Openness in scholarly research: LIS should be leading by open example

Citation

Stuart, D. (2014), "Openness in scholarly research: LIS should be leading by open example", Online Information Review, Vol. 38 No. 6. https://doi.org/10.1108/OIR-08-2014-0179

Openness in scholarly research: LIS should be leading by open example

Article Type: Guest editorial From: Online Information Review, Volume 38, Issue 6.

In recent years there has been a drive for increased openness in scholarly research, but we are still waiting for paradigm-shifting open research to emerge on a large scale. Initially much of the focus has been on open access, but as this has become an increasingly accepted part of scholarly communication (albeit more in word than deed) there is a greater push for other forms of openness: open data, open notebooks, open code and open source science. Of these, open data are not only the most widely applicable, but also potentially offer the opportunity to revolutionise many areas of research as data are shared and reused, potentially automatically. Unfortunately, the benefits to individual researchers (or groups of researchers) can be less obvious, differ considerably between fields, require significant work; as a consequence, the open sharing of data has been widely adopted in just a few fields. The lack of data sharing more widely has been labelled by Borgman (2012, p. 1059) as the “dirty little secret” of open data promotion.

Borgman (2012) offers four rationales for sharing data: so that data/results may be verified, so that the public have access to the results of publicly funded research, so that new questions can be asked of existing data and for the advancement of research and innovation. To these rationales Borrego and Garcia (2013) add the incentive of a possible citation increase. Another potential rationale, to which the Library and Information Science (LIS) community might pay particular attention, is encouraging others to release their data, thereby helping change the norms of science. Whilst the LIS community has probably given the subject more thought than most, that thought has not necessarily led to action.

Motivated by an article in this issue of Online Information Review about data sharing in sociology journals (Zenk-Möltgen and Lepthien, 2014), I examined the current author guidelines of 18 LIS journals in the LIS category on Google Scholar. Author guidelines provide a means of gaining a quick overview of expected behaviour regarding data in a field, as they include information about dealing with supplementary materials. Importantly, the policies are explicit and easily identifiable.

Journal data policies vary considerably, from the non-acceptance of supplementary materials at one end to explicit data requirements at the other. Of the 18 LIS journals considered I found that: 11 continue to have no explicit policy on supplementary materials; one explicitly states that it does not accept supplementary materials; and four state that they accept supplementary material (emphasising that it could enhance a research outcome). Only two have explicit data requirements: the Journal of the Medical Library Association (JMLA) and the International Journal of Digital Curation (IJDC):

Data analyzed for material accepted for publication in the JMLA […] should be retained for at least five years […] so they may be provided, whenever possible, in response to inquiries from interested readers (JMLA, 2014).

[…] If your submission reports results derived from data, that data should be cited […]. If you collected the data, you should deposit it in a custodial environment that gives an appropriate degree of assurance about its longevity. The data should be given a permanent and resolvable identifier and be made openly accessible if possible (IJDC, 2014).

Similar results were found in a far more extensive analysis of 77 LIS journal guidelines published in 2013 (Borrego and Garcia, 2013), where they found that only 26 of the journals had explicit policies on accepting supplementary materials, although many of those that lacked explicit policies nonetheless published such materials. Those supplementary materials were additional methodological explanations and results; notably, they did not “find any example of raw data supplied as supplementary material so as to enable additional analyses to be performed” (Borrego and Garcia, 2013, p. 513).

Of course, journal policies do not necessarily reflect or influence researchers’ actual behaviour. It could be that, despite limited guidance from publishers, most researchers are actively petitioning to submit their data to journals, and where that is not possible they are depositing their materials in the increasing number of online data repositories. But it seems likely that where the journals lead, many researchers will follow, and currently most journals are not leading them anywhere. Reasoning for a lack of explicit data policies and shared data within LIS revolves around fundamental characteristics of the field itself: its interdisciplinary nature and variation in research practices would require a variety of data policies (Cronin, 2013); its small science nature means data are often collected to answer a specific research question (Borrego and Garcia, 2013). But in reality we do not know which data might be reused, and if mandates dictating the sharing of data are considered too extreme, recommendations could nonetheless help to begin nudging researchers in the right direction.

The recent launch of Scientific Data by the Nature Publishing Group, which focuses on the publishing of data descriptors for data sets rather than traditional data articles, demonstrates the wider appetite for the publishing of data and the positive role publishers can have in promoting and facilitating access to the data:

Scientific Data is an open-access, peer-reviewed publication for descriptions of scientifically valuable datasets. Our primary article-type, the Data Descriptor, is designed to make your data more discoverable, interpretable and reusable (Scientific Data, 2014).

The requirements of different fields, and within the same field, will inevitably differ, but if LIS does not even recommend good data practices, it will look increasingly old fashioned. Maybe it is.

There are many players in the scholarly process: publishers, funders, repositories, universities and the researchers themselves. Such a system cannot change overnight, and each of the groups needs to contribute if there is to be a successful open data ecosystem. On current evidence it seems that the publishers have much more to do, and as publishers often provide the same data policies to all their journals, LIS is no more advanced than any other field.

So what is to be done? There is always room for a bit more openness within an individual researcher's work: for some it could be finally putting an article in an institutional repository, whilst for others it could be sharing the code as well as the software they develop. However, leading by example not only requires changes in the behaviour of individuals but also in the scholarly institutions and processes. Cronin (2013) has suggested that the mandating of full open access to data could deter potentially high-impact papers, but I hope a journal's lack of a strong data policy could also deter potentially high-impact papers being submitted to a journal. Researchers should consider the strength of a journal's data policy when choosing a journal in which to publish, irrespective of whether the particular paper they are submitting has an accompanying data set. By favouring those journals with strong data policies over those with weak data policies, researchers can encourage journals to have stronger policies.

There is considerable work to do at every stage of the data sharing process: understanding the data researchers have to share, standardising the processes of sharing data, citing data and establishing metrics to appropriately measure the use and reuse of data. If LIS is to play its part, it needs to be leading more by example.

References

Borgman, C.L. (2012), “The conundrum of sharing research data”, Journal of the American Society for Information Science and Technology, Vol. 63 No. 6, pp. 1059-1078

Borrego, Á. and Garcia, F. (2013), “Provision of supplementary materials in library and information science scholarly journals”, ASLIB Proceedings, Vol. 65 No. 5, pp. 503-514

Cronin, B. (2013), “Thinking about data”, Journal of the American Society for Information Science and Technology, Vol. 64 No. 3, pp. 435-436, available at: http://onlinelibrary.wiley.com/doi/10.1002/asi.22928/pdf

International Journal of Digital Curation (2014), “Submissions: author guidelines”, available at: www.ijdc.net/index.php/ijdc/about/submissions#authorGuidelines (August 12, 2014)

Journal of the Medical Library Association (2014), “JMLA: instructions to authors”, available at: www.mlanet.org/publications/jmla/jmlainfo.html#retention (August 12, 2014)

Scientific Data (2014), “About Scientific Data”, available at: www.nature.com/sdata/ (August 12, 2014)

Zenk-Möltgen, W. and Lepthien, G. (2014), “Data sharing in sociology journals”, Online Information Review, Vol. 38 No. 6, pp. 709-722