Many funding organisations now encourage or require researchers to share their datasets in full upon publication, for example by linking to a data repository such as Dryad or Figshare. It has been said that data sharing improves transparency, makes it easier to identify findings that fail to replicate, and allows researchers to use existing data to address new questions. Journals have a key role to play in encouraging the sharing of data. However, a new study in PeerJ suggests that only a minority of journals currently require their authors to do so.
The study’s authors, Vasilevsky et al., compiled a list of biomedical science journals that ranked in the top quartile, by either journal impact factor or number of articles published, in 2013, and then classified their data-sharing policies as given in the instructions to authors.
Of the 318 journals identified, only 12% explicitly stated that data sharing was essential for publication. A further 9% required data sharing but did not state directly that failure to comply would affect acceptance of the article. Data sharing was encouraged, but not required, by 23% of journals, while approximately one third did not mention data sharing at all. Journals with high impact factors were more likely to require data sharing than those with lower impact factors. However, Vasilevsky et al. argue that even among journals that did promote data sharing, more could be done to improve the quality and specificity of the guidance provided.