Ordered Probability Modeling of Seat Belt Usage

Abstract
This paper develops a new method to examine the factors that influence driver's seat belt usage. A binary logit model of the driver's beliefs in the effectiveness of seat belts is first developed. The seat belt beliefs probability is then used as an input to a seat belt usage model. Unlike the traditional binary approach, this model is formulated using the ordered probability approach, which allows three levels of choices of seat belt usage (rarely, sometimes, and frequently). The seat belt usage model was estimated separately for urban (short) trips and rural intercity (long) trips using a survey of drivers (385 interviews) conducted in Amman, Jordan. The model results show that drivers with homes located in rural or suburban areas, female drivers, and old drivers strongly believe in the effectiveness of seat belts. The presence of children in the household, vehicle ownership, and serious accident involvement also positively influence seat belt beliefs. Seat belt usage is affected by trip type, distance traveled, past involvement in traffic accidents, time of buckling up the seat belt, and extent of seat belt beliefs. Seat belt usage varies with the type of environment (rural or urban). Although older drivers strongly believe in seat belt effectiveness, their seat belt usage frequency is higher for rural trips than for urban trips. For work driving, the usage frequency was lower for urban trips than for rural trips. However, for both trip environments, the usage frequency is influenced greatly by the seat belt beliefs. The empirical results underscore the appropriateness of the proposed seat belt usage model and provide more insights into the pertinent factors affecting seat belt usage.