A survey at 1170 schools in the Sydney metropolitan area in 1978 provided data of pedestrian and vehicular numbers at 6000 street crossings. A Logit model of pedestrian accidents with respect to traffic and crossing type was constructed by the then senior lecturer and statistician at Macquarie University, Dr Ann Eyland. Both the numerical form of the model and the results of using it are of interest. The odds of a pedestrian accident occurring while a person is crossing a street were found to be least where there is a Stop or Give Way sign at an intersection. Generally, mid-block crossings are more dangerous than crossings at an intersection, whether unmarked or marked (zebra), by factor of 1.5, and 1.8 respectively. The benefit of supervision was inconclusive, probably because there were inadequate numbers of supervised crossings in Sydney at the time. The relatively high odds of a pedestrian accident at a signalised intersection which were found to be three times that at a Stop/Give Way sign, could be explained by the tendency to use accidents as a warrant for signals. While signals may reduce accidents, they apparently did not fully compensate for the danger to pedestrians. The hazard was attributed to behavioural factors by implication because traffic factors had been accounted for in the model. A review of subsequent literature has been carried out to assess the validity of the model with some hindsight. The Logit model, although developed 15 years ago, is found to be still relevant today and its form is proposed for an extension of the analysis from specific crossings on school frontages to pedestrian crossings in general. (A)
Abstract