Development and reliability of a streetscape observation instrument for international use: MAPS-global

• Main Authors: Kelli L. Cain, Carrie M. Geremia, Terry L. Conway, Lawrence D. Frank, James E. Chapman, Eric H. Fox, Anna Timperio, Jenny Veitch, Delfien Van Dyck, Hannah Verhoeven, Rodrigo Reis, Alexandre Augusto, Ester Cerin, Robin R. Mellecker, Ana Queralt, Javier Molina-García, James F. Sallis
• Format: Article
• Language: English
• Published: BMC 2018-02-01
• Series: International Journal of Behavioral Nutrition and Physical Activity
• Subjects: Exercise; Walking; Built environment; Walkability; Measurement
• Online Access: http://link.springer.com/article/10.1186/s12966-018-0650-z

Abstract Background Relationships between several built environment factors and physical activity and walking behavior are well established, but internationally-comparable built environment measures are lacking. The Microscale Audit of Pedestrian Streetscapes (MAPS)-Global is an observational measure of detailed streetscape features relevant to physical activity that was developed for international use. This study examined the inter-observer reliability of the instrument in five countries. Methods MAPS-Global was developed by compiling concepts and items from eight environmental measures relevant to walking and bicycling. Inter-rater reliability data were collected in neighborhoods selected to vary on geographic information system (GIS)-derived macro-level walkability in five countries (Australia, Belgium, Brazil, Hong Kong-China, and Spain). MAPS-Global assessments (n = 325) were completed in person along a ≥ 0.25 mile route from a residence toward a non-residential destination, and a commercial block was also rated for each residence (n = 82). Two raters in each country rated each route independently. A tiered scoring system was created that summarized items at multiple levels of aggregation, and positive and negative valence scores were created based on the expected effect on physical activity. The intraclass correlation coefficient (ICC) was computed for scales and selected items using one-way random models. Results Overall, 86.6% of individual items and single item indicators showed excellent agreement (ICC ≥ 0.75), and 13.4% showed good agreement (ICC = 0.60–0.74). All subscales and overall summary scores showed excellent agreement. Six of 123 items were too rare to compute the ICC. The median ICC for items and scales was 0.92 with a range of 0.50–1.0. Aesthetics and social characteristics showed lower ICCs than other sub-scales, but reliabilities were still in the excellent range (ICC ≥ 0.75). Conclusion Evaluation of inter-observer reliability of MAPS-Global across five countries indicated all items and scales had “good” or “excellent” reliability. The results demonstrate that trained observers from multiple countries were able to reliably conduct observations of both residential and commercial areas with the new MAPS-Global instrument. Next steps are to evaluate construct validity in relation to physical activity in multiple countries and gain experience with using MAPS-Global for research and practice applications.