Examining methodological issues on combined RP and SP data

Abstract

Abstract Revealed preference (RP) and stated preference (SP) data have been widely used in transportation studies to understand user's preferences regarding various travel decisions. This paper focuses on investigating the modeling techniques to address various issues associated with pooled estimation using both RP and SP data, such as the scale variances, the user heterogeneity, and the state dependency effect, etc. Various model structures are explored, and the results are compared in terms of the significance level of the key parameters, the model performance, and the implications on the derived values of (Value of travel Time) VOT and (Value Of travel time Reliability) VOR. The model results indicate that when inter-alternative scale parameters are incorporated, the model performance and fit improve significantly, with much more reasonable state dependency factors, while the error components and the standard deviations for the random parameters in Mixed Logit (ML) model lose their influence. It suggests that the inter-alternative scale parameters are able to capture the unobserved heterogeneity across alternatives and individuals and relax the Independence of Irrelevant Alternatives (IIA) of Multinomial Logit (MNL) models. This paper contributes to the literature by adding empirical studies and providing insights on the modeling approaches to handle joined dataset, particularly in the field of examining VOT and VOR based on joint RP and SP survey data.