There is geographical variation in road accident incidence. The variation is spatial and temporal and related to changes in intrinsic risk and risk exposure. Whilst there are road accidents there is a clear imperatve for geographical analysis of road accident incidence that seeks to explain the spatial variation over time.
"Road traffic accidents are an important cause of death and serious injury in developed societies. The pattern of incidence has marked spatial and temporal regularities and the geography of road accidents can inform us better about the nature of the problem and the extent to which traditional solutions may or may not reduce the level of accidents." (Whitelegg, 1986)
Road accidents, especially those that involve personal injury or death are an undeniable problem. In the 1990's it was estimated that 30-40% of all fatal accidents in Britain were road accidents. And in one year, 1997, the cost of road accidents (including; hospital costs, damage to property and vehicles, police and insurance costs, lost output, and a notional sum for pain and suffering) was an estimated £14,814 million (ROSPA, 2000). The number of deaths from road accidents, compared to other causes, is very high, especially for some age groups (Chin and Quek, 1997; Whitelegg, 1986).
Here are some themes I am addressing:
In 1997 I completed a masters in GIS with a dissertation titled An analysis of spatial pattern of road accidents in Leeds using GAM/K and GEM'. GAM/K and GEM were spatial analysis tools being developed in the Smart Spatial Analysis Project by my supervisor Stan Openshaw and his colleague Ian Turton
In March 2000, I began reading for a PhD that had a road safety focus. In this work I was supervised by Ian Turton , Andy Evans and Oliver Carsten
The research was mainly concerned with developing Geographically Weighted Statistics for identifying, visualising and investigating the distribution or clustering of personal injury road accidents as recorded for Great Britain in Stats19 data. The key challenge was to review existing tools and develop new tools to assist the analysis of spatial-temporal-attribute distributions. In addition the thesis was to address the question: How well can models predict road accident incidence for arbitrary geographical regions?
After a year of study I arrived at the following hypotheses to test:
The work focused on the development of tools to analyse changes in the spatial pattern over time. It was hoped that the tools and the results they generate would provide a basis for further efforts to improve road safety.