Almost 40 years ago, R. J. Smeed published two papers that studied the number of road accident fatalities in 20 countries for the years 1930, 1938 and 1950. One of the relations that he derived (originally for 1938) has been glorified by certain authors with the title 'Smeed's Law'. It has been used by various authors as a criterion with which to evaluate national road safety performance. It is thus timely to study rigorously the validity of this empirical relation, and the results are reported in this paper. Other models are developed on the basis of this initial study, and these provide sound methods for predicting fatality and accident totals. It becomes clear, however, that such models are not sufficiently sensitive to detect any but the largest changes in trends. (Author/publisher)
Abstract