Exploring biases in travel behavior patterns in big passively generated mobile data from 11 U.S. cities

Abstract

Passively generated mobile data has increasingly become a crucial source for studying human mobility; however, research addressing potential biases within these datasets remains scarce. This study delves into the critical issue of inherent biases in mobile data, a resource that has transformed the study of human mobility. Using a well-established mobile dataset, we analyze biases in 11 diverse metropolitan statistical areas (MSAs) and spotlight disparities in data quality and mobility metric biases, as compared to the National Household Travel Survey (NHTS). A two-level hierarchical linear regression model unveils the contributing factors to these biases, most notably, data quality, user sociodemographic traits, and city sizes. We further highlight the unexpected introduction of uncertainty by stay-point algorithms during data processing. The findings of our research underscore the necessity of meticulously identifying, understanding, and mitigating such biases in mobile data before its deployment in shaping transportation policies and investments. Ultimately, our study advances our understanding of bias in mobility data, which is a fundamental step towards refining methodologies that can effectively address these biases, thereby enhance the value and accuracy of mobile data in transportation studies.