Audit tools are useful for exploring the urban environment and its association with physical activity. Virtual auditing options are becoming increasingly available potentially reducing the resources needed to conduct these assessments. Only a few studies have explored the use of virtual audit tools. Our objective is to test if the Madrid Systematic Pedestrian and Cycling Environment Scan (M-SPACES) discriminates between areas with different urban forms and to validate virtual street auditing using M-SPACES. Three areas (N = 500 street segments) were selected for variation in population density. M-SPACES was used to audit street segments physically and virtually (Google Street View) by two researchers in 2013-2014. For both physical and virtual audits, all analyzed features score significantly different by area (p < 0.05). Most of the features showed substantial (ICC = 0.6-0.8) or almost perfect (ICC ≥ 0.8) agreement between virtual and physical audits, especially neighborhood permeability walking infrastructure, traffic safety, streetscape aesthetics, and destinations. Intra-rater agreement was generally acceptable (ICC > 0.6). Inter-rater agreement was generally poor (ICC < 0.4). Virtual auditing provides a valid and feasible way of measuring residential urban environments. Comprehensive auditor training may be needed to guarantee good inter-rater agreement.
Keywords: Omnidirectional image; Physical activity; Urban environment; Validation studies; Virtual image.