Factors affecting safety perception of helmet and seat belt use: a bayesian network approach

Abstract

Major growth in economy over the last few years, alongside the increasing trend of ride sharing applications resulted in a significant increase of motorized vehicles in Bangladesh. With the increase of motorized vehicles in national and regional highways, there has been a spike in the number of road fatalities. Being one of the fastest growing economies, Bangladesh is still lagging behind in terms of ensuring road safety and awareness due to its poor law enforcement and improper management from the authorities. Safety restraints, particularly helmet and seatbelt, have been proven to be highly effective in reducing road fatality and accidents. But unfortunately, the number of applications of such safety restraints is not up to the mark compared to the number of motorized vehicles plying in the highways in this region. Previous research on road accidents and application of safety restraints in Bangladesh mostly comprised of volumetric studies and statistical analysis of road fatalities and trends of road accidents in the region. There is a gap in research regarding the socio demographic factors that influenced one’s perception of helmet or seat belt usage and the dependence of the factors among themselves, especially for the case developing nations such as Bangladesh. The following study seeks to establish a causal relationship among the factors that influence the rider or driver’s usage of helmet and seatbelt respectively, in order to identify the variables that has the most impact on safety restraints application and their impact on other influencing factors. In this study, a causal relationship was established among variables that represent both demographic and safety perception to identify the factors that influenced helmet and seat belt use on national and regional highways in Bangladesh. The network was established based on previous research findings and expert knowledge using Bayesian Belief Network. The Bayesian network was formed using PC algorithm that was based on conditional independence. Data were collected from 32 intersections from national and regional highways. The study was conducted on secondary data provided by Roads and Highways Department (RHD). The Bayesian network was used to find the posterior probability for all variables for the presence of helmet and seat belt respectively using the expectation-maximization algorithm and sensitivity analysis. xii Results showed that effectiveness of helmet as safety restraint, motorbike ownership, and helmet avoidance are the major influential factors for helmet use, whereas education level, license type, perception of accident reduction, and perception of injury for not wearing helmet are the main factors behind seat belt usage. In both cases, police intervention did not have any significant effect on helmet and seat belt usage, indicating poor law enforcement in the region. Results from this study can be a useful resource for policy makers and law enforcement authorities to take necessary actions to encourage use of safety restraints among riders and drivers.