Surveys and models to forecast travel behaviour

Abstract

The research project attack the problem of the bias due to serial correlation in Stated Prefernce data using the Jack-knife and the Bootstrap techniques. A simulation study analyses the influence of the error distribution for each respondent individually and across respondents, and the variance estimate of the model coefficients obtained in the course of Maximum-likelihood estimation. These procedures also have been applied to a real study for the city of Copenhagen (DK) which aims to forecast the travel demand both on the existing transport modes and on the new mini-metro service.

Date of Award	1998
Original language	English