Validation of Location-Based Service Data for Tracking Tourist Spatiotemporal Behavior in Grand Canyon National Park

Abstract

Third party location-based service (LBS) data has emerged as a prominent resource for measuring human spatiotemporal behavior. However, efforts to measure the external validity of this type of data are limited, especially at smaller scales such as tourist flow within attractions. To address this limitation, we recruited a random sample (n = 539) of visitors to Grand Canyon National Park over two years in May-June 2022 and 2023 to carry GPS devices from their entry to exit. We calculated their rates and sequence of visitation to 34 unique points of interest (POIs), including visitor services and scenic road and trail segments. We repeated this calculation for digital traces within the park boundaries for the same sampling periods using third-party aggregated LBS device data (n = 39,799 devices). Comparisons in spatiotemporal behaviors for both samples were described, including visitation rates to POIs and average stay time in POIs and the park. Movements between POIs were aggregated into flows for both samples by generating directed networks and subsequently analyzed using quadratic assignment procedure (QAP) correlation over 5,000 permutations. We observed that both samples contained near-identical proportions of day-use and overnight visitors (~79%). Relative to the onsite sample, we observed smaller rates of visitation to POIs and shorter stay times in POIs and the park from the LBS sample. A strong correlation in flows was observed (r = 0.8, p < .001). Our results reveal LBS device data may provide helpful high-level insights on tourist flow and trip characteristics but underestimate POI visitation.