International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
Does negative information weight more on activity-travel choices? An empirical study of pre-trip information impact from the perspective of trip-chains
Li Tang

Last modified: 28 March 2017

Abstract


It is a common sense that people’s travel behavior is largely related to their travel environments, including dynamic factors, such as weather, and static factors, such as road networks. A recent study conducted in Beijing announces that with increasing levels of air pollution, the travel mode preferences of private car owners will change dramatically, with many switching to public transit. Big data from the Internet also reveals that the public pays much more attention to bad news than to good news. This phenomenon – that, all else being equal, negative information weighs more heavily on human cognition and decision making than positive information - has been noticed and discussed as the ‘negative effect’ by social psychologists, whose research can be traced back to the 1960s. The understanding of positive-negative asymmetry has been successfully applied to the marketing area, inspiring many valuable research topics, such as the impact of negative word-of-mouth communication (WOMC) on consumer behavior. On the other hand, in the transportation field, dynamic information like weather or the degree of congestion is usually regarded as symmetrically changed attributes. The psychological influence of negative information on travelers’ decision making processes may be underestimated. Considering the revolutionary expansion and huge potential of the Advanced Traveler Information Service (ATIS) system, in-depth and specific research on this aspect of information negativity-positivity is highly needed.

This paper provides some interesting findings about the ‘negative effect’ of travel information based on an empirical study in Chengdu, China. How pre-trip information influences residents’ activity-travel behavior is discussed, involving choices of trip-chain complexity, trip-chain based travel mode, and departure time. A trip chain that is formed by home and only one out-of-home destination is defined as a simple trip chain. Meanwhile, a trip chain connecting home and multiple activity sites is called a complex trip chain. The study of how complex trip chains generate has great significance for traffic demand management (TDM). First, the formation of complex trip chains may lead to an increased use of motor vehicles. Second and more important, there is a significant chance for commuters to add some non-work activities in their commute trip chains (e.g. shopping on the way home from work). In this way, some events that can be done in the off-peak time period are transferred into peak hour, which puts extra burdens on the urban transportation system.

To study trip chain generation for all the activity types, a basic structure of a complex trip chain is built in pairs of subsistence, maintenance and leisure activity. An RP/SP fusion experiment is designed to reduce hypothetical bias. Individuals’ social-demographic characteristic (SDC) and travel-related attributes are set as the RP data. Interviewees are required to recall their perceived travel time, travel cost and travel distance by walk, bus, car and taxi respectively with different trip purposes, which refers to our preset complex trip chain structure. Four bi-level information attributes including weather, estimated travel delay, traffic accident report and traffic regulation on path are set as the SP data. The reason for choosing these attributes is on the basis of past research which studied individuals’ preference among 38 kinds of pre-trip information. Eight situations are generated and divided into four sections by using D-efficient design. Thus, a single respondent needs to make choices under a non-information situation, as well as two situations where they have information. The survey lasted from December 2014 to March 2015. Finally, 1788 observations for trip chain complexity choice, 2823 observations for mode choice and 2855 observations for departure time choice were obtained from the valid sample.

A joint model which is capable of accommodating the information environment itself and multiple information contents simultaneously is used to analyze trip chain complexity and departure time choice. A dummy variable is introduced to represent information presence, the sum of which is defined as information indicator to represent the level of information contact. The impact of one specific information on utility is formulated as the product of information attribute level and its presence indicator (i.e. 1 or 0). The trip chain-based travel mode choice is built as a mixed model combining a rule-based algorithm and a probability-based model. Assume the home-based tour dominates the whole trip chain, and the mode choices of secondary trips (non-home based) will be largely influenced by the choices in the home-based tour. In this way, the hierarchy and linkage of the mode choices between trips are reflected. Using our RP/SP survey data for parameter estimates, calibration results indicate the validity of the models. Weather information is significant in all the relevant choices. Traffic accident report and traffic regulation on path are significant in trip chain complexity and departure time choice, while information of estimated travel delay on path is only significant in departure time choice. This suggests that although travel mode choice is relatively fixed, more work can be done in adjusting trip chain complexity and departure time choice behavior. Moreover, when we look into these changing decisions, it is surprising to find that people are so sensitive to negative information. Taking the impact of weather information on trip chain complexity choice for example, when provided with good weather information, only 12.3% respondents in the valid sample change their choices made in the non-information situation. However, more than 37% of them changed their decisions when informed with bad weather. Furthermore, among the changed choices influenced by negative weather information, nearly 89.4% of them are changing from complex to simple trip chain generation, including canceling non-subsistence activities or breaking a complex trip chain into multiple simple ones. Considering the disadvantages of generating complex trip chain as stated above, it indicates the great potential of using information negativity in TDM.


Conference registration is required in order to view papers.