| Abstract | Open access to data, as a core principle of open science, is predicated on assumptions that scientific data can be reused by other researchers. We test those assumptions by asking where scientists find reusable data, how they reuse those data, and how they interpret data they did not collect themselves. By conducting a qualitative meta-analysis of evidence on two long-term, distributed, interdisciplinary consortia, we found that scientists frequently sought data from public collections and from other researchers for comparative purposes such as “ground-truthing” and calibration. When they sought others’ data for reanalysis or for combining with their own data, which was relatively rare, most preferred to collaborate with the data creators. We propose a typology of data reuses ranging from comparative to integrative. Comparative data reuse requires interactional expertise, which involves knowing enough about the data to assess their quality and value for a specific comparison such as calibrating an instrument in a lab experiment. Integrative reuse requires contributory expertise, which involves the ability to perform the action, such as reusing data in a new experiment. Data integration requires more specialized scientific knowledge and deeper levels of epistemic trust in the knowledge products. Metadata, ontologies, and other forms of curation benefit interpretation for any kind of data reuse. Based on these findings, we theorize the data creators’ advantage, that those who create data have intimate and tacit knowledge that can be used as barter to form collaborations for mutual advantage. Data reuse is a process that occurs within knowledge infrastructures that evolve over time, encompassing expertise, trust, communities, technologies, policies, resources, and institutions. Keywords: data, science, reuse, biomedicine, environmental sciences, open science, data practices, science policy |
