Explaining heterogeneity in consumers’ preferences towards owning, sharing, and riding Autonomous Vehicles

Autonomous Vehicles (AVs), at the highest level of automation, will strongly influence the way people move, both in the short-term (e.g. mode choice, travel frequency,…) and in the long-term (e.g. residential location, car ownership, …). Several researchers have been dealing with the wide spectrum of impacts of AVs, on environmental sustainability, energy consumption, traffic congestion, and road safety. Based on scenario analysis including assumptions on demand (e.g. modal shares, AVs market penetration, quotas of individual, shared or collective modes, and so on), many researchers are expecting that the widespread use of AVs would bring economic, environmental and social benefits (Anderson et al., 2016) (Clements and Kockelman, 2017) (Milakis, 2019). However, only few authors have tried to assess how likely the expectation about the future of AVs is likely to happen and when. In fact, concerns are growing about AVs and its risks w.r.t. reliability in mixed urban traffic, economic feasibility, legal and insurance liability in the event of accidents, cybersecurity, and privacy of user data (Kyriakidis et al., 2015) (König and Neumayr, 2017) (Zoellick et al., 2019), and it must not be taken for granted neither that the AVs will be accepted as soon as the technology will be ready for the market, nor in which modalities they will be introduced (e.g. private-owned or public vehicles, individual or shared/collective use). The study presented in this paper aims at shedding lights on the individual behavioral aspects that might affect the adoption of AVs, including heterogeneity in preferences according to their socio-economic characteristics, travel habits, lifestyles, and personal attitudes. The research questions are the following ones: "To what extent AVs would lead to changes in users’ travel behavior?", "What are the key factors affecting the mobility demand to own, share or ride AVs?", and finally, “What determinants explain the heterogeneity in consumers’ preferences?”. The methodological approach adopted in this study involved a first stage of data collection through an online questionnaire aimed at gathering both Revealed Preference and Stated Intention (RP/SI) responses, and a second stage of formulation and estimation of behavioral models. In detail, the survey allowed to probe the attitudes and expectations towards AVs of more than 400 potential consumers with different backgrounds and socio-economic characteristics. These data allow for the estimation of Structural Equation Models (SEMs), in order to impute the existence of latent constructs that can explain the random taste variation of individuals, and of Ordinal Regression Models (ORMs), to assess which are the statistically significant variables w.r.t. the preferences expressed by the interviewees through a Likert scale in relation to four stated intentions: desire-to-experience, willingness-to-own, willingness-to-share, and willingness-to-ride AVs. Preliminary results show that individuals’ preferences differ not only for generational and economic factors, in fact the SEM show that it is possible to cluster users on the basis of some aspects of the respondents not directly observable (i.e. latent traits), such as the vocation for technology, the propensity to share goods/services, the pleasure of manual driving, the aversion to Public Transport, and the sensitivity to privacy concerns. Furthermore, the estimated ORMs allowed to identify and quantitatively measure which factors weigh the most, and how these determinants vary significantly as a function of the dependent variables taken into consideration