Abstract
Very few work using repeated cross-sectional travel data has been undertaken in transportation research. This is especially true for travel data gathered every 10 years such as person-trip survey data. In these circumstances, we have confirmed the effectiveness of fixed-effects predicting models unifying zonal (interzonal) heterogeneity and first-order serial correlation of error terms for trip generation and distribution, using three-time points travel data in Hiroshima. However, several problems have still remained concerning interpretation of heterogeneity parameters, effects of different consistent estimates of first-order serial correlation coefficient on parameter estimation and estimation methods of models according to different hypothesis on heterogeneity parameters. This paper deals with these problems for three travel models including modal split for the purpose of developing a new travel demand predicting system considering these two longitudinal factors.
