Association Between User-Generated Commuting Data and Population-Representative Active Commuting Surveillance Data — Four Cities, 2014–2015

Abstract

Creating environments that support all types of physical activity, including active transportation, is a public health priority (1). Public health surveillance that identifies the locations where community members walk and bicycle (i.e., engage in active transportation) can inform such efforts. Traditional population-representative active transportation surveillance incurs a considerable time lag between data collection and dissemination, and often lacks geographic specificity (2). Conversely, user-generated active transportation data from Global Positioning System (GPS)-based activity tracking devices and mobile applications can provide near real-time information, but might be subject to self-selection bias among users. CDC analyzed the association between GPS-based commuting data from a company that allows tracking of activity with a mobile application (Strava, Inc., San Francisco, California) and population-representative commuting data from the U.S. Census Bureau's American Community Survey (ACS) (3) for four U.S. cities. The level of analysis was the Census block group. The number of GPS-tracked commuters in Strava was associated with the number of ACS active commuters (Spearman's rho=0.60), suggesting block groups were ranked similarly based on these distinct but related measurements. The correlation was higher in high population density areas. User-generated active transportation data might complement traditional surveillance systems by providing near real-time, location-specific information on where active transportation occurs.