Check the Score: Field Validation of Street Smart Walk Score in Alberta, Canada ( breakout presentation )

Abstract

Walk Score® is an internet-based tool providing a common standard for measuring walkability between communities that is low-cost, easily accessible, and highly generalizable. Many studies have used Walk Score® to demonstrate how walkable communities promote physical activity, reduce obesity and overweight, and lower the risk of chronic diseases. However, Walk Score® has only been field validated for accuracy in metropolitan regions of the United States, leaving a knowledge gap for walkability researchers and practitioners in Canada and internationally. Our research aimed to address this knowledge gap in Canadian walkability research and practice by field validating Walk Score® with street-level observations from three communities situated along a rural to urban continuum in Alberta, Canada. Starting from the methodological outline and results of our field validation, we would like to begin a conversation with conference attendees about the benefits and disadvantages of using tools like Walk Score® to capture walkability and what it might mean for people living in different municipalities and regions internationally. We reverse-engineered 2181 Walk Scores® for locations in Bonnyville, Medicine Hat, and North Central Edmonton using ground-truth data obtained through systematic street-level observations. To conduct the field validation, we calculated Spearman's Rank Correlation Coefficients (rho, rs), accounting for potential inflation of p-values due to spatial autocorrelation by means of the Clifford-Richardson adjustment, and calculated composite means, 95% confidence intervals, and t-tests (α=0.05), as measures of association between Walk Score® and our reverse-engineered index. Walk Score® was successfully field validated across the rural to urban continuum for three communities in Alberta, Canada. Our field validation produced very high Spearman's rho correlations for Bonnyville (rs=0.950, adjusted p<0.001), and high correlations for Medicine Hat (rs=0.790, adjusted p<0.001) and high correlations for North Central Edmonton (rs=0.763, adjusted p<0.001). Our field validation of Walk Score® addressed a knowledge gap ensuring the reliability of this internet-based metric for Canadian walkability research and practice. However, we caution against interpreting walkability solely using Walk Score®, and encourage conference participants to consider its integration with community-based participatory methods to more inclusively capture resident perspectives on walkability across different municipalities and regions.