Modelling Road Traffic Accidents Counts in Tanzania: A Poisson Regression Approach

Road traffic accidents have become serious threats to Tanzanians in recent years. The outcryemanates from the increasing prevalence of negative effects of accidents on human lives,properties, environments and the economy. Poisson regression model was used to study therelationship between road accidents and the factors facilitating them in Tanzania. Count data onyearly road traffic accidents for Tanzania covering the period 1993 to 2019 were used. Due toover-dispersion of Poisson regression model, quasi-Poisson regression model was found the mostappropriate approach for the analysis of these data. Results indicated that all predictors aresignificant under Poisson regression model with p-value less than 0.05 but high speed was foundinsignificant using quasi-Poisson regression model. All factors causing road accidents predictedminor increase of accidents, showing that current control measures on road accidents are likely tobe effective. Keywords: Road accidents; Poisson regression; Over-dispersion; Deviance; Variance inflationfactor.