Article Preview
Top1. Introduction
Modal share in urban transportation has been one of the major concerns as it affects the very functionality and sustainability of transportation system in growing cities. Promotion and use of public transportation is often emphasised, but at the same time excessive private car ownership and usage has been reported to be one of the persistent challenges (Steg, 2005; Kitamura, 1989). Plenty of research have been done to understand the mode choice behaviour and consequent status of the modal share. For instance, some majority of the studies in this area use the common approach of mode choice analysis such as binary and multinomial logit modelling (Mintesnot and Takano, 2007). Other statistical (Scheiner and Holz-Rau, 2010; Anable and Gatersleben, 2005; Lu and Pas, 1999) and qualitative (Beirao and Cabral, 2007; Stone et al., 2003; Hiscock et al., 2002) approaches have also been used to analyse empirical data on mode choice. Studies also have been done with the use of panel data to observe the mode choice behaviour over time (Kitamura, 2009). These different methodologies and approaches help to develop substantial understanding in the subject area. However the literature also point out some shortcomings related to the methodology, and they also indicate directions for further research. Kitamura (2009) argued that the static approach based on cross-sectional data might lead to unreliable analysis results, and thus panel data could be used for better result and understanding of mode choice behaviour over time. Mokhtarian (2009) observed that the analysis of behaviour over time would provide more meaningful insights but there are limitations of panel data analysis. The forecasting-based panel data analysis tool in fact could not accommodate many variables and the relevant dynamic simulation was based on the replication of past behaviour of limited number of variables. On the side of directions for further research, some of the literature indicate the need to include soft variables in the mode choice behaviour studies. Cao and Mokhtarian (2005) stated that in reality the mode choice behaviours are influenced by a large variety of qualitative and experiential variables that are seldom considered and measured. Scheiner and Holz-Rau (2010) and Fujii and Kitamura (2003) emphasised that the travel behaviour would not only be affected by hard objective factors, but it would also be influenced by subjective and psychological factors. Bamberg et al. (2011) recommended for further research in the line of improving the theory of causal determinants of car use. Stern and Richardson (2005) asserted that most of the travel behaviour models lack a cognitive explanatory mechanism explaining the individual’s choice process. There is a sense of consensus that the whole set of individual level behaviour have not been fully understood in the context of transportation (McFadden, 2007). In this paper, an attempt has been made to study the mode choice behaviour at an individual level in the context of a city. The tool of system dynamics has been used in this study (see Sterman (2000), Morecroft (2007) or Azar (2012) for details on the systems thinking and system dynamics approach). The tool helps to develop understanding on the mode choice behaviour from a holistic systems perspective. Especially the tool helps to conceptualise the mode choice behaviour in terms of feedback loops that comprise circular causal relationships between a number of relevant hard and soft variables. The conceptualised feedback loops can be used to develop a simulation model on which experiments can be done to learn why and how one individual would make choice on the mode of transport over time.