Wafaa Saleh and Séona Farrell, Edinburgh Napier University, UK
Abstract: There is increasing research work on travel demand forecasting using different sources of data including stated preference and revealed preference data. There are a number of problems that can be associated with data collection, sampling methods and types of data used in such models. For example, the problem of repeated measurements in stated preference data has generally been recognised in the literature but ignored in practice (Bates & Terzis, 1997). This problem has been investigated and a number of correction factors have been suggested. Other error sources include the scale factor and taste variations. In this paper five randomly selected samples are drawn from a larger data set and used to investigate the reliability of the forecasting models using the segmented models relative to the full data set model. All segments have the same sample size and the same number of observations per respondent. A reliability indicator has been derived to investigate the variations in the coefficient estimates. This indicator shows a range of variations in the estimates. Further research is clearly needed in this area.
Keywords: Stated preference; error sources; forecasting; reliability and correction factors.
