arrow Economics arrow Bounded rational choice behaviour
< Prev   CONTENTS   Next >

3.3.3. Travel Information and Travel Behaviour

Travel information plays a significant role in influencing trip makers' decisions. Because of incomplete information, trip makers usually sort out not only information sources but also travel alternatives. Consequently, trip makers do not recognise each alternative equally. Due to the incomplete information, the consideration of different alternatives on the choice utility might not be equal. In other words, individuals may show an unequal and asymmetrical evaluation about different alternatives in the choice process. In line with such considerations, Fujiwara, Zhang, and Odaka (2004) and Zhang and Fujiwara (2004) applied the concept of relative utility to evaluate the effects of pre-trip multi-modal travel information on modal change. Fujiwara et al. (2004) built an r_MNL model, defined in Eqs. (3.5) and (3.6), and an r_NL model that follows the modelling approach shown in Section 3.1 but has a three-level nest structure. Zhang and Fujiwara (2004) adopted the r_QNL model that is defined in Eq. (3.11) with a quasi-nested structure, where Eqs. (3.9a) and (3.9b) were further adopted.

In both studies, the same SP survey was adopted, where respondents were asked to choose whether to acquire any travel information, which types of information devices (personal computer, mobile phone and cable TV) in case of information acquisition, and travel mode (car, bus and Astramline (a rail-based transit system)), in the context of commuting and shopping trips. Travel information includes length of road traffic congestion, timetables for bus and Astramline, and total travel time for all the travel modes, where information availability is also included. The level-of-service variables include only travel time for car and bus, and length of car traffic congestion. To reflect individuals' selective behaviours about information contents, respondents were asked to report which information contents they referred to during the choice process. As a result, 1952 valid SP responses were collected from 565 respondents.

In the study by Fujiwara et al. (2004), the accuracy of r_MNL model (i.e. adjusted McFadden's Rho-squared) is 23% higher than that of the MNL model, and that of r_NL model is 7% higher than the NL model, where the r_NL model has the highest accuracy among all the models. Moreover, 7% improvement in the r_NL model accuracy also sufficiently supported the conclusion that by using the concept of relative utility the similarity among alternatives in the nested choice structure can be satisfactorily represented. As for the study by Zhang and Fujiwara (2004), the accuracy of r_MNL model is 23% higher than that of MNL model, but that of r_QNL model is 16% higher than that of r_MNL model.

In the work by Fujiwara et al. (2004), it is found that on average, individuals show the largest interest (its parameter is 0.56) in the Astramline, the second largest interest (its parameter is 0.24) in the car, and the smallest interest (its parameter is 0.20) in the bus. Individuals show almost the similar high interests in cable TV and personal computer (their parameters are 0.50 and 0.43, respectively), however, the lowest interest (its parameter is 0.07) in mobile phone. Considering that 49% of the respondents had mobile phones, the low interest in mobile phone implies that there probably existed higher cognitive barriers in accessing travel information via mobile phone, which was however expected as the most convenient way to access information. Figure 3.1 shows unequal and asymmetric choice structures for travel modes and information devices. For example, the influence of Astramline on car choice is -0.24 while that of car on Astramline is -0.56, the influence of mobile phone on personal computer is -0.43 while that of personal computer on mobile phone is -0.07. Focusing on the existence of heterogeneity, experience with using information devices has a positive influence and age has a negative influence on the choice of information devices, respectively. This suggests that young people and people who have experience accessing information devices are more likely to attach much importance to the choice of information device. A positive parameter for the latent variable attitude for information acquisition suggests that people strongly rely on

The estimated unequal and asymmetric choice structures.

Figure 3.1: The estimated unequal and asymmetric choice structures.

travel information (interest parameter is 0.87) during the choice of travel mode. It is also shown that the higher the frequency of using the Astramline, the larger its interest parameter. However, bus trip frequency works in an opposite direction.

Zhang and Fujiwara (2004) concluded that age, information-accessing attitude and frequency by mode explain the relative interest parameter. Informationaccessing attitude is the factor score obtained from a factor analysis using the collected attitudinal data from the survey, that is the attitudes of travel mode comparison, pre-trip information search, active information acquisition, access of multiple sources of information, dynamic and predicted travel information. The young people show large values of relative interest parameters for information devices. Information-accessing attitude positively influences information acquisition behaviour. The higher the frequency by travel mode, the higher the relative interest parameter for the corresponding travel mode. This supports the assumption of habitual decision-making about travel mode. People prefer to acquire information when making decisions about travel mode choice, since the relative parameters related to information acquisition behaviour are considerably higher than that of nonacquisition behaviour. The r_QNL model estimates the relative parameters of travel modes with smaller variations than the r_MNL model. It seems that the r_MNL model over-evaluates non-acquisition behaviour. The weight parameter related to bus choice group has the highest value with an extremely high /-score. In contrast, those of car and Astramline are almost close to zero. Remembering the quasi-nested choice structure in r_QNL model, it is confirmed that bus choice has the highest negative impact on the choices of car and Astramline, and when choosing the bus, people do not care about the existence of car and Astramline. This suggests that bus users might be more captive than users of car and Astramline.

Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
Business & Finance
Computer Science
Language & Literature
Political science