Utility of global positioning system to measure active transport in urban areas

Purpose: The purpose of this study was to determine test-retest reliability of global positioning system (GPS) units for measuring distance traveled when walking and cycling (AT), and to determine whether GPS unit placement influences accuracy. Methods: Participants (N = 19) completed two walking and cycling trials at self-selected speeds on a measured 1489-m course wearing two Garmin GPS units, worn in lanyard and Waistband placements. GPS estimates of travel distance were compared with actual distance, and test-retest reliability was examined. Data-cleaning protocols were developed to remove signal noise. Results are presented for both raw and cleaned data. Results: For both raw and cleaned data, no significant differences were observed between trials (trial 1 vs trial 2), unit placement (lanyard vs waistband), or AT mode (walk vs cycle) (P >= 0.05). Both lanyard and waistband units significantly overestimated distance traveled during walking trials (P <= 0.05), but not cycling trials (P >= 0.05). The relative technical error of measurement (TEM) of the raw data ranged from 3.74 to 15.51%, and average absolute errors ranged from 5.03 to 8.53% for all trials. A significant position by AT mode interaction was observed for clean data (P < 0.05). Relative TEM for the clean data ranged from 1.42 to 1.98%, and average absolute errors ranged from 0.32 to 1.97%. Intraclass correlations (ICC) were poor to fair for all trials using raw and cleaned data. Conclusion: Signal noise during unit initialization may adversely affect unit performance; however, application of data-cleaning procedures to remove data associated with signal noise improves unit ability to measure distance. Results suggest that the lanyard position is the optimal placement for units during data collection.