Open data: nice people can't share!

Talk for FORCE2018 meeting in Montreal, Qc.

One of the most meaningful ways by which scientists can engage with other scientists and with the broader community is by sharing one of their most treasured possessions: data.  Indeed, publishers and funders increasingly acknowledge the importance of data as a scientific output. As such, many journals and funders now require that scientists make their data openly accessible, both to promote transparency and to accelerate the pace of scientific discovery. To date, journal policies that mandate data sharing have successfully increased the accessibility to datasets underlying scientific publications. However, is the quality of these data sufficient to allow reuse and reanalysis? Even when journals mandate data sharing, our survey of the ecological and evolutionary literature found that 56% of open datasets were incomplete, and 64% were shared in a way that partially or entirely prevented reuse. Given the highly competitive nature of academia, authors might be wary of openly sharing their data for fear of criticism or of others benefiting from their work at their expense. As such, ‘scholarly altruism’ is often cited as a key reason for why some authors are willing to share high quality data that are complete and readily reusable. In this talk, I will discuss our latest (unpublished) work to test this hypothesis by assessing how researchers’ psychological motivations and level of cooperation in real-world situations relate to the quality of their shared datasets.