Unpacking walkability indices and their inherent assumptions.

Walkability indices are used to characterize the relationship between health and place. Indices make assumptions that affect analysis of the built environment and resulting walkability scores. This study compares three walkability indices created by health researchers focusing on the methods, variables, and walkability scores resulting from differences in definitions and methods. This paper deconstructs the walkability algorithms utilized by each index and rebuilds them in Vancouver, Canada. We find that neighbourhoods in the northern core closer to the downtown area have similar walkability scores across all three indices, while the outer peripheral neighbourhoods with moderate to low walkability have more variation in walkability scores across indices. Most walkability variables - residential density, street connectivity, and land-use - lack a rationale for inclusion, often assumed by researchers. Walkability indices used in health research prove to be incongruent with each other and misrepresentative of actual human behavior. We explore the impact of variable selection and methodologies on indices in the interest of more rigorous health research.