Higher rates of school switching by students contribute to achievement disparities and are typically theorized as driven by attributes of individual pupils or families. In contrast the neoclassical-economic account postulates that switching is necessary for competition among schools. We argue that both frames fail to capture social-referential and institutional comparisons that drive student mobility, hypothesizing that pupil mobility stems from the (a) student's time in school and grade; (b) student's race, class, and achievement relative to peers; (c) quality of schooling relative to nearby alternatives; and (4) proximity, abundance, and diversity of local school options. Propositions are tested with discrete-time hazard models using data from Los Angeles, including 6.5 million observations. We find the student's position relative to peers, relative school quality, and proximity to local alternatives contribute significantly to the likelihood of switching schools, beyond the effects of individual pupil or family attributes. Implications for understanding "choice" as a social-referential process within diverse organizational fields like urban education markets are discussed.