Modeling Of Road Traffic Fatalities In Namibia: A Generalized Linear Model Approach

Abstract

Fatalities caused by road traffic accidents have become a major concern worldwide and road traffic accidents happen almost every day on Namibian national roads claiming lives of many road users. It is predicted that by 2030, road crash fatalities would be the fifth most frequent cause of death worldwide. Many efforts have been made to prevent loss of lives and to curb down the MVA expenses on compensations due to road traffic accident; however, road traffic fatalities are still on the increase. Therefore, reliable statistical analysis is required to understand the main determinants of road fatal accidents. The main objective of this study was to identify the best model in modelling Road Traffic Fatalities to establish factors contributing to the number of road traffic fatalities in Namibia.The study was a quantitative cross – sectional study using secondary data of road accidents collected on daily basis by National Road Safety Council (NRSC) for the whole country from January 2012 to December 2013. Descriptive statistics were used to profile background characteristics of the population. The study explored various count models to assess factors influencing road traffic fatalities in Namibia. Six models namely; Poisson, Negative Binomial, Zero- Inflated Poisson, Zero Inflated Negative Binomial, Hurdle Poisson, Hurdle Negative Binomial were explored and their performances were rated based on the AIC. The best model which was the Zero Inflated Negative Binomial (ZINB) was deployed for the estimation of parameters. Results showed that the region with highest number of road accidents from January 2012 to December 2013 with a frequency of (37 %) was Khomas region which is the region with the high population density and road traffic density. Therefore, intervention efforts should focus on more safety awareness campaigns in Khomas region. The ZINB has indicated that, the Road Traffic Fatalities in Namibia over a two calendar year under study were influenced by Khomas region; head on collision and side wipe collision.