Using latent attitudinal variables estimated through a structural equations model for understanding carpooling propensity

<title>Abstract</title> Carpooling systems have never been able to achieve significant reductions in the use of private vehicles. The psychological barriers associated with riding with nonacquaintances and losing the flexibility of using one's private vehicle as a single occupant are often referred to. However, the effect of these issues is hard to determine as they reveal subjective attitudes not easy to quantify. In order to measure these attitudes a survey was conducted in Lisbon (Portugal). In this survey, we collected data for several attitudinal variables and conducted a stated preference experiment comparing driving alone/with family with external carpooling. A simultaneous estimation of the latent attitudinal factors and the probabilities of choice were conducted through a structural equations model. The model results show that attitudes play an important role in the decision to carpool, and they also act as mediators from socio-demographic characteristics. The negative aspects of carpooling may be overcome by the positive aspects; however, our results also stress the difficulties associated with changing from an acquaintance-based carpool to a broader system which may hinder the possibility of expanding the number of carpoolers in big cities.