Hazard-based duration modelling represents a promising approach for examining duration processes in which understanding and accommodating temporal dynamics is important. At the same time, hazard models are sufficiently flexible to handle censoring, time-varying covariates, and unobserved heterogeneity. There are several potential areas of application of duration models in the transportation field. These include the analysis of delays in traffic engineering (e.g., at signalised intersections, at stop-sign controlled intersections, at toll booths), accident analysis (i.e., the personal or environmental factors that affect the hazard of being involved in an accident), incident-duration analysis (e.g., time to detect an incident, time to respond to an incident, time to clear an incident, time for normalcy to return), time for adoption of new technologies or new employment arrangements (electric vehicles, in-vehicle navigation systems, telecommuting, for example), temporal aspects of activity participation (e.g., duration of an activity, travel time to an activity, home-stay duration between activities, time between participating in the same type of activity), and panel-data related durations (e.g., drop-off rates in panel surveys, time between automobile transactions, time between taking vacations involving intercity travel). In contrast to the large number of potential applications of duration models in the transport field, there have been surprisingly very few actual applications. Hensher and Mannering (1994) provide a comprehensive review of applications before 1994, and so attention is confined here to applications after 1993. In this post-1993 period, almost all applications have been in the area of activity and travel-behaviour analysis.
Abstract