Ridesourcing mode choice: a latent class choice model for UberX in Chile

Abstract

As shared mobility expands, ridesourcing has become its most popular manifestation. However, users' mode choice has not yet been sufficiently explored. Thus, this study aims to model ridesourcing mode choice across different latent classes to ascertain who chose ridesourcing and why. We conducted a mode choice study by collecting revealed preference surveys from UberX users in Viña del Mar, Chile, in 2017. We then determined the existence of two latent classes and modeled the mode choice using a latent class choice model. Ultimately, we characterized individuals belonging to each latent class and calculated the subjective value of time (SVT). Most UberX users were highly educated and aged 20?35 years. Further, UberX gained users principally from public transport (80%). Likewise, the two latent classes differed by socioeconomic characteristics and SVTs. A latent class grouped the highest-educated and highest-earning users, who also offered the highest SVT. In summary, two latent classes, differentiated by educational level and income, formed the ridesourcing market. Besides, they offered distinct ridesourcing choice behavior based on the widely dissimilar SVTs. There was also a strong substitution effect between ridesourcing and transit use. The results imply that policymakers and transportation planners could have increased the competitiveness of the public transit system by improving rapidity and safety, having room to increase the fares to defray the improvements. Further, they could have used information related to the latent classes to customize relevant policies and marketing strategies (routes, frequency, fares, etc.) for every latent class.