One reason for researchers to skip their data sharing responsibilities is the difficulty in determining which datasets should be shared: there is a wide gap between the wording of data sharing policies and the actions required for their particular study. The only way to overcome this obstacle is to get specific and tell researchers exactly which of their datasets they need to share, and where they should go. Unfortunately, stakeholders also find it hard to work out which data should be shared, and it’s also far from clear where in the research cycle stakeholders should focus their open data efforts.
Researchers typically encounter data sharing policies at journals, and hence they are accustomed to providing all of the data associated with a particular manuscript. However, manuscripts may not be the most natural ‘unit’ of data sharing – a unit of research effort for which we try to obtain all the underlying data – and we should consider alternatives.
This BoF session considers the strengths and weaknesses of a range of data sharing units:
Focusing attention on delimiting the units of open data is vital because it addresses the ‘What’ aspect of data sharing, and thus complements the ‘How’ described by the FAIR guidelines. More practically, it enables stakeholders to give their researchers specific expectations around open data at all stages of the research cycle.
Our long term goal is to develop a working group that will formulate a set of guidelines for stakeholders to use when determining whether all of the data associated with a particular unit of research have been successfully made public. This BoF meeting will establish the level of broader interest in this work and lay out the initial steps.