Michal Tal-Socher,1 Adrian Zideman1,2
Data sharing is promoted in different avenues, a powerful one being the scholarly publication process. The research examines interdisciplinary differences in journal data-sharing policies and, to a smaller extent, data-sharing policies of major publishers.
The websites of a sample of academic journals and the websites of major journal publishers were examined for information on policies with regard to data sharing (as retrieved from October 2015 to December 2015). The journal sample was selected from 15 disciplines, drawn from 5 main academic discipline categories: biomedical sciences (including life sciences and medicine); physical sciences (including natural sciences without life sciences); social sciences (including economics, psychology, and political science); arts and humanities; and formal sciences (including cognitive science and statistics). Each disciplinary journal listing is ordered according to Scientific Journal Rankings score, and 10 journals were selected from each discipline; the total number of journals sampled is 150. Unlike the usual research strategy in the field, we opted for a broader sample distribution among different disciplines at the expense of depth. For each journal, the journal’s website was searched to see whether there is a text instructing authors on the sharing of data supporting the research. Two definitions of data sharing were adopted for this purpose: enabling data sharing, defined as the sharing of academic article-related research data on an open digital platform, and strong data sharing, where at least some types of data must be deposited for open sharing as a condition for publication.
The central results, relating to journal policies in the 5 discipline categories, are shown in the Table. The results also show the importance of major publishers in promoting data sharing. The 5 leading journal publishers account for 56% of the journals in our sample that enable sharing (and 46% of the strong policy journals), although they publish only one-third of the sample journals. The paper concludes with a presentation of the current preferences for different data-sharing solutions in different fields (ie, specialized repositories, general repositories, or publishers’ hosting area).
Assuming that journal and publisher policies are an important indicator of actual data sharing, the results consolidate the notion of the primacy of biomedical sciences in the implementation of data-sharing norms, the lagging implementation in the arts and humanities, and similar levels of norms adoption in the physical and social sciences. The results also point to the overlooked status of the formal sciences, which demonstrate low levels of data-sharing implementation. However, other tools for encouraging data sharing exist that may be stronger than publication policies in less journal-centric disciplines.
1Consultant, Committee on Publication Ethics, Harleston, UK, firstname.lastname@example.org; 2Economics Department, Bar-Ilan University, Israel
Conflict of Interest Disclosures:
The research was funded by the Committee on Publishing Ethics and conducted within the framework of its Research Subcommittee (Chair: Adrian Ziderman).
The conclusions and views expressed are those of the authors and not necessarily of COPE.