Development of Models for Understanding Causal Relationships Among Activity and Travel Variables

Understanding joint and causal relationships among multiple endogenous variables has been of much interest to researchers in the field of activity and travel behavior modeling. Structural equation models have been widely developed for modeling and analyzing the causal relationships among travel time, activity duration, car ownership, trip frequency and activity frequency. In the model, travel time and activity duration are treated as continuous variables, while car ownership, trip frequency and activity frequency as ordered discrete variables. However, many endogenous variables of interest in travel behavior are not continuous or ordered discrete but unordered discrete in nature, such as mode choice, destination choice, trip chaining pattern and time-of-day choice (it can be classified into a few categories such as AM peak, midday, PM peak and off-peak). A modeling methodology with involvement of unordered discrete variables is highly desired for better understanding the causal relationships among these variables. Under this background, the proposed dissertation study will be dedicated into seeking an appropriate modeling methodology which aids in identifying the causal relationships among activity and travel variables including unordered discrete variables. In this dissertation, the proposed modeling methodologies are applied for modeling the causal relationship between three pairs of endogenous variables: trip chaining pattern vs. mode choice, activity timing vs. duration and trip departure time vs.mode choice. The data used for modeling analysis is extracted from Swiss Travel Microcensus 2000. Such models provide us with rigorous criteria in selecting a reasonable application sequence of sub-models in the activity-based travel demand model system