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

  • Thadei Sagamiko
  • Nyimvua Mbare
Keywords: Road accidents; Poisson regression, Over-dispersion; Deviance, Variance inflation factor.

Abstract

Road traffic accidents have become serious threats to Tanzanians in recent years. The outcry
emanates from the increasing prevalence of negative effects of accidents on human lives,
properties, environments and the economy. Poisson regression model was used to study the
relationship between road accidents and the factors facilitating them in Tanzania. Count data on
yearly road traffic accidents for Tanzania covering the period 1993 to 2019 were used. Due to
over-dispersion of Poisson regression model, quasi-Poisson regression model was found the most
appropriate approach for the analysis of these data. Results indicated that all predictors are
significant under Poisson regression model with p-value less than 0.05 but high speed was found
insignificant using quasi-Poisson regression model. All factors causing road accidents predicted
minor increase of accidents, showing that current control measures on road accidents are likely to
be effective.
Keywords: Road accidents; Poisson regression, Over-dispersion; Deviance, Variance inflation
factor.

Published
2021-02-14
Section
Articles

Journal Identifiers


eISSN: 2507-7961
print ISSN: 0856-1761