Macroscopic Safety Modelling with Spatial Autocorrelation and Unobserved Heterogeneity

This thesis investigated macroscopic safety models while addressing methodological issues of spatial autocorrelation and unobserved heterogeneity. The thesis examined the application of macroscopic safety models for transportation network screening. A number of advanced statistical and spatial tools were employed to achieve the aim of this thesis. Several planning level factors were identified to influence road safety at the macroscopic level. Addressing spatial autocorrelation and unobserved heterogeneity was found to improve the predictive performance of studied safety models. The developed safety models provide safety practitioners with a basis for the critical application of safety models for hot zone identification and network screening.