This paper focuses on the mode choice of workers in Chennai city. The main objectives ofthe study are to investigate potential variation across respondents in the consideration ofalternatives in the choice set; study the preference and response heterogeneity associated withcontextual factors namely, captivity status and the presence of copassengers; examinewhether and how these heterogeneity effects differ based on alternative choice setrepresentations. This study is motivated by the following considerations. Existing studies thataccount for these contextual differences mainly only capture the preference heterogeneity orimpose unrealistic restrictions on the utility of non-automobile modes for joint (copassengers> 0) trips. Disregarding preference or response heterogeneity, when present, can lead to poormodel fit, erroneous forecasts and misleading policy inferences. Second, each of these threesources of heterogeneity has been studied by researchers to some extent, but in isolation. It isplausible that these three contextual effects are inter-dependent, and need to be studiedtogether to disentangle the effect of each. The performance of alternative models with fixed,deterministically variable, and probabilistically variable choice sets is evaluated. Factorsinfluencing the consideration of alternatives are also identified in this empirical context.Unlike many studies, where captivity is represented by a binary variable, behavioraldifferences across three levels of captivity (captive by vehicle ownership, captive by drivingknowledge, and semi-captive) are investigated. Further, the effect of preference and responseheterogeneity is incorporated simultaneously. Finally, the interaction between contextual andchoice set heterogeneity is also empirically analyzed. The results show strong evidence ofchoice set variability and contextual heterogeneity in this empirical context and haveimportant policy implications.
展开▼