The concept of transferability has been much discussed in recent literature. The study is carefully designed to make an important contribution to these discussions. Data collection methods chosen are a home interview of 1000 households, "number-plate", road surveys and train surveys, which are combined to obtain the most cost-effective modelling, but require separate and detailed treatment of their inherent biases. Estimation methods appropriate for deriving models from combined data sets are described. Finally, consideration is given to the appropriate tests of transferability to be made in the study.
Abstract