• Collins Aondona ORTESE
  Department of Mathematics and Computer Sc., Benue State University, Makurdi, Nigeria
  




• Terna Godfrey IEREN
  Department of Mathematics and Computer Sc., Benue State University, Makurdi, Nigeria
  




• Edwin Hart OGWUCHE
  Department of Mathematics and Computer Science, Federal University of Health Sciences, Otukpo, Benue State - Nigeria
  

Road accidents have been one of the leading causes of mortality in Nigeria.rn Some identified risk factors responsible for this ugly trend are; errors from thern driver, number of persons involved, age, vehicles safety and road conditions.rn The aim of this study is to estimate the effect of these factors on road-relatedrn fatalities. The Bayesian Simulation Modeling Approach was employed torn estimate these parameters. This was done using the Markov Chain Monte Carlorn Algorithm implemented on the Windows Bayesian Inference Using Gibbsrn Sampler platform. The gamma and normal prior distributions were assumedrn and the Poisson likelihood as the preferred distribution to estimate the posteriorrn parameters. A Random sample of 32 accidents reported cases at Federal roadrn safety office in Makurdi town was retrieved and used as training dataset for thern Algorithm. Results show that the coefficients for thern parameters; drivers age, mechanism of mobility, driver error, road conditionrn and unsafe vehicle shows a positive significant affect on the fatality rate due torn accident. An increase in the number of persons conveyed on a particular transportrn mechanism increases the expected death rate due to accident by 8%.rn Furthermore, there will be an increase in fatalities caused by driver error, roadrn condition, and unsafe vehicle by 52%, 72% and 99% respectively. Furthermore,rn the expected deaths due to accident on motorbike is twice (0.52) as much as thern corresponding deaths on motor vehicle with same condition of road, humanrn factors and unsafe vehicles. It was also observed that increase in drivers’ agern decreases the rate of deaths due to accident by 1%. Visually inspecting thern history plot has confirmed the model convergence. Density plot reflect thern target distribution which further validates the prior distribution selected.rn Deviance Information Criteria (DIC) of 65.439 verified the model fit andrn adequacy. It was concluded that Bayesian modeling via simulation is suitablern for estimating stochastic parameters in the face of scarce and incomplete data.