Modeling spatio-temporal diffusion of carsharing membership in Quebec City

During the last few years, car sharing has undergone significant growth, both in Canada and around the world. In this type of service, users share access to a fleet of vehicles, thereby giving them most of the advantages of automobile use, such as its temporal and spatial flexibility, without many of the constraints of ownership. This study analyzes the geographical and socio-economic factors that favour membership of a carsharing service in Quebec City. We combined Cervero's and Kockelman's 5D model (density, diversity, design, distance to transit, and destination accessibility) with Hagerstrand's concept of innovation diffusion so as to analyze the evolution of potential car-sharing membership. Zero-inflated negative binomial (ZINB) regression was used to model the spatial diffusion of the number of car-sharing members in Quebec City from 1996 (two years after its inauguration) to 2008 at the local scale, with an annual time step. Results indicate that the carsharing distribution did, indeed, follow Hagerstrand's innovation diffusion model and that, even though some of the 5D model significantly influenced membership, it was socio-economic factors (education, non-motorization, and family structure) that most greatly affected the membership rate in the service area. The model is used to assess and discuss market coverage potential in Quebec City. (C) 2014 Elsevier Ltd. All rights reserved.