Abstract
The paper describes the development of statistical models of road pavement deterioration based on rut progression data from 74 test sections in 8 locations over more than 20 years. Visualisation led to deterioration model forms reflecting engineering experience. A full model was first proposed for each test site including all the significant variables. Reduced models were then systematically compared to isolate and remove variables with less than significant contribution to the predictive capability. A single model for all sections selected "traffic loading" as the sole factor. (A)