Developing policy thresholds for objectively measured environmental features to support active travel

Abstract A novel evidence-based methodology is presented for determining place-based thresholds of objectively measured built environment features’ relationships with active travel. Using an innovative machine-learning based Generalized Additive Modeling framework, systematic heterogeneity fundamental to the development of well-justified and objective environmental thresholds is accounted for. The methodology is employed to model an individual’s likelihood of transport walking as a function of environmental factors using California Household Travel Survey linked with comprehensive built environment data. The results reveal strong and complex non-linear dependencies of likelihood of transport walking on environmental features that cannot be quantified using standard threshold detection methods. Thresholds for key environmental features to enhance active travel vary significantly across different socioeconomic groups. Accounting for strong income-based differences in development of environmental benchmarks is emphasized. The thresholds can serve as a useful guiding tool for policymakers, planners, engineers, and public health officials to track existing environmental conditions and healthy behaviors.