Abstract
Abstract
This paper extends the author's earlier statistical analyses of fatal train accidents on the national railway system of Great Britain to include data on speeds and further data on the rolling stock of the trains involved in the accidents. The objective is to explore whether such information is useful in improving projections of fatalities. It is found that the mean number of fatalities in accidents involving loaded or empty passenger trains does indeed rise with impact speed, with different estimated relationships for multiple unit and locomotive hauled passenger stock. However, the actual numbers of fatalities in individual accidents are very variable. An unexpected finding is that there has been only a slight upward trend over time in the mean speeds of trains in accidents; therefore there is no strong reason to expect increased mean fatalities on this account. A direct estimate of the trend in mean fatalities per accident over time shows that this has been almost flat; for that reason the author's preferred projection is now to assume that mean fatalities per accident will remain constant at the overall mean in the data, currently 4.0. This gives projections about 10 per cent higher than the author's most recent previous projections.
Get full access to this article
View all access options for this article.