Modelling Road Accident Injuries and Fatalities in Suratthani Province of Thailand Using Conway-Maxwell-Poisson Regression

In 2013, Thailand recorded the highest number of casualties in road traffic accidents among ASEAN countries and ranked second in the world, as reported in the 2015 World Health Organization survey (36.2 deaths per 100,000 population). Road safety has become a critical problem in the country, especially in provincial areas. Suratthani province has been one of the top ten provinces facing road traffic accidents for many years, with the largest number of accidents in the southern region. In this study, we investigated factors associated with injury and fatality counts per accident using the Conway -Maxwell -Poisson regression model based on data collected of 2,887 accidents in 2015. The six covariates considered were road type, road surface, road section, weather condition, light condition, and accident month. The results showed that the distribution of injury and fatality count data was underdispersed, which is rare. Road type, light condition, and accident month were statistically significant factors associated with count response. The mean injury and death count were 2.87 times higher for national highways than for rural roads when other variables were held constant. Driving at night, such numbers were reduced by a factor of 0.49 with streetlights, compared to those without streetlights. Our findings also showed that, for September as a reference month, human injuries and deaths on the road reduced for January, February, March, and August. These findings could be useful for establishing preventive measures at the road level in this province, and the method can be applied to wider regions.