The role of road infrastructures in the usage of bikeshare and private bicycle

Differences in travel behaviors and user attributes exist between bikeshare users and regular cyclists. However, little is known about the preference difference in cycling facilities and other built environment elements for regular cyclists and share-bike users, and how the cycling infrastructure and street pattern factors affect the choice between shared bikes and regular bikes. To address this gap, we applied t-test and mode choice generalized additive mixed model (GAMM) to explore the nonlinear effects of cycling facilities and street patterns on the selection of bikeshare and regular bikes. Additionally, we examined the heterogeneous effects across education levels, distinguishing between high and low education. By analyzing bike travel data from Xi'an, China, the GAMM model results reveal the diverse impacts of built environment elements on bike choices at different education levels. Notably, the average geodesic distance (a street pattern indicator) negatively correlated with the likelihood of choosing a shared bike over a regular one. While high intersection density encourages more bikeshare trips compared to regular trips, it has a more pronounced effect on encouraging highly educated cyclists to opt for regular bikes over bikeshare. Bikeshare riders exhibit a preference for riding in areas with a single type of bike lane. Especially, highly educated bikeshare users are more inclined to cycle in areas with a high proportion of bike lanes separated by lines, while less educated cyclists tend to favor areas with a specific type of bike lane without specifying the exact type. Our study has the potential to provide valuable recommendations for planning of cycling facilities to accommodate a substantial number of bikeshare trips.