Over the coming decades, a significant increase in the numbers of elderly people requiring travel will occur as the demographic profile of Canadians shift thereby affecting all aspects of transportation demand. Furthermore, cohort effects are anticipated which may see tomorrow's elderly leading more active lives and travelling to more activities than today's aged. The current lack of a detailed description of elderly travel characteristics and behaviours, particularly one that examines the issue at a level involving activity engagement, was a deficiency addressed by this research. A further product of this study was the development and testing of a simplified activity-based modelling framework. The framework was designed to describe elderly travel characteristics and demand with the added benefit of providing a tool that can evaluate transportation related impacts of proposed policies.

Comparisons of activity participation of the elderly with younger age groups showed that although the daily number of activities remains relatively constant, beginning around age 75 there is a significant decrease in the number to which they travel. There are also significant changes in the types of activities to which the elderly travel compared with the younger age groups. Furthermore, the daily number of trip tours was shown to increase for those 65 to 75 years of age before it steadily declines with advancing age. The average number of activities accessed in each trip tour was found to decrease significantly beginning at about age 65.

Having been traditionally addressed as a relatively homogeneous group by transportation planners, the elderly were shown to possess extremely varied characteristics. Cluster analyses were undertaken to identify subpopulations of the elderly from a sample of 1,150 who responded to an activity-based survey conducted in Portland, Oregon. To identify different lifestyle groups, exploratory analyses were undertaken to delineate clusters based on socio-demographic, travel, and activity engagement variables. The final cluster solution chosen to provide a categorical basis for the modelling framework identified six distinct lifestyle groups based on socio-demographic variables. These clusters were also found to have statistically significant differences in travel behaviour and activity engagement patterns. The clusters identified are characterized as those who remain active in the workforce, the mobility impaired, the elderly who live with their grown offspring, the disabled who drive, and those who either live alone or with a spouse and continue to drive.

The activity-based model was developed using discrete-event, stochastic simulation (or microsimulation) as a platform. Through a sequential process, the model stochastically assigns individuals with a daily itinerary of activities. Trip tours are estimated based on the type and quantity of activities requiring travel. All model assignments are conditioned on each individual's cluster membership. Although the model is operationalized at a relatively rudimentary level, it provides a base structure that can be enhanced in subsequent versions.

The model framework successfully replicated all facets of the base data set used for its development. Elements of travel behaviour synthesized for individuals being modelling included total daily activities (with and without travel), activities engaged in by class (with and without travel), total daily trip tours, and mode splits. Comparing model outputs with observed base data, both the number of activities requiring travel and the total daily trip tours were overestimated by 3.7 percent for all of the elderly combined. The travel model was also applied to a smaller external data set (data from a different study area not used for model development) for validation. The number of activities requiring travel and the number of trip tours were overestimated by 9.2 and 10.5 percent, respectively. Differences between model outputs and observed values are the combined result of the stochastic nature of the modelling framework, aggregation effects (i.e., assigning individuals to clusters with predefined characteristics), model inaccuracies (e.g., use of regression models to predict the number of trip tours), and an incomplete set of constraining rules which govern daily activity itineraries.

Two test applications of the model explored its ability to evaluate the impacts of a road pricing policy and a mandatory license retesting program on the different segments of the elderly. Results from a stated-adaptation survey for road pricing were used to modify the underlying empirical distributions imbedded in the base model. The model was rerun and the results compared with the original outputs. The analysis allowed the varied impacts of increased travel costs to be compared between the six elderly lifestyle clusters. This first test application illustrated the importance of having a statistically significant sample from a stated-response survey to represent each lifestyle cluster. Future applications should rely on stratified sampling techniques for stated-response surveys.

The second test application examined the potential impacts associated with the implementation of a mandatory relicensing program for those older than 80. Given that the clusters were delineated based on several general socio-demographic variables, the model was not able to isolate fully the activity and travel patterns of this target group based only on age and driver's license variables. The test case reinforced the importance of defining clusters based on the end use of the model. For specific uses of the model, defining clusters on dimensions other than general socio-demographic variables will sometimes be necessary.

The research has provided a more comprehensive understanding of the varied lifestyles, activity patterns, and subsequent travel behaviour and needs of the elderly. Furthermore, it has been shown that a categorical approach using lifestyle groups with unique activity and travel characteristics can be successfully combined within an activity-base framework. Although this approach was applied specifically to the elderly, it can be extended to other heterogeneous groups including the population as a whole. The successful development and validation of a simplified activity-based model have given this field of study a much needed demonstration of an operational activity-based modelling framework. It has been shown that even a simplified framework can synthesize the linkages between activity patterns and corresponding trip-making.