A study of work-based travel in Victoria
Date
1985
Authors
Hogan, Patrick Edward
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This thesis documents the development and application of a disaggregate mode choice model calibrated for Victoria, B.C. A major objective of this work is to show that state-of-the-art methods in mode choice forecasting can be undertaken in smaller and medium sized cities of well under half a million population. This objective has been made possible by using a smaller cost effective data base where quality data is the important factor compared with traditionally voluminous and expensive data bases associated with the transport studies of the late 1950's and '60's. The advent of smaller transport data bases affords the use of smaller more accessible computers including full optioned micro computers.
The mode choice model developed is a binary choice LOGIT model using maximum likelihood estimation to derive model coefficients. The dependent variable is the binary choices of transit and auto modes. Six explanatory variables were found to be most useful and were included in four final model designs.
The LOGIT model was calibrated using an origin-destination home interview survey conducted in Victoria during the summer of 1977. The data base was restricted to work-based trips between home and work producing 623 trip records amongst 215 sampled households .
The useful independent variables in order of their significance in the main model are car availability which is based on the household's car ownership and the number of employed persons. The second is relative level-of-service expressed as the ratio between the levels-of-service for transit and auto modes. The third is the out-of-vehicle travel time which is calculated as the difference between the out-of-vehicle times for transit and auto. The fourth variable is the relative out-of-pocket costs of transit and auto. The fifth and sixth variables are age and sex taken directly from the 1977 travel survey. All six variables were found to have satisfactory significance or better for all four model designs. The model designs included a general work-based model, a peak-period work-based model, a downtown-based model and a non-peak, non-downtown work-based model.
An important aspect of the model development is the application of the model as a key part of the transport forecasting process. A scenario analysis is designed to illustrate the model's use in transit planning. In this context it is shown that an optimization problem can be solved by comparing transit options on service levels and various fare charges using the model's estimates. A trip generation and distribution component is included in the example application to best illustrate the practical use of a disaggregate mode choice model in transit planning.