Macro and micro pressures in data sharing

Despite predicted benefits of data integration in the sciences, many factors continue to limit data sharing. In this paper we argue that data practices of individual researchers represent a key bottleneck. Through an interview study across multiple fields, we show that data practices vary as a function of norms learned from senior mentors, where data sharing is more likely when supported and modeled by a lab director. Further, through an analysis of deposits to a data repository, we show that data practices are sensitive to global events, where data sharing is more likely in the face of an emergency or crisis. These results suggest that current strategies to achieve data integration, such as policy directives from federal agencies, are poorly aligned with incentives that matter to scientists. We conclude with a summary of influences that are more likely to shape data practices in the direction of greater data sharing.