This report describes an empirical study investigating the factors influencing the routes drivers choose when presented with alternatives. Sixteen sites at each of which there was a binary route choice, were surveyed for the study. The sites were scattered throughout the UK. Surveys of the route choices being made by drivers were based on the recording of partial registration numbers at pre-defined locations. The estimation of robust and representative journey times by route was based on scatter plots of the journey times of the matched registration pairs against start time. Other characteristics of each route were also measured such as route length and numbers of roundabouts, give ways, zebras and signals. Taking as dependent variables the number of drivers choosing the primary route, generalised linear models were fitted using the GLIM program. These models had a binomial error structure and logit link function, with the linear predictor consisting of terms formed from the route characteristics above. The results from the analyses carried out confirm that a prime determinant is time. More uncertain is the role distance plays in drivers' perceptions as it appears to be correlated with other geophysical variables.
Abstract