An Empirical Model of Expressway Drivers Behaviors on Stopover at sa/pa involving the Location Choice

Drivers on expressway have not only need of high speed and safety but also other various kinds of services for their comfortable traveling. The Japan expressway company (called NEXCO) privatized in 2005 has increasingly attached great importance to improvement of those services. Service and parking areas (SA/PAs) are consequently required to offer not only traditional rest stop services but also additional services, such as refreshments, dining, and souvenir shopping. Previous researches of driver decision-making on stopover at SA/PA and on which SA/PA is chosen have been hardly advanced because of lack of survey data. These data are clearly useful to identify the factors determining SA/PA choice behaviors and also to develop marketing strategies for promoting SA/PA use. This research aims to analyze driver's decision structures on SA/PA choice behaviors and the level of customer satisfaction with SA/PA services. The web-based questionnaire surveys were conducted twice to the drivers who stopped at the studied SA/PAs on the surveyed period. One was done at the section on Chugoku expressway in 2012 and the other was on Sanyo expressway in 2013. Based on the results from our previous study, in which a structural equation model (SEM) was applied to represent the decision structure underlying SA/PA choice behavior using Chugoku expressway data in 2012, a discrete choice behavior model is developed to identify whether a given driver would stop at any of the studied SA/PAs or not and to identify which SA/PA would be chosen. In this modeling, two different types of the nested logit models, in which the upper level is the decision on stopover at SA/PA or not and the lower level is the one on choice of the location, are mainly calibrated using both the data set of Chugoku in 2012 and those of Sanyo in 2013. The goodness of fit of estimated models including statistical significance of estimated parameters are discussed to clarify the differences in determining factors between Chugoku and Sanyo expressways. The results indicate that both two types of nested logit models are significant in common while the goodness of fit for Chugoku models are different from that for Sanyo models. They also indicate that those differences are caused by driver's characteristics such as travel distance and frequency of expressway use, and by SA/PA characteristics such as the contents of services and the CS (customer satisfaction rate) and the location distribution.