Pedestrian fatality and natural light: Evidence from South Africa using a Bayesian approach

- A Poisson-Gamma model on pedestrian fatality is developed.
- Data from the Tshwane municipality with accident time and by road type is used.
- Results indicate that accidents during day time are significantly lower than otherwise.
- We suggest that, relative to other expensive measures, a cheap intervention like effective lighting can lower fatality.

In this paper we use a Bayesian approach to investigate the relationship between pedestrian fatality records from Tshwane and time of fatality. Time of fatality is used as a proxy to reflect the presence of effective lighting, not precluding the presence of any other lighting intervention. In South Africa, for a large proportion (60%) walking is a primary means of transport, with about 45% of all deaths on South African roads being pedestrian. Such reports call for attention to be devoted to analyzing pedestrian fatalities records to locate possible directions of intervention. Results from this analysis reveals that not only does time of day influence pedestrian fatality counts, but also within road types, Municipal roads were the most prone to pedestrian fatalities followed by National roads, while the Regional roads were the least prone to pedestrian fatalities.