A COMPARATIVE CASE STUDY OF WALKING ENVIRONMENT IN MADRID AND PHILADELPHIA USING MULTIPLE SAMPLING METHODS AND STREET VIRTUAL AUDITS

Abstract

The objective of this study is to quantify, using virtual audits in Madrid and Philadelphia, cross-city differences in the walking environment and to test whether differences vary by sampling method. We used two sampling methods; first, a contiguous area combining census units (~15.000 population area for each setting) was selected using the Median Neighborhood Index (MNI). MNI is a summary index that averages Euclidean distances of sociodemographic and urban form features, used to select the median neighborhood for a given city. Second, we selected a population-density stratified sampling of the same number of census units as above. M-SPACES audit tool was deployed, using street virtual audits to measure function, safety, aesthetics, and destinations along each street segment. Madrid streets had lower scores for function (b=-0.29 CI95% -0.55;-0.31) and safety (b=-0.38 CI95% -0.61;-0.14). Madrid had a greater proportion of streets having at least one walking destination in the street segment (PR=1.92 95% CI 1.55; 2.39). We did not find a significant difference between Madrid and Philadelphia in aesthetics. We found an interaction between safety and sampling methods. This approach can reveal which elements of the built environment account for between-city differences, key to mass influences that operate at the city level.

Keywords: built environment; cities; omnidirectional image; virtual image; walkability.

Figure 1.

Map of the census units selected for the study. The left panel shows Madrid, while the right panel corresponds to Philadelphia.

Figure 2.

Predicted values of the mixed-effects regression models comparing street function, safety, aesthetics and destinations between Madrid and Philadelphia and their interactions with the sampling method (MNI vs stratified sampling) (N=936 streets).