A classic problem in managing scientific research is how to measure the value of scientific outputs, in particular, publications. In practice, administrators and other stakeholders measure the value of scientific publications by how often they are cited by other publications. Yet despite the frequency with which citations and related metrics are used throughout academia, the referencing decisions on which citation counts are based are poorly understood. This project addresses this “missing link” by surveying authors of thousands of scientific papers and asking them why they referenced particular works in their papers, how they were influenced by them, and what they think of their quality. With these data we hope to establish whether the assumptions about how scientists write papers are valid and, consequently, whether citations can be meaningfully used for valuing scientific works.