Despite the proven effectiveness of seatbelt use in reducing traffic casualties, not wearing a seatbelt still contributes to a substantial proportion of fatal crashes worldwide. This problem has raised the need to better understand factors contributing to seatbelt use, particularly in multi-occupant vehicles. Among these factors, behavioural determinants of seatbelt use are difficult to measure and their data are not readily available. These behavioural factors may have shared influences on vehicle occupants, causing their seatbelt use choices to be correlated. These complexities have prevented a comprehensive understanding of seatbelt use choices in the literature. This study aims to fill this gap by developing an econometric model that explains seatbelt use choices in multi-occupant vehicles. A set of binary logit models are constructed for seatbelt use choices and their utilities are correlated across vehicle occupants. A new latent variable representing the unobserved factors or ‘atmosphere’ of the vehicle is then incorporated into the model. The model is empirically tested using seatbelt use data from Tennessee, United States. Results indicate that vehicle body type and time of the day are significantly associated with seatbelt use. In addition, the collective seatbelt use in a vehicle is influenced by the unobserved atmosphere in the vehicle. Age, alcohol and drug consumption, higher proportion of old population and white racial mix, higher income per capita, and higher education levels are factors contributing to this latent atmosphere.
- Bayesian inference
- Integrated choice and latent variable
- Seatbelt use