The influence of neighborhood environment and household travel interactions on school travel behavior: an exploration using geographically-weighted models

•Logit models of mode choice for trips to school in Toronto, Canada were estimated.•Spatial autocorrelation was explored by including an auto-covariate parameter.•Built environment near home, school and along travel route was important.•Predicted built environment effects were less accurate in some neighborhoods.•Availability of adults and absence of siblings increased the likelihood of driving.

Professional and popular interest in active school transportation (walking and cycling) is matched by an emerging literature on this topic. This paper explores school travel behavior of 11-year old children in Toronto, Canada. In particular, the effects of the neighborhood environment and caregiver-child travel interactions on travel mode choice were studied. Results indicate that the built environment near both home and school locations was associated with the odds of walking. However, predicted built environment effects were less accurate in some neighborhoods. Availability of adults at the time of school travel likely encouraged driving. School transportation interventions that broadly consider school and neighborhood-oriented policies and enable independent mobility may increase walking rates. Presence of spatial autocorrelation in the prevalence of walking suggests that more research is required to understand inter-household similarities in behaviors that are spatially structured.