Characterizing the spatial and sociodemographic sampling biases found in mobile location data through a multiscale analysis of the United States

Abstract Mobile location data are widely used to inform public health, emergency response, and policy decisions. However, these datasets contain sampling biases tied to sociodemographic factors such as age, income, education, and race/ethnicity, which can obscure mobility patterns and reinforce existing inequities. This study advances understanding of underrepresentation in large-scale mobile location data by identifying regional and sociodemographic patterns in the data’s sampling and representation levels. Using a multiscale analysis, we evaluate sampling rates and bias patterns across states, counties, and census tracts. Results reveal consistent underrepresentation of older adults (65+), low-income households, and individuals without higher education, particularly in the western United States and rural areas. In contrast, biases affecting racial and ethnic subpopulations vary by spatial scale, with aggregation masking underrepresentation at finer scales. These findings highlight the importance of scale-sensitive analysis and equity-aware approaches when using mobile location data to inform decision-making.